get_x_y_z_h_mat_from_obj computes z from bound_box[0], which differs from obj.location.z when the representation is centered at origin (e.g., a PolygonalFaceSet unit cube). The previous fix (24f7629f) removed translate_obj_to_z_location to prevent doubling Z for extrusion representations, but that broke centered ones where z != location.z.
Replace the removed relative translate with an absolute active_obj.location.z = z. This is a no-op for extrusion representations (z == location.z) and corrects the position for centered ones.
Add test_regenerate_space_from_centered_cube_representation to cover the regeneration path with a centered mesh representation.
Generated with the assistance of an AI coding tool.
Add explicit import bonsai.core.tool to satisfy pyright's type checker,
which requires submodules to be explicitly imported rather than relying
on transitive imports from import bonsai.
All 8 space-generation tests now create IFC walls/slabs with real
solid-block representations using IfcExtrudedAreaSolid, instead of
relying on the old Blender-mesh bisection path (broken since 79ee88da5
switched to IFC-geometry-only for boundary detection).
- _BlockHelper provides create_wall (10x10xheight block) and create_slab
(12x12x1.0 block) helpers using standard IfcOpenShell API calls.
- The wall block bisects to a 10x10 polygon at the cutting plane
(matching the old cube-behaviour), and auto-height detects wall_top_z.
- Pre-existing height assertions (z=10) now pass correctly because
auto-height = wall_top_z - base_z = 10 - 0 = 10 (the old values were
wrong for the Blender path where h defaulted to 3).
- test_regenerate_after_wall_height_change modifies the IFC extrusion
depth directly and bumps the geom cache token via
_bump_geom_cache_token() instead of relying on Blender depsgraph.
- No Blender cubes are created except when absolutely needed for
selection/active-object flow (regeneration, apply-height).
- Added ifcopenshell.util.representation to imports.
- Import _bump_geom_cache_token from bonsai.tool.spatial.
Generated with the assistance of an AI coding tool.
Covers all three covering operators with success and error-path tests
using the Prophecy mocking framework. The key assertion verifies that
get_space_polygon_from_context_visible_objects' return value is unpacked
so the polygon (not the tuple of polygon+bounding_elements) reaches
set_covering_representation_from_polygon.
Shapely geometry objects are not JSON-serialisable (Prophecy call
serialisation), so we use the plain integer 42 as a stand-in for the
polygon value.
Generated with the assistance of an AI coding tool.
Three covering core functions (add_instance_flooring_covering_from_cursor,
add_instance_ceiling_covering_from_cursor, regen_selected_covering_object)
used the old single-value assignment from
get_space_polygon_from_context_visible_objects, which now returns a
(polygon, bounding_walls) tuple. The isinstance(str) guard never fired,
causing the tuple to flow into set_covering_representation_from_polygon
and raise a shapely error.
Fix by unpacking space_polygon, _ at all three call sites.
Generated with the assistance of an AI coding tool.
Move Blender-independent space generation algorithms from Bonsai
(GPL) to ifcopenshell.util (LGPL):
- ifcopenshell.util.shape.bisect_mesh_plane_vf: vectorized numpy
triangle/plane intersection for mesh bisection
- ifcopenshell.util.element.iter_top_connections: walker for
IfcRelConnectsElements(TOP) relationships
- ifcopenshell.util.space: new module with get_boundary_lines,
get_space_polygon, get_auto_space_height and height detection
helpers — all operating on IFC geometry without Blender
Bonsai's tool/spatial.py now delegates to these utilities via
thin wrappers, keeping only Blender-specific concerns (cache
management with depsgraph invalidation, UI property reads).
tool/wall.py iter_wall_slab_connections delegates to
ifcopenshell.util.element.iter_top_connections.
Added 22 tests: 6 for bisect_mesh_plane_vf, 10 for space
generation algorithms, 4 for iter_top_connections, 2 Bonsai
integration tests for cache behavior.
Generated with the assistance of an AI coding tool.
Space height is now auto-detected using IFC geometry directly
(ifcopenshell.geom.create_shape + get_shape_bottom/top_elevation)
instead of Blender object bounding boxes. This fixes height detection
when the slab above is not loaded in Blender.
Detection priority:
1. IfcRelConnectsElements(TOP) connections on bounding walls
2. IfcSlab / IfcRoof elements above with XY overlap to space polygon
3. Minimum wall top Z of bounding walls
4. Fallback to space_height property (default 3m)
Added space_height and force_space_height properties to
BIMSpatialDecompositionProperties. The height field is synced to
the active space's height via active_object_callback (msgbus), not
in draw().
Added ApplySpaceHeightToSelection operator to modify
IfcExtrudedAreaSolid.Depth in place without regenerating footprint.
bounding_walls changed from list[tuple[element, obj]] to
list[entity_instance] since Blender objects are no longer needed.
Generated with the assistance of an AI coding tool.
Add a paste button to IfcRelSpaceBoundary specific attributes
(RelatingSpace, RelatedBuildingElement, ParentBoundary,
CorrespondingBoundary, PhysicalOrVirtualBoundary,
InternalOrExternalBoundary) reusing the existing
copy_attribute_to_selection core function.
The core function value type hint is broadened from Union[str, None]
to Any since boundary relation attributes pass IFC entity instances.
Generated with the assistance of an AI coding tool.
Linux "basic dependencies" omitted libeigen3-dev even though
ifcgeom requires Eigen3 (find_package Eigen3 REQUIRED) and the
cmake snippet already passes -DEIGEN_DIR=/usr/include/eigen3.
macOS Homebrew line already installs eigen.
Closes#6903
Generated with the assistance of an AI coding tool.
test_file's parametrize list was filtered with `sys.argv[1] in
os.path.basename(fn)`, reading the raw process argv instead of a
pytest-native option. Under a bare `pytest` invocation sys.argv[1] is
pytest's own first CLI token, never a match, so the 138-fixture EXPRESS
rule corpus in test/fixtures/rules collapses to an empty parametrize and
pytest reports it as a single skipped test rather than an error. Under
CI's actual invocation (pytest -p no:pytest-blender -n $NPROCS test ...)
sys.argv[1] is "-p", which happens to substring-match 47 of the 138
fixtures, so CI has been silently running a coincidental 34% slice of
the corpus with no signal anything was wrong.
Replaced the module-level list comprehension with a pytest_generate_tests
hook plus a --rule CLI option (added via a new test/conftest.py). This
runs the full corpus by default under any pytest invocation, still
allows filtering to one rule for local debugging via --rule, and no
longer collides with pytest's own argv.
Verified all 138 fixtures collect and pass under the fixed harness
(63 fail- fixtures each raise a violation, 75 pass- fixtures raise none).
Generated with the assistance of an AI coding tool.
mcp 2.0.0 (unpinned in CI and in the ifcmcp[mcp] extra) renamed
mcp.server.fastmcp.FastMCP to mcp.server.mcpserver.MCPServer, which
ifcmcp does not support yet. server.py caught the resulting
ModuleNotFoundError with a bare except Exception and silently
reported it as FastMCP not installed, masking the real breakage
until the ifcmcp test suite failed in CI.
Pinned mcp to >=1.0,<2 in both ci.yml and ifcmcp's pyproject.toml
mcp extra, confirmed the full ifcmcp test suite (70 tests) passes
against mcp 1.29.0, and confirmed the genuinely-not-installed path
still raises the expected ImportError. Also narrowed the except
clause to ImportError only so an unrelated future bug in that
import block surfaces instead of being swallowed as "not installed".
Generated with the assistance of an AI coding tool.
create(timestamp=0) computed the FILE_NAME timestring with
`d.get("timestamp") or time.time()`, which treats 0 (a legitimate
epoch timestamp) as unset because 0 is falsy. The header ended up
with the current wall-clock time in FILE_NAME while IFCOWNERHISTORY
correctly stored CreationDate=0, an inconsistent pair of dates in
the same file. Switched to an explicit None check so an explicit
timestamp of 0 is honoured the same way any other explicit
timestamp is.
Generated with the assistance of an AI coding tool.
The profile mapping builds its points as
profile_helper(m4, {
{{-x, -y}, {f2}},
...
where `f2` is a `double` and profile_point's second member is a
`boost::optional<double>`. In recent Boost (somewhere between 1.85 and 1.91)
optional's converting constructor became explicit, and an explicit constructor
cannot be used in copy-initialization — which is what a braced element is. So
every one of these call sites stops compiling:
MSVC 19.4x: error C2664: cannot convert argument 2 from
'initializer list' to 'const std::vector<profile_point>&'
clang-cl 22: error: chosen constructor is explicit in copy-initialization
Twelve translation units are affected (IfcCShapeProfileDef,
IfcIShapeProfileDef, IfcLShapeProfileDef, IfcTShapeProfileDef,
IfcUShapeProfileDef, IfcZShapeProfileDef, IfcAsymmetricIShapeProfileDef,
IfcCraneRailAShapeProfileDef, IfcRectangleProfileDef,
IfcRectangleHollowProfileDef, IfcRoundedRectangleProfileDef,
IfcTrapeziumProfileDef), roughly 100 call sites in total.
Adding one overload that takes the double directly fixes all of them without
touching a single call site, and changes nothing for existing code: the
optional overload still wins wherever an optional is passed.
Verified by building schemas 2x3;4;4x3_add2 with MSVC 2022 against Boost
1.91 and OCCT 7.9.3 — IfcParse, IfcGeom, the schema mappings and
geometry_kernel_opencascade all archive cleanly. Without this, the same build
against Boost 1.85 succeeds, which is what identified Boost as the variable.
Removing translate_obj_to_z_location from the existing-IfcSpace
regeneration branch. The ShapeBuilder rewrite (d8de62308) builds
geometry in local space preserving obj.matrix_world, making the
translate call redundant — it adds z on top of the already-correct
location.z, producing 2*z.
Add test_regenerate_space_preserves_z_location to cover the
regeneration path with a non-zero Z elevation.
Generated with the assistance of an AI coding tool.
Linux "basic dependencies" omitted libeigen3-dev even though
ifcgeom requires Eigen3 (find_package Eigen3 REQUIRED) and the
cmake snippet already passes -DEIGEN_DIR=/usr/include/eigen3.
macOS Homebrew line already installs eigen.
Closes#6903
Generated with the assistance of an AI coding tool.
test_file's parametrize list was filtered with `sys.argv[1] in
os.path.basename(fn)`, reading the raw process argv instead of a
pytest-native option. Under a bare `pytest` invocation sys.argv[1] is
pytest's own first CLI token, never a match, so the 138-fixture EXPRESS
rule corpus in test/fixtures/rules collapses to an empty parametrize and
pytest reports it as a single skipped test rather than an error. Under
CI's actual invocation (pytest -p no:pytest-blender -n $NPROCS test ...)
sys.argv[1] is "-p", which happens to substring-match 47 of the 138
fixtures, so CI has been silently running a coincidental 34% slice of
the corpus with no signal anything was wrong.
Replaced the module-level list comprehension with a pytest_generate_tests
hook plus a --rule CLI option (added via a new test/conftest.py). This
runs the full corpus by default under any pytest invocation, still
allows filtering to one rule for local debugging via --rule, and no
longer collides with pytest's own argv.
Verified all 138 fixtures collect and pass under the fixed harness
(63 fail- fixtures each raise a violation, 75 pass- fixtures raise none).
Generated with the assistance of an AI coding tool.
mcp 2.0.0 (unpinned in CI and in the ifcmcp[mcp] extra) renamed
mcp.server.fastmcp.FastMCP to mcp.server.mcpserver.MCPServer, which
ifcmcp does not support yet. server.py caught the resulting
ModuleNotFoundError with a bare except Exception and silently
reported it as FastMCP not installed, masking the real breakage
until the ifcmcp test suite failed in CI.
Pinned mcp to >=1.0,<2 in both ci.yml and ifcmcp's pyproject.toml
mcp extra, confirmed the full ifcmcp test suite (70 tests) passes
against mcp 1.29.0, and confirmed the genuinely-not-installed path
still raises the expected ImportError. Also narrowed the except
clause to ImportError only so an unrelated future bug in that
import block surfaces instead of being swallowed as "not installed".
Generated with the assistance of an AI coding tool.
FullBufferImpl and PagedFileImpl both open the file and then use the handle
without ever testing it:
auto stream = _wfopen(fn_wide, L"rb"); // null when the file is missing
fseek(stream, 0, SEEK_END); // null goes straight to the CRT
buf_.resize((size_t)ftell(stream));
Opening a path that does not exist therefore hands a null FILE* to the CRT. On
MSVC that does not return an error: the runtime terminates the process
immediately (fastfail, exit code 0xC0000409). No exception is thrown, no stack
unwinding starts, so a caller cannot defend with try/catch — the host
application simply dies. On glibc it is undefined behaviour as well.
This is reachable through the ordinary entry point, because guess_file_type()
answers FT_IFCSPF for a path that does not exist (its own comment calls this
"just weird, but for consistency with earlier behaviour"), so a missing path
flows into the reader rather than being reported.
The fix is to leave the reader empty when the open fails. Both implementations
then behave like a zero-length file: size() is 0 and get() throws out_of_range
for any position, so the parse fails and IfcFile::good() reports it, which is
what a caller can actually handle. PagedFileImpl's destructor already tested
fp_ for null, so the possibility was known — only the constructor did not check.
Verified by reading a non-existent path through IfcParse::IfcFile: the
constructor returns and good() reports the failure, where before the process
died with 0xC0000409 and no output.
create(timestamp=0) computed the FILE_NAME timestring with
`d.get("timestamp") or time.time()`, which treats 0 (a legitimate
epoch timestamp) as unset because 0 is falsy. The header ended up
with the current wall-clock time in FILE_NAME while IFCOWNERHISTORY
correctly stored CreationDate=0, an inconsistent pair of dates in
the same file. Switched to an explicit None check so an explicit
timestamp of 0 is honoured the same way any other explicit
timestamp is.
Generated with the assistance of an AI coding tool.
The profile mapping builds its points as
profile_helper(m4, {
{{-x, -y}, {f2}},
...
where `f2` is a `double` and profile_point's second member is a
`boost::optional<double>`. In recent Boost (somewhere between 1.85 and 1.91)
optional's converting constructor became explicit, and an explicit constructor
cannot be used in copy-initialization — which is what a braced element is. So
every one of these call sites stops compiling:
MSVC 19.4x: error C2664: cannot convert argument 2 from
'initializer list' to 'const std::vector<profile_point>&'
clang-cl 22: error: chosen constructor is explicit in copy-initialization
Twelve translation units are affected (IfcCShapeProfileDef,
IfcIShapeProfileDef, IfcLShapeProfileDef, IfcTShapeProfileDef,
IfcUShapeProfileDef, IfcZShapeProfileDef, IfcAsymmetricIShapeProfileDef,
IfcCraneRailAShapeProfileDef, IfcRectangleProfileDef,
IfcRectangleHollowProfileDef, IfcRoundedRectangleProfileDef,
IfcTrapeziumProfileDef), roughly 100 call sites in total.
Adding one overload that takes the double directly fixes all of them without
touching a single call site, and changes nothing for existing code: the
optional overload still wins wherever an optional is passed.
Verified by building schemas 2x3;4;4x3_add2 with MSVC 2022 against Boost
1.91 and OCCT 7.9.3 — IfcParse, IfcGeom, the schema mappings and
geometry_kernel_opencascade all archive cleanly. Without this, the same build
against Boost 1.85 succeeds, which is what identified Boost as the variable.
Box select resolved hits by reading the depth-tested object_id MRT, so only
the front-most surface in each pixel could ever come back. In x-ray that is
wrong twice over: you can see the geometry behind, and you still cannot
select it.
Add a second box-pick path used only while x-ray is active. It runs the same
vs_pick geometry through fs_boxpick with depth compare Always, no depth write
and no colour targets, scissored to the marquee — so nothing culls a fragment
behind another and the pass's only output is an atomicOr of one bit per
object into a hit bitmask. Reading that back gives every object with geometry
inside the box, occluded or not.
The bitmask rides alongside sel_flags at group(0) binding 2, allocated and
bound by ensureSelectionFlagsBuffer so the two can never disagree about how
many object ids exist. The layout entry is FRAGMENT-visible only: WebGPU
forbids a read_write storage buffer in the vertex stage, and every pipeline
shares this layout. Back-face culling is off for the pass — a box landing
inside a closed solid would otherwise see none of its faces and miss it.
Outside x-ray the depth-tested read stands, so a plain marquee still takes
only what is visible. A failure to build the pipeline falls back to that path
rather than breaking box select.
Tests cover the three properties worth having: x-ray selects strictly more,
its result is a superset of the plain one (a bare count would wave through a
wrong scissor or an off-by-one in the bit decode), and turning x-ray off
restores front-most-only. They need a model with real self-occlusion, which
sidecar_bake cannot currently produce — it segfaults on any input, including
the pristine sample.ifc — so they skip with an explanation until a fixture is
supplied. See the note at the top of the spec.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
ViewportCore has had a preset table (blender / rhino / revit / web) since the
nav bindings were shared with the desktop, and the web input handlers already
classify presses against it — but main() hard-coded "web" and nothing could
reach setNavPreset from JS, so every page was stuck on LMB-orbit.
Export ifcv_set_nav_preset_c and wrap it as IfcViewer.create's `navPreset`
option plus a setNavPreset() method. The preset is only the button/modifier
table, so it needs no GPU state: it applies as soon as the module resolves,
which means the first drag already uses the host's scheme rather than
flipping after a frame or two.
Names are validated in JS against the four the core knows. The core silently
falls back to blender for anything unrecognised, which would turn a typo into
a mystery change of scheme rather than an error.
The default stays "web", so existing pages and the smoke tests are unaffected.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Create file-owned headers after storage is selected but before streaming starts. Let owner-backed streamers use that header directly, and keep owned_header_ exclusively for ownerless streamers.
Generated with the assistance of an AI coding tool.
Three related cleanups to how the bonsaiviewer-autodesk connector is built
and shipped.
build-bonsaiviewer-autodesk.yml no longer builds a bundle. Its four-runner
matrix produced autodesk-<os>-<arch>.zip artifacts that nothing consumed —
shipping happens in the platform pipelines, which each invoke
packaging/build.py themselves. What is left is the crate's only lint and
test coverage, so the workflow is renamed to match what it does and a
header comment records where the shipped binary actually comes from.
build_osx.yml now builds and bundles the connector, which it never did:
macOS users have been getting a Bonsai Viewer with no Autodesk connector at
all. ConnectorDiscovery resolves applicationDirPath()/connectors, which
inside a bundle is Contents/MacOS, so that is where the folder lands.
The tkinter probes in the Windows and Linux workflows are dropped. They
guarded the old PyInstaller connector's Tk GUI (de7520418) and have been
dead since the Rust rewrite (9d9f4054f); python3.11-tkinter goes with them.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The crate had never been run through rustfmt, so the CI job's first step
(`cargo fmt --all -- --check`) failed and masked 12 clippy errors behind
it. Fix both.
Beyond the mechanical reformat and the redundant-closure/div_ceil/Default
lints, two changes carry meaning:
- SettingsDialog::on_reload is UI-thread only, so it becomes an Rc. The
Arc was never shared across threads and clippy rightly flagged it as
an Arc over a non-Send/Sync closure.
- WorkerMsg variants lose their shared `Loaded` postfix; the enum's doc
comment already says these are worker completions.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This branch carried v0.8.0's strict `[tool.ty.rules] all = "error"` config but
not the source fixes that were made upstream to satisfy it, so both ci-lint ty
gates were failing: `poe ty-ios` reported 256 diagnostics and `poe ty-bonsai`
258. Both are now clean.
Most fixes are ported from v0.8.0 and follow two idioms: initialise a name
before a conditional that may not bind it (plus an `assert` where the invariant
is real but not provable), and close an exhaustive `if`/`elif` chain with
`else: assert False, <discriminant>`.
The branch's own newer accessors are preserved throughout - `.file`,
`.declaration`, `file.types()`, `get_max_id()` are kept rather than reverted to
`wrapped_data.*`, and non-ty upstream changes (notably the in-progress geometry
cache removal) are deliberately not pulled in.
Notable fixes that are not straight ports:
* ifcopenshell_wrapper.pyi: `entity_instance.file` was declared as
`def file(self) -> file`, where the property name shadows the `class file`
below it, so the annotation resolved to `Unknown`. Every `element.file` in
the codebase was therefore unchecked. Qualifying it to `ifcopenshell.file`
restores `.schema` to its Literal union and surfaces no new diagnostics.
* model/wall.py: a duplicated merge fragment in the void-straddle path ran an
always-true `if void_straddles:` that read `new_opening` from the mutually
exclusive branch (stale value, or NameError on the first iteration), followed
by an unreachable duplicate `elif`. Removing it makes the file match v0.8.0.
* light/operator.py: upstream's own fix unpacks three targets from two values
and raises ValueError unconditionally; corrected to `None, None, None`.
* assign_system.py, validate.py, geom/main.py: walrus-in-genexp is valid at
runtime (PEP 572 binds in the containing scope) but ty does not model it;
rewritten as explicit loops, matching upstream.
Verified: poe ty-ios, poe ty-bonsai, ruff check src/ nix/, black --check .,
and compileall -W error at py3.10 (ifcopenshell-python) and py3.11 (bonsai).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Removing translate_obj_to_z_location from the existing-IfcSpace
regeneration branch. The ShapeBuilder rewrite (d8de62308) builds
geometry in local space preserving obj.matrix_world, making the
translate call redundant — it adds z on top of the already-correct
location.z, producing 2*z.
Add test_regenerate_space_preserves_z_location to cover the
regeneration path with a non-zero Z elevation.
Generated with the assistance of an AI coding tool.
Both surfaced on the first Windows/WASM CI run of this branch:
- XmlSerializer.cpp: the IfcPropertySetDefinitionSet block used a C-style
cast to convert the set to std::vector<IfcPropertySetDefinition>. GCC
invokes the non-explicit conversion operator; MSVC rejects the cast to a
template type (C2440/C3536/C2661). Use copy-initialisation instead, which
invokes the same implicit conversion portably. (This block was dead until
the SCHEMAS_->SCHEMA_HAS_ typo fix enabled it, so it had never hit MSVC.)
- parse.cpp: the floating-point parse path falls back to strtod_l because
libc++ =deletes the float from_chars overload. That fallback was guarded
for __APPLE__ only; Emscripten uses the same libc++, so WASM hit the
deleted from_chars. Extend the guard to __EMSCRIPTEN__ (its musl provides
strtod_l/newlocale, treating all locales as C).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Both fixes were made on v0.8.0's IfcParse.cpp AFTER the datamodel branch
had already renamed it to parse.cpp. When main was later merged into the
branch, the modify/delete was resolved toward the deletion (merge
8c9c3cde2), so the changes never reached the live parse.cpp. They are
invisible to `git log v0.8.0...HEAD` because the originating commits sit
in the merged-in shared ancestry; only a content sweep of the renamed
files surfaces them.
- format_double now uses the shortest decimal representation that
round-trips exactly (Mac-safe manual implementation, no std::to_chars),
instead of setprecision(digits10) which padded clean REALs with noise
digits (0.0174532925199433 -> 0.017453292519943299) and rewrote every
REAL on re-save. Recovers ee2b357d7 + fa597536e + 821cf7b67, #7696.
- The [SYN004] non-entity-type parse-error branch now resets current_id
to 0 before advancing, matching its sibling error branches, so a
malformed non-entity instance no longer erroneously terminates parsing.
Recovers 7c9df9f98.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
During the v0.8.0->wgpu port replay, ty config changes in pyproject.toml were
deferred (wgpu's whitelist all=ignore kept, v0.8.0's code fixes applied). Now
adopt v0.8.0's stricter blacklist config (all=error with curated ignores) and
pin ty to 0.0.63 to match. The pyproject diff was ty-only, so no wgpu-specific
config is lost.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The example block was copy pasted verbatim from ExtractPropertiesToSQLite,
so it named the wrong recipe and wrote a .sqlite file. These docstrings are
what ifcpatch surfaces as CLI and UI help, so anyone following the example
for AGS2IFC got a recipe name that does not match the one they selected.
Also state that the input file is not read and that a new IFC4X3 model is
built, since that is not obvious from the signature and the recipe creates
its own project rather than patching the one passed in.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 9621388953)
Graph2D::query and arrange_polygons relied on boost::optional reaching
them transitively through CGAL/boost headers. That transitive include no
longer happens on newer toolchains (GCC 14 / newer libstdc++), so the
build breaks with "boost::optional does not name a template type". The
rest of the svgfill module already uses std::optional; migrate these two
holdouts to match and include <optional> explicitly rather than freeload
on a fragile transitive include.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
angle_round_threshold was only assigned inside the `distance > 0`
branch of calculate_distance_and_angle, but read unconditionally
whenever should_round is True. When the mouse sample coincides with
the last placed point (distance == 0), such as the first mouse move
after placing a wall's start point on a YZ plane view, this crashed
the modal wall tool.
angle_round_threshold is a fixed cutoff unrelated to whether distance
is currently zero, so it is now assigned once before the branch.
Fixes#8597.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 89523999b3)
test_returns_none_when_report_file_absent/empty build a MagicMock repo
without configuring index.unmerged_blobs(), so it returned a truthy
MagicMock and git_mergetool's load-bearing "unresolved conflicts remain"
fallback (tool/ifcgit.py:646-647) returned that list instead of None -
failing "assert [] is None". The production fallback is correct and
intentionally left untouched; the tests just misrepresented the
"mergetool resolved cleanly" scenario they are named for. Set
mock_repo.index.unmerged_blobs.return_value = {} in both.
Verified in headless Blender: test/tool/test_ifcgit.py::TestGitMergetool
2 failed / 1 passed -> 3 passed.
This change was made with the assistance of an AI tool.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 51ab38de27)
DeepDiff's dictionary_item_added/set_item_added results are a
deepdiff.helper.SetOrdered instance, which subclasses orderly_set's
StableSetEq rather than the OrderedSet class json_dump_default checked
for, so the property relationship check always crashed export() with
"Object of type SetOrdered is not JSON serializable". Check against
StableSet, the common base class shared by every orderly_set set
flavour, instead.
Fixes#8905
Generated with the assistance of an AI coding tool.
(cherry picked from commit fbe36532a0)
It's ignoring underscore prefixed functions as not actually used.
Removing underscore to keep it happy without adding new exceptions.
(cherry picked from commit 88c8bd032f)
Browsers were caching /static/js and /static/css for the standalone
webui (costing, gantt, drawings, index, demo pages) indefinitely, so a
shipped JS fix (e.g. the Download CSV button) would only reach a user
after a manual hard refresh.
Two changes, applied consistently across all five webui pages.
1. Every locally served link/script tag in the pystache templates now
carries a ?v=<bonsai version> query string, falling back to a static
asset mtime hash when BONSAI_VERSION isn't set (e.g. running
sioserver.py standalone). Since get_bonsai_version() includes the
build's commit hash, the token changes on every shipped update.
2. Responses under /static/ and /jsgantt/ now carry
Cache-Control: no-cache, must-revalidate. This covers what query
stamping alone can't reach: cost.js and gantt.js statically import
utilities/costui.js by a fixed relative path with no query string, so
that nested module still needed server side revalidation to pick up
changes.
Verified against a live aiohttp instance of sioserver.py: rendered
HTML for all five routes shows the stamped URLs, and the token
changes when BONSAI_VERSION changes between two server runs. A
conditional GET against a static file with a stale If-Modified-Since
header confirms the cheap 304 revalidation path still works.
Also used this instance plus a real headless Chromium (Playwright) to
click test the previously untested Download CSV button on the costing
page. The ribbon renders it correctly, and clicking it (with a
synthetic cost-items table injected into the DOM to stand in for a
connected Blender's data) triggers a real Blob download with the
correct filename and CSV content. No bug found, the button works as
intended.
AI-generated with Claude Code.
(cherry picked from commit e759135608)
Stefano's final ask on #6251 was specific: the ODS/XLSX export should
show exactly what the cost panel shows, ID (Identification), Name,
Quantity, Value, Total Cost, no more, no less. The previous fix in
this PR removed the internal bookkeeping columns but still exported
Description, Unit and a per-category cost breakdown (Labor Cost,
Material Cost, etc), none of which appear in the panel.
Presentation formats (.ods/.xlsx) now use an explicit allow-list of
columns instead of a block-list of internal ones, and relabel headers
to match the panel's own wording (ID / Value / Total Cost). The .csv
format is unchanged: csv2ifc still reads back the extra bookkeeping
columns for the import round trip, which is why it keeps them.
Also add a "Download CSV" button to the browser costing view
(Generate spreadsheet browser), which previously only offered a
clipboard-based Copy Selected. It reuses the already-rendered table
(respecting the user's column visibility settings) and triggers a
real file download, dropping only the UI-only Actions column.
AI-generated with Claude Code; reviewed and tested by Petru Conduraru.
(cherry picked from commit 1df738d968)
Three defects reported against the Costing tab export:
1. XLSX export crashed with ModuleNotFoundError: xlsxwriter was never
bundled with Bonsai. Port the writer to openpyxl, which ifccsv
already uses and Bonsai already ships, so it works out of the box.
2. Every ODS cell was written as a string (numbers as text), and the
formula branch was dead code: it compared against 'Total Price' /
'Rate Subtotal' while the headers are 'TotalPrice' / 'RateSubtotal'.
Numeric columns are now typed float cells and TotalPrice becomes a
real formula: Quantity*RateSubtotal on leaf items, SUM over the
direct children's TotalPrice cells on sum items.
3. Internal bookkeeping columns (Id, ItemIsASum, Hierarchy, Index,
Quantities) leaked into the presentation formats. ODS/XLSX now hide
them; CSV keeps them since csv2ifc consumes them for the round trip.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 98c28a1f30)
get_element_value could not reach the members of an IfcPhysicalComplexQuantity
(or IfcComplexProperty) by their natural path. util.element expands a complex
quantity into a dict whose nested members live under a "properties" sub-dict,
but the selector's dict navigation only looked at the top level, so
"Qto_Custom.Layer1.Width" returned None and IfcCsv exported nothing for it.
Only the internal "Qto_Custom.Layer1.properties.Width" path worked.
When a key is not a direct member of the value dict, descend into its
"properties" sub-dict so nested quantities/properties resolve with the
natural "Set.Complex.Nested" path. Direct keys still take priority, so the
explicit ".properties." path stays backward compatible and the regex branch
is untouched.
Verified: Qto_Custom.Layer1.Width -> 0.1 and Layer1.Height -> 2.5 (were
None), the sibling simple NetArea still resolves, the legacy .properties.
path still works, and IfcCsv now exports the nested value. test_selector.py:
38 passed (adds test_selecting_a_nested_complex_quantity).
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 21122c0d28)
* Fix ci-bonsai-daily: ProjectLibraryData duplicate parent-library enum
parent_libraries_enum() adds an explicit entry for get_root_context(),
then loops over cls.data["project_libraries"] (all IfcProjectLibrary
entities) and appends each. For a library-only file (no IfcProject),
get_root_context falls back to the top-level IfcProjectLibrary itself,
so the root is appended twice with the same enum key (its STEP id),
which Blender EnumProperty requires to be unique -> the data load
asserts. Normal project files are unaffected (root is an IfcProject
whose id never collides with a library id).
Skip library_id == root.id() in the loop (dedup by id, the colliding
key). Verified in headless Blender:
test_project_library_data.py::TestLibraryOnlyFile goes from 1 failed /
5 passed to 6 passed.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Bonsai: repair library files missing the required IfcProject, not just the symptom
Per the IFC Project Context concept template, every project data set (library
files included) shall contain exactly one IfcProject, and IfcProjectLibrary
instances are assigned to it via IfcRelDeclares. There is no such thing as a
spec-valid file rooted on IfcProjectLibrary alone.
get_root_context() (added in 260a387069, #8184) treated a missing IfcProject
as license to use the top-level IfcProjectLibrary as the file's root context
instead. That invalid premise is why project_libraries() (which walks every
IfcProjectLibrary, root included) then re-added that same entity, producing
the duplicate, colliding enum key this PR originally papered over with a
dedup guard.
Add tool.Project.ensure_project_context(), which repairs a file missing
IfcProject by creating one and declaring the file's root-level
IfcProjectLibrary instances to it, and tool.Project.open_library_file(),
which opens a library file through that repair. Route all three
IfcStore.library_file load sites in SelectLibraryFile through it. Downstream
code (get_root_context, ProjectLibraryData, RefreshLibrary,
AddProjectLibrary) now always operates on a spec-valid model, so the
duplicate enum entry cannot occur; the previous one-line dedup guard in
parent_libraries_enum() is kept only as cheap defense in depth for callers
that bypass the load-time repair, not as the fix.
Rework test_project_library_data.py: the previous _make_library_only_file()
fixture built an invalid library-only model and asserted that as correct
behaviour. Replace it with a spec-valid fixture (IfcProject + IfcProjectLibrary
declared to it) for the downstream tests, and a malformed fixture used only to
exercise the new repair path.
Verified live in headless Blender (isolated profile): reproduced the original
duplicate-enum-key failure mode, then confirmed ensure_project_context/
open_library_file repair a malformed file and ProjectLibraryData,
refresh_library and add_project_library all operate correctly on the result,
with no duplicate keys and no regression on already-valid files or IFC2X3.
This change was made with the assistance of an AI tool.
* Bonsai: stop supporting library-only files, do not repair them
Per Moult's feedback: if the IFC is invalid, our default position is to not
support it, not to patch around it. A library file with no IfcProject is
invalid IFC (Project Context concept template requires exactly one
IfcProject), and it is not ubiquitous: every library file bonsai ships under
bim/data/libraries has an IfcProject with the IfcProjectLibrary declared to
it via IfcRelDeclares. The single #8183 report is an outlier, not a common
authoring pattern worth accommodating.
Remove tool.Project.ensure_project_context() and open_library_file() (the
load-time repair added in the previous commit here) and revert
SelectLibraryFile's three load sites to plain ifcopenshell.open. Simplify
get_root_context() back to returning ifc_file.by_type("IfcProject")[0]
directly, no IfcProjectLibrary fallback: a file without IfcProject now raises
IndexError instead of being silently treated as valid. AddProjectLibrary's
nest-under-library branch is now dead code (root_context is always an
IfcProject) and is removed. The one-line enum dedup guard from the original
commit here is also removed: since get_root_context can only return an
IfcProject or raise, an IfcProject id can never collide with a library id, so
the guard has nothing left to guard against.
Rework test_project_library_data.py: drop the invalid _make_library_only_file
fixture and its tests, which asserted an unsupported model as correct
behaviour. Replace with a single spec-valid fixture matching bonsai's own
shipped library files (IfcProject + IfcProjectLibrary declared to it), used
for the ci-bonsai-daily regression test and the refresh/add-library
operators, plus one explicit test that get_root_context raises for a file
without IfcProject, documenting that this input is intentionally
unsupported rather than silently tolerated.
Verified live in headless Blender (isolated profile, source-loaded, never
the real profile): confirmed the removed methods are gone, that a
library-only file now raises instead of being handled, that
ProjectLibraryData/refresh_library/add_project_library all work correctly
on a spec-valid model with unique enum keys, and spot-checked that every
library file under bim/data/libraries already has an IfcProject.
This change was made with the assistance of an AI tool.
* Bonsai: inline get_root_context, trim docstrings, confirm get_parent_library unchanged
Per Moult's round 3 review. get_root_context added nothing over
ifc_file.by_type("IfcProject")[0], which is guaranteed by the IFC Project
Context concept template; remove it and inline the call at its three sites
(operator.py's RefreshLibrary and AddProjectLibrary, data.py's
parent_libraries_enum). Trim the get_parent_library docstring to one line;
its logic is untouched by this PR, byte for byte identical to origin/v0.8.0,
and still returns None only when project_library has neither Nests nor
HasContext, never for a library declared directly to IfcProject.
Rework test_project_library_data.py to match: replace the two
get_root_context-specific tests with one that exercises the real call site
(ProjectLibraryData.parent_libraries_enum raising IndexError for a file
without IfcProject), and add an explicit test that get_parent_library
returns None for a genuinely orphaned library. Also drop a long inline
comment that restated what the test body already shows.
Verified live in headless Blender (isolated profile, source-loaded, never
the real profile): all 17 test/bim/module/project tests pass, including the
new get_parent_library None-for-orphan case. Ran the full test/bim suite
before and after on the identical harness: 82 failed/1335 passed both times,
same failing tests (all pre-existing, unrelated to this module).
This change was made with the assistance of an AI tool.
* Bonsai: fix EditProjectLibrary leaving stale declarations after reparenting
Per Moult's round 4 review. The assertion change (get_parent_library(root)
now returns the IfcProject instead of None) is correct: in the old
library-only test model a top-level library had neither IfcRelNests nor
IfcRelDeclares, so None meant "top level". In the new spec-valid model a
top-level library is always declared to the guaranteed IfcProject via
IfcRelDeclares, so get_parent_library correctly resolves it through the
HasContext branch instead of falling through to None. get_project_hierarchy
already keys top-level libraries under the project for exactly this reason,
so the library tree still renders correctly.
Auditing every caller found one real bug in EditProjectLibrary, which
Gorgious56 originally wrote for the library-only model. Its move-library
logic assumed a top-level library (previous_parent_library is None) needed
no cleanup before nesting it under a new parent, and that unnesting a
library back to the project needed no new relationship because it was
"already assigned by default". Both assumptions relied on a top-level
library never actually holding a IfcRelDeclares, which is no longer true.
Reproduced live: moving a project-declared library under another library
left its old IfcRelDeclares dangling alongside the new IfcRelNests (an
invalid double parentage), and moving a nested library back to the project
left it with neither relationship, orphaning it out of the tree entirely.
Fixed by tearing down whichever of IfcRelDeclares/IfcRelNests the library
previously had before establishing whichever one the new parent requires,
instead of assuming which prior state applies.
Added tests: get_parent_library resolving a nested sub-library to its
library parent (the third contract case alongside project-declared and
orphaned), and both EditProjectLibrary reparenting directions, which fail
without the operator.py fix and pass with it.
Verified live in headless Blender (isolated profile, source-loaded, never
the real profile): all 20 test/bim/module/project tests pass. Ran the full
test/bim suite before and after on the identical harness: 123 failed/1294
passed before, 123 failed/1297 passed after, identical failing test names
in both runs (diffed), the extra 3 passes are the new tests above.
This change was made with the assistance of an AI tool.
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit d1f9e5243e)
on_depsgraph_update and on_depsgraph_update_caps are registered together
as persistent depsgraph handlers (bim/module/clip_box/__init__.py:50-52).
on_depsgraph_update guards with `if cls._file_loading: return`, but the
sibling on_depsgraph_update_caps did not, so a depsgraph tick during the
file-load window still ran it. Beyond the failing test, this can re-arm a
cap-rebuild bpy.app.timers callback in the exact load window _on_load_pre
cancels timers for, against regions whose GPU state is not yet wired.
Add the same _file_loading guard as the first check.
Verified in headless Blender:
test_clip_box.py::TestRefreshTimerLifecycle::test_depsgraph_update_no_op_while_loading
1 failed -> passed.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 537317b26f)
Link IFC with 'Use Cache' unchecked crashed with FileNotFoundError
when no .ifc.cache.blend existed yet (a fresh link). Regression from
35e3d9c42, which refactored the cache-clear guard from
'if not self.use_cache and blend_filepath.exists()' into
should_clear_cache() but dropped the existence check on the
not-use_cache path, so os.remove() ran on a non-existent file.
Check blend_filepath.exists() first in should_clear_cache() so the
remove is never attempted when there is nothing to clear, while
keeping the query-mismatch cache invalidation intact.
Fixes#8350
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 4a095f9810)
When replacing a style on an item whose previous IfcStyledItem wraps its styles
in the deprecated IfcPresentationStyleAssignment, and the assignment is not
being reused (use_style_assignment is False, e.g. an IFC4 file authored by
AVEVA E3D), the else branch called remove_same_type_styles(style_assignment)
with style_assignment still None, raising
AttributeError: 'NoneType' object has no attribute 'Styles'. Operate on style_,
the assignment found in the current iteration, instead of the accumulator.
Verified red-green with a minimal IFC4 file using IfcPresentationStyleAssignment.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 82f73c29ea)
compare() recursed into list values passing the negated comparison through,
so != meant "at least one item differs" and both = and != matched the same
elements on any multi-valued property (e.g. an enumerated property with two
values selected). Strip the negation for the per-item comparison and negate
the aggregate instead, so != means "no item equals" and stays the complement
of =. The same applies to !*=.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 209c44db83)
The Optimise recipe imported `toposort`, a third-party PyPI package that
is not bundled with Bonsai, so running the recipe there raised
`ModuleNotFoundError: No module named 'toposort'`.
Replace it with the standard library `graphlib.TopologicalSorter`
(available since Python 3.9), which provides the same dependencies-first
ordering guarantee the recipe relies on: forward-referenced instances are
mapped before the instances that reference them. The dependency-graph
dict format ({node: {predecessors}}) is identical between the two, so the
graph construction is unchanged. Drop `toposort` from ifcpatch's
dependencies since it is no longer used.
Verified with toposort NOT installed: the Optimise recipe now runs and
deduplicates correctly (IfcParseExamples_test.ifc 88 -> 63 instances, all
6 products preserved, output reopens cleanly).
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 382f5e0c21)
tool.Bsdd.identifier_url() (pset/ui.py pset name check in the Property
Sets panel) reads client.baseurl unconditionally, but the test stub
never had that attribute, so any scenario that opens the Property Sets
panel dies with AttributeError under the stub. The boolean.feature
scenarios only surfaced this once their STEP id failures were fixed,
the id failure had been masking it. Mirror the real bsdd.Client
default so identifier_url() resolves to the standard identifier URL.
This change was made with the assistance of an AI tool.
(cherry picked from commit 81a42cec1a)
The two boolean.feature scenarios pinned representation item objects by
absolute STEP id (Item/IfcHalfSpaceSolid/90, the BBIM_Boolean pset text
[91]). Those ids shift every time any earlier entity allocation in an
empty project changes (latest instance: #8577 moved 90 to 86), so this
cluster re-breaks on unrelated commits.
Make the object-name and panel-text BDD steps run their argument through
replace_variables, the same substitution 'the variable' and the
connection steps already use, and have boolean.feature capture the real
ids from the IFC file (by_type(...)[0].id()) into variables at the point
the entities are created. The steps stay strict: the substituted name
must still resolve to exactly the named object, there is no wildcard
matching. Substitution is a no-op for every existing feature string
without a {variable} placeholder.
This change was made with the assistance of an AI tool.
(cherry picked from commit 45fa04a94b)
Two independent test-harness/fixture defects in test/bim/test_feature.py:
- OperatorSpy had no bl_rna, so any BDD step that redraws a panel calling
helper.draw_filter() (which tests "module" in op.bl_rna.properties)
crashed with AttributeError. Give OperatorSpy a bl_rna property that
forwards to the real registered operator class
(bpy.types[bl_idname].bl_rna), matching live UILayout.operator()
semantics. Fixes test_select_all_walls and test_edit_filter_query.
- The shared "I create default MEP types" step looked up
bpy.data.objects["IfcDistributionPort/Port"], but port creation never
sets port.Name, so tool.Loader.get_name deterministically names the
object "IfcDistributionPort/Unnamed". Update the literal. Fixes the MEP
scenarios (connect/transition/bend) that share this setup.
Verified in headless Blender: OperatorSpy scenarios 2 passed (were
AttributeError); MEP test_connect_mep_elements* go from
KeyError 'IfcDistributionPort/Port' to passing.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit e27624c77f)
Pset names containing spaces (e.g. "SOLIDWORKS Custom Properties") were
not quoted when building selector keys in SelectSimilarData, causing
get_element_value to fail when the operator ran. Now wraps pset names
and property names in double quotes if they contain spaces, consistent
with the selector syntax used elsewhere.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 836d57e7ff)
Per aothms's review comment: this file's schema was already changed
independently on v0.8.0 since this branch was created, so this PR's own
edit conflicts with it. Reverting to the current upstream version of the
fixture; the bsdd.py rate-limiting fix is untouched.
(cherry picked from commit 3d7d1ff4f3)
bsdd.py: the Client made every request with a bare requests.get, so a single
429 from the (unauthenticated, aggressively rate limited) bSDD API failed the
whole test. Route requests through a Session with a mounted urllib3 Retry
(5 attempts, backoff, honouring Retry-After) for 429/5xx, matching how a
resilient API client should behave, not just papering over the test.
ColumnPSetsOfSets.ifc: FILE_SCHEMA was accidentally changed from IFC4X3_ADD2
to IFC2X3 in a7738eeb64 (an unrelated logger refactor), a one line collateral
edit to this fixture. The file's DATA section still uses IFCPROPERTYSETDEFINITIONSET,
an IFC4+ only type. Parsing it against IFC2X3 threw "Entity ... not found in
schema", which silently fell back to interpreting the value as a raw nested
aggregate instead of the intended defined-type wrapper, producing the
double-nested tuple that broke test_stream, test_file and test_rocks in
test_streaming_rocksdb_and_simpletyperefs.py. Restoring the original schema
declared when the fixture was added (ff3fa48332) fixes all three.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 6911418c67)
Bsdd.get_dictionaries() unconditionally did cls.client = bsdd.Client(),
replacing whatever client was already set - including the
bSDDClientStub the BDD suite injects at module load
(test_feature.py: tool.Bsdd.client = bSDDClientStub()) to avoid live
network calls. Because "Load bSDD Dictionaries" is the first step of
every bsdd.feature scenario, the stub was discarded before its fixture
data ("LCA", "BonsaiTestDict") could ever be returned.
The re-init is unnecessary: bsdd.Client.__init__ only sets baseurl and
blank tokens, and the next line already updates baseurl defensively via
hasattr. Drop the clobbering assignment; reuse whichever client is
already set.
Verified in headless Blender: bsdd scenarios (load dictionaries, search
all/single dictionary) go from 3 failed ("Could not see LCA/
BonsaiTestDict") to 3 passed.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 5c8eab981c)
The choco dir was renamed from choco/blenderbim to choco/bonsai back in
2024 (Rename choco dir), but choco_release.py's BLENDERBIM_DIR constant
was never updated, so the daily choco release job crashes immediately
with FileNotFoundError trying to os.chdir into the now nonexistent
choco/blenderbim directory.
Release tags also moved from a bare blenderbim-YYMMDD scheme to
bonsai-X.Y.Z-alphaYYMMDDHHMM, so the tag-prefix strip used to build the
package version still looked for the old "blenderbim-" prefix and left
it untouched, embedding the raw tag (including the already-present
"-alpha" segment) into the nuspec version field, which the template
then doubled up with its own "-alpha" suffix, producing an invalid
NuGet version string. Both are fixed together since the second bug
would otherwise surface as soon as the first one is unblocked.
The pre-commit black hook also reformatted pre-existing whitespace
drift in choco_release.py (this file sits outside CI's lint scope, so
it had never been auto-formatted before); that reformatting is
incidental to satisfying the local hook, not part of the fix itself.
Generated with the assistance of an AI coding tool.
(cherry picked from commit a155a1ca80)
replace_attribute() rewrites references inside aggregate attributes via
element.walk(), but never checked whether the replacement value was
already present elsewhere in the same aggregate. For an EXPRESS SET
(e.g. IfcProject.RepresentationContexts, IfcRelAggregates.RelatedObjects)
this can leave the same reference listed twice, which is invalid IFC.
LIST and BAG aggregates legitimately allow duplicates, so a blanket dedup
would be wrong; only SET-typed attributes are deduplicated, determined at
runtime from the schema declaration (IfcOpenShell#8706 review comment).
The SET/LIST/BAG check is cached per (schema, class, attribute index), and
the dedup pass itself only runs when a cheap linear pre-check finds the
replacement value already present in the aggregate, so the common case
(no duplicate produced) pays only that pre-check, not a hash-set rebuild.
Benchmarked against a 23MB (431k entities) and a 104MB (2.4M entities) IFC
model against a large SET attribute: worst case adds well under 1ms per
call; the realistic case (merging duplicate contexts, matching the PR
#8706 scenario) shows no measurable regression.
Fixes the root cause flagged in IfcOpenShell#8706 (Moult), obviating the
need for MergeDuplicateContexts' own manual aggregate-dedup pass for that
scenario.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 8c434b7167)
Fixes issue #3910's documentation gap. IsDefinedBy() returns
IfcRelDefinesByProperties relationship objects, not the property set
itself, and RelatingPropertyDefinition() must be used to reach the
IfcPropertySet or IfcElementQuantity. Properties can also come from an
element's type via IsTypedBy() -> RelatingType() -> HasPropertySets(),
a path that is easy to miss because it works differently. Adds a
worked, beginner-commented, compilable example covering both paths.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 8fb8966094)
The ifcopenshell take-off engine left every Qto_EarthworksFillBaseQuantities
value null, so Bonsai added the qset with no numbers on IFC4X3 models. Map the
geometrically derivable quantities using the slab axis convention: Length on
local X, Width on local Y, Depth on local Z, and the net solid volume as the
compacted (Fill) or undisturbed (Cut) volume. LooseVolume and Weight stay
unmapped because they need soil bulking and density factors absent from
geometry. Bring IfcEarthworksCut to parity with the Blender engine and wire
IfcReinforcedSoil on both engines.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit d506f5df1e)
assign_cost_item_quantity skipped every IfcSpatialElement, which also
swallowed IfcSpace. Spaces are legitimate quantifiable objects, so their
Qto_SpaceBaseQuantities (for example GrossFloorArea) were never picked up
and count based cost items fell back to 0. Keep skipping spatial
containers (site, building, storey) but allow IfcSpace.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit e23142d01a)
The headless "IfcOpenShell" calculator had all Qto_SpaceBaseQuantities
formulas set to null for IfcSpace in both IFC4QtoBaseQuantities.json and
IFC4X3QtoBaseQuantities.json, so qto.py's `if not formula: continue`
skipped every quantity, no geometry task was queued, and spaces never
appeared in results (elements_quantified: 0). The Blender calculator
already computes these; they were just never ported to the
ifcopenshell.util.shape-backed calculator.
Map the eight computable quantities to existing util.shape functions,
mirroring the Blender calculator semantics (no new shape.py code):
GrossFloorArea=gross_get_footprint_area, NetFloorArea=net_get_footprint_area,
GrossCeilingArea=gross_get_top_area, NetCeilingArea=net_get_top_area,
GrossPerimeter=gross_get_footprint_perimeter, GrossVolume=gross_get_volume,
NetVolume=net_get_volume, Height=net_get_z.
Left null (matching the Blender ruleset, not guessed): GrossWallArea,
NetWallArea, NetPerimeter (Blender stub), and FinishFloor/CeilingHeight
(Blender derives these from sibling IfcCovering decomposition geometry,
which this per-element calculator architecture can't reach).
Verified on IFC4 (4x3 space extruded 2.5m): before -> {} / elements_quantified 0;
after -> GrossFloorArea 12, GrossPerimeter 14, Height 2.5, GrossVolume 30,
etc. - all exact matches to the extrusion. IFC4X3 formulas are identical
and the formula->function resolution is schema-agnostic.
Scope: fixes the IfcSpace case (the issue title). The 12 other all-null
classes noted in the issue (IfcDoor, IfcSite, IfcRailing, ...) are left as
follow-up.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 6085894433)
The ifc5d IfcOpenShell (geometry-based) Qto engine computed
Qto_CoveringBaseQuantities using axis-agnostic heuristics:
- GrossArea/NetArea: gross_get_max_side_area / net_get_max_side_area,
the largest of the X/Y/Z projected side areas.
- Width: gross_get_min_xyz, the smallest of the X/Y/Z dimensions.
The Blender Qto engine instead already used
EPset_Parametric.LayerSetDirection (AXIS2 for wall-like coverings,
AXIS3 for floor/ceiling-like coverings) to pick the correct axis via
get_covering_gross_area/get_covering_net_area/get_covering_width in
bonsai/bim/module/qto/calculator.py.
For any covering whose length isn't the largest dimension (e.g. a
short wall-covering strip, or a small covering patch), the two
engines' heuristics can pick different faces/axes entirely, giving
different Width/Area values for the same element - this is what was
reported in #6728.
Fix: give the IfcOpenShell engine the same layer-set-direction
awareness. Added IfcOpenShell.get_covering_parametric_axis/
get_covering_area/get_covering_width (dispatched as internal
functions, like the existing get_weight/get_segment_length), and
wired gross_get_covering_area/net_get_covering_area/
gross_get_covering_width into the IfcCovering rules in
IFC4QtoBaseQuantities.json and IFC4X3QtoBaseQuantities.json.
The AXIS2 area/width formulas (get_side_area, net_get_y) intentionally
match the simpler formulas already used for Qto_WallBaseQuantities in
this same rule set (net_get_side_area/net_get_y), rather than
replicating the Blender engine's more elaborate get_lateral_area/
get_width (min(X,Y)) helpers, consistent with how the two engines
already diverge for regular walls without being considered a bug.
Verified with a standalone script driving ifc5d.qto.IfcOpenShell
directly against synthetic AXIS2/AXIS3 IfcCovering geometry: for
typical proportions old and new formulas agree, and for
disproportionate coverings (thin dimension not the smallest/largest)
the old formulas picked the wrong axis while the new ones correctly
track the covering's LayerSetDirection, matching the Blender engine.
Did not verify through the full Blender/Bonsai UI, as it would have
required registering the addon in the machine's shared Blender
profile, which is unsafe while other agents may have it loaded.
Generated with the assistance of an AI coding tool.
(cherry picked from commit c61ebe3876)
copy_cost_item appends the copy to the inverse relationships of the
original cost item, which for a root cost item includes the source
schedule's IfcRelAssignsToControl. copy_cost_schedule then assigned that
same cost item to the new schedule as well, so the copies showed up in
both schedules and deleting them from one removed them from the other.
Unassign the copy from the source schedule before assigning it to the
new one.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 305f8c6003)
Fixes#4443. The Wall/Slab/other authoring tools (BimTool subclasses)
already bind Shift+Q to bim.perform_quantity_take_off via hotkey_S_Q,
but the Spatial tool has its own separate keymap/operator
(bim.spatial_hotkey) that never registered a Q entry, forcing users to
switch tools just to (re)calculate quantities for a selected element.
Added the same Shift+Q keymap entry and a matching hotkey_S_Q handler
to the Spatial tool, mirroring BimTool's existing behavior exactly
(including the same selected-objects guard).
The other part of the request, a bulk "calculate all quantities"
entry point, already exists today: bim.perform_quantity_take_off
computes quantities for every IfcElement when no objects are
selected, exposed via the Scene > Quantity Take-off panel regardless
of which workspace tool is active, so no change was needed there.
AI-generated, reviewed and tested by BIMvoice.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
(cherry picked from commit 98a25eca96)
can_nest() only permitted IfcElement-to-IfcElement pairs, so nesting
two IfcElementType objects (e.g. an IfcElementAssemblyType nesting a
component IfcDoorType) was silently rejected. IfcRelNests.RelatingObject/
RelatedObjects are typed as the general IfcObjectDefinition in the
schema, so type-to-type nesting is schema legal, IfcOpenShell's core
nest.assign_object API already handles it generically, and the Nest
UI panel is driven purely by ifcopenshell.util.element.get_nest/
get_components (IFC data queries, not Blender collection structure),
so once the relationship exists it displays correctly with no other
changes needed.
Extended is_compatible_class to also accept a same-kind IfcTypeProduct
pair. Mixing an occurrence element with a type is intentionally still
rejected, that isn't a real modeling pattern.
Verified live in headless Blender: type-to-type nesting now creates
a real IfcRelNests and the Nest panel's own data functions reflect
it correctly; mixing an occurrence with a type is still rejected;
existing element-to-element nesting is unaffected.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 99c370b755)
Root cause: the IfcOpenShell-geometry-engine calculator ruleset
(IFC4QtoBaseQuantities.json and IFC4X3QtoBaseQuantities.json) left
IfcWall's GrossFootprintArea/NetFootprintArea mapped to null, so
these two quantities were silently omitted from Qto_WallBaseQuantities
whenever that ruleset was used. The generic gross_get_footprint_area
and net_get_footprint_area formulas already exist and are already
wired up for IfcSlab in the same files, so this was a missing mapping,
not a missing implementation.
Fixes#7029.
Generated with the assistance of an AI coding tool.
(cherry picked from commit f3cb7aea61)
Footings are authored two ways with different local axis conventions. Beam-like
footings (STRIP_FOOTING, FOOTING_BEAM) are a profile extruded along local Z, so
Length is local Z and the cross section sits on local X (Width, horizontal) and
local Y (Height, vertical). Slab-like footings (PAD_FOOTING, PILE_CAP) have their
footprint on local X/Y and their thickness (Height) on local Z.
The engine rule set is keyed per IfcFooting and cannot branch on predefined type,
so the previous static rule (Height=net_get_z, Length=net_get_max_xy, Width=null)
swapped Length and Height for beam-like footings and never emitted Width.
Add predefined-type-aware get_footing_length/width/height to the IfcOpenShell and
Blender calculators, and point the IfcFooting rule at them in all four IFC4/IFC4X3
ios/Blender rule files.
Confirmed by authoring footings through the real Bonsai generators and measuring
world-axis orientation: a beam-like footing with a 0.3 wide by 0.6 tall cross
section and 6.0 run reports Length 6.0, Width 0.3, Height 0.6, with the 0.3
physically horizontal and 0.6 physically vertical; a 2.0x1.5x0.3 pad reports
Length 2.0, Width 1.5, Height 0.3.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 4ceadd8f10)
See #6835. Qto_OpeningElementBaseQuantities came out axis-scrambled for
openings authored in a Z-up local frame (X along the voided wall, Y
through it, Z vertical), which is how Bonsai authors every wall opening:
- The IfcOpenShell engine mapped Height to the local Y extent and Depth
to the local Z extent, so a 0.9 x 2.0 door opening with Bonsai's
default 1.2m void depth reported Height 1.2 and Depth 2.0, and Area
(max side area) picked the through-wall side, 2.4 instead of 1.8.
This matches the wrong Height=1.2/Area=1.2 screenshots reported for a
1x1 window opening in #6835.
- The Blender engine mapped opening Width to get_length, which returns
the longest bounding box edge, i.e. the opening height for typical
door openings (the same defect 4adaf0d fixed for IfcDoor Width), and
get_opening_depth used min(x, y), which returns the opening width
whenever the width is smaller than the void depth.
The IfcOpenShell engine now has opening-aware internal calculators
(get_opening_width/height/depth/area) that detect horizontal (slab
style) openings with the same heuristic as the Blender calculator, so
slab opening depths keep reporting the slab thickness. The Blender
ruleset uses get_x for opening Width, and get_opening_depth measures the
through-element Y extent for vertical openings.
Door and window quantities themselves are addressed separately: the
Blender engine door Width was fixed in 4adaf0d, and the remaining
door/window defects (door not quantified on the IfcOpenShell engine,
inflated areas) are fixed by the attribute-based calculators in #8389.
Generated with the assistance of an AI coding tool.
(cherry picked from commit efac8a0ec0)
falken10vdl reviewed 16b1b4e7b1 on #8843 and pointed out that tagging
every area for redraw was overkill. The actual problem was that the
Object Material panel and the scene Materials list read from plain
python caches (ObjectMaterialData and MaterialsData) that only get
invalidated when the Materials editing UI list is reloaded, which
never happens while you are not in editing mode. The redraw itself was
never the issue, closing the rename dialog already triggers one.
Removed the tag_redraw loop from RenameMaterial and instead call the
existing bonsai.bim.module.material.data.refresh() function from
core.rename_material, unconditionally, through a new tool.Material.refresh()
method. This is the same invalidate-on-next-load mechanism already used
by every other module's Data classes, just wired up for this operator
too, instead of introducing a new one.
Also updates the core tests to prescribe the new unconditional refresh()
call, and adds tool-layer coverage for tool.Material.refresh().
Generated with the assistance of an AI coding tool.
(cherry picked from commit 7ab0628c54)
theoryshaw tested #8843 and asked for the new name to show up right
away instead of needing a manual refresh. The Object Material panel
and the scene Materials list both already re-read live IFC data on
their next draw (tool.Ifc.Operator purges those caches after every
IFC-mutating operator), so the button text was correct on the next
redraw. What was missing was the redraw itself: the material name is
a plain button label, not an RNA property Blender tracks, so nothing
told the Properties editor to repaint after the rename dialog closed.
Tag every area for redraw once the rename completes, the same pattern
used elsewhere in Bonsai for popup-triggered edits that need an
immediate repaint.
Also adds core-layer test coverage for rename_material, which had
none.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 16b1b4e7b1)
Adds a "Rename Material" entry to the context menu that already
extends every button in the properties editor (UI_MT_button_context_menu),
triggered when right-clicking a material name button
(bim.select_by_material) that points to a real IfcMaterial. This
gives a quick entry point to renaming from the Object Material panel
without navigating to the scene Materials list.
This follows the pattern that #6680's thread converged on: theoryshaw
requested a right-click entry (rather than a pencil icon or
double-click) that keeps the existing single-click select-by-material
behaviour intact. falken10vdl is the issue's assignee; this is offered
as a starting point for that discussion, not a replacement for it.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 8c667b8ae0)
Add an "Is Selected" checkbox to each target-view category header in
BIM_UL_drawinglist that toggles selection for all drawings in the
category. The toggle only affects drawings currently visible in the
list (honoring the show_drawings_on_sheets_only filter), and the header
checkbox reflects the aggregate selection state of its drawings.
Also make category headers more obvious: wrap them in a box() for a
distinct inset background and make the header name clickable to
expand/contract the category (same as the disclosure triangle).
Ref: #8825
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit e52e5e2e58)
Adds a "Show Only Drawings on Sheets" toggle below the drawing list. When
enabled, the list is filtered to drawings referenced by at least one sheet
(target-view headers with no sheeted drawings are hidden too), and
bim.select_all_drawings only acts on the visible/filtered drawings.
A drawing is considered sheeted when its drawing document Location matches a
document reference Location on any SHEET-scoped IfcDocumentInformation.
Filtering is computed live so it reflects sheet edits without reloading.
Closes#8823
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 2d59ea1988)
* docs: cover the Blender 5.1 / Python 3.13 transition in installation guides
The system requirements still listed Blender 4.3-4.5 with Python 3.11
only, and nothing documented the pitfall from issue 7623: importing
preferences into a Blender whose Python version changed carries over an
incompatible Bonsai build that silently fails to load. Document the two
Python generations, that Get Extensions picks the matching build
automatically while manual zip installs do not, and the
uninstall-reinstall step that resolves the upgrade case.
Generated with the assistance of an AI coding tool.
* docs: keep it simple, only Blender 5.1 and 5.2 with Python 3.13
Per review, drop the descriptive text and the Python 3.11 line.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 55a2430d71)
* Keep real cause in wrapper ImportError
When the compiled wrapper exists for the current interpreter but fails
to load (for example a glibc version mismatch, as on AWS Lambda in
issue 5927), the bare except rewrote the error into the misleading
"IfcOpenShell not built for '<platform>'" message. Environments such
as AWS Lambda or the Blender add-on dialog only surface the final
exception message, so the actual cause was invisible and undiagnosable.
Keep the "not built for" message only when no matching binary is
present, and otherwise include the original loader error, chaining the
cause in both branches.
This change was AI-generated.
Fixes#5927
* Simplify wrapper import failure to a single message
Per review feedback, drop the filesystem scan and the two message
variants. Always raise the classic "IfcOpenShell not built for
'<platform>'" message with the original exception appended in
parentheses, still chained as the cause. Environments that only show
the final exception message (AWS Lambda, the Blender add-on dialog)
now surface the real loader error, such as the glibc version mismatch
in issue 5927, without any extra logic.
This change was AI-generated.
(cherry picked from commit 04a2535a98)
The jsGantt-improved library that renders Bonsai's Gantt chart already
ships full support for an "Hour" granularity (column width, header
labels in every bundled language, hour-aware rendering math). Bonsai's
config only exposed Day/Week/Month/Quarter, with a comment claiming
Hour caused browser issues even with vUseSingleCell enabled.
Headless Chrome testing against the same library version shows that
claim no longer holds once vUseSingleCell is active (as Bonsai already
configures it at 10000): Hour-format charts render without errors from
typical schedules up through fairly extreme ones (5000 tasks across a
3 year span rendered in about 2.4s). The failure mode the old comment
described only reproduces with vUseSingleCell disabled, which is not
how Bonsai runs it.
Task start/finish times already flow through to the chart unmodified
as raw ISO datetimes (tool/sequence.py create_new_task_json), so any
schedule authored with real hour-level timestamps, for example an
imported MS Project/P6/Excel schedule or one written directly through
ifcopenshell-python, can now be viewed at hour granularity. Verified
live with a night shift schedule crossing midnight, rendered correctly
with no console errors.
Note: Bonsai's own "Edit Task Time" UI currently always snaps
ScheduleStart/ScheduleFinish to 09:00/17:00 regardless of the hour
entered (ifcopenshell/api/sequence/edit_task_time.py), and work
calendars only encode working days, not working hours. So authoring a
genuine hour-precision schedule through that UI is still not possible;
this change only unlocks viewing hour-level data that already exists
in the model. Fixing the editor and calendar model is a separate,
larger design decision for a maintainer.
Addresses #2772.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 727b5f3475)
Point scriptSalome.py at templates/salome/ and the bonded scripts at
_deprecated/, matching the current tree so README links resolve.
Generated with the assistance of an AI coding tool.
(cherry picked from commit a45f2fae61)
* Bonsai: move annotations between drawings when reassigning their group
Assigning an IfcAnnotation to a group that represents another drawing
previously left the annotation in both drawings at once: it stayed in
its old drawing group, its Blender object stayed in the old drawing
collection, and it kept the old camera depth, so the reassignment
appeared to do nothing useful. Issue #2966 documents the seven step
manual workaround users needed instead.
The assign group operator now detects when the target group represents
a drawing (via the new tool.Drawing.get_group_drawing, the inverse of
get_drawing_group), unassigns the annotation from its previous drawing
group, moves its object into the new drawing collection, and places it
on the new drawing camera plane. The target camera is imported on
demand when it has not been loaded yet, matching the pattern used by
the activate drawing operator.
Generated with the assistance of an AI coding tool.
* Bonsai: add one click copy of annotations to another drawing (#2966)
Duplicating an annotation into a different drawing used to require a
seven step manual process: loading groups in scene properties, copying
the object, fixing its group assignment by hand, and repositioning it
onto the target camera plane. A plain Blender duplicate is not enough
because the copy keeps pointing at the same IFC entity, and the Shift D
override, while it does create a genuine new entity through
root.copy_class, leaves the duplicate in the source drawing group,
collection, and camera depth.
The new copy annotation to drawing operator packages the proven recipe
already used by duplicate drawing into one action: duplicate through
tool.Geometry.duplicate_ifc_objects, unassign the copy from the source
drawing group, assign it to the chosen target group, place it on the
target camera plane at the same world XY, and file it into the target
drawing collection. The originals are left untouched and the user's
selection is restored. The target camera is imported on demand when it
has not been loaded yet.
The operator shows a target drawing dropdown and is reachable from the
annotation tool sidebar when an annotation is selected, and from the
drawings panel. Annotations already in the target drawing are skipped
and reported.
The orchestration lives in core.drawing.copy_annotations_to_drawing
with prophecy tests covering the copy, the skip, and the camera import
branches. Verified live in headless Blender 5.1: the copy is a new
IfcAnnotation with its own GlobalId and IfcTextLiteral, both texts are
editable independently, and everything survives save and reload with
each annotation loading in its own drawing.
Generated with the assistance of an AI coding tool.
(cherry picked from commit bc1fb2a88d)
Fixes#4395.
Root cause: the SVG cut-linework merge step that fuses adjacent
elements' cut polygons together (per the pset-driven JoinCriteria
setting) was hardcoded to only IfcWall and IfcSlab. IfcCovering cut
shapes were skipped unconditionally, so adjacent coverings never
joined, leaving a visible seam/broken corner in section drawings
regardless of JoinCriteria.
Fix: added an EPset_Drawing.JoinClasses property, following the
exact same user-overridable pattern already used by
EPset_Drawing.BringToFront - a comma-separated list of IFC classes
to join, defaulting to "IfcWall,IfcSlab" (unchanged behavior) when
unset. Users can override per-drawing to add IfcCovering (or any
other class) when they want it joined too. Kept this opt-in rather
than hardcoding IfcCovering into the default list, since joining a
thin finish layer the same way as a thick wall/slab could produce
unwanted mitring in some cases - the user decides per drawing.
Verified live against the reporter's own attached file
(ifcovering joining.ifc) and its cached section linework: with
JoinClasses unset, two separate closed paths reproduce the reported
seam exactly. With JoinClasses = "IfcWall,IfcSlab,IfcCovering", the
two coverings merge into a single closed polygon with the internal
seam removed. Confirmed IfcSlab join behavior is unchanged in both
runs.
Generated with the assistance of an AI coding tool.
Co-authored-by: Dion Moult <dionmoult@gmail.com>
(cherry picked from commit c55a79b8b5)
reimport_element_representations() built a fresh
ifcopenshell.geom.settings() without copying deflection_tolerance /
angular_tolerance from the IfcImportSettings it had just
constructed, and never passed geometry_library to either the
iterator() or create_shape() calls it makes. As a result, exiting
Item/edit mode (which reaches this function via
switch_representation) silently fell back to IfcOpenShell's
hard-coded mesher defaults (0.001 linear deflection, ~50x finer than
the project's default of 0.05) and the default geometry kernel,
instead of the project's configured tolerance and Geometry Library.
This made geometry visibly change quality after a no-op Tab into and
back out of edit mode, since the reload path was unintentionally far
more precise (and used a different kernel) than the initial import.
Both settings, and geometry_library, are now taken from the
IfcImportSettings instance already built at the top of the function,
so a reload matches the original import.
Refs #5685.
Generated with the assistance of an AI coding tool.
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
(cherry picked from commit b669baf793)
theoryshaw's follow-up on #6944: after the profile/curve reconstruction fix
(previous commit), the arc/circle marker for a freshly Shift+D-duplicated
loop wouldn't appear until leaving and re-entering Edit Mode.
Root cause: ProfileDecorator groups arc/circle vertices purely by
IFCARCINDEX/IFCCIRCLE vertex-group index every draw call (it has no cache
to go stale, it fully recomputes from the live edit-mesh bmesh each frame).
Duplicating a loop copies its vertex-group weights onto the new geometry,
since Blender allocates no new group for a duplicate, so the source loop
and its live duplicate land in the same dict entry. That entry then fails
the "exactly 2 verts per circle / 3 per arc" check and is skipped entirely,
so BOTH the original and the duplicate stop being drawn until the mesh is
reimported and gets fresh, distinct groups.
Verified live in headless Blender: built a bmesh with an IFCCIRCLE loop and
an IFCARCINDEX loop, then ran bmesh.ops.duplicate on each (the same
bmesh-level operation underlying Shift+D) and called ProfileDecorator's
draw method directly. Before this change, duplicating either loop dropped
both the original and the duplicate from the decorator (0 circle/arc
batches drawn instead of 2). After, both draw immediately, with no change
to the non-duplicated case (still 1) or to genuinely distinct loops (5
independent circles still resolve to 5, not merged). 500-circle timing is
unchanged (~14.3ms/draw before and after), so the added connectivity split
is not a hot-path regression.
Added test/bim/module/model/test_profile_decorator_duplicate_loop.py
pinning the new _connected_components helper's behavior for single and
duplicated circle/arc loops.
This contribution was produced with the assistance of an AI coding tool.
(cherry picked from commit 32ac20e8e3)
auto_detect_profiles had the identical defect fixed in the previous
commit: duplicating a circle/arc loop in Edit Mode reuses the same
IFCCIRCLE/IFCARCINDEX vertex group index for the new geometry, and this
sibling function (used for curve/annotation editing rather than profile
voids) tallied group membership across the whole mesh instead of per
loop, so it also rejected a legitimately duplicated loop as malformed.
Applied the identical fix: scope the group-count sanity check to each
connected edge loop, computed after the loops are built rather than in
the initial whole-mesh vertex pass. Kept the existing forked-loop check
(more than 2 edges per vertex) in the first pass since it is unrelated
to group counting.
Verified live in headless Blender: constructed two 2-vertex IFCCIRCLE
loops sharing one vertex group index (the exact state Blender's Edit
Mode duplicate produces) and called auto_detect_curves directly.
Before this change it returned (False, "CIRCLE"); after, it returns two
valid IfcCircle curves.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 0d5ea02169)
Duplicating a circular or filleted-arc void in the profile CAD editor
(Shift+D on the loop's vertices) reused the same IFCCIRCLE/IFCARCINDEX
vertex group index for the new geometry, since Blender's mesh duplicate
copies vertex group weights but does not allocate a new group. On exit
from Edit Mode, auto_detect_profiles tallied group membership across the
whole mesh rather than per loop, so a group meant to hold exactly 2 (circle)
or 3 (arc) vertices ended up with double that, failing its sanity check
and blocking the edit with an "INVALID PROFILE" popup. Fixes#6944.
Scope the sanity check to each connected edge loop instead, matching how
the loops are actually converted into IfcCircle/arc segments below. Also
explicitly reject an arc/circle vertex tagged onto an isolated vertex with
no edges at all, which the old whole-mesh count also caught.
Verified live in headless Blender against the issue's repro file
(IfcFurniture "Slab.004", IfcArbitraryProfileDefWithVoids with three
IfcCircle voids): entering the profile editor, duplicating one void's
2-vertex loop and moving it produced an "INVALID PROFILE" popup before
this change, and now produces a valid profile (the original 3 voids
intact, plus the duplicate as a 4th void or a separate solid profile
depending on whether it still falls inside the outer boundary).
test/tool/test_model.py passes unchanged (32 passed, 1 pre-existing
unrelated failure present on both before and after).
Generated with the assistance of an AI coding tool.
(cherry picked from commit 0a027d47a3)
Both files were merged unformatted and fail the Black formatter step
on every branch, keeping ci-lint red repo-wide.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 6014bbd877)
The ci-lint workflow's ty steps fail on every branch because base
v0.8.0 has four diagnostics.
ty check (bonsai):
- root/operator.py: bpy.data.objects.get() can return None, so
UnlinkObject._execute could put None in its objects list and crash
on the first attribute access when an unknown object name is passed.
Handle the miss explicitly, which also satisfies the declared
list[bpy.types.Object] type.
- tool/sequence.py: ty does not narrow Literal types through
membership tests on list literals, so the assert_never() exhaustive
check was flagged. Use tuple literals, which ty narrows, keeping the
exhaustiveness check intact.
ty check (ios):
- draw.py: arrange_polygons was called through conditional argument
splats that let the same call site work against pre-April-2026
wrappers lacking arrange_polygon_settings and the logger parameter.
No runtime bug for current builds, but the dynamic splats cannot be
typed against the fixed 3-parameter signature. Drop the old-build
workaround and call the current signature directly, following the
precedent of 3d8115ebc5 which dropped similar old-build workarounds
in ifcopenshell.file. Verified against a current wrapper build that
the direct call arranges polygons and serializes to SVG, with and
without a logger.
- Optimise.py: igraph is an optional dependency with a guarded import
and a toposort fallback, but it was missing from the ios type-check
venv so ty could not resolve it. Add it to type-check-requirements
next to the toposort fallback that is already listed.
After this, poe ty-bonsai and poe ty-ios both pass cleanly.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 6d6d92b849)
The script called Polyline.get_mode() on every modelspace entity, but
that method only exists on POLYLINE entities, so any typical DXF
containing lines, circles or text crashed with AttributeError before
converting anything. Test for POLYLINE polyface meshes with
dxftype()/is_poly_face_mesh instead and skip other entities with a
message, only create the spatial containment relation when products
exist, and take the input/output paths from the command line (matching
obj2ifc.py) instead of a hardcoded input.dxf/test.ifc.
Fixes#2151
This change was written with the assistance of an AI coding tool.
(cherry picked from commit bb49822f2e)
When grabbing an array child, the selection now expands to include
the array parent and all sibling children before the move operator
runs. Mirrors existing behavior for aggregates and nests.
Generated with the assistance of an AI coding tool.
(cherry picked from commit b66d8b2c4d)
Ran the new sync_stub.py against a real local build: adds
context.delete_same_facet_edge_pairs (present on the compiled wrapper,
missing from the stub) and drops the module-level logger_or_root
(present in the stub, no longer exists on the wrapper at all).
Nothing else changes - no license header rewrite, no docstring loss,
none of the 14 hand-curated named-parameter constructor/function
signatures touched, unlike the wholesale regeneration this replaces.
Generated with the assistance of an AI coding tool.
(cherry picked from commit b61f809731)
generate_stub.py (this branch's earlier commit) regenerates
ifcopenshell_wrapper.pyi wholesale from the compiled wrapper: it
reliably fixes real drift, but it also discards everything that isn't
mechanically recoverable from the wrapper alone - the license header,
docstrings, and hand-curated named-parameter signatures for
SWIG-overloaded constructors/functions (SWIG itself always emits
generic `*args` for those, so a regenerator can't tell a deliberate
curation from real drift and just overwrites it).
sync_stub.py takes the smaller-blast-radius approach: it only adds
top-level symbols/class members that are genuinely missing, and only
removes ones that are genuinely gone, cross-checking against
validate_stub.py's own full canonicalisation (via the newly-exposed
get_names_tree()) so it never mistakes a property()/staticmethod()-
wrapped member for something absent just because its own narrower
parser skips that form. Anything that exists on both sides under the
same name but with a different signature - exactly where curation
lives - is left untouched and reported for a human to review instead
of guessed at.
Verified against a real local build: applying it to the current
ifcopenshell_wrapper.pyi produces a small, targeted diff (add one
missing method, drop one stale function) with the license header,
docstrings, and all 14 curated constructor/function signatures
preserved byte-for-byte, versus generate_stub.py's ~1000-line
wholesale rewrite for the same underlying fix.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 948ffce7e9)
set_shape_transparency() called AIS_InteractiveContext.SetTransparency(),
whose argument count is inconsistent across pythonocc-core versions
(reported as a TypeError in #1037). Set transparency directly on the AIS
object instead, the same stable pattern already used elsewhere in this
file (display_shape() calls ais.SetTransparency() directly, never through
the Context), then call Context.UpdateCurrentViewer() to refresh.
app.py's viewer used a "SetSelectionPriority(counter)"/"SelectionPriority()"
pair as an ad hoc unique key to map a displayed AIS object back to its IFC
product. On modern pythonocc-core this crashed with AttributeError because
.GetObject() (needed to unwrap the old handle-based API) no longer exists
on AIS objects (#1098, PR #1113 partially patched one of the two call
sites but left the one in HandleSelection unguarded).
Live pythonocc-core 7.9.3 testing showed the GetObject() guard alone is
not sufficient: SetSelectionPriority/SelectionPriority themselves have
been removed from AIS_InteractiveObject entirely in modern OCCT (only
AIS_Trihedron keeps a same-named but unrelated method for datum parts),
so gating the .GetObject() call with the existing USE_OCCT_HANDLE flag
would still crash the first time a shape is selected. Verified live that
AIS objects retain correct __eq__/__hash__ (matching the underlying OCCT
instance) across separate SWIG wrapper instances, so ais_to_product is
now keyed directly by the AIS object itself, removing the dependency on
the removed OCCT API and the GetObject()/handle distinction altogether.
Verified live against pythonocc-core 7.9.3 (conda-forge) using real
AIS_Shape objects obtained from ifcopenshell.geom.occ_utils.display_shape()
and a real IFC file: reproduced both the original TypeError (#1037) and
AttributeError (#1098), confirmed both fixes resolve them, and confirmed
the ais_to_product dict lookup round trips correctly through a real
Context.Select()/SelectedInteractive() call. Could not exercise the full
Qt-embedded viewer.finished()/HandleSelection() flow end to end because
this pythonocc-core build segfaults natively when creating a second GL
context inside a Qt widget on this macOS host, a pre-existing environment
issue unrelated to this diff (reproduces identically with unpatched code,
before any touched line executes).
AI-generated, reviewed and tested by Petru Conduraru.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
(cherry picked from commit 6603c8459a)
create_shape() returns a Python-owned Element (SWIG_POINTER_OWN in the
boost::variant out typemap). Its .geometry property calls Element::geometry(),
which returns a reference into the element's boost::shared_ptr<Representation>
_geometry member. SWIG wraps that reference as a non-owning pointer, so the
returned Triangulation/BRep/Serialization proxy does not keep the element alive.
When a caller keeps only .geometry (e.g. create_shape(s, e).geometry) and drops
the parent element, Python garbage-collects the element, destroying its
shared_ptr and freeing the underlying representation. Subsequent reads of
verts/faces then return freed memory: empty or implausible float/int garbage,
non-deterministically depending on GC and allocator timing. This is silent data
corruption, not a crash, and has bitten users since 2020.
Fix: in the TriangulationElement/SerializedElement/BRepElement pythoncode, wrap
the geometry getter so the returned geometry stores a backreference to its
owning element (result._parent = self). This makes the parent's lifetime at
least as long as the geometry's, automatically and transparently, so no caller
has to remember to hold the element. This is aothms's suggested backreference,
applied generically in the binding rather than left as a workaround.
Reproduced deterministically (washBasin fixture): before, verts len 0 vs 133500
across repeated GC-pressure runs; after, 133500 every run for all three element
types. test_create_shape passes; no regressions.
Note: tree.select_ray()'s ray_intersection_result (2024 follow-up in #1124) is a
separate ownership mechanism (std::vector element reference + std::array member
pointer) and is left as follow-up scope.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
(cherry picked from commit 824c1fc280)
recalculate_cut() and recalculate_fill() each ran is_intersecting_camera(),
which builds a bmesh and scans every vertex. When a redraw recalculated both
(camera moved, cache miss, or the object selected) that was two full
intersection tests per object per frame for the same answer.
Compute it once in decorate() and pass it to both, and skip the test
entirely when neither recalculation is needed. Never more tests than before,
identical result since the camera can't move within a frame.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 705af7ba3a)
is_camera_moved() runs eval()/numpy over the camera matrix and, as a side
effect, refreshes the stored checksum the first time it returns True. It was
called up to twice per object inside decorate(), so on a frame where the
camera actually moved the first call updated the checksum and every later
call - the fill check on the same object, and both checks on all remaining
objects - then saw an already-current checksum and returned False. Only the
first object's cut got recalculated; its fill and every other element stayed
stale until something else invalidated the cache.
Evaluate it once at the top of __call__ and reuse the flag. This halves the
per-object eval overhead on the common path (viewport navigation with the
camera object stationary) and, when the camera does move, correctly
recalculates the cut and fill for every intersecting element instead of just
the first.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 074fc26e8f)
Deleting a type used to strip its occurrences: any that displayed the
type's mapped representation lost their geometry, and inherited material
and presentation styles were dropped too.
The no-SHIFT "Delete Type" path now bakes each occurrence's geometry,
styles, and inherited material onto the occurrence before the type is
removed:
- Refactor UnassignType's unmap logic into a reusable
UnassignType.unassign_and_unmap(), and extend it to re-attach styled
items (copy_deep only follows forward refs, so IfcStyledItem is lost)
and bake down any inherited (non-owned) material.
- Add RemoveType._detach_type_material_set(): unhook the type's
IfcMaterialLayerSet/ProfileSet association cascade-free before deletion,
so remove_product's aggressive unassign_material never fires and the
occurrences' layer/profile-set usages survive intact.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit a90064929b)
Adds a trash button in BIM_PT_type_attributes that deletes the relating
type via bim.remove_type. SHIFT+Click also deletes every occurrence of
the type in the project, behind a confirmation dialog showing the count.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 397f13e71c)
aothms asked for the toposort dependency ordering used by the dedup
walk to try igraph's C-backed topological_sorting() first, since it
should shave off additional time on top of the non-recursive
get_info fix. Falls back to the pure python toposort package with a
warning if igraph is not installed.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 7ebdd046b6)
The 2020 profiling in issue #1043 found the Optimise recipe's dedup
loop spent almost all of its time in entity_instance.get_info(recursive=True):
because the topological sort already guarantees every referenced entity
is folded before the entity that references it, recomputing each
already-folded subtree's canonical value from scratch for every parent
that points to it is wasted work. Confirmed this is still exactly the
bottleneck in the current codebase, unchanged since 2020 (get_info's
recursive path still walks the whole subtree on every call).
Applied aothms's suggested fix from the issue thread: canonicalize each
entity with a non-recursive get_info, and for referenced entities substitute
the already-computed identity of their folded replacement (looked up in
instance_mapping) instead of re-expanding the subtree. Also limited the
toposort dependency graph to direct references (max_levels=1), since a
topological sort only needs direct edges, not the full transitive closure
traverse() was computing for every entity.
Benchmarked before and after on real IFC test fixtures and a larger
synthetic file with heavily shared geometry (thousands of walls sharing
a handful of profile/point subtrees, mirroring the sharing pattern
described in the issue):
- test/input/geometrygym_great_court_roof.ifc (56989 entities): 9.9s -> 1.7s
- test/input/acad2010_objects.ifc (16296 entities): 3.7s -> 0.4s
- synthetic 120083-entity fixture with heavy geometry sharing: 19.2s -> 3.4s
Verified correctness by comparing the full canonical (recursive get_info)
multiset of the optimized output between the old and new implementation on
all three fixtures: identical results, same fold counts.
Added test_Optimise.py covering the core scenario from the issue: entities
built from separate, value-identical non-rooted subtrees fold to a shared
instance, while entities with distinct values do not.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 57cfd9d1fd)
IfcSurfaceCurveSweptAreaSolid regressed in 0.8 for geometry far from the
origin (for example parapets on a georeferenced building), which went
missing or glitched.
The kernel offsets the directrix toward the origin when it is far away
(mean.norm() > 1e2), storing the offset copy in a local curve variable and
setting applied_temporary_offset so the finished solid is translated back by
+mean. But the wire was still built from scs->curve, the un-offset original,
so the offset never took effect and the result was translated by +mean from
its correct location. Build the wire from curve instead. When no offset is
applied curve aliases scs->curve, so near-origin geometry is unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit e333c1c100)
Two related bugs in read_from_stream's reference-resolution
loop, both reachable from malformed input:
- has_attribute_value<IfcBaseClass*> only checks the stored
slot's type, not that it's non-null (e.g. an explicit $
value), so the following get_attribute_value() call could
return null and inst->declaration() crashed on it.
- byid_[ref] default-inserts (and returns) a null pointer
when the owning instance id isn't present, which was then
dereferenced unconditionally via ->data().
Added regression tests using the two minimized crash inputs
that found these.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 56121ca061)
`values` dictionary was missing and variables were never collected to it, so `FormulaEvaluator(values)` was always resulting in missing variable error.
(cherry picked from commit 2e21fc5a98)
Ports #96e2efebc onto wgpu. wgpu's boolean_utils logged via the global
::logger::root() singleton (which IfcConvert never wires to --log-file),
so boolean-op messages were effectively dropped. Thread the caller's
injected logger through instead:
- boolean_settings gains `::logger* logger` + `log()` accessor (falls
back to ::logger::root()); boolean_operation logs via settings.log()
- eliminate_narrow_operands / boolean_subtraction_2d_using_builder take a
`::logger& logger = ::logger::root()` param; boolean_operation passes
settings.log() into them
- OpenCascadeKernel / boolean_result set bst.logger = &logger_ and log
via logger_ (were ::logger::root())
Adapted from v0.8.0's Logger/Logger::Root() to wgpu's ::logger/::logger::root().
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The class-pairing validation added in 10ee5aef4f rejects any type
assignment whose class isn't in the buildingSMART implementer
agreement map. That map only covers physical product occurrence/type
pairs (IfcWallType -> IfcWall, etc); IfcTypeProcess and IfcTypeResource
subtypes such as IfcTaskType, IfcProcedureType and the resource types
have no entry, so previously-valid assignments like
IfcTaskType -> IfcTask were rejected with "allowed occurrence
classes: <none>".
These classes still follow the schema's universal Type-suffix naming
convention, so derive the pairing the same way the existing
ApplicableOccurrence fallback does: strip "Type" from the relating
type's class name and accept it only if the schema actually declares
that entity. This can only add pairings implied by the type's own
class name, so it cannot loosen the existing rejection of genuine
mismatches (e.g. IfcWallType -> IfcWindow).
Generated with the assistance of an AI coding tool.
(cherry picked from commit d188e3beaf)
Ports the 7-commit v0.8.0 SVG edge-classification feature (f0970b90b +
6 follow-ups) onto wgpu's heavily-diverged serializer. Reconstructed
block-by-block rather than merged, because both sides rewrote
SvgSerializer (v0.8.0 +419, wgpu +778) and git's conflict alignment was
misleading.
Key wgpu adaptations reasoned per block:
- IfcUtil::IfcBaseEntity* (pointer identity) -> express::Base (value),
incl. as a std::map key in draw_hlr (express::Base has operator<);
nullptr fallback -> express::Base{}
- boost::optional -> std::optional (css_class, dash_array)
- hlr_calc::result_type pair -> 3-tuple (adds per-edge class label)
- draw_hlr restructured with a group_by_product map: ONE path_object per
product so multiple class buckets share a group and per-path classes
survive Bonsai's merge (NOT naive per-item threading, which fragments
groups -- caught during visual verification)
- settings wired into wgpu's apply_settings() (ctor-called), NOT the
feature's added ready() call which wgpu already solved differently
- logger_.Warning -> logger().warning in write(geometry_data)
Verified end-to-end via IfcConvert on a curved-geometry stress scene
(Suzanne/Torus/Sphere/Cube/Cone/...): edges classify into outline/sharp/
crease/boundary/flush with correct per-product grouping and CSS styling.
Requires the model's storey to carry an Elevation and --svg-project to
trigger the HLR projection.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The Reassign Class operator refused to reassign an element to a different
IFC product family unless it was an IfcElement <-> IfcElementType swap, so a
piece of geometry mistakenly hosted on IfcSite could not be turned into
IfcFurniture even though root.reassign_class handles it fine.
Loosen the guard: only block the case that actually matters - a spatial
element (IfcSpatialElement / IfcSpatialStructureElement for IFC2X3) with no
geometry, which would be a real containment-hierarchy container rather than
a stray modelled object. Everything else reassigns freely.
Closes#8664
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit b5a0f1fc74)
When converting a wall representation to a parametric extrusion via the
Representation Utilities buttons, an element that has openings would report
"has openings - representation cannot be updated" and stop, without telling
the user there is an ALT+click path that bakes the openings into the new
representation. Point the message at that path so the error is actionable.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 25441bd816)
* Bonsai: place auto-generated opening boundaries at their real position #8237
auto_generate_boundaries (single-space mode) built each opening/filling boundary
from the opening's LOCAL geometry (get_vertices) but first did
mat.translation = (0, 0, 0) on its placement matrix. Because the vertices are
local, that placement translation is exactly what carries the opening to its
real location, so zeroing it collapsed every window/door boundary onto the
origin. This is why the auto path misplaced window boundaries while the
single-element path (create_element_boundary) placed them correctly, as
@MDHering observed with the two modes. Keep the full placement matrix.
Verified on the reporter's file: the opening's real placement is (0.1, 1.5, 1.0);
a vertex went from (0.6, 0, 0) under the old code to (0.7, 1.5, 1.0) with the fix,
i.e. moved by exactly the (0.1, 1.5, 1.0) that was being discarded.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Remove superfluous comment from #8237 fix
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: CyrilWaechter <cyril@biminsight.ch>
(cherry picked from commit 65811ac7c9)
Last reference to this file was dropped in 7ae685dbf, though the ref was
pointing to `/patches/opencollada/pr622.patch`, so IIUC
`patches/pr622.patch` was never used.
(cherry picked from commit c013b9aca7)
Introduced in e21277e80, reference removed
in 683cadeb7 when occt was bumped to 7.3.0 and switched to git-tag based
download.
(cherry picked from commit 24e454ce0c)
Surveyor test was failing because `get_z_rotation` and `set_z_rotation` were not implemented.
The code was added in 230cbe1fd8, but it was never used.
(cherry picked from commit 9e25c12b16)
My resolution of #25 (e38993909) over-took v0.8.0's explicit
aggregate_of<...>::ptr type, which the rewrite renamed. Every other
get_related call in this file uses auto (it's a deduced-return template);
wgpu's HEAD already used auto here. Only the SCHEMA_HAS_ macro typo fix
was actually needed. Revert the type to auto.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
versionURLs in brand.html used http:// while the docs sites are
served over https://, so currentURL.includes(url) never matched and
the <select> never reflected/switched to Unstable. Fixes#8023.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 780739719f)
Two TestImplementsTool failures on v0.8.0:
- test_cost.py: Cost could not be instantiated because
core.tool.Cost declared abstract get_direct_cost_item_products, which
tool.cost.Cost never implements. The method is dead (zero call sites;
get_cost_item_products(is_deep=False) already covers the 'direct'
case), so remove the abstract declaration.
- test_ifcgit.py: tool.ifcgit.IfcGit was not declared as a subclass of
its core.tool.IfcGit interface (unlike every sibling tool class), so
the isinstance check failed. Add the base class (and the
bonsai.core.tool import it needs). All 50 interface methods are
already implemented on the concrete class.
No behaviour change. Verified in headless Blender: isinstance(Cost(), core.tool.Cost) and isinstance(IfcGit(), core.tool.IfcGit) both True (were TypeError / False); repo abstract-vs-impl diff confirms all IfcGit abstracts are implemented.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 4a717ca7ff)
black (the version CI's psf/black@stable resolves to) flags three spots
in util/selector.py: the chained .replace() in FormatTransformer.number,
the suppress_zero_inches kwarg in format_length, and the long
`elif key in (...) and hasattr(...)` placement-key tuple in
set_element_value. Reformat all three to black's multi-line style.
Formatting only, no behavioural change (all keys preserved).
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 5a831e3d21)
generate_annotation built the annotation list from a set union and sorted it by
ZIndex and TEXT-ness only. Annotations that tied on that key kept set iteration
order, which follows entity hash (step id plus the process memory address), so
the order of tied annotations (for example a label and its background fill)
shuffled between Blender restarts and flipped their draw order.
Add the stable IFC step id as a final tiebreaker so the order is total and
session independent. Behavior preserving, no z-layer semantics changed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit d30286225c)
coerce_value assumed value_str was always a CLI string, but ifcmcp
passes JSON-decoded native types (int, None) straight through. Guard
the Union/Optional "none" check so it only calls .lower() on strings,
and handle native None explicitly.
(cherry picked from commit 65695fb878)
IfcSpace is not a subtype of IfcElement, so quantify.run_quantify()'s
default selector silently skipped all spaces, reporting
elements_quantified: 0 with no error or warning.
Generated with the assistance of an AI coding tool.
(cherry picked from commit ab15750747)
Importing a Primavera P6 XML crashed with
`AttributeError: 'NoneType' object has no attribute 'text'` in
P62Ifc.parse_activity_xml, which read
activity.find("pr:CalendarObjectId").text unconditionally. CalendarObjectId
is optional on a P6 Activity; when omitted, the activity inherits the
project's ActivityDefaultCalendarObjectId.
Capture the project default in parse_xml and fall back to it when an
activity has no CalendarObjectId (`calendar_id or self.default_calendar_id`).
Verified on the reporter's attached file (20241021 Cronograma.xml): 3 of 14
activities lack a CalendarObjectId and reproduced the exact crash on
v0.8.0; after the fix parse_xml completes and those activities resolve to
the project default calendar "2" (a valid calendar in the file). An
activity with an explicit CalendarObjectId keeps its own value.
Fixes the P6 re-import crash reported in #5617 (that issue tracks several
Gantt items; this addresses the import AttributeError).
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 694a44e638)
get_pset and get_psets assumed RelatingPropertyDefinition is a single property
definition and read definition.Name directly. When it is an
IfcPropertySetDefinitionSet (a defined type wrapping a list of property set
definitions) that attribute access raised AttributeError, so an element whose
psets are grouped in a set returned none of them.
Unpack IfcPropertySetDefinitionSet into its members in both loops and process
each one. Single property definitions and the psets_only and qtos_only filters
are unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit a3950ac191)
Several BDD scenarios hardcode absolute representation-item object names
whose trailing number is the IFC STEP line id
(f"Item/{item.is_a()}/{item.id()}"). Those ids drift when file-creation
order changes; a recent shift moved all of them by a uniform -4, so the
scenarios failed with "Item/.../NN does not exist".
The failing step (the_object_name_exists in test_feature.py) dumps the
full bpy.data.objects listing on failure, so the correct current ids are
recoverable directly from the CI log (run 29208793599, tested commit
36e21e882f, an ancestor of HEAD with only a .gitignore commit between).
Renumber to match:
IfcExtrudedAreaSolid/77->73, IfcPolygonalFaceSet/76->72,
IfcVertexPoint/69->65, IfcEdge/72->68, IfcFace/74->70.
Verified against the CI failure dump (a local build produces different
ids, so this is validated by CI's own object listing rather than a local
run). boolean.feature also hardcodes IfcHalfSpaceSolid/90 and panel text
[91] downstream of the failing assertion, which CI never reached and so
never dumped; left as-is to avoid guessing - they will print a fresh dump
next run for a follow-up if still stale.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 6f90badda8)
Two host-environment bugs in the build-env scripts that break on
macOS/Apple Silicon hosts, independent of target architecture:
- Dockerfile: groupadd fails outright when USER_GID collides with an
existing system group in the rockylinux9 base image (e.g. macOS
default user GID 20 "staff" collides with RHEL's GID 20 "games").
Guard with getent so useradd attaches to the existing group instead.
- ifcos_env: `sed -si` is GNU-only syntax and errors under BSD/macOS
sed. Do the UNIQUE_ID substitution via a portable temp-file + mv.
Per sboddy's review on the original PR: dropped the linux/amd64
platform-pin additions from this change. The stack already targets
Rocky9/x64 build outputs by design, and Docker Desktop on macOS has
no native container runtime regardless (it's a Linux VM either way),
so forcing the image to run under emulation doesn't produce anything
that's actually loadable into a native macOS Blender/Bonsai install.
That's a separate, harder problem worth solving via a native build
path instead (mirroring build_osx.yml), not by fighting emulation
here. These two fixes stand on their own merits on any host.
This change was made with the assistance of an AI tool.
(cherry picked from commit 8b05510d6c)
The ccache named volume had no explicit name, so Docker Compose
namespaced it under the per-checkout project name (derived from
UNIQUE_ID), giving each checkout its own cache even though
docker/README.md already documented them as shared. Give the volume
a fixed name so all checkouts attach the same one.
Measured cache size after a full build (IfcParse+IfcGeom+IfcConvert+
wrapper, one Python version) is ~300MB, only ~5% of the previous 5G
cap. Shrink CCACHE_MAXSIZE to 2G, which comfortably covers the shared
baseline plus per-branch deltas from several diverging checkouts.
Generated with the assistance of an AI coding tool.
(cherry picked from commit b1470223d3)
Three host-portability fixes to the docker/ toolchain from #8564 so it
runs on macOS as well as Linux. All three are no-ops on native amd64
Linux.
1. Dockerfile: only groupadd when the target GID is free. macOS's default
primary group `staff` is GID 20, which already exists as `games` in
rockylinux:9, so `groupadd -g 20` aborted the image build. Guard with
`getent group "${USER_GID}" || groupadd ...`; useradd -g accepts the
existing GID.
2. ifcos_env unique(): replace GNU-only `sed -si` (BSD/macOS sed errors
"illegal option -- s") with a portable `sed > tmp && mv` rewrite of the
UNIQUE_ID line. Verified against macOS BSD sed.
3. create() + compose.yaml: build with an explicit `--platform linux/amd64`
so the locally built image's platform matches the `platform:
linux/amd64` pin in compose.yaml. Without it, on arm64 the local image
is tagged linux/arm64, compose treats the platform-mismatched image as
absent and tries to pull `ifcopenshell-build-env:updated` from Docker
Hub (which does not exist -> access denied). Also add `pull_policy:
never` as a safety net so a future mismatch surfaces as a clear "image
not found" rather than a registry auth error.
Note: on Apple Silicon the amd64 build runs under emulation and a cold
full build is slow; ccache makes incremental rebuilds tolerable. A native
Linux/Intel host or CI remains the better choice for routine use, but these
fixes turn "hard broken" into "works with a caveat" on macOS.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit f25b072fa0)
Reported: Blender segfaults when clicking Cancel on the "newer
autosave found" recovery popup shown by LoadProject at startup.
Root cause: LoadProject.execute()/invoke() triggered the recovery
popup via bpy.ops.bim.load_autosaved_recovery_popup("INVOKE_DEFAULT",
...) and returned that call's result ({'RUNNING_MODAL'}) as their own
return value, without LoadProject itself ever calling
modal_handler_add(). Blender's window manager takes a RUNNING_MODAL
return as a promise the operator registered its own modal handler;
since it hadn't, the WM's operator bookkeeping was left corrupted -
silently, since this is heap/state corruption rather than an
immediate crash. It only surfaced later, when the real modal operator
(the popup) closed and the WM reconciled its modal stack, which lines
up with the crash occurring specifically on dialog close regardless
of which button was pressed. check_autosave_recovery() now returns a
plain bool and fires the popup fire-and-forget; LoadProject reports
its own honest {"FINISHED"}.
Also hardened, as defense in depth: LoadAutosavedRecoveryPopup's
execute()/cancel() call back into bim.load_project(...), which (with
should_start_fresh_session) calls wm.read_homefile() and tears down
the window manager/screens. Doing that synchronously from inside this
popup's own execute()/cancel() - itself invoked from deep inside
Blender's modal handling for the popup's button click - risks the
same class of use-after-free as the timer bug fixed in the previous
commit. The reload is now deferred by one timer tick so it runs after
the popup's modal handling has fully unwound, and the deferred
callback closes over plain values rather than `self`, since the
operator instance may not survive past cancel()/execute() returning.
This defer-only change was tried and tested first, on the (incorrect)
assumption it was the root cause: it produced a byte-for-byte
identical crash backtrace on retest, which is what pointed at the
RUNNING_MODAL bug above as the actual cause - the defer change alone
was insufficient because the corruption happens when the popup is
first shown, not when it's closed.
Generated with the assistance of an AI coding tool.
(cherry picked from commit d0eca6fa90)
The periodic autosave timer called reset_timer() at the end of its
own callback, which unregistered the timer that was still executing
(itself). Blender frees the timer's internal registry entry on that
manual unregister, then frees it again when the callback returns
None - a double free that corrupts the heap and can crash Blender
later, once the corrupted memory is reused.
Reschedule by returning the next interval from the callback instead,
which is the safe, documented way to repeat a bpy.app.timers
callback. External reset_timer() calls (from SaveProject,
LoadProject, AutosavePrompt) are unaffected since they run from a
separate call stack (UI events), not from inside the timer.
Found while investigating a segfault reported when cancelling the
autosave recovery popup; not itself the cause of that crash (see the
following commit), but the same reentrant-unregister pattern and a
real, independent latent bug in the periodic reminder path.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 6306ce0f80)
Dockerfile (renamed from Dockerfile_init, Dockerfile_update removed):
- Run as a non-root `builder` user matching the host UID/GID (passed as
--build-arg by create() from id -u/id -g), so build output under the
bind mount stays owned by the host user instead of root.
- Fix CCACHE_MAXSIZE: `ccache -M 5G` wrote its limit to a config file
under /ccache at image-build time, but /ccache is a volume mount
point, so that file gets shadowed by the (empty) volume the moment
the container actually runs - the cap never took effect. Set
CCACHE_MAXSIZE=5G as an image ENV instead.
- Dedupe ccache/libffi-devel, add --setopt=install_weak_deps=False
--setopt=tsflags=nodocs, add `git lfs install --system`, combine the
dnf update+install into one layer.
- Drop Dockerfile_update: it built FROM its own previous output, so
every `update` call made the image strictly larger forever (Docker
layers are append-only, `dnf clean` in a later layer can't shrink an
earlier one). `update` now just calls create(), which already runs
`dnf update -y` FROM a clean rockylinux:9 every time.
compose.yaml: pin platform: linux/amd64 so this doesn't silently run
under emulation on an ARM host.
ifcos_env:
- Split the previously-conflated stop/down into six distinct,
Compose-native lifecycle commands: up (create-or-start), down
(remove), stop, start, restart (stop+start, same container),
recreate (down+up, fresh container). Previously `stop` was aliased
to `down`, which silently removed the container instead of pausing
it.
- Implement try(): copies the built wrapper into a real Blender/Bonsai
install for manual testing, reading the target from a new
BLENDER_USER_RESOURCE .env variable and auto-detecting the built
Python version (disambiguating via PY_TGT for multi-version builds).
Deliberately kept human-only - it mutates a live Blender install, so
it shouldn't run unattended as part of an automated/AI workflow,
which should instead copy the wrapper into the repo's own
src/ifcopenshell-python/ifcopenshell/ (documented in SKILL.md).
- Fix unique(): the "has .env already got a UNIQUE_ID line" check
referenced an unset $FILE instead of $ENV_FILE, so it always
evaluated true and appended a fresh "UNIQUE_ID=dummy" line to .env
on every single `up`.
- Minor: differentiate remove()'s log message from down()'s (no longer
identical now that they're distinct operations), tidy help text
alignment and a stray double-space typo in clean().
SKILL.md: rewritten as current-state documentation (no more "fixed in
this copy" changelog framing) covering the above, plus a migration
note for anyone hitting root-owned leftovers from an older image.
Verified by actually building the image and driving every new
lifecycle command (stop/start/restart keep the same container ID;
down+up and recreate produce a new one) and try() (including the
quoted-tilde BLENDER_USER_RESOURCE edge case) against the real container.
Generated with the assistance of an AI coding tool.
(cherry picked from commit 92c50ed3b4)
First functional version, but it needs some improvements and fixes
identified as I've used it personally on one thing, and when an AI
(Claude) used it to work through the CI test errors.
I had the AI make a SKILL.md file. If the AI indicates it needs to
build the ifcopenshell binary, use this and let it rip.
(cherry picked from commit fa98aad469)
The project-unit to Blender-unit mapping in format_distance only knew
FOOT/INCH/METRE/DECIMETRE/CENTIMETRE/MILLIMETRE, so creating a project
with Kilometers or Miles in the New Project Wizard crashed with
KeyError: 'KILOMETRE' (or 'MILE') as soon as the spatial tree formatted
an elevation. Add the missing Blender-supported units (kilometre, mile,
micrometre) and fall through gracefully for anything else (for example
HECTOMETRE) so unknown units use the adaptive formatting branch instead
of raising.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 980988f208)
Expose the Euler rotation of an element's placement in degrees through
get_element_value, alongside the existing x/y/z and easting/northing/
elevation keys. This makes element rotation exportable through ifccsv,
e.g. for placing oriented symbols in GIS.
Adopts the approach agreed in the review of the stale PR #6272 by
@TZwielehner: reuse util.shape_builder.np_matrix_to_euler and do the
degree conversion inside get_element_value.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit d4805387ef)
Project loading set scene length_unit to f"{Prefix}METERS", but Blender's
enum only defines KILOMETERS, CENTIMETERS, MILLIMETERS and MICROMETERS.
A model with a DECIMETRE (or HECTO/DECA/etc.) length unit therefore raised
on the enum assignment and the file failed to open. Guard with the set of
supported values and fall back to ADAPTIVE display for the rest.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 69a4be68e8)
The quickstart ended with three empty sections whose bodies were only
"TODO" (placing occurrences, changing locations, modeling a building),
which read as a dead end on docs.bonsaibim.org. The page now ends on the
completed save-and-view flow.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit 06da416b8f)
The guard that avoids re-assigning the same object to the same resource
tested is_a("IfclRelAssignsToResource") (stray "l"), so it never matched.
A repeat assignment therefore fell through and appended the related object
to RelatedObjects a second time. Corrected to "IfcRelAssignsToResource".
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 21ae78fbc2)
IfcDiff defaulted to relationships=["geometry"], so a plain diff only ever
compared geometry. Attribute-only edits on an element that kept its GlobalId
(a modified or removed PredefinedType, a renamed element, etc.) were silently
missed. The CLI made this worse: --relationships did not list "attributes" or
"geometry" as valid values, so there was no documented way to enable it.
The default is now ["attributes", "geometry"], so a plain `ifcdiff old new`
reports attribute changes alongside geometry changes. The CLI help and the
IfcDiff docstring now document all valid relationship values.
Added a regression test covering a PredefinedType change detected with the
default configuration.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
(cherry picked from commit 0a8ae14789)
FormatTransformer.round() called Decimal() directly on the input value,
which raises decimal.InvalidOperation when the value is a non-numeric
string (a text property, or a value carrying a unit suffix like "12.5 m").
In a spreadsheet export this crashed the entire operation as soon as one
element carried such a value.
Now round() catches InvalidOperation and returns the value unchanged, the
same graceful-fallback convention used by add(). Numeric rounding is
unaffected.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
(cherry picked from commit 2eea7728d2)
In the cobie24 Coordinate sheet, Floor rows use get_local_placement, whose values
are in the project length unit, but Space rows come from ifcopenshell.geom
create_shape, whose vertices are in SI metres, and the space branch never scaled
them back. So on a non metre model (for example millimetres) the Coordinate sheet
mixed units a thousandfold apart and disagreed with the Facility sheet's declared
LinearUnits.
Scale the space bounding box by the project unit scale so the whole Coordinate
sheet is consistent. A metre model is unchanged since the scale is 1.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 6b3cc54afc)
The filter_elements selector grammar had no way to comment out part of a
query, so users had to delete and retype text to temporarily toggle a
facet. Add a /* ... */ block comment terminal that is ignored by the
lexer, and tolerate a trailing "+" so that commenting out the final
operand (e.g. "IfcWall + /* IfcSlab */") parses cleanly. Comments may
span multiple lines; a /* sequence inside a quoted string is not treated
as a comment. Only the filter grammar is affected, not get_element or
format which use "/" for regex and division.
Adds a regression test and documents the syntax.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 5c11946470)
The location and parent filters both match at any depth in the spatial
hierarchy, which surprises users who want only the elements immediately
under a given container. Document that the parent query key resolves the
direct parent only (e.g. query:"parent.Name"="My Site"), add a matching
filter example, and note the immediacy on the parent value key.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 0b7e25a3ef)
SHIFT+CTRL+CLICK on Activate Drawing now imports the
annotations of all selected drawings without switching
the active view or camera, then selects their cameras with
the first as active. SHIFT+CTRL+ALT+CLICK also selects the
loaded annotation objects. The drawing camera is imported
when missing so annotations land in the correct collection.
Loading is idempotent.
Generated with the assistance of an AI coding tool.
(cherry picked from commit d16c283aef)
Converting to a path whose directory does not exist (or is not writable)
failed silently: the serializer's ready() check correctly returned false,
but IfcConvert deleted the temp file and returned EXIT_FAILURE without any
message, so the user saw no reason for the failure.
Log a SYS error naming the output file before returning, matching the
existing "Unable to open output file" reporting used elsewhere.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit a0f493b471)
Property sets contained in an IfcPropertySetDefinitionSet were exported as
an empty element in XML. The XmlSerializer already had a block to expand
such a set into its member property sets, but it was gated behind
emits SCHEMA_HAS_IfcPropertySetDefinitionSet (singular). The plural spelling
is defined nowhere, so the block was dead code and a RelatingPropertyDefinition
holding a set produced nothing.
Correct the macro name so the set is expanded and its property sets are
serialized. The parse layer already reads these nested sets (they are
reachable from util.element), so this only completes the XML path.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit e389939092)
escape_xml escaped the five XML metacharacters but passed control
characters (0x00 to 0x1F other than tab, newline and carriage return)
through unchanged. Those bytes are illegal in XML 1.0 and cannot be
represented even as numeric character references, so any IFC string
containing them produced non-well-formed XML and SVG output.
Strip those illegal control characters before escaping. Bytes belonging to
a valid UTF-8 multibyte sequence are always >= 0x80, so filtering on the low
control range leaves real text intact. This is the shared helper used by the
SVG serializer text and attribute sites (audited: all route through it) and
by the XML/Collada paths, so both reports are resolved at one place.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 380675e214)
Fixes clean-but-broken breakage from replayed v0.8.0 commits that
compiled on v0.8.0's API but not wgpu's renamed one (caught by the
checkpoint build, not by any merge conflict):
- face.cpp: logger().Warning -> warning (from #527)
- IfcAsymmetricIShapeProfileDef.cpp (from #1367): map_impl takes a
reference not a pointer (matches wgpu's BIND convention); inst-> -> inst.;
boost get_value_or -> std::optional value_or; logger_.Message/Logger:: ->
message/::logger::
- IfcTriangulatedFaceSet.cpp: inst->PnIndex() -> inst. (my own port slip;
wgpu's triangulated map_impl is also a reference)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
IfcTriangulatedFaceSet and IfcPolygonalFaceSet used CoordIndex values to
index Coordinates.CoordList directly, ignoring the optional PnIndex
attribute. When PnIndex is present it remaps point references, so a
CoordIndex value i must resolve as CoordList[PnIndex[i-1]-1] (both 1-based).
Without the indirection any model carrying a PnIndex was built from the wrong
points.
Add a resolve() helper in both mappings that applies the PnIndex indirection
when present and is a plain bounds-checked lookup otherwise, with bounds
checks at both index levels. When PnIndex is absent the behavior is unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 3e55c5126c)
Address maintainer request on #8368: instead of a deflection floor on top
of a fixed CircleSegments count, use one mode or the other. When
CircleSegments == 0 (the new default) the CGAL kernel derives the conic
segment count from MesherLinearDeflection, matching the deflection based
meshing OpenCascade already does and fixing #8051. When CircleSegments is
non zero it is used directly as a fixed, radius independent count.
CircleSegments is only read by the CGAL kernel; OpenCascade meshes by
deflection and never reads it, so the new default has no effect there.
Update the setting description and the ifcconvert / geometry-settings docs.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 0d70812641)
The CGAL kernels (cgal and cgal-simple) allocate arc segments as a
fraction of the full circle via CircleSegments, ignoring the radius.
A large-radius arc that spans a small angle therefore collapsed to a
single chord, turning curved curtain-wall mullions straight while the
OpenCascade kernel (which meshes by deflection) kept them curved.
evaluate_conic now also enforces a deflection-based floor on the number
of segments, keeping the chord deviation within mesher-linear-deflection,
matching OpenCascade. Small circles are unchanged (CircleSegments floor
still dominates); only large-radius curves get denser.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit dd9fa65629)
IfcConvert returned a success exit code even when geometry conversion logged
errors and silently dropped elements (for example a failed TopoDS::Shell build
under layerset slicing produced valid looking output with most objects
missing), so CI and scripts could not detect a partial conversion.
Add an opt-in --fail-on-error flag that makes IfcConvert exit non-zero when any
error was logged during processing, reusing the existing MaxSeverity based
failure check already used for --validate. The default exit behaviour is
unchanged, so pipelines that tolerate individual element failures are
unaffected.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit eb7324e7fc)
A face whose inner boundary crosses the outer boundary (or another inner
boundary) is invalid per the schema. Open Cascade silently heals or drops
such a face, so the intended hole is lost or the face is corrupted with no
diagnostic at all (the 2018 report saw a dropped face; on the current line
the face survives as wrong geometry, still silently).
After the wires are collected, if a face has inner boundaries, measure the
BRepExtrema distance between each inner wire and every earlier wire. Two
non intersecting loops have strictly positive distance, so a distance at
or below the modelling precision means the boundaries touch or cross; emit
a warning (GEO 402) naming the offending face. This is diagnostic only, no
geometry change.
The message is emitted via the kernel logger() rather than Logger::Root():
IfcConvert configures a local Logger and worker logs merge into it, while
Logger::Root() is a separate unconfigured singleton whose messages are
discarded (a latent issue affecting some existing GEO messages too).
Verified on OCC 7.9.2 with synthesized IFC4 faces: an inner triangle
crossing the outer edge, and one straddling the bottom edge, each emit one
GEO 402; a valid 4x4 hole emits none and triangulates identically (area
84.0), in both sequential and multithreaded runs. Pure inner self
intersection and full containment are distinct classes and intentionally
left untouched.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 061bb90d50)
Comply with AGENTS.md: new AI-generated files must carry a top-of-file
comment indicating AI assistance.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit a8d0ef3437)
In IFC2X3 IfcAsymmetricIShapeProfileDef is a subtype of
IfcIShapeProfileDef, so the IfcIShapeProfileDef mapping dispatched it by
inheritance. From IFC4 onwards it is a standalone subtype of
IfcParameterizedProfileDef, so nothing mapped it and the extruded solid
came out empty (GEO326, 0 verts).
Add a dedicated map_impl that builds the twelve-point asymmetric section
(independent bottom/top flange widths, thicknesses, fillet/edge radii and
flange slopes), plus a guarded BIND. Both are wrapped in
SCHEMA_IfcAsymmetricIShapeProfileDef_HAS_BottomFlangeWidth, which is only
defined where the type is standalone, so IFC2X3 keeps its existing
subtype route unchanged.
Verified on OCC 7.9.2: an IFC4 asymmetric extrusion goes from 0 verts to
a correct 72-vert solid (bottom flange wider than top); IFC2X3 output is
unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
(cherry picked from commit 438c0955f2)
This allows a file that will be automatically picked up by Claude. It can either
be a copy of a CLAUDE.md, or a one line file pointing to a shared common file. i.e.
@~/.claude/conventions-ifcopenshell.md
(cherry picked from commit b9deb9c63d)
- autosave.py: black formatting (blank line) and ruff's
collections.abc.Callable import fix.
- project/__init__.py, tool/__init__.py: ruff import-sort fixes. The
autosave import in tool/__init__.py is deliberately kept last (must
come after tool.drawing, per its existing comment) via `# isort: skip`
rather than letting ruff move it, which would reintroduce that bug.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
(cherry picked from commit c0d2c2ea24)
The recovery popup used invoke_popup, which is dismissed the instant
the mouse leaves its bounds - closing the prompt without loading
either file, and with no visible feedback that anything happened.
Switches to invoke_props_dialog, which blocks the rest of the UI and
is only dismissed by an explicit action. Since Blender always renders
both a fixed "Cancel" button and one labelled by confirm_text on that
dialog type, the prompt is reframed as a direct Yes/Cancel question
("Do you want to load the autosaved version instead?") instead of
adding separate Load Original/Load Autosave buttons on top of those.
Folds the load logic directly into the popup's execute()/cancel(), so
the now-redundant LoadAutosavedRecovery operator is removed.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
(cherry picked from commit 0ce6e94352)
Previously the autosaved copy was only ever overwritten, never removed,
so a deliberate quit (whether the user saved or chose "don't save")
still nagged with a recovery prompt on next startup.
Registers an atexit cleanup that removes the active IFC's autosave
file(s) on a graceful interpreter shutdown. atexit never runs on an
actual crash, so a genuine crash still leaves the recovery file in
place as before.
The cleanup reads a cached plain-string path kept up to date by
reset_timer(), rather than looking it up live via bpy.context - by
the time atexit fires, Blender's C++ side is torn down far enough
that even a read-only bpy.context.scene access aborts the process
(std::bad_optional_access) instead of raising a catchable exception.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
(cherry picked from commit be55400ec6)
Implemented as described in #5753, with two options:
- A nag dialog with save or cancel options.
- An autosaved file.
Settings are in preference to activate the feature (default: off), the period before prompting/saving,
and choosing between the two methods.
Prevent the autosave file being added to the recent files list when the user opens the original, but selects to open the autosaved version.
black/ruff
This commit was created using AI assistance. Cursor for the initial code, then Grok and I fixing all the errors
that Cursor made. Finally Copilot did a code review.
I have reviewed and tested the code, and I understand it, and it works and does not introduce any obvious bugs.
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Grok
Co-authored-by: Cursor
(cherry picked from commit 6f1737bb58)
- add_stationing_referent.py: black reformat (new drift from v0.8.0).
- update_fallback_position.py: v0.8.0's changes to this file made the
ifcopenshell.util.unit import (added in an earlier commit here) unused;
removed per ruff.
(cherry picked from commit c4605f2a8f)
poe ty's sequence only reaches ty-ios once ty-bonsai passes, so these
never surfaced until now:
- util/alignment.py: drop the stale `include_referent=False` kwarg from
add_zero_length_segment() - that parameter was removed from the function's
signature in 45ea5eb07 but this caller in a different file was missed,
leaving a latent TypeError if this code path is ever exercised.
- ifcopenshell_wrapper.pyi: add the optional trailing `logger` parameter to
parse_ifcxml/open/construct_iterator*, matching the real SWIG signatures
in src/ifcwrap/*.i (all declare `Logger& logger = Logger::Root()`) that
the hand-maintained stub never picked up.
- ifcopenshell/__init__.py: remove a stale `ty: ignore[unknown-argument]`
comment that ty confirms is no longer suppressing anything.
- assign_cost_item_quantity.py: OPERATORS mixes 2-arg binary operators with
the 1-arg `operator.neg` (for ast.USub), but FormulaEvaluator has no
visit_UnaryOp so USub can never reach this lookup via visit_BinOp.
Suppressed at the call site rather than touching the dict, since this
looks like scaffolding for unary-minus support rather than dead code.
- Explicit submodule imports (ifcopenshell.geom / api.alignment / util.unit
/ api.aggregate / api.context / api.spatial) added where accessed but
only reachable by accident of import order.
(cherry picked from commit d5e890bccd)
- tool.py: drop the `-> int` annotation on the Parametric interface's
get_geom_generation stub; its `pass` body implicitly returns None, which
ty can't reconcile with the runtime @interface/@abstractmethod rewriting
it never sees statically. Matches the file's other stubs (-> None).
- railing.py: qualify the "BIMRailingProperties" string annotations as
"prop.BIMRailingProperties" on the two functions using it, since the bare
name was never imported into this module's namespace.
- product.py: suppress ty's missing-argument errors on
copy_z_rotation_to_selected's Surveyor.get_z_rotation/set_z_rotation
calls with targeted ty: ignore comments. The function is unused and its
two dependencies were never implemented on the concrete Surveyor tool;
left as-is rather than deleted or implemented.
(cherry picked from commit 9f848a73e1)
- gizmos.py: TYPE_CHECKING-guard `import bmesh` for the string-literal
annotation in build_schematic_mesh; suppress the still-unresolved
gizmo_textures import in TexturedQuadGizmoMixin (WIP dependency, not dead
code).
- model/__init__.py: register the `decorator` submodule, which unregister()
already calls (would have raised NameError on addon disable).
- mep.py / tool/model.py: add explicit imports for bonsai.core.geometry and
bonsai.core.model, previously only reachable by accident of import order.
- Test files: add explicit ifcopenshell.api.pset / ifcopenshell.util.element
submodule imports used but not imported.
(cherry picked from commit 4fb8af2278)
The example block was copy pasted verbatim from ExtractPropertiesToSQLite,
so it named the wrong recipe and wrote a .sqlite file. These docstrings are
what ifcpatch surfaces as CLI and UI help, so anyone following the example
for AGS2IFC got a recipe name that does not match the one they selected.
Also state that the input file is not read and that a new IFC4X3 model is
built, since that is not obvious from the signature and the recipe creates
its own project rather than patching the one passed in.
Generated with the assistance of an AI coding tool.
angle_round_threshold was only assigned inside the `distance > 0`
branch of calculate_distance_and_angle, but read unconditionally
whenever should_round is True. When the mouse sample coincides with
the last placed point (distance == 0), such as the first mouse move
after placing a wall's start point on a YZ plane view, this crashed
the modal wall tool.
angle_round_threshold is a fixed cutoff unrelated to whether distance
is currently zero, so it is now assigned once before the branch.
Fixes#8597.
Generated with the assistance of an AI coding tool.
test_returns_none_when_report_file_absent/empty build a MagicMock repo
without configuring index.unmerged_blobs(), so it returned a truthy
MagicMock and git_mergetool's load-bearing "unresolved conflicts remain"
fallback (tool/ifcgit.py:646-647) returned that list instead of None -
failing "assert [] is None". The production fallback is correct and
intentionally left untouched; the tests just misrepresented the
"mergetool resolved cleanly" scenario they are named for. Set
mock_repo.index.unmerged_blobs.return_value = {} in both.
Verified in headless Blender: test/tool/test_ifcgit.py::TestGitMergetool
2 failed / 1 passed -> 3 passed.
This change was made with the assistance of an AI tool.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
DeepDiff's dictionary_item_added/set_item_added results are a
deepdiff.helper.SetOrdered instance, which subclasses orderly_set's
StableSetEq rather than the OrderedSet class json_dump_default checked
for, so the property relationship check always crashed export() with
"Object of type SetOrdered is not JSON serializable". Check against
StableSet, the common base class shared by every orderly_set set
flavour, instead.
Fixes#8905
Generated with the assistance of an AI coding tool.
Browsers were caching /static/js and /static/css for the standalone
webui (costing, gantt, drawings, index, demo pages) indefinitely, so a
shipped JS fix (e.g. the Download CSV button) would only reach a user
after a manual hard refresh.
Two changes, applied consistently across all five webui pages.
1. Every locally served link/script tag in the pystache templates now
carries a ?v=<bonsai version> query string, falling back to a static
asset mtime hash when BONSAI_VERSION isn't set (e.g. running
sioserver.py standalone). Since get_bonsai_version() includes the
build's commit hash, the token changes on every shipped update.
2. Responses under /static/ and /jsgantt/ now carry
Cache-Control: no-cache, must-revalidate. This covers what query
stamping alone can't reach: cost.js and gantt.js statically import
utilities/costui.js by a fixed relative path with no query string, so
that nested module still needed server side revalidation to pick up
changes.
Verified against a live aiohttp instance of sioserver.py: rendered
HTML for all five routes shows the stamped URLs, and the token
changes when BONSAI_VERSION changes between two server runs. A
conditional GET against a static file with a stale If-Modified-Since
header confirms the cheap 304 revalidation path still works.
Also used this instance plus a real headless Chromium (Playwright) to
click test the previously untested Download CSV button on the costing
page. The ribbon renders it correctly, and clicking it (with a
synthetic cost-items table injected into the DOM to stand in for a
connected Blender's data) triggers a real Blob download with the
correct filename and CSV content. No bug found, the button works as
intended.
AI-generated with Claude Code.
Stefano's final ask on #6251 was specific: the ODS/XLSX export should
show exactly what the cost panel shows, ID (Identification), Name,
Quantity, Value, Total Cost, no more, no less. The previous fix in
this PR removed the internal bookkeeping columns but still exported
Description, Unit and a per-category cost breakdown (Labor Cost,
Material Cost, etc), none of which appear in the panel.
Presentation formats (.ods/.xlsx) now use an explicit allow-list of
columns instead of a block-list of internal ones, and relabel headers
to match the panel's own wording (ID / Value / Total Cost). The .csv
format is unchanged: csv2ifc still reads back the extra bookkeeping
columns for the import round trip, which is why it keeps them.
Also add a "Download CSV" button to the browser costing view
(Generate spreadsheet browser), which previously only offered a
clipboard-based Copy Selected. It reuses the already-rendered table
(respecting the user's column visibility settings) and triggers a
real file download, dropping only the UI-only Actions column.
AI-generated with Claude Code; reviewed and tested by Petru Conduraru.
Three defects reported against the Costing tab export:
1. XLSX export crashed with ModuleNotFoundError: xlsxwriter was never
bundled with Bonsai. Port the writer to openpyxl, which ifccsv
already uses and Bonsai already ships, so it works out of the box.
2. Every ODS cell was written as a string (numbers as text), and the
formula branch was dead code: it compared against 'Total Price' /
'Rate Subtotal' while the headers are 'TotalPrice' / 'RateSubtotal'.
Numeric columns are now typed float cells and TotalPrice becomes a
real formula: Quantity*RateSubtotal on leaf items, SUM over the
direct children's TotalPrice cells on sum items.
3. Internal bookkeeping columns (Id, ItemIsASum, Hierarchy, Index,
Quantities) leaked into the presentation formats. ODS/XLSX now hide
them; CSV keeps them since csv2ifc consumes them for the round trip.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
get_element_value could not reach the members of an IfcPhysicalComplexQuantity
(or IfcComplexProperty) by their natural path. util.element expands a complex
quantity into a dict whose nested members live under a "properties" sub-dict,
but the selector's dict navigation only looked at the top level, so
"Qto_Custom.Layer1.Width" returned None and IfcCsv exported nothing for it.
Only the internal "Qto_Custom.Layer1.properties.Width" path worked.
When a key is not a direct member of the value dict, descend into its
"properties" sub-dict so nested quantities/properties resolve with the
natural "Set.Complex.Nested" path. Direct keys still take priority, so the
explicit ".properties." path stays backward compatible and the regex branch
is untouched.
Verified: Qto_Custom.Layer1.Width -> 0.1 and Layer1.Height -> 2.5 (were
None), the sibling simple NetArea still resolves, the legacy .properties.
path still works, and IfcCsv now exports the nested value. test_selector.py:
38 passed (adds test_selecting_a_nested_complex_quantity).
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* Fix ci-bonsai-daily: ProjectLibraryData duplicate parent-library enum
parent_libraries_enum() adds an explicit entry for get_root_context(),
then loops over cls.data["project_libraries"] (all IfcProjectLibrary
entities) and appends each. For a library-only file (no IfcProject),
get_root_context falls back to the top-level IfcProjectLibrary itself,
so the root is appended twice with the same enum key (its STEP id),
which Blender EnumProperty requires to be unique -> the data load
asserts. Normal project files are unaffected (root is an IfcProject
whose id never collides with a library id).
Skip library_id == root.id() in the loop (dedup by id, the colliding
key). Verified in headless Blender:
test_project_library_data.py::TestLibraryOnlyFile goes from 1 failed /
5 passed to 6 passed.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Bonsai: repair library files missing the required IfcProject, not just the symptom
Per the IFC Project Context concept template, every project data set (library
files included) shall contain exactly one IfcProject, and IfcProjectLibrary
instances are assigned to it via IfcRelDeclares. There is no such thing as a
spec-valid file rooted on IfcProjectLibrary alone.
get_root_context() (added in 260a387069, #8184) treated a missing IfcProject
as license to use the top-level IfcProjectLibrary as the file's root context
instead. That invalid premise is why project_libraries() (which walks every
IfcProjectLibrary, root included) then re-added that same entity, producing
the duplicate, colliding enum key this PR originally papered over with a
dedup guard.
Add tool.Project.ensure_project_context(), which repairs a file missing
IfcProject by creating one and declaring the file's root-level
IfcProjectLibrary instances to it, and tool.Project.open_library_file(),
which opens a library file through that repair. Route all three
IfcStore.library_file load sites in SelectLibraryFile through it. Downstream
code (get_root_context, ProjectLibraryData, RefreshLibrary,
AddProjectLibrary) now always operates on a spec-valid model, so the
duplicate enum entry cannot occur; the previous one-line dedup guard in
parent_libraries_enum() is kept only as cheap defense in depth for callers
that bypass the load-time repair, not as the fix.
Rework test_project_library_data.py: the previous _make_library_only_file()
fixture built an invalid library-only model and asserted that as correct
behaviour. Replace it with a spec-valid fixture (IfcProject + IfcProjectLibrary
declared to it) for the downstream tests, and a malformed fixture used only to
exercise the new repair path.
Verified live in headless Blender (isolated profile): reproduced the original
duplicate-enum-key failure mode, then confirmed ensure_project_context/
open_library_file repair a malformed file and ProjectLibraryData,
refresh_library and add_project_library all operate correctly on the result,
with no duplicate keys and no regression on already-valid files or IFC2X3.
This change was made with the assistance of an AI tool.
* Bonsai: stop supporting library-only files, do not repair them
Per Moult's feedback: if the IFC is invalid, our default position is to not
support it, not to patch around it. A library file with no IfcProject is
invalid IFC (Project Context concept template requires exactly one
IfcProject), and it is not ubiquitous: every library file bonsai ships under
bim/data/libraries has an IfcProject with the IfcProjectLibrary declared to
it via IfcRelDeclares. The single #8183 report is an outlier, not a common
authoring pattern worth accommodating.
Remove tool.Project.ensure_project_context() and open_library_file() (the
load-time repair added in the previous commit here) and revert
SelectLibraryFile's three load sites to plain ifcopenshell.open. Simplify
get_root_context() back to returning ifc_file.by_type("IfcProject")[0]
directly, no IfcProjectLibrary fallback: a file without IfcProject now raises
IndexError instead of being silently treated as valid. AddProjectLibrary's
nest-under-library branch is now dead code (root_context is always an
IfcProject) and is removed. The one-line enum dedup guard from the original
commit here is also removed: since get_root_context can only return an
IfcProject or raise, an IfcProject id can never collide with a library id, so
the guard has nothing left to guard against.
Rework test_project_library_data.py: drop the invalid _make_library_only_file
fixture and its tests, which asserted an unsupported model as correct
behaviour. Replace with a single spec-valid fixture matching bonsai's own
shipped library files (IfcProject + IfcProjectLibrary declared to it), used
for the ci-bonsai-daily regression test and the refresh/add-library
operators, plus one explicit test that get_root_context raises for a file
without IfcProject, documenting that this input is intentionally
unsupported rather than silently tolerated.
Verified live in headless Blender (isolated profile, source-loaded, never
the real profile): confirmed the removed methods are gone, that a
library-only file now raises instead of being handled, that
ProjectLibraryData/refresh_library/add_project_library all work correctly
on a spec-valid model with unique enum keys, and spot-checked that every
library file under bim/data/libraries already has an IfcProject.
This change was made with the assistance of an AI tool.
* Bonsai: inline get_root_context, trim docstrings, confirm get_parent_library unchanged
Per Moult's round 3 review. get_root_context added nothing over
ifc_file.by_type("IfcProject")[0], which is guaranteed by the IFC Project
Context concept template; remove it and inline the call at its three sites
(operator.py's RefreshLibrary and AddProjectLibrary, data.py's
parent_libraries_enum). Trim the get_parent_library docstring to one line;
its logic is untouched by this PR, byte for byte identical to origin/v0.8.0,
and still returns None only when project_library has neither Nests nor
HasContext, never for a library declared directly to IfcProject.
Rework test_project_library_data.py to match: replace the two
get_root_context-specific tests with one that exercises the real call site
(ProjectLibraryData.parent_libraries_enum raising IndexError for a file
without IfcProject), and add an explicit test that get_parent_library
returns None for a genuinely orphaned library. Also drop a long inline
comment that restated what the test body already shows.
Verified live in headless Blender (isolated profile, source-loaded, never
the real profile): all 17 test/bim/module/project tests pass, including the
new get_parent_library None-for-orphan case. Ran the full test/bim suite
before and after on the identical harness: 82 failed/1335 passed both times,
same failing tests (all pre-existing, unrelated to this module).
This change was made with the assistance of an AI tool.
* Bonsai: fix EditProjectLibrary leaving stale declarations after reparenting
Per Moult's round 4 review. The assertion change (get_parent_library(root)
now returns the IfcProject instead of None) is correct: in the old
library-only test model a top-level library had neither IfcRelNests nor
IfcRelDeclares, so None meant "top level". In the new spec-valid model a
top-level library is always declared to the guaranteed IfcProject via
IfcRelDeclares, so get_parent_library correctly resolves it through the
HasContext branch instead of falling through to None. get_project_hierarchy
already keys top-level libraries under the project for exactly this reason,
so the library tree still renders correctly.
Auditing every caller found one real bug in EditProjectLibrary, which
Gorgious56 originally wrote for the library-only model. Its move-library
logic assumed a top-level library (previous_parent_library is None) needed
no cleanup before nesting it under a new parent, and that unnesting a
library back to the project needed no new relationship because it was
"already assigned by default". Both assumptions relied on a top-level
library never actually holding a IfcRelDeclares, which is no longer true.
Reproduced live: moving a project-declared library under another library
left its old IfcRelDeclares dangling alongside the new IfcRelNests (an
invalid double parentage), and moving a nested library back to the project
left it with neither relationship, orphaning it out of the tree entirely.
Fixed by tearing down whichever of IfcRelDeclares/IfcRelNests the library
previously had before establishing whichever one the new parent requires,
instead of assuming which prior state applies.
Added tests: get_parent_library resolving a nested sub-library to its
library parent (the third contract case alongside project-declared and
orphaned), and both EditProjectLibrary reparenting directions, which fail
without the operator.py fix and pass with it.
Verified live in headless Blender (isolated profile, source-loaded, never
the real profile): all 20 test/bim/module/project tests pass. Ran the full
test/bim suite before and after on the identical harness: 123 failed/1294
passed before, 123 failed/1297 passed after, identical failing test names
in both runs (diffed), the extra 3 passes are the new tests above.
This change was made with the assistance of an AI tool.
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
SidecarBuilder::build (the one-shot bake used by the models-panel export
command in bonsaiviewer) never called reorderSidecarByMorton, unlike the
live streaming loader. The chunk table was therefore left empty, so the
exported .ifcview had its geometry laid out non-contiguously and only the
metadata blocks compressed. Reorder before writeSidecar to match the
loader so exported sidecars stream correctly.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
on_depsgraph_update and on_depsgraph_update_caps are registered together
as persistent depsgraph handlers (bim/module/clip_box/__init__.py:50-52).
on_depsgraph_update guards with `if cls._file_loading: return`, but the
sibling on_depsgraph_update_caps did not, so a depsgraph tick during the
file-load window still ran it. Beyond the failing test, this can re-arm a
cap-rebuild bpy.app.timers callback in the exact load window _on_load_pre
cancels timers for, against regions whose GPU state is not yet wired.
Add the same _file_loading guard as the first check.
Verified in headless Blender:
test_clip_box.py::TestRefreshTimerLifecycle::test_depsgraph_update_no_op_while_loading
1 failed -> passed.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Link IFC with 'Use Cache' unchecked crashed with FileNotFoundError
when no .ifc.cache.blend existed yet (a fresh link). Regression from
35e3d9c42, which refactored the cache-clear guard from
'if not self.use_cache and blend_filepath.exists()' into
should_clear_cache() but dropped the existence check on the
not-use_cache path, so os.remove() ran on a non-existent file.
Check blend_filepath.exists() first in should_clear_cache() so the
remove is never attempted when there is nothing to clear, while
keeping the query-mismatch cache invalidation intact.
Fixes#8350
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
driveStreamingLoads could deadlock: a chunk waiting on asynchronous
pool growth parks in a frame-counted backoff cooldown, but once the
render loop quiesced after the settle burst the frame index froze, so
the cooldown never expired and streaming stalled part-loaded until the
user moved the camera. Keep the loop alive while growth may still land
(growth_pending() || can_grow()), exposed via a new BufferPool accessor.
loadSidecarMetadataWeb put the model in the scene before reading the
element-metadata block header, leaving a window where the locator was
still zero. A getObjects() landing in that window could not distinguish
"locator not read yet" from "sidecar has no element block" and latched
the model as permanently empty. Read the 16-byte header first, then
apply; carry the locator through applyCachedModel so it is set before
any web element-metadata fetch can run.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
When replacing a style on an item whose previous IfcStyledItem wraps its styles
in the deprecated IfcPresentationStyleAssignment, and the assignment is not
being reused (use_style_assignment is False, e.g. an IFC4 file authored by
AVEVA E3D), the else branch called remove_same_type_styles(style_assignment)
with style_assignment still None, raising
AttributeError: 'NoneType' object has no attribute 'Styles'. Operate on style_,
the assignment found in the current iteration, instead of the accumulator.
Verified red-green with a minimal IFC4 file using IfcPresentationStyleAssignment.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
compare() recursed into list values passing the negated comparison through,
so != meant "at least one item differs" and both = and != matched the same
elements on any multi-valued property (e.g. an enumerated property with two
values selected). Strip the negation for the per-item comparison and negate
the aggregate instead, so != means "no item equals" and stays the complement
of =. The same applies to !*=.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Right-clicking a model in the models panel now offers "View Selected Model",
which frames the camera on just that model's geometry — View All, scoped to
one model. With several models selected the action reads "View Selected
Models" and frames their union, matching how the panel's existing Move to
Group already treats a multi-selection.
The AABB fold behind viewAll moves into InstanceCompose, which exists so this
kind of logic is unit-testable without a Qt window or a wgpu device (populating
ViewportCore's model map needs a real GPU, so the fold was previously
untestable in place). It splits in two:
- sceneWorldAabb — every VISIBLE model, what viewAll frames.
- modelsWorldAabb — only the named models, hidden or not. A model the caller
named explicitly is framed even if hidden; second-guessing
that is worse than honouring it. Models with no loaded
geometry contribute nothing, and if none of them do the
camera is left alone rather than flying to the origin.
Both are covered by six new cases in test_instance_compose (131 total).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Give host pages a real API over the web viewer, not just "embed it and listen
for picks": read/set the camera, read/set multi-selection, enumerate every
object with its IFC identity, drive per-object visibility, and override
object colours.
The wasm boundary keeps to object_ids (u32 arrays marshalled through the heap,
with an "ask twice" convention on the getters); web/ifcviewer.js layers IFC
GlobalId resolution on top, from the element table getObjects() fetches. Every
id-taking call accepts an objectId, a GlobalId, or an element object.
Colour override needed no new mechanism: color_override_rgba8 was already
plumbed through the sidecar, the instance SSBO, the WGSL shader and the
opaque/transparent cull classifier, but nothing ever wrote a non-zero value
into it. setObjectsColor is the missing writer, which is why an alpha below 255
correctly reclassifies the instance into the transparent pass.
Two bugs surfaced while wiring this up:
- wgpu_initialized_ was only ever set by the Qt desktop host, so on web every
upload guarded on it was a silent no-op — including the pre-existing
recomposeAndUploadModel that federation transforms depend on. The core now
latches it in its own web init.
- The demo pages were copied into the build dir by a POST_BUILD command on the
wasm target, so they only refreshed when the wasm itself relinked; editing a
page left a stale copy that the dev server (and the Playwright suite) kept
serving. Each page now has its own copy rule with a real dependency, and
sample.ifcview is a LINK_DEPENDS so regenerating it forces a relink.
applyCachedModel also now keeps the element metadata it already parses on the
path-based load (it was being dropped), so the embedded sample has GUIDs and
the demo works with no file to pick.
The sample model was three coincident cubes, which made per-object hide and
colour look like no-ops — whatever you hid was still drawn by the box behind
it. make_sample.py regenerates it as a slab, a wall and a beam in distinct
places, so the fixture is reproducible rather than an opaque blob.
Demoed by web/scripting.html (linked from the index; viewer is on
window.viewer) and covered by tests/scripting.spec.mjs — 6 cases against a real
GPU, asserting visibility and colour at the pixels, not just at the API.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The Optimise recipe imported `toposort`, a third-party PyPI package that
is not bundled with Bonsai, so running the recipe there raised
`ModuleNotFoundError: No module named 'toposort'`.
Replace it with the standard library `graphlib.TopologicalSorter`
(available since Python 3.9), which provides the same dependencies-first
ordering guarantee the recipe relies on: forward-referenced instances are
mapped before the instances that reference them. The dependency-graph
dict format ({node: {predecessors}}) is identical between the two, so the
graph construction is unchanged. Drop `toposort` from ifcpatch's
dependencies since it is no longer used.
Verified with toposort NOT installed: the Optimise recipe now runs and
deduplicates correctly (IfcParseExamples_test.ifc 88 -> 63 instances, all
6 products preserved, output reopens cleanly).
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
tool.Bsdd.identifier_url() (pset/ui.py pset name check in the Property
Sets panel) reads client.baseurl unconditionally, but the test stub
never had that attribute, so any scenario that opens the Property Sets
panel dies with AttributeError under the stub. The boolean.feature
scenarios only surfaced this once their STEP id failures were fixed,
the id failure had been masking it. Mirror the real bsdd.Client
default so identifier_url() resolves to the standard identifier URL.
This change was made with the assistance of an AI tool.
The two boolean.feature scenarios pinned representation item objects by
absolute STEP id (Item/IfcHalfSpaceSolid/90, the BBIM_Boolean pset text
[91]). Those ids shift every time any earlier entity allocation in an
empty project changes (latest instance: #8577 moved 90 to 86), so this
cluster re-breaks on unrelated commits.
Make the object-name and panel-text BDD steps run their argument through
replace_variables, the same substitution 'the variable' and the
connection steps already use, and have boolean.feature capture the real
ids from the IFC file (by_type(...)[0].id()) into variables at the point
the entities are created. The steps stay strict: the substituted name
must still resolve to exactly the named object, there is no wildcard
matching. Substitution is a no-op for every existing feature string
without a {variable} placeholder.
This change was made with the assistance of an AI tool.
Two independent test-harness/fixture defects in test/bim/test_feature.py:
- OperatorSpy had no bl_rna, so any BDD step that redraws a panel calling
helper.draw_filter() (which tests "module" in op.bl_rna.properties)
crashed with AttributeError. Give OperatorSpy a bl_rna property that
forwards to the real registered operator class
(bpy.types[bl_idname].bl_rna), matching live UILayout.operator()
semantics. Fixes test_select_all_walls and test_edit_filter_query.
- The shared "I create default MEP types" step looked up
bpy.data.objects["IfcDistributionPort/Port"], but port creation never
sets port.Name, so tool.Loader.get_name deterministically names the
object "IfcDistributionPort/Unnamed". Update the literal. Fixes the MEP
scenarios (connect/transition/bend) that share this setup.
Verified in headless Blender: OperatorSpy scenarios 2 passed (were
AttributeError); MEP test_connect_mep_elements* go from
KeyError 'IfcDistributionPort/Port' to passing.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Pset names containing spaces (e.g. "SOLIDWORKS Custom Properties") were
not quoted when building selector keys in SelectSimilarData, causing
get_element_value to fail when the operator ran. Now wraps pset names
and property names in double quotes if they contain spaces, consistent
with the selector syntax used elsewhere.
Generated with the assistance of an AI coding tool.
Per aothms's review comment: this file's schema was already changed
independently on v0.8.0 since this branch was created, so this PR's own
edit conflicts with it. Reverting to the current upstream version of the
fixture; the bsdd.py rate-limiting fix is untouched.
bsdd.py: the Client made every request with a bare requests.get, so a single
429 from the (unauthenticated, aggressively rate limited) bSDD API failed the
whole test. Route requests through a Session with a mounted urllib3 Retry
(5 attempts, backoff, honouring Retry-After) for 429/5xx, matching how a
resilient API client should behave, not just papering over the test.
ColumnPSetsOfSets.ifc: FILE_SCHEMA was accidentally changed from IFC4X3_ADD2
to IFC2X3 in a7738eeb64 (an unrelated logger refactor), a one line collateral
edit to this fixture. The file's DATA section still uses IFCPROPERTYSETDEFINITIONSET,
an IFC4+ only type. Parsing it against IFC2X3 threw "Entity ... not found in
schema", which silently fell back to interpreting the value as a raw nested
aggregate instead of the intended defined-type wrapper, producing the
double-nested tuple that broke test_stream, test_file and test_rocks in
test_streaming_rocksdb_and_simpletyperefs.py. Restoring the original schema
declared when the fixture was added (ff3fa48332) fixes all three.
Generated with the assistance of an AI coding tool.
Bsdd.get_dictionaries() unconditionally did cls.client = bsdd.Client(),
replacing whatever client was already set - including the
bSDDClientStub the BDD suite injects at module load
(test_feature.py: tool.Bsdd.client = bSDDClientStub()) to avoid live
network calls. Because "Load bSDD Dictionaries" is the first step of
every bsdd.feature scenario, the stub was discarded before its fixture
data ("LCA", "BonsaiTestDict") could ever be returned.
The re-init is unnecessary: bsdd.Client.__init__ only sets baseurl and
blank tokens, and the next line already updates baseurl defensively via
hasattr. Drop the clobbering assignment; reuse whichever client is
already set.
Verified in headless Blender: bsdd scenarios (load dictionaries, search
all/single dictionary) go from 3 failed ("Could not see LCA/
BonsaiTestDict") to 3 passed.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The choco dir was renamed from choco/blenderbim to choco/bonsai back in
2024 (Rename choco dir), but choco_release.py's BLENDERBIM_DIR constant
was never updated, so the daily choco release job crashes immediately
with FileNotFoundError trying to os.chdir into the now nonexistent
choco/blenderbim directory.
Release tags also moved from a bare blenderbim-YYMMDD scheme to
bonsai-X.Y.Z-alphaYYMMDDHHMM, so the tag-prefix strip used to build the
package version still looked for the old "blenderbim-" prefix and left
it untouched, embedding the raw tag (including the already-present
"-alpha" segment) into the nuspec version field, which the template
then doubled up with its own "-alpha" suffix, producing an invalid
NuGet version string. Both are fixed together since the second bug
would otherwise surface as soon as the first one is unblocked.
The pre-commit black hook also reformatted pre-existing whitespace
drift in choco_release.py (this file sits outside CI's lint scope, so
it had never been auto-formatted before); that reformatting is
incidental to satisfying the local hook, not part of the fix itself.
Generated with the assistance of an AI coding tool.
replace_attribute() rewrites references inside aggregate attributes via
element.walk(), but never checked whether the replacement value was
already present elsewhere in the same aggregate. For an EXPRESS SET
(e.g. IfcProject.RepresentationContexts, IfcRelAggregates.RelatedObjects)
this can leave the same reference listed twice, which is invalid IFC.
LIST and BAG aggregates legitimately allow duplicates, so a blanket dedup
would be wrong; only SET-typed attributes are deduplicated, determined at
runtime from the schema declaration (IfcOpenShell#8706 review comment).
The SET/LIST/BAG check is cached per (schema, class, attribute index), and
the dedup pass itself only runs when a cheap linear pre-check finds the
replacement value already present in the aggregate, so the common case
(no duplicate produced) pays only that pre-check, not a hash-set rebuild.
Benchmarked against a 23MB (431k entities) and a 104MB (2.4M entities) IFC
model against a large SET attribute: worst case adds well under 1ms per
call; the realistic case (merging duplicate contexts, matching the PR
#8706 scenario) shows no measurable regression.
Fixes the root cause flagged in IfcOpenShell#8706 (Moult), obviating the
need for MergeDuplicateContexts' own manual aggregate-dedup pass for that
scenario.
Generated with the assistance of an AI coding tool.
Fixes issue #3910's documentation gap. IsDefinedBy() returns
IfcRelDefinesByProperties relationship objects, not the property set
itself, and RelatingPropertyDefinition() must be used to reach the
IfcPropertySet or IfcElementQuantity. Properties can also come from an
element's type via IsTypedBy() -> RelatingType() -> HasPropertySets(),
a path that is easy to miss because it works differently. Adds a
worked, beginner-commented, compilable example covering both paths.
Generated with the assistance of an AI coding tool.
The ifcopenshell take-off engine left every Qto_EarthworksFillBaseQuantities
value null, so Bonsai added the qset with no numbers on IFC4X3 models. Map the
geometrically derivable quantities using the slab axis convention: Length on
local X, Width on local Y, Depth on local Z, and the net solid volume as the
compacted (Fill) or undisturbed (Cut) volume. LooseVolume and Weight stay
unmapped because they need soil bulking and density factors absent from
geometry. Bring IfcEarthworksCut to parity with the Blender engine and wire
IfcReinforcedSoil on both engines.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
assign_cost_item_quantity skipped every IfcSpatialElement, which also
swallowed IfcSpace. Spaces are legitimate quantifiable objects, so their
Qto_SpaceBaseQuantities (for example GrossFloorArea) were never picked up
and count based cost items fell back to 0. Keep skipping spatial
containers (site, building, storey) but allow IfcSpace.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The headless "IfcOpenShell" calculator had all Qto_SpaceBaseQuantities
formulas set to null for IfcSpace in both IFC4QtoBaseQuantities.json and
IFC4X3QtoBaseQuantities.json, so qto.py's `if not formula: continue`
skipped every quantity, no geometry task was queued, and spaces never
appeared in results (elements_quantified: 0). The Blender calculator
already computes these; they were just never ported to the
ifcopenshell.util.shape-backed calculator.
Map the eight computable quantities to existing util.shape functions,
mirroring the Blender calculator semantics (no new shape.py code):
GrossFloorArea=gross_get_footprint_area, NetFloorArea=net_get_footprint_area,
GrossCeilingArea=gross_get_top_area, NetCeilingArea=net_get_top_area,
GrossPerimeter=gross_get_footprint_perimeter, GrossVolume=gross_get_volume,
NetVolume=net_get_volume, Height=net_get_z.
Left null (matching the Blender ruleset, not guessed): GrossWallArea,
NetWallArea, NetPerimeter (Blender stub), and FinishFloor/CeilingHeight
(Blender derives these from sibling IfcCovering decomposition geometry,
which this per-element calculator architecture can't reach).
Verified on IFC4 (4x3 space extruded 2.5m): before -> {} / elements_quantified 0;
after -> GrossFloorArea 12, GrossPerimeter 14, Height 2.5, GrossVolume 30,
etc. - all exact matches to the extrusion. IFC4X3 formulas are identical
and the formula->function resolution is schema-agnostic.
Scope: fixes the IfcSpace case (the issue title). The 12 other all-null
classes noted in the issue (IfcDoor, IfcSite, IfcRailing, ...) are left as
follow-up.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The ifc5d IfcOpenShell (geometry-based) Qto engine computed
Qto_CoveringBaseQuantities using axis-agnostic heuristics:
- GrossArea/NetArea: gross_get_max_side_area / net_get_max_side_area,
the largest of the X/Y/Z projected side areas.
- Width: gross_get_min_xyz, the smallest of the X/Y/Z dimensions.
The Blender Qto engine instead already used
EPset_Parametric.LayerSetDirection (AXIS2 for wall-like coverings,
AXIS3 for floor/ceiling-like coverings) to pick the correct axis via
get_covering_gross_area/get_covering_net_area/get_covering_width in
bonsai/bim/module/qto/calculator.py.
For any covering whose length isn't the largest dimension (e.g. a
short wall-covering strip, or a small covering patch), the two
engines' heuristics can pick different faces/axes entirely, giving
different Width/Area values for the same element - this is what was
reported in #6728.
Fix: give the IfcOpenShell engine the same layer-set-direction
awareness. Added IfcOpenShell.get_covering_parametric_axis/
get_covering_area/get_covering_width (dispatched as internal
functions, like the existing get_weight/get_segment_length), and
wired gross_get_covering_area/net_get_covering_area/
gross_get_covering_width into the IfcCovering rules in
IFC4QtoBaseQuantities.json and IFC4X3QtoBaseQuantities.json.
The AXIS2 area/width formulas (get_side_area, net_get_y) intentionally
match the simpler formulas already used for Qto_WallBaseQuantities in
this same rule set (net_get_side_area/net_get_y), rather than
replicating the Blender engine's more elaborate get_lateral_area/
get_width (min(X,Y)) helpers, consistent with how the two engines
already diverge for regular walls without being considered a bug.
Verified with a standalone script driving ifc5d.qto.IfcOpenShell
directly against synthetic AXIS2/AXIS3 IfcCovering geometry: for
typical proportions old and new formulas agree, and for
disproportionate coverings (thin dimension not the smallest/largest)
the old formulas picked the wrong axis while the new ones correctly
track the covering's LayerSetDirection, matching the Blender engine.
Did not verify through the full Blender/Bonsai UI, as it would have
required registering the addon in the machine's shared Blender
profile, which is unsafe while other agents may have it loaded.
Generated with the assistance of an AI coding tool.
copy_cost_item appends the copy to the inverse relationships of the
original cost item, which for a root cost item includes the source
schedule's IfcRelAssignsToControl. copy_cost_schedule then assigned that
same cost item to the new schedule as well, so the copies showed up in
both schedules and deleting them from one removed them from the other.
Unassign the copy from the source schedule before assigning it to the
new one.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Fixes#4443. The Wall/Slab/other authoring tools (BimTool subclasses)
already bind Shift+Q to bim.perform_quantity_take_off via hotkey_S_Q,
but the Spatial tool has its own separate keymap/operator
(bim.spatial_hotkey) that never registered a Q entry, forcing users to
switch tools just to (re)calculate quantities for a selected element.
Added the same Shift+Q keymap entry and a matching hotkey_S_Q handler
to the Spatial tool, mirroring BimTool's existing behavior exactly
(including the same selected-objects guard).
The other part of the request, a bulk "calculate all quantities"
entry point, already exists today: bim.perform_quantity_take_off
computes quantities for every IfcElement when no objects are
selected, exposed via the Scene > Quantity Take-off panel regardless
of which workspace tool is active, so no change was needed there.
AI-generated, reviewed and tested by BIMvoice.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
can_nest() only permitted IfcElement-to-IfcElement pairs, so nesting
two IfcElementType objects (e.g. an IfcElementAssemblyType nesting a
component IfcDoorType) was silently rejected. IfcRelNests.RelatingObject/
RelatedObjects are typed as the general IfcObjectDefinition in the
schema, so type-to-type nesting is schema legal, IfcOpenShell's core
nest.assign_object API already handles it generically, and the Nest
UI panel is driven purely by ifcopenshell.util.element.get_nest/
get_components (IFC data queries, not Blender collection structure),
so once the relationship exists it displays correctly with no other
changes needed.
Extended is_compatible_class to also accept a same-kind IfcTypeProduct
pair. Mixing an occurrence element with a type is intentionally still
rejected, that isn't a real modeling pattern.
Verified live in headless Blender: type-to-type nesting now creates
a real IfcRelNests and the Nest panel's own data functions reflect
it correctly; mixing an occurrence with a type is still rejected;
existing element-to-element nesting is unaffected.
Generated with the assistance of an AI coding tool.
Root cause: the IfcOpenShell-geometry-engine calculator ruleset
(IFC4QtoBaseQuantities.json and IFC4X3QtoBaseQuantities.json) left
IfcWall's GrossFootprintArea/NetFootprintArea mapped to null, so
these two quantities were silently omitted from Qto_WallBaseQuantities
whenever that ruleset was used. The generic gross_get_footprint_area
and net_get_footprint_area formulas already exist and are already
wired up for IfcSlab in the same files, so this was a missing mapping,
not a missing implementation.
Fixes#7029.
Generated with the assistance of an AI coding tool.
Error was:
```
configure: error: could not find a working compiler, see config.log for details
```
config.log:
```
conftest.c: In function 'f':
conftest.c:12:48: error: too many arguments to function 'g'; expected 0, have 6
12 | for(i=0;i<1;i++){if(e(got,got,9,d[i].n)==0)h();g(i,d[i].src,d[i].n,got,d[i].want,9);if(d[i].n)h();}}
| ^ ~
```
It was passing `IFC4X3` directly to `schema_by_name` which is expecting
schema identifier (e.g. IFC4X3_ADD2, not IFC4X3 allowed by `IFC_SCHEMA`
- IFC4X3 is one of the IFC4X3 iterations while it was in development,
not the final one).
Noticed by tests failing:
FAILED
test/util/test_schema.py::TestGeometryClassesIntroducedAfter::test_ifc4x3_to_ifc2x3_is_superset_of_ifc4_to_ifc2x3
- RuntimeError: No schema named IFC4X3
FAILED
test/util/test_schema.py::TestGeometryClassesIntroducedAfter::test_ifc4_to_ifc4x3_is_empty
- RuntimeError: No schema named IFC4X3
This reverts commit b61f809731.
This commit was probably using not updated build, currently latest build is e333c1c and can confirm that it has `logger_or_root` added and `delete_same_facet_edge_pairs` removed.
AST parser has changed a bit and there are some minor differences in the
.py output. Updating files just to avoid seeing these diffs when
rerunning rule compiler.
Example error:
```
ast.Str(s=node.attr),
^^^^^^^
AttributeError: module 'ast' has no attribute 'Str'
```
`ast.Str` was deprecated since 3.8 and was removed in 3.14, see
https://docs.python.org/3/whatsnew/3.14.html#id9
Because uv was always trying to install when starting a venv in `ifcopenshell` folder, though they might be already available globally. And also they were listed twice - in pyproject and in the ci-lint.yml, now there's a single source of truth.
Footings are authored two ways with different local axis conventions. Beam-like
footings (STRIP_FOOTING, FOOTING_BEAM) are a profile extruded along local Z, so
Length is local Z and the cross section sits on local X (Width, horizontal) and
local Y (Height, vertical). Slab-like footings (PAD_FOOTING, PILE_CAP) have their
footprint on local X/Y and their thickness (Height) on local Z.
The engine rule set is keyed per IfcFooting and cannot branch on predefined type,
so the previous static rule (Height=net_get_z, Length=net_get_max_xy, Width=null)
swapped Length and Height for beam-like footings and never emitted Width.
Add predefined-type-aware get_footing_length/width/height to the IfcOpenShell and
Blender calculators, and point the IfcFooting rule at them in all four IFC4/IFC4X3
ios/Blender rule files.
Confirmed by authoring footings through the real Bonsai generators and measuring
world-axis orientation: a beam-like footing with a 0.3 wide by 0.6 tall cross
section and 6.0 run reports Length 6.0, Width 0.3, Height 0.6, with the 0.3
physically horizontal and 0.6 physically vertical; a 2.0x1.5x0.3 pad reports
Length 2.0, Width 1.5, Height 0.3.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
See #6835. Qto_OpeningElementBaseQuantities came out axis-scrambled for
openings authored in a Z-up local frame (X along the voided wall, Y
through it, Z vertical), which is how Bonsai authors every wall opening:
- The IfcOpenShell engine mapped Height to the local Y extent and Depth
to the local Z extent, so a 0.9 x 2.0 door opening with Bonsai's
default 1.2m void depth reported Height 1.2 and Depth 2.0, and Area
(max side area) picked the through-wall side, 2.4 instead of 1.8.
This matches the wrong Height=1.2/Area=1.2 screenshots reported for a
1x1 window opening in #6835.
- The Blender engine mapped opening Width to get_length, which returns
the longest bounding box edge, i.e. the opening height for typical
door openings (the same defect 4adaf0d fixed for IfcDoor Width), and
get_opening_depth used min(x, y), which returns the opening width
whenever the width is smaller than the void depth.
The IfcOpenShell engine now has opening-aware internal calculators
(get_opening_width/height/depth/area) that detect horizontal (slab
style) openings with the same heuristic as the Blender calculator, so
slab opening depths keep reporting the slab thickness. The Blender
ruleset uses get_x for opening Width, and get_opening_depth measures the
through-element Y extent for vertical openings.
Door and window quantities themselves are addressed separately: the
Blender engine door Width was fixed in 4adaf0d, and the remaining
door/window defects (door not quantified on the IfcOpenShell engine,
inflated areas) are fixed by the attribute-based calculators in #8389.
Generated with the assistance of an AI coding tool.
falken10vdl reviewed 16b1b4e7b1 on #8843 and pointed out that tagging
every area for redraw was overkill. The actual problem was that the
Object Material panel and the scene Materials list read from plain
python caches (ObjectMaterialData and MaterialsData) that only get
invalidated when the Materials editing UI list is reloaded, which
never happens while you are not in editing mode. The redraw itself was
never the issue, closing the rename dialog already triggers one.
Removed the tag_redraw loop from RenameMaterial and instead call the
existing bonsai.bim.module.material.data.refresh() function from
core.rename_material, unconditionally, through a new tool.Material.refresh()
method. This is the same invalidate-on-next-load mechanism already used
by every other module's Data classes, just wired up for this operator
too, instead of introducing a new one.
Also updates the core tests to prescribe the new unconditional refresh()
call, and adds tool-layer coverage for tool.Material.refresh().
Generated with the assistance of an AI coding tool.
theoryshaw tested #8843 and asked for the new name to show up right
away instead of needing a manual refresh. The Object Material panel
and the scene Materials list both already re-read live IFC data on
their next draw (tool.Ifc.Operator purges those caches after every
IFC-mutating operator), so the button text was correct on the next
redraw. What was missing was the redraw itself: the material name is
a plain button label, not an RNA property Blender tracks, so nothing
told the Properties editor to repaint after the rename dialog closed.
Tag every area for redraw once the rename completes, the same pattern
used elsewhere in Bonsai for popup-triggered edits that need an
immediate repaint.
Also adds core-layer test coverage for rename_material, which had
none.
Generated with the assistance of an AI coding tool.
Adds a "Rename Material" entry to the context menu that already
extends every button in the properties editor (UI_MT_button_context_menu),
triggered when right-clicking a material name button
(bim.select_by_material) that points to a real IfcMaterial. This
gives a quick entry point to renaming from the Object Material panel
without navigating to the scene Materials list.
This follows the pattern that #6680's thread converged on: theoryshaw
requested a right-click entry (rather than a pencil icon or
double-click) that keeps the existing single-click select-by-material
behaviour intact. falken10vdl is the issue's assignee; this is offered
as a starting point for that discussion, not a replacement for it.
Generated with the assistance of an AI coding tool.
Add an "Is Selected" checkbox to each target-view category header in
BIM_UL_drawinglist that toggles selection for all drawings in the
category. The toggle only affects drawings currently visible in the
list (honoring the show_drawings_on_sheets_only filter), and the header
checkbox reflects the aggregate selection state of its drawings.
Also make category headers more obvious: wrap them in a box() for a
distinct inset background and make the header name clickable to
expand/contract the category (same as the disclosure triangle).
Ref: #8825
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Adds a "Show Only Drawings on Sheets" toggle below the drawing list. When
enabled, the list is filtered to drawings referenced by at least one sheet
(target-view headers with no sheeted drawings are hidden too), and
bim.select_all_drawings only acts on the visible/filtered drawings.
A drawing is considered sheeted when its drawing document Location matches a
document reference Location on any SHEET-scoped IfcDocumentInformation.
Filtering is computed live so it reflects sheet edits without reloading.
Closes#8823
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* docs: cover the Blender 5.1 / Python 3.13 transition in installation guides
The system requirements still listed Blender 4.3-4.5 with Python 3.11
only, and nothing documented the pitfall from issue 7623: importing
preferences into a Blender whose Python version changed carries over an
incompatible Bonsai build that silently fails to load. Document the two
Python generations, that Get Extensions picks the matching build
automatically while manual zip installs do not, and the
uninstall-reinstall step that resolves the upgrade case.
Generated with the assistance of an AI coding tool.
* docs: keep it simple, only Blender 5.1 and 5.2 with Python 3.13
Per review, drop the descriptive text and the Python 3.11 line.
Generated with the assistance of an AI coding tool.
* Keep real cause in wrapper ImportError
When the compiled wrapper exists for the current interpreter but fails
to load (for example a glibc version mismatch, as on AWS Lambda in
issue 5927), the bare except rewrote the error into the misleading
"IfcOpenShell not built for '<platform>'" message. Environments such
as AWS Lambda or the Blender add-on dialog only surface the final
exception message, so the actual cause was invisible and undiagnosable.
Keep the "not built for" message only when no matching binary is
present, and otherwise include the original loader error, chaining the
cause in both branches.
This change was AI-generated.
Fixes#5927
* Simplify wrapper import failure to a single message
Per review feedback, drop the filesystem scan and the two message
variants. Always raise the classic "IfcOpenShell not built for
'<platform>'" message with the original exception appended in
parentheses, still chained as the cause. Environments that only show
the final exception message (AWS Lambda, the Blender add-on dialog)
now surface the real loader error, such as the glibc version mismatch
in issue 5927, without any extra logic.
This change was AI-generated.
Per aothms's request on #8605: QtViewer is being superseded by the new
Bonsai Viewer, so its remains are deleted here (src/qtviewer, its
BUILD_QTVIEWER cmake option and add_subdirectory, and its references in
ci.yml's path filter, .gitignore, the conda recipe's license table, and
README's library table).
src/ifcopenshell-python/ifcopenshell/geom/app.py's qtViewer3d is
unrelated (pythonocc-core's own OCC.Display widget class, a name
coincidence) and is untouched.
Generated with the assistance of an AI coding tool.
The jsGantt-improved library that renders Bonsai's Gantt chart already
ships full support for an "Hour" granularity (column width, header
labels in every bundled language, hour-aware rendering math). Bonsai's
config only exposed Day/Week/Month/Quarter, with a comment claiming
Hour caused browser issues even with vUseSingleCell enabled.
Headless Chrome testing against the same library version shows that
claim no longer holds once vUseSingleCell is active (as Bonsai already
configures it at 10000): Hour-format charts render without errors from
typical schedules up through fairly extreme ones (5000 tasks across a
3 year span rendered in about 2.4s). The failure mode the old comment
described only reproduces with vUseSingleCell disabled, which is not
how Bonsai runs it.
Task start/finish times already flow through to the chart unmodified
as raw ISO datetimes (tool/sequence.py create_new_task_json), so any
schedule authored with real hour-level timestamps, for example an
imported MS Project/P6/Excel schedule or one written directly through
ifcopenshell-python, can now be viewed at hour granularity. Verified
live with a night shift schedule crossing midnight, rendered correctly
with no console errors.
Note: Bonsai's own "Edit Task Time" UI currently always snaps
ScheduleStart/ScheduleFinish to 09:00/17:00 regardless of the hour
entered (ifcopenshell/api/sequence/edit_task_time.py), and work
calendars only encode working days, not working hours. So authoring a
genuine hour-precision schedule through that UI is still not possible;
this change only unlocks viewing hour-level data that already exists
in the model. Fixing the editor and calendar model is a separate,
larger design decision for a maintainer.
Addresses #2772.
Generated with the assistance of an AI coding tool.
Point scriptSalome.py at templates/salome/ and the bonded scripts at
_deprecated/, matching the current tree so README links resolve.
Generated with the assistance of an AI coding tool.
* Bonsai: move annotations between drawings when reassigning their group
Assigning an IfcAnnotation to a group that represents another drawing
previously left the annotation in both drawings at once: it stayed in
its old drawing group, its Blender object stayed in the old drawing
collection, and it kept the old camera depth, so the reassignment
appeared to do nothing useful. Issue #2966 documents the seven step
manual workaround users needed instead.
The assign group operator now detects when the target group represents
a drawing (via the new tool.Drawing.get_group_drawing, the inverse of
get_drawing_group), unassigns the annotation from its previous drawing
group, moves its object into the new drawing collection, and places it
on the new drawing camera plane. The target camera is imported on
demand when it has not been loaded yet, matching the pattern used by
the activate drawing operator.
Generated with the assistance of an AI coding tool.
* Bonsai: add one click copy of annotations to another drawing (#2966)
Duplicating an annotation into a different drawing used to require a
seven step manual process: loading groups in scene properties, copying
the object, fixing its group assignment by hand, and repositioning it
onto the target camera plane. A plain Blender duplicate is not enough
because the copy keeps pointing at the same IFC entity, and the Shift D
override, while it does create a genuine new entity through
root.copy_class, leaves the duplicate in the source drawing group,
collection, and camera depth.
The new copy annotation to drawing operator packages the proven recipe
already used by duplicate drawing into one action: duplicate through
tool.Geometry.duplicate_ifc_objects, unassign the copy from the source
drawing group, assign it to the chosen target group, place it on the
target camera plane at the same world XY, and file it into the target
drawing collection. The originals are left untouched and the user's
selection is restored. The target camera is imported on demand when it
has not been loaded yet.
The operator shows a target drawing dropdown and is reachable from the
annotation tool sidebar when an annotation is selected, and from the
drawings panel. Annotations already in the target drawing are skipped
and reported.
The orchestration lives in core.drawing.copy_annotations_to_drawing
with prophecy tests covering the copy, the skip, and the camera import
branches. Verified live in headless Blender 5.1: the copy is a new
IfcAnnotation with its own GlobalId and IfcTextLiteral, both texts are
editable independently, and everything survives save and reload with
each annotation loading in its own drawing.
Generated with the assistance of an AI coding tool.
Fixes#4395.
Root cause: the SVG cut-linework merge step that fuses adjacent
elements' cut polygons together (per the pset-driven JoinCriteria
setting) was hardcoded to only IfcWall and IfcSlab. IfcCovering cut
shapes were skipped unconditionally, so adjacent coverings never
joined, leaving a visible seam/broken corner in section drawings
regardless of JoinCriteria.
Fix: added an EPset_Drawing.JoinClasses property, following the
exact same user-overridable pattern already used by
EPset_Drawing.BringToFront - a comma-separated list of IFC classes
to join, defaulting to "IfcWall,IfcSlab" (unchanged behavior) when
unset. Users can override per-drawing to add IfcCovering (or any
other class) when they want it joined too. Kept this opt-in rather
than hardcoding IfcCovering into the default list, since joining a
thin finish layer the same way as a thick wall/slab could produce
unwanted mitring in some cases - the user decides per drawing.
Verified live against the reporter's own attached file
(ifcovering joining.ifc) and its cached section linework: with
JoinClasses unset, two separate closed paths reproduce the reported
seam exactly. With JoinClasses = "IfcWall,IfcSlab,IfcCovering", the
two coverings merge into a single closed polygon with the internal
seam removed. Confirmed IfcSlab join behavior is unchanged in both
runs.
Generated with the assistance of an AI coding tool.
Co-authored-by: Dion Moult <dionmoult@gmail.com>
reimport_element_representations() built a fresh
ifcopenshell.geom.settings() without copying deflection_tolerance /
angular_tolerance from the IfcImportSettings it had just
constructed, and never passed geometry_library to either the
iterator() or create_shape() calls it makes. As a result, exiting
Item/edit mode (which reaches this function via
switch_representation) silently fell back to IfcOpenShell's
hard-coded mesher defaults (0.001 linear deflection, ~50x finer than
the project's default of 0.05) and the default geometry kernel,
instead of the project's configured tolerance and Geometry Library.
This made geometry visibly change quality after a no-op Tab into and
back out of edit mode, since the reload path was unintentionally far
more precise (and used a different kernel) than the initial import.
Both settings, and geometry_library, are now taken from the
IfcImportSettings instance already built at the top of the function,
so a reload matches the original import.
Refs #5685.
Generated with the assistance of an AI coding tool.
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
theoryshaw's follow-up on #6944: after the profile/curve reconstruction fix
(previous commit), the arc/circle marker for a freshly Shift+D-duplicated
loop wouldn't appear until leaving and re-entering Edit Mode.
Root cause: ProfileDecorator groups arc/circle vertices purely by
IFCARCINDEX/IFCCIRCLE vertex-group index every draw call (it has no cache
to go stale, it fully recomputes from the live edit-mesh bmesh each frame).
Duplicating a loop copies its vertex-group weights onto the new geometry,
since Blender allocates no new group for a duplicate, so the source loop
and its live duplicate land in the same dict entry. That entry then fails
the "exactly 2 verts per circle / 3 per arc" check and is skipped entirely,
so BOTH the original and the duplicate stop being drawn until the mesh is
reimported and gets fresh, distinct groups.
Verified live in headless Blender: built a bmesh with an IFCCIRCLE loop and
an IFCARCINDEX loop, then ran bmesh.ops.duplicate on each (the same
bmesh-level operation underlying Shift+D) and called ProfileDecorator's
draw method directly. Before this change, duplicating either loop dropped
both the original and the duplicate from the decorator (0 circle/arc
batches drawn instead of 2). After, both draw immediately, with no change
to the non-duplicated case (still 1) or to genuinely distinct loops (5
independent circles still resolve to 5, not merged). 500-circle timing is
unchanged (~14.3ms/draw before and after), so the added connectivity split
is not a hot-path regression.
Added test/bim/module/model/test_profile_decorator_duplicate_loop.py
pinning the new _connected_components helper's behavior for single and
duplicated circle/arc loops.
This contribution was produced with the assistance of an AI coding tool.
auto_detect_profiles had the identical defect fixed in the previous
commit: duplicating a circle/arc loop in Edit Mode reuses the same
IFCCIRCLE/IFCARCINDEX vertex group index for the new geometry, and this
sibling function (used for curve/annotation editing rather than profile
voids) tallied group membership across the whole mesh instead of per
loop, so it also rejected a legitimately duplicated loop as malformed.
Applied the identical fix: scope the group-count sanity check to each
connected edge loop, computed after the loops are built rather than in
the initial whole-mesh vertex pass. Kept the existing forked-loop check
(more than 2 edges per vertex) in the first pass since it is unrelated
to group counting.
Verified live in headless Blender: constructed two 2-vertex IFCCIRCLE
loops sharing one vertex group index (the exact state Blender's Edit
Mode duplicate produces) and called auto_detect_curves directly.
Before this change it returned (False, "CIRCLE"); after, it returns two
valid IfcCircle curves.
Generated with the assistance of an AI coding tool.
Duplicating a circular or filleted-arc void in the profile CAD editor
(Shift+D on the loop's vertices) reused the same IFCCIRCLE/IFCARCINDEX
vertex group index for the new geometry, since Blender's mesh duplicate
copies vertex group weights but does not allocate a new group. On exit
from Edit Mode, auto_detect_profiles tallied group membership across the
whole mesh rather than per loop, so a group meant to hold exactly 2 (circle)
or 3 (arc) vertices ended up with double that, failing its sanity check
and blocking the edit with an "INVALID PROFILE" popup. Fixes#6944.
Scope the sanity check to each connected edge loop instead, matching how
the loops are actually converted into IfcCircle/arc segments below. Also
explicitly reject an arc/circle vertex tagged onto an isolated vertex with
no edges at all, which the old whole-mesh count also caught.
Verified live in headless Blender against the issue's repro file
(IfcFurniture "Slab.004", IfcArbitraryProfileDefWithVoids with three
IfcCircle voids): entering the profile editor, duplicating one void's
2-vertex loop and moving it produced an "INVALID PROFILE" popup before
this change, and now produces a valid profile (the original 3 voids
intact, plus the duplicate as a 4th void or a separate solid profile
depending on whether it still falls inside the outer boundary).
test/tool/test_model.py passes unchanged (32 passed, 1 pre-existing
unrelated failure present on both before and after).
Generated with the assistance of an AI coding tool.
Both files were merged unformatted and fail the Black formatter step
on every branch, keeping ci-lint red repo-wide.
Generated with the assistance of an AI coding tool.
The ci-lint workflow's ty steps fail on every branch because base
v0.8.0 has four diagnostics.
ty check (bonsai):
- root/operator.py: bpy.data.objects.get() can return None, so
UnlinkObject._execute could put None in its objects list and crash
on the first attribute access when an unknown object name is passed.
Handle the miss explicitly, which also satisfies the declared
list[bpy.types.Object] type.
- tool/sequence.py: ty does not narrow Literal types through
membership tests on list literals, so the assert_never() exhaustive
check was flagged. Use tuple literals, which ty narrows, keeping the
exhaustiveness check intact.
ty check (ios):
- draw.py: arrange_polygons was called through conditional argument
splats that let the same call site work against pre-April-2026
wrappers lacking arrange_polygon_settings and the logger parameter.
No runtime bug for current builds, but the dynamic splats cannot be
typed against the fixed 3-parameter signature. Drop the old-build
workaround and call the current signature directly, following the
precedent of 3d8115ebc5 which dropped similar old-build workarounds
in ifcopenshell.file. Verified against a current wrapper build that
the direct call arranges polygons and serializes to SVG, with and
without a logger.
- Optimise.py: igraph is an optional dependency with a guarded import
and a toposort fallback, but it was missing from the ios type-check
venv so ty could not resolve it. Add it to type-check-requirements
next to the toposort fallback that is already listed.
After this, poe ty-bonsai and poe ty-ios both pass cleanly.
Generated with the assistance of an AI coding tool.
The script called Polyline.get_mode() on every modelspace entity, but
that method only exists on POLYLINE entities, so any typical DXF
containing lines, circles or text crashed with AttributeError before
converting anything. Test for POLYLINE polyface meshes with
dxftype()/is_poly_face_mesh instead and skip other entities with a
message, only create the spatial containment relation when products
exist, and take the input/output paths from the command line (matching
obj2ifc.py) instead of a hardcoded input.dxf/test.ifc.
Fixes#2151
This change was written with the assistance of an AI coding tool.
When grabbing an array child, the selection now expands to include
the array parent and all sibling children before the move operator
runs. Mirrors existing behavior for aggregates and nests.
Generated with the assistance of an AI coding tool.
Ran the new sync_stub.py against a real local build: adds
context.delete_same_facet_edge_pairs (present on the compiled wrapper,
missing from the stub) and drops the module-level logger_or_root
(present in the stub, no longer exists on the wrapper at all).
Nothing else changes - no license header rewrite, no docstring loss,
none of the 14 hand-curated named-parameter constructor/function
signatures touched, unlike the wholesale regeneration this replaces.
Generated with the assistance of an AI coding tool.
generate_stub.py (this branch's earlier commit) regenerates
ifcopenshell_wrapper.pyi wholesale from the compiled wrapper: it
reliably fixes real drift, but it also discards everything that isn't
mechanically recoverable from the wrapper alone - the license header,
docstrings, and hand-curated named-parameter signatures for
SWIG-overloaded constructors/functions (SWIG itself always emits
generic `*args` for those, so a regenerator can't tell a deliberate
curation from real drift and just overwrites it).
sync_stub.py takes the smaller-blast-radius approach: it only adds
top-level symbols/class members that are genuinely missing, and only
removes ones that are genuinely gone, cross-checking against
validate_stub.py's own full canonicalisation (via the newly-exposed
get_names_tree()) so it never mistakes a property()/staticmethod()-
wrapped member for something absent just because its own narrower
parser skips that form. Anything that exists on both sides under the
same name but with a different signature - exactly where curation
lives - is left untouched and reported for a human to review instead
of guessed at.
Verified against a real local build: applying it to the current
ifcopenshell_wrapper.pyi produces a small, targeted diff (add one
missing method, drop one stale function) with the license header,
docstrings, and all 14 curated constructor/function signatures
preserved byte-for-byte, versus generate_stub.py's ~1000-line
wholesale rewrite for the same underlying fix.
Generated with the assistance of an AI coding tool.
Follow-up to the scalar-only fix in #8754, per aothms's direct request on
that PR ("Please do make all int types consistent") and his own original
2023 design intent on issue #3058 ("make all integers (incl. schema
namespaces) an int64_t"). Widens the remaining inconsistent spots now that
compatibility isn't a constraint on this v0.9 branch:
- Integer aggregates (IfcTriangulatedFaceSet.CoordIndex and similar
List<int> attributes), including the SWIG to_vec_int/to_vec_vec_int
helpers, which previously silently truncated via static_cast<int> on the
Python-set path - the same bug class as the original scalar issue.
- The schema code generator (express/mapping.py's integer type mapping),
and all 12 generated schema header/source pairs regenerated to match, so
every schema-typed getter/setter (e.g. IfcOwnerHistory::CreationDate) is
int64_t end to end, not just the dynamic attribute-value path.
Instance/reference identifiers (STEP #123 ids) are deliberately left at
32-bit: they're a file-local index into internal maps, not an EXPRESS
domain value an application chooses, and no realistic STEP file has
billions of entities. The lexer's Token_IDENTIFIER parsing still funnels
through a 32-bit int for this reason - flagged as a known, low-risk gap
rather than fixed, since fixing it would mean touching indexing/hashing
code for no realistic benefit.
Verified: original PR's round-trip tests extended with aggregate cases
(IfcTriangulatedFaceSet.CoordIndex, InnerCoordIndices) at 64-bit boundary
values, in memory and through STEP text, IFC2X3 and IFC4. A standalone C++
program exercising the generated schema API directly (Ifc4::IfcOwnerHistory
::setCreationDate/CreationDate, IfcTriangulatedFaceSet::setCoordIndex/
CoordIndex) confirms int64_t end to end, bypassing SWIG. Full build
(BUILD_IFCGEOM, WITH_OPENCASCADE, BUILD_IFCPYTHON, IFC2X3+IFC4) clean.
test/util/test_attribute.py and test_file.py pass unchanged.
This contribution was produced with the assistance of an AI coding tool.
Setting an IfcInteger/IfcTimeStamp typed attribute (e.g. IfcOwnerHistory.CreationDate)
outside the signed 32-bit range corrupted the value instead of raising, since the
Python wrapper's set_attribute_value_py() truncated it with a plain static_cast<int>
before handing it to the C++ storage. Unix timestamps before 1901-12-13 or after
2038-01-19 silently wrapped around (e.g. 3000000000 became -1294967296) rather than
being rejected or stored correctly. Fixes#3058, equivalent to PR #8683 but ported to
this branch's rewritten ifcparse (snake_case files, variant_array/instance_data
storage, SWIG PyObject-based attribute setter) instead of the old IfcEntityInstanceData
sources, which no longer exist here.
The scalar slot of the attribute variant (Argument_INT) becomes int64_t. Integer
aggregates (Argument_AGGREGATE_OF_INT, e.g. CoordIndex) and instance/reference
identifiers stay 32-bit, since neither is the value that overflows here; this narrow
scope is kept on its own technical merits (aggregates and identifiers were never the
source of the bug, and widening them would be a much larger, riskier change for no
benefit) even though aothms said compatibility isn't a concern on this v0.9-track
branch. express::Base::set_attribute_value promotes the schema-generated int to
int64_t at a single choke point, so the generated setters keep compiling unchanged.
The STEP lexer, writer, and SWIG wrapper (set_attribute_value_py, pythonize) are all
widened together, since widening only the Python-facing setter would have silently
wrapped the value on file write instead of raising.
Verified in a build (IFC2X3 and IFC4, BUILD_IFCGEOM off, no kernels): pre-1901,
post-2038, both 32-bit boundaries, and a 9e12 value all round trip exactly both in
memory and through STEP text serialization (write then reopen). A value outside the
64-bit range now raises a clean exception instead of corrupting data. Ordinary
in-range integers and integer aggregates (e.g. IfcTriangulatedFaceSet.CoordIndex) are
unaffected. The existing util/test_attribute.py and test_file.py suites pass
unchanged; test_entity_instance.py has 5 pre-existing failures unrelated to this
change (confirmed identical on an unfixed build of this branch, caused by a missing
get_info_2 binding and _patch_swig_comparisons never being implemented here).
Generated with the assistance of an AI coding tool.
When a STEP instance has fewer attribute tokens than its schema declares
(commonly from corrupted/malformed syntax), parse_context::construct()
sized the in-memory attribute storage to the smaller token count instead
of the schema's attribute count. This left the storage's last N attribute
slots simply nonexistent rather than blank, so any later read of one of
those trailing attributes by index threw an uncaught IfcParse::IfcException
("Index N is out of range for storage of size N") that terminated the
whole process (SIGABRT) instead of being handled as a parse warning.
Fix: when the schema declaration is known, size the storage to the
schema's attribute count. Indices beyond the number of tokens found are
left at their existing default-constructed blank value (the storage
constructor already blank-initializes every slot), so a truncated
instance now degrades to blank values for its missing trailing
attributes, matching the parser's existing "expected N attribute values,
found M" warning intent instead of crashing.
Reproduced with the fuzzing script attached to #5679: single-byte
mutations of a minimal IFC4 file that corrupt the IFCPROJECT instance's
token stream reliably aborted IfcConvert with this exact exception before
the fix, and now parse with a logged syntax error and exit code 0.
Fixes#5679
Generated with the assistance of an AI coding tool.
set_shape_transparency() called AIS_InteractiveContext.SetTransparency(),
whose argument count is inconsistent across pythonocc-core versions
(reported as a TypeError in #1037). Set transparency directly on the AIS
object instead, the same stable pattern already used elsewhere in this
file (display_shape() calls ais.SetTransparency() directly, never through
the Context), then call Context.UpdateCurrentViewer() to refresh.
app.py's viewer used a "SetSelectionPriority(counter)"/"SelectionPriority()"
pair as an ad hoc unique key to map a displayed AIS object back to its IFC
product. On modern pythonocc-core this crashed with AttributeError because
.GetObject() (needed to unwrap the old handle-based API) no longer exists
on AIS objects (#1098, PR #1113 partially patched one of the two call
sites but left the one in HandleSelection unguarded).
Live pythonocc-core 7.9.3 testing showed the GetObject() guard alone is
not sufficient: SetSelectionPriority/SelectionPriority themselves have
been removed from AIS_InteractiveObject entirely in modern OCCT (only
AIS_Trihedron keeps a same-named but unrelated method for datum parts),
so gating the .GetObject() call with the existing USE_OCCT_HANDLE flag
would still crash the first time a shape is selected. Verified live that
AIS objects retain correct __eq__/__hash__ (matching the underlying OCCT
instance) across separate SWIG wrapper instances, so ais_to_product is
now keyed directly by the AIS object itself, removing the dependency on
the removed OCCT API and the GetObject()/handle distinction altogether.
Verified live against pythonocc-core 7.9.3 (conda-forge) using real
AIS_Shape objects obtained from ifcopenshell.geom.occ_utils.display_shape()
and a real IFC file: reproduced both the original TypeError (#1037) and
AttributeError (#1098), confirmed both fixes resolve them, and confirmed
the ais_to_product dict lookup round trips correctly through a real
Context.Select()/SelectedInteractive() call. Could not exercise the full
Qt-embedded viewer.finished()/HandleSelection() flow end to end because
this pythonocc-core build segfaults natively when creating a second GL
context inside a Qt widget on this macOS host, a pre-existing environment
issue unrelated to this diff (reproduces identically with unpatched code,
before any touched line executes).
AI-generated, reviewed and tested by Petru Conduraru.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
create_shape() returns a Python-owned Element (SWIG_POINTER_OWN in the
boost::variant out typemap). Its .geometry property calls Element::geometry(),
which returns a reference into the element's boost::shared_ptr<Representation>
_geometry member. SWIG wraps that reference as a non-owning pointer, so the
returned Triangulation/BRep/Serialization proxy does not keep the element alive.
When a caller keeps only .geometry (e.g. create_shape(s, e).geometry) and drops
the parent element, Python garbage-collects the element, destroying its
shared_ptr and freeing the underlying representation. Subsequent reads of
verts/faces then return freed memory: empty or implausible float/int garbage,
non-deterministically depending on GC and allocator timing. This is silent data
corruption, not a crash, and has bitten users since 2020.
Fix: in the TriangulationElement/SerializedElement/BRepElement pythoncode, wrap
the geometry getter so the returned geometry stores a backreference to its
owning element (result._parent = self). This makes the parent's lifetime at
least as long as the geometry's, automatically and transparently, so no caller
has to remember to hold the element. This is aothms's suggested backreference,
applied generically in the binding rather than left as a workaround.
Reproduced deterministically (washBasin fixture): before, verts len 0 vs 133500
across repeated GC-pressure runs; after, 133500 every run for all three element
types. test_create_shape passes; no regressions.
Note: tree.select_ray()'s ray_intersection_result (2024 follow-up in #1124) is a
separate ownership mechanism (std::vector element reference + std::array member
pointer) and is left as follow-up scope.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
recalculate_cut() and recalculate_fill() each ran is_intersecting_camera(),
which builds a bmesh and scans every vertex. When a redraw recalculated both
(camera moved, cache miss, or the object selected) that was two full
intersection tests per object per frame for the same answer.
Compute it once in decorate() and pass it to both, and skip the test
entirely when neither recalculation is needed. Never more tests than before,
identical result since the camera can't move within a frame.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
is_camera_moved() runs eval()/numpy over the camera matrix and, as a side
effect, refreshes the stored checksum the first time it returns True. It was
called up to twice per object inside decorate(), so on a frame where the
camera actually moved the first call updated the checksum and every later
call - the fill check on the same object, and both checks on all remaining
objects - then saw an already-current checksum and returned False. Only the
first object's cut got recalculated; its fill and every other element stayed
stale until something else invalidated the cache.
Evaluate it once at the top of __call__ and reuse the flag. This halves the
per-object eval overhead on the common path (viewport navigation with the
camera object stationary) and, when the camera does move, correctly
recalculates the cut and fill for every intersecting element instead of just
the first.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Deleting a type used to strip its occurrences: any that displayed the
type's mapped representation lost their geometry, and inherited material
and presentation styles were dropped too.
The no-SHIFT "Delete Type" path now bakes each occurrence's geometry,
styles, and inherited material onto the occurrence before the type is
removed:
- Refactor UnassignType's unmap logic into a reusable
UnassignType.unassign_and_unmap(), and extend it to re-attach styled
items (copy_deep only follows forward refs, so IfcStyledItem is lost)
and bake down any inherited (non-owned) material.
- Add RemoveType._detach_type_material_set(): unhook the type's
IfcMaterialLayerSet/ProfileSet association cascade-free before deletion,
so remove_product's aggressive unassign_material never fires and the
occurrences' layer/profile-set usages survive intact.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adds a trash button in BIM_PT_type_attributes that deletes the relating
type via bim.remove_type. SHIFT+Click also deletes every occurrence of
the type in the project, behind a confirmation dialog showing the count.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
aothms asked for the toposort dependency ordering used by the dedup
walk to try igraph's C-backed topological_sorting() first, since it
should shave off additional time on top of the non-recursive
get_info fix. Falls back to the pure python toposort package with a
warning if igraph is not installed.
Generated with the assistance of an AI coding tool.
The 2020 profiling in issue #1043 found the Optimise recipe's dedup
loop spent almost all of its time in entity_instance.get_info(recursive=True):
because the topological sort already guarantees every referenced entity
is folded before the entity that references it, recomputing each
already-folded subtree's canonical value from scratch for every parent
that points to it is wasted work. Confirmed this is still exactly the
bottleneck in the current codebase, unchanged since 2020 (get_info's
recursive path still walks the whole subtree on every call).
Applied aothms's suggested fix from the issue thread: canonicalize each
entity with a non-recursive get_info, and for referenced entities substitute
the already-computed identity of their folded replacement (looked up in
instance_mapping) instead of re-expanding the subtree. Also limited the
toposort dependency graph to direct references (max_levels=1), since a
topological sort only needs direct edges, not the full transitive closure
traverse() was computing for every entity.
Benchmarked before and after on real IFC test fixtures and a larger
synthetic file with heavily shared geometry (thousands of walls sharing
a handful of profile/point subtrees, mirroring the sharing pattern
described in the issue):
- test/input/geometrygym_great_court_roof.ifc (56989 entities): 9.9s -> 1.7s
- test/input/acad2010_objects.ifc (16296 entities): 3.7s -> 0.4s
- synthetic 120083-entity fixture with heavy geometry sharing: 19.2s -> 3.4s
Verified correctness by comparing the full canonical (recursive get_info)
multiset of the optimized output between the old and new implementation on
all three fixtures: identical results, same fold counts.
Added test_Optimise.py covering the core scenario from the issue: entities
built from separate, value-identical non-rooted subtrees fold to a shared
instance, while entities with distinct values do not.
Generated with the assistance of an AI coding tool.
IfcSurfaceCurveSweptAreaSolid regressed in 0.8 for geometry far from the
origin (for example parapets on a georeferenced building), which went
missing or glitched.
The kernel offsets the directrix toward the origin when it is far away
(mean.norm() > 1e2), storing the offset copy in a local curve variable and
setting applied_temporary_offset so the finished solid is translated back by
+mean. But the wire was still built from scs->curve, the un-offset original,
so the offset never took effect and the result was translated by +mean from
its correct location. Build the wire from curve instead. When no offset is
applied curve aliases scs->curve, so near-origin geometry is unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
references_to_resolve is never set while parsing header
entities, so binding a reference to it was UB, caught by
UBSan on any file with a header.
Generated with the assistance of an AI coding tool.
Two related bugs in read_from_stream's reference-resolution
loop, both reachable from malformed input:
- has_attribute_value<IfcBaseClass*> only checks the stored
slot's type, not that it's non-null (e.g. an explicit $
value), so the following get_attribute_value() call could
return null and inst->declaration() crashed on it.
- byid_[ref] default-inserts (and returns) a null pointer
when the owning instance id isn't present, which was then
dereferenced unconditionally via ->data().
Added regression tests using the two minimized crash inputs
that found these.
Generated with the assistance of an AI coding tool.
`values` dictionary was missing and variables were never collected to it, so `FormulaEvaluator(values)` was always resulting in missing variable error.
The fixture declared FILE_SCHEMA(('IFC2X3')) but used
IFCPROPERTYSETDEFINITIONSET(...), a defined type that only exists in
IFC4+ (confirmed absent from the generated Ifc2x3-schema.cpp/
Ifc2x3-definitions.h, present in the IFC4 equivalents). The file's own
FILE_NAME record ('Column_4x3.ifc') suggests it was originally
exported as IFC4X3 and the schema tag was later miscopied to IFC2X3.
Traced with an instrumented parser build: on encountering the
unrecognized keyword, declaration_by_name() correctly throws
"Entity with name 'IFCPROPERTYSETDEFINITIONSET' not found in schema
'IFC2X3'", caught by the existing IfcException handler in
in_memory_file_storage::load(). The parser then falls back to parsing
the trailing (#136,#138) as a plain nested SET rather than the typed
value, so RelatingPropertyDefinition ends up as a bare tuple instead
of an IfcPropertySetDefinitionSet-wrapped value with .is_a(). This is
correct, expected behavior for content that doesn't match its
declared schema - not a parser bug. Fixing the header to IFC4 (which
does declare the type) resolves test_stream, test_file, and test_rocks
in test_streaming_rocksdb_and_simpletyperefs.py.
Generated with the assistance of an AI coding tool.
src/ifcgeom/kernels/opencascade/boolean_utils.cpp, OpenCascadeKernel.cpp,
and boolean_result.cpp logged diagnostics (including the "Processed
fully in 2D" family of messages) through the global Logger::Root()
singleton. IfcConvert's main() constructs its own Logger and wires it
to --log-file via SetOutput(), then threads that instance through
Converter/kernel constructors as logger_ (see AbstractKernel). Since
Logger::Root() is never itself configured with an output stream, every
Notice/Warning/Message call through it was silently dropped instead of
reaching the log file - Logger::Message's log1_/log2_ null checks just
no-op.
This made src/ifcopenshell-python/test/test_wall_opening.py fail: it
asserts on specific log messages that the underlying boolean-op code
was still emitting correctly, just to nowhere. The geometry itself was
never wrong.
Add a Logger*, defaulting to null, to boolean_settings (with a log()
accessor falling back to Logger::Root() for the few remaining
call sites with no injected logger available), thread it through
eliminate_narrow_operands and boolean_subtraction_2d_using_builder,
and have OpenCascadeKernel/boolean_result.cpp populate it from their
inherited logger_ member instead of relying on the global singleton.
Generated with the assistance of an AI coding tool.
The class-pairing validation added in 10ee5aef4f rejects any type
assignment whose class isn't in the buildingSMART implementer
agreement map. That map only covers physical product occurrence/type
pairs (IfcWallType -> IfcWall, etc); IfcTypeProcess and IfcTypeResource
subtypes such as IfcTaskType, IfcProcedureType and the resource types
have no entry, so previously-valid assignments like
IfcTaskType -> IfcTask were rejected with "allowed occurrence
classes: <none>".
These classes still follow the schema's universal Type-suffix naming
convention, so derive the pairing the same way the existing
ApplicableOccurrence fallback does: strip "Type" from the relating
type's class name and accept it only if the schema actually declares
that entity. This can only add pairings implied by the type's own
class name, so it cannot loosen the existing rejection of genuine
mismatches (e.g. IfcWallType -> IfcWindow).
Generated with the assistance of an AI coding tool.
The crease and sharp weighting seemed flipped to my sensibilities, so
now crease is heavier than sharp. I also added a commented out block
for debug colours in case someone wants to quickly use bright colours
to diagnose future problems.
Add UseEdgeClassification, RenderCreases, ValleyAngleMinDegrees,
RenderSharp, RidgeAngleMinDegrees, and RenderFlush to EPset_Drawing,
following the existing HasUnderlay/DPI/PerspectiveShiftX pattern.
The master toggle defaults off, preserving current linework output;
the three dependent controls only show in the panel once it's on.
Removes the previous dormant, transient operator-redo properties for
the ridge/valley thresholds and flush-edge toggle, which were never
persisted per-drawing or exposed in any panel, replacing them with
the persistent camera properties read in setup_serialiser().
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Add svg-use-edge-classification (default off, preserving today's
linework), svg-render-crease-edges, and svg-render-sharp-edges
settings, gating the existing 5-class classification feature so it
can be disabled entirely (falling back to the pre-classification
whole-shape output) or have individual classes suppressed.
Also fixes a bug uncovered while wiring this into Bonsai: ready(),
where geometry_settings() actually gets read into the serializer,
was only ever invoked explicitly by IfcConvert's CLI driver and
isn't exposed to Python. Every Svg* setting -- including the three
from previous rounds -- silently stayed at its hardcoded constructor
default when the serializer was constructed directly through the
Python bindings, as Bonsai does. Fixed by calling ready() from
SvgSerializer's own constructor, safe since it only reads
geometry_settings() with no other side effects, and settings are
always finalized before construction in every call path.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Circular-profile IfcColumn/IfcPile elements produce a genuine
analytic cylindrical BRep face (via BRepPrimAPI_MakePrism), not a
tessellated facet. The edge classification/extraction pipeline is
edge-identity-based end to end, but a smooth surface's silhouette is
synthesized by HLR on the fly and has no corresponding pre-existing
edge to bucket, so it was silently dropped once any edge in the
product had been classified. Add a face-level pass that includes any
non-planar face directly in the outline bucket, giving HLR's
per-face OutLine reconstruction a face identity to correlate
against. Purely additive: diffing the whole test scene's output
before and after shows only the two previously-missing tangent
lines appear, nothing else changes.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Re-enable the view-relative sign flip for folds seen through an
opening (e.g. a box with a face removed), reverted in the previous
commit after it corrupted unrelated geometry. The earlier revert's
diagnosis was slightly off: bucket reassignment can't affect HLR's
own visibility computation, so the corruption was actually an
asymmetric-threshold artifact -- an unconditional flip re-tested
small, correctly-flush deviations against the much smaller valley
threshold instead of the ridge one. Gating the flip so it only
reinterprets folds that already clear their own pre-flip threshold
fixes the box case while leaving every other test object's
classification unchanged (verified against the full test scene).
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Fixes three bugs in classify_edge_from_faces() found via real-world
testing against a dedicated stress-test scene (icosphere, Suzanne,
cylinders/cones at various orientations, a dihedral-angle sweep rig):
- The outline (silhouette) test used a bare sign comparison, so a face
at or near exactly edge-on to the camera could land on the wrong
side of zero and fall through to angle-based classification instead
of being drawn as outline. Now uses a tolerance band around zero,
matching an equivalent check already used elsewhere in this file.
- The signed deviation-from-flat formula was inverted (180 - angle
instead of angle), so small, genuinely near-flat facet angles came
out with a large computed deviation and always classified as
sharp/crease, never flush. This is why thresholds appeared to have
no effect. Also replaced the edge/wire-orientation-based convexity
sign (unreliable on real BRep topology, verified wrong against a
known fully-convex icosphere) with a simpler position-based test.
- A specific edge that was previously missing entirely (not just
misclassified) reappears correctly as a side effect of the outline
fix above; no separate change was needed for it.
A fourth issue (folds viewed through an opening, e.g. a box missing a
face, should read as crease rather than sharp) was attempted via a
back-facing sign flip, but reverted: it broke the fixes above broadly,
since "both faces back-facing" isn't a rare look-through-a-hole case
once HLR has already filtered to visible edges only. Documented in a
code comment for whoever picks this up next.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Adds boundary/outline/sharp/crease/flush classification of HLR
projection edges in SvgSerializer, so CSS can style silhouettes,
ridges, and valleys differently instead of drawing every edge
identically (fixes the "ugly faceted sphere" problem from #3668).
Classification happens pre-HLR on the original solid's real face
topology (three prior attempts tried to classify HLR's own output,
which carries no face topology at all and can't be correlated back
by edge identity). Each class's visible portion is then extracted via
HLRBRep_HLRToShape::VCompound(S)/OutLineVCompound(S), the same
per-shape filtering mechanism already used for per-product
segmentation, applied per class instead. Classes are tagged directly
on individual <path> elements so Bonsai's merge_linework_and_add_metadata
group-level class rewrite in operator.py never touches them.
New settings: svg-ridge-angle-min-degrees, svg-valley-angle-min-degrees,
svg-emit-flush-edges (ConversionSettings.h), wired through Bonsai's
CreateDrawing operator and exposed via its redo panel.
Refs #3668.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The Reassign Class operator refused to reassign an element to a different
IFC product family unless it was an IfcElement <-> IfcElementType swap, so a
piece of geometry mistakenly hosted on IfcSite could not be turned into
IfcFurniture even though root.reassign_class handles it fine.
Loosen the guard: only block the case that actually matters - a spatial
element (IfcSpatialElement / IfcSpatialStructureElement for IFC2X3) with no
geometry, which would be a real containment-hierarchy container rather than
a stray modelled object. Everything else reassigns freely.
Closes#8664
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
When converting a wall representation to a parametric extrusion via the
Representation Utilities buttons, an element that has openings would report
"has openings - representation cannot be updated" and stop, without telling
the user there is an ALT+click path that bakes the openings into the new
representation. Point the message at that path so the error is actionable.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* Bonsai: place auto-generated opening boundaries at their real position #8237
auto_generate_boundaries (single-space mode) built each opening/filling boundary
from the opening's LOCAL geometry (get_vertices) but first did
mat.translation = (0, 0, 0) on its placement matrix. Because the vertices are
local, that placement translation is exactly what carries the opening to its
real location, so zeroing it collapsed every window/door boundary onto the
origin. This is why the auto path misplaced window boundaries while the
single-element path (create_element_boundary) placed them correctly, as
@MDHering observed with the two modes. Keep the full placement matrix.
Verified on the reporter's file: the opening's real placement is (0.1, 1.5, 1.0);
a vertex went from (0.6, 0, 0) under the old code to (0.7, 1.5, 1.0) with the fix,
i.e. moved by exactly the (0.1, 1.5, 1.0) that was being discarded.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Remove superfluous comment from #8237 fix
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: CyrilWaechter <cyril@biminsight.ch>
It was passing `IFC4X3` directly to `schema_by_name` which is expecting
schema identifier (e.g. IFC4X3_ADD2, not IFC4X3 allowed by `IFC_SCHEMA`
- IFC4X3 is one of the IFC4X3 iterations while it was in development,
not the final one).
Noticed by tests failing:
FAILED
test/util/test_schema.py::TestGeometryClassesIntroducedAfter::test_ifc4x3_to_ifc2x3_is_superset_of_ifc4_to_ifc2x3
- RuntimeError: No schema named IFC4X3
FAILED
test/util/test_schema.py::TestGeometryClassesIntroducedAfter::test_ifc4_to_ifc4x3_is_empty
- RuntimeError: No schema named IFC4X3
Error was:
```
configure: error: could not find a working compiler, see config.log for details
```
config.log:
```
conftest.c: In function 'f':
conftest.c:12:48: error: too many arguments to function 'g'; expected 0, have 6
12 | for(i=0;i<1;i++){if(e(got,got,9,d[i].n)==0)h();g(i,d[i].src,d[i].n,got,d[i].want,9);if(d[i].n)h();}}
| ^ ~
```
Last reference to this file was dropped in 7ae685dbf, though the ref was
pointing to `/patches/opencollada/pr622.patch`, so IIUC
`patches/pr622.patch` was never used.
AST parser has changed a bit and there are some minor differences in the .py output. Updating files just to avoid seeing these diffs when rerunning rule compiler.
Example error:
```
ast.Str(s=node.attr),
^^^^^^^
AttributeError: module 'ast' has no attribute 'Str'
```
`ast.Str` was deprecated since 3.8 and was removed in 3.14, see https://docs.python.org/3/whatsnew/3.14.html#id9
Because uv was always trying to install when starting a venv in `ifcopenshell` folder, though they might be already available globally. And also they were listed twice - in pyproject and in the ci-lint.yml, now there's a single source of truth.
versionURLs in brand.html used http:// while the docs sites are
served over https://, so currentURL.includes(url) never matched and
the <select> never reflected/switched to Unstable. Fixes#8023.
Generated with the assistance of an AI coding tool.
Two TestImplementsTool failures on v0.8.0:
- test_cost.py: Cost could not be instantiated because
core.tool.Cost declared abstract get_direct_cost_item_products, which
tool.cost.Cost never implements. The method is dead (zero call sites;
get_cost_item_products(is_deep=False) already covers the 'direct'
case), so remove the abstract declaration.
- test_ifcgit.py: tool.ifcgit.IfcGit was not declared as a subclass of
its core.tool.IfcGit interface (unlike every sibling tool class), so
the isinstance check failed. Add the base class (and the
bonsai.core.tool import it needs). All 50 interface methods are
already implemented on the concrete class.
No behaviour change. Verified in headless Blender: isinstance(Cost(), core.tool.Cost) and isinstance(IfcGit(), core.tool.IfcGit) both True (were TypeError / False); repo abstract-vs-impl diff confirms all IfcGit abstracts are implemented.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
black (the version CI's psf/black@stable resolves to) flags three spots
in util/selector.py: the chained .replace() in FormatTransformer.number,
the suppress_zero_inches kwarg in format_length, and the long
`elif key in (...) and hasattr(...)` placement-key tuple in
set_element_value. Reformat all three to black's multi-line style.
Formatting only, no behavioural change (all keys preserved).
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
generate_annotation built the annotation list from a set union and sorted it by
ZIndex and TEXT-ness only. Annotations that tied on that key kept set iteration
order, which follows entity hash (step id plus the process memory address), so
the order of tied annotations (for example a label and its background fill)
shuffled between Blender restarts and flipped their draw order.
Add the stable IFC step id as a final tiebreaker so the order is total and
session independent. Behavior preserving, no z-layer semantics changed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
coerce_value assumed value_str was always a CLI string, but ifcmcp
passes JSON-decoded native types (int, None) straight through. Guard
the Union/Optional "none" check so it only calls .lower() on strings,
and handle native None explicitly.
IfcSpace is not a subtype of IfcElement, so quantify.run_quantify()'s
default selector silently skipped all spaces, reporting
elements_quantified: 0 with no error or warning.
Generated with the assistance of an AI coding tool.
Importing a Primavera P6 XML crashed with
`AttributeError: 'NoneType' object has no attribute 'text'` in
P62Ifc.parse_activity_xml, which read
activity.find("pr:CalendarObjectId").text unconditionally. CalendarObjectId
is optional on a P6 Activity; when omitted, the activity inherits the
project's ActivityDefaultCalendarObjectId.
Capture the project default in parse_xml and fall back to it when an
activity has no CalendarObjectId (`calendar_id or self.default_calendar_id`).
Verified on the reporter's attached file (20241021 Cronograma.xml): 3 of 14
activities lack a CalendarObjectId and reproduced the exact crash on
v0.8.0; after the fix parse_xml completes and those activities resolve to
the project default calendar "2" (a valid calendar in the file). An
activity with an explicit CalendarObjectId keeps its own value.
Fixes the P6 re-import crash reported in #5617 (that issue tracks several
Gantt items; this addresses the import AttributeError).
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
get_pset and get_psets assumed RelatingPropertyDefinition is a single property
definition and read definition.Name directly. When it is an
IfcPropertySetDefinitionSet (a defined type wrapping a list of property set
definitions) that attribute access raised AttributeError, so an element whose
psets are grouped in a set returned none of them.
Unpack IfcPropertySetDefinitionSet into its members in both loops and process
each one. Single property definitions and the psets_only and qtos_only filters
are unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Several BDD scenarios hardcode absolute representation-item object names
whose trailing number is the IFC STEP line id
(f"Item/{item.is_a()}/{item.id()}"). Those ids drift when file-creation
order changes; a recent shift moved all of them by a uniform -4, so the
scenarios failed with "Item/.../NN does not exist".
The failing step (the_object_name_exists in test_feature.py) dumps the
full bpy.data.objects listing on failure, so the correct current ids are
recoverable directly from the CI log (run 29208793599, tested commit
36e21e882f, an ancestor of HEAD with only a .gitignore commit between).
Renumber to match:
IfcExtrudedAreaSolid/77->73, IfcPolygonalFaceSet/76->72,
IfcVertexPoint/69->65, IfcEdge/72->68, IfcFace/74->70.
Verified against the CI failure dump (a local build produces different
ids, so this is validated by CI's own object listing rather than a local
run). boolean.feature also hardcodes IfcHalfSpaceSolid/90 and panel text
[91] downstream of the failing assertion, which CI never reached and so
never dumped; left as-is to avoid guessing - they will print a fresh dump
next run for a follow-up if still stale.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Two host-environment bugs in the build-env scripts that break on
macOS/Apple Silicon hosts, independent of target architecture:
- Dockerfile: groupadd fails outright when USER_GID collides with an
existing system group in the rockylinux9 base image (e.g. macOS
default user GID 20 "staff" collides with RHEL's GID 20 "games").
Guard with getent so useradd attaches to the existing group instead.
- ifcos_env: `sed -si` is GNU-only syntax and errors under BSD/macOS
sed. Do the UNIQUE_ID substitution via a portable temp-file + mv.
Per sboddy's review on the original PR: dropped the linux/amd64
platform-pin additions from this change. The stack already targets
Rocky9/x64 build outputs by design, and Docker Desktop on macOS has
no native container runtime regardless (it's a Linux VM either way),
so forcing the image to run under emulation doesn't produce anything
that's actually loadable into a native macOS Blender/Bonsai install.
That's a separate, harder problem worth solving via a native build
path instead (mirroring build_osx.yml), not by fighting emulation
here. These two fixes stand on their own merits on any host.
This change was made with the assistance of an AI tool.
The ccache named volume had no explicit name, so Docker Compose
namespaced it under the per-checkout project name (derived from
UNIQUE_ID), giving each checkout its own cache even though
docker/README.md already documented them as shared. Give the volume
a fixed name so all checkouts attach the same one.
Measured cache size after a full build (IfcParse+IfcGeom+IfcConvert+
wrapper, one Python version) is ~300MB, only ~5% of the previous 5G
cap. Shrink CCACHE_MAXSIZE to 2G, which comfortably covers the shared
baseline plus per-branch deltas from several diverging checkouts.
Generated with the assistance of an AI coding tool.
Three host-portability fixes to the docker/ toolchain from #8564 so it
runs on macOS as well as Linux. All three are no-ops on native amd64
Linux.
1. Dockerfile: only groupadd when the target GID is free. macOS's default
primary group `staff` is GID 20, which already exists as `games` in
rockylinux:9, so `groupadd -g 20` aborted the image build. Guard with
`getent group "${USER_GID}" || groupadd ...`; useradd -g accepts the
existing GID.
2. ifcos_env unique(): replace GNU-only `sed -si` (BSD/macOS sed errors
"illegal option -- s") with a portable `sed > tmp && mv` rewrite of the
UNIQUE_ID line. Verified against macOS BSD sed.
3. create() + compose.yaml: build with an explicit `--platform linux/amd64`
so the locally built image's platform matches the `platform:
linux/amd64` pin in compose.yaml. Without it, on arm64 the local image
is tagged linux/arm64, compose treats the platform-mismatched image as
absent and tries to pull `ifcopenshell-build-env:updated` from Docker
Hub (which does not exist -> access denied). Also add `pull_policy:
never` as a safety net so a future mismatch surfaces as a clear "image
not found" rather than a registry auth error.
Note: on Apple Silicon the amd64 build runs under emulation and a cold
full build is slow; ccache makes incremental rebuilds tolerable. A native
Linux/Intel host or CI remains the better choice for routine use, but these
fixes turn "hard broken" into "works with a caveat" on macOS.
This change was made with the assistance of an AI tool.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Reported: Blender segfaults when clicking Cancel on the "newer
autosave found" recovery popup shown by LoadProject at startup.
Root cause: LoadProject.execute()/invoke() triggered the recovery
popup via bpy.ops.bim.load_autosaved_recovery_popup("INVOKE_DEFAULT",
...) and returned that call's result ({'RUNNING_MODAL'}) as their own
return value, without LoadProject itself ever calling
modal_handler_add(). Blender's window manager takes a RUNNING_MODAL
return as a promise the operator registered its own modal handler;
since it hadn't, the WM's operator bookkeeping was left corrupted -
silently, since this is heap/state corruption rather than an
immediate crash. It only surfaced later, when the real modal operator
(the popup) closed and the WM reconciled its modal stack, which lines
up with the crash occurring specifically on dialog close regardless
of which button was pressed. check_autosave_recovery() now returns a
plain bool and fires the popup fire-and-forget; LoadProject reports
its own honest {"FINISHED"}.
Also hardened, as defense in depth: LoadAutosavedRecoveryPopup's
execute()/cancel() call back into bim.load_project(...), which (with
should_start_fresh_session) calls wm.read_homefile() and tears down
the window manager/screens. Doing that synchronously from inside this
popup's own execute()/cancel() - itself invoked from deep inside
Blender's modal handling for the popup's button click - risks the
same class of use-after-free as the timer bug fixed in the previous
commit. The reload is now deferred by one timer tick so it runs after
the popup's modal handling has fully unwound, and the deferred
callback closes over plain values rather than `self`, since the
operator instance may not survive past cancel()/execute() returning.
This defer-only change was tried and tested first, on the (incorrect)
assumption it was the root cause: it produced a byte-for-byte
identical crash backtrace on retest, which is what pointed at the
RUNNING_MODAL bug above as the actual cause - the defer change alone
was insufficient because the corruption happens when the popup is
first shown, not when it's closed.
Generated with the assistance of an AI coding tool.
The periodic autosave timer called reset_timer() at the end of its
own callback, which unregistered the timer that was still executing
(itself). Blender frees the timer's internal registry entry on that
manual unregister, then frees it again when the callback returns
None - a double free that corrupts the heap and can crash Blender
later, once the corrupted memory is reused.
Reschedule by returning the next interval from the callback instead,
which is the safe, documented way to repeat a bpy.app.timers
callback. External reset_timer() calls (from SaveProject,
LoadProject, AutosavePrompt) are unaffected since they run from a
separate call stack (UI events), not from inside the timer.
Found while investigating a segfault reported when cancelling the
autosave recovery popup; not itself the cause of that crash (see the
following commit), but the same reentrant-unregister pattern and a
real, independent latent bug in the periodic reminder path.
Generated with the assistance of an AI coding tool.
Dockerfile (renamed from Dockerfile_init, Dockerfile_update removed):
- Run as a non-root `builder` user matching the host UID/GID (passed as
--build-arg by create() from id -u/id -g), so build output under the
bind mount stays owned by the host user instead of root.
- Fix CCACHE_MAXSIZE: `ccache -M 5G` wrote its limit to a config file
under /ccache at image-build time, but /ccache is a volume mount
point, so that file gets shadowed by the (empty) volume the moment
the container actually runs - the cap never took effect. Set
CCACHE_MAXSIZE=5G as an image ENV instead.
- Dedupe ccache/libffi-devel, add --setopt=install_weak_deps=False
--setopt=tsflags=nodocs, add `git lfs install --system`, combine the
dnf update+install into one layer.
- Drop Dockerfile_update: it built FROM its own previous output, so
every `update` call made the image strictly larger forever (Docker
layers are append-only, `dnf clean` in a later layer can't shrink an
earlier one). `update` now just calls create(), which already runs
`dnf update -y` FROM a clean rockylinux:9 every time.
compose.yaml: pin platform: linux/amd64 so this doesn't silently run
under emulation on an ARM host.
ifcos_env:
- Split the previously-conflated stop/down into six distinct,
Compose-native lifecycle commands: up (create-or-start), down
(remove), stop, start, restart (stop+start, same container),
recreate (down+up, fresh container). Previously `stop` was aliased
to `down`, which silently removed the container instead of pausing
it.
- Implement try(): copies the built wrapper into a real Blender/Bonsai
install for manual testing, reading the target from a new
BLENDER_USER_RESOURCE .env variable and auto-detecting the built
Python version (disambiguating via PY_TGT for multi-version builds).
Deliberately kept human-only - it mutates a live Blender install, so
it shouldn't run unattended as part of an automated/AI workflow,
which should instead copy the wrapper into the repo's own
src/ifcopenshell-python/ifcopenshell/ (documented in SKILL.md).
- Fix unique(): the "has .env already got a UNIQUE_ID line" check
referenced an unset $FILE instead of $ENV_FILE, so it always
evaluated true and appended a fresh "UNIQUE_ID=dummy" line to .env
on every single `up`.
- Minor: differentiate remove()'s log message from down()'s (no longer
identical now that they're distinct operations), tidy help text
alignment and a stray double-space typo in clean().
SKILL.md: rewritten as current-state documentation (no more "fixed in
this copy" changelog framing) covering the above, plus a migration
note for anyone hitting root-owned leftovers from an older image.
Verified by actually building the image and driving every new
lifecycle command (stop/start/restart keep the same container ID;
down+up and recreate produce a new one) and try() (including the
quoted-tilde BLENDER_USER_RESOURCE edge case) against the real container.
Generated with the assistance of an AI coding tool.
First functional version, but it needs some improvements and fixes
identified as I've used it personally on one thing, and when an AI
(Claude) used it to work through the CI test errors.
I had the AI make a SKILL.md file. If the AI indicates it needs to
build the ifcopenshell binary, use this and let it rip.
The project-unit to Blender-unit mapping in format_distance only knew
FOOT/INCH/METRE/DECIMETRE/CENTIMETRE/MILLIMETRE, so creating a project
with Kilometers or Miles in the New Project Wizard crashed with
KeyError: 'KILOMETRE' (or 'MILE') as soon as the spatial tree formatted
an elevation. Add the missing Blender-supported units (kilometre, mile,
micrometre) and fall through gracefully for anything else (for example
HECTOMETRE) so unknown units use the adaptive formatting branch instead
of raising.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Expose the Euler rotation of an element's placement in degrees through
get_element_value, alongside the existing x/y/z and easting/northing/
elevation keys. This makes element rotation exportable through ifccsv,
e.g. for placing oriented symbols in GIS.
Adopts the approach agreed in the review of the stale PR #6272 by
@TZwielehner: reuse util.shape_builder.np_matrix_to_euler and do the
degree conversion inside get_element_value.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Project loading set scene length_unit to f"{Prefix}METERS", but Blender's
enum only defines KILOMETERS, CENTIMETERS, MILLIMETERS and MICROMETERS.
A model with a DECIMETRE (or HECTO/DECA/etc.) length unit therefore raised
on the enum assignment and the file failed to open. Guard with the set of
supported values and fall back to ADAPTIVE display for the rest.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The quickstart ended with three empty sections whose bodies were only
"TODO" (placing occurrences, changing locations, modeling a building),
which read as a dead end on docs.bonsaibim.org. The page now ends on the
completed save-and-view flow.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The guard that avoids re-assigning the same object to the same resource
tested is_a("IfclRelAssignsToResource") (stray "l"), so it never matched.
A repeat assignment therefore fell through and appended the related object
to RelatedObjects a second time. Corrected to "IfcRelAssignsToResource".
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
IfcDiff defaulted to relationships=["geometry"], so a plain diff only ever
compared geometry. Attribute-only edits on an element that kept its GlobalId
(a modified or removed PredefinedType, a renamed element, etc.) were silently
missed. The CLI made this worse: --relationships did not list "attributes" or
"geometry" as valid values, so there was no documented way to enable it.
The default is now ["attributes", "geometry"], so a plain `ifcdiff old new`
reports attribute changes alongside geometry changes. The CLI help and the
IfcDiff docstring now document all valid relationship values.
Added a regression test covering a PredefinedType change detected with the
default configuration.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
FormatTransformer.round() called Decimal() directly on the input value,
which raises decimal.InvalidOperation when the value is a non-numeric
string (a text property, or a value carrying a unit suffix like "12.5 m").
In a spreadsheet export this crashed the entire operation as soon as one
element carried such a value.
Now round() catches InvalidOperation and returns the value unchanged, the
same graceful-fallback convention used by add(). Numeric rounding is
unaffected.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
In the cobie24 Coordinate sheet, Floor rows use get_local_placement, whose values
are in the project length unit, but Space rows come from ifcopenshell.geom
create_shape, whose vertices are in SI metres, and the space branch never scaled
them back. So on a non metre model (for example millimetres) the Coordinate sheet
mixed units a thousandfold apart and disagreed with the Facility sheet's declared
LinearUnits.
Scale the space bounding box by the project unit scale so the whole Coordinate
sheet is consistent. A metre model is unchanged since the scale is 1.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The filter_elements selector grammar had no way to comment out part of a
query, so users had to delete and retype text to temporarily toggle a
facet. Add a /* ... */ block comment terminal that is ignored by the
lexer, and tolerate a trailing "+" so that commenting out the final
operand (e.g. "IfcWall + /* IfcSlab */") parses cleanly. Comments may
span multiple lines; a /* sequence inside a quoted string is not treated
as a comment. Only the filter grammar is affected, not get_element or
format which use "/" for regex and division.
Adds a regression test and documents the syntax.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The location and parent filters both match at any depth in the spatial
hierarchy, which surprises users who want only the elements immediately
under a given container. Document that the parent query key resolves the
direct parent only (e.g. query:"parent.Name"="My Site"), add a matching
filter example, and note the immediacy on the parent value key.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
SHIFT+CTRL+CLICK on Activate Drawing now imports the
annotations of all selected drawings without switching
the active view or camera, then selects their cameras with
the first as active. SHIFT+CTRL+ALT+CLICK also selects the
loaded annotation objects. The drawing camera is imported
when missing so annotations land in the correct collection.
Loading is idempotent.
Generated with the assistance of an AI coding tool.
Converting to a path whose directory does not exist (or is not writable)
failed silently: the serializer's ready() check correctly returned false,
but IfcConvert deleted the temp file and returned EXIT_FAILURE without any
message, so the user saw no reason for the failure.
Log a SYS error naming the output file before returning, matching the
existing "Unable to open output file" reporting used elsewhere.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Property sets contained in an IfcPropertySetDefinitionSet were exported as
an empty element in XML. The XmlSerializer already had a block to expand
such a set into its member property sets, but it was gated behind
#ifdef SCHEMAS_HAS_IfcPropertySetDefinitionSet while the schema generator
emits SCHEMA_HAS_IfcPropertySetDefinitionSet (singular). The plural spelling
is defined nowhere, so the block was dead code and a RelatingPropertyDefinition
holding a set produced nothing.
Correct the macro name so the set is expanded and its property sets are
serialized. The parse layer already reads these nested sets (they are
reachable from util.element), so this only completes the XML path.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
escape_xml escaped the five XML metacharacters but passed control
characters (0x00 to 0x1F other than tab, newline and carriage return)
through unchanged. Those bytes are illegal in XML 1.0 and cannot be
represented even as numeric character references, so any IFC string
containing them produced non-well-formed XML and SVG output.
Strip those illegal control characters before escaping. Bytes belonging to
a valid UTF-8 multibyte sequence are always >= 0x80, so filtering on the low
control range leaves real text intact. This is the shared helper used by the
SVG serializer text and attribute sites (audited: all route through it) and
by the XML/Collada paths, so both reports are resolved at one place.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
IfcTriangulatedFaceSet and IfcPolygonalFaceSet used CoordIndex values to
index Coordinates.CoordList directly, ignoring the optional PnIndex
attribute. When PnIndex is present it remaps point references, so a
CoordIndex value i must resolve as CoordList[PnIndex[i-1]-1] (both 1-based).
Without the indirection any model carrying a PnIndex was built from the wrong
points.
Add a resolve() helper in both mappings that applies the PnIndex indirection
when present and is a plain bounds-checked lookup otherwise, with bounds
checks at both index levels. When PnIndex is absent the behavior is unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Boost.System has been header-only since Boost 1.69 and its compiled stub
library was removed in newer Boost, so listing system in the requested
find_package components makes configuration fail on Boost 1.70 and up (for
example Boost 1.90 errors with "Could not find boost_system"). Boost.System
is still pulled in transitively by thread / iostreams where it is needed, so
drop it from the explicit component list.
Verified: with this change IfcOpenShell configures and builds IfcConvert
cleanly against Homebrew Boost 1.90 and OpenCASCADE 7.9.2.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Address maintainer request on #8368: instead of a deflection floor on top
of a fixed CircleSegments count, use one mode or the other. When
CircleSegments == 0 (the new default) the CGAL kernel derives the conic
segment count from MesherLinearDeflection, matching the deflection based
meshing OpenCascade already does and fixing #8051. When CircleSegments is
non zero it is used directly as a fixed, radius independent count.
CircleSegments is only read by the CGAL kernel; OpenCascade meshes by
deflection and never reads it, so the new default has no effect there.
Update the setting description and the ifcconvert / geometry-settings docs.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The CGAL kernels (cgal and cgal-simple) allocate arc segments as a
fraction of the full circle via CircleSegments, ignoring the radius.
A large-radius arc that spans a small angle therefore collapsed to a
single chord, turning curved curtain-wall mullions straight while the
OpenCascade kernel (which meshes by deflection) kept them curved.
evaluate_conic now also enforces a deflection-based floor on the number
of segments, keeping the chord deviation within mesher-linear-deflection,
matching OpenCascade. Small circles are unchanged (CircleSegments floor
still dominates); only large-radius curves get denser.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
IfcConvert returned a success exit code even when geometry conversion logged
errors and silently dropped elements (for example a failed TopoDS::Shell build
under layerset slicing produced valid looking output with most objects
missing), so CI and scripts could not detect a partial conversion.
Add an opt-in --fail-on-error flag that makes IfcConvert exit non-zero when any
error was logged during processing, reusing the existing MaxSeverity based
failure check already used for --validate. The default exit behaviour is
unchanged, so pipelines that tolerate individual element failures are
unaffected.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A face whose inner boundary crosses the outer boundary (or another inner
boundary) is invalid per the schema. Open Cascade silently heals or drops
such a face, so the intended hole is lost or the face is corrupted with no
diagnostic at all (the 2018 report saw a dropped face; on the current line
the face survives as wrong geometry, still silently).
After the wires are collected, if a face has inner boundaries, measure the
BRepExtrema distance between each inner wire and every earlier wire. Two
non intersecting loops have strictly positive distance, so a distance at
or below the modelling precision means the boundaries touch or cross; emit
a warning (GEO 402) naming the offending face. This is diagnostic only, no
geometry change.
The message is emitted via the kernel logger() rather than Logger::Root():
IfcConvert configures a local Logger and worker logs merge into it, while
Logger::Root() is a separate unconfigured singleton whose messages are
discarded (a latent issue affecting some existing GEO messages too).
Verified on OCC 7.9.2 with synthesized IFC4 faces: an inner triangle
crossing the outer edge, and one straddling the bottom edge, each emit one
GEO 402; a valid 4x4 hole emits none and triangulates identically (area
84.0), in both sequential and multithreaded runs. Pure inner self
intersection and full containment are distinct classes and intentionally
left untouched.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Comply with AGENTS.md: new AI-generated files must carry a top-of-file
comment indicating AI assistance.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
In IFC2X3 IfcAsymmetricIShapeProfileDef is a subtype of
IfcIShapeProfileDef, so the IfcIShapeProfileDef mapping dispatched it by
inheritance. From IFC4 onwards it is a standalone subtype of
IfcParameterizedProfileDef, so nothing mapped it and the extruded solid
came out empty (GEO326, 0 verts).
Add a dedicated map_impl that builds the twelve-point asymmetric section
(independent bottom/top flange widths, thicknesses, fillet/edge radii and
flange slopes), plus a guarded BIND. Both are wrapped in
SCHEMA_IfcAsymmetricIShapeProfileDef_HAS_BottomFlangeWidth, which is only
defined where the type is standalone, so IFC2X3 keeps its existing
subtype route unchanged.
Verified on OCC 7.9.2: an IFC4 asymmetric extrusion goes from 0 verts to
a correct 72-vert solid (bottom flange wider than top); IFC2X3 output is
unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This allows a file that will be automatically picked up by Claude. It can either
be a copy of a CLAUDE.md, or a one line file pointing to a shared common file. i.e.
@~/.claude/conventions-ifcopenshell.md
Feature #5753 - Autosave for ifc files
Merging because it could be a life saver. It is hidden behind an option and is off by default.
- Provides the option have an autosave file created periodically (duration in prefs).
- Can be set to save immediately or a dialog prompt to save, but can be dismissed.
- Removes the autosave when Blender quits cleanly.
- If the autosave file exists at startup, it will prompt which file to load.
_Every_ AI had a hand in this, but I have reviewed, understood and tested it. AI Credits go to:
Cursor, Grok, Copilot, and Claude.
- autosave.py: black formatting (blank line) and ruff's
collections.abc.Callable import fix.
- project/__init__.py, tool/__init__.py: ruff import-sort fixes. The
autosave import in tool/__init__.py is deliberately kept last (must
come after tool.drawing, per its existing comment) via `# isort: skip`
rather than letting ruff move it, which would reintroduce that bug.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The recovery popup used invoke_popup, which is dismissed the instant
the mouse leaves its bounds - closing the prompt without loading
either file, and with no visible feedback that anything happened.
Switches to invoke_props_dialog, which blocks the rest of the UI and
is only dismissed by an explicit action. Since Blender always renders
both a fixed "Cancel" button and one labelled by confirm_text on that
dialog type, the prompt is reframed as a direct Yes/Cancel question
("Do you want to load the autosaved version instead?") instead of
adding separate Load Original/Load Autosave buttons on top of those.
Folds the load logic directly into the popup's execute()/cancel(), so
the now-redundant LoadAutosavedRecovery operator is removed.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Previously the autosaved copy was only ever overwritten, never removed,
so a deliberate quit (whether the user saved or chose "don't save")
still nagged with a recovery prompt on next startup.
Registers an atexit cleanup that removes the active IFC's autosave
file(s) on a graceful interpreter shutdown. atexit never runs on an
actual crash, so a genuine crash still leaves the recovery file in
place as before.
The cleanup reads a cached plain-string path kept up to date by
reset_timer(), rather than looking it up live via bpy.context - by
the time atexit fires, Blender's C++ side is torn down far enough
that even a read-only bpy.context.scene access aborts the process
(std::bad_optional_access) instead of raising a catchable exception.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Implemented as described in #5753, with two options:
- A nag dialog with save or cancel options.
- An autosaved file.
Settings are in preference to activate the feature (default: off), the period before prompting/saving,
and choosing between the two methods.
Prevent the autosave file being added to the recent files list when the user opens the original, but selects to open the autosaved version.
black/ruff
This commit was created using AI assistance. Cursor for the initial code, then Grok and I fixing all the errors
that Cursor made. Finally Copilot did a code review.
I have reviewed and tested the code, and I understand it, and it works and does not introduce any obvious bugs.
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Grok
Co-authored-by: Cursor
- add_stationing_referent.py: black reformat (new drift from v0.8.0).
- update_fallback_position.py: v0.8.0's changes to this file made the
ifcopenshell.util.unit import (added in an earlier commit here) unused;
removed per ruff.
poe ty's sequence only reaches ty-ios once ty-bonsai passes, so these
never surfaced until now:
- util/alignment.py: drop the stale `include_referent=False` kwarg from
add_zero_length_segment() - that parameter was removed from the function's
signature in 45ea5eb07 but this caller in a different file was missed,
leaving a latent TypeError if this code path is ever exercised.
- ifcopenshell_wrapper.pyi: add the optional trailing `logger` parameter to
parse_ifcxml/open/construct_iterator*, matching the real SWIG signatures
in src/ifcwrap/*.i (all declare `Logger& logger = Logger::Root()`) that
the hand-maintained stub never picked up.
- ifcopenshell/__init__.py: remove a stale `ty: ignore[unknown-argument]`
comment that ty confirms is no longer suppressing anything.
- assign_cost_item_quantity.py: OPERATORS mixes 2-arg binary operators with
the 1-arg `operator.neg` (for ast.USub), but FormulaEvaluator has no
visit_UnaryOp so USub can never reach this lookup via visit_BinOp.
Suppressed at the call site rather than touching the dict, since this
looks like scaffolding for unary-minus support rather than dead code.
- Explicit submodule imports (ifcopenshell.geom / api.alignment / util.unit
/ api.aggregate / api.context / api.spatial) added where accessed but
only reachable by accident of import order.
- tool.py: drop the `-> int` annotation on the Parametric interface's
get_geom_generation stub; its `pass` body implicitly returns None, which
ty can't reconcile with the runtime @interface/@abstractmethod rewriting
it never sees statically. Matches the file's other stubs (-> None).
- railing.py: qualify the "BIMRailingProperties" string annotations as
"prop.BIMRailingProperties" on the two functions using it, since the bare
name was never imported into this module's namespace.
- product.py: suppress ty's missing-argument errors on
copy_z_rotation_to_selected's Surveyor.get_z_rotation/set_z_rotation
calls with targeted ty: ignore comments. The function is unused and its
two dependencies were never implemented on the concrete Surveyor tool;
left as-is rather than deleted or implemented.
- gizmos.py: TYPE_CHECKING-guard `import bmesh` for the string-literal
annotation in build_schematic_mesh; suppress the still-unresolved
gizmo_textures import in TexturedQuadGizmoMixin (WIP dependency, not dead
code).
- model/__init__.py: register the `decorator` submodule, which unregister()
already calls (would have raised NameError on addon disable).
- mep.py / tool/model.py: add explicit imports for bonsai.core.geometry and
bonsai.core.model, previously only reachable by accident of import order.
- Test files: add explicit ifcopenshell.api.pset / ifcopenshell.util.element
submodule imports used but not imported.
The datamodel-v1.0 merge (cf05bbd1b) overwrote build_rocky.yml with a
version that switched python3 -> uv run but dropped the --shared flag that
a91b1da28 ("Reduce Rocky package size") had added. build_osx.yml kept it
(ddee88bed).
Without --shared, nix/build-all.py builds IfcOpenShell as static libs, so
each of the ~40 plug-in .so files (schemas x8, kernels, mappings,
serializers, writers) statically embeds a full copy of libIfcParse +
libIfcGeom. The data-model rewrite made those base libs much larger, so the
duplication ballooned the Linux packages (~2-3x). With --shared the plug-ins
dynamically reference the shared libIfcParse/libIfcGeom instead. Restores the
same size reduction macOS already has.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adds a 'Rename' action to the model (non-group) context menu, mirroring
renameGroup: prompts via QInputDialog, trims, and calls
setModelDisplayName + notifyFederationChanged. The Federation already emits
modelChanged, so the panel item text updates in place.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
With dbus-devel added, the connector's Rust code compiles fully and reaches
the final link, which fails: the bundled FLTK GUI toolkit needs the X11
extension, pango and cairo shared libs, plus libsupc++.a. ld reports every
unresolved -l at once, so this is the complete set:
-lXext -lXinerama -lXcursor -lXrender -lXfixes -lXft
-lpango-1.0 -lpangoxft-1.0 -lpangocairo-1.0 -lcairo -lsupc++
The X/pango/cairo -devel packages are in AppStream; libstdc++-static
(libsupc++.a) is in CRB, so enable it for the transaction. GitHub's ubuntu
runners ship all of this, which is why the dedicated connector workflow
never needed it.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
cargo build of bonsaiviewer-autodesk pulls dbus-secret-service (the Linux
OS-keyring backend for credential storage) -> dbus -> libdbus-sys, whose
build.rs needs dbus-1.pc via pkg-config. GitHub's ubuntu runners ship
libdbus-1-dev, so the dedicated connector workflow never needed it; the
minimal Rocky container doesn't. Add dbus-devel to both Rocky jobs
(pkg-config is already present). This was the last step after a fully
successful C++ build + cargo compile.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The community rockylinux/rockylinux:10 image omits PATH from its image
config, unlike the old Docker Official arm64v8/rockylinux:9. GitHub Actions
derives each run step's PATH from that config, so with no PATH the shell
exec (docker exec ... sh -e {0}) fails with exit 127, 'exec: sh: not found'
— it broke before any build logic ran. Restore a standard PATH via the
container env; the runner still layers GITHUB_PATH additions (uv, cargo)
on top.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Two gaps left when BonsaiViewer and its Rust connector were newly added to
the Rocky CI jobs (May–Jun), neither previously exercised there:
1. Rust: the autodesk connector was rewritten from a PyInstaller Python
app to a Rust crate, so packaging/build.py now runs 'cargo build
--release'. Neither Rocky workflow installed a toolchain. Add rustup
(stable, matching the dedicated dtolnay/rust-toolchain@stable workflow)
to both x86 and ARM.
2. ARM glibc: aqt's official Qt6 ARM binaries link glibc 2.38, which Rocky
9 (glibc 2.34) can't load — moc fails, breaking IfcViewer_autogen. Move
the ARM job to Rocky 10 (glibc 2.39). The legacy arm64v8/rockylinux
image stopped at 9, so use rockylinux/rockylinux:10 (multi-arch, has
arm64). Rocky 10 defaults to Python 3.12 and drops python3.11, so the
script-runner references move python3.11 -> python3 (system Python only
runs helper scripts; ifcopenshell is built against uv's Python). Bump
the ccache key to rockylinux10. x86 stays on Rocky 9 to keep its lower
glibc floor for end users.
The rockylinux9-arm64 build-outputs deps branch is kept as-is: Rocky 9
deps are forward-compatible on Rocky 10, and no rocky10 branch exists yet.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
createWgpuSurface() calls wgpu_macos_attach_metal_layer() in the Q_OS_MAC
branch at the top of the file, but the only #include of MetalSurface_mac.h
sat ~450 lines below the call site, so macOS builds failed with 'use of
undeclared identifier'. The header self-guards on __APPLE__, so move the
include up into the early platform block next to <Windows.h>; the lone
call site is the sole consumer, so the late include was dead.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The data-model branch's spf_header::set_file_description/name/schema take a
const shared_pointer_type& (an internal instance_data* storage handle). SWIG
wraps them and emits the alias unqualified into the global-scope wrapper,
which MSVC rejects (C2065 'shared_pointer_type': undeclared identifier). The
matching getters are already %ignore'd and re-exposed via %extend; the raw
setters are not a usable Python API, so ignore them the same way.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Iterator::next() now returns express::Base (data-model branch). Comparing
it against 0 is ambiguous: 0 converts to Base via the pointer ctor while
Base converts to int via operator bool, so both operator!=(int,int) and
Base::operator!= are candidates. Use an explicit truthiness test — an
empty Base signals end-of-iteration.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
parse_num_ used std::from_chars for both integers and doubles, but the
floating-point from_chars overload is =deleted in Apple clang's libc++, so
the macOS build failed to compile (parse.cpp:136, instantiated for double).
Split parse_num_ with `if constexpr`: integers keep std::from_chars
everywhere; on macOS, doubles parse via strtod_l with a cached "C" locale
(locale-independent, restoring the pre-charconv Apple path). libstdc++ and
the MSVC STL have working float from_chars and are left unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
build-all.py's install_qt6 runs `sys.executable -m aqt`, but the build now
runs under `uv run`, whose isolated env never got aqtinstall — it was pip
installed into the system Python. `uv run --with typing_extensions --with
aqtinstall` puts them where the script actually executes.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A .rdbview is a zip of model.rdb/ (the lossy IFC data DB — the rdb
serializer skips IfcRepresentationItem) + model.ifcview (baked geometry).
The viewer could produce them but not open them.
- extractRdbview(): unzip a .rdbview (QZipReader) into a session temp dir
keyed by a hash of path+mtime+size (reused on re-open), returning the
extracted model.rdb. The producer's layout means sidecarPath(model.rdb)
resolves the sibling model.ifcview automatically, so it then loads exactly
like any pure .rdb: geometry from the sidecar, data from the .rdb via
ifcopenshell::file(FT_AUTODETECT). No SceneLoader/engine changes.
- detail::loadModels() resolves each source path through it before
queueModels (both fresh-open and project reload go through here), so the
Federation persists the .rdbview while the loader gets the extracted .rdb.
- cleanupRdbviewCache() clears stale extractions at startup.
- .rdbview is offered under "Add Geometry" (the file picker; "Add IFC
Database" is a directory picker), not "Add IFC File" — it's a lossy viewer
bundle, not a source IFC.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Opening a fresh .ifc streams geometry to the GPU, then bakes the .ifcview
cache. That bake — reorder + per-chunk zstd (level 19) — ran synchronously
in SceneLoader::onStreamerFinished, which is a QueuedConnection slot on the
main thread, so it froze the UI right as the progress bar hit 100% (≈15s of
zstd for a 130 MB-geometry model).
- Move the compress + writeSidecar onto a background thread. The geometry is
already resident and the sidecar is only a cache for the next open, so the
viewport is interactive the instant streaming finishes; the write is joined
before the next write and in the destructor.
- Parallelise the per-chunk zstd across hardware_concurrency threads (compress
all chunks, then write serially to keep contiguous offsets) so the
background write also finishes quickly.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The "Backface Culling" checkbox persisted a value and reflected it, but
nothing consumed AppSettings::backfaceCulling — the opaque pipeline
hardcoded cullMode = Back, so toggling had no effect.
Build a second opaque pipeline (cullMode None) alongside the culled one and
pick between them per-frame from a backface_culling_ flag; setBackfaceCulling
flips the flag and requests a redraw (no rebuild). ViewportWindow forwards
it, and MainWindow applies the persisted value at startup and re-applies on
change — same wiring as the nav preset.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Restructure the web viewer so the wasm is a reusable module and add a
second example that drives it from ordinary page DOM.
Build:
- Emit IfcViewerWeb.js (a `createIfcViewer` factory, MODULARIZE) + .wasm
instead of a single baked page (dropped --shell-file); copy the static
example pages next to it at build time.
- Unbreak the web build: CameraMath.h / ViewportCore.cpp used
boost::math::constants::pi just for pi, pulling all of boost/math into a
header shared with the Emscripten build (no Boost in its sysroot). Replace
with a constexpr kPiF — identical value, no dependency, desktop unaffected.
JS integration (web/ifcviewer.js):
- A small helper wraps the factory: boots the viewer on a canvas, runs the
RAF loop from onRuntimeInitialized (NOT a post-await .then, which stalls
Dawn-web's device callback and leaves the device half-initialised), and
exposes addFile/addUrl, clearScene, model list/progress, and onSelect(...).
- ViewportCore/main_web emit each pick to JS via Module.__ifcvOnSelect
(object id + IFC GlobalId + model index; empty on deselect); onSelect also
dispatches an 'ifcviewer:select' DOM event.
- Fix input coords for a non-fullscreen canvas: mousemove/mouseup are
window-targeted, so convert their coords to canvas-relative via the canvas
client-rect origin (marquee + box-pick were offset when embedded).
Examples:
- IfcViewerWeb.html: the fullscreen viewer (same DOM/behaviour as before,
now loading the module) — the Playwright smoke suite still targets it.
- embedded.html: a sized viewer with DOM outside it to add models (file or
URL), list loaded models with streaming progress, and show the model +
GlobalId of the clicked object. Starts empty (drops the wasm's embedded
sample, which the fullscreen page/tests still use).
- index.html links both.
Federation note: the web viewer already streams multiple models into one
scene (a byte-source per file/URL); it doesn't need the desktop Federation
document for this. Verified: 11/11 web smoke tests pass; embedded example
loads models, reports the picked model + GUID, and the marquee aligns.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Theme QInputDialog (the New Group / Rename Group popup) so it follows
the dark theme instead of rendering light.
- Theme the generic QTabBar that QMainWindow creates for tabbed docks
(previously bright white). The app's own #appTabBar keeps its look via
more specific selectors.
- Reduce QHeaderView::section vertical padding (7px -> 4px) so table/tree
header rows match the body row height throughout the UI.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
initWgpu() runs on the first exposeEvent, after MainWindow has already
applied the nav preset saved in Settings. It then unconditionally
re-applied "blender" whenever WGPU_NAV_PRESET was unset, silently
overriding the user's saved choice — so the applied navigation didn't
match what Settings showed.
Only apply the preset from WGPU_NAV_PRESET when that env override is
actually set; otherwise leave the current preset (MainWindow's persisted
choice, or the blender default). The startup log now reports the effective
orbit/pan bindings rather than a hardcoded name.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Previously Del always removed the most recently added section plane. Now a
plane can be picked and deleted individually:
- ViewportCore tracks a selected plane index, kept valid as planes are
added (the new one becomes selected), removed, or cleared.
- Clicking a gizmo with the section tool active selects that plane.
- The section gizmo geometry is baked white and coloured via its per-plane
tint, so the selected plane draws in a bright amber highlight while the
rest stay red (unchanged look).
- Del/Backspace removes the selected plane, falling back to the most recent
one when nothing is selected.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The filter field now actually filters. Typing shows only the sets whose
name matches, or that contain a matching property name/value — and when
it's a property/value match, only the matching rows are kept (neighbouring
rows are dropped). Matching is case-insensitive.
Set widgets live in a per-section container that's rebuilt from the raw
data on each keystroke, so filtering never recreates the filter field (its
focus and cursor are preserved). Placeholder reads "No properties/
quantities" with no data, "No matching properties/quantities" when the
filter excludes everything.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Convert bare Mock() to Mock(spec=bpy.types.Object) for Blender-object
stand-ins in TestRecalculateWallsWithNewConnections, TestMEPActionGuards,
and TestRecreateAggregateIteratesAllNew so typos on the Blender API
fail loudly instead of silently returning a MagicMock.
Replace the ad-hoc Mock() ConnectionRecord stand-in in
TestRecreateConnectionsZipsPairs with a real ConnectionRecord instance,
which pins field names at construction and catches drift if the
dataclass fields ever get renamed.
IFC entity mocks remain bare Mock() intentionally: entity_instance
attributes are schema-driven at runtime rather than defined statically
on the class, so spec= would refuse the .GlobalId / .HasFillings /
.ConnectedTo attribute writes the tests need.
Relates to #8088.
Generated with the assistance of an AI coding tool.
Rewrite tool.Array.select_only_parent as a thin call to
tool.Blender.select_and_activate_single_object; drop the ad-hoc
per-child deselect loop and the unused parent_element parameter.
Replace the tail parent_obj.select_set(True) in _regenerate_array_body
with tool.Blender.set_object_selection, which wraps select_set in the
hidden-object try/except the utility already owns.
Relates to #8088.
Generated with the assistance of an AI coding tool.
Add tool.Array.is_array_child helper. Port decorator and MEP
action gizmos (lock, pen, join) hide on array children — writes
on children get wiped by the next regen, and the port topology
is inherited from the parent.
Introduce tool.Array.select_only_parent and wire it into both
bim.regenerate_array and bim.finish_editing_array so post-regen
state converges on parent-only-selected + active. Grow and shrink
paths otherwise diverge (grow left new children selected alongside
the parent; shrink left only the parent).
Relates to #8088.
Generated with the assistance of an AI coding tool.
Sweep [0]-indexing in recreate_aggregate, recreate_connections,
and recreate_port_connections so batched N-child duplicates
recreate relationships on every new child, not just the first.
Single-source callers unaffected (loop collapses to one iteration
on 1-element lists).
Relates to #8088.
Generated with the assistance of an AI coding tool.
Replace N sequential duplicate_ifc_objects([parent]) calls in
_regenerate_array_body with one duplicate_ifc_object_n_times call
per layer, batching the fixed per-call overhead (snapshot gather,
UI refresh, decorator reload).
Guard batch_host_recut drain against dead StructRNA refs and prune
orphan array-child GUIDs at regen so outliner-delete of a
Bonsai-managed child cannot crash subsequent regenerate_array.
Recalculate walls after recreate_connections so Shift+D of
connected walls produces correct junction geometry without a
manual regen step.
Relates to #8088.
Generated with the assistance of an AI coding tool.
- Replace the mock IfcWall placeholder with a "No item selected" empty
state; the panel only fills in class/attributes/relationships/psets from
a resolved object, and safely stays empty otherwise.
- Show "No properties" / "No quantities" placeholders (muted, themed via
secondary_text) when those sets are empty.
- Drop the base application font 10pt -> 9pt to fit more data. Panel titles
keep their own explicit size and are unaffected.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- helpers/placement: port get_storey_elevation (placement Z, falling back
to the Elevation attribute), matching ifcopenshell.util.placement.
- Add a secondary column: the storey elevation for IfcBuildingStorey,
otherwise the LongName when filled. Elevations are right-aligned.
- Columns: Name is drag-resizable (interactive) and defaults to 20% of the
width, Long Name stretches to fill the rest, and the eye is pinned to the
right at a fixed width. Header shown so the divider can be grabbed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The streaming settle burst re-armed the render loop whenever a
non-resident chunk was frustum-visible, but the enqueue only fetches
chunks that are contribution-visible (big enough on screen) and not in a
blocked cooldown. A chunk that is in the frustum but sub-pixel is never
loaded, so visible_pending stayed true forever and the loop spun at full
frame rate with no input.
Match visible_pending to the enqueue's eligibility test: a non-resident
chunk keeps the loop alive only if it's actively loading, or is
contribution-visible and past its cooldown. Sub-pixel / cooldown-blocked
chunks no longer prevent idle; they still stream in when a camera move or
eviction requests a frame.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
IfcSurfaceFeature (e.g. road markings) adheres to a host element through
IfcRelAdheresToElement, a [1:1] cardinality hierarchical relationship in the
same family as aggregation, containment and nesting since IFC4.3. The spatial
traversal never followed it, so surface features had no resolvable parent or
container: on import they landed in the Unsorted collection instead of the
host's spatial collection, and were dropped entirely in DECOMPOSITION filter
mode.
Add get_adhered_element (feature to host) to the get_parent resolver chain and
walk HasSurfaceFeatures in get_decomposition, plus a get_surface_features helper
mirroring get_parts/get_contained. With get_parent resolving adherence,
get_container now returns the host's spatial container, so tool.Collector places
surface features under the host. Also follow HasSurfaceFeatures in the Bonsai
DECOMPOSITION filter path so they load in that mode.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Build the spatial hierarchy panel from the loaded model's real IFC
spatial structure instead of mock data:
- helpers/element: add get_spatial_children (IsDecomposedBy -> RelatedObjects,
filtered to spatial elements) to walk IfcProject -> IfcSite -> IfcBuilding
-> IfcBuildingStorey -> IfcSpace.
- SessionState: relay dataSourceReady as modelDataSourceReady (the .ifc for a
sidecar hit loads asynchronously, so the tree can only build once it arrives).
- spatial_hierarchy/View: walk the active model's IFC file into a TreeNode
tree, naming nodes by Name (fallback to class), mapping site/building/storey
kinds; siblings sorted with natural (numeric) collation.
- spatial_hierarchy/Panel: tree now fills the panel height (setBodyExpanding +
Expanding size policy); right-click menu for recursive Expand/Collapse
Subtree and Expand/Collapse All.
Add the concept of an active model:
- SessionState: activeModelId / setActiveModelId / activeModelChanged; the
first loaded model is active by default; reassigns/clears on removal.
- Models panel: clicking a model makes it active; its cube icon is drawn with
the accent colour (makeAccentSvgIcon) via FederationItemModel::setActiveModelId.
- The spatial hierarchy reflects only the active model.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Populate the Properties and Quantities sections from the pset helper:
get_psets(psets_only) for Pset_*, get_psets(qtos_only) for Qto_* /
BaseQuantities, inheriting occurrence-over-type values. A toPropertySets
converter drops the internal "id" key and non-scalar values, formats
scalars for single-line cells, and skips empty sets. Placeholders are
cleared once a project is loaded.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add helpers to src/helpers/element: get_scalar_attributes (primitive
EXPRESS attributes only — entity refs / aggregates omitted), get_type
and get_container (ports of ifcopenshell.util.element), and a public
get_string_attribute for safe by-name reads.
Wire them into the properties panel:
- Attributes section shows the element's direct primitive attributes for
live entities, or cached GlobalId / Name for geometry-only elements.
- Relationships section shows the construction Type and spatial Container
by name (falling back to the class when unnamed).
- Placeholders are cleared once a project is loaded, so no mock data leaks.
Also: a deselect (click on empty space -> object_id 0) no longer resets
the panel; it keeps showing the last active object. Project reset/open
still clear it.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add src/helpers/element.{h,cpp}, a schema-dispatched C++ port of
ifcopenshell.util.element.get_predefined_type: prefers the associated
type element's predefined type (IsTypedBy / IsDefinedBy), falls back to
ElementType / ProcessType when USERDEFINED, then the occurrence's own
PredefinedType / ObjectType. Attribute reads are by-name so they work
across the IfcElement / IfcType* subtypes that carry these attributes.
Wire it into the properties panel entity summary: live IFC entities show
their real predefined type; geometry-only elements (a .ifcview loaded
without its .ifc/.rdb) show "N/A". Clears the placeholder so a stale
predefined type no longer leaks once a project is loaded.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Rename the two overloaded model identifiers and make object_id
assignment single-authority, fixing a pick -> properties mismatch.
Identifiers:
- Per-model UUID fed_id -> model_id; the uint32 runtime handle
model_id -> session_model_id (SessionState accessors + mirror hashes
renamed to match). "fed_id" was a misnomer -- the federation is the
whole collection, not one model.
object_id assignment (fixes wrong class on click):
- Producers (GeometryStreamer, .ifcview sidecar) now stamp model-LOCAL
object_ids; ViewportCore::applyCachedModel is the sole authority that
assigns the session-global id (base + local). Removed
SceneLoader::next_object_id_, GeometryStreamer::lastObjectId(), and the
streamer's start_object_id parameter.
- The element table is stamped by the same base on both load paths
(applySidecarData and onStreamerFinished), so registry ids match the
ids pick returns. Previously the sidecar path double-rebased instances
vs the registry (click IfcSite -> showed IfcDoor); the live-stream path
had the same latent mismatch. Both closed.
Naming / cleanup:
- SceneLoader::addFiles -> queueModels; startStreamLoadFor ->
loadFromGeometryStreamer; readSidecarMetadataOnly -> readSidecarMetadata.
- Federation::addModel takes an explicit display_name (no QFileInfo
fallback); callers pass QFileInfo(path).fileName().
- Disambiguate cryptic short locals (d->sidecar, m->model, c->chunk, ...)
in SceneLoader, Federation, ViewportWindow, AreaMeasurement,
SectionGizmoRenderer, and the SidecarData/SidecarReadPlan spots in
ViewportCore.
Tests: 125/125 pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
file.write() streamed directly onto the target path, so a crash mid-write
left a truncated file with dangling STEP references. Serialize to a temp
file in the same directory, then atomically rename it onto the target.
- New IfcUtil::path::atomic_rename_file: std::rename on POSIX, MoveFileExW
with MOVEFILE_REPLACE_EXISTING on Windows. Unlike rename_file it never
unlinks the destination first, so there is no window where it goes missing.
- Fully in C++/swig (per aothms), so the FILE_NAME header is untouched: it
comes from the model header, not the output path (verified empirically).
- Temp lives next to the target so the rename stays on one filesystem.
- Stream is closed before the rename (Windows cannot move an open file).
- On any write error the temp is removed and the original target is intact.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
IfcConvert --version reported 0.8.0 on a plain source build even though the
VERSION file says 0.8.6 (#8164). buildinfo.cpp already falls back to the
IFCOPENSHELL_VERSION_STRING macro and CMake already passes it as
${RELEASE_VERSION}, but RELEASE_VERSION was only read from the VERSION file
when VERSION_OVERRIDE was on. A default build (VERSION_OVERRIDE off,
ADD_COMMIT_SHA off, as the nixpkgs package builds it) fell through to the
hardcoded "0.8.0", so the fallback macro carried the stale value.
Read the VERSION file unconditionally so RELEASE_VERSION is always the real
version. VERSION_OVERRIDE still governs the branch name embedded when
ADD_COMMIT_SHA is on, and project()/CPack now also reflect the true version.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
roughnessFactor was computed as 1/specularity. An IfcSpecularExponent of
0 produced infinity, which nlohmann::json serialises as null and makes
the glTF invalid; exponents below 1 produced values above 1, which glTF
also forbids. Map exponents <= 1 to full roughness and keep 1/exponent
above that, so the factor always lands in [0, 1].
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The georeferencing fix (e6dc582) iterates IfcProject.RepresentationContexts
unconditionally, but the attribute is OPTIONAL and None on projects without
contexts, crashing every extraction on such files with
TypeError: 'NoneType' object is not iterable.
Also extend the #8199 regression test to assert element placements are
copied verbatim, so extraction can never bake map coordinates into local
placements.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
IfcGeometricRepresentationContext.Precision is typed as a plain IfcReal
but is interpreted in the project length unit, so the IfcLengthMeasure
traversal in convert_file_length_units never touched it. A model
converted from mm to m kept a Precision of e.g. 0.01 (fine in mm, huge
in m), which breaks downstream geometry interpretation such as
IfcConvert boolean cleanup.
Subcontexts derive Precision from their parent, so only root contexts
are scaled.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
get_info_2 advertises the same signature as get_info but raised a bare
AssertionError for anything the C++ fast path does not implement --
including its own default arguments (recursive=False).
Use the fast path when recursive=True, return_type=dict and ignore=()
hold, and delegate to the pure Python get_info otherwise. As noted in
the issue, without recursion there is no meaningful performance gain to
lose by delegating.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
std::to_chars is locale-independent, so the ostringstream and imbue(locale)
are no longer needed. Build the REAL string with plain std::string operations.
Output is unchanged (verified in standalone compile: same shortest values, all
round-trip). Addresses review feedback on #8309.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
format_double formatted doubles with setprecision(max_digits10) (17 digits),
which padded clean values with noise: 0.0174532925199433 was rewritten as
0.017453292519943299 and 1.E-05 as 1.0000000000000001E-05. Every REAL in a file
changed on save, producing enormous diffs for anyone version-controlling IFC.
Use std::to_chars, which emits the shortest string that round-trips exactly
(like Python's repr), then keep the existing mantissa/exponent formatting.
Verified in a standalone compile of the exact function logic: the reporter's
values become 0.0174532925199433 and 1.E-05, 0.1 stays 0.1, and every tested
value (including a denormal) round-trips back to the identical double.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
check_aggregate_of_type used an exact type comparison (element->ob_type ==
type_obj), so a numpy array was rejected because its elements are numpy scalars
(numpy.float64) rather than direct float instances. For the numeric types,
accept subclasses: PyFloat_Check for double (numpy.float64 subclasses float) and
PyLong_Check (excluding bool) for int. The SPF REAL vs INTEGER distinction is
kept, so a float is not accepted where an int is expected and vice versa.
This replaces the earlier Python-side walk() approach, which the maintainer
preferred not to take since walk() is removed in v0.9. Verified with a runtime
red-green (built as a shared lib, called via ctypes): the old check rejects
np.array([3.0, 4.0]) and the new one accepts it, plain lists still work, an int
list is still rejected where a REAL is expected, and bool is rejected for INTEGER.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
asStringRef removes the first and last characters of a string, enumeration
or binary token to drop the delimiters, guarded only by !str.empty(). A
malformed single-character token (e.g. a bare '.' left when a fuzzer turns
'.PHYSICAL.' into '.)HYSICAL.') has length 1, so the first erase empties the
string and the second erase(str.begin()) runs on an empty string. That is
undefined behaviour: benign on a normal build, but it aborts (or throws
std::length_error from a later append) under a hardened libstdc++ with
_GLIBCXX_ASSERTIONS, which is why this file only segfaulted on the Fedora
build. Require at least two characters before stripping.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
When ADD_COMMIT_SHA is off (the default for release tarballs), buildinfo.cpp
fell back to a hardcoded "0.8.0", so a 0.8.5/0.8.6 build reported 0.8.0 from
IfcConvert --version and in written file headers. Pass CMake's RELEASE_VERSION
(read from the VERSION file) to IfcParse as IFCOPENSHELL_VERSION_STRING and use
it as the fallback, mirroring how the branch/commit defines are handled. The
commit-sha build and the last-resort literal are unchanged.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
generate_linework() loops over the main file plus any linked models
(added in 5db955d4), reassigning drawing_elements each iteration. The
SHAPELY fill pass builds elements_with_faces/raycast_objs from
drawing_elements, but did so once *after* that loop finished, so it
only ever saw whichever file was processed last -- silently dropping
surface fills for every other file, including the main model whenever
any link was loaded.
Accumulate elements_with_faces/raycast_objs across every file inside
the loop instead of capturing drawing_elements once after it ends.
Generated with the assistance of an AI coding tool.
The underlying bug (has_material_styles bypassing the tool layer) was
already fixed by e76455913, which added the required mock expectation
here, but left the test under its quarantine placeholder name
test_AAAAAAAAAAAA. Restore the real name now that it genuinely passes.
test_memusage_partial_open was silently skipped in CI (psutil was
never installed there). Add psutil so it actually runs, and run the
RSS measurement in a subprocess so the fixture file isn't already in
the page cache from earlier tests, which was making both deltas read
as zero.
Generated with the assistance of an AI coding tool.
mathutils only ships pre-built wheels for Python 3.13+ (verified
against PyPI's file list); on CI's Python 3.11, `pip install
mathutils` falls back to a slow/unreliable source build. Skip the
mathutils-dependent tests when the interpreter is too old instead.
These packages already have their own pytest suites (ifc5d, ifcedit,
ifcmcp, ifcquery) but were never run in CI, so regressions in them
went unnoticed. Add path triggers and test steps for all four, plus
odfpy and xlsxwriter which ifc5d's spreadsheet export tests need and
mcp which ifcmcp's server tests need.
Generated with the assistance of an AI coding tool.
ItemIsASum and Quantities are exporter columns that were missing from
MAIN_CSV_HEADER_COLUMNS, causing them to be misidentified as numeric cost
value categories on re-import. Also initialise rate_cost_schedule to None
before the search loop to avoid UnboundLocalError when no match is found.
Generated with the assistance of an AI coding tool.
is_drawing_active() required an open VIEW_3D area purely as a poll()
gate for bim.create_drawing, even though SVG generation is
ifcopenshell.geom-based with no viewport dependency; skip that check
when bpy.app.background is true, since a viewport is neither
obtainable nor meaningful there. Interactive behaviour is unchanged.
Generated with the assistance of an AI coding tool.
Linked IFC files were included in SVG output but without their
world transform, causing geometry to appear at wrong coordinates.
Falls back to no transform if the link cache is unavailable.
Generated with the assistance of an AI coding tool.
Link empty handles were missing from visible_objects, so linked
models were always hidden when activating a drawing.
Generated with the assistance of an AI coding tool.
Rewrite the BonsaiViewer viewport architecture page to match the current
renderer (updated type names, streaming/sidecar flow). Add a dedicated
.ifcview sidecar format reference page and link it from the ifcopenshell
formats toctree, and polish the Bonsai intro copy.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Rename the streamer/sidecar transfer and record types to describe what
they are rather than how they move:
MeshChunk -> StreamedMesh
InstanceChunk -> StreamedInstance
InstanceCpu -> InstanceInfo
PackedElementInfo -> ElementTableRecord
uploadMeshChunk -> uploadStreamedMesh
uploadInstanceChunk -> uploadStreamedInstance
buildMeshChunk -> buildStreamedMesh
and the two post-index sidecar metadata blocks:
"critical" metadata -> "geometry" metadata (meshes/instances/georef/TOC)
"deferred" metadata -> "element" metadata (elements + string table)
parseSidecarCritical -> parseSidecarGeometryMetadata
parseSidecarDeferred -> parseSidecarElementMetadata
The one behavioural change: the element hierarchy (parent_id) was
carried through ElementInfo, ElementTableRecord, and the sidecar element
table but never consumed, so drop it and bump SIDECAR_VERSION 16 -> 17.
No back-compat: regenerate sidecars. sample.ifcview is regenerated at v17.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Full section tool for the web viewport, with the gizmo + interaction shared with
desktop from one codebase.
- True-face surface pick. pickSurfaceAt had always ray-cast the instance AABB (to
skip a depth readback), so cuts sat in front of the real surface. The pick
fragment already computes the exact world_pos (it clips sections with it); now
it OUTPUTS it to a 3rd pick MRT (RGBA32F) that every pick path renders, and
pickSurfaceAt / pickSurfaceAtAsync read it back (decodeMappedPickPosition;
ray-AABB kept only as a fallback). The web async pick chains id -> normal ->
position spontaneous staging maps.
- Web tool: LMB drops a cut at the picked surface (LMB drag still orbits), K
toggles, Shift+K clears; oriented to the real MRT surface normal. Exports + a
Section / Clear cuts toolbar pair.
- Shared gizmo: lifted the section-gizmo renderer (SECTION_WGSL + thick-line AA +
quad+arrow VBO + pack + screen-space hit-test) out of the Qt-coupled
OverlayRenderer into a Qt-free SectionGizmoRenderer that ViewportCore::render
draws for BOTH desktop and web (both already render via render()). One identical
gizmo; OverlayRenderer's now-dead section code removed. Fixed 1 m size (matches
the desktop constant).
- Interaction (shared): hitTestSectionGizmo (SectionGizmoRenderer::hitTest) +
beginSectionDrag / updateSectionDrag / endSectionDrag live in ViewportCore.
Drag a gizmo arrow to slide the plane along its normal; Del/Backspace removes
the most recent cut. Desktop's ViewportWindow dropped its duplicate hit-test /
drag math + state and delegates to the core; web wires the same calls.
Tests: sectionPlaneCount add/clear/cap (Catch2, 125); web smoke "click a surface
cuts geometry, clear restores" exercises the shared gizmo + 3-MRT pick (11/11).
Desktop object-pick / marquee unaffected; BonsaiViewer builds.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add IfcOpenShell-Python and IfcViewer test-running documentation, including desktop CTest targets and web Playwright smoke tests. Move the web test README content into the Sphinx docs.\n\nGenerated with the assistance of an AI coding tool.
Update Bonsai viewer headers to identify Bonsai and GPL licensing, add Bonsai Viewer documentation, and document debug output capture.
Generated with the assistance of an AI coding tool.
Bring rubber-band box-select to the web on the Web preset's select button (RMB).
- Core: factor the pick-pass encode + rect copy out of picksInRect into
encodeBoxPickToStaging (mirroring how single-pick shares
encodePickReadbackToStaging), shared by the sync picksInRect (desktop) and a
new async picksInRectAsync (web) — the latter maps the staging buffer via a
spontaneous callback because the sync spin-map hangs the JS loop. New
applyMarqueeToSelection (plain replace / Shift add / Ctrl remove).
- Web main_web: a select-button drag past the click threshold draws a marquee
rubber-band (a plain DOM <div> positioned in CSS px — no GPU overlay pass,
which the web lib lacks) and on release box-picks the rect (device px) and
applies it to the selection. A click (no drag) still single-picks.
- Web shell.html: the #marquee div + styling, and — the reported bug — a
contextmenu preventDefault on the canvas so RMB (now the select button) doesn't
pop the browser menu. (Firefox still forces its native menu on Shift+RightClick;
that's a browser escape hatch pages can't override.)
Tests: applyMarqueeToSelection replace/add/remove + id-0 (Catch2, 124 total);
web smoke marquee drag → rubber-band shown → selection changes → hidden (10/10).
Desktop picksInRect unchanged in behaviour; BonsaiViewer builds.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Drop the obsolete diagnostic macOS WGPU workflow that referenced removed standalone viewer paths and targets.\n\nGenerated with the assistance of an AI coding tool.
Rename short local variables and parameters in the viewer loading, sidecar, and BonsaiViewer command paths to make their responsibilities clearer.\n\nGenerated with the assistance of an AI coding tool.
Make orbit/pan/select mouse bindings pure data owned by ViewportCore so both
hosts and every preset share one source of truth, and add a "Web" preset. This
rounds out the matrix: the desktop gains a web-style scheme and the web inherits
all presets, with no per-platform hardcoding.
- Core: NavBindings { orbit, pan, select button + modifier } + setNavPreset
("blender" default | "rhino" | "revit" | "web") + navBindings(). Select is
preset-driven too (was hardcoded LMB) so "web" moves it to RMB. web = orbit
LMB, pan MMB, select RMB (LMB drag orbits with no click/drag ambiguity; RMB
click-selects / drag-marquees). NavMod uses "Plain" not "None" (X11 #defines
None to 0L).
- Desktop ViewportWindow: applyNavPreset sources the core table (mapped to Qt);
marquee-arm / single-pick dispatch keys off select_button_. Default stays
blender → no behaviour change.
- Desktop config: AppSettings::NavPreset gains Web + navPresetName(); the
Settings dialog lists it. This also FIXES a pre-existing gap — the preset combo
was persisted but never applied (only WGPU_NAV_PRESET env worked). MainWindow
now applies the persisted preset at startup (env override still wins) and live
on navPresetChanged, so all four presets actually work from the dialog.
- Web main_web: classifyPress routes the pressed button through navBindings()
(orbit/pan/select), defaulting to the "web" preset; context menu already
suppressed so RMB is free.
Tests: setNavPreset table (Catch2, 123 total); web smoke select tests use RMB.
BonsaiViewer builds; 9/9 web smoke.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
bpy.ops.bim.reload_link(link_index=...) from a script skips invoke(),
so self.query stayed at its empty default and execute() overwrote the
link's stored query, reloading everything. Only update link.query when
the property was explicitly set (dialog or script argument), and reload
using the stored query.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Store the selector query used at Link IFC time on the Link
PropertyGroup, restore it from the sidecar cache JSON on host
IFC reopen, and forward it through LoadLink and ReloadLink so
subsequent reloads replay the original filter instead of loading
every element. ReloadLink now opens a small dialog pre-populated
with the current query, allowing the user to edit it in place
without unlink-and-relink.
Also swap TestCalculateLinkMatrix off NamedTemporaryFile(delete=True)
which held an exclusive Windows handle and blocked the
code-under-test from reopening the sidecar path.
Closes#8219
Generated with the assistance of an AI coding tool.
Fold two related cleanups from post-PR review into one commit:
Shared filling predicate — the gizmo poll and AddOpening._add_openings
both need to decide whether an IFC entity is a Bonsai-supported filling
(IfcDoor / IfcWindow, the classes the opening generator can derive
geometry from). Centralise the check in bim.module.model.opening as
is_filling_supported so a schema-broadening tomorrow only edits one
predicate. The gizmo's own predicate is renamed
is_supported_filling_or_opening to reflect its wider domain (also
accepts None for raw meshes and IfcOpeningElement for reassignment).
Aggregate-host guard — regenerate_filling_opening_body returns the
voided host Blender object so callers can recut it. Aggregates have
no mesh data; returning them made callers hit switch_representation
against a None data-block. Guard on voided_obj.data is None and
return None so callers can skip cleanly.
Adds a direct position_gizmos test asserting host-at-index-1 (filling
active) still anchors on the slab — pins the class-based dispatch's
selection-order independence.
Generated with the assistance of an AI coding tool.
Both files interleaved bpy.types imports with ifcopenshell.util
imports, which ruff's I001 rejects for standard-library / third-party
ordering. Running ruff check --fix on the two files reorders them into
the isort-canonical shape with no behaviour change.
Generated with the assistance of an AI coding tool.
EditOpenings.edit_openings unconditionally walked sibling wall sets
twice on every processed opening — once by mapped source id via
get_similar_openings_building_objs, once by filling type via
get_all_building_objects_of_similar_openings — and unioned both into
the building_objs recut set. reload_body_representation then hit every
one of those walls with a switch_representation call, even for the
show/hide toggle path where nothing about the opening changed.
Move both sibling-wall unions inside the is_edited / is_moved branch.
Pure show/hide (no shape edit, no move) now touches only the wall(s)
directly hosting the toggled openings. The edit and move paths still
refresh siblings the same as before, since a mapped-source rewrite
propagates the new shape to every sharing wall and each one needs a
recut.
Generated with the assistance of an AI coding tool.
recalculate_walls commits the wall's own placement to IFC before
recreating its geometry but did not touch its fillings. A door moved
along the wall's reference line therefore stayed cut at its old
position when the user pressed SHIFT+G on the wall, because the wall
recut ran against the still-stale opening placement in IFC.
Walk each wall's HasOpenings and, for every filling whose Blender
matrix_world differs from its committed IFC placement (tool.Ifc.is_moved),
commit the filling's placement and propagate the new matrix to the
enclosing opening via ifcopenshell.api.geometry.edit_object_placement.
The subsequent recreate_wall pass then sees the fresh opening positions
and cuts at the right spot.
Generated with the assistance of an AI coding tool.
Commit 82dd1d94d switched RecalculateFill from
bonsai.core.geometry.switch_representation to the surgical
tool.Geometry.recut_host to speed up batched host recuts. The trade-off
was intentional for that scope but dropped the implicit opening body
refresh that switch_representation used to provide: SHIFT+G on a door
whose parametric dimensions had drifted from its opening no longer
resized the opening, so the wall recut still hit a stale mapped source.
Extract a targeted single-source helper on tool.Model
(regenerate_filling_opening_body) that regenerates one filling's
mapped opening body via the existing FilledOpeningGenerator and
inverse-substitutes the new representation across every filling that
shares the mapped source. Refactor the family-wide caller
(update_simple_openings, used by the parametric-edit finish path) to
delegate to the same helper, deduped by source id so fragmented type
families where multiple mapped sources coexist all get refreshed.
Call the targeted helper at the top of RecalculateFill._recalculate_fills
for each distinct source among the selected fillings. All body-
representation lookups go through tool.Geometry.get_body_representation
rather than inlining the ("Model", "Body", "MODEL_VIEW") triple. An AST
forward-compat guard pins the call site.
Generated with the assistance of an AI coding tool.
The + gizmo previously appeared whenever a fillable host and any
non-host object were selected, so clicking it against an IfcCovering
crashed the geometry kernel when the opening generator tried to derive
a shape it couldn't build (AttributeError on 'NoneType.wrapped_data').
Tighten the gizmo poll to require the secondary selection to be a
class the operator can dispatch on: IfcDoor, IfcWindow,
IfcOpeningElement, or a non-IFC mesh. Make the poll selection-order-
independent so either click order activates it. Validate the same
class set at the operator boundary so keymap or scripted invocations
report a clear warning instead of crashing.
The narrower Door/Window support in the opening generator is a Bonsai
implementation limit, not an IFC schema restriction —
IfcRelFillsElement.RelatedBuildingElement is typed as IfcElement and
the schema permits any subtype. The tooltip and inline comment on the
validation branch note this so a future reader knows the gate is
future-work, not schema-mandated.
Rewrite the operator's bl_description to end-user-friendly wording that
drops the internal terms matrix_world and rl1/rl2.
Fixes#8215.
Generated with the assistance of an AI coding tool.
Coverage had lagged the recent feature work. Add both layers:
- test_viewport_camera (Catch2, headless): constructs ViewportCore with a mock
ViewportHost — construction/teardown touch no GPU (the wgpu teardown lives in
releaseWgpuModelGpuData, only reached with models loaded), and the camera ops
are pure — so it unit-tests the SHARED fly math fast and deterministically:
flyMove (forward step, 5x boost, opposing-key cancel, dt clamp, QE along +Z,
degenerate-pitch stays finite), flyLook (turn-in-place pins the eye, pitch
clamp), flyAdjustSpeed (x1.25/notch, [0.05,1000] clamp), toggleXray, and
hideSelected/showAll. Links the built IfcViewerCore (ViewportCore.cpp is the
monster TU that can't compile standalone). +9 cases → 122 desktop.
- smoke.spec.mjs (Playwright): fly (enter → W moves the camera → Esc exits),
x-ray (toggle translucency on/off), and hide-after-pick, driving the exported
C hooks end-to-end. +3 cases → 9 web.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Lift the visibility + X-ray ops into ViewportCore so desktop and web share them.
The cull already reads visibility_ (hidden objects skipped) and the frame uniform
reads xray_alpha_cap_, both per frame, so each op just mutates state + schedules
a frame — no GPU buffers to rebuild, no rendering work:
- hideSelected (hide the selection, then deselect), isolateSelected (hide every
non-selected object in a visible model), showAll (clear the hidden set),
toggleXray (flip the alpha cap 1.0<->0.3; cull routes all to the transparent
pass), xrayActive().
- Desktop ViewportWindow: the H / Shift+H / Alt+H / Alt+X keys and the menu-action
wrappers now call the core; the four inline/duplicated implementations are gone.
- Web main_web: same keys (H hide · Shift+H isolate · Alt+H show all · Alt+X
x-ray) + toolbar buttons (Hide / Isolate / Show all / X-ray, the last reflecting
active state).
Verified on web: x-ray toggles and visibly translucent-izes the scene, hide after
a pick removes geometry, 0 GPU errors. 113/113 desktop + 6/6 web smoke; desktop
app builds.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
applyCachedModel already does a one-time viewAll gated by initial_view_applied_
(desktop + web) — it frames the FIRST model loaded and never re-frames as more
arrive. Two web-only divergences from the desktop viewer, fixed here:
- loadSidecarMetadataWeb called viewAll() AGAIN, unconditionally, per model, so
every federated model that streamed in yanked the camera back to fit the whole
scene. Drop the redundant call; the shared gate handles first-model framing.
- Nothing ever cleared initial_view_applied_. On web the embedded sample sets it
at startup, so after clear_scene_c (a ?model= federation load) the gate was
already tripped and the loaded models never got framed — the camera stayed on
the prior view. (The redundant per-model viewAll above masked this until it was
removed.) Clear the flag in resetScene: a fresh scene should auto-frame its
first model — correct for desktop fresh-opens too.
113/113 desktop + 6/6 web smoke pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
DRY the fly camera into ViewportCore so desktop and web share identical math:
- flyMove(fwd,back,right,left,up,down,boost,dt): WASD in the view plane, QE
along world +Z, forward = eye→target (no snap after orbiting), Shift = 5x,
dt clamped to 0.1s. flyLook(dx,dy): turn in place (yaw/pitch, eye pinned,
0.2 deg/px, pitch ±89.9). flyAdjustSpeed(notches): wheel scales speed x1.25.
fly_move_speed_ + orbitEye now live in the core (orbitEye's duplicate removed
from ViewportWindow).
- Desktop ViewportWindow: fpsIntegrate / mouse-look / wheel-speed call the core
methods; behaviour unchanged, BonsaiViewer builds clean.
- Web main_web: Shift+F (or the Fly toolbar button) enters and pointer-locks the
canvas; W/A/S/D/Q/E + Shift are held-tracked (keydown/keyup) and integrated
each RAF frame with wall-clock dt; pointer-lock mouse deltas drive flyLook;
wheel tunes speed. Exit on Esc OR a canvas click (matches desktop) — a
pointerlockchange handler catches the browser eating the first Esc to release
the lock (so a single Esc exits), guarded by fly_locked so a denied lock on
entry doesn't insta-exit. Fly button reflects/ syncs active state.
Verified: web enter→WASD moves→click/Esc exits, 0 GPU errors; 113/113 desktop
core + 6/6 web smoke; desktop app builds.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The .ifcview data is hugely redundant (repeated double instance matrices,
patterned indices) — measured 12x zstd whole-file. Server Content-Encoding
can't be used (it breaks HTTP Range), so compress PER-CHUNK into the format.
Format (v16): geometry becomes per-chunk zstd(vertices)+zstd(indices) frames —
each independently Range-fetchable, so streaming is intact — and the critical +
deferred metadata blocks are single zstd frames. SidecarChunk carries the
compressed blob offsets/sizes; applyStreamedChunk (render/upload) is UNCHANGED —
decompression slots into the fetch. Full readSidecar (test/tooling) reconstructs
by decompress+scatter. zstd: desktop links libzstd (also compresses at bake);
the web build (Emscripten has no zstd port) FetchContent's the pinned zstd
source and compiles its decompress-only subset for wasm — no vendored blob,
same version as desktop. New SidecarCompress wraps it (compress guarded off
under Emscripten). Both stream paths — desktop StreamingThread worker + sync
fallback (readChunkGeometryCompressed) and web beginWebChunkLoad — decompress;
readSidecarMetadataOnly / the web bootstrap / loadDeferredMetadataWeb decompress
the metadata blocks. streamingByteProgress reports COMPRESSED bytes. MEASURED: a
752 MB v15 federation → 75 MB v16 (10x; per-file 6.7-15.3x); PP-PLP 118→15 MB,
loads 13/13 chunks on web, 0 errors.
Three fixes found while testing big federations on a real server:
- Web-streamed race: streaming_from_web was set in the deferred-header callback
(a round-trip after the model+chunks exist), so driveStreamingLoads could take
the sync fopen path meanwhile → "failed to read/decompress chunk 0". Now set
immediately after applyCachedModel.
- OOM abort on 18 models: the pool grew unbounded until an alloc failed, but on
web that's an uncatchable bad_alloc abort. Cap total pool capacity
(setMaxTotalCapacity, 3 GB) so it stops before the heap ceiling, and raise
MAXIMUM_MEMORY 2→4 GB (wasm32 max) for headroom.
- Web never evicted (grow-or-block only). At the hard budget, fall through to the
LRU/priority evictor so a big federation stays navigable (highest-contribution
chunks win) instead of freezing with holes.
113/113 desktop + 6/6 web smoke pass. No back-compat: regenerate sidecars
(desktop bakes v16; scratch conv tool migrates v15→v16).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Companion to the assign_type.py fix. The same class-pairing validation
added in 10ee5aef4f also gates the Bonsai-side type assignment UI via
tool.Type.is_relating_type_compatible, which the AssignType operator
uses to filter selectable objects. For annotation types (abstract
IfcTypeProduct), get_applicable_types(IfcAnnotation) is empty, so every
annotation was skipped with "No selected object can be typed by
IfcTypeProduct."
Honor the type's ApplicableOccurrence attribute as a fallback, matching
the core API fix. occurrence.is_a() handles subtypes and returns False
for unknown tokens, so free-form text is not trusted blindly.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The class-pairing validation added in 10ee5aef4f rejected every typed
annotation with "IfcTypeProduct cannot type IfcAnnotation ... (allowed
occurrence classes: <none>)".
The check derived allowed occurrence classes solely from the
buildingSMART implementer-agreement map, which has no entry for the
abstract IfcTypeProduct that Bonsai uses for annotation types (IFC4 has
no IfcAnnotationType). The intended occurrence class is declared in the
type's ApplicableOccurrence attribute (e.g. "IfcAnnotation/TEXT"), the
schema-defined mechanism for exactly this purpose.
Augment the allow-list with the ApplicableOccurrence class, but only
when its leading token resolves to a real entity in the schema so
free-form text is not trusted blindly. Genuine mismatches (e.g.
IfcWallType -> IfcWindow) are still rejected.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Streaming picked candidates on raw frustum visibility, so viewAll over a big
federation (all models in-frustum) fetched every chunk — even fine chunks that
project to sub-pixel and the renderer never draws. Gate streaming on the SAME
contribution decision the render path already makes.
- New per-chunk contribution_visible_count: instances that passed frustum AND
the contribution cull (projected radius >= min_radius_px), counted BEFORE HiZ
— so it's stable while the camera is still (unlike the HiZ-post counters,
which flip frame-to-frame and would thrash the loader) and shifts only on
navigation, when the working set should. The candidate gate skips chunks with
count 0: the network pulls only what's resolvable now; the rest stream in as
you approach. Verified: fit view needs the geometry, zoomed-out needs 0.
- Loading UI reworked from per-model segments to a combined bar over the whole
federation: dark track = not needed for this view, dim = needed-but-unloaded,
bright = loaded. "loaded / needed" = how done THIS view is; "needed / total" =
how much of the model the view requires — "Loading 45 / 90 MB for this view ·
12% of 718 MB total". Driven by ViewportCore::streamingByteProgress + ifcv_bytes_*.
Two fixes from testing: (1) show a distinct overhead phase while total==0
("Loading model data — X MB · Y/N models ready") so the metadata download
isn't a dead-looking bar; (2) re-assert display:block every active frame so the
bar REAPPEARS when navigation reveals new chunks (was only set in
beginLoadProgress → stayed hidden after the first catch-up). Per-model exports
kept for a future detailed view.
111/111 unit + 6/6 web smoke pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The camera math already lived in the shared ViewportCore; the desktop just
bound keys to it. The web build had no keyboard handler and no camera UI, so
none of it was reachable. Wire it up (Tier 1 + 2 of nav parity; fly mode is a
separate follow-up).
Shared core:
- setStandardView(StandardView) — named Front/Back/Left/Right/Top/Bottom wrapper
over setStandardView(yaw,pitch), so the axis→angle mapping lives in one place.
- frameSelection() — lifts the desktop's "union selected AABBs → frameAabb(1.30)"
focus logic out of ViewportWindow into the core. Desktop's focusOnSelectedObject
and the X/Y/Z hotkeys now call these (DRY, behaviour unchanged).
Web:
- main_web gains a keydown handler matching the desktop bindings — Home=view all,
F=zoom to selected, P=ortho toggle, X/Y/Z (+Shift=negative)=standard views —
plus exported entry points (view_all_c / frame_selection_c / toggle_projection_c
/ projection_is_ortho_c / standard_view_c) for the toolbar.
- shell.html adds a bottom nav toolbar (Fit / Focus / Persp-Ortho / the six views)
with the hotkeys in tooltips; the ortho button reflects state.
Verified: Z key and Front button both move the camera, ortho toggles render +
label, zero GPU errors; 111/111 unit + 6/6 web smoke pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Query string: auto-load a whole federation, not one model. Accepts repeated
params (?model=a&model=b&…) and/or a comma list (?models=a,b,c); each becomes
its own streamed byte-source, clearing the embedded sample once then streaming
concurrently into one scene. A failed URL is reported without aborting the rest.
Loading UI: the bare aggregate chunk count didn't reveal the federation state,
so surface per-model progress. ViewportCore::streamingModelCount() +
streamingModelProgress(idx,…) (ordered by model_id = load order) feed
main_web's ifcv_model_count_c / ifcv_model_{resident,total}_c. shell.html draws
a panel: "Loading N models — X done · Y streaming · Z waiting · N MB" plus one
segment per model (blue fill while streaming, green when resident, grey while
its metadata is still pending) — so parallel loading and how many remain are
visible at a glance.
Verified: 10 models from a 10x ?model= query stream in parallel; the panel
steps 10 waiting → streaming → all done. 111/111 unit + 6/6 web smoke pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The scene core is already multi-model — models_gpu_ is a map, applyCachedModel
APPENDS, and per-model model_id / object_id rebasing / georef+transformation
are how the desktop federates today. The only web-specific gap was the byte
source: web had ONE global source (__ifcvFile/__ifcvUrl) and reset the scene on
every load, so it could show one file at a time. Desktop meanwhile carries a
per-model source (streaming_file_path).
Mirror that on web: give each model its own web_source_id into a JS source
registry (Module.__ifcvSources[id] = a picked File or a sized remote URL).
beginWebChunkLoad, the metadata bootstrap, and the on-demand deferred fetch all
read from the owning model's source, so several files stream concurrently into
one federated scene — reusing all the shared machinery (viewAll, picking, the
GUID fetch) untouched.
- webReadRangesAsync / ifcvReadRangeInto / ifcvSourceSize take a source id.
- loadSidecarMetadataWeb(source_id, …) appends (no resetScene); main_web
exposes load_sidecar_from_source_c(id) + clear_scene_c().
- URL size resolution moved to JS (shell.html registers + sizes sources via
HEAD/Range), retiring the C-side ifcvBeginUrlSource / ifcv_source_ready dance.
- shell.html: source registry + "Open" (replace) / "Add" (append) buttons,
multi-file selection; ?model= registers a URL source then loads.
Verified: two sidecars from two sources stream into one scene, both fully
resident, zero GPU errors. 111/111 unit + 6/6 web smoke pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
First real consumer of the v15 deferred property block, and an end-to-end
demonstration that on-demand property loading works. On a left-click pick,
logSelectedObjectGuidWeb ensures the owning model's deferred block is loaded
(loadDeferredMetadataWeb — a network fetch the FIRST time, cached after) and
logs the picked object's GUID to the console.
Fix uncovered while wiring it: applyCachedModel rebases instance object_ids to
a per-model global base (object_id_base) to keep them unique across models,
but the deferred elements carry the sidecar's original local ids — so a lookup
by the picked (global) id missed. Store object_id_base on the model and rebase
the elements by it when the deferred block loads.
Verified: with a streamed model, the deferred block is fetched ONLY after the
first pick (not at load), and the pick logs a valid 22-char IFC GUID. 111/111
unit + 6/6 web smoke pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Users had no feedback during the seconds before/while a model streams. Add a
top progress strip + caption driven by the streaming state:
- "Loading model… N MB" while the critical metadata downloads (no chunks yet),
- "Loading geometry — R / T chunks · N MB" as chunks go resident,
- "Loaded — T chunks · N MB", then it hides.
ViewportCore::streamingProgress(resident, total) sums chunk residency across
models; main_web exports ifcv_chunks_resident_c / ifcv_chunks_total_c for
shell.html to poll each RAF. The EM_JS range reader accumulates
Module.__ifcvBytesLoaded so the caption can show MB downloaded. Shown only for
streamed loads (?model= URL / picked file), not the embedded sample.
6/6 web smoke pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
First-paint over a network is metadata-bound: the whole post-index metadata
(~10 MB on a 118 MB model) had to download before any geometry. But ~25% of
it — elements + string_table, the IFC element tree (names/GUIDs/hierarchy) —
is used only for UI/picking, never for rendering (ViewportCore never touches
it).
v15 splits the post-index metadata into a render-CRITICAL block (meshes,
instances, georef, chunk TOC) preceded by its byte length, then a DEFERRED
block (elements + string_table). The web loader reads only the critical block
before painting; the deferred block sits at a known, self-describing offset
([critical end, EOF)) and is fetched on demand. Desktop reads both (local).
Web on-demand path is wired and complete (not yet called — no UI consumer):
loadDeferredMetadataWeb(model_id) range-fetches + parses the deferred block
into ModelGpuData.elements/string_table, at most once; the first consumer
will be "show the selected object's name" on pick. No background prefetch —
view-only sessions never download the property data (saves 2.64 MB on this
model).
parseSidecarTail split into parseSidecarCritical + parseSidecarDeferred (pure,
unit-tested); StreamingSidecar gains the critical-block locator. Measured
(118 MB model): critical metadata 10.35 -> 7.71 MB, deferred 2.64 MB off the
path; first paint 10.5 -> 9.6 s @ 24 Mbps. (Instances still dominate the
critical block — the next metadata lever.) Format -> v15, no back-compat;
regenerate sidecars. 111/111 unit + 6/6 web smoke pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
SHIFT+CTRL+CLICK on Activate Drawing now imports the
annotations of all selected drawings without switching
the active view or camera. The drawing camera is imported
when missing so annotations are collected into the correct
drawing collection. Loading is idempotent.
Generated with the assistance of an AI coding tool.
The drawing Include filter replaced the camera-view element set
entirely, so elements outside the camera boundary were drawn.
Intersect the filter results with the camera-view set instead.
Generated with the assistance of an AI coding tool.
The georef orientation gizmo (crosshair, project-north arrow,
grid-north arrow, true-north arrow, WCS leader) is a coordinate-
system overlay: its purpose is to communicate orientation regardless
of what the model contains, so it must remain visible regardless of
whether 3D geometry occupies the gizmo's z=0 footprint.
Wrap GeoreferenceDecorator.draw_geometry's draw cycle in a
gpu.state.depth_test_set("ALWAYS") / restore pair so the overlay
draws on top of any 3D geometry between the camera and the gizmo.
Matches the precedent set by the dashed-line overlay in
bim/module/model/opening.py.
Generated with the assistance of an AI coding tool.
Three crashes that surfaced when viewport decorators ran against
selected non-IFC blender objects or top-level objects with no
aggregate parent:
- WallAxisDecorator.draw_wall_axis: tool.Ifc.get_entity(obj) returns
None for a non-IFC selection (default cube, lamp, camera). The
subsequent element.is_a("IfcWall") raised AttributeError on every
redraw. Guard with `element and element.is_a(...)`.
- _ConnectedNetworkPathDecorator flow-segment loop: same shape;
iterates entries that may be None, calls .is_a("IfcFlowSegment")
unconditionally. Same guard.
- AggregateDecorator.draw_aggregate: indexes aggregates_list[-1]
unconditionally in the else branch; raises IndexError when the
selected element has no aggregate parent. Also leaves `aggregate`
unbound across loop iterations in the `in_aggregate_mode` branch
when `index <= 0`. Define `aggregate = None` per loop iteration
and guard the [-1] indexing with `elif aggregates_list:`.
Generated with the assistance of an AI coding tool.
Migrate 17 legacy viewport decorators (ClashDecorator, SolarDecorator,
MeasureDecorator, ItemDecorator, GeoreferenceDecorator, NestDecorator,
NestModeDecorator, GridDecorator, LoadsDecorator, AggregateDecorator,
AggregateModeDecorator, PolylineDecorator, ProductDecorator,
WallAxisDecorator, SlabDirectionDecorator, FaceAreaDecorator,
BoundingBoxDecorator) from hand-rolled install/uninstall lifecycles
onto the canonical tool.Blender.ViewportDecorator base. The legacy
uninstall removed each handler from Blender but never cleared
cls.handlers, growing a stale-reference list across enable/disable
cycles. The base's uninstall clears the list correctly.
State-derived install methods (ItemDecorator, ProductDecorator,
LoadsDecorator, PolylineDecorator) keep an install override per the
base's documented contract.
Drop the now-redundant per-class draw_batch copies and the module-
or method-scope transparent_color defs in favour of the base helpers
introduced in the preceding commit. system/decorator.py and
boundary/decorator.py keep their installed-flag lifecycle (different
pattern, no leak) but consume tool.Blender.transparent_color.
Add an AST forward-compat guard pinning the contract structurally:
any class declaring handlers = [] (Assign or AnnAssign) must subclass
tool.Blender.ViewportDecorator. Add a runtime regression on
ClashDecorator's install/uninstall cycle.
Generated with the assistance of an AI coding tool.
Add two helpers to tool.Blender that 17+ existing viewport decorators
re-implement byte-identically:
- ViewportDecorator.draw_batch(shader_type, content_pos, color, indices=None)
collapses the validate + batch_for_shader + uniform_float + draw cycle
every shader-driven decorator needs.
- Blender.transparent_color(color, alpha=0.1) is the RGBA-alpha-override
helper duplicated across nest, project, aggregate, model, system module
scopes plus six nested-def copies inside draw methods.
Pure additions on the tool/ layer with direct unit tests covering the
default-alpha, explicit-alpha, non-mutation, new-list-instance, and
validation-guard branches.
Generated with the assistance of an AI coding tool.
Makes large-model streaming over a network actually good — fixing read
amplification, then first-paint latency — building on the byte-range work.
v14 layout + TOC (SidecarLayout, pure + unit-tested)
The loader chunks meshes by spatial Morton order, but the sidecar stored
geometry in mesh-id order, so a chunk's meshes were scattered through the
file: streaming one chunk meant either hundreds of tiny range requests or
reading (and discarding) everything between them — a 113 MB model fetched
~340 MB, a 531 MB model 2.25 GB (4.2x). Fix: at bake, reorder meshes into
the loader's chunk order and rebuild vertex/index(LOD0+LOD1)/instance
sections so each chunk is one CONTIGUOUS byte range, and bake a chunk TOC
({first_mesh, mesh_count}). The loader builds chunks straight from the TOC
rather than re-deriving the plan — the float Morton quantisation isn't
bit-identical across toolchains (x86 baker vs wasm loader), so a re-derived
plan scatters the chunks. Format bumped to v14 (regenerate sidecars). The
reorder buckets instances by per-instance mesh_id (the baker never sets
MeshInfo.first_instance — trusting it scrambled every transform → geometry
at the origin). Multiset-verified on a 28,900-instance model: every
instance's placement + geometry preserved. Result: 531 MB fetches 531 MB
(1.0x) in 72 requests (was 2036).
Progressive streaming (concurrency cap + small chunks)
Even at 1x, geometry appeared only after ~the whole model arrived: the
browser multiplexes every in-flight Range request over one HTTP/2 conn, so
unbounded concurrency (9 in flight) split the bandwidth and nothing finished
until the end (measured: first paint after 113 of 118 MB / 35 s @ 24 Mbps).
Cap concurrent chunk loads (kMaxWebInflightChunks=2): the priority-sorted
top chunks finish and paint first, then the next → first paint 9 s. Chunk
size dropped 16->4 MB (cheap now that each chunk is one read; matches Cesium
3D Tiles / xeokit / SVF2) for smoother progression. First-paint is now
metadata-bound (~10 MB tail) — the next lever.
111/111 unit (new test_sidecar_layout: geometry preserved, contiguous layout,
Morton-identity) + 6/6 web smoke pass; desktop bake (SceneLoader) reorders
before writeSidecar; embedded web sample regenerated to v14.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A pending RegionView3D.update() timer registered before wm.open_mainfile()
fires during the load against freshly-allocated regions whose GPU contexts
are not yet wired, CTD-ing inside GPU_matrix_ortho_set. Cancel both the
refresh and cap-rebuild timers in a new load_pre handler, hold a
_file_loading gate from load_pre through the first on_pre_view tick (first
paint = GPU ready), and short-circuit on_depsgraph_update during the
window so its IFC-reload schedule_refresh + apply_clip_planes_direct
branches can't re-arm against unready regions.
Generated with the assistance of an AI coding tool.
Lets tooling read the wasm heap size (e.g. to verify a large sidecar
streams by byte range instead of loading whole, and to watch memory while
battle-testing big models over the network). Standard emscripten runtime
method, zero size cost.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Battle-testing real sidecars over HTTP Range exposed severe thrash: a
531 MB model re-fetched 2.25 GB (4×) and never converged — viewAll puts
the whole model in frustum, so every chunk wants to be resident, and the
web async path made it worse two ways:
- A web load only consumes pool space when it COMPLETES (async), so the
per-frame issuance over-committed the pool; completions then failed
applyStreamedChunk on a full pool, the chunk re-candidated with no
cooldown, and re-fetched every frame.
- Pool growth is itself async on web (provisional sub-buffers validated
off the JS event loop), so even fetched chunks failed to alloc until
the pool caught up, and re-fetched.
Fix: gate web chunk issuance on VALIDATED free space + in-flight
reservation, and grow the pool BEFORE fetching:
- streaming_web_inflight_bytes_ reserves each in-flight load's footprint
so we never have more bytes in flight than the pool can place.
- When a visible chunk doesn't fit validated free, don't fetch — call
pool_.requestGrowth() (BufferPool: drives the async provisional grow
without allocating) and short-back-off; the chunk is fetched once,
after space exists. When the pool is saturated (model > GPU memory),
long-cooldown so a never-fitting chunk isn't re-fetched. Gating before
the evictor also kills phase-2 visible↔visible swap thrash.
- On async load failure, cool down (short if the pool can still grow,
long if saturated) instead of re-candidating next frame.
Result (manual battle tool, host.mjs + real files): 531 MB now loads
23/23 chunks, 322 MB loads 14/14 — resident climbs monotonically with
ZERO thrash warnings and a stable resident set, vs the old re-fetch loop.
The whole model resides on the GPU and stays. (Remaining ~3× ramp
over-fetch — per-chunk re-loads during the async-growth ramp + read
amplification from chunk byte-locality — is a separate efficiency
follow-up, not thrash.) 6/6 web smoke + 107/107 unit pass; desktop
unaffected (the gate is web-only; requestGrowth is a no-op wrapper there).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
beginWebChunkLoad read the vertex ranges, then in the completion callback
read the index ranges, then applied — two serial round trips per chunk.
On a network that's the dominant per-chunk latency. Now both reads fire
at once and a small shared join (payloads + per-read done/ok flags) runs
the apply when the second lands, halving per-chunk RTT. Model re-lookup
still happens at apply time, so a resetScene mid-flight is dropped safely.
Per-chunk concurrency stacks with the existing across-chunk concurrency
(driveStreamingLoads issues several loads per frame); the browser caps
simultaneous connections per origin, so no explicit in-flight cap is
needed. 6/6 web smoke + 107/107 unit pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adds a network byte-source alongside the local Blob one. The async-chunk
infra is source-agnostic — only the two JS primitives knew it was a Blob —
so this generalises them and reuses everything else:
- ifcvReadRangeInto: local → Blob.slice; remote → fetch() with a Range
header (206). If a server ignores Range and returns 200, the requested
window is sliced out so it still works (without the bandwidth saving).
- ifcvFileSize: Blob size, or the URL's total length resolved up front.
- ifcvBeginUrlSource: resolves total size (HEAD Content-Length, else a
0-0 ranged GET's Content-Range) then fires _ifcv_source_ready.
- The metadata bootstrap is extracted into a source-agnostic
loadSidecarMetadataWeb(label); loadSidecarFromBlobWeb / FromUrlWeb are
thin entries. streaming_from_blob → streaming_from_web (now covers both).
main_web exports load_sidecar_from_url_c(url); shell.html reads a
?model=URL query param and ccalls it once the app is live (same-origin
needs no CORS; cross-origin hosts must send CORS + Accept-Ranges).
Test: serve.mjs now answers HEAD + Range (206) and falls back to the
ifcviewer-web source dir for sample.ifcview (embedded in the wasm, not in
build-web). New smoke case loads ?model=/sample.ifcview and asserts it
renders via the Range path. 6/6 web smoke + 107/107 unit pass; desktop
unaffected (web-guarded; only the shared field rename touches it).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Reported: open a model in the web viewer, then launch the desktop
BonsaiViewer and open another model — the browser tab freezes, and a
fresh web tab then fails with "RequestDevice failed: Not enough memory
left". Root cause is GPU-memory contention: two heavy GPU clients on one
GPU, and the desktop app's allocations starve the browser's WebGPU
process, which reclaims our device.
We can't conjure GPU memory, but we were amplifying the symptom: with no
device-lost handler, render() kept driving a dead device —
wgpuSurfaceGetCurrentTexture returns Lost every frame and the
reconfigure + requestFrame retry becomes a tight per-RAF loop that hangs
the tab. Now the web device descriptor wires a device-lost callback that
latches device_lost_ (ignoring the intentional Destroyed reason from our
own shutdown); render() bails while set, so the loop goes idle instead of
spinning, and the console logs guidance to reload. The fresh-tab
RequestDevice OOM is genuine GPU exhaustion — surfaced as before, now with
a clearer message.
Verified the lost callback doesn't disturb Dawn-web's RequestDevice (all
5 web smoke tests still init + pass); desktop unaffected (device_lost_
stays false). The real contention path can't be reproduced in the headless
harness, so the loss handler itself is covered by review, not a test.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The fragment shader pre-decodes sRGB→linear to cancel the surface's
automatic linear→sRGB write encoding, so the final bytes match the GL
backend. That only holds when the render target is an sRGB format. On
desktop the surface's preferred format already is (e.g. BGRA8UnormSrgb),
but the browser canvas only offers plain BGRA8Unorm — so nothing
re-encoded and the whole image (background + models) rendered ~3× too
dark (authored bg 0.125,0.137,0.161 → ~32,35,41 collapsed to ~3,4,6).
Fix: when the surface format isn't sRGB, render through an sRGB *view* of
it — the standard WebGPU canvas pattern. surface_view_format_ is the sRGB
sibling of surface_format_ (unchanged when already sRGB, so desktop is a
no-op); configureSurface advertises it via viewFormats, the colour
pipelines (main, MSAA target, edge) target it, and render() creates the
surface view with it. The screenshot path still reads the base texture, so
its BGRA byte-order check stays on surface_format_.
Regression test: sample a 1x1 background pixel and assert it isn't crushed
dark (R,B > 20). Verified visually too — bg is now the correct dark
blue-gray and the cube is properly lit. 5/5 web smoke + 107/107 unit pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
On first web load the sample stayed blank until a click/drag, then popped
in. Root cause: the main draw + cull run before driveStreamingLoads in
render(), so a chunk that becomes resident there is only painted a frame
later. On desktop the streaming thread keeps inFlightApprox() > 0 during a
load, so the render loop keeps ticking and the next frame paints it. On web
the sync MEMFS / Blob load finishes instantly (inFlightApprox stays 0), so
the single post-load requestFrame fired once and the on-demand loop went
idle before the geometry was ever drawn — until some input re-armed it.
Fix: arm a bounded settle burst (kStreamingSettleFrames) whenever there's
streaming activity — a load this frame, work still queued, or a visible
chunk not yet resident — and bleed it down over the next few frames, each
requesting one more. Covers the cull→display latency under an on-demand
loop and still quiesces at idle (no busy-rendering). General, not web-only.
Regression test: the sample must render with NO pointer input — a centred
patch (the framed cube) differs from a corner patch (background); a blank
stall leaves both as background. Verified empirically with a no-interaction
probe (canvas went from a static blank hash to a stable rendered one).
107/107 unit tests pass; 4/4 web smoke tests pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Third Playwright case: click dead-centre on the framed sample, assert the
canvas changes (selection highlight rendered) with zero WebGPU errors. A
broken async pick would hang init or leave the canvas unchanged. All three
cases pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Desktop pick reads the pick staging buffer back with a blocking
`while(!done) waitTickInstance()` spin. On web that spin is a no-op
(Asyncify is off) and hangs the JS event loop, so click-to-select was
dead on web. This adds an async sibling that uses the spontaneous
map-callback pattern (AllowSpontaneous + the browser microtask loop)
already proven by the HiZ readback — no blocking.
- encodePickReadbackToStaging(x,y,want_normal): the pick-pass render +
copy-texel-to-staging, extracted from pickObjectAt verbatim and shared
by both readbacks (desktop sync path unchanged).
- pickObjectAtAsync(x,y,cb) [web]: encode, then map the staging buffer
with a spontaneous callback that delivers object_id to cb. One pick in
flight at a time (a pick issued mid-map is dropped → cb(0)).
- applyPickToSelection(id, add, remove): routes a pick result through the
selection state machine (replace / Shift-add / Ctrl-remove / empty-click
clear), mirroring the desktop ViewportWindow click semantics. selection_
marks dirty so the next render's uploadSelectionFlagsIfDirty flushes the
highlight.
main_web wires it: a left release under a 4px drag threshold (no orbit) is
a pick at down-position * devicePixelRatio, with Shift/Ctrl modifiers;
the result callback applies selection and requests a frame. Web + desktop
build clean; 107/107 unit tests pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Collapse two parallel hand-maintained lists in the addon-preferences
PropertyGroups into derivations from `tool.Parametric.EDIT_TYPES`:
- GizmoPreferences: the 10 `<name>: BoolProperty` annotations now
generated from the full EDIT_TYPES list.
- DefaultParameters: add `has_default_parameters` flag to
ParametricObject (set True on door/window/stair/railing/roof);
derive the 5 `<name>: PointerProperty(type=BIM<X>Properties)`
annotations and collapse the 5 hand-written `draw_expandable_panel`
blocks in `draw_default_parameters` into loops driven by the flag.
Existing `test_gizmo_preferences_field_per_registry_entry` pinned the
GizmoPreferences contract; new
`test_default_parameters_field_per_registry_entry_with_defaults`
pins the DefaultParameters contract (one-directional: flag=True
implies field present, flag=False allows absence).
Generated with the assistance of an AI coding tool.
Refs gh#8088. Array regen + multi-opening drops fan out N+1 wall recuts
per operator (one per child filling deletion + the final mirror recut),
making CSG opening-subtraction O(N^2) for a linear UX action.
Introduces tool.Geometry.batch_host_recut() — a context manager that
coalesces switch_representation and bpy.ops.bim.update_representation
calls per voided element within one operator transaction. The drain
re-reads the active representation so the recut reflects current IFC.
Wraps 7 entry points (regenerate_array, RegenerateArray, RemoveArray,
AddOpening, RecalculateFill, CloneOpening, regenerate_from_type) and
rewires 7 leaf call sites in opening.py, void/operator.py, and
mirror_parent_void_fillings_to_children.
An AST forward-compat guard pins the rewire contract: no direct
switch_representation or bpy.ops.bim.update_representation in the
three target files outside the helper definitions.
A 16-child array regen now recuts the wall once instead of 17 times.
The CSG cost per recut is unchanged; only the count is reduced.
21 new tests across three lanes (helper unit, entry-point coalescing,
AST guard) — all green.
Generated with the assistance of an AI coding tool.
Mirror the icon-click entry into wall parametric edit on the TAB key.
The dispatch in Modifier.try_applying_edit_mode had no branch for fresh
LAYER2 walls, so TAB landed in item mode instead of the parametric
draft + gizmos. Add the missing entry leg of the toggle, placed after
the generic is_object_editing branch so the finish leg still fires
when a wall is already in edit mode.
Generated with the assistance of an AI coding tool.
Adds a second Playwright case that picks the sample sidecar through the
file input, waits for the C side to confirm the blob load (console),
then asserts an orbit drag changes the canvas with zero WebGPU errors.
This exercises the #88 path distinctly from the embedded MEMFS sample —
a broken metadata-head/tail or chunk range read renders blank and fails
the orbit-changed-canvas check. Both cases pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Picked files are no longer copied whole into the wasm heap. The browser
File object stays in JS (Module.__ifcvFile) and is read lazily through
Blob.slice byte ranges, so a 200-500 MB sidecar never enters wasm linear
memory — only chunk-sized slices do.
Mechanism (web-only, #if __EMSCRIPTEN__):
- JS glue (EM_JS): ifcvFileSize + ifcvReadRangeInto — slice [off,off+n)
of the File and copy it into a caller-provided heap pointer, then call
back _ifcv_on_range_done. No malloc across the boundary; C pre-sizes
the destination from the read plan.
- webReadRangesAsync: reuses planSidecarReadRanges to coalesce a range
set into Blob.slice reads (1 MB gap — each slice is an async hop),
scatters them into a destination laid out in input order, and fires a
continuation when the whole set lands. An in-flight map keyed by id
survives unordered_map rehash (scratch buffers are heap-owned).
- loadSidecarFromBlobWeb: async metadata load — head (16 B) -> index
count -> tail-to-EOF -> parseSidecarHead/Tail -> applyCachedModel, then
tags the model streaming_from_blob and frames it.
- driveStreamingLoads: blob-sourced models route to beginWebChunkLoad
(async vertex+index range reads -> applyStreamedChunk in the callback),
holding is_loading until the bytes arrive. The embedded MEMFS sample
keeps the synchronous fopen path.
shell.html stashes the File and calls _load_sidecar_from_blob_c instead of
FS.writeFile'ing the whole thing; EXPORTED_RUNTIME_METHODS=['FS'] dropped.
Desktop is untouched (the new members + driveStreamingLoads branch are all
emscripten-guarded). Web links clean; desktop rebuilds; 107/107 unit tests
pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Splits the v13 metadata wire-format knowledge out of the FILE*-bound
streaming reader into pure, buffer-based functions so the web byte-range
path (#88) can reuse it without loading the whole sidecar into the wasm
heap:
- parseSidecarHead — validates the 16-byte head, yields num_vertex_bytes
- parseSidecarTail — parses meshes/instances/georef/elements/strings
from an in-memory tail buffer, bounds-checked
- planSidecarReadRanges + SidecarReadPlan — the range-coalescing /
scatter planner, promoted out of the anonymous namespace
readSidecarMetadataOnly and the range readers now call these; desktop
behaviour is unchanged (head + tail are small, the bulk is still skipped
via seek). The metadata tail is split from the head around the bulk
sections, so a blob-backed loader just slices those two regions and
hands the bytes to the same parsers.
Closes a coverage gap: StreamingLoader had no unit tests. Adds
test_streaming_loader.cpp (7 cases: metadata round-trip, corrupt/truncated
rejection, vertex+index range scatter, head validation, tail truncation,
read-plan coalescing). 107/107 unit tests pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Drives the built web page in a real Chrome (channel:'chrome', so no
`playwright install`): waits for wgpu init, then asserts an orbit drag
changes the composited canvas — one check that simultaneously proves the
scene rendered, mouse input is wired, and the log overlay isn't eating
events — and that zero uncaptured WebGPU errors were logged. Every web
bring-up bug so far (blank render, error-buffer cascade, overlay
swallowing input) is this shape; this would have caught them.
serve.mjs statically serves build-web; the config launches headed
against the real GPU (--use-angle=vulkan + --ignore-gpu-blocklist are
load-bearing for a non-null adapter on Linux Chrome). node_modules and
results are gitignored. See README.md to run.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
On web we skip the desktop error-scope spin-wait (it blocks the JS event
loop and hangs the page). The old web addSubBuffer then judged success by
`buf != nullptr` — but Dawn-web returns a NON-NULL error buffer on OOM,
so the pool committed an invalid sub-buffer, alloc handed out slices in
it, and every chunk_bind_group built against it failed ("BindGroup is
invalid" spam + a wgpuQueueSubmit panic). Loading a model larger than the
browser's WebGPU budget triggered exactly this.
Add the grown sub-buffer as *provisional* (alloc and the capacity/free
tallies skip it) and validate it through a non-blocking AllowSpontaneous
PopErrorScope. resolveProvisionalGrowth() clears the flag when it's good,
or drops the sub-buffer and latches growth_disabled_ on a real OOM — at
which point the streaming evictor bounds the working set to what fits
instead of cascading. Only one provisional grow is in flight at a time
(growth_pending_). Desktop keeps its synchronous halve-retry path
unchanged. All 100 unit tests still pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The status overlay was position:fixed top/left/right with max-height
80vh and pointer-events:auto — a near-fullscreen div that both hid the
model and swallowed mouse events, so orbit drags over most of the canvas
did nothing. Move it to a small bottom-left box, set pointer-events:none
so it never intercepts navigation, and collapse it to a few dimmed lines
once the app goes live (errors re-expand it). It still auto-scrolls to
the newest line.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adds an "Open .ifcview…" button that opens the browser's native file
chooser. The picked file's bytes are written into MEMFS and a new
exported entry point, load_uploaded_model_c, reads them back via the
existing loadSidecarFromPath: it resetScene()s the current model
(replace, not append), loads the sidecar, and viewAll()s it. Geometry
becomes resident over subsequent frames through render()'s inline
driveStreamingLoads, same as the embedded sample.
Deliberately a file picker, not drag-drop: browser file drag-drop needs
an X11 drag source (a file manager) to drag *from*, which a minimal WM
(ratpoison) doesn't provide. The native chooser is WM-independent.
Exports _load_uploaded_model_c and the FS runtime method; URL/byte-range
fetch of remote models stays for #88.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The desktop pool walked down from 4 GB looking for the largest single
buffer the runtime would grant, then used that as the pool's first
sub-buffer and growth increment. On a stack that advertises an
effectively unbounded maxBufferSize (1 TB observed on wgpu-native here)
the walk lands on 2 GB, so loading even a 1.2 MB model allocated a 2 GB
sub-buffer. That plus the depth/MSAA/HiZ/pick attachments exhausted
VRAM, and the next tiny allocation — the ~4 KB selection_flags buffer —
failed with "Not enough memory left", invalidating its bind group and
panicking wgpuQueueSubmit.
Drop the probe and size the pool to demand, mirroring the web path:
configure a modest initial sub-buffer and let BufferPool::addSubBuffer
grow it lazily (halve-retrying to its 64 MB floor on constrained
devices). A single chunk is capped at 16 MB, so the initial sub-buffer
only has to clear that. Web keeps its 64 MB initial (Chrome contends on
large first allocations); desktop uses 256 MB to keep sub-buffer count
low for big models. The two platforms now share one createPool() (was
probeAndCreatePool — it no longer probes).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Register emscripten HTML5 pointer/wheel handlers that translate raw
events into ViewportCore::orbitBy / panBy / dollyBy: left-drag orbits,
middle/right-drag pans, wheel zooms. mousedown binds to the canvas;
mousemove/up bind to the window so a drag keeps tracking off-canvas. A
contextmenu suppressor lets right-drag pan without the browser menu.
Pure callbacks — no Asyncify, no sync spin — so none of the web init
gotchas apply. Pick stays deferred until the async buffer-readback
rewrite. The embedded sample is now navigable on Chrome + Firefox.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Replace the inline orbit/pan/wheel math in ViewportWindow's mouse
handlers with calls to ViewportCore::orbitBy / panBy / dollyBy. The core
methods request the frame (via the host), so the now-redundant
requestUpdate() calls drop out; the pivot-indicator afterglow and 3px
drag-promotion stay in the Qt layer where they belong. Behaviour is
unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The orbit navigation math lived in the Qt ViewportWindow, operating on
ViewportCore's camera fields through references. That left the web host
with no way to drive the camera — orbit/pan/zoom were Qt-only.
Lift the three pixel-delta moves into ViewportCore as orbitBy / panBy /
dollyBy so every host (Qt desktop + web) shares one implementation and
the math can't drift between platforms. panBy takes the viewport height
as a parameter (the one Qt coupling: pan's world-units-per-pixel needs
it) so the core stays toolkit-free.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Merging more than two IFC models with the MergeProjects recipe leaves
duplicated IfcGeometricRepresentationContext entities behind. All elements
are kept, but the accumulated contexts cause later disciplines to appear
"not merged" in viewers.
This test merges three projects and asserts the elements are kept, a single
IfcProject remains, and the geometric contexts are reused rather than
accumulated. It currently fails on the context assertion, reproducing #7973.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
EnableEditingExtrusionAxis and EnableEditingExtrusionProfile both
import a mesh from the IFC representation into obj.data as their
first real action. That mesh-import overwrites any in-memory
parametric (gizmo) draft on the object, silently discarding the
user's pending dimension edits.
Concrete reproduction: drag a wall's length gizmo (draft pending),
then click "Edit Axis" before validating the draft. The axis-edit
imports the wall axis mesh; the in-flight draft vanishes; on
cancel the wall snaps back to its pre-drag length.
Both call sites now commit the active draft via
tool.Parametric.commit_object_draft before the mesh import, gated
on tool.Parametric.is_object_editing so the guard is a no-op when
no draft is in flight.
Generated with the assistance of an AI coding tool.
Multi-instance cache-lock, "Opening Cuts Skipped", and "Arrays With
Missing Children" banners now draw in BIM_PT_tabs alongside the
existing global error / outdated-model banners, so they remain
visible regardless of the active Bonsai tab. The corresponding
blocks are removed from BIM_PT_project. Dead imports
(is_cache_locked_by_other_process, draw_multiline_text) dropped
from project/ui.py.
Generated with the assistance of an AI coding tool.
regenerate_wall_representation returns None for walls without an
IfcMaterialLayerSet (the only mode it knows how to rebuild). Feeding
None to switch_representation crashes deep in resolve_representation
on .Items. Document the None return on the API side and bail in
tool.Model.recreate_wall when it hits.
Generated with the assistance of an AI coding tool.
Schema-illegal IfcDoor->IfcWallType pairings parse cleanly but propagate
into operators that fan out by type and eventually crash the wrapper.
Block the pairing at its source: API guard in ifcopenshell.api.type.
assign_type, per-object partition in BIM_OT_assign_type + DuplicateType,
new tool.Type.is_relating_type_compatible helper, AST forward-compat
guard. Files in the wild are still loaded unchanged.
Generated with the assistance of an AI coding tool.
The "mep-pair-fitting" kind added in the previous commit carries an
IfcFlowFitting (the entity whose deletion disconnects the pair), not a
relationship entity, in the dispatch slot — but the slot was named ``rel``
across the function signature and every call site. Rename to ``subject``
so the parameter name reflects the uniform intent: "the entity whose
teardown effects the disconnect", regardless of whether that's a rel or
a fitting.
Sweep covers:
- core.connection.disconnect_rel signature + body
- tool.Connection.find_rels / find_rels_for_element / find_rel docstrings
- The cascade-on-delete call site in tool.Geometry.delete_ifc_object
- DisconnectElements operator in bim.module.model.wall
- All affected test kwargs and AST forward-compat docstring
- Error message: "Unknown rel kind" -> "Unknown kind"
No behaviour change.
Generated with the assistance of an AI coding tool.
Three user-facing fixes for the MEP-system gizmo surface:
1. MEP pair-disconnect no longer crashes Blender. The
MEPSystemPathDecorator cached entity_instance references in
_cached_walk; deleting a bridging fitting via the gizmo left a freed
SWIG handle in the list, and the next _build_geometry pass segfaulted
on .is_a. The cache now stores STEP integer ids and re-resolves via
ifc_file.by_id on each draw, plus folds tool.Parametric.get_geom_generation
into the cache key — ifcopenshell.api mutations invalidate before the
next frame regardless of how the deletion was routed.
2. Bend re-edit pen icon stays reachable. The bend creation path
tessellates the swept-disk body (upstream geometry-kernel workaround),
so tool.System.has_parametric_body correctly returns False for a
freshly-committed bend. _active_is_bend_fitting and
GizmoMEPActions.is_eligible_object now fall back to the type's
BBIM_Fitting pset — the same source bim.enable_bend_preview_from_bend
reads parameters from — keeping the pen icon eligible.
3. MEP pair / per-port unjoin icons unified through bim.disconnect_elements.
The MEP gizmo group's three unjoin icons (pair, start, end) now share
the wall-disconnect surface: same VIEW3D_GT_wall_link_toggle icon, same
bim.disconnect_elements operator. tool.Connection.find_rels learned a
new "mep-pair-fitting" kind that returns the bridging fitting as the
disconnect target; core.connection.disconnect_rel grew the matching
dispatch arm. The old MEPUnjoinAtPort and MEPUnjoinPair operators are
removed.
Also registered wall.GizmoPairDisconnect (previously declared but never
in the classes tuple, so dead code) for the wall+slab pair-disconnect
surface, and extracted MEP port-topology helpers (find_bridging_fitting,
is_disconnectable_fitting, neighbours_at_ports) onto tool.System so the
canonical walk has a single home.
Generated with the assistance of an AI coding tool.
Two new UX features in the IFC Patch panel, both backed by helpers on
bonsai.tool.Patch.
Lossy-downgrade confirmation popup. When the user picks the Migrate
recipe with a target schema older than the source's (IFC4 -> IFC2X3,
IFC4X3 -> IFC2X3), ExecuteIfcPatch.invoke shows a properties dialog
listing what's preserved vs lost: IfcIndexedPolyCurve flattened with
arcs approximated, IfcPolygonalFaceSet / IfcTriangulatedFaceSet
converted to IfcFacetedBrep, IFC4-only IfcElement subclasses (IfcLamp,
IfcPipeSegment, IfcGeographicElement, ...) demoted to
IfcBuildingElementProxy with the original class + PredefinedType
encoded into ObjectType, and PredefinedType enum values absent from
IFC2X3 dropped. The user explicitly approves before the recipe runs.
The popup is gated on tool.Patch.migration_is_lossy_downgrade() which
resolves the source schema via header-only parsing
(tool.Patch._patch_source_schema reads the first ~2KB and matches a
FILE_SCHEMA regex, then normalises via ifcopenshell.util.schema.
get_fallback_schema). Avoids a full ifcopenshell.open() on every
Execute click — multi-second saving on large files. The target schema
is looked up by argument name rather than position so it survives
recipe-parameter reordering.
Per-recipe preset menu. New BIM_MT_ifc_patch_presets + AddIfcPatchPreset
wire Blender's standard preset system into the panel. Each recipe gets
its own preset subdirectory (bonsai/ifc_patch/<RecipeName>/), so a
preset saved for ExtractElements does not pollute the Migrate preset
list. The preset operator uses Attribute.get_value_name() (single
source of truth for data_type -> storage-field mapping) to build the
preset_values list dynamically per recipe.
The recipe-change callback resets
BIM_MT_ifc_patch_presets.bl_label to the canonical title — Blender's
script.execute_preset mutates the menu's bl_label to the loaded
preset's name as a "currently-selected" indicator, and without an
explicit reset the previous recipe's preset name would falsely advertise
itself in the new recipe's menu.
tool.Patch gains get_preset_subdir, migration_is_lossy_downgrade,
_patch_source_schema as cross-cutting helpers. _SCHEMA_AGE module
constant provides the ordering used by the downgrade-detection
predicate.
Test coverage: 12 bim-lane tests under test/bim/module/patch/. The
truth table for migration_is_lossy_downgrade covers IFC4/IFC4X3 source
x downgrade/upgrade/same-schema target x Migrate/non-Migrate recipe.
The schema-sniffing tests write a real IFC4X3_ADD2 file to disk and
assert the helper resolves it to IFC4X3 (regression for the original
startswith iteration-order bug). An end-to-end test drives
bpy.ops.bim.execute_ifc_patch with an in-memory IfcLamp source and
verifies the on-disk IFC2X3 file contains a single
IfcBuildingElementProxy with ObjectType "IfcLamp/COMPACTFLUORESCENT"
and the original GlobalId preserved.
Generated with the assistance of an AI coding tool.
ExtractElements: expand the `query` docstring to cover the exclusion
syntax (`!` on entity classes, `!=` on attribute / pset / material /
classification / location / group facets) and the "seed with a broad
include before subtracting" gotcha — entity-class exclusion does not
auto-seed from "all elements", so a bare `! IfcSlab` query returns
nothing.
FixArchiCADToRevitDoorSwings: guard the `IfcIndexedPolyCurve.Segments`
loop against the IFC4 case where Segments is absent (a polyline
through all coords in declared order). Previously crashed on
`None.__iter__`.
Generated with the assistance of an AI coding tool.
The Migrate recipe previously crashed mid-loop with the cryptic
`RuntimeError: Entity with name '' not found in schema 'IFC2X3'` when
asked to downgrade an IFC4 or IFC4X3 file to IFC2X3 — the
class_4_to_2x3 mapping marks IFC4-only geometry / element classes with
an empty-string sentinel and the old code blindly forwarded that to
create_entity. Real files routinely contain IfcPolygonalFaceSet,
IfcTriangulatedFaceSet, IfcIndexedPolyCurve, IfcLamp, IfcPipeSegment,
IfcGeographicElement, etc.
The recipe now runs a preprocessing pipeline when the target is IFC2X3
and the source is IFC4 or IFC4X3:
- DowngradeIndexedPolyCurve flattens IfcIndexedPolyCurve to IfcPolyline
for the whole file (arcs included — see below).
- IfcPolygonalFaceSet / IfcTriangulatedFaceSet are converted directly
to IfcFacetedBrep at the entity level via
ifcopenshell.util.shape_builder.polygonal_face_set_to_faceted_brep,
preserving topology including IfcIndexedPolygonalFaceWithVoids inner
bounds. IfcShapeRepresentation carriers have their RepresentationType
tag updated from "Tessellation" to "Brep".
- Orphan source-only geometry instances (left over after the rewires)
are purged iteratively via
geometry_classes_introduced_after(target, source).
The Migrator is invoked with fallback_element_to_proxy=True so
IFC4-only IfcElement subclasses (IfcLamp, IfcPipeSegment,
IfcGeographicElement, ...) become IfcBuildingElementProxy in the
output. A post-pass encodes "<OriginalClass>/<PredefinedType>" into
ObjectType (e.g. "IfcLamp/COMPACTFLUORESCENT") when ObjectType is
empty, so the lost subclass identity survives the downgrade as
searchable text.
The migration loop now collects per-entity failures into a list rather
than crashing on the first; a summary RuntimeError fires at end if any
failed, naming up to 20 with their inverse references. Successful
migrations log a single count line via self.logger.
DowngradeIndexedPolyCurve extended:
- Arc segments (IfcArcIndex) are flattened via
ifcopenshell.util.shape_builder.arc_to_polyline_points with
ARC_SUBDIVISION=16 chord points per arc.
- Multi-index IfcLineIndex segments handled correctly.
- Absent Segments list (IFC4 polyline-through-all-coords case) handled.
Test coverage: 11 tests across the two recipes covering all four
preprocessing branches, the IFC4X3 source gate, the ObjectType
encoding (incl. author-supplied ObjectType preservation), the summary
RuntimeError shape, and the arc subdivision.
Generated with the assistance of an AI coding tool.
Adds the IFC-library primitives the ifcpatch Migrate recipe needs for a
defensive IFC4 / IFC4X3 -> IFC2X3 downgrade without each caller
reinventing the wheel.
In ifcopenshell.util.schema:
- Migrator(fallback_element_to_proxy=False) opt-in: when True, IFC4-only
IfcElement subclasses (IfcLamp, IfcPipeSegment, IfcGeographicElement,
...) migrate to IfcBuildingElementProxy instead of raising. Default
preserves the strict failure-on-unmappable contract for existing
callers (classification API, etc.).
- geometry_classes_introduced_after(target, source) derives the
IfcRepresentationItem subclasses present in `source` but absent in
`target` directly from the loaded schemas. Cached per pair. Replaces
hand-curated class lists that drift with each IFC update.
ifc4_only_geometry_classes() retained as an alias.
- generate_default_value synthesises a unit IfcAxis2Placement2D /
IfcAxis2Placement3D when downgrading entities whose Position became
required in the target schema (IfcIShapeProfileDef and friends in
IFC2X3).
- Enum-mismatch detection upgraded from string-matched RuntimeError to a
structural check via ifcopenshell.util.attribute.get_enum_items so
upgrade paths still surface real bugs loudly.
In ifcopenshell.util.shape_builder:
- polygonal_face_set_to_faceted_brep converts IfcPolygonalFaceSet /
IfcTriangulatedFaceSet (IFC4-only) directly to IfcFacetedBrep,
preserving topology including IfcIndexedPolygonalFaceWithVoids inner
bounds. Validates inputs at the boundary.
- arc_to_polyline_points approximates a circular arc through three
points with a chord polyline of configurable subdivisions. Tolerates
floating-point noise on planar Z. Raises on non-planar or invalid
inputs.
Test coverage: 47 unit tests across schema + shape_builder lanes
covering each helper directly (no transitive-only coverage), including
regression pins for the IFC4X3-prefix ordering invariant in
get_fallback_schema and the strict-default Migrator contract.
Generated with the assistance of an AI coding tool.
Add include_linked_ifc toggle on BIMSceneClipBoxProperties so the cap
pipeline can also bisect meshes inside Project > Links collection-instance
empties. Off by default - linked IFCs may carry the entire site or
structural backbone, and capping them adds per-mesh bisect cost on every
clip-box edit.
The new iterator composes instance.matrix_world @ inner.matrix_world as
the effective world placement so caps land in the active scene rather
than at the linked library's local origin. Linked-mesh cache entries are
namespaced with a 'link:' prefix to avoid collisions with top-level
scene objects.
Generated with the assistance of an AI coding tool.
Promotes the private _fill_quads_alpha helper from
bim/module/model/decorator.py to tool.Blender.draw_quads so any feature
decorator can reuse the same TRIS-batch path.
The new utility accepts an optional outline_color so callers can draw
fill, outline, or both in a single call. Migrates the only existing
caller (WallGizmoPreviewDecorator in model/wall.py) to the public API
and removes the local helper.
Generated with the assistance of an AI coding tool.
Add source-based clip box presets — a dropdown menu next to the Add
Clip Box button lets the user pre-size a clip box to the bounding box
of a chosen IFC source: a spatial element, IFC class, type, material,
profile, drawing camera frustum, status, system, group, or zone. The
picker dialog uses prop_with_search so files with hundreds of materials
or types remain browsable.
Add interactive face resize handles — six near-invisible click-target
quads render on the active clip box when its empty is the active
object. Dragging a face grows or shrinks the box one-sided on that
axis; the opposite face stays fixed. Ctrl+Click on a face aligns the
viewport to look at that face, following Blender's numpad-view
convention applied to the box's local axes so rotated boxes align
orthogonally to the screen. The gizmos honour negative-scale empties
so the visible cube and the clickable handles stay aligned.
Add settings and info menus — a gear-icon menu next to the Enable
Clipping / Show Caps toggles exposes per-file preferences (cap only
IFC products, show face handles); an info-icon menu adjacent documents
the gizmo gestures. A quick-access toggle row also appears in the
viewport Overlay popover, greyed out when no clip box exists, and
orphaned clip-box list entries now expose an X button so users can
recover from external host-empty deletions.
Plumbing: cap rebuild fires synchronously on gizmo release and
clip-box selection change, so the cross-section overlay re-forms
without waiting for the depsgraph debounce; cap eligibility honours
the "Only IFC Products" toggle. Includes 121 tests covering source
resolution, drag math, face visibility, gizmo registration, and the
view-alignment up-axis convention.
Generated with the assistance of an AI coding tool.
Adds a viewport overlay that traces the connected element path from
the selected wall or MEP element. Walls follow IfcRelConnectsPathElements
and draw each connected wall's reference axis with endpoint dots; MEP
elements follow IfcRelConnectsPorts and draw each segment's axis plus
a port-to-port spider for each fitting.
The new BIMModelProperties.show_paths toggle (Element Paths in the
Bonsai Decorators group of Blender's viewport overlay popover) drives
install / uninstall of both decorators on flip and on file load,
mirroring the show_slab_direction wiring. The popover row also
surfaces the pre-existing BIMSystemProperties.should_draw_decorations
toggle (System Decorations) so both connectivity overlays sit
together.
Dot colors split free endpoints (decorator_color_special, blue by
default) from junction nodes (decorator_color_selected, green by
default) so dangling chain tips read apart from interior joins. Walls
classify endpoints by IFC topology: rels expose RelatingConnectionType /
RelatedConnectionType and ATPATH dots use tool.Wall.path_connection_location_world
for the canonical T-meets join. MEP keeps the geometric classifier
because port positions coincide exactly across fitting + segment.
Generated with the assistance of an AI coding tool.
Edit-mode click-select rejected verts inside the clip volume
because clip_bb stayed at the first-arm view; the depsgraph and
PRE_VIEW handlers updated clip_planes but skipped the
view3d.clip_border call that refreshes clip_bb. Schedule a full
re-arm at transform-commit, IFC reload, and view drift.
The empty's wireframe was clipped by its own planes when rotated
at non-trivial scale because the 1e-6 absolute margin can't
absorb float-precision drift that scales with the box's world
half-extent. Add a 1e-5 relative expand.
Generated with the assistance of an AI coding tool.
Per IFC4+, IfcContext is the abstract supertype of IfcProject and
IfcProjectLibrary; library-only files legitimately contain only
IfcProjectLibrary as their root context. Bonsai assumed an IfcProject
was always present at three crash sites: the parent-library enum
(reported in #8183), RefreshLibrary's tree view, and AddProjectLibrary.
Introduce tool.Project.get_root_context() that prefers IfcProject and
falls back to IfcProjectLibrary, and route the three sites through it.
get_parent_library() now returns None for a root IfcProjectLibrary;
get_project_hierarchy() and the EditProjectLibrary parent-swap branch
handle that. AddProjectLibrary creates the nested sub-library via
IfcRelNests when the root is an IfcProjectLibrary, matching the
existing convention for library-under-library nesting.
For the separate "Open IFC Project" path, abort with a friendly error
pointing users to Project Setup -> Project Library -> Select Library
File instead of letting set_units() crash deep in the importer.
Closes#8183.
Partly generated with the assistance of an AI coding tool.
IfcQuantity* carries an optional Formula (IfcLabel) documenting how a
quantity was derived. Export it alongside each quantity so it survives
in the Quantities column.
The per-quantity entry shape grows from [name, value] to
[name, value, formula], which stays backward compatible for positional
consumers reading index 0/1. Formula is read with a schema-safe getattr
(it does not exist on IfcPhysicalComplexQuantity, nor in IFC2X3) and is
coalesced to "" when absent.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Adds a viewport clip box feature to Bonsai: a user-controllable oriented bounding box that hides everything outside its 6 faces and draws filled cross-section caps where IFC product geometry intersects the planes.
Quality and coordination > Sandbox > Clip Box
A clip box hides everything outside a user-controllable oriented
bounding box, with cross-section caps drawn where IFC product
geometry intersects the planes. The box is hosted on a Blender
empty (CUBE display); its matrix_world is the single source of
truth — G/R/S edits the empty and the viewport clip planes track.
State persists through IFC save/load via a project-level pset
(IfcProject.BBIM_ClipBoxes) so the boxes survive without binding
to any IfcRoot entity (avoids the IFC scale-lock / strip).
UI: BIM_PT_clip_box under the Sandbox tab. Prominent Enable
Clipping + Show Caps toggles at top, then Add, then a UIList with
per-row duplicate / remove icons. Scene-level enabled / show_caps
so the "hide everything outside" intent applies file-wide;
enabled is intentionally not persisted to the pset so reopening
an IFC never silently hides geometry. Adding a clip box arms
clipping so the user immediately sees the cut.
Default spawn at the 3D cursor with scale 10 (a 20 m cube) so the
volume covers a typical building storey or two rather than the
meaningless 2 m unit cube.
Modal-aware: depsgraph + draw-handler paths gate per-frame side
effects on tool.Blender.is_transform_modal_active so dragging
G/R/S on the box only writes the pset once on commit, not per
frame. Shift+D / Alt+D / Ctrl+Shift+D on a clip box gets adopted
as a first-class entry via the collection-to-list sync.
Cap eligibility is gated on IfcElement (walls, slabs, doors, …)
so spatial structure (IfcSpace, IfcBuildingStorey, IfcSite) and
annotations / grids never sprout solid fills at clip boundaries.
Cap rebuild is debounced behind a 1 s quiet window so external
gizmo drags (and any other burst of non-Bonsai depsgraph updates)
collapse to one rebuild on release. Bonsai's own G/R/S keeps the
snappy on-release feel via a modal-end fast-path. The relevance
filter compares a per-Object matrix hash against a baseline so a
plain selection click — which Blender quirkily flags as a
transform update — doesn't churn the cache or flash the caps off.
Edit mode short-circuits both the rebuild scheduler and the draw
handler entirely.
Caps use the evaluated mesh (modifier stack applied) and a
session/matrix/clip-box-hash cache so a typical scene only
re-bisects meshes whose geometry actually changed.
Performance: every per-frame poller (refresh, depsgraph handlers,
draw handlers) short-circuits on the cheapest available check
first — cap_cache emptiness for the post-view draw handler,
scene_props.enabled for the rest — so a session with clipping
disabled pays only one boolean read per tick.
Known v1 limitations documented in tests / docstrings: hollow
profiles cap as solid discs (single-ring tessellation only),
non-watertight inputs may produce degenerate caps, quad-view
untested, Cycles / EEVEE render not supported (GPU-overlay only).
Generated with the assistance of an AI coding tool.
Bisects a BMesh against a set of planes (clear_outer per plane),
then fills the cut edges as cap faces tagged via a BMesh int layer
so the tag survives subsequent bisects. Cut edges are grouped into
connected components before filling so a hollow profile's outer +
inner loops produce two separate cap faces instead of a single
welded outer face that hides the hole. Pre-welds T-junctions
introduced by IFC Boolean meshes so the cut closes into a fillable
loop.
Callers are responsible for input mesh quality. Non-watertight
inputs (terrain, single-shell surfaces) may produce degenerate
cap faces — that's an accepted user-supplied data limitation
which can be revisited if real-world feedback shows it matters.
Used by the clip-box feature to compute cross-section caps per IFC
product mesh.
Generated with the assistance of an AI coding tool.
Adds geometry primitives the viewport clip-box feature needs:
- obb_world_clip_planes / obb_clip_planes_from_matrix: derive the 6
inward clip planes of an oriented bounding box (or unit cube under
a matrix_world) in RegionView3D.clip_planes form. expand / expand_rel
margins let callers visualising the box with overlapping geometry
(an empty CUBE display sharing edges with the planes) keep the box's
own wireframe inside the clip volume.
- point_is_inside_clip_planes / corners_might_cross_clip_planes: cheap
reject tests for the per-mesh capping pass to skip the expensive
bisect when an object's AABB is fully outside the box.
- newell_normal / plane_basis: robust planar-ring normal for thin
near-degenerate cap rings where a two-edge cross product is unstable.
- tessellate_ring_planar: triangulate [outer, *inners] 3D rings in the
outer ring's best-fit plane, with a shapely constrained-Delaunay
fallback for the known failure mode of mathutils.tessellate_polygon
on complex concave polygons-with-holes.
Tests cover unit-box, translated, rotated, and scaled cases for the
OBB-from-matrix builder + the rejection helpers.
Generated with the assistance of an AI coding tool.
The transform-modal active check (Bonsai keymap macros + Blender's
TRANSFORM_OT_* family) was a module-local helper in drawing/gizmos.py
used by per-gizmo poll callbacks. It needs to be shared with other
features that gate per-frame side effects on whether a drag is in
progress (clip box plane re-arming, future modal-aware decorators).
Move BONSAI_TRANSFORM_MACROS and the gate into tool.Blender as
is_transform_modal_active classmethod; widen its window scan to all
WM windows for callers without a window-bound context (depsgraph
callbacks). Leave a thin module-local alias in drawing/gizmos.py so
AST scans and existing call sites stay decoupled from the helper's
home module.
Also add generic Blender helpers needed by the clip-box feature
(reusable by any future feature):
- iter_view3d_regions: yield (area, region, region_3d) for every
WINDOW region in every 3D viewport — for clip-plane / draw-handler
fanout.
- get_or_create_collection: idempotent named-collection lookup +
link to a scene.
- is_in_edit_mode: True iff the active object is in any EDIT_*
mode — for features that need to suspend per-tick work during
vert/edge/face manipulation.
- serialize_matrix / deserialize_matrix / hash_matrix: round-trip a
4x4 matrix as a 16-float CSV string for IFC pset persistence + a
matching hash for cache keys.
Generated with the assistance of an AI coding tool.
* Tolerate stale array child/parent GUIDs
A real-world IFC project (an arrayed door whose host got deleted
externally) crashed Bonsai's project load with "Instance with
GlobalId not found" inside setup_arrays.
tool.Blender.get_object_from_guid declared Optional return but let
RuntimeError propagate; callers iterating BBIM_Array child lists then
crashed instead of skipping. Honour the documented contract by
returning None on miss, matching the convention used by every other
by_guid lookup helper in tool/array.py, tool/ifc.py, tool/geometry.py.
Sweep the four user-action sites that resolve array child/parent
GUIDs without a guard - they shared the same bug class but were
reachable from different operators (regenerate_array, RegenerateArray
clear, duplicate_ifc_objects, process_arrays). An already-missing
entity is the desired terminal state for each, so the fix is
try/except RuntimeError: continue/skip.
setup_arrays now also collects each parent with at least one stale
child GUID into IfcImporter.broken_arrays, surfaced via a new Project
panel banner mirroring the existing pending_opening_recut UX. The
banner reports the count and offers "Select Elements" to navigate to
the affected array parents and a Dismiss button.
constrain_children_to_parent was being called once per layer inside
setup_arrays' for loop even though it always iterates all layers
internally - lifted out of the loop (pre-existing N x perf bug
that the stale-GUID print exposed).
Regression tests:
- test_returns_none_when_guid_not_in_file pins the get_object_from_guid
Optional contract.
- test_remove_array_tolerates_stale_child_guid injects a fake child
GUID into BBIM_Array.Data and asserts bim.remove_array completes
cleanly.
Generated with the assistance of an AI coding tool.
* Black: wrap long bl_description in dismiss_pending_array_repair
Generated with the assistance of an AI coding tool.
serialise_cost_quantities built the "Quantities" JSON string by manual
concatenation, inserting quantity.Name and the related element's Name
without any escaping. A name containing a double quote, backslash or
newline produced invalid JSON, breaking any downstream parser (e.g. a
Typst json.decode consumer reporting "failed to parse JSON"). It also
crashed with a TypeError when a name was None (str += None).
Build a Python list and serialise it with json.dumps instead, keeping
the exact same [[name, value], ...] output shape, the element-name
prefix and the unsupported-type behaviour. None names are coalesced to
"" and quantity values are defensively coerced to float.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Wraps three over-length lines black wanted on the merge-filter +
fillet-lock commit (wall.py's ``either_is_fillet`` chain rewraps the
right-hand ``or`` operand; test_disconnect_elements.py patch-stacks
break each ``patch(`` onto its own continuation line).
Adds per-test docstrings to test_wall_props_resync_on_dim_change.py
and test_wall_split_filled_opening.py so the contract each pins is
visible on grep / on test-run failure output without scrolling to
the module-level docstring. Drops a flip_object sibling-symbol
mention from the module docstring per CLAUDE.md §4a.
Generated with the assistance of an AI coding tool.
DumbWallJoiner.merge previously crashed on walls with a slab underside
clip because the ConnectedTo / ConnectedFrom migration loops assumed
every rel was an IfcRelConnectsPathElements. The slab's
IfcRelConnectsElements(TOP) rel has no RelatingConnectionType /
RelatedConnectionType and raised AttributeError mid-migration. Filter
on rel class; the slab rel dies with element2 via the trailing
delete_ifc_object cascade.
The fillet preview pen icon now also flips the corner's
BIMWallProperties.is_editing so the connection-disconnect gizmos
surface in parallel with the radius drag. CancelWallFilletPreview
clears the flag before tearing the preview state down so both UIs
hide together. GizmoWallUnjoinSingle.poll inlines the viewport +
array-child guards from the topology gate so the gizmo can show
during preview — its own is_editing check is the real gate.
Fillet-to-source-wall path connection icons render in a muted gray
(LOCKED_COLOR) instead of the active disconnect tone, and the
bim.disconnect_elements operator early-returns with an INFO report
("Fillet wall path connections can't be unjoined — delete the fillet
wall element to remove the corner.") when either side resolves to a
fillet corner. The slab clip rel kind stays disconnect-able since
its identity is separate from the fillet's chord-axis reference.
Drive-by /improve polish on adjacent wall.py code: 3 comment tightenings
dropping sibling-symbol names + a defensive ``if opening.ObjectPlacement:``
guard in the merge opening migration matching the pattern used elsewhere
in the same file.
Generated with the assistance of an AI coding tool.
bim.merge_wall now consumes the non-active selection into the active
one — matching Blender's OBJECT_OT_join (Ctrl+J) and MESH_OT_merge
"at last" convention. The wall the user clicks last absorbs the
other; users following Blender muscle-memory get the result they
expect. DumbWallJoiner.merge is already structurally asymmetric
(wall1 = survivor); only the caller in MergeWall._perform needed
flipping. Audit confirmed the previous call site was the sole
caller of DumbWallJoiner.merge in production code.
Drive-by tidies on adjacent code: collapse two over-length comprehensions
under black's 120-char budget, and switch ``any(True for _ in gen)`` to
``any(gen)`` since the iterable yields tuples that are always truthy.
Generated with the assistance of an AI coding tool.
Two bugs in DumbWallJoiner.split's filled-opening branch:
1. Side classification read filling_obj.matrix_world.translation —
flip-fragile because flip_object rotates the filler 180° + translates
so the bbox stays visually in place, moving the door origin to the
opposite bbox corner. A flipped door centred over the cut could be
classified on the wrong side. Switch to the opening's axis-projected
midpoint, which the unfilled-opening loop already uses.
2. When the void straddles the cut and the filling moves to element2,
the void copy for element1 was taken from the rebound new_opening
whose PlacementRelTo had been swapped to element2 — the new void on
element1 then sat in element2's local frame. Reorder so the copy
reads from the original opening (still hosted by element1) before
remove_feature destroys it.
Generated with the assistance of an AI coding tool.
ChangeExtrusionDepth, ChangeExtrusionXAngle, and ChangeLayerLength
mutate IFC extrusion / axis but never re-prime BIMWallProperties from
the post-mutation state. Gizmo icons that position from props.height
then sit at the pre-mutation elevation even though the wall mesh shows
the new one — visible asymmetry against the workspace header H field
which redraws live. Add the existing _resync_walls_after_mutation
call to each operator's epilogue. _maybe_resync_wall_props_from_ifc
already skips non-walls and walls in edit mode, so calling on the raw
selection list is safe.
Generated with the assistance of an AI coding tool.
DumbWallJoiner.merge cascade-deletes element2's HasOpenings via
delete_ifc_object, which previously dropped every IfcOpeningElement
(and any IfcDoor / IfcWindow filling) hosted by the discarded wall.
Re-host each void rel onto the survivor BEFORE the delete fires, and
re-apply the opening's captured world matrix via edit_object_placement
so the void doesn't drift when the two walls have different
placements — a PlacementRelTo swap alone would fail this when origins
differ along the shared axis.
Generated with the assistance of an AI coding tool.
Schematic dimensions float in billboarded viewport space; their labels
carry the value, not the bar length. Snapping the dragged tip to scene
vertices produces nonsensical value jumps when the mouse crosses
unrelated meshes. Add an opt-out flag on the parametric gizmo group
base and override it on the schematic base — every schematic subclass
inherits no-snap behaviour, and in-place parametric gizmos (door,
window, wall, stair, roof, mep) keep the existing Ctrl-toggleable
snap because the default stays True.
GizmoDimension.invoke also forces tool_settings.use_snap = False for
schematic gizmos so the header magnet visibly switches off for the
drag's duration. The existing exit path restores the user's previous
setting on release.
Generated with the assistance of an AI coding tool.
Wall body rebuild + slab underside re-clip are now unified behind
tool.Model.regenerate_wall and called from split / merge / extend
operators. Fillet corner walls accept extend-to-underside (poll +
operator partition switched to is_path_connectable_wall) and surface
the wall-unjoin gizmo without the parametric-edit gate, since
fillets cannot enter that lifecycle. DumbWallJoiner.split strips the
duplicate's inherited slab-trim booleans up front so wall2 lands at
the cut point. regenerate_fillet_corner_wall re-clips after the body
rewrite so a prior extend-to-slab survives neighbour recalcs.
Drive-by bug sweep: tuple typo in hotkey_S_G's IfcSpace check,
defensive .get() in draw_regen_operations for partial AuthoringData
loads, and a try/except in get_active_representation matching the
existing convention for stale mesh ifc_definition_ids after a
representation rebuild.
Tests cover the regenerate_wall branching, the get_active_representation
stale-id contract, and the GizmoWallExtendVertically fillet acceptance.
Generated with the assistance of an AI coding tool.
Add a "Copy" option to bim.cad_offset. When enabled (the
default) it offsets a new copy of the selected edges as
before; when disabled it moves the existing edges to the
offset location instead. The toggle is exposed in the CAD
tool's Offset panel and the operator redo panel.
Generated with the assistance of an AI coding tool.
Deleting a slab that was connected to a wall via IfcRelConnectsElements(TOP)
left the wall holding orphan IfcBooleanResult items + a stale BBIM_Boolean
pset. The disconnect operator already runs the right cleanup; element delete
just never invoked it.
Extract the per-kind cleanup into core.connection.disconnect_rel so the
operator (bim.disconnect_elements) and a new cascade in
tool.Geometry.delete_ifc_object share one dispatch table. Adding a future
rel kind to tool.Connection.find_rels now flows into both call sites
automatically; an AST forward-compat guard enforces coverage.
Other adjustments:
- regenerate_wall_to_underside zero-slab branch now removes stale clip
booleans instead of silently skipping, so disconnecting the last TOP
slab also reverts the wall correctly.
- duplicate_ifc_objects (Shift+D) calls strip_underside_booleans on copied
walls so the duplicate doesn't carry over the source's slab trim, then
reloads the body representation when something was stripped so the
viewport reflects the change without waiting on Shift+G.
- batch_being_deleted_ids threads through OverrideDelete so the cascade
can suppress partner-side regenerate when both endpoints are queued for
deletion in the same batch.
This file was generated with the assistance of an AI coding tool.
Wires slabs into the parametric edit framework (tool.Parametric
.EDIT_TYPES) so the wall-slab disconnect UI gets ESC handling, red
cancel icon, mutual exclusion with other parametric edits, and
per-feature gizmo prefs — all from BaseParametricGizmoGroup — without
duplicating the lifecycle.
Adds:
- ParametricObject("slab") registry entry + tool.Parametric.is_slab
predicate (any IfcSlab).
- BIMSlabProperties with is_editing flag; PointerProperty wired by
the framework's register_object_properties.
- bim.enable_editing_slab / bim.finish_editing_slab /
bim.cancel_editing_slab operators on tool.Ifc.Operator so they
flow through tool.Parametric.run_bim_op cleanly. No IFC mutation
— slab edit is a pure UI gate; finish and cancel share the body.
- tool.Model.get_slab_props accessor.
- GizmoSlabEdition inheriting BaseParametricGizmoGroup with the
pen / validate / cancel triad. is_element_type narrows to
IfcSlab with at least one wall clipped to its underside.
The disconnect-icon group GizmoSlabUnjoinWalls polls behind
_slab_connection_gizmo_poll_gate(require_editing=True), which now
reads is_editing through tool.Model.get_slab_props.
Drops the standalone GizmoSlabConnectionAccess + the
setup_pen_cancel_icons helper added earlier in this branch — both
superseded by the framework integration.
Also folds in the wall + multi-slab gizmo polish requested live:
- Wall side: stack the per-slab unjoin icons vertically (up to 5)
so multi-slab connections each get a distinct clickable icon;
hover-highlight reveals which slab will disconnect.
- GizmoPairDisconnect activates when 2 elements with an
IfcRelConnectsElements(TOP) rel are selected, with the icon at
the wall-slab connection world anchor.
- Wall-slab anchor moved from slab clip Z to wall top +
WALL_SLAB_CONNECTION_Z_CLEARANCE so the disconnect icon perches
above the extend-vertical / slope gizmo instead of overlapping.
- Shared _resolve_active_partner_pair helper for 2-selection
gizmos; _slab_connection_gizmo_poll_gate added to
_REQUIRED_CALLEES + GizmoSlabEdition added to the AST
forward-compat allowlist.
Build note: wall.py's DisconnectElements._perform imports
bonsai.core.connection.disconnect_rel — that core module is being
added in a parallel-session commit. Until that lands the addon
import will fail.
Generated with the assistance of an AI coding tool.
Single generic dispatcher replaces UnjoinWallPathConnection +
DisconnectWallSlab. Takes two GlobalIds, looks up every supported
rel between them via tool.Connection.find_rels, dispatches the right
cleanup by rel kind:
- path (IfcRelConnectsPathElements): remove_connection on every rel
in both orientations + recreate both walls + resync drafts.
- element-top (IfcRelConnectsElements with Description=="TOP"):
disconnect_element + regenerate_wall_to_underside on the wall side
via orient_element_top to recover which input is wall vs slab.
- element (other IfcRelConnectsElements): plain disconnect_element.
tool.Connection lands as a new tool module with two helpers:
- find_rels(a, b): every supported rel between two elements, walking
both ConnectedTo + ConnectedFrom (catches both authoring
orientations and dedups by id).
- find_rel(a, b): first-match convenience.
- orient_element_top(rel, a, b): recovers (wall, slab) from a TOP
rel regardless of which input came first.
Updates GizmoWallUnjoinSingle to target bim.disconnect_elements with
both element_a_guid + element_b_guid pre-filled per icon. Adds the
single registration in tool/__init__.py and the classes-tuple entry
in bim/module/model/__init__.py. Drops the two retired classes.
Tests cover both cleanup branches (path + element-top), missing
endpoints, no-rel-found, and registration smoke.
Generated with the assistance of an AI coding tool.
Counterpart to UnjoinWallPathConnection on the wall-slab side: takes
a wall + slab GlobalId pair, locates the IfcRelConnectsElements(TOP)
between them via tool.Wall.find_wall_slab_rel, removes it via
ifcopenshell.api.geometry.disconnect_element, then re-runs
core.regenerate_wall_to_underside so the wall re-clips against any
remaining connected slabs (the disconnected slab is excluded
naturally because the helper walks tool.Model.get_connected_slab_objs
which filters by the rel set).
Defensive reports replace silent CANCELLED on three error paths the
UI can hit when the gizmo dispatches against stale state: unknown
GlobalIds, wall entity without a Blender object, no rel found
between the resolved pair.
Tests cover all four control flows (happy path + three error paths)
plus a registration smoke that catches a forgotten classes-tuple
update.
A follow-up commit will retrofit this + UnjoinWallPathConnection +
the MEP port disconnects through a unified bim.disconnect_elements
dispatcher with a small connection-type registry; that lands as a
separate single-concern commit so the typed operator can be
reviewed first.
Generated with the assistance of an AI coding tool.
Four classmethods enable the new wall-slab connection gizmo work:
- iter_wall_slab_connections(wall): yields (slab, rel) tuples for
every IfcRelConnectsElements(TOP) on wall.ConnectedFrom — the rel
kind extend_walls_to_underside creates.
- iter_slab_wall_connections(slab): mirror, walks slab.ConnectedTo
so a slab-side gizmo can enumerate every wall clipped to its
underside.
- find_wall_slab_rel(wall, slab): locates the specific rel between
a wall + slab pair so a disconnect operator knows what to remove.
- wall_slab_connection_location_world(wall_obj, slab_obj): returns
the world-space icon anchor — wall axis midpoint X/Y lifted to
the slab's mesh-bbox underside Z. Approximate (uses slab bbox vs
reconstructing the slab's clip plane) but adequate for icon
placement on a wall whose top meets the slab; returns None when
the wall has no IFC Axis representation.
Tests (11) pin the rel-shape contract (class + Description=="TOP",
non-TOP and non-IfcRelConnectsElements rels skipped, None relating
defensively skipped) plus the icon-anchor math (axis-mid lifted to
slab-bbox bottom; None for axisless walls).
Generated with the assistance of an AI coding tool.
update_railing skipped the bmesh rebuild for WALL_MOUNTED_HANDRAIL
railings because the only mesh source available at the time mutated
IFC. The viewport-only preview helper that lands with the parametric
gizmo work (generate_wall_mounted_handrail_preview) sidesteps IFC
entirely, so the WALL_MOUNTED_HANDRAIL branch can join the
FRAMELESS_PANEL path and trigger update_railing_modifier_bmesh on
every property write. Gizmo drag now repaints the viewport in real
time instead of waiting for Finish Editing.
Generated with the assistance of an AI coding tool.
Switches the IfcRailingType terminal-type selector from cycle-on-click
to a popup menu of all terminal-type literals — 5+ values trip the
§2.8 menu-pick threshold. Updates classes registration; removes
EditRailingTerminalType in favour of PickRailingTerminalType which
inherits PickTypeMixin.
Adapts the cherry-pick from db016d881 to post-PR5 framework state:
- Imports CycleTypeMixin / PickTypeMixin / PathPreservingEditMixin
from bim.parametric_lifecycle (PR5 moved them off gizmos.py).
- Routes is_railing through tool.Parametric (predicates moved off
tool.Blender.Modifier between PR3-PR5).
Skips the parametric_lifecycle.py framework refactor the source
commit shipped — HEAD has the more-evolved post-PR5 framework that
already covers it.
Adds the _FakePropsBase + make_lifecycle_obj test helpers to
test/bim/conftest.py so the new test_railing_lifecycle.py can
exercise the edit triad without a real bpy.types.Object. Brings the
test_railing_schematic.py marker in line with the rest of the model
lane.
Generated with the assistance of an AI coding tool.
IFC2X3 representations have no HasShapeAspects inverse; opening the
Geometry & Materials subpanel on an IFC2X3 object raised AttributeError
and left the items list empty. Wrap the access with a getattr default
so pre-IFC4 schemas return an empty iterable, and pin the contract with
an AST forward-compat guard that scans bim/, tool/, and core/ for any
future direct .HasShapeAspects access.
Closes#8157
Generated with the assistance of an AI coding tool.
The check `len(bytedata) == n * 2` was wrong: float64 is 8 bytes per
element, not 2. Legacy float64 checksums fell through to the float32
reader and produced a (2n,)-shaped array, breaking is_moved() and
is_camera_moved() with `ValueError: operands could not be broadcast`
on .blend files saved by Blender <5.0.
Adds a parametrized regression test covering both n=3 (translation)
and n=9 (rotation) for both dtypes.
Generated with the assistance of an AI coding tool.
GizmoWallEdition.position_gizmos used props.anchor_x / props.length
for the in-range check (split icon visibility) and perpendicular
gizmo placement. Those props mirror IFC and are re-primed by
_maybe_resync_wall_props_from_ifc — any operator path that skips
the re-sync leaves the perpendicular gizmo clamped to the previous
wall extent, so the icon parks at the old wall end instead of the
cursor's orthogonal projection. Visible after a wall mutation as
the perpendicular icon landing way off the cursor in top-down view.
Switch to the mesh bbox along local X. recreate_wall rebuilds the
mesh to match the current IFC body on every wall mutation, so
bound_box is authoritative without an explicit props sync.
Generated with the assistance of an AI coding tool.
MEP elements imported as tessellation / brep (no IfcExtrudedAreaSolid
or IfcSweptDiskSolid in their body representation) can't be
parametrically edited — the gizmos offer affordances the geometry
kernel has no path to honour. tool.System.has_parametric_body
inspects the Model/Body/MODEL_VIEW representation and returns True
only when at least one item resolves to one of the two
profile-sweep primitives.
The gate is wired into:
- GizmoMEPActions.is_eligible_object (the action icon group)
- _active_is_flow_segment / _active_is_bend_fitting visibility
predicates the icon row consults per-icon
- GizmoPipeSegmentEdition / GizmoDuctSegmentEdition is_element_type
tool.Parametric.is_pipe_segment / is_duct_segment stay IFC-class-only
so their truth-table contract test keeps reading a single concern.
Generated with the assistance of an AI coding tool.
Pure-math parallelism check (value ≡ 0 mod π within VTX_PRECISION)
that lived as a module-private helper in mep.py belongs next to
tool.Cad.is_x — same comparator family, no MEP-specific knowledge.
Other features with rotation-difference checks (wall fillet, roof
slope, railing terminus) now have a sanctioned spelling.
Generated with the assistance of an AI coding tool.
Four standalone test files pinning contracts the production code
already honours:
- test_mep_actions_cache.py: GizmoMEPActions visibility-predicate
cache evicts on selection or generation change.
- test_mep_bend_preview_cache.py: bend decorator polyline cache
re-uses within a generation and rebuilds on generation bump.
- test_mep_distribution_fit_smoke.py: bim.fit_flow_segments
round-trips a 3-segment polyline without raising.
- test_preview_cancel_ops_forward_compat.py: AST scan ensures every
preview Enable* operator has a paired Cancel* operator with the
matching prop reset.
Generated with the assistance of an AI coding tool.
Two small refactors:
- apply_transform_modal_draw_gate(group, context) replaces the
three-line _is_transform_modal_active + _hide_all_non_modal_gizmos
pair that BillboardingGizmoGroupMixin, BaseParametricGizmoGroup
and BaseSchematicGizmoGroup all repeat in draw_prepare.
- decorator.py renames _stroke_lines_alpha to a public-scope
draw_polyline_segments and drops the no-longer-private companion
docstring reference; the function is now usable by sibling
decorators that draw polyline overlays.
Plus a few one-liner tweaks in tool/model.py and opening.py
following the helper rename.
Generated with the assistance of an AI coding tool.
Two hot paths the gizmo polls fire every viewport event memoise
their result against tool.Parametric.get_geom_generation():
- tool.Blender.Modifier.any_selected_array_child caches the
per-selection scan against the selection identity-set + the
IFC generation token so a stable selection during a drag
doesn't re-walk every selected object's BBIM_Array pset every
frame.
- bim/module/model/wall.py grows a pair-predicate + connection
cache that the wall topology gizmos hit; both keyed on
(pair_uids, predicate_kind, generation) so a wall split or
axis edit invalidates correctly via the generation bump.
Behavioural contract is unchanged — stale entries are evicted
on generation bump; cache miss returns the same value the
un-cached path returned.
Generated with the assistance of an AI coding tool.
bim/module/model/conftest.py exposes the autouse _require_real_bpy
skip-guard, four make_* factories (obj / element / context /
ifc_file), and a patched_tool context-manager factory that wires
the half-dozen tool.* boundary patches every gizmo + decorator
test was repeating.
Existing test files in the directory drop their local copies of
_require_real_bpy and adopt the patched_tool / make_* fixtures
where the call site simplifies — test_mep_port_operators.py is
the biggest beneficiary (−89 LOC).
No production behaviour change.
Generated with the assistance of an AI coding tool.
Three concerns bundled by file boundary (all in mep.py):
- Extract bend preview operators + GizmoBendPreview into a focused
mep_bend_preview.py module; preview_base.py grows the shared helper
set both bend and other previews now consume; classes tuple in
model/__init__.py updated to register the new module.
- Surface ERROR reports on five silent CANCELLED returns in
MEPUnjoinAtPort / MEPRemoveTerminalFitting / MEPUnjoinPair so a
degenerate IFC file ("fitting has no Blender object", "connected
port leads nowhere") shows up in the popup instead of looking like
a no-op.
- DRY: _resolve_active_mep_segment + _require_port_state factor the
segment-id-or-active-object resolve + port-state guard out of every
port operator's prologue; _wire_anchored_icon_targets pulls the
GizmoMEPActions setup() body into an exercise-without-MRO helper so
the wiring-contract tests can hit it without instantiating the
GizmoGroup.
Drops the now-unused preview_base import that the extraction left
behind.
Generated with the assistance of an AI coding tool.
The opening preview's outline used a single-batch two-pass scheme that
dimmed the occluded back pass via alpha=0.25. The visible front pass also
inherited the source decorator color's modest alpha, so the outline read
as subtle on both sides.
Replace with a CAD hidden-line convention: solid full-alpha front pass on
the visible side, world-space dashed back pass on the occluded side. Both
passes use POLYLINE_UNIFORM_COLOR so depth and line-weight paths match.
The dashed batch is built once per object epoch by a new pure helper
tool.Blender.build_dashed_line_segments (pre-segments edges into world-
space dash chunks), then cached via the existing batch-cache mechanism
under "<uid>_dashed".
The solid front pass is rendered at a slightly wider line width than the
dashed back pass so its halo overpowers Blender's WIRE-display overlay
bias at outline pixels — without the asymmetry the wire's anti-z-fight
forward bias makes the LESS_EQUAL comparison narrowly fail and the
dashed pass wins on visible edges too.
Generated with the assistance of an AI coding tool.
MEPConnectElements took obj1_name/obj2_name (Blender object names),
which break when objects are renamed or replicated by array
duplication. Switch to obj1_guid/obj2_guid resolved via
ifc_file.by_guid, with by_guid RuntimeError surfaced as an operator
error rather than a stack trace. DrawPolylineProfile (the sole
in-tree caller) updates to pass GlobalIds.
Generated with the assistance of an AI coding tool.
Pins the geometry contracts the hand-meshed bend body relies on
while IfcSweptDiskSolid round-trip is broken upstream (#8106):
- profile cross-section sampling: circle returns 16 evenly-spaced
points starting at (radius, 0); rectangle returns the four
canonical corners; anything else returns None so the rep swap
is skipped rather than meshed against the wrong section
- parallel-transport framing keeps the cross-section continuous
around L-shaped corners — pinned via start / end ring planes
- initial_basis override seeds the first ring with the source
segment's local +X / +Y axes, fixing the asymmetric-rectangle
twist the world-Z seed produces
Generated with the assistance of an AI coding tool.
Pins which IFC mutation each port operator commits and which
inputs each refuses with CANCELLED:
- MEPUnjoinAtPort removes the fitting + reconnects the two free
ports; refuses if the named port is free or terminal
- MEPRemoveTerminalFitting deletes the terminal element + leaves
the segment's port free; refuses on bridged fittings
- SelectMEPPathMembers walks IfcRelConnectsPorts in both
directions from the active segment and selects every fitting /
segment reachable through the port graph
Boundary mocks for tool.Ifc, tool.System and MEPGenerator stand
in for the IFC fixture; tests assert against the recorded
ifcopenshell.api.* calls.
Generated with the assistance of an AI coding tool.
Pins two regressions the live MEP gizmo group can hit:
- per-icon setup() must write `position` (and `mode` on open-lock
icons) onto every target_set_operator result; the test stands in
for the AttributeError on bim.mep_add_obstruction that surfaced
when a field was dropped from the operator declaration
- each visibility_condition lambda must stay total against None /
non-IFC inputs, since a single raising predicate silently disables
every sibling icon in the group
Generated with the assistance of an AI coding tool.
ProfileDecorator.draw_faces (used by the roof path-edit overlay) and
SystemDecorator.draw_faces called bmesh.ops.triangulate on the live
bmesh — both mutated the input and produced ear-clip fans that rendered
as visible streaks across n-gon roof faces at alpha 0.1. The opening
DecorationsHandler edit-mode branch had a separate bug: it computed
triangles from obj.data.calc_loop_triangles() while iterating the
edit-mode bmesh, so any topology added mid-edit desynced the indices.
Centralise the correct draw path on tool.Blender.draw_bmesh_face_tris
(wraps bm.calc_loop_triangles, non-mutating, beauty triangulator) and
route all three call-sites through it. A forward-compat AST guard walks
every *Decorator / DecorationsHandler class under bim/module/ and pins
the no-bmesh.ops.triangulate rule against future regressions.
Generated with the assistance of an AI coding tool.
A user clicking the pen icon on a typed-product occurrence whose body
representation is mapped from its type would silently mutate every
sibling occurrence's geometry. Add a confirmation dialog at the pen-icon
dispatcher (the single chokepoint every feature routes through) showing
the sibling count, with a session-scoped suppress checkbox.
The check is read-only: tool.Model.get_sibling_occurrence_count wraps
tool.Geometry.get_elements_by_representation against the resolved body
rep and subtracts self + type. A forward-compat AST guard pins the
dispatcher monopoly so any future feature that binds pen_gizmo directly
to a feature-specific enable op fails the test before merge.
Generated with the assistance of an AI coding tool.
Once a bend was created, the only way to retune start_length /
end_length / radius was to delete and recreate from scratch.
EnableBendPreviewFromBend re-opens the preview on an existing
parametric bend: it walks the bend's ports to resolve the two
connected segments, reads start / end length and radius from the
bend type's BBIM_Fitting pset, and sets editing_bend_id on the
preview props. MEPAddBend then deletes the old bend + its port
connections (single undo step) before the recreate path runs, so
finish replaces the bend in place and cancel discards the edit
without touching the original.
GizmoMEPActions surfaces a pen icon on single bend-fitting
selections via the new _active_is_bend_fitting predicate; the icon
dispatches the new operator. Mirror of the wall fillet re-edit
flow (EnableWallFilletPreviewFromCorner + editing_corner_id in
CreateWallFillet).
Test coverage: registration probe for the new operator, an attached
editing_bend_id field probe on the preview umbrella, and a
parametrized truth-table for the _is_bend_fitting predicate
(IfcFlowFitting with BEND PredefinedType, with other PredefinedType,
with no type, IfcFlowSegment, IfcWall, None).
Generated with the assistance of an AI coding tool.
Wall topology mutations (merge / join / extend-to-wall / unjoin /
fillet) applied to a Bonsai array child are silently overwritten by
the next ``regenerate_array``; merge also orphans a GUID listed in
the parent's ``BBIM_Array.Data``. Add a central
``tool.Blender.Modifier.any_selected_is_array_child`` predicate and
gate the five wall topology gizmo groups plus the six bound operators
behind it. Operator gating is defence in depth against keymap / F3
invocation paths that bypass the gizmo.
The base ``_wall_gizmo_poll_gate`` keeps its loose two-check shape
(viewport gizmos + no preview). A new
``_wall_topology_gizmo_poll_gate`` wraps it with the array-child
filter and is what the topology gizmos use. Host-opening gizmos
deliberately stay on the loose gate: openings authored on a child
are preserved through ``regenerate_array`` and track with the
replicated instance.
A forward-compat AST guard walks wall.py for ``GizmoGroup`` subclasses
and asserts each routes its poll through the tighter gate or the
central predicate, with an allow-list for the parametric-edit and
preview-owner exceptions. New wall topology gizmos inherit the
contract by construction.
Generated with the assistance of an AI coding tool.
Selecting a Bonsai-parametric IfcDoor now shows the swing arc(s)
without entering edit mode. A new viewport decorator polls on the
active object, reads the door's BBIM_Door pset, and draws the same
arcs the parametric door swing gizmo would draw — matching the
hinge / panel-width / x-mirror contract minus the is_editing gate.
A forward-compat test walks every door operation type and cross-
checks the readonly decorator's arc selection against the gizmo's
swing-arc config table, so future enum additions fail in both
surfaces simultaneously.
Also disables the inherited 8-pass dark halo on GizmoArc: an open
curve has no enclosed silhouette, so the offset passes read as
ghost arcs rather than a uniform outline. The arc's own cross-
section thickness keeps it legible without the halo.
Generated with the assistance of an AI coding tool.
The MEP bend feature's IfcSweptDiskSolid representation produces
geometrically correct output but fails to round-trip through the
OpenCascade geometry kernel (upstream issue #8106) — the body is
dropped on the next file load. Until upstream is fixed, MEPAddBend
captures the bend centerline in world space before the segments are
extended (otherwise the post-extension axes no longer reach the
original intersection and arc reconstruction is wrong), then after
the fitting is placed it hand-meshes the bend body and swaps the
type's swept-disk representation for an IfcTessellatedFaceSet via
tool.Geometry.export_mesh_to_tessellation + tool.Model.
replace_object_ifc_representation.
The centerline includes the straight start_length / end_length legs
in addition to the arc so the bend covers the full segment-to-
segment span. Sweep uses parallel-transport framing — each ring's
(right, up) basis is rotated by the minimum rotation that maps the
previous tangent to the current one, eliminating the twist a fixed
world-axis reference produces when the tangent crosses the
reference. Cross-section orientation seeds from the source segment's
matrix_world local +X / +Y so asymmetric IfcRectangleProfileDef
ducts land with XDim / YDim on the same axes the segment expects;
parallel transport then preserves that alignment around the arc.
Centerline radius is radius + profile_dim[lateral_axis] to match
MEPAddBend's ref_point_radius — without this offset, the bend legs
fall short of the extended segments by profile_dim * tan(angle/2).
Face winding is left to the caller to correct via
bmesh.ops.recalc_face_normals on the closed bend tube.
Two FIXME(#8106) markers (capture site + helper call site) so both
can be dropped once upstream lands a swept-disk round-trip fix.
Generated with the assistance of an AI coding tool.
GizmoWallEdition gains a fourth cursor-anchored icon that
spawns a perpendicular branch wall from the cursor's
orthogonal projection on the source wall axis. Click forms
a T-junction; shift+click forms an L-corner with the source
wall trimmed at the projection, keeping its longer portion.
The branch inherits the source's spatial container and
centerline baseline so its authored axis matches the source's
alignment rather than the type's default.
Also includes a floor-plane preview quad for the new gizmo,
a floor-Z cross line on the split preview for top-down
visibility, a small bump to QUAD_ALPHA for clearer preview
fills, and a stacking-offset helper that centralises the
cursor-row screen-up step across three call sites.
Generated with the assistance of an AI coding tool.
The MEP one-shot operators (join, unjoin variants, terminal removal,
path-select, obstruction add/remove) had no viewport surface. This
commit adds GizmoMEPActions — the icon-action gizmo group that
surfaces them as billboarded icons around selected MEP elements.
Three anchor regions: a horizontal row above the bbox top
(selection-cardinality icons), per-port endpoints for the three-state
lock / unjoin icons (open lock for PORT_FREE, closed for
PORT_TERMINAL, unjoin for PORT_JOINED — resolved per-frame from
port_connection_state), and the predicted join location
(compute_mep_join_location, shared with the bend preview) for the
join / unjoin_pair pair. Unjoin icons render at full
DEFAULT_BILLBOARD_SCALE with warning-red hover; endpoint lock icons
shrink so the lock row stays subordinate to the row icons. The
group hides itself entirely while a bend preview is active.
MEPAddObstruction grew a position enum (CURSOR / START / END) and a
mode enum (ADD / REMOVE / TOGGLE) so the gizmo can target a specific
port without touching the cursor and dispatch ADD or REMOVE based on
the click target — the lock_open icons drive ADD with position
pinned, the lock_closed icons drive bim.mep_remove_terminal_fitting.
Without the new fields the gizmo wiring (op_props.position = ...)
crashed at setup() with AttributeError on the obstruction operator.
validate_bend_preconditions extracts the type-match and profile-kind
checks MEPAddBend enforces so EnableBendPreview surfaces the
rejection immediately — the user no longer tunes a preview only to
learn at commit time that the segments use an unsupported profile
(e.g. IfcArbitraryClosedProfileDef).
Generated with the assistance of an AI coding tool.
Four discrete one-shot operators driven by the MEP segment's port
state. mep_unjoin_at_port deletes the IfcFlowFitting bridging a
segment's named port to a second element when the port is in the
JOINED state. mep_remove_terminal_fitting deletes the terminal
fitting at a port (closed-lock state) and dispatches by fitting
type — OBSTRUCTION fittings go through MEPGenerator.remove_obstruction
so the segment absorbs the freed length, other terminal fittings go
through the standard delete path. mep_unjoin_pair finds the single
fitting bridging two selected MEP segments and deletes it.
select_mep_path_members walks the connected MEP network from the
active element via IfcRelConnectsPorts and replaces the selection
with every reachable member. Foundation for the MEP Actions gizmo
group which surfaces these operators as icon affordances around
selected segments.
Generated with the assistance of an AI coding tool.
The extend icon used a pure screen-space billboard that always
pointed +X across the screen — the arrow ran horizontally
regardless of the pipe / duct's orientation. The new
billboarded_along_axis helper rotates the gizmo about the camera-
forward axis so its local +X aligns with the segment's local +Z
projected onto the screen, keeping the icon camera-facing but
visually following the extrusion direction. The flip-mirror branch
now reads from cursor-vs-current-end along the segment axis (not
screen-X), so the arrow points away from the current endpoint
regardless of viewport orientation. The split icon stacks
perpendicular to the rotated extend arrow in screen space so the
two don't overlap.
The decorator's green preview line no longer clamps the cursor
projection to min_projected_length — it follows the raw projection
so the line stays visible when the cursor crosses behind the
segment origin (the user still sees where they're pointing even
though the operator floors the actual commit).
Generated with the assistance of an AI coding tool.
The bend preview gizmo group (commit 2) populated a Scene draft but
the user saw nothing in the viewport until they hit finish — they
had to commit blindly. This commit ports the BendPreviewDecorator
(centerline arc + two leg projections on valid geometry, warning-red
axes on invalid in-segment intersections) and the interactive
GizmoBendPreview group (three dimension widgets for start_length /
end_length / radius plus validate / cancel icons). The bend axis
math lives in a pure compute_bend_preview_polylines helper, fed
into both the gizmo group's per-frame positioning and the GPU
decorator's draw path. MEPSegmentExtendPreviewDecorator lands at
the same time because it shares the decorator install / uninstall
plumbing — renders the extend-to-cursor preview line for the
GizmoPipeSegmentEdition / GizmoDuctSegmentEdition extend icons
when hovered, clamping the projected endpoint to the operator's
minimum so the preview matches where the commit lands. The
MEPJoinSegments dispatcher routes two selected MEP segments to
mep_add_transition (parallel) or enable_bend_preview (non-parallel)
— the F3 search entry point that makes the bend preview testable
before the gizmo-icon dispatch lands.
11 new tests in test_mep_bend_preview.py cover the geometry helper
truth table (parallel rejection, right-angle happy path, near-
collinear rejection, in-segment invalid_axes), the
_intersection_past_near parametrized boundary, registration probes
for the lifecycle operators / join dispatcher / gizmo group /
decorator, and the FinishBendPreview RuntimeError catch contract.
6 extend-preview-line tests (deferred from commit 3) join the
existing 35 in test_mep_segment_edition.py.
Generated with the assistance of an AI coding tool.
Spatial.get_root_element walks aggregate / nest / filled-void /
voided-element chains and core.assign_container assigns the container
to whatever the walk returns. For an IfcDoor the filled-void hop
redirects to the IfcOpeningElement, then voided-element to the host
wall, so a user who selects a door and runs bim.assign_container ends
up targeting the wall — and silently no-ops on the door if the wall is
already in the target storey.
Per IFC4 / IFC4.3 (IfcDoor, IfcWindow): the spatial containment of a
filling is defined independently of the filling relationship. Major
exporters (Revit, ArchiCAD, Tekla, Allplan) emit independent
ContainedInStructure on doors / windows accordingly. Drop the
filled-void / voided-element hops from the walk; aggregate and nest
remain — those are true sub-part relationships where the parent
legitimately owns the container.
New TestGetRootElement in test/tool pins the new contract (filling
resolves to itself) plus the retained aggregate / nest / loose-element
paths so a future PR that re-adds either hop is caught. Two new
TestAssignContainer cases in test/core pin filling-to-self through the
core layer and per-element can_contain filtering.
Generated with the assistance of an AI coding tool.
Pipe and duct segments had no parametric-edit affordance — the only
length edit path was a property panel value with no live preview.
This commit ports the per-segment parametric edit triad
(enable / finish / cancel) plus a cursor-anchored extend operator
and a cursor-projected split operator into one gizmo group per
segment type. The two PropertyGroups (BIMPipeSegmentProperties,
BIMDuctSegmentProperties) host the draft length plus snap fields
so cancel / no-op-finish restore the segment to its exact pre-edit
visual state including a non-identity pre-edit scale. Length
commits are written through DumbProfileJoiner.set_depth and
auto-dispatch bim.regenerate_distribution_element so adjacent
fittings track the port move. The split operator preserves
downstream port connectivity and runs through tool.Ifc.run for
single-step undo. The two segment types are now first-class
entries in tool.Parametric.EDIT_TYPES, which resolves the FIXME
on auto-commit-on-save dispatch.
35 unit tests cover predicate truth tables, segment_world_length
geometry, preview-via-scale / restore-scale helpers, gizmo class
wiring, lifecycle operator registration, dimension matrix_position
rotation respect, and lifecycle drift-handling. The 6 extend-
preview-line decorator tests stay deferred until the bend preview
decorator commit lands MEPSegmentExtendPreviewDecorator.
Generated with the assistance of an AI coding tool.
Parametric gizmos (wall/door/window/stair/roof/array/MEP) recompute
matrix_basis every frame from obj.matrix_world. While Blender's
transform modal (G/R/S and the Bonsai macro overrides) drags the
matrix, the gizmos slide off-cursor and fight the transform overlay.
Detect via context.window.modal_operators (Blender 4.2+) — the
collection of running modal operators. Gate poll() (forward-compat)
and draw_prepare() (production path: gizmo.hide=True preserves the
GizmoGroup across the drag instead of destroying it). Cover the
Bonsai macro override for G key (and Shift/Alt/Ctrl+Shift+D) by
matching the BIM_OT_* macro idnames that surface in modal_operators.
Forward-compat test walks every parametric-edit module for GizmoGroup
subclasses and asserts poll returns False with the detector mocked,
so new gizmo groups inherit the hide automatically.
Generated with the assistance of an AI coding tool.
Every preview operator (commit + cancel for both bend and wall
fillet) was inlining the same 3-4 line cleanup: set is_active to
False, zero every *_id IntProperty. The new clear_preview_state
helper in preview_base.py introspects bl_rna and applies that
contract generically — adopters become a single call. Two new tests
pin the contract: every *_id IntProperty zeroes, non-id fields stay.
Generated with the assistance of an AI coding tool.
Extract the pending_opening_recut tracking, three operators (apply /
dismiss / select), Project-panel banner, and the sibling
multi-instance warning banner (its backend helpers already landed
on this branch) from commit a85ed6032 on gizmos-8088.
All tool.* dependencies (Geometry.reimport_element_representations,
Blender.set_objects_selection, Array.*) and IfcImporter.gross_elements
are already on this branch -- no other diffs from a85ed6032 are
pulled.
The source's narrow except-tuple paraphrase comments are trimmed
to keep only the durable "don't swallow programmer errors" note,
per CLAUDE.md s4a.
Tests: 5 bim-lane tests in test/bim/module/project/
test_pending_opening_cuts.py covering apply happy-path + missing
entity, dismiss, select happy-path + cancellation.
Generated with the assistance of an AI coding tool.
MEPAddBend exists on the main flow but commits bend geometry with
hardcoded defaults (start_length=0.1, end_length=0.1, radius=0.2)
with no opportunity to tune before commit. The new scene-level
BIMBendPreviewProperties hosts a draft (start_segment_id,
end_segment_id, start_length, end_length, radius); EnableBendPreview
populates it from the two selected MEP segments after asserting they
are non-parallel, FinishBendPreview dispatches MEPAddBend with the
tuned values and clears the draft, CancelBendPreview discards it.
Scene-level placement follows CLAUDE.md 2.9: a bend creates a new
fitting entity between two segments, so neither segment alone owns
the draft. Foundation for the upcoming bend preview gizmo group and
decorator.
Generated with the assistance of an AI coding tool.
When an operator mutated IFC then raised mid-execute the user was left
staring at a raw traceback with the IFC graph captured by the active
transaction but the Blender side stale. Blender does not push an undo
step for a raised operator (the same gap that the CANCELLED-modal arm
patches via bpy.ops.ed.undo_push), so the WARNING the framework can
emit is only honest if it pushes that undo step too. The framework
now detects partial state via ifc_file.transaction.operations,
pushes a Recover undo step, then reports a WARNING naming Ctrl+Z so
the recovery path is discoverable. The bespoke try/except wrapper in
UnjoinWallPathConnection becomes redundant and is retired in the
same change.
Generated with the assistance of an AI coding tool.
IfcTriangulatedFaceSet/IfcPolygonalFaceSet were introduced in
Fix#7992: IFC4 and do not exist in IFC2X3. Previously, requesting an
IfcTessellatedFaceSet representation in an IFC2X3 file silently
fell back to a faceted brep after unassigning material sets.
Add a guard in the update_representation operator (user-facing
error) and in the add_representation API (ValueError) so the
unsupported request is caught instead of failing silently.
Generated with the assistance of an AI coding tool.
Three files flagged by black --check on the lint-formatting job:
* bim/module/geometry/operator.py — single-arg `.update(...)` rejoined
onto one line under the 120-char budget.
* test/bim/module/model/test_wall_gizmos.py — same join on a
_make_path_rel call.
* test/modal/test_modal.py — pre-existing baseline noise picked up
via the upstream merge: PEP-8 blank-line separators between top-
level functions, `0.68+` → `0.68 +`, double quotes, trailing
whitespace stripped.
No behavioural change; pure whitespace.
Generated with the assistance of an AI coding tool.
Some distributions (e.g. Fedora) ship only a shared RocksDB that exports
RocksDB::rocksdb-shared rather than RocksDB::rocksdb. The CMake target
selection now falls back to the shared target when the static one is absent.
Newer RocksDB also changed DB::Open and DB::OpenForReadOnly to take
std::unique_ptr<DB>* instead of DB**. IfcFile.cpp uses SFINAE tag dispatch
to build against both old and new APIs without version detection.
CGAL 6.x deleted operator< from Point_d, so std::map<Point_d, ...>
no longer compiles. Adds a custom lexicographic comparator and updates
the three affected maps in snap_halfspaces and snap_halfspaces_2.
Fixes builds with newer GCC/libstdc++ that no longer provide <cstdint>,
<cstring>, <cfloat>, <memory>, <algorithm> etc. transitively. Also
disambiguates visit<> calls in taxonomy.h with the full namespace and
casts the character value in IfcCharacterDecoder to uint32_t to silence
ambiguous overload warnings.
The toggle_openings icon lived outside the IconSlot layout — each
host (wall, roof) declared an ad-hoc setup_pen_row_toggle_openings_icon
+ update_pen_row_toggle_openings_icon pair, and GizmoArrayEdition
queried a hardcoded _FEATURE_IDLE_MAX_X dict to position past it.
On an arrayed wall the dict was shadowed: find_for_element returns
"array" before "wall" in EDIT_TYPES order, the wall reservation was
never consulted, and the first per-layer ARRAY icon (local X=0.37)
landed 13cm from the wall's toggle_openings (X=0.50) — visually on
top of each other.
Promote idle-row icons into the slot system instead of patching the
dict:
* IconSlot gains an Optional visible_when predicate for state-driven
visibility (toggle_openings only when the host carries openings).
* BaseParametricGizmoGroup gains idle_slots: ClassVar[tuple[IconSlot]]
+ _idle_slot_x_positions() + _idle_row_right_edge() helpers; the
setup + idle-branch positioning loops mirror the existing
feature_slots path.
* Wall and roof declare toggle_openings as an idle_slot and drop
their ad-hoc setup/update calls.
* GizmoArrayEdition's _resolve_feature_idle_max_x walks
BaseParametricGizmoGroup.REGISTRY and takes the max
_idle_row_right_edge() across peers whose poll passes — no more
hardcoded dict, no more find_for_element-order shadowing.
* setup_pen_row_toggle_openings_icon + update_pen_row_toggle_openings_icon
helpers deleted from drawing/gizmos.py.
* 3 forward-compat AST guards pin the new contract.
Also bundles an unrelated array-test fix: TestUsingArrays in
test/tool/test_model.py was asserting against bpy.context.selected_objects
which is a fragile signal after remove_array / apply_array. A new
_array_objects() helper filters bpy.data.objects via the BIM_Array
pset's IfcActuator type instead.
Layout on an arrayed wall after the fix:
pen X = 0.00
toggle X = 0.50 (idle_slot 0)
array[0] X = 0.87 (one ICON_ARRAY_GAP past idle row)
array[1] X = 1.27
All separated by the standard inter-icon spacing.
Generated with the assistance of an AI coding tool.
In plan view world-Z collapses to zero on screen, so every wall-edit
icon anchored on the floor — the projected 3D cursor, wall endpoints,
wall-to-wall corners, IfcRelConnectsPathElements connection points —
projects onto the click target it represents. The result on a typical
extend / split / unjoin action: the icon sits on top of the cursor
crosshair (or the corner the user wants to click), defeating precise
positioning.
Add shared ``gizmo.top_down_clearance(context, billboard_rot)`` to
bim/module/drawing/gizmos.py: returns a screen-up Vector in top-down
view (cosine cone around world Z, matching ``is_view_top_down``) and a
zero Vector elsewhere, so call sites apply it unconditionally before
``billboarded_at``. Default distance 0.4 m aligns with the inter-icon
stack spacing already used by GizmoWallJoinIntersection so single
icons and stack bases land at consistent screen-up positions when
multiple groups render around the same wall endpoint.
Apply at the seven wall-edit anchor sites:
* GizmoWallEdition cursor stack (top-down branch only — non-top-down
already stacks along world-Z at structural points clear of the
cursor).
* GizmoWallExtendVertically (single icon at wall origin endpoint,
active-object Z elevation).
* GizmoWallJoinIntersection corner stack base + merge midpoint.
* GizmoWallUnjoinSingle link-toggle pool (one icon per IFC path
connection, previously sitting exactly on the connection point).
* GizmoWallFilletReedit pen icon at fillet corner.
* GizmoWallFilletToggleOpenings.
The clearance is a pure visual offset — bound operators still read
the world-space anchor (cursor / endpoint / connection point) at
execute time, so the action's target is unaffected.
Also tighten GizmoWallUnjoinSingle: gate poll on ``props.is_editing``
so the link-toggle icons only surface during the wall edit lifecycle
(matching every other edit-row icon), and downsize them via a new
``ICON_SCALE = 0.35`` constant since 16 of them at default scale
cluttered the viewport on path-heavy walls.
ruff + black clean. Wall gizmos test lane 14/14 pass.
Generated with the assistance of an AI coding tool.
bim/handler.py was importing two feature-module internals
(wall_offset_gizmos.clear_caches, preview_base.discard_pending_previews)
to drain load-transient parametric state alongside the existing
tool.Parametric.heal_stale_edit_flags() call inside
_apply_save_file_invariants. Each new parametric drain added one
top-level import and one inline call — every load_post drain leaked
into handler.py's namespace.
Hide all three drains behind tool.Parametric.on_load_post(scene),
sited adjacent to heal_stale_edit_flags. The two feature-module
imports become late imports inside on_load_post — same pattern as
refresh_post_commit's existing `import bonsai.bim.handler` — which
sidesteps the tool.parametric -> bim.module.model.preview_base ->
bonsai.tool registration-time cycle.
The forward-compat AST contract that pinned "every module-scope
GenerationKeyedCache + clear_caches MUST be drained on load_post"
follows the call site to its new home — the test now walks
tool.Parametric.on_load_post instead of _apply_save_file_invariants.
No behaviour change. 45/45 affected bim tests pass
(test_handler_forward_compat, test_preview_base,
test_wall_offset_gizmos, test_parametric_registry).
ruff + black clean on all touched files.
Generated with the assistance of an AI coding tool.
Three feature-specific decorators previously lived in
bim/module/model/decorator.py despite owning state only their
home module reads:
* ArrayPreviewDecorator + ArraySelectionHighlightDecorator +
draw_array_layer_children_bbox -> array.py (read array
edit-state props and walk BBIM_Array psets)
* WallGizmoPreviewDecorator + draw_wall_partner_bbox -> wall.py
(dereference wall.py-private classes and helpers via lazy
imports)
decorator.py keeps cross-cutting infrastructure
(BoundingBoxDecorator, SlabDirectionDecorator, WallAxisDecorator,
WallFilletPreviewDecorator, PolylineDecorator, ProductDecorator)
and the shared bbox primitives (bbox_world_edges,
draw_polyline_segments, _BBOX_EDGES, _stroke_lines_alpha,
_fill_quads_alpha) that several feature files now import.
handler.py and gizmos.py update their import paths; the
wall-feature lazy imports inside WallGizmoPreviewDecorator
methods collapse to direct references now that the decorator
lives in wall.py.
No behaviour change. Wall lane 37/37, array lane 15/15, wall
forward-compat 6/6, parametric-registry 8/8 still pass.
Generated with the assistance of an AI coding tool.
Three concerns sharing the same architectural theme (collapse inline
bbox / edit-state lookups, drop overrides that re-do base-class work):
== Bbox helpers and array operator DRY ==
* tool/blender.py: add a "dimensions" tuple key to both
get_object_bounding_box and get_object_world_bounding_box return
dicts. The (max - min) per-axis extent — which callers previously
computed via local helpers — is now a key alongside min_x / max_x
/ min_point / max_point / center. Distinct from Blender's built-in
obj.dimensions (which folds object-level scale): the local variant
is the intrinsic mesh bbox extent; the world variant is the
matrix_world-applied AABB.
* bim/module/model/array.py: drop the local _bbox_dims helper; the
two callers now read tool.Blender.get_object_bounding_box["dimensions"]
directly.
* Rename _parent_geometry_changed -> _array_children_need_rebuild.
The old name suggested "did the parent change just now", implying
the function was a parent-edit-finish trigger. It actually runs
only inside the array-edit-finish path as a drift safety net (the
upstream-deliberate design — see commit 83d97d7e9 "Fix #7616. Make
regenerate array an operator instead of an array preference" —
means the array doesn't auto-regen when its parent geometry edits
finish). New name matches the call-site phrasing
``if X: _wipe_array_children(layers)`` and clarifies that this is
a children-state check, not a parent-edit trigger.
* Extract _resolve_array_edit_props(context) — returns the active
object's array props during an active edit lifecycle, or None.
Collapses the obj-active-then-is-editing prologue (3 lines + return)
to one resolver call across 4 sites: ToggleArrayMethod.execute,
AdjustArrayCount.execute, RemoveArrayLayerFromEdit._execute and
.poll. Each call site shrinks from 7 lines to 3.
* Migrate two inline bbox reads inside GizmoArrayEdition to the new
dict keys: get_axis_world_face_center collapses the manual
xs/ys/zs min/max + center math to bbox["center"] + bbox["max_x"] /
["max_y"] / ["max_z"]; get_element_height collapses
``max(corner[2] for corner in obj.bound_box)`` to
tool.Blender.get_object_bounding_box(obj)["max_z"].
The _BBOX_EQUALITY_EPS = 1e-5 tolerance stays inline as a single-
consumer constant — no other call site needs tolerance-equality on
dimension tuples, so extracting it to a shared util would be
speculative abstraction.
== Drop dead code ==
* GizmoArrayEdition.update_editing_gizmos override + its
_has_other_parametric_type helper: redundant with
hide_pen_button = True at line 1024. The base class already hides
the pen in every idle case (when hide_pen_button is truthy) AND in
every editing case (unconditionally). The override's conditional
hide-when-parametric only re-hid a pen that was already hidden in
both branches. Removes the only remaining path that could re-show
the array's pen icon; array-edit entry is now uniformly via the
per-layer ARRAY icons (which is the documented preferred
affordance, see the hide_pen_button comment).
* _wall_fillet_preview_active in wall.py: defined but never called.
_wall_fillet_props (the sibling thin-wrapper around
preview_base.get_preview_props) is heavily used; the
is_preview_active wrapper was added speculatively and never picked
up a consumer.
Generated with the assistance of an AI coding tool.
Bundled bug fixes + the forward-compat AST guard that prevents the
underlying class of bug from coming back.
* bim/module/model/wall.py: FinishEditingWall._execute early-returns
CANCELLED when props.is_editing is False. Without this guard, a
failed enable (e.g. on a wall without IfcMaterialLayerSetUsage)
leaves is_editing False but a press on finish still walked the
sub-ops below, which dereferenced layer-set-dependent state and
crashed.
* tool/model.py: Model.offset_wall now guards against
ifcopenshell.util.element.get_material returning None before
calling .is_a("IfcMaterialLayerSetUsage"). Fixes the pre-existing
test/bim/module/model/test_wall_header_refresh.py crash that has
been the only failing test in the wall lane since this branch
started.
* bim/handler.py: _apply_save_file_invariants drains
wall_offset_gizmos.clear_caches() on load_post. The module-scope
GenerationKeyedCache instance survives the .blend reload; without
the drain the cache may serve entries whose bpy_struct references
point into the freed bpy.data of the previous file.
* test/bim/test_handler_forward_compat.py: AST-walk test that
enumerates every bim/module/model/*.py source declaring both a
module-scope GenerationKeyedCache assignment AND a top-level
clear_caches function, and asserts each module appears as a
<module>.clear_caches() call in _apply_save_file_invariants. Pins
the contract: any future module-scope geom cache that exposes
clear_caches must wire into the load_post drain.
* test/bim/feature/model.feature + test/bim/test_feature.py: wall
edit-lifecycle scenarios switch from "add cube + assign as
IfcWallType" to "load the demo construction library + add an
occurrence of the WAL100 wall type", so the parametric edit runs
against a real LAYER2 wall with IfcMaterialLayerSetUsage rather
than a vanilla-mesh promotion that lacks one. The demo-library
step also picks the schema-matching library file (IFC2X3 /
IFC4 / IFC4X3) so the appended types remain valid across schemas.
Door saved-height assertion updates from 2.5 → 2500 to reflect
that BBIM_Door pset stores project units (METRIC_MM in the
empty-project fixture).
Generated with the assistance of an AI coding tool.
bpy.ops.bim.new_project(preset='demo') crashed in
refresh_bim_tool_headers: the post-commit hook fired for every
nested bpy.ops.bim.append_library_element during template
loading, and the operator context Blender hands to
programmatically-invoked nested operators is stripped of the
view-layer attributes the refresh reads.
Two changes resolve it.
Gate the header refresh in tool.Parametric.refresh_post_commit
on operator.bl_idname being one of the EDIT_TYPES finish_op
idnames. Only validate-gizmo commits (bim.finish_editing_<name>)
now trigger the refresh; demo-loader and other non-edit
operators skip it. Querying the registry directly is the
canonical signal — string-prefix matching would silently drift
if ParametricObject.finish_op changes derivation.
Harden tool.Blender.get_active_object so its view_layer fallback
also uses getattr; the 150+ callers routed through it now
tolerate stripped contexts. _resolve_bim_tool_context applies
the same defensive pattern to mode / workspace.
Tests:
- test_handler_restricted_context covers get_active_object's
defensive path and the BimTool-family whitelist (excludes
annotation, spatial, structural).
- test_handler_forward_compat AST-pins that the gate consults
EDIT_TYPES (not a string prefix).
- test_wall_header_refresh rewritten — three tests cover the
gated-by-registry contract: counter bumps for every commit,
finish_op operators refresh headers, others don't.
Hotkey-driven in-place edits (S_E / C_E) no longer trigger the
refresh — they were caught by the pre-refactor "every commit"
design. Left out of scope; the new skip-non-finish test pins
this as intentional.
Generated with the assistance of an AI coding tool.
Two related cleanups bundled because each was too small on its own.
== Drop dead Geometry.has_material_styles duplicate ==
Two parallel has_material_styles implementations existed on HEAD:
* Geometry.has_material_styles (tool/geometry.py:853, added by
3483683cb "Add tool.Geometry helpers for body representation +
placement"): checks each material via tool.Material.get_style
for an IfcSurfaceStyle. This is the implementation gizmos-8088
uses — its core/root.py:58 calls geometry.has_material_styles.
* Root.has_material_styles (tool/root.py:75, added by e76455913
"Route _has_material_styles through tool.Root.has_material_styles"):
checks each material for a HasRepresentation inverse. Added to
fix the test/core/test_root.py::TestCopyClass::test_AAAAAAAAAAAA
failure by routing the check through a Prophecy-mockable seam.
HEAD's core/root.py:59 calls root.has_material_styles. The Geometry
version became orphaned by that migration — zero callers historically
(git log -S "Geometry.has_material_styles" returns nothing). The
Root placement is the right architectural home: has_material_styles
pairs with assign_body_styles in the copy_class flow as "is there
material-defined styling? if not, apply body styling" — both
decisions live on the same interface, called in sequence from the
same caller.
The semantic delta (HasRepresentation vs IfcSurfaceStyle) is a close
approximation in real IFC files where HasRepresentation almost always
indicates a styled material; if precision becomes necessary, the
Root impl can be tightened independently of this cleanup.
Drop the Geometry method + its abstract declaration in core/tool.py.
== Sanitation sweep per CLAUDE.md §4a ==
Eight rot-prone references in code we authored on this branch get
their first-draft mistakes cleaned up. The §4a rule (no sibling
symbol names, no test paths, no motivation history in docstrings)
got added during this branch, so older commits sometimes named their
siblings in prose; this is a focused cleanup of the worst offenders.
* bim/module/model/wall.py:201 — _CommitWallDraftsFirstMixin
docstring carried motivation history ("...that every multi-wall
operator … used to repeat at the top of _execute"). Rewrite to
describe only the current contract.
* bim/module/model/wall.py:1910 — cycle_type_operator comment named
two sibling methods. Rephrase to describe what happens at the slot.
* bim/module/model/wall.py:2025 — _active_instances ClassVar comment
named WallGizmoPreviewDecorator. Rephrase to "the wall-gizmo
preview decorator" (role, not class).
* bim/module/drawing/gizmos.py:3402 — GizmoFillet hit_uses_bbox
comment named GizmoWallJoinIntersection. Rephrase to "the wall-join
gizmo group".
* bim/module/drawing/gizmos.py:3887 — GizmoCountLabel docstring had
a :meth:`set_count` cross-reference. Drop — reader sees the method
next to the class.
* bim/module/model/host_add_opening_gizmo.py:201 — poll-exclusion
comment named GizmoWallEdition + GizmoRoofEdition. Rephrase to
describe why we skip ("walls and parametric roofs both render
their own toggle in the pen row").
* bim/module/void/operator.py:45 — preserve_placement comment named
FilledOpeningGenerator.generate. Rephrase to "the filling-opening
generator gates its snap-to-wall-axis block on this flag".
* bim/parametric_lifecycle.py:64 — module docstring named the test
file path (test/bim/test_parametric_registry.py). Rewrite to
"enforced by the registry contract tests".
Sweep otherwise clean: no third-party software names in this-branch-
authored comments (upstream Revit / Tekla / ArchiCAD references are
legitimate external-constraint workarounds, §4a-allowed). No
PR/issue numbers we authored except the FIXME(PR5) in
tool/parametric.py:150, deliberately preserved until PR6's MEP slice
resolves it.
Generated with the assistance of an AI coding tool.
The 7 multi-wall operators (UnjoinWalls, UnjoinWallPathConnection,
ExtendWallsToUnderside, ExtendWallsToWall, SplitWall, MergeWall,
JoinWallsIntersection) each opened their _execute with an identical
prologue:
_commit_pending_wall_edits_for_selection(context)
# ... operator-specific logic
— flushing any in-progress wall parametric drafts so the operator
acts on committed IFC state rather than the draft preview box.
Extract that prologue into _CommitWallDraftsFirstMixin: its _execute
calls the commit helper, then delegates to a subclass-supplied
_perform. Subclasses inherit the mixin first in their bases tuple so
the mixin's _execute resolves first via the MRO. The IFC transaction
opened by tool.Ifc.Operator.execute still wraps both the commit and
the perform.
Behaviour-equivalent — same call, same order, same selection scope.
Architectural cleanup only: a future multi-wall operator can no
longer forget the commit step. The named helper
_commit_pending_wall_edits_for_selection stays as the single
encapsulation of the names=("wall",) filter; its docstring loses
the stale "every multi-wall operator calls it at the top of
_execute" sentence and now just describes the filter contract.
Matches gizmos-8088's _CommitWallDraftsFirstMixin pattern.
Generated with the assistance of an AI coding tool.
The regular bim.add_opening click on the host-add-opening gizmo
(wall + door/window co-selected) routes through
FilledOpeningGenerator.generate, which snaps the filling to the
wall's reference-line axis, optionally rotates 180° when the
filling sits on the opposite side, and re-applies an rl1 / rl2
Z-elevation default. That is the right default for "drag a fresh
door onto a wall and let the model place it for me", but defeats
the workflow where the user has already positioned the filling
precisely (e.g. snapped to a window in an adjacent wall, copy-
pasted at an exact Z, aligned to a reference object).
Holding SHIFT while clicking the gizmo now opts into a
"preserve placement" mode: the filling stays at its current
matrix_world and the opening is created at the filling's existing
position. The opening / filling rels and representation work are
unchanged — only the snap-to-axis branch is skipped, so the IFC
graph is identical to the regular click; only the spatial
position of the filling differs (user-chosen vs auto-snapped).
Implementation:
* bim/module/void/operator.py: AddOpening gains a hidden
preserve_placement BoolProperty + an invoke() that sets it from
event.shift. The call into FilledOpeningGenerator.generate
forwards the flag. bl_description documents the SHIFT modifier
so it surfaces in F3 search / hover tooltip.
* bim/module/model/opening.py: FilledOpeningGenerator.generate
accepts preserve_placement (default False — backwards-compatible
with the other caller, tool.Model.add_filled_opening). The
voided_obj.data-gated snap block (raycast + axis projection +
rl-Z default + filling_obj.matrix_world write) skips entirely
when the flag is True. The opening's matrix_world reads from
filling_obj.matrix_world below the gate, so the opening lands
at the filling's preserved position automatically.
Generated with the assistance of an AI coding tool.
The door / window / stair edit-row's type-cycle icon advanced one
type per click (CycleDoorType / CycleWindowType / CycleStairType
bound to cycle_type_operator). DoorType has 8 IFC variants,
WindowType 9, StairType 3 — so cycling past the target was the norm.
Threshold rule for cycle-vs-menu: cycle is appropriate for exactly 2
values (advance-one-per-click stays predictable). Three or more
values warrants a popup menu. Door / window / stair all qualify;
roof (RoofGenerationMethod has 2 values) keeps cycle. Wall has no
type cycle. Array is unaffected.
Swap to the popup-menu pattern (PickTypeMixin already on HEAD at
bim/parametric_lifecycle.py:442): clicking the icon opens a menu
listing all type_literal values; selecting one applies it in a
single undo step. The hamburger icon (VIEW3D_GT_menu) is wired into
BaseParametricGizmoGroup.setup_editing_gizmos whenever
pick_type_operator is set (mutually exclusive with
cycle_type_operator). Matches gizmos-8088's pattern exactly.
Per-feature shape:
* door.py: PickDoorType replaces CycleDoorType.
GizmoDoorEdition.cycle_type_operator → pick_type_operator.
* window.py: PickWindowType replaces CycleWindowType. Same swap.
* stair.py: PickStairType replaces CycleStairType (no
tool.Ifc.Operator inheritance — stair-type changes
BIMStairProperties only, no IFC mutation). Same swap.
* bim/module/model/__init__.py: registration entries renamed
Cycle* → Pick*.
* bim/module/drawing/gizmos.py: drop the
CycleTypeMixin / PickTypeMixin / TypeAccessorBase shim re-export —
its own docstring already noted "PR5 cleanup drops these" and the
three callers (door / window / stair Cycle*Type) it served are
gone. Roof's CycleTypeMixin import was already direct from
bim.parametric_lifecycle. Also update GizmoMenu docstring to
reflect the 2-vs-3+ threshold.
Generated with the assistance of an AI coding tool.
When entering parametric edit on a door or window that fills a
wall opening, four dimension gizmos now measure the distances
from the wall edges to the filling's jambs and from the wall's
base/top to the sill/header. Dragging any gizmo translates the
filling along the wall's local axis; 180°-flipped fillings and
slanted LAYER2 walls round-trip correctly. The has_host_wall
predicate hides all four when the filling → opening → wall
chain cannot be resolved.
Generated with the assistance of an AI coding tool.
Four concerns that together make the cursor-anchored gizmos
(extend_x_gizmo, extend_z_gizmo, split_gizmo on GizmoWallEdition)
fully functional and visually informative without entering parametric
edit mode first:
* Drop the props.is_editing gate in _update_cursor_gizmos. The three
bound operators (bim.extend_wall_to_cursor,
bim.extend_wall_height_to_cursor, bim.split_wall_at_cursor) already
poll on wall-selected and commit any pending wall edit before
acting, so single-click without entering edit mode is now the
canonical flow. Matches gizmos-8088's always-on behaviour.
* Register GizmoWallEdition instances in a per-region weakref map
(_active_instances) populated at setup_element_specific_gizmos
time. The WallGizmoPreviewDecorator dereferences this map to read
live is_highlight state off the cursor icons. Without the
registration its _cursor_icon_hovered always returned False and
the hover-gated GPU previews silently never drew. Mirrors the
same pattern already in place on GizmoWallJoinIntersection.
* Add post-operator resync to all three cursor operators
(_maybe_resync_wall_props_from_ifc for the single-wall split /
extend-height paths, _resync_walls_after_mutation for the
selection-wide extend-X path). Without this, props.length /
props.height stayed stale after the operator ran, so the
orientation flips _apply_wall_extend_flips computes from
cursor_local vs wall dimensions kept using the pre-extend values
until the next selection change. Matches gizmos-8088's pattern.
* Hover-gated GPU previews per icon:
- extend-X: filled Z=0 floor quads spanning the wall's offset to
offset+thickness Y band, visible from plan view without side-
view clutter. Grow case (cursor beyond either endpoint): one
green decorator_color_selected quad over the extension. Shrink
case (cursor inside extent): green quad for the portion that
REMAINS + red decorator_color_error quad for the portion the
operator REMOVES.
- extend-Z: vertical lines at the cursor's projected X in the
wall's y=0 reference-line plane. Grow case (cursor above wall
top): one green segment from z=height to z=cursor.z. Shrink
case: green from z=0 to z=cursor.z (REMAINS) + red from
z=cursor.z to z=height (REMOVES).
- split: one red vertical line at the cursor's projected X from
base to wall top — the cut plane.
Quads use QUAD_ALPHA=0.25 so the underlying wall body stays
visible.
* New module-level _fill_quads_alpha helper next to
_stroke_lines_alpha, plus a per-decorator _fill convenience method
and a _wall_floor_quad corner builder.
Modal-active gizmo hiding (is_gizmo_hidden_by_modal) is preserved.
Generated with the assistance of an AI coding tool.
Add openings — GizmoWallAddOpening only fired when a wall was active +
co-selected with a non-host; slabs and roofs got no in-viewport handle.
GizmoHostAddOpening covers all three host types via is_supported_host,
dispatching walls to the axis-projection anchor and slabs/roofs to a
world-Z anchor lifted just above the host's top face (predictable
height regardless of the void's vertical position).
Show openings on hosts with their own parametric-edit toolbar —
GizmoRoofEdition gains an idle-row toggle_openings_gizmo parallel to
the wall's, parked at the cancel-slot X next to the pen. Visible only
when the host carries HasOpenings and the edit triad is idle. Roof
overrides get_element_height to return the mesh's world-AABB top in
object-local Z, so the WHOLE pen-row anchors visibly above sloped or
stepped roof bodies. The wall's idle-row toggle now also hides when
HasOpenings is empty.
Show openings on hosts WITHOUT a parametric-edit toolbar —
GizmoHostToggleOpenings scoped strictly to the fallback case: a single
host selected, HasOpenings non-empty, NOT a path-connectable wall, NOT
a parametric roof. Covers slabs today plus any foreign-authored IfcRoof
without BBIM_Roof. Anchored at object origin XY + world-AABB top Z.
When slab parametric-edit eventually lands, the slab predicate joins
the exclusion list and this gizmo's poll narrows automatically.
Operator move — ToggleWallOpenings was already host-agnostic; renamed
to ToggleHostOpenings in opening.py (bl_idname bim.toggle_host_openings).
Three callers (the wall idle-row binding, GizmoWallFilletToggleOpenings,
and workspace.py's hotkey_A_O for Alt+O) now route through the renamed
operator. The Alt+O binding is surfaced in the operator's
bl_description so it appears in F3 search and hover tooltips.
DRY refactors —
* GizmoWallAddOpening deleted (subsumed by GizmoHostAddOpening)
* tool.Blender.get_object_world_bounding_box added as the world-AABB
sibling of the existing local helper; 3 inline call sites in
tool/misc.py (set_object_origin_to_bottom, scale_object_to_height)
and gizmos.py adopt it (2 other sites in drawing/operator.py and
project/operator.py inherently need raw transformed corners for
per-corner plane / NDC tests — not AABB candidates)
* BaseParametricGizmoGroup gains setup_pen_row_toggle_openings_icon +
update_pen_row_toggle_openings_icon; wall + roof + any future host
gizmo wire up the idle-row toggle with two one-line calls
* _resolve_active_host shared poll prologue between the two host
gizmos (gate + selection count + active-in-selected + entity lookup
+ supported-host check)
* HasOpenings non-empty checks at 3 sites route through
tool.Geometry.has_openings
* hotkey_A_O body collapsed to bpy.ops.bim.toggle_host_openings()
The forward-compat AST guard pinning "must accept fillet-corner walls"
retargets from GizmoWallAddOpening.poll to is_supported_host.
Generated with the assistance of an AI coding tool.
The extend-X / extend-Z / split icons share the cursor's projected X
on the wall axis, separated only by world Z (floor / cursor / wall
top). World Z collapses to a single screen point in plan view, so
every icon piled onto extend-X's hit target and only the topmost was
clickable.
Two refinements ported from gizmos-8088:
* When ``tool.Blender.is_view_top_down(context)`` reports the camera
is near plan-view, swap world-Z stacking for screen-up stacking:
anchor all icons at the floor world position and offset each by
``index * CURSOR_STACK_OFFSET`` along ``tool.Blender.get_screen_up_world(context)``.
Each icon lands in its own screen-space slot regardless of view
rotation.
* In the same top-down branch, drop ``extend_z_gizmo`` entirely. A
vertical-intent gizmo has no readable cue when looking down +Z —
clicking it would mutate the wall in a direction the user can't
see change.
* Bonus: split's local Z now goes through
``core.extrusion_depth_from_vertical_height(props.height, props.x_angle)``
so the icon lands on the slanted top edge of sloped walls (x_angle
!= 0) instead of the vertical-height target the wall isn't at.
All three helpers (``is_view_top_down``, ``get_screen_up_world``,
``extrusion_depth_from_vertical_height``) already on HEAD from PR2/PR3.
Non-top views unchanged — same world-Z stacking + cascading bumps as
before.
Generated with the assistance of an AI coding tool.
Hover-gated viewport preview lines that show where a wall-join /
extend / split operator would land before the user clicks. Four
preview paths, each gated on a specific icon's ``is_highlight`` state:
* **Join intersection** — two LAYER2 walls selected in the ``intersect``
state (non-joined, non-collinear, non-parallel). Draws four lines:
each wall's axis at both base and top Z, extending from the wall's
nearer endpoint to the projected XY intersection. The pair of lines
per wall communicates the full plane the join welds at, not just
the floor edge.
* **Cursor extend** — single LAYER2 wall, hover on ``extend_x_gizmo``.
One line from the wall's nearer X endpoint to the cursor's projected
X on the wall axis.
* **Cursor extend-Z** — hover on ``extend_z_gizmo``. Vertical line at
the cursor's projected X from wall base to cursor Z (the new total
height).
* **Cursor split** — hover on ``split_gizmo``. Vertical line at the
cursor's projected X from wall base to wall top — the cut plane.
Warning-red colour matches the icon's destructive-action signal.
Hover colour rules for the join preview:
* **Join or Fillet hover** → all four lines highlight in
``decorator_color_selected``. Both icons commit a symmetric corner
meet, so every line is part of the operation.
* **Extend-to-Wall hover** → only the non-active wall's two lines
(base + top) highlight. The default-direction extend operator
moves the non-active wall into the active one's axis; only that
wall's preview should signal motion.
* No hover → all four lines in ``decorations_colour``.
Three coordinated changes:
* ``bim/module/model/wall.py`` gains the ``_classify_wall_join_state``
wrapper over ``core.classify_wall_join_state`` (feeds the
``_are_walls_joined`` flag the core helper expects) AND a
``_active_instances`` per-region weakref ClassVar on
``GizmoWallJoinIntersection`` populated in ``setup()``. Without the
weakref registration, the decorator's ``_lookup_active_instance``
call returns None every frame and the hover gates silently
evaluate False — the symptom would be preview lines that never
switch colour. Both pieces ported from gizmos-8088.
* ``bim/module/model/decorator.py`` gains
``WallGizmoPreviewDecorator`` (~280 LOC across the four preview
paths + shared helpers ``_stroke`` /
``_active_layer2_wall_for_gizmo_preview`` /
``_join_group_hover_state`` / ``_extended_wall_index``). All
cross-file dependencies (``core.classify_wall_join_state``,
``core.wall_join_preview_lines``, ``_stroke_lines_alpha``,
``_cursor_icon_hovered``, ``_lookup_active_instance``,
``tool.Parametric.is_path_connectable_wall``,
``_wall_axis_world_segment_from_geom``) already on HEAD.
* ``bim/handler.py`` wires ``WallGizmoPreviewDecorator.install()`` /
``.uninstall()`` alongside the other always-on preview decorators.
The decorator self-polls every frame; cost is one selection-count
check + one ``is_highlight`` read when no eligible state is active.
Verified: headless smoke green, ruff + black clean. Live testing
confirms the four preview paths fire correctly when hovering each
icon.
Generated with the assistance of an AI coding tool.
GizmoWallJoinIntersection used to place its icons at state-specific
world points: join at floor Z, extend-to-wall at the active wall's
top Z, fillet stacked screen-up above join. Same XY at different Z
collapses to a single screen pixel in plan / top view, so two icons
became one hit target — invisible from above.
* position_gizmos now always-stacks along screen-up at a wall-top
anchor in both the joined (unjoin + fillet) and the intersecting
(extend + join + fillet) states. Order bottom-up is
extend / L / fillet. Collinear-merge keeps its single boundary
icon (no stack needed).
* New _stack_anchor_z picks the active wall's top Z (or the taller
of the two on mid-selection-transition frames). New _stack_at
lays a tuple of icons along screen-up at the resolved anchor.
* Glyph swap: join_icon -> VIEW3D_GT_wall_corner (L), extend_to_wall_icon
-> VIEW3D_GT_wall_tee (T). Both classes already existed in
bim/module/drawing/gizmos.py from an earlier commit; only the
setup() bl_idname strings changed. The previous arrow-merge /
arrow-extend pair read as the same direction once stacked.
Forward-compat AST contracts in test_wall_gizmos_forward_compat.py
pin the new invariants: the L and T bl_idnames must appear in
setup(), and position_gizmos must route through _stack_at so a
regression that reintroduces a direct billboarded_at write for any
state-specific icon fails CI before it flattens the stack again.
Also folds in a one-line typo fix in core/spatial.py:
assign_container's per-element can_contain check iterated `e` but
predicate-tested `root_element` (the outer for-loop variable), so
every element in the comprehension was tested against the same
container/element pair. Switch the argument to `e`.
Generated with the assistance of an AI coding tool.
When validate_type selected a preferred_item from remaining_items
(e.g. the sole IfcBooleanResult in a representation), it left that
item in the list. The subsequent Items filter removed every item,
leaving Items=[] and causing guess_type to return
"MappedRepresentation" — silently corrupting the representation.
Also removes temporary debug print statements added during
investigation of the wall-to-slab extension workflow.
Generated with the assistance of an AI coding tool.
Re-running the operator on the same wall/slab pair created
additional IfcPolygonalFaceSet booleans each time. Now each
wall's existing booleans are removed before re-clipping, and
previously connected slabs are merged with the new selection
so no earlier clips are silently discarded.
Generated with the assistance of an AI coding tool.
When Shift+G is pressed on a LAYER3 element, any LAYER2 walls
connected via IfcRelConnectsElements(TOP) are now re-clipped
to the slab's updated geometry after recalculate_slab runs.
Generated with the assistance of an AI coding tool.
extend_walls_to_underside now accepts multiple slab/roof
objects in a single operation — all selected non-LAYER2 IFC
elements are treated as clip targets, all LAYER2 elements as
walls. Placement sync is done once upfront; each wall is then
clipped against every selected slab before reloading.
Also adds bim.regenerate_wall_to_underside (Shift+G): after
moving a slab, re-clips connected walls using the existing
IfcRelConnectsElements(TOP) relationship. Old booleans are
removed via remove_representation_item before re-clipping.
Generated with the assistance of an AI coding tool.
When extend_walls_to_underside is applied to a wall and the
roof/slab is later moved, pressing Shift+G now re-clips the
wall to the slab's new position.
The IFC relationship created by connect_wall_to_slab
(IfcRelConnectsElements, Description="TOP") is used to look
up which slabs a wall is clipped to. On regeneration, the
existing manual booleans (IfcPolygonalFaceSet operands) are
cleanly removed via remove_representation_item, then
clip_wall_to_slab is re-applied for each connected slab.
Shift+G on a LAYER2 wall that has a TOP connection now calls
bim.regenerate_wall_to_underside; walls without a connection
continue to call bim.recalculate_wall as before.
Generated with the assistance of an AI coding tool.
When the operator was called twice on the same wall for a
ridge roof, the two IfcPolygonalFaceSet clip solids shared
an exact ridge edge (kissing-solid). OCCT produced spurious
extra vertices at the coincident boundary.
Fix by building the clip solid from a rectangle on the slope
plane that extends slightly past the face edge (1 project
unit margin) rather than the exact face footprint. Adjacent
slope solids now volumetrically overlap at the ridge instead
of sharing a boundary face, which OCCT handles correctly.
Generated with the assistance of an AI coding tool.
ifc.get_object(element) can return None for IFC elements that aren't loaded as Blender objects (e.g., decomposed sub-elements).
The loop now skips those instead of passing None into collector.assign().
Cheers!
The temporary-offset workaround (#7408, commit bd57cc8735) subtracts the
directrix centroid (`mean`) from the curve points before building the
sweep near the origin, then must add it back to restore the original
location. The restore negated the sign — `Move(-mean)` instead of
`Move(+mean)` — placing the swept solid at -mean (mirrored through the
origin) rather than its true position.
Only triggers for polyline directrixes (`is_polyhedron()`) whose centroid
is more than 100 m from the origin (`mean.norm() > 1e2`), so models
centered near the origin are unaffected. Models that keep absolute site
coordinates (e.g. many Revit/ODA IFC exports) render affected swept
solids — reinforcing bars, pipes — at a mirrored phantom location far
from the rest of the model.
Completes the PR4/PR5 cleanup the FIXME at tool/blender.py
flagged: every is_<type> / Array.<helper> shim on
tool.Blender.Modifier delegated one-for-one to tool.Parametric /
tool.Array. Callers now reach the canonical home directly, and the
shim block — seven is_<type> classmethods plus the inner class Array
— comes out.
Renames (no semantic change):
* tool.Blender.Modifier.is_<door|railing|roof|stair|wall|window>
→ tool.Parametric.is_<x>
13 sites across tool/loader.py, bim/import_ifc.py,
bim/module/geometry/{data,operator}.py, bim/module/model/{door,
railing,roof,stair,ui,wall,window}.py.
* tool.Blender.Modifier.Array.<helper> → tool.Array.<helper>
4 sites across tool/root.py, bim/import_ifc.py,
bim/module/geometry/operator.py.
* test_parametric_registry.py: the two getattr probes that hunt
predicates by name now look on tool.Parametric. Docstring + the
test function name (test_every_entry_has_modifier_predicate →
test_every_entry_has_parametric_predicate) follow the move.
Kept on tool.Blender.Modifier (non-shim, no equivalent on
tool.Parametric): try_applying_edit_mode,
try_canceling_editing_modifier_parameters_or_path,
is_eligible_for_<x>_modifier (×5), is_array_child, is_slab.
Verified: 109 model-lane tests + 8 parametric-registry tests pass
(the one pre-existing failure in test_wall_header_refresh.py is
unrelated — it patches handler.update_bim_tool_props which has been
renamed). git grep for tool\.Blender\.Modifier\.(is_<type>|Array\.)
returns empty. black + ruff clean on every touched file.
Generated with the assistance of an AI coding tool.
Ports roof parametric edit gizmo group from gizmos-8088 and folds in
three roof-mesh bug fixes surfaced during live testing.
Port:
* ``CycleRoofGenerationMethod`` operator (bim.cycle_roof_generation_method)
cycles props.generation_method between "HEIGHT" and "ANGLE". Shift+click
cycles in reverse via the ``CycleTypeMixin`` contract.
* ``GizmoRoofEdition`` gizmo group: 3 dimension gizmos for height
(visible in HEIGHT mode) / slope angle with tan/atan2 rise round-trip
+ degree formatter (ANGLE mode) / roof_thickness. All three handles
anchor at the object's local origin and separate visually via their
declared axes (height/slope +Z, thickness -Z) — height + slope are
mutually exclusive via ``visibility_condition`` so they never paint
at the same time. Anchoring at the origin sidesteps the first-click
default-identity-matrix symptom that footprint-derived anchoring
would have hit on a stale ``RoofData`` cache.
* Lifecycle factory swap: explicit ``EnableEditingRoof / CancelEditingRoof
/ FinishEditingRoof`` classes replaced by ``tool.Parametric.build_edit_lifecycle("roof", _RoofEditMixin, ...)``.
Same bl_idnames out, no external caller changes.
* Registration: ``CycleRoofGenerationMethod`` + ``GizmoRoofEdition``
added to ``bim/module/model/__init__.py`` classes tuple.
* Tests: ``test_roof_gizmos.py`` covering slope round-trip, visibility
gates, cycle operator metadata, and origin-anchored positioning.
Bug fixes:
* ``generate_hipped_roof_bmesh`` flipped the bottom slab face's normal
at low slope angles. The kernel's outward-inference becomes
ambiguous on near-flat geometry once ``remove_doubles`` and
internal-face deletion run, and the early ``recalc_face_normals``
pass at line 389 ran BEFORE the topology was final. A second pass
on the final closed mesh fixes the eave plane (now reliably points
down regardless of slope).
* ``bpypolyskel.polygonize`` can emit a face whose vertex list
contains the same index twice on certain footprint/slope
combinations (a straight-skeleton ridge collapse). ``bm.faces.new``
rejects those with ``found the same (BMVert) used multiple times``,
aborting the whole rebuild. Filter the degenerate faces out so the
rest of the roof renders.
* ``_RoofEditMixin._restore_viewport_after_cancel`` now rebuilds the
bmesh from the just-restored draft via ``update_roof_modifier_bmesh``.
The hook was abstract on ``PathPreservingEditMixin`` and raised
``NotImplementedError`` on cancel-after-edit, leaving the user
stranded.
Also folds in a parallel ``tool/loader.py`` swap from
``tool.Blender.Modifier.is_railing`` to ``tool.Parametric.is_railing``
(consistent with the rest of the loader using ``tool.Parametric.*``).
Verified: headless smoke green, test_parametric_registry.py 8/8,
test_roof_gizmos.py 15/15. ruff + black clean on the touched files.
Generated with the assistance of an AI coding tool.
Three wall-gizmo fixes:
* GizmoWallFilletPreview crashed on every draw_prepare after the
DRY-colors refactor moved decoration lookups onto
self.get_decoration_colors() — that method lives on
BillboardingGizmoGroupMixin / BaseParametricGizmoGroup, but
GizmoWallFilletPreview inherited only from bpy.types.GizmoGroup.
setup() AttributeError'd silently, leaving radius_dim and friends
unset. Add the mixin to the bases; rename _position_gizmos to
position_gizmos so the mixin's refresh/draw_prepare dispatch lands
correctly and drop the now-redundant overrides.
* GizmoWallAddOpening's poll gated on the strict is_wall predicate,
which rejects fillet-corner walls (no LAYER2 usage by IFC spec).
Switch to is_path_connectable_wall on both the active and the
partner-exclusion checks so the add-opening icon surfaces over
curved corners — matching every other wall-state gizmo's host gate.
* Show / hide openings was only available on LAYER2 walls because
GizmoWallEdition's parametric edit pipeline (which carries the
toggle) refuses fillet bodies. Add GizmoWallFilletToggleOpenings,
a dedicated single-icon group that polls on is_fillet_corner_wall
and reuses bim.toggle_wall_openings — the body stays untouched.
Forward-compat AST guards in test_wall_gizmos_forward_compat.py pin
both invariants: every wall GizmoGroup that calls
self.get_decoration_colors() must inherit a mixin that provides it,
and GizmoWallAddOpening.poll must keep using the looser predicate.
Generated with the assistance of an AI coding tool.
tool.Parametric.refresh_post_commit was calling update_bim_tool_props
after every IFC mutation. The function does two things — refresh
read-only header values (extrusion_depth/length/x_angle) and re-target
user-intent enums (ifc_class, relating_type_id) from the active object.
Doing both on the commit path crashed on IfcAnnotation actives (the
type isn't in the bim_tool ifc_class enum) and silently overwrote the
user's "what to build next" choice on every other element.
Split the function: update_bim_tool_props remains selection-driven and
does both halves; new refresh_bim_tool_headers is header-only and is
what refresh_post_commit now calls. Behaviour on selection change is
preserved. Also ports the upstream PR #8136 try/except guard onto the
props.ifc_class write for the selection-driven path. Adds
test_handler_forward_compat.py to pin both contracts via AST.
Generated with the assistance of an AI coding tool.
Two bugs in BIM_PT_array:
1. The "is this layer in edit mode" predicate compared a BoolProperty
against an int (props.is_editing == i). Python evaluates False == 0
as True, so layer 0 always rendered the per-layer edit form even
when no edit was active — clicking validate/cancel then dispatched
against a phantom edit state. Switched to
props.editing_item_index == i, which defaults to -1 and matches
exactly one layer when an edit is active.
2. The panel's CHECKMARK and CANCEL buttons called bim.edit_array /
bim.disable_editing_array, a parallel lifecycle that only cleared
editing_item_index. Entering edit mode via the viewport gizmo
(bim.enable_editing_array, the triad enter) sets is_editing=True
and hides array children; the legacy panel exit unwound neither —
so committing or cancelling from the panel left is_editing=True
with children hidden, and the viewport gizmo thought the edit was
still in progress. Re-bound both panel buttons to the canonical
triad operators (bim.finish_editing_array /
bim.cancel_editing_array), which _ArrayEditMixin already owns and
which the viewport gizmo group already uses. Panel and gizmo now
share one exit path.
The three now-unreachable operators are deleted with their
registration entries: EditArray (bim.edit_array), DisableEditingArray
(bim.disable_editing_array), and EnableEditingArrayItem
(bim.enable_editing_array_item, never called from any UI). The two
test/tool/test_model.py sites that drove bim.edit_array as a commit
step are switched to bim.finish_editing_array.
External scripts or user keymaps bound to bim.edit_array /
bim.disable_editing_array will need to update — the replacements are
bim.finish_editing_array and bim.cancel_editing_array, both taking no
parameters (the layer is read from props.editing_item_index).
Partly generated with the assistance of an AI coding tool.
The recent per-gizmo-prefs cleanup left ``update_swing_gizmos`` with a
stale ``prefs`` reference that raised NameError mid-refresh, so the flip
arc's ``matrix_basis`` was never reassigned and the gizmo drifted to the
world origin. SINGLE_SWING_RIGHT also lacked an X-mirror on the primary
arc, so the swing extended past the door's right edge instead of
sweeping back over the panel.
Five related fixes / additions:
* Drop the leftover ``prefs.decorations_colour[:3]`` per-frame colour
override (the setup-time ``decorator_color_special`` is the durable
contract — there's no reason to overwrite it every refresh).
* Add X-mirror to RIGHT-hinged single-panel transforms so the arc
sweeps back over the door rather than past the right edge.
* Treat DOUBLE_DOOR_SINGLE_SWING as a two-panel layout: 4 arcs total
(left + right panels, each with its own Y-mirrored flip) scaled to
``overall_width / 2``.
* Hide all swing arcs for SLIDING_TO_LEFT / SLIDING_TO_RIGHT /
DOUBLE_DOOR_SLIDING — sliding doors don't swing. A slide-direction
indicator is deferred to a separate change.
* Pin ``select_bias = -1000.0`` on every arc gizmo so the big
quarter-arc hit shapes don't steal clicks from the smaller dimension
and edit gizmos drawn on top.
Architectural cleanup driven by the same diff: the imperative
4-create + 50-line update block is replaced by a declarative
``swing_arc_props`` list of ``SwingArcConfig`` entries (mirrors the
existing ``dimension_gizmo_props`` pattern). Setup iterates the list
and creates one (main, flip) pair per entry under
``gizmo_swing_arc_<name>`` / ``gizmo_swing_arc_<name>_flip``; update
iterates the same list and positions each pair via the lambdas. Adding
a hypothetical multi-panel variant becomes a config entry rather than
two more attribute names plus a transform branch.
``ToggleDoorSwing`` gets a ``description`` classmethod that returns
user-facing wording per ``flip_geometry`` branch so the tooltip on
hover stops reading like operator internals.
``test/bim/module/model/test_door_gizmos.py`` (new) pins the
per-door-type contract: 11 cases covering LEFT / RIGHT hinge positions,
DOUBLE_SWING parity with SINGLE_SWING, DOUBLE_DOOR 4-arc layout, the
sliding-types hide invariant, ``is_editing=False`` hide invariant,
flip-arc matrix re-assignment, and world-matrix pre-multiplication.
Verified: ``pytest test/bim/module/model/test_door_gizmos.py`` 11/11
green; combined wall + stair + door gizmo lanes 37/37 green; ruff +
black clean on the three touched files.
Generated with the assistance of an AI coding tool.
Add a clickable "xN" badge to GizmoStairEdition's edit row, mirroring
the array's popup-input UX: click opens a number dialog (no more
shift+click-into-modal). Text-only — no 2x2 grid glyph.
Structural changes that enable this cleanly:
* IconSlot.placeholder=True: slots reserve an X position in the row
without auto-creating a gizmo. Subclasses resolve the reserved X via
_slot_x_positions()[name] to place their own dynamic gizmos. Drops
the brittle "remember to add extra_gap_before" workaround that would
silently rot on slot reorders.
* Array bug fix: the count badge collided with the "-" icon because
the slot manager placed count_minus at the cycle position (X=0.87)
where ICON_NUMBER_X also lives. Migrating the badge to a placeholder
slot lets the manager allocate the X naturally and the "-" no longer
overlaps. ICON_NUMBER_X constant removed.
* IntegerInputDialogMixin in parametric_lifecycle.py: extracts the
popup-dialog plumbing shared between InputArrayCount and the new
InputStairTreads. Subclasses declare an IntProperty + attr_name +
props_getter; the mixin owns invoke/execute. _resolve_props helper
factors the common obj/props/requires_editing prologue.
Tests: BIM_GT_count_label registration; IconSlot placeholder contract
(no gizmo_idname required; gizmo_attrs() returns empty); the stair
edit-row slot layout reserves the label position between tread_lock
and plus at one ICON_ARRAY_GAP each; visibility propagates from
props.is_editing.
Partly generated with the assistance of an AI coding tool.
Three related cleanups in one pass:
* **Per-gizmo preferences removed.** The ``visibility_pref`` field on
IconSlot, the ``prefs.gizmos.<feature>.<icon>`` PropertyGroups, and
the dispatcher that surfaced them in the addon preferences UI are
all gone. ``update_gizmo_visibility`` loses its ``pref_enabled``
parameter — visibility is now driven purely by editing state and
modal gating. bim/ui.py drops ~257 lines of dead PropertyGroup
definitions; bim/__init__.py and tool/parametric.py shed their
matching wiring; door / wall slot declarations stop referencing
the now-nonexistent prefs.
* **Dynamic-wall face normals fixed.** ``regenerate_wall_mesh_from_props``
in wall.py now calls ``bmesh.ops.recalc_face_normals`` before writing
the mesh. Without it, walls regenerated from the parametric edit
draft could ship with inward-facing normals on some faces, which
rendered as visual holes under any backface-cull or normal-aware
shading. ``test/bim/module/model/test_wall_preview_mesh.py`` pins
the invariant (every face's normal points away from the wall centre).
* **Color constants DRY.** ``COLOR_RED`` / ``COLOR_GREEN`` /
``COLOR_BLUE`` / ``COLOR_NEUTRAL`` now live at module scope in
gizmos.py; the BaseParametricGizmoGroup class attributes alias the
same tuples so ``self.COLOR_GREEN`` keeps working. IconSlot
declarations in stair.py (plus / minus) and array.py (count_minus /
count_plus / delete) now reference the named constants instead of
duplicating the RGB tuples inline.
Verified: headless smoke green at 1267 BIM_OT_ classes,
test_parametric_registry.py 8/8, wall lane 31/31 (includes the new
preview-mesh test). ruff + black clean on the touched files.
Generated with the assistance of an AI coding tool.
Hovering a wall-junction link-toggle icon today only swaps the icon
shape — the user doesn't see which wall the click will disconnect from
until after they click. ATPATH (T-junction) configurations especially
make the partner ambiguous when multiple connections sit close together.
On hover, paint a wireframe bbox around the partner wall using the same
shader, constants and color the array module already established for
its layer-children highlight (POLYLINE_UNIFORM_COLOR, decorator_color_special,
line width 1.8, alpha 0.8). The line-width / alpha constants in decorator.py
are renamed from _ARRAY_LAYER_BBOX_LINE_* to _BBOX_HIGHLIGHT_LINE_* and
shared between draw_array_layer_children_bbox and the new
draw_wall_partner_bbox so the two highlights stay in lockstep.
The trigger lives in a new GizmoWallLinkToggle subclass in wall.py
which keeps the base gizmos.GizmoLinkToggle generic (per the
generic-naming convention for shared widgets). The subclass's draw()
calls super().draw(context) then on self.is_highlight outlines its
partner_obj via the shared decorator helper. Same trigger pattern as
GizmoArrayLayerIndicator.
Blender's Gizmo API exposes target_set_operator but no symmetric
getter, so the partner reference can't be read back from the bound
operator handle. Instead GizmoWallUnjoinSingle.position_gizmos
mirrors the resolved partner_obj onto each visible icon every frame
next to the existing other_wall_guid write — the icon's draw() reads
from its own __slots__-declared attribute.
A forward-compat AST test pins the contract: GizmoWallLinkToggle.draw
must reference is_highlight and call draw_wall_partner_bbox. Catches
the regression where someone tidies the draw() override into super()
or replaces the shared helper with an ad-hoc draw call.
Generated with the assistance of an AI coding tool.
The parametric edit toolbar row used to assign each feature icon its
own ICON_<NAME>_X constant, with a separate FEATURE_ICON_MAX_X override
each subclass had to bump whenever a new icon was added. Forgetting the
bump silently collided icons — wall's rotate icon and the array button
both landed at X=1.24 in edit mode.
The new IconSlot dataclass + feature_slots tuple replace the
constants-and-override pattern with order-driven positioning: the
layout manager assigns each slot an X from its tuple index plus a
uniform ICON_ARRAY_GAP. Adding an icon is now a one-line append; the
"forget to bump" failure mode is structurally impossible.
Slot capabilities cover every existing icon-row shape:
* Single icon (wall rotate, array delete).
* N-variant slots — N gizmos at the same X with one visible per frame
via a subclass picker (stair tread-lock open/closed, wall baseline
exterior/center/interior). Pair becomes the N=2 case; triplet the
N=3 case. Variant idnames can be authored either as a tuple of
explicit names or as a string prefix that auto-suffixes _<variant>.
* Visibility prefs gate slot rendering without reflowing the row —
hidden slots still consume their X position.
* Extra per-slot gap before for visual separation (array's delete
trails the routine controls by an extra 0.2 m).
* Operator props forwarded to target_set_operator so adjusters
(+/-, increment) and generic toggles (property_name=...) work.
When the cycle slot is unused, feature slots collapse into the cycle
position so the row stays tight — that's how wall's baseline triplet
sits at X=0.87 without a gap before it.
Three subclasses migrate to the new system:
* wall.py — rotate icon + baseline triplet variants. Drops
ICON_ROTATE_X, _BASELINE_GIZMO_ATTRS, the manual triplet creation
loop, and the matching positioning block in _update_icon_row_extras
(it now just picks variant visibility).
* stair.py — tread_lock pair (open/closed) + plus + minus.
_update_editing_icon_positions reads slot X via _slot_x_positions
instead of three hardcoded constants. Also fixes the standalone
total_length_lock gizmo, which was broken since PR4 split
VIEW3D_GT_lock into open/closed pair (caller wasn't updated).
* array.py — count_minus + count_plus + method + delete (with
extra_gap_before=0.20 to separate the destructive action).
Drops the manual edit-row positioning loop entirely; the base
loop handles it. GizmoArrayChild now inherits BillboardingGizmoGroupMixin
and uses the shared setup_icon_gizmo helper, dropping its
duplicated _make_icon wrapper.
Two helpers added on BillboardingGizmoGroupMixin to fold the duplicated
prefs/color preamble that appeared at the top of six wall gizmo setups
plus the array-child setup:
* get_decoration_colors() — (decorations_colour, decorator_color_selected),
the active-state pair.
* get_unselected_decoration_colors() — (decorator_color_unselected,
decorator_color_selected) for gizmos surfaced on already-selected
geometry that should not pull focus.
Verified: headless smoke green at 1267 BIM_OT_ classes,
test_parametric_registry.py 8/8 pass, wall lane 29/29 pass,
model lane unchanged at 135 pass + 7 pre-existing v0.8.0 failures
(no regressions). ruff + black clean.
Generated with the assistance of an AI coding tool.
The previous single-wall unjoin gizmo used a bracket-pair icon
(VIEW3D_GT_unjoin) that reads as "unjoin" only after you know what
it is, with no clear "linked" inverse — closing the brackets to
suggest the connected state collapses to a hollow square that
doesn't read as a link at all.
Add GizmoLinkToggle (VIEW3D_GT_link_toggle): two filled dots joined
by a horizontal connector in the default state. On hover the two
halves shear vertically apart — left dot+stub slip down as a unit,
right dot+stub slip up — with a horizontal gap at the centre,
signalling that a click will sever the underlying connection. The
glyph lives next to the generic icon classes (GizmoLockOpen/Closed,
GizmoArc) so any path / link / pair-of-connected-items context can
reuse it; it isn't wall-specific despite the first caller.
The class keeps its own per-state GPUBatch cache so the shape swap
on hover doesn't allocate per frame. The hit-shape is sourced from
the broken form (the larger bbox of the two states) so the cursor
doesn't lose hover at the offset dots' outer edges and flicker
between states.
GizmoWallUnjoinSingle.setup() now requests VIEW3D_GT_link_toggle.
The operator binding (bim.unjoin_wall_path_connection), the
POOL_SIZE, and the per-frame partner-GUID write are unchanged.
Generated with the assistance of an AI coding tool.
When the two source walls were placed at different elevations, the
fillet corner wall ended up with sub-degree X and Y Euler rotations
even though both source walls had only a Z rotation.
Cause: _apply_fillet_corner_geometry derived the corner's local X
axis from `chord = tangent_b - tangent_a` (a 3D vector). With walls
at different Z, `chord.z` was non-zero, so `x_dir = chord.normalized()`
inherited that Z component. The Z axis was already hardcoded to world
Z, so x_dir and z_dir were no longer orthogonal — the resulting
matrix_world was non-orthonormal, and Blender's Euler decomposition
surfaced the skew as the visible X/Y rotation drift.
Project the chord to the XY plane before normalising so x_dir is
strictly XY-aligned and orthogonal to z_dir. The corner wall is now
placed at wall A's elevation with a pure Z rotation, which matches
the user's expectation when both inputs are Z-aligned regardless of
their relative elevation.
Generated with the assistance of an AI coding tool.
GizmoWallUnjoinSingle.poll accepts fillet-corner walls via the looser
tool.Parametric.is_path_connectable_wall predicate (fillet corners
have no LAYER2 usage by IFC spec, but they still participate in
IfcRelConnectsPathElements). The partner filter inside
_iter_path_connections used the stricter tool.Blender.Modifier.is_wall
(LAYER2-only), so adjacent LAYER2 walls silently dropped their
fillet-corner partners from the connection list — the unjoin icon
appeared when the fillet wall itself was selected but not on either
of its LAYER2 neighbours.
Switch the partner filter to is_path_connectable_wall so host and
partner predicates match. Add a regression test for the fillet case
and an AST forward-compat guard pinning the predicate symbol so a
future "tidy the imports" can't silently re-introduce the asymmetry.
Generated with the assistance of an AI coding tool.
Ports the array parametric-edit lifecycle, gizmo group, child guard,
per-layer ARRAY entry icons, and the array bbox decorators
(preview + selection highlight + layer-children) from gizmos-8088.
Restores the array_gizmo icon's positioning + visibility in the
framework's parametric edit row.
Registry (tool/parametric.py):
* EDIT_TYPES adds ParametricObject("array", supports_build_edit_lifecycle=True).
_ArrayEditMixin in array.py feeds build_edit_lifecycle which auto-
generates EnableEditingArray / FinishEditingArray / CancelEditingArray
with the conventional bl_idnames the gizmo references.
tool/blender.py:
* Adds is_array predicate wrapper around tool.Parametric.is_array.
The registry contract test test_every_entry_has_modifier_predicate
enforces every EDIT_TYPES entry has a matching is_<name> wrapper on
tool.Blender.Modifier.
array.py (+1130 LOC port from gizmos-8088):
* _ArrayEditMixin(ParametricEditMixinBase) drives the auto-generated
enable / finish / cancel lifecycle.
* GizmoArrayEdition: validate + cancel + count display + +/- adjusters
+ method toggle + delete button + per-layer ARRAY entry icons
(preallocated pool of MAX_LAYER_GIZMOS=8).
* GizmoArrayChild: child-array gizmo for the array-replica case.
* EditArrayFromChild: resolves the spawning layer via
tool.Array.get_child_layer_index so clicking a child's array gizmo
opens the layer that produced that child rather than always layer 0
(the gizmos-8088 source itself hardcoded item=0; HEAD has the helper
to do it right).
* New operators: EnableEditingArrayItem, ArrayParentGizmoClick,
ArrayGizmoClick, ToggleArrayMethod, RemoveArrayLayerFromEdit,
InputArrayCount, AdjustArrayCount.
prop.py: BIMArrayProperties gets per_child_opening BoolProperty
(when the array parent fills a host, give each child its own
opening + filling pair).
Bug fix: guard update_relating_array_from_object against the
cleanup-time None set. _finish_one writes relating_array_object = None
to clear the source-array reference; that fired the update callback,
which dispatched bpy.ops.bim.enable_editing_array(item=self.is_editing).
With is_editing just flipped to False, the bool coerced to 0 and
re-opened layer-0 edit immediately after every validate. The guard
short-circuits on None; item is also fixed to 0 (the bool-as-layer-
index was always meaningless for the legitimate user-pick path).
decorator.py (+312 LOC, all ports from gizmos-8088):
* bbox_world_edges / draw_polyline_segments / _BBOX_EDGES - shared
geometry helpers usable across array decorators.
* draw_array_layer_children_bbox - green wireframe bbox per child of
one array layer, drawn inline from a gizmo's draw() so the highlight
tracks the hover cursor without POST_VIEW lag.
* ArrayPreviewDecorator - faint cyan ghost bboxes at each future
array instance during the edit lifecycle (offset math mirrors
Model.regenerate_array, gated on props.is_editing).
* ArraySelectionHighlightDecorator - bounding-box overlay surfacing
the array family of the selected object. Child selected -> parent
in special color + siblings in unselected color; parent selected
(idle) -> all children in unselected color. TokenCache-backed.
handler.py: imports + uninstall/install the 2 always-on decorators in
_install_viewport_overlays. Both self-poll, so installation has no
cost when no array is selected / in edit mode.
Registration (bim/module/model/__init__.py):
* Adds the 3 lifecycle classes generated by build_edit_lifecycle
(CancelEditingArray, EnableEditingArray, FinishEditingArray) -
they exist as module-level names but are only visible to Blender's
operator registry when included in the classes tuple.
* Adds the 8 new operators + 2 new gizmo groups in alphabetical order.
gizmos.py: restores the array_gizmo icon position + visibility block
in BaseParametricGizmoGroup.update_editing_gizmos. Was force-hidden
in c250b2c1a because no array gizmo existed; the icon's plumbing
comes back online now that GizmoArrayEdition is registered.
Verified by test/bim/test_parametric_registry.py: all 8 tests pass -
enable/finish/cancel ops resolve, PropertyGroup attached, is_array
predicate present, predicate is total on non-matching elements.
Generated with the assistance of an AI coding tool.
PR3 shipped tool.Wall.path_connection_location_world; the local
_path_connection_location_world added in PR4 commit 70845e4dd
duplicated the same logic. The only caller in wall.py already uses
the tool method (line 3687 area), so the local helper has been
dead code since the migration in 7e5e7b8d6 routed _get_wall_geom_cached
to tool.Wall.read_geometry. Drop it.
Generated with the assistance of an AI coding tool.
Pre-existing architectural smell on v0.8.0: core/root.py.copy_class
called a module-level _has_material_styles helper that did
ifcopenshell.util.element.get_materials() directly, bypassing the
Prophecy mock seam that every other branch in copy_class flowed
through. Symptom: test/core/test_root.py::TestCopyClass::
test_AAAAAAAAAAAA passed mock strings into copy_class, the helper
called .is_a() on the string, AttributeError.
Move the check to tool.Root.has_material_styles (paired with
assign_body_styles — they're called in sequence as "is there a
material style? if not, assign body style"). core/root.py now
calls root.has_material_styles(new) like every other dependency,
fixing the test failure and dropping the ifcopenshell.util.element
import that was the only consumer of the ifcopenshell import at
module load in core/root.py.
* core/tool.py: add abstract has_material_styles to Root interface.
* tool/root.py: add concrete classmethod near assign_body_styles.
* core/root.py: replace _has_material_styles helper call site with
root.has_material_styles; drop the local helper and its import.
* test/core/test_root.py: add the new mock expectation
root.has_material_styles("element").will_return(False) before the
existing assign_body_styles expectation.
Generated with the assistance of an AI coding tool.
Three test files covering PR4's new surfaces — preview registry,
wall-gizmo poll behaviour, fillet operator registration. Every test
walks the live registry or class hierarchy instead of hard-coding
preview keys, operator names, or helper function names, so adding a
new preview / wall gizmo group / fillet operator exercises the same
invariants without test edits.
test_preview_base.py (6 tests):
* RegistryContract: every PREVIEW_CANCEL_OPS entry resolves to a
callable cancel operator on bpy.ops.bim.
* GetPreviewPropsTolerance: get_preview_props returns None for
contexts without a scene (regression guard for the SimpleNamespace
bug fixed in commit ee63137c6).
* ActivationCycle (registry-driven loop): any_preview_active toggles
with each registered preview's is_active flag;
discard_pending_previews clears every active flag across every
registered preview.
* SaveOnDiscardWired: locates the bim.save_project operator
dynamically and verifies its execute path references the discard
helper by its actual __name__.
test_wall_gizmo_poll_gate.py (4 tests):
* WallGizmoGroupsHideDuringPreview: walks the wall module for
bpy.types.GizmoGroup subclasses (skips preview-owner exceptions
whose bl_idname contains 'preview'), mocks any_preview_active to
True, and asserts every discovered gizmo's poll returns False.
* BaseParametricGizmoPollHidesDuringPreview: mirrors the test for
the cross-feature parametric framework base class.
test_fillet_operators.py (3 tests):
* FilletOperatorsRegistered: at-least-four-ops + every-discovered-op-
is-callable. Catches accidental deregistration.
* EnableRejectsIneligibleSelection: poll returns False without a
selection so the operator is greyed-out in menus.
State-clearing tests via bpy.ops.bim.cancel_wall_fillet_preview() are
deliberately omitted — the operator early-returns when context.screen
is unattached and prior tests in the model lane can leave the screen
in that state, making the dispatch path inherently flaky. Live testing
covers the behaviour.
Net: 13 tests pass cleanly in both single-file and full model lane.
Generated with the assistance of an AI coding tool.
Save-path:
* SaveProject._execute (project/operator.py) now calls
preview_base.discard_pending_previews(context.scene) right after
tool.Parametric.commit_pending_edits(). Previews are session-
transient — discard rather than commit. Sibling gizmo polls gate
on each preview's is_active flag; a stuck flag persisted through
the save would silently hide them on reload. Mirrors the pattern
already in gizmos-8088.
Preview-active gate hardening:
* preview_base.get_preview_props tolerates contexts without a
``scene`` attribute. Pre-existing tests use SimpleNamespace mocks
for the context; the previous getattr(context.scene, ...) raised
AttributeError before the inner default kicked in.
Test update:
* test_wall_header_refresh.test_geom_generation_invalidates_wall_geom_cache
patches tool.Wall.read_geometry instead of the now-deleted local
wall._read_wall_geometry (commit 7e5e7b8d6 migrated the call site).
Generated with the assistance of an AI coding tool.
Two cohesive cleanups in one commit.
A. Migrate wall.py to PR3-absorbed tool methods (fixes bug 4: pen icon
missing on fillet corner walls):
PR3 shipped tool.Wall.read_geometry + tool.Wall.validate_for_parametric_edit
but wall.py kept local duplicates predating that work. The local
_read_wall_geometry guards on tool.Blender.Modifier.is_wall (LAYER2-only)
while the tool method guards on tool.Parametric.is_path_connectable_wall
(LAYER2 OR fillet corner). Consequence: _get_wall_geom_cached → local
_read_wall_geometry returned None for every fillet corner →
GizmoWallFilletReedit.position_gizmos hit `if geom is None: hide` →
pen icon was unreachable for every fillet corner the user created.
Three _read_wall_geometry callers migrated to tool.Wall.read_geometry
(_read_wall_state_into_props, _get_wall_geom_cached,
GizmoWallJoinIntersection.position_gizmos). Two
_validate_wall_for_parametric_edit callers migrated to
tool.Wall.validate_for_parametric_edit (_maybe_resync_wall_props_from_ifc,
EnableEditingWall._execute). Local helpers deleted; docstring references
updated.
B. Drop over-restrictive gizmo gates (fixes bug 1: join icons missing
when walls intersect away from endpoints):
GizmoWallJoinIntersection.position_gizmos no longer hides itself when
the projected intersection lands further than MAX_DISTANCE_TO_ENDPOINT_
FACTOR (0.75 wall lengths) from any endpoint. The remaining
PARALLEL_DOT_THRESHOLD (cos 2°) gate via project_axis_intersection
returns None for near-parallel walls and is the only correctness bound;
distance from endpoints is a UI concern, not a geometric one.
GizmoWallFilletReedit.poll drops the has_a / has_b ConnectedFrom +
ConnectedTo guard — the IsFilletCorner pset is the authoritative signal.
EnableWallFilletPreviewFromCorner.execute already separately validates
both neighbour connections and reports a user-facing error if either
side is disconnected.
Generated with the assistance of an AI coding tool.
Three live-session regressions surfaced after the fillet feature
landed.
Sister gizmos competed with the active preview:
* preview_base.any_preview_active(context): new helper iterates the
PREVIEW_CANCEL_OPS registry and returns True if any preview is open.
Future previews registered there automatically gate sister gizmos.
* BaseParametricGizmoGroup.poll (gizmos.py): short-circuits on
any_preview_active so every parametric gizmo (door/window/stair/
roof/railing/wall edition) hides during ANY preview.
* The 4 wall gizmo groups with explicit polls (GizmoWallAddOpening,
GizmoWallExtendVertically, GizmoWallJoinIntersection,
GizmoWallUnjoinSingle) + GizmoWallFilletReedit gain the same gate.
DRY: extract _wall_gizmo_poll_gate(context):
* 5 wall gizmo polls each duplicated the 2 pre-flight checks
(viewport-gizmos enabled + no preview active). The helper centralises
them — each poll becomes a single short-circuit line followed by its
per-feature selection inspection.
ESC cancels the active preview:
* try_cancel_active_preview already existed in preview_base since PR3
but had no caller. Hooked into OverrideEscape.execute (geometry/
operator.py) as a new elif branch — same keymap that already cancels
pen gizmo edit mode + item mode + edit mode + aggregate mode. Order
in the branch chain matters: try preview cancel before falling back
to try_canceling_editing_modifier_parameters_or_path so the in-
flight preview wins over a stale modifier-edit cancel attempt.
Generated with the assistance of an AI coding tool.
End-to-end fillet flow on top of the helpers + recreate_wall hook
(landed in the previous commit). Users select two LAYER2 walls, click
the fillet entry icon, drag the live radius widget, and validate to
replace the corner with a curved LAYER2 corner wall (banana body).
Operators (5):
* EnableWallFilletPreview: 2-wall selection → validates LAYER2 +
straight axis + zero-slope + intersect-or-joined state → seeds the
preview props with a default radius computed from the shorter
available leg.
* FinishWallFilletPreview: dispatches CreateWallFillet with the tuned
radius; clears preview state on FINISHED, preserves it on failure so
the user can re-tune without re-selecting.
* CancelWallFilletPreview: clears preview state, no IFC mutation.
* EnableWallFilletPreviewFromCorner: pen-icon re-edit on an existing
fillet corner — pre-fills the preview from the corner's BBIM_Wall
pset + walks the inverse graph to recover wall A and wall B.
* CreateWallFillet: deletes any prior corner + A↔B path connection,
shortens A and B to the tangent points, instantiates a corner wall
from A's type, unassigns the swept-layer material/type (the explicit
banana body MUST own its geometry), assigns the dominant material,
rebuilds the body, sets a straight 2-point chord axis, stores
BBIM_Wall.IsFilletCorner+FilletRadius, reconnects A and B to the
corner with NOTDEFINED on the corner's side.
Gizmo groups (2 new + entry icon on existing):
* GizmoWallFilletPreview: visible while a preview is active. Bundles
a radius_dim widget at the arc apex, a trim_dim widget along wall A
expressing the same DOF via the leg setback distance
(trim = |radius| * tan(sweep/2)), and validate / cancel icons
anchored above the apex in screen-up.
* GizmoWallFilletReedit: pen-icon entry on an existing fillet corner
wall (single-selection, BBIM_Wall.IsFilletCorner set, both neighbour
connections present). Mutually exclusive with an active preview.
* GizmoWallJoinIntersection now stacks a fillet entry icon
(VIEW3D_GT_fillet → bim.enable_wall_fillet_preview) above the
existing join/unjoin icon in the joined and intersect state branches.
Property + decorator infrastructure:
* prop.py: BIMWallFilletPreviewProperties (Scene-level draft) +
BIMPreviewProperties umbrella with only the wall_fillet pointer.
The umbrella is the seam preview_base.py (landed in PR3) already
reads via getattr(scene, "BIMPreviewProperties", None).
* decorator.py: _stroke_lines_alpha helper + WallFilletPreviewDecorator.
Polls is_active; renders leg projections + arc + arc-center
construction lines from tool.Wall.compute_wall_fillet_geometry.
* __init__.py: registers operators + gizmo groups + property groups +
wires Scene.BIMPreviewProperties.
* handler.py: WallFilletPreviewDecorator.install/uninstall in
_install_decorators — always installed, self-polls on is_active.
Drive-by: extract gizmo.get_screen_up(billboard_rot) helper —
the local +Y of a billboard rotation is the camera's screen-up world
direction. Replaces 4 inline `billboard_rot @ Vector((0.0, 1.0, 0.0))`
sites added across the fillet feature's gizmo groups.
Generated with the assistance of an AI coding tool.
Eleven module-level helpers in wall.py that the upcoming wall-fillet
operators + gizmo groups depend on. Each is self-contained or
references only helpers earlier in the file; the operators and
gizmos themselves land in follow-up commits.
* _wall_fillet_props / _wall_fillet_preview_active /
_wall_fillet_preview_walls: thin read-side accessors over the
BIMPreviewProperties.wall_fillet pointer (added with the
operators commit). Safe today: get_preview_props returns None
until the pointer is attached.
* _walls_have_zero_slope_for_fillet: validates that input walls
are vertical (x_angle ~ 0); slanted-extrusion fillets require
swept-along-curve geometry the banana profile builder doesn't
support.
* _build_curved_corner_body_representation: builds the banana
(annular sector) IfcExtrudedAreaSolid as a polyline-tessellated
IfcIndexedPolyCurve.
* _apply_fillet_corner_geometry: positions the corner wall at
tangent_a and rebuilds its body. Shared by the creation operator
and the regenerate path.
* _resolve_two_walls: pulls (active, other) from a 2-wall
selection, validates both as LAYER2 + straight-axis + not-already-
a-fillet-corner.
* _pick_dominant_wall_material: returns the thickest layer's
material from an element's IfcMaterialLayerSet / Usage.
* regenerate_fillet_corner_wall: re-runs the geometry build from
BBIM_Wall.FilletRadius + current neighbour layer parameters.
Called by tool.Model.recreate_wall when the IsFilletCorner pset
is set; the FIXME(PR4) placeholder in recreate_wall is dropped.
* _wall_fillet_gizmo_x_matrix: 4x4 placement matrix with local +X
aligned to a world-space direction; used by the fillet preview
gizmo group.
Centralises the IsFilletCorner pset read as
tool.Parametric.is_fillet_corner_wall — replaces 3 inline
get_pset(element, "BBIM_Wall", "IsFilletCorner") sites
(tool.Model.recreate_wall, tool.Model.recalculate_walls,
tool.Parametric.is_path_connectable_wall) plus the new
_resolve_two_walls call.
Generated with the assistance of an AI coding tool.
The framework's parametric-edit icon row currently binds an array
icon to bim.add_array_from_feature_edit, but the supporting per-
feature add-array flow and gizmo positioning haven't fully landed.
Showing the icon today lets the user click it and trigger a half-
wired flow.
Force the icon hidden inside the props.is_editing branch of
BaseParametricGizmoGroup.update_editing_gizmos. The else-branch
(not editing) already hides it, so this just mirrors that behavior
during edit mode. Drop this gate when array integration completes
to re-enable the icon position + visibility plumbing.
Generated with the assistance of an AI coding tool.
The extend-X / extend-Z icons in GizmoWallEdition's cursor row are
billboarded toward the camera; without orientation polish they
always point in the same screen-space direction regardless of which
wall endpoint the click will move (or whether the cursor sits above
or below the wall top). New helper mirrors the icon's local-X (extend-X)
or local-Y (extend-Z) axis so each arrow points toward the end it
will move:
* Extend-X: walk wall midpoint to figure out which endpoint stays
fixed (cursor past midpoint → ATSTART stays; cursor before midpoint
→ ATEND stays). Project the fixed endpoint into screen-space and
flip the arrow when the gizmo's anchor sits on the same side.
* Extend-Z: flip when the cursor is below the wall top (within
EXTEND_FLIP_EPSILON tolerance).
Called once per resolved cursor gizmo from
``GizmoWallEdition._update_cursor_gizmos``, after the gizmo's
``matrix_basis`` is set by ``gizmo.billboarded_at``. Reuses
``gizmo.should_flip_extend_arrow`` + ``EXTEND_FLIP_MIRROR_X/Y`` +
``EXTEND_FLIP_EPSILON`` already on tool.
Generated with the assistance of an AI coding tool.
GizmoWallJoinIntersection's unjoin only fires when exactly two walls
are selected and surfaces one icon at their shared corner — useless
when the wall has 3+ joins and the user wants to disconnect just one.
* UnjoinWallPathConnection: surgical counterpart to UnjoinWalls.
Disconnects the active wall from a single partner wall identified
by IFC GlobalId (invariant under Blender-object renames + file
save/reload + undo). Walks both inverse arrays of the active wall
for the specific IfcRelConnectsPathElements joining the pair —
matches DumbWallJoiner.split's pattern and avoids disconnect_path's
direction-sensitivity. Resyncs both walls' draft props after the
recreate_wall pass.
* GizmoWallUnjoinSingle: activates on exactly-one selected
LAYER2 wall. Preallocates a pool of 16 unjoin icons (Blender forbids
gizmo allocation outside setup(); ATSTART + ATEND + ATPATH rels are
rarely more than a handful). Per-frame, iterates _iter_path_connections,
positions one billboarded icon at each join via
tool.Wall.path_connection_location_world, and hides the rest. Each
visible icon's bound operator carries the partner GlobalId, so a
click removes only that one rel.
* model/__init__.py: register both classes alphabetically.
Mutually exclusive with GizmoWallJoinIntersection via poll() — that
group requires len(selected) == 2; this one requires 1.
Generated with the assistance of an AI coding tool.
The single-wall unjoin gizmo needs to enumerate every
IfcRelConnectsPathElements a wall participates in, regardless of which
side of the rel the wall was authored on, and place an icon at each
join's physical location. Two helpers carry that work:
_path_connection_location_world wraps core.compute_path_connection_location
at the Vector boundary. _iter_path_connections walks ConnectedTo +
ConnectedFrom, normalises orientation to (other, self_ct, other_ct),
and filters non-wall partners + None refs so per-frame gizmo positioning
survives malformed IFC.
Generated with the assistance of an AI coding tool.
After a one-shot wall IFC mutation (unjoin / split / merge / extend /
join-at-corner …) the always-visible gizmos on the OTHER side of the
join can be left reading stale ``BIMWallProperties`` — the IFC
geometry moved but the draft props that drive the gizmo handles still
point at the pre-mutation numbers, so a subsequent edit-mode enter
shows the wall at its old length / position.
* New ``_maybe_resync_wall_props_from_ifc(obj)``: re-primes a single
wall's draft props from current IFC, with guards for non-walls,
non-parametric walls, and walls in an active draft session (the
draft is then the source of truth, not IFC). Must run from an
operator ``_execute`` — ID writes from gizmo refresh raise.
* New ``_resync_walls_after_mutation(objs)``: iterates the above
across a selection.
* Six existing mutation operators gain a resync call after their
``core.*`` / ``DumbWallJoiner`` mutation completes:
UnjoinWalls, ExtendWallsToUnderside, ExtendWallsToWall, SplitWall,
MergeWall, JoinWallsIntersection. MergeWall resyncs only the
surviving wall — the active wall is the deletion target.
Generated with the assistance of an AI coding tool.
Bundle of bugs surfaced when exercising the new gizmo framework
end-to-end in a live Blender session after the
bim/module/drawing/gizmos.py refactor + TypeAccessor/CycleType/PickType
mixins landed.
Register / annotation resolution
* parametric_lifecycle.py: hoist `entity_instance` import out of
TYPE_CHECKING so typing.get_type_hints resolves the
Callable[[entity_instance], bool] annotation at operator registration
(CycleDoorType, CycleWindowType, CycleStairType failed with NameError).
Clarify the INTERFACE return contract on the picker entry-point so
readers see why the gizmo step stays off the undo stack.
Framework callable contracts
* model/wall.py, door.py, window.py, stair.py: migrate `props_getter`
and `element_checker` from bl_idname strings to bound classmethods
on tool.Model / tool.Parametric. BaseParametricGizmoGroup.get_props
expects a callable; the string form raised TypeError on first
gizmo poll.
* model/door.py, model/stair.py: drop the dead `prop_path=` operator
kwarg from create_arc_gizmo / create_icon_gizmo call sites. The
framework helper blindly setattrs every kwarg onto the operator's
OperatorProperties, but ToggleDoorSwing / ToggleStairProperty don't
declare prop_path — the setattr raised mid-setup_element_specific_gizmos,
so self.gizmo_door_type / self.lock_gizmo never got assigned and
every subsequent draw_prepare tornadoed AttributeError. Nothing
reads op.prop_path anywhere; the kwarg was dead data.
Dispatcher operators
* model/array.py: add EnableEditingParametric (the framework pen-icon
dispatcher that routes to a per-feature edit operator by bl_idname
string) and AddArrayFromFeatureEdit (binds the framework's array
icon to bim.add_array on the current parametric draft).
* model/__init__.py: register both new operators.
Per-frame robustness
* drawing/gizmos.py: guard BaseParametricGizmoGroup.draw_prepare with
is_setup_complete() — matches the existing guard in refresh() and
in BaseSchematicGizmoGroup.draw_prepare(). Defense-in-depth: when
any subclass's setup raises mid-way, draw_prepare now no-ops cleanly
instead of per-frame AttributeError-tornadoing on whatever attribute
the failed setup phase was meant to populate.
* model/decorator.py: guard ProfileDecorator.__call__ against
context.active_object is None. The decorator is a per-frame
viewport draw handler; deselecting or deleting the active object
while it's installed crashed on obj.mode access. Treat None the
same as "no longer in edit mode" — uninstall + fire the exit
callback if present.
* geometry/data.py: ViewportData.load() populates `data` before
flipping `is_loaded`, so a raise from cls.mode() no longer leaves
the class flag-set but data-empty for subsequent reads.
Generated with the assistance of an AI coding tool.
Three concerns bundled into one cohesive refactor of gizmos.py
(splitting them surgically requires intermediate commits with
duplicate same-named classes that Python can't parse):
1. Framework primitives — StaticTrisGizmoMixin + TexturedQuadGizmoMixin
replace the older TrisGizmoMixin. New module-level helpers:
_get_static_tris_shader / _get_static_tris_batch / clear_static_
tris_cache for cached GPU batch reuse, _draw_outline_and_body for
the shared outline-then-body render path, draw_tris_with_outline
as the public wrapper. billboarded_at(world_pos, billboard_rot,
scale) is the canonical billboard-matrix helper; should_flip_extend_
arrow encapsulates the view-aware mirror decision for extend
gizmos; get_warning_color_from_prefs reads the user's warning
color.
2. Config classes — BaseValueGizmoConfig (shared visibility + dimension-
text contract), CountGizmoConfig (array N indicator),
DimensionGizmoConfig (length / height / depth labels), IconActionConfig
(icon-only gizmos that invoke an operator on click). DimensionRenderer
draws the actual numeric label using BLF.
3. Icon classes — each rewritten on StaticTrisGizmoMixin so they share
the cached GPU batch + outline-then-body render path:
GizmoLockOpen / GizmoLockClosed (replacing the single-state
GizmoLock), GizmoArc, GizmoFillet, GizmoWallCornerIcon,
GizmoWallTeeIcon, GizmoPen / GizmoValidate / GizmoCancel (the
parametric-edit triad), GizmoPlus / GizmoMinus / GizmoTrash,
GizmoArrayParent / GizmoArrayAll / GizmoArrayLayerIndicator (array
context indicators with a small digit-rendering helper for the "xN"
count label), GizmoMerge / GizmoSplit / GizmoUnjoin (wall-join
icons), and GizmoMenu (textured-quad icon-action menu trigger).
The legacy TrisGizmoMixin, GizmoLock, and DimensionDrawConfig are
removed; downstream callers in subsequent PR4 commits swap to the
new mixin and config classes when their feature operators land.
CycleTypeMixin / PickTypeMixin / TypeAccessorBase live in
bim.parametric_lifecycle (previous commit). The three mixins are
re-exported from gizmos.py here so feature-module access via
``gizmo.<MixinName>`` keeps working until PR5 cleanup drops the
re-exports.
bim/module/drawing/__init__.py is updated in the same commit to
register the 11 new gizmo classes (GizmoLockOpen / GizmoLockClosed /
GizmoFillet / GizmoWallCornerIcon / GizmoWallTeeIcon / GizmoTrash /
GizmoArrayParent / GizmoArrayAll / GizmoArrayLayerIndicator /
GizmoUnjoin / GizmoMenu) — without that, the new classes exist in
gizmos.py but aren't usable as bpy gizmo types.
Generated with the assistance of an AI coding tool.
Three operator mixins for type-selection ops on parametric features
(door type-cycle, window type-pick, stair type-cycle, railing
type-pick, roof type-cycle, etc.). Each shares the same contract:
* ``element_checker`` validates the active object is the expected
IFC type
* ``props_getter`` resolves the BIM<Name>Properties group
* ``type_literal`` is the Literal type whose args drive the enum
* ``type_attr`` is the PropertyGroup field to read/write
* ``skip_element_check=True`` bypasses element validation (for
operators that target a non-IFC context)
CycleTypeMixin shift-click reverses direction (forward by default).
PickTypeMixin opens a popup menu and routes the picked value
through execute() so F6 redo / EXEC_DEFAULT reach the apply path.
The PickType modal-handler dance waits for LEFTMOUSE release before
opening the menu when invoked mid-click (e.g. from a gizmo's
target_set_operator) so Blender's drag-through-pick gesture doesn't
commit an accidental item.
Ships standalone — the next commit's gizmos.py framework refactor
re-exports these names from bonsai.bim.parametric_lifecycle so
gizmo modules can spell ``gizmo.CycleTypeMixin`` / ``gizmo.PickTypeMixin``.
Concrete operator subclasses land in subsequent PR4 commits per
feature (door / window / stair / railing / roof).
Generated with the assistance of an AI coding tool.
deepdiff 9.1.0 added cachebox<6,>=5.2 as a direct runtime dep.
cachebox 5.2.3 only publishes macOS x86_64 wheels for macosx_10_12+,
incompatible with the macos py311 build's --platform macosx_10_10_x86_64.
The daily build's linux-wheel safeguard fires when the resulting
cachebox-*-manylinux_*.whl leaks into the macOS / windows wheels folder
(builds run on ubuntu-latest and cross-build via pip download --platform).
Pin deepdiff to <9.1 (resolves to 9.0.0, no cachebox transitive dep) as
the minimal hotfix. Long-term cleanup: bump the macos py311 platform tag
from 10_10 to 10_13 (matching py312/py313) and re-flag this line with the
standard \$(PYPI_PLATFORM) --only-binary=:all: pattern used by brickschema
and python-socketio.
Partly generated with the assistance of an AI coding tool.
When override_mode_set_edit encounters an unsupported profile (Couldn't
import profile), deselect the object so Tab continues to cycle cleanly.
Also restores the selection that existed before entering aggregate mode
when finally tabbing out, via save/restore_previous_selection().
Following the pattern from 586f9be077, deselect the active object after
exiting item mode so Tab continues to cycle cleanly. Also deselects
parametric LAYER1/LAYER2 items that cannot be edited directly, avoiding
the need to manually deselect before Tab-cycling out of aggregate mode.
In BIM_PT_type_attributes and BIM_PT_object_attributes (when
the active object is a type), attribute value buttons now use
"type.<Attr>" as the selector key so the operator finds
matching occurrences via their relating type rather than the
occurrence's own (often unset) attributes.
Generated with the assistance of an AI coding tool.
After selecting objects in the same container, copy a `location="Name"`
filter query to the clipboard and report it — consistent with the same
behaviour in SelectSimilarType, SelectSimilarAggregate, SelectIfcClass,
and SelectSimilarMaterial.
Generated with the assistance of an AI coding tool.
When a `+`-separated filter group returns no results, `FacetTransformer.facet_list`
was skipping the reset of `has_additive_facet_in_current_list` because the reset
was inside the `if self.elements:` guard. The stale flag caused the next group's
`add_default_elements()` to bail out early, leaving its element set empty and
silently dropping every subsequent group from the result.
Move the flag reset outside the guard so it always fires regardless of whether
the group produced any results.
In the Placement panel, show Location and Rotation X/Y/Z
each on their own row beneath a header label. When the IFC
file uses imperial units, display a read-only feet-and-inches
label alongside each Location input field.
Generated with the assistance of an AI coding tool.
Display general and calculated stair parameters (Width,
Height, Tread Run, Tread Rise, Length, etc.) formatted
to the IFC file's configured length unit rather than
raw numeric values.
Generated with the assistance of an AI coding tool.
DecorationsHandler now caches dissolved edges (mesh-keyed), world-space
draw payload, and GPUBatch objects with per-object epoch invalidation —
moving one wall doesn't wipe 50 opening caches. Object-mode dissolve
removes triangulation noise; 2-pass depth-test split dims occluded lines
instead of hiding them. Edit-mode behavior unchanged.
Also: disable viewport shadows for IfcFeatureElementSubtraction objects,
and wire DecorationsHandler.uninstall() into the model module's
unregister() so the new persistent handlers don't leak on addon disable.
Generated with the assistance of an AI coding tool.
EditAssignedMaterial propagated layer set usage attributes
to all selected objects but skipped this loop for profile
set usage. Add the same loop so CardinalPoint and
ReferenceExtent are copied to each selected object's
IfcMaterialProfileSetUsage on save.
Generated with the assistance of an AI coding tool.
When the user enters `-0' - 10"`, Python parses feet as -0.0.
The check `feet < 0` is False for negative zero, so the sign was
silently dropped. Use math.copysign to detect it correctly.
Generated with the assistance of an AI coding tool.
Two paired test files for the framework infrastructure landed
earlier in this PR.
test_decorator_cache.py (11 tests):
* The 4-hook invalidation list (depsgraph_update_post + undo_post +
redo_post + load_post) is symmetrically managed by
install_decorator_cache_handlers / uninstall_decorator_cache_handlers.
A future edit that drops a hook from one side without the other
would land as a Blender segfault when a cached bpy.types.Object
ref outlives its underlying ID block — the regression must surface
as a test failure first.
* install is idempotent (calling twice doesn't double-register).
* uninstall when not installed doesn't raise.
* The bump handler accepts Blender's variadic args.
* The depsgraph predicate gates correctly: bumps on Object geometry
or transform updates, silently skips on Material / NodeTree / Image
updates (which would otherwise rebuild every cache on every node
edit).
* TokenCache.get_or_compute short-circuits on key+token match and
recomputes when the token bumps.
test_undo_resync_parametric_drafts.py (3 tests):
* UNDO_REGENERATORS keys must all be in tool.Parametric.EDIT_TYPES.
A typo would silently no-op on Ctrl+Z, restoring the desync the
helper is meant to prevent.
* The dispatcher skips objects with no active parametric edit
(undo_post fires for every undo, most of which touch zero drafts).
* The dispatcher silently skips parametric types that have no
UNDO_REGENERATORS entry (door / window / array are IFC-derived
with no draft preview mesh — they don't need a regenerator).
Mocks use spec=bpy.types.Depsgraph / spec=bpy.types.DepsgraphUpdate
/ spec=tool.parametric.ParametricObject so typos in mocked-attribute
access fail loudly (CLAUDE.md test discipline).
Generated with the assistance of an AI coding tool.
Splitting a wall through a door orphaned the door (door.FillsVoids
became empty). The fill rel was being reassigned by setting its
RelatedBuildingElement slot — schema-wise that's the filling slot, not
the wall slot — so when remove_feature deleted the old opening it
also cascade-removed the rel. Transferring via RelatingOpeningElement
keeps the rel pointing at the new opening so the door stays
associated. Pre-existing bug from 5a6476a57, surfaced by ef144dce2.
Generated with the assistance of an AI coding tool.
DumbWallJoiner.split assigned openings by projecting the opening's
centre-point onto the wall axis, so any opening whose footprint
straddled the cut was silently dropped from whichever wall its centre
missed. Now the full axis-projected extent (via ifcopenshell.geom.
create_shape) drives the assignment; for filled openings whose void
straddles the cut, a pure-void copy is added back to the neighbour
wall so its body is also cut.
Generated with the assistance of an AI coding tool.
Five inline copies of the same defensive pattern lived across
``tool/parametric.py``, ``bim/parametric_lifecycle.py``,
``bim/module/model/preview_base.py`` (twice), and as a near-twin
in ``tool/blender.py:update_all_viewports`` itself.
``tool.Blender.update_all_viewports`` already covered the
``tag_redraw`` job but used an ``assert context.screen`` that would
raise during background-mode operators or early-load_post calls
where ``screen`` legitimately is None. Relax to a defensive
``getattr(context, "screen", None)`` + silent return so the helper
fits every caller's needs, then collapse the 4 inline copies to
single calls.
Net -9 LOC. The helper now describes its contract ("silent no-op
when no screen attached") rather than naming specific callers, so
moving a caller doesn't rot the docstring.
Generated with the assistance of an AI coding tool.
Ctrl+Z / Ctrl+Shift+Z on an in-progress parametric draft (wall /
stair / roof) used to leave the preview mesh frozen in its
pre-undo shape — the IFC mutation rolls back but the bmesh built
from draft props doesn't repaint.
Add a registry of per-type regenerator functions
(``UNDO_REGENERATORS``) that re-build each type's preview mesh
from its current props. The dispatcher
``resync_parametric_drafts_after_undo`` walks all objects, skips
any without an active parametric edit, looks up the regenerator
by feature name, and calls it. Tagged 3D viewports for redraw.
Types without an entry (door / window / railing / etc.) are
intentionally absent — they're IFC-derived, so the undo's
representation rollback + next-frame refresh already repaints
correctly without a draft-side regenerator.
Undo/redo wiring is self-installed by
``bonsai.bim.parametric_lifecycle``: a ``@persistent``
``_resync_on_undo`` callback dispatches into the registry, and
``install_parametric_lifecycle_handlers()`` /
``uninstall_parametric_lifecycle_handlers()`` append/remove it
from ``bpy.app.handlers.undo_post`` and ``redo_post``.
``bim/__init__.py``'s ``register()`` calls the install function
*after* the central ``handler.undo_post`` / ``redo_post`` appends
so the regenerators see restored IFC state — ``bpy.app.handlers``
fire in append order. ``handler.py`` itself stays ignorant of the
parametric subsystem. The lazy function-local imports in each
regenerator break the addon-load cycle —
``bonsai.bim.parametric_lifecycle`` loads before
``bim/module/model/*``.
Generated with the assistance of an AI coding tool.
Three changes to the shared Enable/Finish/Cancel mixins:
1. Always-on drift triad on ParametricEditMixinBase. The base now
provides ``_handle_drift_on_enable`` / ``_handle_drift_on_finish``
/ ``_handle_drift_on_cancel`` classmethods, called from the
per-mixin ``_enable_one`` / ``_finish_one`` / ``_cancel_one``.
Pre-edit Blender-side translations commit to IFC on Enable
(apply_scale=False — only translation/rotation, not the user's
accidental scale), in-edit drag commits on Finish (apply_scale=True),
and Cancel restores the committed IFC placement via
``restore_or_rebaseline_placement``. Prevents the
"uncommitted drag disappears on Finish" and "preview snaps back
on Cancel" UX bugs.
2. ``_ParametricEditMixinBase`` renamed to ``ParametricEditMixinBase``
(public). Per-feature mixins that need to subclass directly
(e.g., when neither FeatureModifier nor PathPreserving fits)
can do so without reaching into a private name.
3. ``_update_modifier_bmesh`` (PathPreserving) renamed to
``_restore_viewport_after_cancel``. The old name was inaccurate
for subclasses that load a different IFC representation on
Cancel rather than rebuilding a bmesh preview from props.
Plus two polish changes:
* ``_mark_type_thumbnail_dirty`` helper on the base centralises the
``ifcopenshell.util.element.get_type`` + thumbnail-mark pattern
that both mixins repeated inline.
* ``FeatureModifierEditMixin._cancel_one`` and
``PathPreservingEditMixin._cancel_one`` wrap the restore in
``try/finally`` so ``props.is_editing = False`` flips even on
partial restore failure. Without this, a Cancel that raised
mid-restore would leave the user locked out of the edit lifecycle.
* ``PathPreservingEditMixin._finish_one`` / ``_cancel_one`` skip the
pset commit + viewport rebuild when the draft equals the stored
pset (no-op Enable→Finish round-trip should not pollute the
representation list or burn an undo entry).
``FeatureModifierEditMixin._finish_one`` now routes the pset commit
through ``tool.Pset.write_bbim_data`` instead of inlining the
``createIfcText(json.dumps(...))`` + ``ifcopenshell.api.pset.edit_pset``
dance. Two test assertions updated to match.
Generated with the assistance of an AI coding tool.
Three concerns folded into ``load_post`` argue for separation:
1. Save-file invariants every load must re-establish (msgbus
subscription, owner-settings, thumbnail cache, draft-flag healing,
blend-warning flag, H5 lock probe).
2. User-preference-driven UI setup (toolbar, workspace, viewport
shading, panel hijack, snap defaults).
3. Viewport overlay sync (every decorator's install/uninstall).
Pull each into its own function (``_apply_save_file_invariants`` /
``_apply_user_preferences`` / ``_install_viewport_overlays``). The
``load_post`` callback becomes a 3-line orchestrator. Each phase
is independently call-able from tests and from PR4 features that
need to re-trigger one phase without the others.
Two new hooks land with the decompose:
* ``tool.Parametric.heal_stale_edit_flags()`` + ``discard_pending_previews(scene)``
fire in ``_apply_save_file_invariants``. The first clears
object-level ``BIM<Name>Properties.is_editing`` flags that lost
their backing IFC element across a load; the second clears
scene-level ``BIMPreviewProperties.<x>.is_active`` so saved
preview state never resurfaces with no UI to interact with it.
* ``install_decorator_cache_handlers`` / ``uninstall_decorator_cache_handlers``
wrap the decorator install/install pass in
``_install_viewport_overlays``. The bump handlers append to
``depsgraph_update_post`` + ``undo_post`` + ``redo_post`` +
``load_post`` so the previous commit's ``TokenCache`` in
``tool.System.get_decoration_data`` finally invalidates on
structural scene changes.
Generated with the assistance of an AI coding tool.
Shared helpers for Bonsai's Scene-level parametric preview flows.
Two PR4 features will consume this — MEP bend preview and wall
fillet preview — both following the same shape:
Enable<X>Preview — populates draft on Scene.BIMPreviewProperties.<x>
Gizmo<X>Preview — polls on is_active, surfaces tunable widgets
<X>PreviewDecorator — GPU lines while is_active is True
Finish<X>Preview — bpy.ops.bim.<verb>(...) with draft kwargs
Cancel<X>Preview — pure state reset
The module hosts the cross-cutting accessors (``get_preview_props``,
``is_preview_active``), lazy-closure factories for gizmo dimension
callbacks (``make_props_callback`` / ``make_dim_getter`` /
``make_dim_setter`` — defensive against missing scene / freed RNA
struct on file open / undo), the Enable-time IFC-placement sync
(``sync_uncommitted_moves``), and the Esc + load_post discard
machinery (``PREVIEW_CANCEL_OPS`` registry, ``try_cancel_active_preview``,
``discard_pending_previews``).
Ships standalone — the consumer features land in PR4 (preview
PropertyGroups, Enable/Finish/Cancel operators, gizmo groups,
decorators, Esc keymap binding). All accessors are defensive
against missing PropertyGroups / operators on v0.8.0 — calling
``discard_pending_previews(scene)`` from the next commit's
load_post hook is a no-op until PR4 attaches BIMPreviewProperties.
Generated with the assistance of an AI coding tool.
System decoration draws on every viewport refresh — the
``_build_decoration_data`` body walks every distribution element,
resolves connected ports, builds the vert/edge arrays for the GPU
batch. A bare call per frame burns time on an unchanged scene.
Add a single-entry cache keyed on ``(decorator_cache_token,
id(decorated_elements_set))``. Reads short-circuit when neither
component moved:
* ``decorator_cache_token`` from ``bim.decorator_cache`` invalidates
on depsgraph / undo / redo / load via the bump handler.
* ``id(decorated_elements_set)`` invalidates when
``SystemDecorationData.load()`` reassigns the set (e.g. when the
user changes the set of decorated systems via the panel).
The handler that bumps the token is installed in the next commit
(bim/handler.py decompose). Until then the token stays at 0, so
the cache only hits when ``id()`` also matches — degraded behaviour
during the bisect window but not incorrect.
Generated with the assistance of an AI coding tool.
New helper module for POST_VIEW decorators. Exports:
* ``get_decorator_cache_token()`` — global int counter consumers
include in their cache key so the value invalidates on structural
scene changes.
* ``_bump_decorator_cache_token()`` — ``@bpy.app.handlers.persistent``
callback that increments the token. Gates on the depsgraph payload
so animation playback / driver evaluation doesn't churn the token.
* ``install_decorator_cache_handlers`` / ``uninstall_…`` — idempotent
append / remove against depsgraph_update_post + undo_post + redo_post
+ load_post. Called once from ``bim.register`` / ``unregister``.
* ``TokenCache[T]`` — single-entry memoiser keyed on ``(caller_key,
token)``. Cached ``bpy.types.Object`` references can't outlive the
underlying ID blocks because any depsgraph / undo / load bumps the
token and forces a recompute.
This commit ships the module standalone. The next commits in this
PR wire it: tool/system.py adds the cache wrap on get_decoration_data
and bim/handler.py installs the bump callbacks. Until both land,
the module is intentionally dead code — keeps the diff narrow and
the commit history bisectable.
Generated with the assistance of an AI coding tool.
Five code paths in slim PR2 referenced symbols that don't exist in
v0.8.0's bim layer, raising at first call. Plus three type
annotations that ty flagged as unresolved.
1. tool/system.py:get_decoration_data — drop the cache layer that
keyed on a token from a bim/decorator_cache.py module. The cache
is dead-or-broken in slim: the depsgraph bump handler that would
invalidate the token lives in PR3's bim/handler.py decompose, so
the token stays at 0 forever. Either the cache never hits
(decorated_elements rebuilt → new id() per call) or returns
stale data (list reused). Revert to direct
`_build_decoration_data()` calls. PR3 reintroduces the cache
atomically: decorator_cache module + handler install + cache
wrap + tests. Keeps `_build_decoration_data` extraction
(cleaner than v0.8.0's monolithic version regardless of cache).
2. tool/spatial.py — add `get_host_element` + `get_host_wall`.
The interface stubs in `core/tool.py:1037-1038` were declared
but never implemented. `tool/duplicate.py:99` (object duplication
with fills) and `tool/model.py:1260` (array per-child opening
mirror) call these and would raise AttributeError.
3. tool/model.py:recreate_wall — drop the fillet-corner branch
that function-locally imports `regenerate_fillet_corner_wall`
from `bim/module/model/wall`. The function lands with PR4; fall
through to the straight-extrusion path preserves v0.8.0
behaviour for fillet walls until then. Tag FIXME(PR4).
4. tool/model.py — drop `get_pipe_segment_props` /
`get_duct_segment_props` accessors. Their return types reference
`BIMPipeSegmentProperties` / `BIMDuctSegmentProperties` which
land with PR4's prop.py; calling either accessor on v0.8.0 would
AttributeError on `obj.BIM<X>SegmentProperties`. Zero callers in
slim — PR4 reintroduces both accessors together with the
PropertyGroups they wrap. Also drops the matching TYPE_CHECKING
imports.
5. tool/blender.py:557 — `Mapping[type[ViewportDecorator], bool]`
needs the qualified `Blender.ViewportDecorator` because the
annotation is on a method INSIDE the same nested class; the
bare name doesn't resolve at type-check time.
6. core/tool.py Surveyor — drop the `obj: "bpy.types.Object"` /
`z: float` / `-> float` / `-> None` annotations on
`get_z_rotation` / `set_z_rotation`. The `@interface` decorator
wraps each method as `classmethod(abstractmethod(...))` at
import time, but ty doesn't track the wrap and flags every
call site as `missing-argument` plus the `pass` body as
`empty-body` against the declared return type, plus the
`bpy.types.Object` forward-ref as `unresolved-reference`.
Reverting to v0.8.0's untyped style (matching the sibling
`get_absolute_matrix(cls, obj)` stub) clears six ty errors at
the cost of zero runtime semantics — the abstract stubs only
serve as registry markers, concrete `tool.Surveyor.*` carries
the real signatures.
Generated with the assistance of an AI coding tool.
Surfaces any regression in:
* the modules dict in bim/__init__.py (added a folder, forgot the entry)
* PointerProperty wiring on bpy.types.{Scene,Object,...}
* registry-driven GizmoPreferences<Name> auto-registration in
tool.Parametric.iter_gizmo_preference_classes
* bpy.app.handlers append/remove balance
* every register()/unregister() across the 45+ feature modules
as a single PASSED/FAILED test instead of the silent "addon failed to
enable" users encounter in a fresh Blender. Paired with the existing
test_parametric_registry.py contract tests, this catches both the
registry-shape regressions (operators/PropertyGroups/predicates) and
the registration-mechanics regressions (PointerProperty types not
registered before their owners).
Generated with the assistance of an AI coding tool.
Three corrective fixes folded into one commit. All surface as
addon-load / save-time exceptions on v0.8.0's bim layer because
PR2's tool.Parametric refactor over-committed to the PR4 contract.
1. iter_gizmo_preference_classes — the previous implementation
returned only the shared GizmoPreferencesFeature class. v0.8.0's
bim/ui.py declares PointerProperty fields ('door', 'window', ...)
on GizmoPreferences that point at per-feature
GizmoPreferences<Name> classes; those must be registered BEFORE
GizmoPreferences itself. The shared-class-only return broke
addon registration with:
'door' PointerProperty could not register (see previous error)
Restore the v0.8.0 per-feature lookup (iterate EDIT_TYPES, look
up each GizmoPreferences<Capitalize(name)> on ui_module) and
keep the shared-class lookup as forward-compat. Tag FIXME(PR5).
2. EDIT_TYPES — drop the array / pipe_segment / duct_segment
entries from the registry. Their bim.finish_editing_<name>
operators land with PR4. Registering them in PR2's EDIT_TYPES
without the operators makes auto-commit-on-save dispatch a
non-existent finish_op for any object whose
BIM<Name>Properties.is_editing flag is True, raising:
RuntimeError: 'bim.finish_editing_array' must be a registered
tool.Ifc.Operator subclass for undo-safe IFC mutation
PR4 re-adds the three entries together with their operators.
Tag FIXME(PR4).
3. tool.Blender.Modifier shim block — upgrade the prose comment to
a formal FIXME(PR5) marker so the PR5 cleanup sweep finds it via
grep alongside every other tagged shim site.
Generated with the assistance of an AI coding tool.
Fixes addon-load ImportError that surfaces when tool/geometry.py
and tool/model.py (extracted in C8 / C9) reference symbols that
don't exist on v0.8.0:
* bim/ifc.py: get_cache_or_detect_lock — IfcStore.get_cache
variant that tracks the multi-instance-cache-locked-by-other-
process flag, sets it on PermissionError, clears it (along with
the dismiss flag) on subsequent success. Used by
tool.Geometry.* to gate IFC cache reads without crashing when
another Blender instance holds the cache lock.
* tool/cad.py: WELD_TOLERANCE constant + paired CAD helpers
(auto-detect-curves vertex precision, polyline normal helpers,
etc.) used by tool.Model.* + by the parametric model operators
that land in PR4.
Both modules had zero upstream commits since the gizmos-8088 fork
point — safe bulk extraction. PR4 has no caller-line work for
either file (the additions are pure additions, no existing API
removed); the v0.8.0 callers of get_cache_or_detect_lock and
WELD_TOLERANCE are the PR2-scope files that needed them.
Generated with the assistance of an AI coding tool.
The previous commit moved is_<type> predicates off tool.Blender.Modifier
onto tool.Parametric, and earlier C4 moved the Array helper bag off
tool.Blender.Modifier.Array onto tool.Array. PR4 will migrate every
caller; this commit keeps the OLD entry points alive as thin delegates
so PR2 ships without breaking ~30 caller sites that still spell the
old API in v0.8.0:
* tool.Blender.Modifier.is_door / is_railing / is_roof / is_stair /
is_wall / is_window — delegate to tool.Parametric.is_<type>.
* tool.Blender.Modifier.Array.bake_children_transform / constrain_
children_to_parent / get_all_children_objects / get_all_objects /
get_children_objects / get_modifiers_data / remove_constraints /
set_children_lock_state — delegate to tool.Array.<same name>.
These shims are removed in PR5's cleanup commit once PR4 has rewritten
the call sites in bim/import_ifc.py, bim/module/geometry/operator.py,
bim/module/geometry/data.py, bim/module/model/array.py + the per-feature
operators (door, wall, window, railing, roof, stair, ui).
Generated with the assistance of an AI coding tool.
tool.Parametric becomes the central registry for Bonsai's parametric
features (wall, slab, door, window, railing, roof, stair, plus
mep-segment variants). Each feature registers a ParametricObject spec
declaring its enable/finish/cancel op names, props accessor, regen
callback, and is_element_type predicate.
Public surface:
* tool.Parametric.WALL / SLAB / DOOR / WINDOW / RAILING / ROOF /
STAIR / PIPE_SEGMENT / DUCT_SEGMENT — typed accessors per feature.
* tool.Parametric.is_wall / is_door / is_window / is_railing /
is_roof / is_stair — element-type predicates that move off
tool.Blender.Modifier into the parametric registry. The next
commit adds backward-compat shims on tool.Blender.Modifier so
v0.8.0 callers keep working.
* tool.Parametric.is_object_editing(obj) — returns the registered
feature an object is currently editing, or None.
* tool.Parametric.run_bim_op(op_name) — invoke a parametric op by
bl_idname.
* tool.Parametric.heal_stale_edit_flags — clear is_editing flags
on file load so a saved-mid-edit project doesn't leave gizmos
poll-locked.
* supports_build_edit_lifecycle field on ParametricObject — declares
whether the feature implements the build/edit/cancel triad.
The previous bare `print(f"Bonsai: commit of {obj.name!r} via
{finish_op} failed: {e}")` exception-handler is replaced with
logger.warning(..., exc_info=True). Same channel (Bonsai configures
logging to the Blender console at WARNING level), strictly more
information (full traceback), correct idiom for an error-path
message. A second logger.warning is added for parametric predicate
failures, also exception-handler scope.
Generated with the assistance of an AI coding tool.
tool.Model gains:
* get_pipe_segment_props / get_duct_segment_props — typed prop accessors
for the MEP-segment edit lifecycle.
* resolve_active_props_for_edit — picks the right BIM*Properties to
drive a parametric edit triad based on the active object's IFC class.
* mirror_parent_void_fillings_to_children — when an array parent has
hosted fillings (door/window in a wall), replicate the same fill
rels onto each array child. Uses tool.Array.get_parametric_propagation_
targets so the propagation stays within the array family (the old
get_all_element_occurrences over-propagated to standalone occurrences
of the same type, which silently mutated unrelated arrays).
* unshare_opening_representation — fork a shared IfcShapeRepresentation
so editing one opening doesn't mutate its array sibling.
* duplicate_ifc_objects gains a post-condition select-restore on the
array parent so callers don't get a deselected parent for N>=2 arrays.
sync_object_ifc_position is kept as a thin delegate to
tool.Geometry.commit_placement_if_moved (the new home, added in C8) so
the 6 v0.8.0 callers in mep / product / system don't AttributeError;
PR4 migrates each caller and removes the delegate.
tool.Pset gains:
* upsert_pset — get-or-add-or-edit in one call.
* write_bbim_data — JSON-encode + write BBIM_* metadata in one call.
tool.Slab is new — slab-specific reads (active extrusion, axis
direction) used by the slab gizmos, pure-IFC, no PropertyGroup mutation.
Generated with the assistance of an AI coding tool.
Adds:
* get_body_representation(element) — DRY of the repeated
ifcopenshell.util.representation.get_representation(element, "Model",
"Body", "MODEL_VIEW") call across slab / wall / opening / stair /
roof / door / window / mep. One central place to read the body rep;
every caller stops re-spelling the four magic strings.
* has_axis_representation(element) — predicate for elements with a
GRAPH_VIEW Axis representation. Used by the wall/MEP path decorators
to skip elements without an unambiguous 1D path.
* has_material_styles(element) — predicate for whether the element
carries IfcStyledItem material assignments.
* restore_placement_from_ifc(obj, element) — snap obj.matrix_world back
to element's committed IFC placement + rebaseline the drift checksum.
* restore_or_rebaseline_placement(obj, element) — Cancel-flow helper:
restores if ObjectPlacement exists, just rebaselines the checksum if
not.
* detach_representation(product) — remove the active representation
from a product without deleting the entity (used by parametric
rebuilds that wipe + re-add).
commit_placement_if_moved docstring expanded with a "drop-in scope"
note so callers don't redundantly wrap it in an is_moved check that
the helper already does.
Switches the duplicate-aware helper calls (formerly tool.Root.*) to
tool.Duplicate.* now that the service exists (C6).
Generated with the assistance of an AI coding tool.
Adds:
* ViewportDecorator base class — install/uninstall/draw lifecycle for
3D viewport gpu overlays, with handler-rollback-on-failure so a
partial install can't leave dangling draw handlers.
* sync_all classmethod — drive each listed ViewportDecorator subclass
to its desired install state in one call.
* is_view_top_down + top_down_factor — viewport-camera orientation
predicates used by gizmo billboarding and decorator layout.
* get_screen_up_world — screen-up vector in world space for gizmo
text orientation.
* are_viewport_gizmos_enabled — central gate for the global
draw_gizmos_in_3d_viewport pref, replacing duplicated prefs reads.
* DecoratorColors NamedTuple + get_decorator_colors — single source
for the colour palette every viewport decorator binds.
Preserves Ryan Schultz's add_layout_hotkey_operator polish (719309571,
2026-05-25): the row-position move + separator(factor=1) between the
modifier and key icons stay intact in this extraction.
Generated with the assistance of an AI coding tool.
Extract the duplicate-aware relationship-walk + restoration logic
(get_decomposition_relationships, get_connection_relationships,
get_port_connection_relationships, recreate_decompositions,
recreate_connections, recreate_port_connections, consume_warnings)
out of tool.Root into its own service.
tool.Root's responsibility is identity and addressing of IFC roots;
the duplicate-aware bookkeeping of "before duplication, what relations
did this graph have, and how do I restore them on the new copies?"
deserves its own home. The split was already declared on core/tool.py
(C2); this commit lands the concrete tool.Duplicate implementation.
tool.Root keeps its own copies of the methods on v0.8.0's tool/root.py
during this PR so callers in bim/module/spatial/operator.py keep
working at runtime; the Root cleanup lands in PR4 alongside the
caller updates.
Generated with the assistance of an AI coding tool.
Adds:
* direction_from_port_pair(port_a, port_b) — derive the connect_port
direction kwarg from each port's FlowDirection (NOTDEFINED for
non-canonical pairs). Centralises a pattern that callers were
inlining inconsistently.
* tool.System.walk_connected_mep_elements — BFS over connected MEP
flow elements via IfcRelConnectsPorts.
* tool.System.get_port_world_position — port placement → world-space
Vector, used by the MEP path decorator.
* tool.System._build_decoration_data — cached decoration metadata
for the MEP system-path overlay.
Plus a get_port_relating_element return-type tightening (Union with
None) and a partial-init cycle workaround on bim.module.system.data
imports (now function-local — top-level import triggered the cycle
through tool.Ifc.Operator).
Generated with the assistance of an AI coding tool.
Top-level array-domain service extracted out of tool.Blender.Modifier.Array.
Owns the BBIM_Array pset graph navigation (constrain_children_to_parent,
remove_constraints, get_modifiers_data, get_children_objects,
get_all_children_objects, get_child_layer_index, bake_children_transform),
plus the Blender-side CHILD_OF constraint lifecycle that ties each child
replica to its parent's transform.
Array's own module gives the parent/child semantics a clean home — array
behaviour was previously scattered between tool.Blender.Modifier and ad-hoc
helpers in bim/module/model/array.py. The relocation eliminates the inline
duplication and gives Bonsai callers a single import surface.
Generated with the assistance of an AI coding tool.
Bpy-permitted wall reads — get_axis_local_extent, get_length_and_height,
get_x_angle, get_path_connection_location, walk_connected_walls — used
by gizmo lambdas that need wall dimensions and join topology without
the side effect of loading the wall's draft BIMWallProperties (the
loader mutates PropertyGroup state and would clobber the wall's own
gizmo state when both the wall and a hosted filling are selected).
All reads go through ifcopenshell.util.representation / .util.element
so the IFC graph stays the source of truth. tool.Wall consumes
core.model's PARALLEL_DOT_THRESHOLD + collinearity helpers (no inline
magic numbers).
Generated with the assistance of an AI coding tool.
Declares the bpy-free contract for tool services landing in subsequent
commits — tool.Wall, tool.Array, tool.System, tool.Duplicate (extracted
from tool.Root), tool.Parametric, plus minor additions on existing
interfaces (tool.Spatial.get_host_element / get_host_wall,
tool.Geometry.has_axis_representation / has_material_styles,
tool.Surveyor.get_z_rotation / set_z_rotation).
The @interface declarations are empty-bodied; concrete implementations
land in the per-service tool/* commits below. Keeping the contract in
core lets core/* helpers and tests reference the surface without
importing the concrete tool modules.
Moves get_decomposition_relationships + recreate_decompositions off
tool.Root onto the new tool.Duplicate (extraction of duplicate-aware
behaviour into its own service).
Generated with the assistance of an AI coding tool.
core/model.py gains:
* Three calibrated dot-product / distance thresholds — PARALLEL_DOT_THRESHOLD
(~2° from parallel, cos(2°) ≈ 0.9994), COLLINEAR_LINE_TOLERANCE (50mm
perpendicular distance for two parallel wall axes to share a line),
BASELINE_OFFSET_TOLERANCE — replacing inline magic numbers that the
wall-join classifier, fillet-state machine, and gizmo preview decorator
all read from.
* Pure wall-join geometry helpers (project_axis_intersection,
are_axes_collinear, classify_wall_join_state, wall_join_preview_lines,
resolve_extend_walls_target, extrusion_depth_from_vertical_height,
length_and_height_from_extrusion). They take primitive tuples + floats,
no bpy, no ifcopenshell — testable in the core lane.
core/product.py is new — pure-Python aggregate-walk helpers (resolve_host_
of_product, collect_decomposed_products) that downstream tool/spatial and
tool/aggregate consumers can call without importing ifcopenshell at module
load.
Generated with the assistance of an AI coding tool.
add_railing_representation now factors into two parts:
* compute_wall_mounted_handrail_geometry returns a pure-geometry
WallMountedHandrailGeometry dataclass (handrail polyline + support
list + terminal caps), no IFC mutation.
* add_railing_representation wraps that dataclass into an
IfcShapeRepresentation as before.
Downstream consumers that want the same math without round-tripping
through an IFC file (Blender gizmo previews, viewport drafts) now
drive compute_X directly. Future add_X_representation work in the
geometry API is encouraged to follow the same shape — a sibling
compute_X function + thin IFC wrapper.
The railing_type parameter is dropped from the signature — only
WALL_MOUNTED_HANDRAIL was ever supported, so the kwarg was dead.
The Bonsai railing-modifier caller is updated in the same commit
to stop passing it; without that update Bonsai's
finish_editing_railing_path raises TypeError on the first edit.
RailingSupport and WallMountedHandrailGeometry use @dataclass(slots=True)
— they're constructed N-per-cap during arc sampling, so the per-instance
overhead matters.
Public symbols (RailingSupport, TERMINAL_TYPE,
WallMountedHandrailGeometry, compute_wall_mounted_handrail_geometry,
add_railing_representation) re-exported from ifcopenshell.api.geometry.
New test/api/geometry/test_add_railing_representation.py covers the
compute/wrap contract.
Generated with the assistance of an AI coding tool.
Drops the module-local ``mm()`` helper in favour of the centralised
``ifcopenshell.util.unit.mm_to_m`` (added earlier in this PR). The
``as mm`` import alias preserves the existing call sites' readability.
Generated with the assistance of an AI coding tool.
Drops the module-local ``mm()`` helper in favour of the centralised
``ifcopenshell.util.unit.mm_to_m`` (added earlier in this PR). The
``as mm`` import alias preserves the existing call sites' readability.
Generated with the assistance of an AI coding tool.
ShapeBuilder gains module-level NP_X / NP_Y / NP_Z / NP_XY / NP_XZ /
NP_YZ / NP_YX axis-index constants. Downstream geometry builders had
been redefining local copies for indexing np.ndarray vectors of shape
(3,) or (N, 3); centralising removes the duplication.
mep_transition_length and mep_transition_calculate verbose default
flipped from True to False. The prints are diagnostic-only output;
True-by-default spammed the console on every transition computation,
which fires per-fitting on IFC load.
Generated with the assistance of an AI coding tool.
Centralises the millimetre-to-metre conversion shortcut that
add_door_representation and add_window_representation each defined
locally. Subsequent commits in this PR switch both call sites to
import this from util.unit, removing the duplicate definitions.
Generated with the assistance of an AI coding tool.
Use add_layout_hotkey_operator for draw_regen_operations so the Regen
button shows text and shortcut icons in the sidebar like all other
panel buttons. Add a separator between modifier and key icons for
readability.
test_parametric_lifecycle.py covers the door/window/railing/roof
state-transition contracts (enable/finish/cancel; no-op on
non-matching elements; draft preserved on finish-time failure)
that the registry smoke test never exercised.
test_parametric_registry.py gains a check that every is_<name>
predicate stays total (never raises on a non-matching IFC entity)
— a raising predicate would break the save path for unrelated
types. Also rewrites the gizmo-prefs check to read __annotations__
instead of hasattr, which depended on Blender registration timing.
Generated with the assistance of an AI coding tool.
Aligns with the test/bim/ convention: heavy imports go inside test
functions so the autouse _require_real_bpy fixture skips cleanly
when bpy is mocked, rather than module-level imports failing at
collection time and erroring out the whole file.
Generated with the assistance of an AI coding tool.
PowerShell and some wrapper scripts on Windows occasionally strip
or reorder the `--` separator before Blender sees it, dropping the
pytest args into Blender's positional file-load slot ("File format
is not supported"). The env var carries the same args via a
shell-evaluation-free channel. Default `--` path is byte-identical
to the pre-change behaviour.
Generated with the assistance of an AI coding tool.
Docstrings naming sibling methods, private helpers, test files, or
historical symbols silently go wrong on rename. Strip Sphinx :meth:
/ :class: / :func: / :attr: markup that mostly added noise (no
Sphinx in this project), and rewrite five docstrings that cited
specific test paths or private hooks to describe the behaviour
instead.
Generated with the assistance of an AI coding tool.
The dialog's only outcomes were "Apply & Save" (same as silent save)
or "Cancel" (same as not saving) — net friction with no actual choice.
Auto-commit stays as the safety net; the count now suffixes the
existing save-success report so it isn't immediately overwritten.
Generated with the assistance of an AI coding tool.
When two GizmoDimension hit regions overlap (a short dimension
nested inside a longer one along the same axis), the larger one
used to win because hit boxes are scaled by world-space length —
the long box fully contains the short one, leaving the short
gizmo unreachable. The larger gizmo stays clickable at its
exposed ends, so smaller-wins is the right UX default.
Sets self.select_bias = -self._dimension_length inside
GizmoDimension.set_dimension_length. The smaller gizmo writes a
less-negative depth value in the GPU select buffer and wins the
tie-break. select_bias is unused elsewhere in the codebase, so
icon and arrow gizmos keep bias=0 and are unaffected (icons
correctly still win against dimensions, since 0 > -length).
Adds test/bim/module/drawing/test_dimension_gizmo_priority.py
with 5 cases: direct ordering, monotonicity across length ranges,
abs() handling for signed dimensions, and NaN/Inf safety.
Generated with the assistance of an AI coding tool.
set_icon_gizmo_position computed
``mw @ (Translation @ billboard_rot @ Scale)`` — the object's world
matrix was applied AFTER the billboard rotation, so any non-trivial
object rotation (e.g. a wall rotated in plan, a stair rotated to
match a corridor) carried over into the icon's transform and tilted
it edge-on to the camera instead of facing it.
Switch to ``billboarded_at(world_pos, billboard_rot, scale)`` where
``world_pos = mw @ local_pos``: translate to world space first, then
apply the billboard rotation independently of the object's rotation.
This matches the manual pattern the base class's
``update_editing_gizmos`` already uses for validate/cancel/cycle for
exactly this reason.
Drops the now-stale workaround docstring on
``GizmoWallEdition._update_icon_row_extras`` that documented why it
bypassed ``set_icon_gizmo_position`` — the helper does the right
thing now.
Adds ``test/bim/module/model/test_stair_gizmos.py`` as the regression
guard: parametrised over six rotation angles, asserts that the rotation
part of the resulting matrix equals ``billboard_rot`` (no contribution
from ``mw``'s rotation) and that the translation lands at
``world_pos``. Also exercises ``set_icon_gizmo_position`` end-to-end via
a stub gizmo to catch the exact shape of the previously-broken call
site.
Generated with the assistance of an AI coding tool.
Walls gain in-viewport parametric editing matching the door/window/stair
UX: drag handles for length, height, slope (x-angle), layer baseline
cycle, plus cursor-anchored quality-of-life operators (split at cursor,
extend to cursor, extend height, rotate 90, toggle openings) and
two-object state-machine gizmos (unjoin / merge / join-corner /
extend-to-wall / extend-vertically / add-opening).
Wall enters tool.Parametric.EDIT_TYPES, so save-time auto-commit,
GizmoPreferencesWall registration, and the in-progress-edit predicates
all light up automatically through the registry plumbing landed two
commits back.
The three-layer commit model (drag -> BIMWallProperties -> bmesh
preview -> Finish -> single ifc.run) means dragging a handle through
hundreds of intermediate values produces zero extra IFC entities. A
no-op enable->finish round-trip is byte-identical. The snapshot diff
in FinishEditingWall skips unchanged params.
_commit_active_wall_edit_if_any ensures cursor-anchored operators see
committed geometry, not the draft preview box.
Also lands the `prompt_auto_commit_parametric_edits` BoolProperty on
BIM_ADDON_preferences (consumed by the auto-commit dialog landed in
the framework commit) and refactors
`draw_{door,window,stair}_gizmo_parameters` into a shared
`_draw_parametric_gizmo_parameters` helper that the new
`draw_wall_gizmo_parameters` reuses. This commit and the framework
commit are stacked - the framework commit references the BoolProperty
defined here, so they must land together.
Tests cover pure math (core/test_model.py), DimensionGizmoConfig text
formatter, GizmoWallExtendVertically.poll() preconditions, and the
refresh_post_commit cache-invalidation regression. BDD scenarios in
model.feature cover the edit triad, auto-commit on save, and the
two-object gizmos. Documentation added to creating_walls.rst.
Generated with the assistance of an AI coding tool.
Three small post-landing cleanups against the parametric framework commit:
* core/model.py had `are_axes_collinear` and `closest_endpoint_midpoint`
each defined twice — Python silently kept the second copy, the first
was dead code. Removed the dead copies; runtime behavior unchanged
(the live versions were already the kept ones).
* bim/__init__.py's `_parametric_gizmo_preference_classes` docstring
named the wrong link in the import chain (`tool.blender → bim.ifc`).
The real chain is `tool/ifc.py` (and ~6 other tool/* modules) which
import `from bonsai.bim.ifc import IfcStore` at module load. Updated
docstring to cite that root cause and the architectural fix (move
`IfcStore` out of `bim/`).
* tool/blender.py's `from bonsai.bim.ifc import IFC_CONNECTED_TYPE`
carried a 5-line comment claiming it was "lazy" to avoid a circular
load. The import sits inside an `if TYPE_CHECKING:` block with
`from __future__ import annotations` — it never runs at runtime
regardless. Comment removed; the TYPE_CHECKING guard is
self-explanatory.
Generated with the assistance of an AI coding tool.
Establish a single source of truth for parametric element types (door,
window, stair, railing, roof). tool.Parametric.EDIT_TYPES drives:
- BIM<Name>Properties PointerProperty attachment via the registry
- GizmoPreferences<Name> class registration in bim/__init__.py
- save-time auto-commit of pending draft edits
- the refresh_post_commit epilogue called from IfcStore after every IFC
mutation, which fixes the stale-header bug where in-place hotkey
mutations (S_E / C_E) left BIMModelProperties and the gizmo cache
pointing at obsolete values.
Refactors door/window/railing/roof onto shared mixins from
bim/parametric_lifecycle.py (FeatureModifierEditMixin and
PathPreservingEditMixin); stair gets the lock-gizmo refactor and
frame-cache integration. Behavior preserved.
Adds BaseParametricGizmoGroup._prime_frame_caches so the parametric
gizmos stop re-deriving preferences, view direction, and billboard
rotation per frame; reorders poll() to short-circuit on the cheapest
predicate first. Adds the icon library + BillboardingGizmoGroupMixin
that the wall feature in the next commit will consume.
Generated with the assistance of an AI coding tool.
When a project has a ifc file associated, selecting non-ifc objects and duplicating them with SHIFT + D now correctly both duplicate them, keep the new objects selected and starts the transform modal. IFC objects behaviour is unaffected.
The x-angle transformation for LAYER3 slabs assumed SweptArea
is always IfcArbitraryClosedProfileDef (which has OuterCurve),
but composite profiles use IfcCompositeProfileDef instead.
Apply the coord scaling to each sub-profile individually.
Generated with the assistance of an AI coding tool.
generate_section_reference_points had no handler for
MODEL_VIEW target view, causing it to silently return
None. Add MODEL_VIEW branch that clips the section line
to XY camera bounds while preserving the Z coordinate
for correct 3D placement.
Generated with the assistance of an AI coding tool.
Guard the int() cast on CardinalPoint in
BIM_OT_edit_assigned_material so a None value (no cardinal
point set) no longer raises a TypeError.
Generated with the assistance of an AI coding tool.
Extracted from the font file like so:
python3 -c "
from fontTools.ttLib import TTFont
tt = TTFont('src/bonsai/bonsai/bim/data/fonts/OpenGost Type B TT.ttf')
for record in tt['name'].names:
if record.nameID == 13:
print(record.toUnicode())
"
IFC2X3 representations have no HasShapeAspects inverse; opening the
Geometry & Materials subpanel on an IFC2X3 object raised AttributeError
and left the items list empty. Wrap the access with a getattr default
so pre-IFC4 schemas return an empty iterable, and pin the contract with
an AST forward-compat guard that scans bim/, tool/, and core/ for any
future direct .HasShapeAspects access.
Closes#8157
Generated with the assistance of an AI coding tool.
Port gizmos-8088's railing gizmo block to v0.8.0:
- _RailingEditMixin (PathPreservingEditMixin specialisation) +
EnableEditingRailing / CancelEditingRailing / FinishEditingRailing
edit triad
- CycleRailingType (2-value type cycler) + ToggleRailingUseManualSupports
one-shot + EditRailingTerminalType
- FlipRailingPathOrder + EnableEditingRailingPath /
CancelEditingRailingPath / FinishEditingRailingPath path-edit
operators (mutually exclusive with the schematic frame)
- GizmoRailingSchematic (BaseSchematicGizmoGroup specialisation) —
axonometric schematic frame with per-attribute dimension gizmos
for FRAMELESS_PANEL + WALL_MOUNTED_HANDRAIL railing types;
hover-on-attr highlights the schematic edges tagged with the
matching feature
Tests: test_railing_lifecycle.py (280 LOC) +
test_railing_schematic.py (272 LOC).
Drops the per-feature GizmoPreferences{Door,Window,Stair,Wall,Roof,
Railing} PropertyGroups that the source commit added to bim/ui.py
— that finer-grained per-attribute toggle model was deliberately
collapsed to flat per-feature bools in the PR5b prefs sweep, and
GizmoRailingSchematic gates on the flat ``prefs.gizmos.railing``
bool via ``gizmo_pref_name`` so no functionality is lost.
Generated with the assistance of an AI coding tool.
The check `len(bytedata) == n * 2` was wrong: float64 is 8 bytes per
element, not 2. Legacy float64 checksums fell through to the float32
reader and produced a (2n,)-shaped array, breaking is_moved() and
is_camera_moved() with `ValueError: operands could not be broadcast`
on .blend files saved by Blender <5.0.
Adds a parametrized regression test covering both n=3 (translation)
and n=9 (rotation) for both dtypes.
Generated with the assistance of an AI coding tool.
GizmoWallEdition.position_gizmos used props.anchor_x / props.length
for the in-range check (split icon visibility) and perpendicular
gizmo placement. Those props mirror IFC and are re-primed by
_maybe_resync_wall_props_from_ifc — any operator path that skips
the re-sync leaves the perpendicular gizmo clamped to the previous
wall extent, so the icon parks at the old wall end instead of the
cursor's orthogonal projection. Visible after a wall mutation as
the perpendicular icon landing way off the cursor in top-down view.
Switch to the mesh bbox along local X. recreate_wall rebuilds the
mesh to match the current IFC body on every wall mutation, so
bound_box is authoritative without an explicit props sync.
Generated with the assistance of an AI coding tool.
MEP elements imported as tessellation / brep (no IfcExtrudedAreaSolid
or IfcSweptDiskSolid in their body representation) can't be
parametrically edited — the gizmos offer affordances the geometry
kernel has no path to honour. tool.System.has_parametric_body
inspects the Model/Body/MODEL_VIEW representation and returns True
only when at least one item resolves to one of the two
profile-sweep primitives.
The gate is wired into:
- GizmoMEPActions.is_eligible_object (the action icon group)
- _active_is_flow_segment / _active_is_bend_fitting visibility
predicates the icon row consults per-icon
- GizmoPipeSegmentEdition / GizmoDuctSegmentEdition is_element_type
tool.Parametric.is_pipe_segment / is_duct_segment stay IFC-class-only
so their truth-table contract test keeps reading a single concern.
Generated with the assistance of an AI coding tool.
Pure-math parallelism check (value ≡ 0 mod π within VTX_PRECISION)
that lived as a module-private helper in mep.py belongs next to
tool.Cad.is_x — same comparator family, no MEP-specific knowledge.
Other features with rotation-difference checks (wall fillet, roof
slope, railing terminus) now have a sanctioned spelling.
Generated with the assistance of an AI coding tool.
Four standalone test files pinning contracts the production code
already honours:
- test_mep_actions_cache.py: GizmoMEPActions visibility-predicate
cache evicts on selection or generation change.
- test_mep_bend_preview_cache.py: bend decorator polyline cache
re-uses within a generation and rebuilds on generation bump.
- test_mep_distribution_fit_smoke.py: bim.fit_flow_segments
round-trips a 3-segment polyline without raising.
- test_preview_cancel_ops_forward_compat.py: AST scan ensures every
preview Enable* operator has a paired Cancel* operator with the
matching prop reset.
Generated with the assistance of an AI coding tool.
Two small refactors:
- apply_transform_modal_draw_gate(group, context) replaces the
three-line _is_transform_modal_active + _hide_all_non_modal_gizmos
pair that BillboardingGizmoGroupMixin, BaseParametricGizmoGroup
and BaseSchematicGizmoGroup all repeat in draw_prepare.
- decorator.py renames _stroke_lines_alpha to a public-scope
draw_polyline_segments and drops the no-longer-private companion
docstring reference; the function is now usable by sibling
decorators that draw polyline overlays.
Plus a few one-liner tweaks in tool/model.py and opening.py
following the helper rename.
Generated with the assistance of an AI coding tool.
Two hot paths the gizmo polls fire every viewport event memoise
their result against tool.Parametric.get_geom_generation():
- tool.Blender.Modifier.any_selected_array_child caches the
per-selection scan against the selection identity-set + the
IFC generation token so a stable selection during a drag
doesn't re-walk every selected object's BBIM_Array pset every
frame.
- bim/module/model/wall.py grows a pair-predicate + connection
cache that the wall topology gizmos hit; both keyed on
(pair_uids, predicate_kind, generation) so a wall split or
axis edit invalidates correctly via the generation bump.
Behavioural contract is unchanged — stale entries are evicted
on generation bump; cache miss returns the same value the
un-cached path returned.
Generated with the assistance of an AI coding tool.
bim/module/model/conftest.py exposes the autouse _require_real_bpy
skip-guard, four make_* factories (obj / element / context /
ifc_file), and a patched_tool context-manager factory that wires
the half-dozen tool.* boundary patches every gizmo + decorator
test was repeating.
Existing test files in the directory drop their local copies of
_require_real_bpy and adopt the patched_tool / make_* fixtures
where the call site simplifies — test_mep_port_operators.py is
the biggest beneficiary (−89 LOC).
No production behaviour change.
Generated with the assistance of an AI coding tool.
Three concerns bundled by file boundary (all in mep.py):
- Extract bend preview operators + GizmoBendPreview into a focused
mep_bend_preview.py module; preview_base.py grows the shared helper
set both bend and other previews now consume; classes tuple in
model/__init__.py updated to register the new module.
- Surface ERROR reports on five silent CANCELLED returns in
MEPUnjoinAtPort / MEPRemoveTerminalFitting / MEPUnjoinPair so a
degenerate IFC file ("fitting has no Blender object", "connected
port leads nowhere") shows up in the popup instead of looking like
a no-op.
- DRY: _resolve_active_mep_segment + _require_port_state factor the
segment-id-or-active-object resolve + port-state guard out of every
port operator's prologue; _wire_anchored_icon_targets pulls the
GizmoMEPActions setup() body into an exercise-without-MRO helper so
the wiring-contract tests can hit it without instantiating the
GizmoGroup.
Drops the now-unused preview_base import that the extraction left
behind.
Generated with the assistance of an AI coding tool.
The opening preview's outline used a single-batch two-pass scheme that
dimmed the occluded back pass via alpha=0.25. The visible front pass also
inherited the source decorator color's modest alpha, so the outline read
as subtle on both sides.
Replace with a CAD hidden-line convention: solid full-alpha front pass on
the visible side, world-space dashed back pass on the occluded side. Both
passes use POLYLINE_UNIFORM_COLOR so depth and line-weight paths match.
The dashed batch is built once per object epoch by a new pure helper
tool.Blender.build_dashed_line_segments (pre-segments edges into world-
space dash chunks), then cached via the existing batch-cache mechanism
under "<uid>_dashed".
The solid front pass is rendered at a slightly wider line width than the
dashed back pass so its halo overpowers Blender's WIRE-display overlay
bias at outline pixels — without the asymmetry the wire's anti-z-fight
forward bias makes the LESS_EQUAL comparison narrowly fail and the
dashed pass wins on visible edges too.
Generated with the assistance of an AI coding tool.
MEPConnectElements took obj1_name/obj2_name (Blender object names),
which break when objects are renamed or replicated by array
duplication. Switch to obj1_guid/obj2_guid resolved via
ifc_file.by_guid, with by_guid RuntimeError surfaced as an operator
error rather than a stack trace. DrawPolylineProfile (the sole
in-tree caller) updates to pass GlobalIds.
Generated with the assistance of an AI coding tool.
Pins the geometry contracts the hand-meshed bend body relies on
while IfcSweptDiskSolid round-trip is broken upstream (#8106):
- profile cross-section sampling: circle returns 16 evenly-spaced
points starting at (radius, 0); rectangle returns the four
canonical corners; anything else returns None so the rep swap
is skipped rather than meshed against the wrong section
- parallel-transport framing keeps the cross-section continuous
around L-shaped corners — pinned via start / end ring planes
- initial_basis override seeds the first ring with the source
segment's local +X / +Y axes, fixing the asymmetric-rectangle
twist the world-Z seed produces
Generated with the assistance of an AI coding tool.
Pins which IFC mutation each port operator commits and which
inputs each refuses with CANCELLED:
- MEPUnjoinAtPort removes the fitting + reconnects the two free
ports; refuses if the named port is free or terminal
- MEPRemoveTerminalFitting deletes the terminal element + leaves
the segment's port free; refuses on bridged fittings
- SelectMEPPathMembers walks IfcRelConnectsPorts in both
directions from the active segment and selects every fitting /
segment reachable through the port graph
Boundary mocks for tool.Ifc, tool.System and MEPGenerator stand
in for the IFC fixture; tests assert against the recorded
ifcopenshell.api.* calls.
Generated with the assistance of an AI coding tool.
Pins two regressions the live MEP gizmo group can hit:
- per-icon setup() must write `position` (and `mode` on open-lock
icons) onto every target_set_operator result; the test stands in
for the AttributeError on bim.mep_add_obstruction that surfaced
when a field was dropped from the operator declaration
- each visibility_condition lambda must stay total against None /
non-IFC inputs, since a single raising predicate silently disables
every sibling icon in the group
Generated with the assistance of an AI coding tool.
ProfileDecorator.draw_faces (used by the roof path-edit overlay) and
SystemDecorator.draw_faces called bmesh.ops.triangulate on the live
bmesh — both mutated the input and produced ear-clip fans that rendered
as visible streaks across n-gon roof faces at alpha 0.1. The opening
DecorationsHandler edit-mode branch had a separate bug: it computed
triangles from obj.data.calc_loop_triangles() while iterating the
edit-mode bmesh, so any topology added mid-edit desynced the indices.
Centralise the correct draw path on tool.Blender.draw_bmesh_face_tris
(wraps bm.calc_loop_triangles, non-mutating, beauty triangulator) and
route all three call-sites through it. A forward-compat AST guard walks
every *Decorator / DecorationsHandler class under bim/module/ and pins
the no-bmesh.ops.triangulate rule against future regressions.
Generated with the assistance of an AI coding tool.
A user clicking the pen icon on a typed-product occurrence whose body
representation is mapped from its type would silently mutate every
sibling occurrence's geometry. Add a confirmation dialog at the pen-icon
dispatcher (the single chokepoint every feature routes through) showing
the sibling count, with a session-scoped suppress checkbox.
The check is read-only: tool.Model.get_sibling_occurrence_count wraps
tool.Geometry.get_elements_by_representation against the resolved body
rep and subtracts self + type. A forward-compat AST guard pins the
dispatcher monopoly so any future feature that binds pen_gizmo directly
to a feature-specific enable op fails the test before merge.
Generated with the assistance of an AI coding tool.
Once a bend was created, the only way to retune start_length /
end_length / radius was to delete and recreate from scratch.
EnableBendPreviewFromBend re-opens the preview on an existing
parametric bend: it walks the bend's ports to resolve the two
connected segments, reads start / end length and radius from the
bend type's BBIM_Fitting pset, and sets editing_bend_id on the
preview props. MEPAddBend then deletes the old bend + its port
connections (single undo step) before the recreate path runs, so
finish replaces the bend in place and cancel discards the edit
without touching the original.
GizmoMEPActions surfaces a pen icon on single bend-fitting
selections via the new _active_is_bend_fitting predicate; the icon
dispatches the new operator. Mirror of the wall fillet re-edit
flow (EnableWallFilletPreviewFromCorner + editing_corner_id in
CreateWallFillet).
Test coverage: registration probe for the new operator, an attached
editing_bend_id field probe on the preview umbrella, and a
parametrized truth-table for the _is_bend_fitting predicate
(IfcFlowFitting with BEND PredefinedType, with other PredefinedType,
with no type, IfcFlowSegment, IfcWall, None).
Generated with the assistance of an AI coding tool.
Wall topology mutations (merge / join / extend-to-wall / unjoin /
fillet) applied to a Bonsai array child are silently overwritten by
the next ``regenerate_array``; merge also orphans a GUID listed in
the parent's ``BBIM_Array.Data``. Add a central
``tool.Blender.Modifier.any_selected_is_array_child`` predicate and
gate the five wall topology gizmo groups plus the six bound operators
behind it. Operator gating is defence in depth against keymap / F3
invocation paths that bypass the gizmo.
The base ``_wall_gizmo_poll_gate`` keeps its loose two-check shape
(viewport gizmos + no preview). A new
``_wall_topology_gizmo_poll_gate`` wraps it with the array-child
filter and is what the topology gizmos use. Host-opening gizmos
deliberately stay on the loose gate: openings authored on a child
are preserved through ``regenerate_array`` and track with the
replicated instance.
A forward-compat AST guard walks wall.py for ``GizmoGroup`` subclasses
and asserts each routes its poll through the tighter gate or the
central predicate, with an allow-list for the parametric-edit and
preview-owner exceptions. New wall topology gizmos inherit the
contract by construction.
Generated with the assistance of an AI coding tool.
Selecting a Bonsai-parametric IfcDoor now shows the swing arc(s)
without entering edit mode. A new viewport decorator polls on the
active object, reads the door's BBIM_Door pset, and draws the same
arcs the parametric door swing gizmo would draw — matching the
hinge / panel-width / x-mirror contract minus the is_editing gate.
A forward-compat test walks every door operation type and cross-
checks the readonly decorator's arc selection against the gizmo's
swing-arc config table, so future enum additions fail in both
surfaces simultaneously.
Also disables the inherited 8-pass dark halo on GizmoArc: an open
curve has no enclosed silhouette, so the offset passes read as
ghost arcs rather than a uniform outline. The arc's own cross-
section thickness keeps it legible without the halo.
Generated with the assistance of an AI coding tool.
The MEP bend feature's IfcSweptDiskSolid representation produces
geometrically correct output but fails to round-trip through the
OpenCascade geometry kernel (upstream issue #8106) — the body is
dropped on the next file load. Until upstream is fixed, MEPAddBend
captures the bend centerline in world space before the segments are
extended (otherwise the post-extension axes no longer reach the
original intersection and arc reconstruction is wrong), then after
the fitting is placed it hand-meshes the bend body and swaps the
type's swept-disk representation for an IfcTessellatedFaceSet via
tool.Geometry.export_mesh_to_tessellation + tool.Model.
replace_object_ifc_representation.
The centerline includes the straight start_length / end_length legs
in addition to the arc so the bend covers the full segment-to-
segment span. Sweep uses parallel-transport framing — each ring's
(right, up) basis is rotated by the minimum rotation that maps the
previous tangent to the current one, eliminating the twist a fixed
world-axis reference produces when the tangent crosses the
reference. Cross-section orientation seeds from the source segment's
matrix_world local +X / +Y so asymmetric IfcRectangleProfileDef
ducts land with XDim / YDim on the same axes the segment expects;
parallel transport then preserves that alignment around the arc.
Centerline radius is radius + profile_dim[lateral_axis] to match
MEPAddBend's ref_point_radius — without this offset, the bend legs
fall short of the extended segments by profile_dim * tan(angle/2).
Face winding is left to the caller to correct via
bmesh.ops.recalc_face_normals on the closed bend tube.
Two FIXME(#8106) markers (capture site + helper call site) so both
can be dropped once upstream lands a swept-disk round-trip fix.
Generated with the assistance of an AI coding tool.
IfcTriangulatedFaceSet/IfcPolygonalFaceSet were introduced in
Fix#7992: IFC4 and do not exist in IFC2X3. Previously, requesting an
IfcTessellatedFaceSet representation in an IFC2X3 file silently
fell back to a faceted brep after unassigning material sets.
Add a guard in the update_representation operator (user-facing
error) and in the add_representation API (ValueError) so the
unsupported request is caught instead of failing silently.
Generated with the assistance of an AI coding tool.
GizmoWallEdition gains a fourth cursor-anchored icon that
spawns a perpendicular branch wall from the cursor's
orthogonal projection on the source wall axis. Click forms
a T-junction; shift+click forms an L-corner with the source
wall trimmed at the projection, keeping its longer portion.
The branch inherits the source's spatial container and
centerline baseline so its authored axis matches the source's
alignment rather than the type's default.
Also includes a floor-plane preview quad for the new gizmo,
a floor-Z cross line on the split preview for top-down
visibility, a small bump to QUAD_ALPHA for clearer preview
fills, and a stacking-offset helper that centralises the
cursor-row screen-up step across three call sites.
Generated with the assistance of an AI coding tool.
The MEP one-shot operators (join, unjoin variants, terminal removal,
path-select, obstruction add/remove) had no viewport surface. This
commit adds GizmoMEPActions — the icon-action gizmo group that
surfaces them as billboarded icons around selected MEP elements.
Three anchor regions: a horizontal row above the bbox top
(selection-cardinality icons), per-port endpoints for the three-state
lock / unjoin icons (open lock for PORT_FREE, closed for
PORT_TERMINAL, unjoin for PORT_JOINED — resolved per-frame from
port_connection_state), and the predicted join location
(compute_mep_join_location, shared with the bend preview) for the
join / unjoin_pair pair. Unjoin icons render at full
DEFAULT_BILLBOARD_SCALE with warning-red hover; endpoint lock icons
shrink so the lock row stays subordinate to the row icons. The
group hides itself entirely while a bend preview is active.
MEPAddObstruction grew a position enum (CURSOR / START / END) and a
mode enum (ADD / REMOVE / TOGGLE) so the gizmo can target a specific
port without touching the cursor and dispatch ADD or REMOVE based on
the click target — the lock_open icons drive ADD with position
pinned, the lock_closed icons drive bim.mep_remove_terminal_fitting.
Without the new fields the gizmo wiring (op_props.position = ...)
crashed at setup() with AttributeError on the obstruction operator.
validate_bend_preconditions extracts the type-match and profile-kind
checks MEPAddBend enforces so EnableBendPreview surfaces the
rejection immediately — the user no longer tunes a preview only to
learn at commit time that the segments use an unsupported profile
(e.g. IfcArbitraryClosedProfileDef).
Generated with the assistance of an AI coding tool.
Four discrete one-shot operators driven by the MEP segment's port
state. mep_unjoin_at_port deletes the IfcFlowFitting bridging a
segment's named port to a second element when the port is in the
JOINED state. mep_remove_terminal_fitting deletes the terminal
fitting at a port (closed-lock state) and dispatches by fitting
type — OBSTRUCTION fittings go through MEPGenerator.remove_obstruction
so the segment absorbs the freed length, other terminal fittings go
through the standard delete path. mep_unjoin_pair finds the single
fitting bridging two selected MEP segments and deletes it.
select_mep_path_members walks the connected MEP network from the
active element via IfcRelConnectsPorts and replaces the selection
with every reachable member. Foundation for the MEP Actions gizmo
group which surfaces these operators as icon affordances around
selected segments.
Generated with the assistance of an AI coding tool.
The extend icon used a pure screen-space billboard that always
pointed +X across the screen — the arrow ran horizontally
regardless of the pipe / duct's orientation. The new
billboarded_along_axis helper rotates the gizmo about the camera-
forward axis so its local +X aligns with the segment's local +Z
projected onto the screen, keeping the icon camera-facing but
visually following the extrusion direction. The flip-mirror branch
now reads from cursor-vs-current-end along the segment axis (not
screen-X), so the arrow points away from the current endpoint
regardless of viewport orientation. The split icon stacks
perpendicular to the rotated extend arrow in screen space so the
two don't overlap.
The decorator's green preview line no longer clamps the cursor
projection to min_projected_length — it follows the raw projection
so the line stays visible when the cursor crosses behind the
segment origin (the user still sees where they're pointing even
though the operator floors the actual commit).
Generated with the assistance of an AI coding tool.
Wire up wgpu init + scene load + RAF render so the embedded sample
sidecar paints on the canvas in both browsers.
Root-cause fix: BufferPool::addSubBuffer spin-waited on
PopErrorScope, which resolves via JS microtask on Dawn-web. The
spin blocked the JS event loop, so the microtask never fired and
the first allocation hung the page indefinitely. Skip the
error-scope dance on Emscripten; trust the buffer pointer.
ViewportCore: add initWgpuAsyncWeb (nested-callback adapter→device
chain with AllowSpontaneous mode, no spin) and loadSidecarFromPath
(Qt-free entry point). waitTickInstance becomes a no-op shim on
web; cull-threads / streaming_thread_ / wgpuSurfacePresent gated
off; chunk I/O runs inline.
main_web.cpp: AppState + initWgpuAsyncWeb → buildPipelines (+
HiZ/edge/pick) → loadSidecarFromPath → ready flag → Module._app_ptr
handoff. The RAF loop lives in shell.html (NOT here) because any
RAF helper called from inside Dawn-web's wgpu Promise.then chain
stalls the device callback.
CMakeLists.txt: EXIT_RUNTIME=0 + Module.noExitRuntime=true (shell)
keeps wasm alive past main() so the device promise lands; no
Asyncify; export _raf_tick_c so shell.html's RAF can call it.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The bend preview gizmo group (commit 2) populated a Scene draft but
the user saw nothing in the viewport until they hit finish — they
had to commit blindly. This commit ports the BendPreviewDecorator
(centerline arc + two leg projections on valid geometry, warning-red
axes on invalid in-segment intersections) and the interactive
GizmoBendPreview group (three dimension widgets for start_length /
end_length / radius plus validate / cancel icons). The bend axis
math lives in a pure compute_bend_preview_polylines helper, fed
into both the gizmo group's per-frame positioning and the GPU
decorator's draw path. MEPSegmentExtendPreviewDecorator lands at
the same time because it shares the decorator install / uninstall
plumbing — renders the extend-to-cursor preview line for the
GizmoPipeSegmentEdition / GizmoDuctSegmentEdition extend icons
when hovered, clamping the projected endpoint to the operator's
minimum so the preview matches where the commit lands. The
MEPJoinSegments dispatcher routes two selected MEP segments to
mep_add_transition (parallel) or enable_bend_preview (non-parallel)
— the F3 search entry point that makes the bend preview testable
before the gizmo-icon dispatch lands.
11 new tests in test_mep_bend_preview.py cover the geometry helper
truth table (parallel rejection, right-angle happy path, near-
collinear rejection, in-segment invalid_axes), the
_intersection_past_near parametrized boundary, registration probes
for the lifecycle operators / join dispatcher / gizmo group /
decorator, and the FinishBendPreview RuntimeError catch contract.
6 extend-preview-line tests (deferred from commit 3) join the
existing 35 in test_mep_segment_edition.py.
Generated with the assistance of an AI coding tool.
Spatial.get_root_element walks aggregate / nest / filled-void /
voided-element chains and core.assign_container assigns the container
to whatever the walk returns. For an IfcDoor the filled-void hop
redirects to the IfcOpeningElement, then voided-element to the host
wall, so a user who selects a door and runs bim.assign_container ends
up targeting the wall — and silently no-ops on the door if the wall is
already in the target storey.
Per IFC4 / IFC4.3 (IfcDoor, IfcWindow): the spatial containment of a
filling is defined independently of the filling relationship. Major
exporters (Revit, ArchiCAD, Tekla, Allplan) emit independent
ContainedInStructure on doors / windows accordingly. Drop the
filled-void / voided-element hops from the walk; aggregate and nest
remain — those are true sub-part relationships where the parent
legitimately owns the container.
New TestGetRootElement in test/tool pins the new contract (filling
resolves to itself) plus the retained aggregate / nest / loose-element
paths so a future PR that re-adds either hop is caught. Two new
TestAssignContainer cases in test/core pin filling-to-self through the
core layer and per-element can_contain filtering.
Generated with the assistance of an AI coding tool.
Pipe and duct segments had no parametric-edit affordance — the only
length edit path was a property panel value with no live preview.
This commit ports the per-segment parametric edit triad
(enable / finish / cancel) plus a cursor-anchored extend operator
and a cursor-projected split operator into one gizmo group per
segment type. The two PropertyGroups (BIMPipeSegmentProperties,
BIMDuctSegmentProperties) host the draft length plus snap fields
so cancel / no-op-finish restore the segment to its exact pre-edit
visual state including a non-identity pre-edit scale. Length
commits are written through DumbProfileJoiner.set_depth and
auto-dispatch bim.regenerate_distribution_element so adjacent
fittings track the port move. The split operator preserves
downstream port connectivity and runs through tool.Ifc.run for
single-step undo. The two segment types are now first-class
entries in tool.Parametric.EDIT_TYPES, which resolves the FIXME
on auto-commit-on-save dispatch.
35 unit tests cover predicate truth tables, segment_world_length
geometry, preview-via-scale / restore-scale helpers, gizmo class
wiring, lifecycle operator registration, dimension matrix_position
rotation respect, and lifecycle drift-handling. The 6 extend-
preview-line decorator tests stay deferred until the bend preview
decorator commit lands MEPSegmentExtendPreviewDecorator.
Generated with the assistance of an AI coding tool.
Parametric gizmos (wall/door/window/stair/roof/array/MEP) recompute
matrix_basis every frame from obj.matrix_world. While Blender's
transform modal (G/R/S and the Bonsai macro overrides) drags the
matrix, the gizmos slide off-cursor and fight the transform overlay.
Detect via context.window.modal_operators (Blender 4.2+) — the
collection of running modal operators. Gate poll() (forward-compat)
and draw_prepare() (production path: gizmo.hide=True preserves the
GizmoGroup across the drag instead of destroying it). Cover the
Bonsai macro override for G key (and Shift/Alt/Ctrl+Shift+D) by
matching the BIM_OT_* macro idnames that surface in modal_operators.
Forward-compat test walks every parametric-edit module for GizmoGroup
subclasses and asserts poll returns False with the detector mocked,
so new gizmo groups inherit the hide automatically.
Generated with the assistance of an AI coding tool.
Every preview operator (commit + cancel for both bend and wall
fillet) was inlining the same 3-4 line cleanup: set is_active to
False, zero every *_id IntProperty. The new clear_preview_state
helper in preview_base.py introspects bl_rna and applies that
contract generically — adopters become a single call. Two new tests
pin the contract: every *_id IntProperty zeroes, non-id fields stay.
Generated with the assistance of an AI coding tool.
Extract the pending_opening_recut tracking, three operators (apply /
dismiss / select), Project-panel banner, and the sibling
multi-instance warning banner (its backend helpers already landed
on this branch) from commit a85ed6032 on gizmos-8088.
All tool.* dependencies (Geometry.reimport_element_representations,
Blender.set_objects_selection, Array.*) and IfcImporter.gross_elements
are already on this branch -- no other diffs from a85ed6032 are
pulled.
The source's narrow except-tuple paraphrase comments are trimmed
to keep only the durable "don't swallow programmer errors" note,
per CLAUDE.md s4a.
Tests: 5 bim-lane tests in test/bim/module/project/
test_pending_opening_cuts.py covering apply happy-path + missing
entity, dismiss, select happy-path + cancellation.
Generated with the assistance of an AI coding tool.
MEPAddBend exists on the main flow but commits bend geometry with
hardcoded defaults (start_length=0.1, end_length=0.1, radius=0.2)
with no opportunity to tune before commit. The new scene-level
BIMBendPreviewProperties hosts a draft (start_segment_id,
end_segment_id, start_length, end_length, radius); EnableBendPreview
populates it from the two selected MEP segments after asserting they
are non-parallel, FinishBendPreview dispatches MEPAddBend with the
tuned values and clears the draft, CancelBendPreview discards it.
Scene-level placement follows CLAUDE.md 2.9: a bend creates a new
fitting entity between two segments, so neither segment alone owns
the draft. Foundation for the upcoming bend preview gizmo group and
decorator.
Generated with the assistance of an AI coding tool.
When an operator mutated IFC then raised mid-execute the user was left
staring at a raw traceback with the IFC graph captured by the active
transaction but the Blender side stale. Blender does not push an undo
step for a raised operator (the same gap that the CANCELLED-modal arm
patches via bpy.ops.ed.undo_push), so the WARNING the framework can
emit is only honest if it pushes that undo step too. The framework
now detects partial state via ifc_file.transaction.operations,
pushes a Recover undo step, then reports a WARNING naming Ctrl+Z so
the recovery path is discoverable. The bespoke try/except wrapper in
UnjoinWallPathConnection becomes redundant and is retired in the
same change.
Generated with the assistance of an AI coding tool.
Three files flagged by black --check on the lint-formatting job:
* bim/module/geometry/operator.py — single-arg `.update(...)` rejoined
onto one line under the 120-char budget.
* test/bim/module/model/test_wall_gizmos.py — same join on a
_make_path_rel call.
* test/modal/test_modal.py — pre-existing baseline noise picked up
via the upstream merge: PEP-8 blank-line separators between top-
level functions, `0.68+` → `0.68 +`, double quotes, trailing
whitespace stripped.
No behavioural change; pure whitespace.
Generated with the assistance of an AI coding tool.
The toggle_openings icon lived outside the IconSlot layout — each
host (wall, roof) declared an ad-hoc setup_pen_row_toggle_openings_icon
+ update_pen_row_toggle_openings_icon pair, and GizmoArrayEdition
queried a hardcoded _FEATURE_IDLE_MAX_X dict to position past it.
On an arrayed wall the dict was shadowed: find_for_element returns
"array" before "wall" in EDIT_TYPES order, the wall reservation was
never consulted, and the first per-layer ARRAY icon (local X=0.37)
landed 13cm from the wall's toggle_openings (X=0.50) — visually on
top of each other.
Promote idle-row icons into the slot system instead of patching the
dict:
* IconSlot gains an Optional visible_when predicate for state-driven
visibility (toggle_openings only when the host carries openings).
* BaseParametricGizmoGroup gains idle_slots: ClassVar[tuple[IconSlot]]
+ _idle_slot_x_positions() + _idle_row_right_edge() helpers; the
setup + idle-branch positioning loops mirror the existing
feature_slots path.
* Wall and roof declare toggle_openings as an idle_slot and drop
their ad-hoc setup/update calls.
* GizmoArrayEdition's _resolve_feature_idle_max_x walks
BaseParametricGizmoGroup.REGISTRY and takes the max
_idle_row_right_edge() across peers whose poll passes — no more
hardcoded dict, no more find_for_element-order shadowing.
* setup_pen_row_toggle_openings_icon + update_pen_row_toggle_openings_icon
helpers deleted from drawing/gizmos.py.
* 3 forward-compat AST guards pin the new contract.
Also bundles an unrelated array-test fix: TestUsingArrays in
test/tool/test_model.py was asserting against bpy.context.selected_objects
which is a fragile signal after remove_array / apply_array. A new
_array_objects() helper filters bpy.data.objects via the BIM_Array
pset's IfcActuator type instead.
Layout on an arrayed wall after the fix:
pen X = 0.00
toggle X = 0.50 (idle_slot 0)
array[0] X = 0.87 (one ICON_ARRAY_GAP past idle row)
array[1] X = 1.27
All separated by the standard inter-icon spacing.
Generated with the assistance of an AI coding tool.
1) Async-wait spin loops in initWgpu / probeAndCreatePool / pick /
screenshot finalize all called wgpuInstanceProcessEvents in a tight
while. wgpu-native drives queued callbacks from there; Dawn-web
queues the callback for an event-loop tick that never happens
because wasm doesn't yield back to JS. Page hung at the first
await (RequestAdapter) and Firefox flagged the tab as slow.
New `waitTickInstance` helper calls emscripten_sleep(0) on
Emscripten (Asyncify unwinds wasm, JS resolves WebGPU promises,
resume) before wgpuInstanceProcessEvents drains completions into
our callback. Desktop path is unchanged. All five
`while (!done) ProcessEvents` sites switch to the helper.
2) streaming_thread_.start() inside initWgpu spawns a std::thread.
On Emscripten that needs -pthread + COOP/COEP headers from the
hosting page. None of that is wired yet, so the start is gated
`#if !defined(__EMSCRIPTEN__)`. The sync chunk-load fallback
already inside driveStreamingLoads carries the load until #88
replaces it with emscripten_fetch.
3) wgpuSurfacePresent at the end of render() aborted with
"wgpuSurfacePresent is unsupported (use requestAnimationFrame via
html5.h instead)" — Dawn-web composites the canvas at the end of
the RAF tick automatically. Call skipped on Emscripten;
WebViewportHost::requestFrame is the RAF-tick driver.
Page now loads, brings up wgpu, configures the surface, and renders
one frame with the configured background. The frame log in the
status overlay shows the first-frame startup spike (~1000 fps from
a 1 ms tick); subsequent frames are paint-on-demand, which on the
empty scene means no frames at all — consistent with the desktop
event-driven model.
Replaces the standalone wgpu-only clear-color spike in main_web.cpp
with a real WebViewportHost implementation: surface creation via the
emdawnwebgpu canvas-selector source, framebufferSize through
emscripten_get_element_css_size + dpr, requestFrame as a deferred
flag the RAF main_loop consumes, quit through emscripten_force_exit.
main_web.cpp now does the same lifecycle the desktop initWgpu shell
does: core_.initWgpu(web_limits=true) → buildPipelines → buildHiz/
Edge/Pick. The render loop runs core_.render() once per RAF tick when
the host has flagged a frame pending, with a surface reconfigure on
size changes.
Builds clean under emcc 6.0 + emdawnwebgpu (1.4 MB wasm, 278 KB JS
glue). Renders an empty scene with the configured background — the
plumbing is end-to-end through the same ViewportCore code path the
desktop build uses. No sidecar load yet: that lands with the
emscripten_fetch streaming backend (#88).
addSectionPlaneAtSurface (camera-facing auto-flip + kMaxSectionPlanes
cap check), removeSectionPlane, and clearSectionPlanes all move to
ViewportCore. ViewportWindow keeps tiny forwarders so the section
tool's input handlers (still VW) call through without seeing the move.
Each method now calls host_->requestFrame() in place of the
isExposed() + requestUpdate() gate, which means the section tool path
becomes the next piece that could exercise the WebViewportHost: a
click-to-add over WebGPU will work as soon as the host is wired,
without further core-side changes.
The frame loop — surface acquisition, parallel cull dispatch, streaming
drive, two-pass main render, HiZ resolve, edge pass, screenshot
capture, FrameStats emission, interactive / bench heartbeat, bench
summary + auto-quit — all live in ViewportCore now.
ViewportWindow::render() shrinks to the Qt-only prelude:
isExposed() guard, fpsIntegrate() (fly-mode WASD step), then
core_.render().
The overlay renderer stays Qt-bound (OverlayRenderer.h carries
QString labels). Core reaches it via two new ViewportHost virtuals:
encodeOverlaysInMainPass (section gizmos, highlights, pivot, lines,
points — in-MSAA-pass) and encodeOverlaysPostMain (corner axis,
marquee, labels — on the resolved surface). The QtViewportHost
implementation in ViewportWindow forwards each to overlays_.X().
FrameStats moves to its own Qt-free header (FrameStats.h) with
ViewportWindow::FrameStats re-exported as a using-alias so the
bonsai-side signal binding keeps working. ViewportHost::onFrameStats
replaces the placeholder 4-double signature with the typed POD.
OverlayFrame moves alongside (OverlayFrame.h) so the host overlay
callbacks can carry it without dragging Qt into core.
Bench + frame-stats + cull-tuning state (min_pixel_radius_,
motion_min_pixel_radius_, lod1_pixel_threshold_, cull_threads_enabled_,
prev_camera_*, has_prev_camera_, last_cull_was_motion_, last_visible_*,
last_cull_*ms_, last_stream_ms_, bench_*, interactive_frame_count_,
frame_time_ms_window_/_sum_/_count_/_head_) move to ViewportCore; VW
keeps reference aliases so the env-var prelude, setBenchmarkFrames,
and the various tool keybind setters keep compiling unchanged.
Smoke checks: --screenshot path renders + saves a clean PNG;
--benchmark 10 runs the warm gate, prints the per-frame log + summary,
and exits cleanly via host_->quit().
The inline screenshot path inside render() — surface-to-buffer copy,
async map, BGRA->RGBA swap into a tightly-packed RGBA8 image — moves
into core_.encodeScreenshotCapture / finalizeScreenshotCapture. render()
calls the two new helpers around its existing queueSubmit.
The PNG write itself stays Qt-bound, but it's now reached through a
new ViewportHost::saveScreenshotRgba8 virtual. The QtViewportHost
override (ViewportWindow::saveScreenshotRgba8) constructs a QImage
around the host-buffer and calls QImage::save("PNG") with the same
log lines as before; a future WebViewportHost will route the bytes
through stb_image_write or a download-URL emit instead. Either way,
the wgpu-side capture path never touches Qt again.
quit-after-screenshot now goes through host_->quit() too, so the
--screenshot CLI exit no longer reaches QCoreApplication::quit()
from render() directly.
Two small helpers + the screenshot-quit flag flip across; the render-
side capture encode + readback + PNG save itself stays in VW (those
need a stbi-style PNG writer to replace QImage::save before they can
move, and that's its own commit).
VW keeps tiny forwarders so bonsai's SceneLoader + the CLI screenshot
path don't see the move. The streaming sync-fallback gate already
reads the core-side pending_screenshot_path_, so capture timing is
unchanged.
The swapchain configuration path — WGPU_PRESENT_MODE handling,
mode-preference order (Mailbox / Immediate / FifoRelaxed / Fifo),
caps probe, cfg.alphaMode wiring, and the on-resize depth + MSAA +
HiZ texture reallocation + bind-group invalidation — all move to
ViewportCore. The only VW-side concern was a QString::fromLatin1 in
the advertised-modes log, which becomes a std::string concat with the
same output.
ViewportWindow's render() / resizeEvent / surface-outdated retry paths
call core_.configureSurface() now. createSurface() (which returns the
platform-specific WGPUSurface from the host) stays VW because Qt
platform discovery has to happen on the Qt side.
The whole pick pipeline (R32UInt + RGBA16F MRT, depth attachment,
ping-pong staging, single-pixel + rect readback) plus the public
pickObjectAt / pickSurfaceAt / picksInRect / pickMeshLocalAt / raycast
API and the rayAabbSlab / rayTriMT / rayAABBHit helpers all move to
ViewportCore. ViewportWindow keeps tiny forwarder methods so the
bonsai input + tool callers (mouseRelease, marquee, section tool,
Length/Area refinement) stay compiling.
MeshLocalPick + RaycastHit follow as nested types on ViewportCore;
ViewportWindow re-exports them as using-aliases to preserve the
ViewportWindow::MeshLocalPick / ViewportWindow::RaycastHit names
existing callers (and a couple of bonsai tests) reach for.
State migrated: pick_color_texture_/_view_, pick_normal_texture_/_view_,
pick_depth_texture_/_view_, pick_staging_buffer_, pick_normal_staging_buffer_,
pick_w_/_h_, box_pick_staging_buffer_/_capacity_. The pick_pipeline_
itself was already aliased.
The pick path no longer reaches into VW for any GPU state, so the
render() / shutdown() callers become core_.X() forwards and the pick
infrastructure can be exercised by the future web build without going
through Qt.
buildEdgePipeline, encodeEdgePass, releaseEdgeResources + the EDGE_WGSL
shader source all move to ViewportCore. The edge_bind_group_ + the
edges_enabled_ flag come along too (the latter aliased on VW so the
edge-toggle keybind keeps compiling).
The pass binds the now-core-side depth_view_ directly, so there's no
remaining cross-side state dependency for edge rendering. render()
still calls core_.encodeEdgePass(enc, surface_view) — once render()
itself moves, the call collapses to a sibling method invocation.
The whole HiZ occlusion-cull pipeline (resolve pass, ping-pong async
readback, CPU mip pyramid, per-instance AABB lookup, WGPU_HIZ_TRACE
diagnostic) moves to ViewportCore. The main render-pass depth
attachment and MSAA color attachment come along too — they're shared
between render() (still VW) and the HiZ resolve pass (now core).
Methods migrated: buildHizPipeline, ensureHizTextures,
releaseHizResources, encodeHizResolve, startHizMap, drainHizReadbacks,
aabbOccludedByHiz, ensureDepthTexture, releaseDepthTexture,
ensureMsaaColorTexture, releaseMsaaColorTexture. HIZ_WGSL moves with
them into ViewportCore.cpp's anon namespace.
State migrated: hiz_enabled_, hiz_valid_, hiz_vp_, hiz_pyramid_,
hiz_mip_offset_/_w_/_h_, hiz_reject_count_, hiz_trace_budget_,
hiz_uniform_buffer_, hiz_bind_group_, hiz_resolve_texture_/_view_/_w_/_h_,
hiz_padded_bpr_, hiz_staging_buffers_[2], hiz_slot_vp_[2],
hiz_slot_state_[2], hiz_write_idx_, depth_texture_/_view_/_w_/_h_,
msaa_color_texture_/_view_/_w_/_h_, plus the HizSlotState enum +
HIZ_SLOTS + HIZ_BASE_W constants. ViewportWindow keeps reference
aliases on every field VW.cpp still touches so the render path
compiles unchanged.
The HizOccludedFn shim in render() now wraps core_.aabbOccludedByHiz
directly. Once the render path itself moves into core, that shim
disappears and cull can call aabbOccludedByHiz as a sibling method.
applyCachedModel, uploadMeshChunk, uploadInstanceChunk, finalizeModel
all live in ViewportCore now. The bonsai-facing public entry points on
ViewportWindow are one-line forwarders that keep
SceneLoader → ViewportWindow* binding intact.
State + helpers that came along:
- pending_direct_loads_ (the SidecarData staging map keyed by model_id)
- initial_view_applied_ (auto-viewAll suppression; aliased on VW so
setCamera can still flip it)
- getOrCreateDirectStaging + createBufferWithData (anon namespace
helpers on the core side)
The Qt-bound isExposed() / requestUpdate() pair on the
applyCachedModel tail becomes host_->requestFrame() — the
QtViewportHost forwards to requestUpdate(); a WebViewportHost will
forward to requestAnimationFrame.
The sidecar load path is now fully core-side. ViewportWindow no
longer owns any of the model-creation machinery; everything from
"here's a parsed sidecar" to "fully-built models_gpu_ entry with
empty pool slices waiting on streaming" runs through ViewportCore.
CPU cull (frustum + contribution + LOD + opaque/transparent partition)
and its companion GPU-upload step now live in ViewportCore. The HiZ
occlusion test stays in VW — the pyramid + async readback machinery
hasn't migrated yet — and is plumbed through a
ViewportCore::HizOccludedFn callback the render path binds when HiZ
is enabled-and-fresh. Null callback means "no occlusion test", which
keeps the cull path host-agnostic.
extractFrustumPlanes + aabbInFrustum moved up into CameraMath.h so
both VW's render() (where the planes are extracted) and core's cull
(where they're tested) can share without one #including the other.
LOD-debug counters (lod1_dbg_count_, lod0_dbg_eligible_count_,
lod0_dbg_no_lod1_count_, lod1_dbg_tris_saved_) moved to core too —
they're written by cull and read/reset by VW's still-here per-frame
[frame] heartbeat through reference aliases.
The per-frame streaming residency driver — LRU/priority eviction,
worker-result drain, candidate selection, click-and-track diagnostic,
sync-fallback for screenshot capture — now lives in ViewportCore.
ViewportWindow::driveStreamingLoads is a one-line forwarder.
Streaming-related state moves to core with reference aliases on VW:
streaming_{loads,more_pending,candidates,evictions_{lru,pri},drained,
blocked_oom}_this_frame_, streaming_debug_, tracked_{object_id,
chunk_mid,chunk_idx,was_resident}_, and pending_screenshot_path_. The
pick handler and bench-warm gate (still VW) read/write through the
aliases unchanged.
Qt-isms in the body were replaced en route:
- QFileInfo(...).completeBaseName() → std::filesystem::path::stem()
- requestUpdate() → host_->requestFrame()
- QString::number(x, 'f', N) in numeric logs → raw double / int (we lose
fixed-precision in a couple of diag lines; acceptable tradeoff).
host_->requestFrame() means the streaming loop is now host-agnostic:
the WebViewportHost will provide its own requestAnimationFrame
equivalent when it lands.
The chunk-state machine that mediates between the streaming pool and the
per-chunk WGPU bind groups now lives in ViewportCore. ViewportWindow's
remaining streaming code (driveStreamingLoads, finalizeModel) calls
through to core_.applyStreamedChunk / core_.unloadChunk /
core_.loadChunkBytesAndUploadGpu, and the chunk request builder is a
static helper on ViewportCore so VW's still-here driveStreamingLoads can
enqueue requests against streaming_thread_ without reimplementing it.
streaming_frame_idx_ moved to core (alongside the residency clock),
aliased on VW so the inline streaming logic stays compiling. The
mesh-volume side effect inside applyStreamedChunk now fires a
std::function<void()> callback (core_.on_volume_dirty_) instead of
reaching into ViewportWindow::updateVolumeReadout — VW wires the
callback in its ctor, non-Qt hosts leave it null and pay nothing.
computeMeshLocalVolumeQuantised moved to ViewportCore.cpp's anonymous
namespace; it was only called by applyStreamedChunk.
In plan view world-Z collapses to zero on screen, so every wall-edit
icon anchored on the floor — the projected 3D cursor, wall endpoints,
wall-to-wall corners, IfcRelConnectsPathElements connection points —
projects onto the click target it represents. The result on a typical
extend / split / unjoin action: the icon sits on top of the cursor
crosshair (or the corner the user wants to click), defeating precise
positioning.
Add shared ``gizmo.top_down_clearance(context, billboard_rot)`` to
bim/module/drawing/gizmos.py: returns a screen-up Vector in top-down
view (cosine cone around world Z, matching ``is_view_top_down``) and a
zero Vector elsewhere, so call sites apply it unconditionally before
``billboarded_at``. Default distance 0.4 m aligns with the inter-icon
stack spacing already used by GizmoWallJoinIntersection so single
icons and stack bases land at consistent screen-up positions when
multiple groups render around the same wall endpoint.
Apply at the seven wall-edit anchor sites:
* GizmoWallEdition cursor stack (top-down branch only — non-top-down
already stacks along world-Z at structural points clear of the
cursor).
* GizmoWallExtendVertically (single icon at wall origin endpoint,
active-object Z elevation).
* GizmoWallJoinIntersection corner stack base + merge midpoint.
* GizmoWallUnjoinSingle link-toggle pool (one icon per IFC path
connection, previously sitting exactly on the connection point).
* GizmoWallFilletReedit pen icon at fillet corner.
* GizmoWallFilletToggleOpenings.
The clearance is a pure visual offset — bound operators still read
the world-space anchor (cursor / endpoint / connection point) at
execute time, so the action's target is unaffected.
Also tighten GizmoWallUnjoinSingle: gate poll on ``props.is_editing``
so the link-toggle icons only surface during the wall edit lifecycle
(matching every other edit-row icon), and downsize them via a new
``ICON_SCALE = 0.35`` constant since 16 of them at default scale
cluttered the viewport on path-heavy walls.
ruff + black clean. Wall gizmos test lane 14/14 pass.
Generated with the assistance of an AI coding tool.
Some distributions (e.g. Fedora) ship only a shared RocksDB that exports
RocksDB::rocksdb-shared rather than RocksDB::rocksdb. The CMake target
selection now falls back to the shared target when the static one is absent.
Newer RocksDB also changed DB::Open and DB::OpenForReadOnly to take
std::unique_ptr<DB>* instead of DB**. IfcFile.cpp uses SFINAE tag dispatch
to build against both old and new APIs without version detection.
CGAL 6.x deleted operator< from Point_d, so std::map<Point_d, ...>
no longer compiles. Adds a custom lexicographic comparator and updates
the three affected maps in snap_halfspaces and snap_halfspaces_2.
bim/handler.py was importing two feature-module internals
(wall_offset_gizmos.clear_caches, preview_base.discard_pending_previews)
to drain load-transient parametric state alongside the existing
tool.Parametric.heal_stale_edit_flags() call inside
_apply_save_file_invariants. Each new parametric drain added one
top-level import and one inline call — every load_post drain leaked
into handler.py's namespace.
Hide all three drains behind tool.Parametric.on_load_post(scene),
sited adjacent to heal_stale_edit_flags. The two feature-module
imports become late imports inside on_load_post — same pattern as
refresh_post_commit's existing `import bonsai.bim.handler` — which
sidesteps the tool.parametric -> bim.module.model.preview_base ->
bonsai.tool registration-time cycle.
The forward-compat AST contract that pinned "every module-scope
GenerationKeyedCache + clear_caches MUST be drained on load_post"
follows the call site to its new home — the test now walks
tool.Parametric.on_load_post instead of _apply_save_file_invariants.
No behaviour change. 45/45 affected bim tests pass
(test_handler_forward_compat, test_preview_base,
test_wall_offset_gizmos, test_parametric_registry).
ruff + black clean on all touched files.
Generated with the assistance of an AI coding tool.
Per-frame uniform packing now lives in ViewportCore::updateFrameUniforms,
which reads the camera (via the already-migrated buildViewProj), the
section_planes_ vector, and the xray_alpha_cap_ scalar — all of which
have moved into ViewportCore alongside frame_uniform_buffer_.
ViewportWindow keeps reference-aliases on section_planes_ and
xray_alpha_cap_ so the section-tool and X-ray toggle (still Qt-input-
bound, still living in VW) keep compiling unchanged. The render-path
caller in VW::render now does core_.updateFrameUniforms().
Extracted SectionPlane into its own Qt-free header (SectionPlane.h)
so ViewportCore doesn't have to include OverlayRenderer.h's QString /
QHash. OverlayRenderer.h re-exports it.
Three feature-specific decorators previously lived in
bim/module/model/decorator.py despite owning state only their
home module reads:
* ArrayPreviewDecorator + ArraySelectionHighlightDecorator +
draw_array_layer_children_bbox -> array.py (read array
edit-state props and walk BBIM_Array psets)
* WallGizmoPreviewDecorator + draw_wall_partner_bbox -> wall.py
(dereference wall.py-private classes and helpers via lazy
imports)
decorator.py keeps cross-cutting infrastructure
(BoundingBoxDecorator, SlabDirectionDecorator, WallAxisDecorator,
WallFilletPreviewDecorator, PolylineDecorator, ProductDecorator)
and the shared bbox primitives (bbox_world_edges,
draw_polyline_segments, _BBOX_EDGES, _stroke_lines_alpha,
_fill_quads_alpha) that several feature files now import.
handler.py and gizmos.py update their import paths; the
wall-feature lazy imports inside WallGizmoPreviewDecorator
methods collapse to direct references now that the decorator
lives in wall.py.
No behaviour change. Wall lane 37/37, array lane 15/15, wall
forward-compat 6/6, parametric-registry 8/8 still pass.
Generated with the assistance of an AI coding tool.
Three concerns sharing the same architectural theme (collapse inline
bbox / edit-state lookups, drop overrides that re-do base-class work):
== Bbox helpers and array operator DRY ==
* tool/blender.py: add a "dimensions" tuple key to both
get_object_bounding_box and get_object_world_bounding_box return
dicts. The (max - min) per-axis extent — which callers previously
computed via local helpers — is now a key alongside min_x / max_x
/ min_point / max_point / center. Distinct from Blender's built-in
obj.dimensions (which folds object-level scale): the local variant
is the intrinsic mesh bbox extent; the world variant is the
matrix_world-applied AABB.
* bim/module/model/array.py: drop the local _bbox_dims helper; the
two callers now read tool.Blender.get_object_bounding_box["dimensions"]
directly.
* Rename _parent_geometry_changed -> _array_children_need_rebuild.
The old name suggested "did the parent change just now", implying
the function was a parent-edit-finish trigger. It actually runs
only inside the array-edit-finish path as a drift safety net (the
upstream-deliberate design — see commit 83d97d7e9 "Fix #7616. Make
regenerate array an operator instead of an array preference" —
means the array doesn't auto-regen when its parent geometry edits
finish). New name matches the call-site phrasing
``if X: _wipe_array_children(layers)`` and clarifies that this is
a children-state check, not a parent-edit trigger.
* Extract _resolve_array_edit_props(context) — returns the active
object's array props during an active edit lifecycle, or None.
Collapses the obj-active-then-is-editing prologue (3 lines + return)
to one resolver call across 4 sites: ToggleArrayMethod.execute,
AdjustArrayCount.execute, RemoveArrayLayerFromEdit._execute and
.poll. Each call site shrinks from 7 lines to 3.
* Migrate two inline bbox reads inside GizmoArrayEdition to the new
dict keys: get_axis_world_face_center collapses the manual
xs/ys/zs min/max + center math to bbox["center"] + bbox["max_x"] /
["max_y"] / ["max_z"]; get_element_height collapses
``max(corner[2] for corner in obj.bound_box)`` to
tool.Blender.get_object_bounding_box(obj)["max_z"].
The _BBOX_EQUALITY_EPS = 1e-5 tolerance stays inline as a single-
consumer constant — no other call site needs tolerance-equality on
dimension tuples, so extracting it to a shared util would be
speculative abstraction.
== Drop dead code ==
* GizmoArrayEdition.update_editing_gizmos override + its
_has_other_parametric_type helper: redundant with
hide_pen_button = True at line 1024. The base class already hides
the pen in every idle case (when hide_pen_button is truthy) AND in
every editing case (unconditionally). The override's conditional
hide-when-parametric only re-hid a pen that was already hidden in
both branches. Removes the only remaining path that could re-show
the array's pen icon; array-edit entry is now uniformly via the
per-layer ARRAY icons (which is the documented preferred
affordance, see the hide_pen_button comment).
* _wall_fillet_preview_active in wall.py: defined but never called.
_wall_fillet_props (the sibling thin-wrapper around
preview_base.get_preview_props) is heavily used; the
is_preview_active wrapper was added speculatively and never picked
up a consumer.
Generated with the assistance of an AI coding tool.
Bundled bug fixes + the forward-compat AST guard that prevents the
underlying class of bug from coming back.
* bim/module/model/wall.py: FinishEditingWall._execute early-returns
CANCELLED when props.is_editing is False. Without this guard, a
failed enable (e.g. on a wall without IfcMaterialLayerSetUsage)
leaves is_editing False but a press on finish still walked the
sub-ops below, which dereferenced layer-set-dependent state and
crashed.
* tool/model.py: Model.offset_wall now guards against
ifcopenshell.util.element.get_material returning None before
calling .is_a("IfcMaterialLayerSetUsage"). Fixes the pre-existing
test/bim/module/model/test_wall_header_refresh.py crash that has
been the only failing test in the wall lane since this branch
started.
* bim/handler.py: _apply_save_file_invariants drains
wall_offset_gizmos.clear_caches() on load_post. The module-scope
GenerationKeyedCache instance survives the .blend reload; without
the drain the cache may serve entries whose bpy_struct references
point into the freed bpy.data of the previous file.
* test/bim/test_handler_forward_compat.py: AST-walk test that
enumerates every bim/module/model/*.py source declaring both a
module-scope GenerationKeyedCache assignment AND a top-level
clear_caches function, and asserts each module appears as a
<module>.clear_caches() call in _apply_save_file_invariants. Pins
the contract: any future module-scope geom cache that exposes
clear_caches must wire into the load_post drain.
* test/bim/feature/model.feature + test/bim/test_feature.py: wall
edit-lifecycle scenarios switch from "add cube + assign as
IfcWallType" to "load the demo construction library + add an
occurrence of the WAL100 wall type", so the parametric edit runs
against a real LAYER2 wall with IfcMaterialLayerSetUsage rather
than a vanilla-mesh promotion that lacks one. The demo-library
step also picks the schema-matching library file (IFC2X3 /
IFC4 / IFC4X3) so the appended types remain valid across schemas.
Door saved-height assertion updates from 2.5 → 2500 to reflect
that BBIM_Door pset stores project units (METRIC_MM in the
empty-project fixture).
Generated with the assistance of an AI coding tool.
The instance/adapter/device/queue/pool/surface-format wgpu lifecycle now
lives in ViewportCore — including the OOM-scoped pool size probe and
the worker-thread startup. ViewportWindow::initWgpu becomes a Qt shell
that handles env-var tuning + nav-button preset wiring, then delegates
to core_.initWgpu(); the VW-only pipeline builders (HiZ, edge, overlays,
pick) still run after. ViewportWindow::shutdown drops the VW-only
resources (depth, msaa, hiz, edge, overlays, pick) and lets
core_.shutdown() release the shared wgpu handles it now owns.
The wgpu-native log callback (wgpuSetLogCallback / WGPULogLevel) is
gated on !__EMSCRIPTEN__: it's not part of the W3C spec header, and
the emdawnwebgpu port doesn't ship wgpu.h — validation errors there
land in the browser console regardless.
Drive-by: update test_federation to compare HomeView::target as
Eigen::Vector3f (left stale by #79 when QVector3D was retired).
Fixes builds with newer GCC/libstdc++ that no longer provide <cstdint>,
<cstring>, <cfloat>, <memory>, <algorithm> etc. transitively. Also
disambiguates visit<> calls in taxonomy.h with the full namespace and
casts the character value in IfcCharacterDecoder to uint32_t to silence
ambiguous overload warnings.
When HDF5 is found via its CMake config file, the code previously hardcoded
the hdf5_cpp-static target. On distributions that ship only shared HDF5
(e.g. Fedora rawhide where the config file was added in a newer package),
this caused a link failure. Now checks for hdf5_cpp-static, hdf5_cpp-shared,
and hdf5::hdf5_cpp-shared in order, falling back to module-mode discovery.
Move the main render pipeline construction + the selection flags
buffer/bind group lifecycle. Both buildPipelines and the selection
flags methods produce/consume state ViewportCore already owns
(main_pipeline_, frame_bgl_, etc.) plus a handful of "frame
infrastructure" fields this commit also brings across.
State moved (7 fields):
WGPUBuffer frame_uniform_buffer_
WGPUBindGroup frame_bind_group_
WGPUBuffer selection_flags_buffer_
uint32_t selection_flags_capacity_
std::vector<u32> selection_flags_scratch_
SelectionState selection_
VisibilityState visibility_
Methods moved:
buildPipelines (~150 lines + 320-line MAIN_WGSL string)
ensureSelectionFlagsBuffer (~60 lines)
uploadSelectionFlagsIfDirty (~10 lines)
Plus the MAIN_WGSL constant + the svFromCStr helper into
ViewportCore.cpp's anonymous namespace. ViewportWindow.cpp keeps its
own svFromCStr copy (still used by 50+ label fields in the not-yet-
moved pipeline builders + render encoders).
Shared constants extracted to ViewportCore.h:
kMaxSectionPlanes (was OverlayRenderer::kMaxSectionPlanes — assert
in VW.cpp keeps them in sync)
kViewportSampleCount (was SAMPLE_COUNT in VW; VW keeps a static
constexpr alias for the existing callsites)
struct FrameUniforms (canonical layout for the per-frame UBO,
consumed by both core's buildPipelines and
VW's still-in-flight updateFrameUniforms)
Builds: desktop / bonsai / web all green. Tests 100/100.
Tiny followup to #84-i — move the const-lookup volume helpers used by
bonsai's measurement HUD:
double volumeOfObjects(const std::vector<uint32_t>&) const
vector<pair<uint32_t, double>> volumesPerObject(
const std::vector<uint32_t>&) const
The det3OfPlacement static helper moves with them into ViewportCore.cpp's
anonymous namespace (the original kept its mirror in
ViewportWindow.cpp; ViewportWindow's own internal callers are gone now
since these methods moved).
Pure read of models_gpu_ + mesh_local_volumes — all in core already.
Trivial transplant.
Builds: desktop / bonsai / web all green. Tests 100/100.
Move the cluster of camera-state mutators + per-object AABB helpers
now that the camera fields all live in ViewportCore. CameraState
struct is canonical in core; ViewportWindow keeps a `using` alias
so bonsai's HomeView round-trip (Commands.cpp setHome / restoreHome)
compiles unchanged.
Moved:
void viewAll()
void setCamera(...) — pitch + distance clamping included
void setStandardView(yaw, pitch) — bypasses clamp for ±90°
void toggleProjection()
std::string cameraString() const
CameraState cameraState() const
void frameAabb(mn, mx, padding)
bool computeObjectAabb(id, float[3], float[3]) const
bool computeObjectAabb(id, Eigen::Vector3f&, Eigen::Vector3f&) const
ViewportWindow keeps thin forwarders for the public ones (bonsai
calls them). setCamera additionally flips initial_view_applied_
on the VW side — the auto-viewAll suppression flag isn't in core
yet because the trigger for auto-viewAll lives in the still-in-VW
applyCachedModel path.
The isExposed()+requestUpdate() Qt pattern inside the moved bodies
becomes host_->requestFrame(); two viewAll/toggleProjection diagnostic
prints become fprintf since Log::info() doesn't reach into core.cpp
through the Qt logging surface.
Builds: desktop / bonsai / web all green. Tests 100/100.
Move the three camera-math methods that compute view/projection
matrices, scene bounds, and per-chunk screen footprint for the
streaming priority signal:
void buildViewProj(Eigen::Matrix4f&, Eigen::Matrix4f&) const
bool computeSceneAabb(float[3], float[3]) const
float chunkScreenAreaPx(const ModelGpuData::Chunk&,
const Eigen::Matrix4f&) const
Plus the orbitEye helper (anonymous namespace in ViewportCore.cpp;
the qDegreesToRadians dep got swapped for an inline M_PI/180 constant).
ViewportWindow.cpp's 9 internal callers (cull, streaming, pick,
render, debug) updated to use core_.buildViewProj() etc. The
buildViewProj forwarder stays out of ViewportWindow.h since no
external caller needs it — bonsai/minimal both go through
public API methods like viewAll which still wrap core_ access
on the VW side.
Builds: desktop / bonsai / web all green. Tests 100/100.
Move the camera/projection/clear-color fields that buildViewProj,
updateFrameUniforms, the cull screen-area projector, and the bonsai-
side cameraState/setCamera/viewAll surface depend on. Same alias
pattern; no method bodies move in this commit — the next one moves
the camera math methods now that all their state is in core.
State moved (12 fields):
int configured_w_, configured_h_
float camera_target_[3], camera_distance_
float camera_yaw_deg_, camera_pitch_deg_, camera_fov_y_deg_
float camera_near_, camera_far_
bool projection_ortho_
Eigen::Vector4f background_color_
ViewportWindow keeps reference aliases for each (including a proper
`float (&camera_target_)[3]` reference-to-array binding) so the
~150 call sites that touch camera state stay unchanged. Aliases
collapse when their owning methods migrate.
Builds: desktop / bonsai / web all green. Tests 100/100.
Move the eight scene-mutation methods that drive bonsai's load/unload
and georeference setters, plus the per-model GPU teardown helper.
All are mechanical transplants — no logic change — so behaviour stays
identical; only the owner has changed.
Methods moved (ViewportWindow public-API methods stay as forwarders
to keep the bonsai-side callers compiling):
removeModel / resetScene / hideModel / showModel
setFederatedFalseOrigin
setModelCoordinateOperation
setModelTransformation
recomposeAndUploadModel
State moved:
bool wgpu_initialized_ (storage → core_, alias kept in VW for
the initWgpu call site that still flips
it; goes when initWgpu moves)
Free function moved:
releaseWgpuModelGpuData(ModelGpuData&, BufferPool&) → ViewportCore.cpp
(must live in IfcViewerCore now that ViewportCore.cpp's
removeModel / resetScene call it; ViewportWindow.cpp's remaining
two call sites continue to resolve through ModelGpuData.h's
declaration — same linker view, different definition TU)
The `if (isExposed()) requestUpdate()` Qt pattern inside the moved
bodies became `host_->requestFrame()` since ViewportCore can't see
QWindow; the desktop ViewportHost override at the bottom of
ViewportWindow.cpp continues to translate that into requestUpdate().
Builds: desktop / bonsai / web all green. Tests 100/100.
Move two pure-read methods (no GPU touch, no Qt) that the bonsai
measurement / federation-origin paths use:
bool findInstance(uint32_t, InstanceLookup&) const
bool firstGeometryPointWorldM(uint32_t, Vector3d&) const
ViewportWindow keeps both public-API method names — they now forward
to core_ for the implementation so existing callers in
bonsaiviewer/Measurement.cpp + Federation hooks don't have to change.
The InstanceLookup type also stays a `using` alias in ViewportWindow
(was added in #74).
Both methods were already de-Qt'd (`findInstance` delegates to
InstanceCompose; `firstGeometryPointWorldM` is pure Eigen). The move
is a straight transplant — no behaviour change.
Builds: desktop / bonsai / web all green. Tests 100/100.
First method-body migration. composeInstanceFromPlacement composes the
federated-false-origin × model-transformation × coordinate-operation ×
placement chain and re-derives the world AABB; it's a small,
self-contained method that only reads scene state and one matrix.
Moved:
Eigen::Matrix4d federated_false_origin_meters_ (storage → core_)
void composeInstanceFromPlacement(InstanceCpu&, ...) (body → core_)
ViewportWindow keeps:
- alias reference to federated_false_origin_meters_ (existing
setFederatedFalseOrigin call site still writes through it)
- no method declaration — internal callers route through core_
Internal caller (recomposeAndUploadModel) now invokes
core_.composeInstanceFromPlacement; once recomposeAndUploadModel
itself moves into ViewportCore the call shortens back.
Pattern for the rest of #84: state moves, then method body moves,
then internal callers update. Each commit leaves desktop / bonsai /
web green and tests 100/100. This is one of many such steps.
Move the five scene-state fields that drive per-model GPU upload + the
streaming residency loop into ViewportCore:
BufferPool pool_ — vertex+index sub-allocator
StreamingThread streaming_thread_ — background chunk reader
std::unordered_map<uint32_t, ModelGpuData> models_gpu_
— per-model state
uint32_t next_model_id_ — model-id allocator
uint32_t next_object_id_ — globally-unique object-id allocator
ViewportCore.h gains transitive includes for BufferPool / StreamingThread
/ ModelGpuData; ViewportWindow keeps the same names as reference aliases
so existing method bodies that touch them don't have to change.
Same risk profile as #84-a and #84-b: the storage moved but the values
are still set and consumed by the same code paths, so behaviour stays
identical.
Builds: desktop / bonsai / web all green. Tests 100/100.
Move the 15 pipeline + bind-group-layout + shader-module handles
that buildPipelines / buildEdgePipeline / buildPickPipeline write to.
Same pattern as #84-a: storage lives in ViewportCore, ViewportWindow
keeps reference aliases so existing builder-method bodies don't
have to acquire a `core_.` prefix at every touch point.
Moved fields:
Main render group:
main_shader_module_, frame_bgl_ (group 0), model_bgl_ (group 1),
pipeline_layout_, main_pipeline_, main_pipeline_transparent_
HiZ occlusion-cull group:
hiz_shader_module_, hiz_bgl_, hiz_pipeline_layout_, hiz_pipeline_
Edge silhouette group:
edge_shader_module_, edge_bgl_, edge_pipeline_layout_, edge_pipeline_
Pick pass:
pick_pipeline_ (reuses pipeline_layout_ — same set of bindings)
ViewportWindow's constructor binds 16 new alias references after
the 7 lifecycle ones from #84-a; member-init order matches
declaration order so core_ is constructed before any alias binds.
Builds: desktop / bonsai / web all green. Tests 100/100.
First chunk of the #84 ViewportCore extraction. The seven wgpu lifecycle
handles (instance, adapter, device, queue, surface, surface_format,
surface_configured) now live as ViewportCore members; ViewportWindow
keeps reference aliases pointing at ViewportCore's storage so its
existing render-method bodies don't need a `core_.` prefix added at
every call site — 230+ touches deferred until each method moves
across.
Member init order in ViewportWindow's constructor:
core_(this) → constructs ViewportCore with host_=this
instance_(core_.instance_) → binds the alias to core_'s field
… same for adapter/device/queue/surface/…
Friend declaration on ViewportCore::ViewportWindow lets the references
bind to its private fields. The friend bond shrinks each commit as
render methods (and their `device_` / `queue_` references) migrate into
ViewportCore proper; the goal state is no friend and no aliases.
Next #84 chunks (separate commits) move pipelines, models_gpu_, pool_,
streaming_thread_, then the render/cull/encode methods. Each leaves
the desktop build green.
Builds: desktop / bonsai / web all green. Tests 100/100.
Last round of straight-swap Qt value types in ViewportWindow + its
overlay co-pilot.
setBackgroundColor(const QColor&) → (float r, float g, float b, float a)
QColor background_color_ → Eigen::Vector4f (linear, 0..1)
QPoint {nav_,box_select_,fps_, } → Eigen::Vector2i
{section_drag_start_mouse_}
QSet<int> fps_keys_held_ → std::unordered_set<int>
QElapsedTimer fps_last_tick_, → Stopwatch (new header in
fly_render_clock_, IfcViewerCore — std::chrono-
render_thread_local_ backed, exposes the existing
timers in render() QElapsedTimer .start/.restart/
.elapsed/.nsecsElapsed surface)
Also propagates the QPoint → Eigen::Vector2i change through
OverlayRenderer::encodeMarquee since the marquee corner coords flow
through that interface.
API-level helpers:
toV2i(QPoint) — small inline in ViewportWindow.cpp, isolates
the QMouseEvent→Vector2i conversion at the
five mouse-event handlers
Stopwatch.h — new file, IfcViewerCore. Same call shape as
QElapsedTimer; backed by std::chrono::steady_clock.
QSet method swaps:
.isEmpty() → .empty()
.contains(k) → .count(k) (C++17, no std contains() until C++20)
.remove(k) → .erase(k)
Eigen::Vector2i doesn't have .manhattanLength(); the box-select drag
threshold uses std::abs(diff.x()) + std::abs(diff.y()) inline.
Bonsai side: View.cpp's setBackgroundColor wrapper now decomposes the
QColor into floats at the call site (kept locally so the bonsai UI
keeps its QColor-driven theming).
Closes#81 + the QElapsedTimer half of #83. QTimer
(pivot_indicator_hide_timer_) still uses Qt — it needs the host's
scheduleOnce mechanism that lands with #85.
Builds: desktop / bonsai / web all green. Tests 100/100.
Take QString out of ViewportWindow's outward-facing surface so it can
eventually move into a Qt-free ViewportCore:
void queueLoadSidecar(const QString&) → (const std::string&)
uint32_t loadSidecar(const QString&) → (const std::string&)
QString cameraString() const → std::string …
void captureNextFrameToPng(const QString&, bool)
→ (const std::string&, bool)
void setHudText(const QString&) → (const std::string&)
Internal members also moved off QString:
std::deque<QString> pending_sidecars_ → std::deque<std::string>
QString pending_screenshot_path_ → std::string
Implementation strategy: convert at the boundary where ViewportWindow
still leans on Qt internals — `loadSidecar` bridges to QString once
for QFile/QDir/QFileInfo path handling; the screenshot save path
constructs a QString locally for QImage::save; the OverlayRenderer's
HUD setter still takes QString so setHudText converts before calling
through. Each of those bridges goes away when ViewportCore lands and
OverlayRenderer / SceneLoader / SidecarBuilder get their own de-Qt
sweeps. cameraString now produces its CSV via snprintf — no QString
ever instantiated.
Bonsai-side updates (compile-only):
ifcviewer-minimal/main.cpp — queueLoadSidecar / captureNextFrameToPng
callers add .toStdString() on the QString
parser result
modules/viewport/View.cpp — setHudText callers add .toStdString() to
their `QString::arg(...)` formatter chains;
two `QString()` empty sentinels become
`std::string()`
Measurement.cpp — same pattern, two setHudText sites
ifcviewer/LengthMeasurement.cpp — same, three sites
The cameraString string-streaming fix-up in ViewportWindow.cpp drops
the temporary .toUtf8().constData() bridge from #82 — Log::Stream's
std::string overload now handles it directly.
Builds: desktop / bonsai / web all green. Tests 100/100. Closes#80.
Add Log.h (in IfcViewerCore) — a tiny stream-style logger that backs
fprintf(stderr,...), with overloads for the common primitives + char
strings. Mimics qInfo()/qWarning()'s syntax surface enough that
mass-replacing qInfo()→Log::info() and qWarning()→Log::warn() keeps
existing call sites parsing unchanged; .noquote() / .nospace() exist
as compat no-ops so chained qInfo().noquote()<<x<<y patterns survive.
QString streaming is a transitional concern — the QString → std::string
sweep (#80) hasn't landed yet, so ViewportWindow and friends still
construct QStrings for log payloads. LogQt.h (in IfcViewer, not Core)
adds the QString / QStringView operator<< overloads so those streaming
sites work without source changes during the in-flight Qt removal.
When #80 retires QString, LogQt.h drops out.
ViewportWindow.cpp: 132 qInfo/qWarning callsites converted. The two
printf-style qInfo("fmt %s", ...) callsites get fprintf with explicit
[info]/[warn] prefixes to keep the output discoverable.
Also de-Qt'd:
AreaMeasurement.cpp — 1 qInfo("fmt", …) → fprintf
SceneLoader.cpp — 4 qDebug + 1 qWarning printf-style → fprintf
GeometryStreamer.cpp — 2 qDebug printf-style → fprintf
ifcviewer-minimal/main.cpp — 2 qWarning << → Log::warn
Drops <QDebug> from each. Closes#82.
Builds: desktop / bonsai / web all green. Tests 100/100 pass.
Replace Qt math wrappers with Eigen across ViewportWindow, OverlayRenderer,
Federation, and the bonsai-side viewport modules. Eigen was already the
canonical type for the actually-important matrix work (InstanceCompose,
ModelGpuData, federation matrices); QVector3D/QVector4D/QMatrix4x4 were
leftover from when Qt was the path of least resistance. They offered
nothing over Eigen for our use case beyond a few graphics helpers
(lookAt / perspective / ortho) which were 30 lines to write.
Substitutions:
QMatrix4x4 → Eigen::Matrix4f
QVector2D → Eigen::Vector2f
QVector3D → Eigen::Vector3f
QVector4D → Eigen::Vector4f
API rewrites:
.lengthSquared() → .squaredNorm()
.length() → .norm()
.isNull() → .isZero()
.setToIdentity() → .setIdentity()
.constData() → .data()
.toVector3D() → .head<3>()
.inverted(&ok) → tryInvert4f(M, out)
Q::dotProduct(a,b) → a.dot(b)
Q::crossProduct(a,b) → a.cross(b)
QMat4x4(... row-major) → Eigen::Map<const Matrix4f>(col-major buf)
QMat4x4().lookAt(...) → lookAtRH(eye, target, up)
QMat4x4().perspective(.) → perspectiveYFovGL(fovy, aspect, n, f)
QMat4x4().ortho(...) → orthoGL(l, r, b, t, n, f)
Default-init divergence handled explicitly (QMatrix4x4() = identity,
QVector3D() = zero; Eigen leaves both uninitialized). Public API
(CameraState, HomeView, ViewportWindow::computeObjectAabb, the
addSectionPlaneAtSurface / pickSurfaceAt / raycast signatures) follows
through to Eigen too; bonsai-side View.cpp and Commands.cpp updated to
match.
Camera helpers (lookAtRH, perspectiveYFovGL, orthoGL, tryInvert4f)
extracted to a new CameraMath.h so OverlayRenderer's gizmo MVP and
ViewportWindow's buildViewProj share the same definitions. Federation
drops its <QVector3D> include in favour of <Eigen/Dense> (already had
the latter for the georef matrices).
Builds: desktop IfcViewerMinimal ✓, BonsaiViewer ✓, web IfcViewerWeb ✓.
Tests: 100/100 pass. Closes#78 + #79; opens the door for #80-#83.
Define the boundary the Path-A web-bring-up refactor will move things
across:
- ViewportHost.h is the embedder interface — surface creation,
framebuffer geometry, frame scheduling, quit, and notification
callbacks (onObjectPicked, onToolModeChanged, …). Desktop hosts
forward notifications to Q_SIGNALS; the future web host pushes
them to JS callbacks.
- ViewportCore.{h,cpp} is the platform-agnostic render-core target.
Empty today — the body fills in across the #78-#86 sequence as
each Qt subsystem (matrices, vectors, strings, timers, render
path, input) gets de-Qt'd and moved over.
- ViewportWindow now multiply-inherits ViewportHost alongside QWindow
and implements the host overrides as thin forwarders: createSurface
returns the cached surface_, requestFrame -> requestUpdate, quit ->
QCoreApplication::quit, onObjectPicked -> emit objectPicked.
Renamed the DPR accessor `dpr()` (vs `devicePixelRatio`) to avoid
the inherited-virtual clash with QWindow's qreal-returning version.
No method movement yet — this is purely the architectural scaffold so
subsequent commits have a destination.
First Emscripten target. main_web.cpp brings up a wgpu instance against
a <canvas id="viewer-canvas">, requests adapter+device asynchronously
via the standard webgpu.h callback chain, configures the surface, and
clears to the BonsaiViewer slate background on each RAF tick. No
sidecar load, no pipelines, no scene state yet — the goal is to end-
to-end verify the build + canvas + wgpu plumbing.
src/ifcviewer-web/ is a separate CMake root (not a subdir under the
desktop cmake/CMakeLists.txt) so the web build doesn't have to opt out
of Qt / OpenCASCADE / IfcGeom find_packages it can't satisfy. It adds
src/ifcviewer EXCLUDE_FROM_ALL and consumes only IfcViewerCore.
src/ifcviewer/CMakeLists.txt now gates the wgpu-native fetch + the
Qt-using IfcViewer target + install commands behind NOT EMSCRIPTEN.
The wgpu_native link target still resolves under Emscripten as an
INTERFACE library that activates --use-port=emdawnwebgpu (Dawn's
webgpu.h, replaces the legacy -sUSE_WEBGPU=1).
Build:
source path/to/emsdk_env.sh
emcmake cmake -S src/ifcviewer-web -B build-web -G Ninja
ninja -C build-web
python3 -m http.server --directory build-web 8080
# open http://localhost:8080/IfcViewerWeb.html in a WebGPU-capable
# browser (Chrome 113+, Edge 113+).
Phase B step 3 of #45.
bpy.ops.bim.new_project(preset='demo') crashed in
refresh_bim_tool_headers: the post-commit hook fired for every
nested bpy.ops.bim.append_library_element during template
loading, and the operator context Blender hands to
programmatically-invoked nested operators is stripped of the
view-layer attributes the refresh reads.
Two changes resolve it.
Gate the header refresh in tool.Parametric.refresh_post_commit
on operator.bl_idname being one of the EDIT_TYPES finish_op
idnames. Only validate-gizmo commits (bim.finish_editing_<name>)
now trigger the refresh; demo-loader and other non-edit
operators skip it. Querying the registry directly is the
canonical signal — string-prefix matching would silently drift
if ParametricObject.finish_op changes derivation.
Harden tool.Blender.get_active_object so its view_layer fallback
also uses getattr; the 150+ callers routed through it now
tolerate stripped contexts. _resolve_bim_tool_context applies
the same defensive pattern to mode / workspace.
Tests:
- test_handler_restricted_context covers get_active_object's
defensive path and the BimTool-family whitelist (excludes
annotation, spatial, structural).
- test_handler_forward_compat AST-pins that the gate consults
EDIT_TYPES (not a string prefix).
- test_wall_header_refresh rewritten — three tests cover the
gated-by-registry contract: counter bumps for every commit,
finish_op operators refresh headers, others don't.
Hotkey-driven in-place edits (S_E / C_E) no longer trigger the
refresh — they were caught by the pre-refactor "every commit"
design. Left out of scope; the new skip-non-finish test pins
this as intentional.
Generated with the assistance of an AI coding tool.
Remove the bonsaiviewer-autodesk Cargo examples that were used for local UI and dialog experiments.
Generated with the assistance of an AI coding tool.
Pull the Qt-free / OpenCASCADE-free files out of the IfcViewer target
into a new IfcViewerCore static lib: BufferPool, ChunkPlanner,
InstanceCompose, SidecarCache, StreamingLoader, StreamingThread,
LodBuilder, plus the header-only InstancedGeometry / ModelGpuData /
VertexQuantization / Selection / Visibility headers. IfcViewer PUBLIC-
links IfcViewerCore so existing consumers see no change.
This is the boundary the Emscripten web target will link against —
keeps Qt, IfcGeom, OpenCASCADE, CGAL, and Boost out of the wasm build.
Explicit file list, not glob, because the boundary is the whole point.
Replace qInfo() growth-event logging with fprintf(stderr,...) so
BufferPool.cpp has no Qt touchpoints. Lets the test target drop its
Qt6::Core link too. Prerequisite for the IfcViewerCore library boundary
the web target will link against.
The chunk planner (Morton sort + greedy pack) and instance composition
(federation × placement matrix chain + world-AABB derive) were inline
helpers in ViewportWindow.cpp. Pulled both out as free-function modules
so the math + lookup logic can be exercised without a Qt window or a
wgpu device. ViewportWindow now delegates; InstanceLookup is a using-
alias to InstanceCompose::InstanceLookup.
Also added an addSubBufferForTesting / clearSubPoolsForTesting seam to
BufferPool so the sub-allocator invariants can be pinned with fake
WGPUBuffer handles. The fakes are never dereferenced; the guard drops
the sub-pools before destructor would call wgpuBufferRelease.
Three new test binaries under src/ifcviewer/tests/, 33 cases / 173
assertions: BufferPool first-fit + alignment + coalescing + multi-
sub-pool isolation; ChunkPlanner Morton split / interleave / stable
sort / greedy-pack monotonicity and single-mesh-oversize; InstanceCompose
identity / translation / order-of-multiplication / large-placement
cancellation against federation false origin / column-major writeback /
findInstance lookup paths.
Two related cleanups bundled because each was too small on its own.
== Drop dead Geometry.has_material_styles duplicate ==
Two parallel has_material_styles implementations existed on HEAD:
* Geometry.has_material_styles (tool/geometry.py:853, added by
3483683cb "Add tool.Geometry helpers for body representation +
placement"): checks each material via tool.Material.get_style
for an IfcSurfaceStyle. This is the implementation gizmos-8088
uses — its core/root.py:58 calls geometry.has_material_styles.
* Root.has_material_styles (tool/root.py:75, added by e76455913
"Route _has_material_styles through tool.Root.has_material_styles"):
checks each material for a HasRepresentation inverse. Added to
fix the test/core/test_root.py::TestCopyClass::test_AAAAAAAAAAAA
failure by routing the check through a Prophecy-mockable seam.
HEAD's core/root.py:59 calls root.has_material_styles. The Geometry
version became orphaned by that migration — zero callers historically
(git log -S "Geometry.has_material_styles" returns nothing). The
Root placement is the right architectural home: has_material_styles
pairs with assign_body_styles in the copy_class flow as "is there
material-defined styling? if not, apply body styling" — both
decisions live on the same interface, called in sequence from the
same caller.
The semantic delta (HasRepresentation vs IfcSurfaceStyle) is a close
approximation in real IFC files where HasRepresentation almost always
indicates a styled material; if precision becomes necessary, the
Root impl can be tightened independently of this cleanup.
Drop the Geometry method + its abstract declaration in core/tool.py.
== Sanitation sweep per CLAUDE.md §4a ==
Eight rot-prone references in code we authored on this branch get
their first-draft mistakes cleaned up. The §4a rule (no sibling
symbol names, no test paths, no motivation history in docstrings)
got added during this branch, so older commits sometimes named their
siblings in prose; this is a focused cleanup of the worst offenders.
* bim/module/model/wall.py:201 — _CommitWallDraftsFirstMixin
docstring carried motivation history ("...that every multi-wall
operator … used to repeat at the top of _execute"). Rewrite to
describe only the current contract.
* bim/module/model/wall.py:1910 — cycle_type_operator comment named
two sibling methods. Rephrase to describe what happens at the slot.
* bim/module/model/wall.py:2025 — _active_instances ClassVar comment
named WallGizmoPreviewDecorator. Rephrase to "the wall-gizmo
preview decorator" (role, not class).
* bim/module/drawing/gizmos.py:3402 — GizmoFillet hit_uses_bbox
comment named GizmoWallJoinIntersection. Rephrase to "the wall-join
gizmo group".
* bim/module/drawing/gizmos.py:3887 — GizmoCountLabel docstring had
a :meth:`set_count` cross-reference. Drop — reader sees the method
next to the class.
* bim/module/model/host_add_opening_gizmo.py:201 — poll-exclusion
comment named GizmoWallEdition + GizmoRoofEdition. Rephrase to
describe why we skip ("walls and parametric roofs both render
their own toggle in the pen row").
* bim/module/void/operator.py:45 — preserve_placement comment named
FilledOpeningGenerator.generate. Rephrase to "the filling-opening
generator gates its snap-to-wall-axis block on this flag".
* bim/parametric_lifecycle.py:64 — module docstring named the test
file path (test/bim/test_parametric_registry.py). Rewrite to
"enforced by the registry contract tests".
Sweep otherwise clean: no third-party software names in this-branch-
authored comments (upstream Revit / Tekla / ArchiCAD references are
legitimate external-constraint workarounds, §4a-allowed). No
PR/issue numbers we authored except the FIXME(PR5) in
tool/parametric.py:150, deliberately preserved until PR6's MEP slice
resolves it.
Generated with the assistance of an AI coding tool.
Two long-stale docs that described the deleted OpenGL backend are
replaced with current-state rewrites under `src/bonsaiviewer/docs/`
and wired into the toctree. The originals are removed.
## env-vars.rst (replaces src/ifcviewer/settings.rst)
The orphan `src/ifcviewer/settings.rst` was written for the OpenGL
backend (`IFC_*` prefix, MDI-specific knobs) and was never wired into
any Sphinx toctree — it sat as a one-off file in the C++ source tree,
undiscoverable from a normal docs build.
* **Dead — dropped entirely.** `IFC_SKIP_MDI`, `IFC_MAX_SUBDRAWS`,
`IFC_SUBDRAW_DIAG` were GL-only `glMultiDrawElementsIndirect`
instrumentation. wgpu has no MDI. `IFC_FPS_HITCH_MS` no longer
exists in source.
* **Renamed.** `IFC_HIZ_MOTION` → `WGPU_HIZ_MOTION`,
`IFC_CULL_THREADS` → `WGPU_CULL_THREADS`.
* **New, previously undocumented.** Ten `WGPU_*` vars added during
the port + bring-up (WGPU_HIZ, WGPU_HIZ_TRACE, WGPU_MIN_PX,
WGPU_MIN_PX_MOTION, WGPU_FLY_DEBUG, WGPU_NAV_PRESET,
WGPU_PRESENT_MODE, WGPU_STREAM_DEBUG, WGPU_STREAM_DEEP_DEBUG,
WGPU_STREAM_EVICT_LOG). Descriptions written from each variable's
use-site so wording matches actual behaviour.
* **LOD-build section kept verbatim.** IFC_LOD_ERROR, IFC_LOD_RATIO,
IFC_LOD_MIN_SAVINGS, IFC_LOD_DEBUG — sidecar-bake knobs,
backend-agnostic.
* **GUI-promoted "old IFC_* graveyard" section dropped.** The file
is an env-var reference, not a record of historical spellings.
## viewport_architecture.rst (replaces src/ifcviewer/README.md)
The 994-line `src/ifcviewer/README.md` was an archive of the GL-era
phase-by-phase perf narrative. ~95% of it described deleted code:
OpenGL 4.5 Core, `glMultiDrawElementsIndirect`, VAO/VBO/EBO,
`GL_ARB_shader_draw_parameters`, BVH-per-model, sidecar v5/v7/v9
(current is v13), the now-non-existent `./IfcViewer` binary, Phase
3F "static batching next" plans superseded by the chunk-pool
architecture, Phase 3E "GPU compute culling removed" since re-added
as task #17 pending. Salvaging the ~50 lines of still-correct
content would have left a Frankenstein doc internally contradicting
itself.
Replaced with a focused architecture page covering current reality:
consumer split (BonsaiViewer shell vs IfcViewerMinimal standalone),
stack (wgpu-native v29, Qt6, IfcOpenShell, IfcUtil, Eigen3,
meshoptimizer), five core ideas (unique-mesh instancing, quantized
12 B vertex, chunked streaming on a probed VRAM pool, sidecar v13
fast path, event-driven rendering), per-frame pipeline (cull →
upload → streaming → opaque pass → transparent pass → edge → overlay
→ present), federation + false-origin compose, file map limited to
files that actually exist in `src/ifcviewer/` today, build/run via
`build_viewer.sh`, cross-refs to env-vars.rst, debug-output.rst,
and connectors/.
## Toctree
`src/bonsaiviewer/docs/index.rst` gains `env-vars` and
`viewport_architecture` entries alongside the existing
`connectors/index` and `debug-output`.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The Python implementation of the Autodesk Forma connector
(bonsaiviewer_autodesk/) is deprecated. The Rust port that's been
maturing under src/bonsaiviewer-autodesk-rs/ is now the connector
and takes over the original folder name.
## File operations
* `git rm -r src/bonsaiviewer-autodesk` — drop the 18 tracked Python
source/test/packaging files. (~6.5k untracked build artefacts in
venv/build/dist/egg-info are removed too, but those were never in
the index.)
* `mv src/bonsaiviewer-autodesk-rs src/bonsaiviewer-autodesk` —
the Rust impl takes over the canonical folder name.
* `rm -rf src/bonsaiviewer-autodesk-rs-egui` — abandoned egui-based
experiment, never committed.
* `src/bonsaiviewer-autodesk/.gitignore` extended with `/dist` to
keep packaging output out of the index alongside the existing
`/target` rule.
The Rust binary in Cargo.toml already has `name = "bonsaiviewer-
autodesk"` and `connector.json`'s `exec` field already points at that
name — so the connector loader, build_viewer.sh symlink, and
win/build-all-win.py CONNECTOR_DIR all keep working without edits.
## Packaging shape preserved
`packaging/build.py` is rewritten to:
* shell out to `cargo build --release` instead of pyinstaller,
* copy the produced binary + connector.json into the same
`dist/autodesk/` layout the PyInstaller flow produced,
* zip into `dist/autodesk-<os>-<arch>.zip` with the same
naming pattern (CI artifact uploads keep working).
The Rust binary statically links its deps, so unlike PyInstaller
there's no `_internal/` directory — single executable inside
`dist/autodesk/`. Everything downstream (`build_viewer.sh` symlink,
`win/build-all-win.py collect_connector_files`, the zip step in
`build_rocky.yml`) only cares that `dist/autodesk/` exists, so the
on-disk contract is preserved.
Verified locally: `python3 src/bonsaiviewer-autodesk/packaging/build.py`
produces `dist/autodesk/{bonsaiviewer-autodesk, connector.json}`
(3.9 MB stripped ELF) and `dist/autodesk-linux-x86_64.zip` (~1.5 MB
compressed).
## CI updates
* `.github/workflows/build_rocky.yml` and `build_rocky_arm.yml`:
drop the `pip install ".[build]"` step — `packaging/build.py` is
stdlib-only now, the cargo build wrapped inside it does the work.
* `.github/workflows/build_win.yml`: same — drop pip install,
packaging script handles cargo internally.
* `.github/workflows/build-bonsaiviewer-autodesk.yml`: full rewrite
of the dedicated connector test/build workflow. Replaces the
Python {3.11, 3.13} test matrix with `cargo fmt --check`,
`cargo clippy --all-targets -- -D warnings`, and `cargo test
--all-features`. The OS/arch build matrix is unchanged
(linux-x86_64, macos-arm64, macos-x86_64, windows-x86_64) but
installs a Rust toolchain via dtolnay/rust-toolchain@stable and
caches target/ via Swatinem/rust-cache.
`win/build-all-win.py` and `build_viewer.sh` are unchanged — they
only reference the `dist/autodesk/` path, which the new
`packaging/build.py` populates identically.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Unit / Geolocation / Placement are schema-agnostic IFC helpers ported
from ifcopenshell.util.{unit,geolocation,placement}. Nothing about
them is viewer-specific: pure IfcParse + Eigen, no Qt, no IfcGeom, no
renderer. Living under src/ifcviewer/ implies an unwanted dependency
direction every time a non-viewer caller (test_federation, the bonsai
SettingsView georef readout, a future standalone IFC tool) wants to
use them.
Move them to a new `src/ifcutil/` static lib (IfcUtil). The lib has
PUBLIC `target_include_directories(${CMAKE_CURRENT_SOURCE_DIR})` so
callers that link IfcUtil can keep `#include "Unit.h"` etc. without
relative-path adjustments — the include dir propagates transitively
via IfcViewer's PUBLIC link.
## Changes
* `git mv src/ifcviewer/{Geolocation,Placement,Unit}.{h,cpp}
→ src/ifcutil/` (history follows the rename).
* `src/ifcutil/CMakeLists.txt`: IfcUtil static lib, PUBLIC links
IfcParse + Eigen3::Eigen, PUBLIC include dir.
* `cmake/CMakeLists.txt`: `add_subdirectory(../src/ifcutil ifcutil)`
before ifcviewer/ so the link target exists when IfcViewer's
CMakeLists runs.
* `src/ifcviewer/CMakeLists.txt`: IfcUtil added to IfcViewer's PUBLIC
link_libraries.
* `src/ifcviewer/tests/CMakeLists.txt`: test_federation drops the
explicit `${IFCVIEWER_SRC}/{Unit,Geolocation,Placement}.cpp`
source list and links `IfcUtil` instead (matches how production
code resolves the symbols).
* `src/bonsaiviewer/modules/models/SettingsView.cpp`: the two
explicit `#include "../../../ifcviewer/{Geolocation,Unit}.h"`
paths swap to `../../../ifcutil/…`. All other callers use bare
`#include "Unit.h"` style and continue to work via the propagated
include dir.
## Verification
* `ninja -C build-viewer` builds clean: IfcUtil + IfcViewer +
IfcViewerMinimal + BonsaiViewer + all four pre-existing
ifcviewer tests + the two from-wgpu tests.
* `test_federation` runs green: 226 assertions in 22 test cases
pass with IfcUtil linked instead of the explicit-source compile.
* `git log --follow` traces e.g. `Geolocation.cpp` back through the
rename to its prior location in src/ifcviewer/.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
d390911d75 ("build_osx: build BonsaiViewer on macOS via build-all.py",
2026-06-01) accidentally committed a local single-version pin
(`PYTHON_VERSIONS = ["3.11.8"]`) intended only for fast iteration
during macOS bring-up. With macOS / CI green and the Python wrapper
fix from 748b4e72a landed, restore the full multi-version list so
both Rocky and macOS CI publish wrappers for 3.10/3.11/3.12/3.13/3.14
again.
Cost is ~5 from-source Python builds per CI run; the cache-deps
plumbing in nix/cache_dependencies.py already memoises these so
repeated runs only pay it once per Python release bump.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
## The bug
FZK-Haus windows rendered fully opaque despite every piece of the
data path carrying alpha correctly: vertex format is RGBA u8x4,
InstanceCpu/InstanceGpu carry color_override_rgba8 with its alpha
byte, fs_main returns vec4(rgb, in.color.a). Cause: the main render
pipeline's color target had `blend = nullptr`, which in wgpu disables
the blend stage entirely — fragment RGBA overwrites the back buffer
unmodified, alpha discarded.
## Why "just enable blend" isn't enough
Two failure modes that don't go away with a one-liner:
1. `depthWriteEnabled = True` on the main pipeline would make a
transparent window-frame pane occlude geometry behind it in
depth, so the wall behind the window then fails the depth test
and never draws — you'd see the silhouette of the window with
whatever colour was in the back buffer before, not the wall.
2. Order-dependent blending across transparent surfaces in arbitrary
cull order — overlapping transparent surfaces would shift colours
as the camera moves.
Standard fix for a BIM viewer is two-pass opaque-then-transparent.
## What this commit adds
### Per-mesh "has any alpha < 255" classifier
* `ModelGpuData::mesh_has_alpha` (uint8_t vector, parallel to meshes).
* Sized in `applyCachedModel`.
* Populated in `applyStreamedChunk` by scanning each in-chunk mesh's
vertex bytes for a vertex's alpha byte < 255 (offset 11 within
the 12-byte vertex record — the 4th byte of the third u32, which
the shader reads as `w2 >> 24`). Single chunk-arrival site covers
both sidecar streaming and the worker-result drain. First-load
IFC-without-sidecar geometry still routes opaque until the sidecar
bake completes; A-path scan is deferred.
### Per-chunk opaque/transparent partition during cull
* `Chunk::opaque_visible_vertices` / `opaque_visible_draws`
(per-frame counts).
* Transient `visible_draws_scratch_transparent` +
`transparent_per_draw_vertex_counts` filled alongside the existing
opaque half during the cull walk. Post-walk concat appends
transparent entries onto the opaque half and continues the
cumulative prefix-sum sequence — single buffer, single bind
group, no doubling.
* Classifier inside the cull lambda:
`xray_active ? always_transparent
: override_active ? (override.alpha < 255)
: mesh_has_alpha[mesh_id]`
### Per-chunk uniform layout extension
From `[total_draws, total_verts, 0, 0]` to
`[total_draws, total_verts, opaque_verts, opaque_draws]`. The third
slot is what `render()` passes as `firstVertex` to the transparent-
pass draw call so the shader's vid lands in the transparent range of
the same visible_draws_scratch buffer.
### `main_pipeline_transparent_`
Copy of `main_pipeline_` with `color_target.blend = SrcAlpha /
OneMinusSrcAlpha`. depthWriteEnabled stays True (see below).
### Two-pass `render()`
Opaque pass (`main_pipeline_`, firstVertex=0,
vertexCount=opaque_visible_vertices) then transparent pass
(`main_pipeline_transparent_`, firstVertex=opaque_visible_vertices,
vertexCount=total - opaque). Each loop skips empty halves so an
opaque-only chunk costs one draw call, transparent-only one draw,
mixed chunks two.
### depth_transparent.depthWriteEnabled = True (NOT off)
Initially set False (standard "let further-back geometry paint
through transparent front faces" trick) but that broke the edge-
detect pass: edge detection reads the depth buffer to find
silhouette discontinuities, and windows-without-depth meant the
glass had no silhouette at all (panes looked like framed holes) and
the edges of opaque geometry behind the glass painted through at
full intensity. Keeping the write avoids that — trade-off is depth-
test occlusion between transparent surfaces (closer occludes
farther), which for BIM panes that don't overlap in screen space
is invisible. Real fix for the overlap case is OIT or sort-back-
to-front, not depth-write toggling.
## Alt+X global X-ray (drops in basically free)
* `xray_alpha_cap` field on FrameUniforms + WGSL counterpart, default
1.0 (no effect). fs_main clamps `out.a = min(in.color.a, cap)`.
* `ViewportWindow::xray_alpha_cap_` member, default 1.0. Alt+X
toggles between 1.0 and 0.3.
* Cull classifier sees `xray_alpha_cap_ < 1.0` and forces every
instance into the transparent pass so the blend stage actually
fires (an opaque-pass fragment with capped alpha would still
overwrite the back buffer).
* No per-instance state mutation needed — toggle is a single float
in a uniform plus a re-cull. Excluding objects from x-ray later
would mean tagging them so the classifier skips the force-
transparent branch for them, also small.
Stress-tested on FZK-Haus: window glass visibly translucent with
correct silhouette edges; Alt+X turns the whole scene to a tinted
ghost of itself and back without artefact.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Three coupled fixes that close the macOS bring-up loop:
## 1. ifcwrap: fix INSTALL_RPATH on Apple
The ifcopenshell_wrapper Python module had `INSTALL_RPATH "$ORIGIN"` set
for "NOT WIN32 AND NOT WASM_BUILD" — but `$ORIGIN` is a Linux ld.so
placeholder, not a macOS dyld one. macOS dyld doesn't expand it; it
bakes the literal string `$ORIGIN` into LC_RPATH, which resolves to
nothing at runtime. The wrapper's hard-link `@rpath/ifcopenshell
.document.rdb.dylib` then fails to load even though INSTALL(TARGETS …
LIBRARY DESTINATION "${python_package_dir}/ifcopenshell") above had
already dropped the plug-in dylib right next to the wrapper.
Split the rpath assignment: `@loader_path` on Apple (the dyld
equivalent of `$ORIGIN`), `$ORIGIN` elsewhere.
This is what b0ef47819 (the build_osx IFCOS_BUILD_PYTHON_WRAPPER=off
gate) was working around. The gate is removed below.
## 2. ifcviewer-minimal: stage IfcOpenShell + wgpu_native into the .app
IfcViewerMinimal.app was building on macOS via the cmake
`BUILD_BONSAIVIEWER → BUILD_BONSAIVIEWER_WGPU` promotion, but had no
bundle staging — Contents/Frameworks/ only contained the Qt
frameworks macdeployqt deposited, so the .app would refuse to start
("Library not loaded: @rpath/libwgpu_native.dylib").
Mirror what src/bonsaiviewer/CMakeLists.txt does for BonsaiViewer.app:
* Set INSTALL_RPATH to `@executable_path/../Frameworks` so the exe
knows where to look for @rpath/* deps.
* install(FILES) libwgpu_native.dylib into the bundle's Frameworks/
(globbed from WGPU_NATIVE_LIB_DIR rather than hard-coded so it
covers any future versioned name).
* install(CODE) staging block that copies every `*.dylib` from
<prefix>/lib/ into the bundle's Frameworks/, excluding the
geometry-writer plug-ins (same EXCLUDE regex as BonsaiViewer.app —
viewer doesn't need OBJ/glTF/DAE/STP/IGS/SVG/TTL export converters).
Same long-form rationale + caveats apply (macdeployqt doesn't follow
non-Qt @rpath deps, lib-prefixed core libs vs ifcopenshell.* plug-in
naming split, Linux's equivalent lives in build_rocky.yml workflow
bash via patchelf + stage_runtime_payload). See src/bonsaiviewer/
CMakeLists.txt for the full version.
## 3. build_osx.yml: drop IFCOS_BUILD_PYTHON_WRAPPER=off
With (1) fixed, the Python wrapper smoke test should pass again. The
gate goes away; the comment block in build_osx.yml is replaced with a
short note pointing at the ifcwrap rpath fix as the underlying change
that re-enables this.
Together, (1)+(2)+(3) close the standalone IfcViewerMinimal-on-macOS
gap (task #43) and re-enable IfcOpenShell-Python on the macOS arm64 CI.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The GL backend is gone (task #53). The wgpu/non-wgpu folder split and
the Wgpu* class prefix were both disambiguation artefacts from the
overlap period — now pure dead weight.
## Folder + library merge
* `src/ifcviewer-wgpu/` → folded into `src/ifcviewer/` (git mv tracks
every file as a rename so blame/log history survives).
* `src/ifcviewer-wgpu-minimal/` → `src/ifcviewer-minimal/` (the exe was
already named `IfcViewerMinimal`; this just brings the folder + CMake
target name into line).
* `src/ifcviewer-wgpu/tests/test_wgpu_{selection,visibility}.cpp` →
`src/ifcviewer/tests/test_{selection,visibility}.cpp`, folded into
the existing `add_ifcviewer_unit_test(...)` helper.
* The `IfcViewerWgpu` static library is dissolved — its sources become
part of the unified `IfcViewer` static library, which now bundles
scene/loader + renderer in one target. The pre-merge circular
dependency (IfcViewer linking IfcViewerWgpu just to get the
ViewportWindow.h include path that SceneLoader.h needs) goes away.
* The wgpu-native FetchContent block, the Cocoa/QuartzCore link on
Apple, the OBJCXX-enabled `.mm` source, and the wgpu-native runtime
install all move into `src/ifcviewer/CMakeLists.txt` unchanged.
## Type renames (Wgpu prefix dropped from every Wgpu* identifier)
WgpuAreaMeasurement → AreaMeasurement
WgpuBufferPool → BufferPool
WgpuLengthMeasurement → LengthMeasurement
WgpuMetalSurface → MetalSurface
WgpuModelGpuData → ModelGpuData
WgpuOverlayFrame → OverlayFrame
WgpuOverlayRenderer → OverlayRenderer
WgpuSectionPlane → SectionPlane
WgpuSelectionState → SelectionState
WgpuStreamingLoader → StreamingLoader
WgpuStreamingThread → StreamingThread
WgpuViewportWindow → ViewportWindow
WgpuVisibilityState → VisibilityState
CMake target IfcViewerWgpuMinimal → IfcViewerMinimal (exe name was
already this since wgpu shipped as default).
Deliberately kept: `onWgpuLog` (wgpu-native log callback — names a
binding to an external API, not one of *our* types), and the WGPU*
enum/struct prefixes from wgpu-native's own headers. `WgpuMemProbe`
lives in the separate `src/wgpu-mem-probe/` standalone diagnostic
project and isn't touched.
## Include-path updates
Every `#include "../ifcviewer-wgpu/Wgpu<X>.h"` → `"../ifcviewer/<X>.h"`,
every in-directory `#include "Wgpu<X>.h"` → `"<X>.h"`. Includes from
sibling subdirectories (modules/, etc.) are updated to point at
`../../../ifcviewer/` instead of `../../../ifcviewer-wgpu/`.
## cmake/CMakeLists.txt simplification
The redundant `add_subdirectory(ifcviewer-wgpu)` blocks (one inside
the BUILD_BONSAIVIEWER fan-in, one in the BONSAIVIEWER-less standalone
block) collapse into a single unconditional
`add_subdirectory(../src/ifcviewer ifcviewer)`. The standalone block
keeps only `wgpu-mem-probe` (the diagnostic tool, unrelated to the
viewer lib).
## Verification
* Full build green: `IfcViewer` static lib, `IfcViewerMinimal` exe,
`BonsaiViewer` exe, all four pre-existing ifcviewer unit tests, and
the two new-location tests (`test_selection`, `test_visibility`).
* No stray `Wgpu<X>` identifier remains across `src/ifcviewer/`,
`src/bonsaiviewer/`, `src/ifcviewer-minimal/` (verified by grep).
* Renames tracked by git as `R` entries — `git log --follow` on
ViewportWindow.cpp etc. continues to show history through the move.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Two independent threads that landed on this branch.
## 1. Federation false-origin guess: arm-on-add + frame-on-origin
The cc54237f5 fix moved the guess off the refresh() fan-in (terminating
the stack-overflow recursion on Holter Tower), but the gating was still
`modelIds().size() == 1` — broken for batch-add-into-empty-session
because the session registry is populated synchronously by all
addModel calls before any modelGeometryReady lands. Result: load 3
files at once → none of the per-model geometry-ready events ever
finds `size() == 1` → guess never fires → camera + federation origin
stay at surveyor coords.
Rework:
* **Arm/consume at the command boundary.** modules/models/Commands
exposes `armFederatedFalseOriginGuess` / `consumeFederatedFalseOrigin
Guess` (function pair — arming is a one-shot, raw-bool would let a
peek-without-clear silently break the contract). `addModel` and the
cloud-callback in `addModelFromCloud` arm if `modelIds().isEmpty()`
at the moment they're about to register federation entries. The
first geometry-ready then consumes the arm and runs the guess —
batch add or single, works the same way.
* **Lazy first geometry point from mesh AABB centre.** Drop
SceneLoader's `firstPlacement(mid)` / `first_placement` /
`has_first_placement` and the two capture sites entirely. The
viewport keeps CPU-side MeshInfo + InstanceCpu for picking /
measurement; compute the anchor on demand via new const accessor
`WgpuViewportWindow::firstGeometryPointWorldM(mid, out)` =
instance0.placement × meshes[instance0.mesh_id].aabb_centre.
This is more representative than the placement translation
(placements often live far from the actual geometry due to long
ObjectPlacement chains / intermediate local frames), and lighter
storage-wise (lazy, vs. 128 B per model held just-in-case).
* **`guessFederatedFalseOrigin` math signature: Matrix4d → Vector3d.**
The function only ever consumed `.block<3,1>(0,3)`; the Matrix4d
API surface was a strictly-larger-than-necessary contract. Vector3d
matches what the function actually needs.
* **`WgpuViewportWindow::frameOnFederatedOrigin(mid, max_distance_m)`**
replaces the post-shift use of viewAll() that the ViewportView
almost reached for. The federated false origin sits at (0,0,0) in
post-shift space by construction, so the camera targets there
directly; distance fits the model's post-shift AABB diagonal with
viewAll's padding math, clamped to `max_distance_m` so a model
with one crazy-coord outlier vertex can't pull the camera back so
far the real geometry becomes a pixel. Called with 100 m cap from
the guess. Unlike viewAll() this only iterates the one model the
guess fired for — the "load 10 models, viewAll shows nothing"
failure mode is structurally avoided.
* **ViewportView::tryGuessFirstModelFalseOrigin** renamed to
`guessFederatedFalseOriginFromFirstModel` and the body restructured
to consume the arm, look up the anchor via the viewport, mutate
the federation origin (which propagates through SessionState's
federatedFalseOriginChanged relay → refresh() → recompose all
instance world AABBs), then `frameOnFederatedOrigin(mid, 100)`.
* **Internal guards preserved.** filePath skip (project files own
the origin), current==defaults skip (don't clobber a user who
set the origin manually then removed the model), placement /
georef availability checks — defense-in-depth around the arm, not
the primary gate. The arm-only flow means re-arming on add-into-
empty-session is naturally re-firable: add → remove → add will
retry if the previous guess returned defaults.
## 2. wgpu present-mode: prefer Immediate above FifoRelaxed
On Linux Vulkan stacks where the driver / compositor doesn't advertise
Mailbox (confirmed on the user's setup — capability log added in this
patch reports just `fifo, fifo_relaxed, immediate`), Fifo's 2–3 frame
queue doubles input-to-photon latency the moment WASD activates
(~16 ms render-body fully consumes the budget, so the queue is held
deep). On a 60 Hz display this reads as "less smooth than the 100 fps
HUD suggests" during fly-mode mouse-look-while-moving — confirmed by
WGPU_FLY_DEBUG dt traces (rock-solid 16-17 ms cadence, so it's not
frame pacing — it's latency).
Promote Immediate above FifoRelaxed in the preference order so that
when Mailbox is unavailable we pick the no-queue option (can tear
under fast motion, but tearing on architectural geometry is usually
invisible while the latency win is immediately felt). Also log the
full advertised capability list on first configure so future "why
isn't Mailbox available?" diagnostics don't need a code patch.
Mailbox remains first preference; Fifo remains the spec-guaranteed
final fallback.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The 7 multi-wall operators (UnjoinWalls, UnjoinWallPathConnection,
ExtendWallsToUnderside, ExtendWallsToWall, SplitWall, MergeWall,
JoinWallsIntersection) each opened their _execute with an identical
prologue:
_commit_pending_wall_edits_for_selection(context)
# ... operator-specific logic
— flushing any in-progress wall parametric drafts so the operator
acts on committed IFC state rather than the draft preview box.
Extract that prologue into _CommitWallDraftsFirstMixin: its _execute
calls the commit helper, then delegates to a subclass-supplied
_perform. Subclasses inherit the mixin first in their bases tuple so
the mixin's _execute resolves first via the MRO. The IFC transaction
opened by tool.Ifc.Operator.execute still wraps both the commit and
the perform.
Behaviour-equivalent — same call, same order, same selection scope.
Architectural cleanup only: a future multi-wall operator can no
longer forget the commit step. The named helper
_commit_pending_wall_edits_for_selection stays as the single
encapsulation of the names=("wall",) filter; its docstring loses
the stale "every multi-wall operator calls it at the top of
_execute" sentence and now just describes the filter contract.
Matches gizmos-8088's _CommitWallDraftsFirstMixin pattern.
Generated with the assistance of an AI coding tool.
The regular bim.add_opening click on the host-add-opening gizmo
(wall + door/window co-selected) routes through
FilledOpeningGenerator.generate, which snaps the filling to the
wall's reference-line axis, optionally rotates 180° when the
filling sits on the opposite side, and re-applies an rl1 / rl2
Z-elevation default. That is the right default for "drag a fresh
door onto a wall and let the model place it for me", but defeats
the workflow where the user has already positioned the filling
precisely (e.g. snapped to a window in an adjacent wall, copy-
pasted at an exact Z, aligned to a reference object).
Holding SHIFT while clicking the gizmo now opts into a
"preserve placement" mode: the filling stays at its current
matrix_world and the opening is created at the filling's existing
position. The opening / filling rels and representation work are
unchanged — only the snap-to-axis branch is skipped, so the IFC
graph is identical to the regular click; only the spatial
position of the filling differs (user-chosen vs auto-snapped).
Implementation:
* bim/module/void/operator.py: AddOpening gains a hidden
preserve_placement BoolProperty + an invoke() that sets it from
event.shift. The call into FilledOpeningGenerator.generate
forwards the flag. bl_description documents the SHIFT modifier
so it surfaces in F3 search / hover tooltip.
* bim/module/model/opening.py: FilledOpeningGenerator.generate
accepts preserve_placement (default False — backwards-compatible
with the other caller, tool.Model.add_filled_opening). The
voided_obj.data-gated snap block (raycast + axis projection +
rl-Z default + filling_obj.matrix_world write) skips entirely
when the flag is True. The opening's matrix_world reads from
filling_obj.matrix_world below the gate, so the opening lands
at the filling's preserved position automatically.
Generated with the assistance of an AI coding tool.
The door / window / stair edit-row's type-cycle icon advanced one
type per click (CycleDoorType / CycleWindowType / CycleStairType
bound to cycle_type_operator). DoorType has 8 IFC variants,
WindowType 9, StairType 3 — so cycling past the target was the norm.
Threshold rule for cycle-vs-menu: cycle is appropriate for exactly 2
values (advance-one-per-click stays predictable). Three or more
values warrants a popup menu. Door / window / stair all qualify;
roof (RoofGenerationMethod has 2 values) keeps cycle. Wall has no
type cycle. Array is unaffected.
Swap to the popup-menu pattern (PickTypeMixin already on HEAD at
bim/parametric_lifecycle.py:442): clicking the icon opens a menu
listing all type_literal values; selecting one applies it in a
single undo step. The hamburger icon (VIEW3D_GT_menu) is wired into
BaseParametricGizmoGroup.setup_editing_gizmos whenever
pick_type_operator is set (mutually exclusive with
cycle_type_operator). Matches gizmos-8088's pattern exactly.
Per-feature shape:
* door.py: PickDoorType replaces CycleDoorType.
GizmoDoorEdition.cycle_type_operator → pick_type_operator.
* window.py: PickWindowType replaces CycleWindowType. Same swap.
* stair.py: PickStairType replaces CycleStairType (no
tool.Ifc.Operator inheritance — stair-type changes
BIMStairProperties only, no IFC mutation). Same swap.
* bim/module/model/__init__.py: registration entries renamed
Cycle* → Pick*.
* bim/module/drawing/gizmos.py: drop the
CycleTypeMixin / PickTypeMixin / TypeAccessorBase shim re-export —
its own docstring already noted "PR5 cleanup drops these" and the
three callers (door / window / stair Cycle*Type) it served are
gone. Roof's CycleTypeMixin import was already direct from
bim.parametric_lifecycle. Also update GizmoMenu docstring to
reflect the 2-vs-3+ threshold.
Generated with the assistance of an AI coding tool.
When entering parametric edit on a door or window that fills a
wall opening, four dimension gizmos now measure the distances
from the wall edges to the filling's jambs and from the wall's
base/top to the sill/header. Dragging any gizmo translates the
filling along the wall's local axis; 180°-flipped fillings and
slanted LAYER2 walls round-trip correctly. The has_host_wall
predicate hides all four when the filling → opening → wall
chain cannot be resolved.
Generated with the assistance of an AI coding tool.
Four concerns that together make the cursor-anchored gizmos
(extend_x_gizmo, extend_z_gizmo, split_gizmo on GizmoWallEdition)
fully functional and visually informative without entering parametric
edit mode first:
* Drop the props.is_editing gate in _update_cursor_gizmos. The three
bound operators (bim.extend_wall_to_cursor,
bim.extend_wall_height_to_cursor, bim.split_wall_at_cursor) already
poll on wall-selected and commit any pending wall edit before
acting, so single-click without entering edit mode is now the
canonical flow. Matches gizmos-8088's always-on behaviour.
* Register GizmoWallEdition instances in a per-region weakref map
(_active_instances) populated at setup_element_specific_gizmos
time. The WallGizmoPreviewDecorator dereferences this map to read
live is_highlight state off the cursor icons. Without the
registration its _cursor_icon_hovered always returned False and
the hover-gated GPU previews silently never drew. Mirrors the
same pattern already in place on GizmoWallJoinIntersection.
* Add post-operator resync to all three cursor operators
(_maybe_resync_wall_props_from_ifc for the single-wall split /
extend-height paths, _resync_walls_after_mutation for the
selection-wide extend-X path). Without this, props.length /
props.height stayed stale after the operator ran, so the
orientation flips _apply_wall_extend_flips computes from
cursor_local vs wall dimensions kept using the pre-extend values
until the next selection change. Matches gizmos-8088's pattern.
* Hover-gated GPU previews per icon:
- extend-X: filled Z=0 floor quads spanning the wall's offset to
offset+thickness Y band, visible from plan view without side-
view clutter. Grow case (cursor beyond either endpoint): one
green decorator_color_selected quad over the extension. Shrink
case (cursor inside extent): green quad for the portion that
REMAINS + red decorator_color_error quad for the portion the
operator REMOVES.
- extend-Z: vertical lines at the cursor's projected X in the
wall's y=0 reference-line plane. Grow case (cursor above wall
top): one green segment from z=height to z=cursor.z. Shrink
case: green from z=0 to z=cursor.z (REMAINS) + red from
z=cursor.z to z=height (REMOVES).
- split: one red vertical line at the cursor's projected X from
base to wall top — the cut plane.
Quads use QUAD_ALPHA=0.25 so the underlying wall body stays
visible.
* New module-level _fill_quads_alpha helper next to
_stroke_lines_alpha, plus a per-decorator _fill convenience method
and a _wall_floor_quad corner builder.
Modal-active gizmo hiding (is_gizmo_hidden_by_modal) is preserved.
Generated with the assistance of an AI coding tool.
Add openings — GizmoWallAddOpening only fired when a wall was active +
co-selected with a non-host; slabs and roofs got no in-viewport handle.
GizmoHostAddOpening covers all three host types via is_supported_host,
dispatching walls to the axis-projection anchor and slabs/roofs to a
world-Z anchor lifted just above the host's top face (predictable
height regardless of the void's vertical position).
Show openings on hosts with their own parametric-edit toolbar —
GizmoRoofEdition gains an idle-row toggle_openings_gizmo parallel to
the wall's, parked at the cancel-slot X next to the pen. Visible only
when the host carries HasOpenings and the edit triad is idle. Roof
overrides get_element_height to return the mesh's world-AABB top in
object-local Z, so the WHOLE pen-row anchors visibly above sloped or
stepped roof bodies. The wall's idle-row toggle now also hides when
HasOpenings is empty.
Show openings on hosts WITHOUT a parametric-edit toolbar —
GizmoHostToggleOpenings scoped strictly to the fallback case: a single
host selected, HasOpenings non-empty, NOT a path-connectable wall, NOT
a parametric roof. Covers slabs today plus any foreign-authored IfcRoof
without BBIM_Roof. Anchored at object origin XY + world-AABB top Z.
When slab parametric-edit eventually lands, the slab predicate joins
the exclusion list and this gizmo's poll narrows automatically.
Operator move — ToggleWallOpenings was already host-agnostic; renamed
to ToggleHostOpenings in opening.py (bl_idname bim.toggle_host_openings).
Three callers (the wall idle-row binding, GizmoWallFilletToggleOpenings,
and workspace.py's hotkey_A_O for Alt+O) now route through the renamed
operator. The Alt+O binding is surfaced in the operator's
bl_description so it appears in F3 search and hover tooltips.
DRY refactors —
* GizmoWallAddOpening deleted (subsumed by GizmoHostAddOpening)
* tool.Blender.get_object_world_bounding_box added as the world-AABB
sibling of the existing local helper; 3 inline call sites in
tool/misc.py (set_object_origin_to_bottom, scale_object_to_height)
and gizmos.py adopt it (2 other sites in drawing/operator.py and
project/operator.py inherently need raw transformed corners for
per-corner plane / NDC tests — not AABB candidates)
* BaseParametricGizmoGroup gains setup_pen_row_toggle_openings_icon +
update_pen_row_toggle_openings_icon; wall + roof + any future host
gizmo wire up the idle-row toggle with two one-line calls
* _resolve_active_host shared poll prologue between the two host
gizmos (gate + selection count + active-in-selected + entity lookup
+ supported-host check)
* HasOpenings non-empty checks at 3 sites route through
tool.Geometry.has_openings
* hotkey_A_O body collapsed to bpy.ops.bim.toggle_host_openings()
The forward-compat AST guard pinning "must accept fillet-corner walls"
retargets from GizmoWallAddOpening.poll to is_supported_host.
Generated with the assistance of an AI coding tool.
The .app bundle's Frameworks/ staging rule was only globbing
ifcopenshell.*.dylib (the dlopen-only plug-ins) on the assumption
that macdeployqt would follow BonsaiViewer's link-time @rpath deps
for the lib-prefixed core shared libs. In practice it doesn't —
non-Qt @rpath deps whose source path is outside the standard system
/ Qt prefixes get skipped silently.
In a static build this didn't matter: libifcopenshell.geometry,
libIfcParse, libIfcViewer, etc. were statically embedded in
BonsaiViewer.exe, so there was no runtime dep. With --shared (added
in ddee88bed for the bundle-size win) they're separate dylibs that
must physically live in Frameworks/, or the binary won't even start:
dyld: Library not loaded: @rpath/libifcopenshell.geometry.dylib
Reason: tried '…/BonsaiViewer.app/Contents/Frameworks/
libifcopenshell.geometry.dylib' (no such file)
Broaden the install(CODE) glob to *.dylib so both flavours land:
* lib-prefixed linked core libs (libifcopenshell.geometry.dylib,
libIfcParse.dylib, libIfcViewer.dylib, lib<mapping/kernel>.dylib)
* non-prefixed plug-ins (ifcopenshell.parse.schema.ifcXxX.dylib,
ifcopenshell.geometry.mapping.ifcXxX.dylib, etc.)
<prefix>/lib/ is IfcOpenShell-exclusive (Qt / boost / eigen live in
their own brew / build prefixes), so the broad glob doesn't risk
sweeping in unrelated dylibs. The geometry-writer EXCLUDE regex is
preserved so the size win from the writer-skip side of ddee88bed
stays in place.
In a static build the lib/ directory simply has no *.dylib files
that match, so the rule no-ops cleanly — same code path is safe for
both --shared and the (unused but possible) default static config.
Linux didn't need a counterpart: build_rocky.yml already does the
equivalent in workflow bash (`patchelf --set-rpath '$ORIGIN'` +
`stage_runtime_payload`), and local Linux dev uses CMake's
BUILD_RPATH which auto-resolves to the build subdirectories. macOS
.app bundles are treated as opaque by the packaging step (no
stage_runtime_payload against Contents/), so the staging has to
happen at CMake install time when the bundle is being assembled.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Loading a model whose first placement sits at the world origin
stack-overflowed BonsaiViewer instantly on Windows (and macOS).
WinDbg trace was a 5-frame Qt signal-slot cycle hitting the guard
page ~1400 levels deep; Linux escaped only because that machine's
iterator order put a non-origin instance first, which made the guess
return a non-default value and naturally terminated the recursion
after one step.
Root cause is the architecture, not the specific guard inside the
guess function. `ViewportView::refresh()` was connected to six
SessionState signals (projectReset, projectOpened, modelsChanged,
federationChanged, visibilityChanged, modelGeometryReady) and was
calling `maybeGuessFederatedFalseOrigin` on every model on every
fire. That helper called `session_state_->notifyFederationChanged()`
unconditionally after the mutation, which re-emitted
SessionState::federationChanged, which re-entered refresh(), which
re-entered the guess — a hidden emit-in-slot loop. The "current ==
defaults" guard at the top of the guess prevented further mutations
once the value moved off defaults, but on machines where the guess
itself returned defaults the guard never fired and the loop ran
forever.
Cleanup:
* refresh() is now terminal: it reads federation state, pushes it to
the viewport, and returns. No mutations, no signal emissions.
maybeGuessFederatedFalseOrigin is removed from its for-loop.
* The guess is renamed to `tryGuessFirstModelFalseOrigin(uint32_t)`
and is now invoked only from the modelGeometryReady connection,
not from refresh(). Conditions:
1. modelIds().size() == 1 (the just-loaded model is the only
model — i.e. this is the "first model added" edge)
2. federation->federatedFalseOrigin() == defaults (nobody has
set the origin yet — possibly because the previous attempt
guessed defaults and no-op'd, in which case we deliberately
want to retry next time a model lands)
No one-shot flag: add→remove→add cycles re-attempt the guess
precisely while the origin is still default, which is the right
semantics.
* SessionState now relays Federation::federatedFalseOriginChanged
onto its own bus via notifyFederationChanged. This replaces the
manual `session_state_->notifyFederationChanged()` call the old
guess made post-mutation. With the relay in place, any future
mutation site (commands, settings dialog, project load) will
propagate to views automatically — the emit point lives at the
data change, not at every caller. Views still subscribe to
SessionState only; Federation stays a back-end detail.
Reproduced with ISSUE_053_20181220Holter_Tower_10.ifcview on
Windows (build 21d3945, WinDbg `kn30` showed the recurring cycle
explicitly). Diagnosis confirmed on Linux by adding tracing prints
in refresh() and the guess body: same model, same code, but the
iterator's first-placement happened to be non-origin so the loop
terminated after one step.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Mirrors the Rocky workflow's two-part size reduction (27249770e
"Reduce Rocky package size") on macOS:
1. Pass `--shared` to nix/build-all.py. The default builds
IfcOpenShell as static libs, which means every plug-in dylib
(schemas × 8, kernels × 3, mappings × 8, writers × 8, document
serializers × ~4, linework processing) statically embeds a full
copy of libIfcParse + libIfcGeom. With --shared the plug-ins
reference @rpath/libIfcParse.dylib + @rpath/libIfcGeom.dylib and
the per-plug-in dylib drops from ~30-50 MB to a few MB each.
Dominant size win.
2. Filter `ifcopenshell.geometry.writer.*.dylib` out of BonsaiViewer's
plug-in staging step in src/bonsaiviewer/CMakeLists.txt. These are
the per-schema OBJ / glTF / DAE / STP / IGS / SVG / TTL export
converters — heavy because each one inlines the full schema, and
BonsaiViewer is a viewer, never an exporter, so they're pure
deadweight inside the bundle. Additive on top of --shared.
Bundle went 300 MB → expected ~100 MB, in line with Linux (~100 MB)
and Windows (~80 MB).
The IFCOPENSHELL_BUILD_PYTHON_WRAPPER=off gate is unchanged for now
— once we confirm BonsaiViewer.app size + functionality look sane,
we can ungate the Python wrapper and see if shared-builds-on-macOS
shake out its install issues too.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The extend-X / extend-Z / split icons share the cursor's projected X
on the wall axis, separated only by world Z (floor / cursor / wall
top). World Z collapses to a single screen point in plan view, so
every icon piled onto extend-X's hit target and only the topmost was
clickable.
Two refinements ported from gizmos-8088:
* When ``tool.Blender.is_view_top_down(context)`` reports the camera
is near plan-view, swap world-Z stacking for screen-up stacking:
anchor all icons at the floor world position and offset each by
``index * CURSOR_STACK_OFFSET`` along ``tool.Blender.get_screen_up_world(context)``.
Each icon lands in its own screen-space slot regardless of view
rotation.
* In the same top-down branch, drop ``extend_z_gizmo`` entirely. A
vertical-intent gizmo has no readable cue when looking down +Z —
clicking it would mutate the wall in a direction the user can't
see change.
* Bonus: split's local Z now goes through
``core.extrusion_depth_from_vertical_height(props.height, props.x_angle)``
so the icon lands on the slanted top edge of sloped walls (x_angle
!= 0) instead of the vertical-height target the wall isn't at.
All three helpers (``is_view_top_down``, ``get_screen_up_world``,
``extrusion_depth_from_vertical_height``) already on HEAD from PR2/PR3.
Non-top views unchanged — same world-Z stacking + cascading bumps as
before.
Generated with the assistance of an AI coding tool.
Hover-gated viewport preview lines that show where a wall-join /
extend / split operator would land before the user clicks. Four
preview paths, each gated on a specific icon's ``is_highlight`` state:
* **Join intersection** — two LAYER2 walls selected in the ``intersect``
state (non-joined, non-collinear, non-parallel). Draws four lines:
each wall's axis at both base and top Z, extending from the wall's
nearer endpoint to the projected XY intersection. The pair of lines
per wall communicates the full plane the join welds at, not just
the floor edge.
* **Cursor extend** — single LAYER2 wall, hover on ``extend_x_gizmo``.
One line from the wall's nearer X endpoint to the cursor's projected
X on the wall axis.
* **Cursor extend-Z** — hover on ``extend_z_gizmo``. Vertical line at
the cursor's projected X from wall base to cursor Z (the new total
height).
* **Cursor split** — hover on ``split_gizmo``. Vertical line at the
cursor's projected X from wall base to wall top — the cut plane.
Warning-red colour matches the icon's destructive-action signal.
Hover colour rules for the join preview:
* **Join or Fillet hover** → all four lines highlight in
``decorator_color_selected``. Both icons commit a symmetric corner
meet, so every line is part of the operation.
* **Extend-to-Wall hover** → only the non-active wall's two lines
(base + top) highlight. The default-direction extend operator
moves the non-active wall into the active one's axis; only that
wall's preview should signal motion.
* No hover → all four lines in ``decorations_colour``.
Three coordinated changes:
* ``bim/module/model/wall.py`` gains the ``_classify_wall_join_state``
wrapper over ``core.classify_wall_join_state`` (feeds the
``_are_walls_joined`` flag the core helper expects) AND a
``_active_instances`` per-region weakref ClassVar on
``GizmoWallJoinIntersection`` populated in ``setup()``. Without the
weakref registration, the decorator's ``_lookup_active_instance``
call returns None every frame and the hover gates silently
evaluate False — the symptom would be preview lines that never
switch colour. Both pieces ported from gizmos-8088.
* ``bim/module/model/decorator.py`` gains
``WallGizmoPreviewDecorator`` (~280 LOC across the four preview
paths + shared helpers ``_stroke`` /
``_active_layer2_wall_for_gizmo_preview`` /
``_join_group_hover_state`` / ``_extended_wall_index``). All
cross-file dependencies (``core.classify_wall_join_state``,
``core.wall_join_preview_lines``, ``_stroke_lines_alpha``,
``_cursor_icon_hovered``, ``_lookup_active_instance``,
``tool.Parametric.is_path_connectable_wall``,
``_wall_axis_world_segment_from_geom``) already on HEAD.
* ``bim/handler.py`` wires ``WallGizmoPreviewDecorator.install()`` /
``.uninstall()`` alongside the other always-on preview decorators.
The decorator self-polls every frame; cost is one selection-count
check + one ``is_highlight`` read when no eligible state is active.
Verified: headless smoke green, ruff + black clean. Live testing
confirms the four preview paths fire correctly when hovering each
icon.
Generated with the assistance of an AI coding tool.
GizmoWallJoinIntersection used to place its icons at state-specific
world points: join at floor Z, extend-to-wall at the active wall's
top Z, fillet stacked screen-up above join. Same XY at different Z
collapses to a single screen pixel in plan / top view, so two icons
became one hit target — invisible from above.
* position_gizmos now always-stacks along screen-up at a wall-top
anchor in both the joined (unjoin + fillet) and the intersecting
(extend + join + fillet) states. Order bottom-up is
extend / L / fillet. Collinear-merge keeps its single boundary
icon (no stack needed).
* New _stack_anchor_z picks the active wall's top Z (or the taller
of the two on mid-selection-transition frames). New _stack_at
lays a tuple of icons along screen-up at the resolved anchor.
* Glyph swap: join_icon -> VIEW3D_GT_wall_corner (L), extend_to_wall_icon
-> VIEW3D_GT_wall_tee (T). Both classes already existed in
bim/module/drawing/gizmos.py from an earlier commit; only the
setup() bl_idname strings changed. The previous arrow-merge /
arrow-extend pair read as the same direction once stacked.
Forward-compat AST contracts in test_wall_gizmos_forward_compat.py
pin the new invariants: the L and T bl_idnames must appear in
setup(), and position_gizmos must route through _stack_at so a
regression that reintroduces a direct billboarded_at write for any
state-specific icon fails CI before it flattens the stack again.
Also folds in a one-line typo fix in core/spatial.py:
assign_container's per-element can_contain check iterated `e` but
predicate-tested `root_element` (the outer for-loop variable), so
every element in the comprehension was tested against the same
container/element pair. Switch the argument to `e`.
Generated with the assistance of an AI coding tool.
Bring in 13 commits from upstream v0.8.0 (tip f158ae737):
- Add regenerate_wall_to_underside operator + has_underside_connection
Model interface (closes#7943)
- Extend/regenerate walls to multiple undersides
- Fix duplicate booleans in extend_walls_to_underside
- Fix extend_walls_to_underside ridge artifact
- Regenerate connected walls when recalculating a slab
- Fix validate_type corruption; remove debug prints
- Lazy BVH tree construction in SnapObj + early-terminate solid raycasts
in non-xray mode + optimize 2D projection in ray_cast_by_proximity_2d
- Fix crash in update_bim_tool_props when selected type isn't a valid
ifc_class
- Fix assign_container in spatial.py (#8079)
- Fix sign of temporary offset restore in sweep_along_curve
Auto-merge resolved all overlap files cleanly:
- bim/handler.py: work branch's update_bim_tool_props refactor and
upstream's try/except hardening converged on identical try/except
around props.ifc_class assignment (no net change).
- bim/module/model/__init__.py: upstream's wall.RegenerateWallToUnderside
entry and work branch's roof gizmo entries occupy disjoint sections.
- bim/module/model/wall.py: upstream's RegenerateWallToUnderside operator
and work branch's GizmoWallEdition/IconSlot refactors occupy disjoint
sections.
- core/tool.py: upstream's four new Model stubs and work branch's
Root.has_material_styles stub occupy different classes.
Partly generated with the assistance of an AI coding tool.
# What
Two upstream-shaped fixes to ifcopenshell's runtime plug-in discovery
so the schema/serializer plug-ins are findable in deployment layouts
other than a flat \`<prefix>/lib/\` (specifically: macOS .app bundles).
## (1) Stable anchor variable instead of a function pointer
\`schema_plugin_directory()\` used \`&load_schema_plugins\` as the
anchor whose containing module \`dladdr\` is asked to resolve. Function
addresses are not reliably equal to a single canonical location across
toolchains — on macOS arm64 with BonsaiViewer.app, \`&load_schema_plugins\`
took the address of a PLT/stub inside the consumer binary rather than
the actual symbol inside \`libIfcParse.dylib\`. \`dladdr\` then dutifully
returned the consumer's path and the loader started searching
\`BonsaiViewer.app/Contents/MacOS/\` for plug-ins that were never
installed there.
A variable doesn't suffer from this — it has exactly one canonical
address inside its defining dylib. Add \`ifcopenshell_libifcparse_anchor\`
(exported via IFC_PARSE_API) and use \`&that\` instead. Standard pattern
used by Boost.DLL, GStreamer, \`_dyld_get_image_*\`, etc.
## (2) Bundle-aware fallback search paths
The primary search path is \`dirname(libIfcParse)\`. That works for
flat installs (Linux \`lib/\`, Windows \`bin/\`) where plug-ins are
siblings of libIfcParse. macOS app bundles split the layout:
\`macdeployqt\` puts non-Qt @rpath deps in \`Contents/Frameworks/\`,
Apple convention asks for \`Contents/PlugIns/\`, and some install rules
co-locate libIfcParse with the exe in \`Contents/MacOS/\`. Plug-ins
typically end up in a sibling directory, not the same one.
In \`add_search_paths_or_default\`, after registering the primary path,
also register \`<parent>/PlugIns\`, \`<parent>/Frameworks\`, and
\`<parent>/MacOS\` on Apple platforms. \`discover_exact\` short-circuits
on the first hit so duplicates and missing directories are harmless.
# What this does NOT do
The plug-in dylibs still need to actually be inside the app bundle
somewhere for these fallbacks to find them — the upstream install
rules (\`install(TARGETS …)\` in \`src/ifcparse/CMakeLists.txt\`,
\`src/serializers/CMakeLists.txt\`, etc.) put them in \`<prefix>/lib/\`
which lives outside \`BonsaiViewer.app\`. That side of the fix is a
follow-up — either an explicit bundle-aware install destination on
the plug-in targets, or an install(CODE) sweep that mirrors them
into the bundle.
# What this also un-does
Reverts the BonsaiViewer-only \`install(CODE)\` hack that was about to
copy \`ifcopenshell.*.dylib\` from \`lib/\` into
\`BonsaiViewer.app/Contents/MacOS/\` — superseded by the loader-side
fix above, which lets us put the plug-ins anywhere sane inside the
bundle without further consumer-side stitching.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Completes the PR4/PR5 cleanup the FIXME at tool/blender.py
flagged: every is_<type> / Array.<helper> shim on
tool.Blender.Modifier delegated one-for-one to tool.Parametric /
tool.Array. Callers now reach the canonical home directly, and the
shim block — seven is_<type> classmethods plus the inner class Array
— comes out.
Renames (no semantic change):
* tool.Blender.Modifier.is_<door|railing|roof|stair|wall|window>
→ tool.Parametric.is_<x>
13 sites across tool/loader.py, bim/import_ifc.py,
bim/module/geometry/{data,operator}.py, bim/module/model/{door,
railing,roof,stair,ui,wall,window}.py.
* tool.Blender.Modifier.Array.<helper> → tool.Array.<helper>
4 sites across tool/root.py, bim/import_ifc.py,
bim/module/geometry/operator.py.
* test_parametric_registry.py: the two getattr probes that hunt
predicates by name now look on tool.Parametric. Docstring + the
test function name (test_every_entry_has_modifier_predicate →
test_every_entry_has_parametric_predicate) follow the move.
Kept on tool.Blender.Modifier (non-shim, no equivalent on
tool.Parametric): try_applying_edit_mode,
try_canceling_editing_modifier_parameters_or_path,
is_eligible_for_<x>_modifier (×5), is_array_child, is_slab.
Verified: 109 model-lane tests + 8 parametric-registry tests pass
(the one pre-existing failure in test_wall_header_refresh.py is
unrelated — it patches handler.update_bim_tool_props which has been
renamed). git grep for tool\.Blender\.Modifier\.(is_<type>|Array\.)
returns empty. black + ruff clean on every touched file.
Generated with the assistance of an AI coding tool.
Ports roof parametric edit gizmo group from gizmos-8088 and folds in
three roof-mesh bug fixes surfaced during live testing.
Port:
* ``CycleRoofGenerationMethod`` operator (bim.cycle_roof_generation_method)
cycles props.generation_method between "HEIGHT" and "ANGLE". Shift+click
cycles in reverse via the ``CycleTypeMixin`` contract.
* ``GizmoRoofEdition`` gizmo group: 3 dimension gizmos for height
(visible in HEIGHT mode) / slope angle with tan/atan2 rise round-trip
+ degree formatter (ANGLE mode) / roof_thickness. All three handles
anchor at the object's local origin and separate visually via their
declared axes (height/slope +Z, thickness -Z) — height + slope are
mutually exclusive via ``visibility_condition`` so they never paint
at the same time. Anchoring at the origin sidesteps the first-click
default-identity-matrix symptom that footprint-derived anchoring
would have hit on a stale ``RoofData`` cache.
* Lifecycle factory swap: explicit ``EnableEditingRoof / CancelEditingRoof
/ FinishEditingRoof`` classes replaced by ``tool.Parametric.build_edit_lifecycle("roof", _RoofEditMixin, ...)``.
Same bl_idnames out, no external caller changes.
* Registration: ``CycleRoofGenerationMethod`` + ``GizmoRoofEdition``
added to ``bim/module/model/__init__.py`` classes tuple.
* Tests: ``test_roof_gizmos.py`` covering slope round-trip, visibility
gates, cycle operator metadata, and origin-anchored positioning.
Bug fixes:
* ``generate_hipped_roof_bmesh`` flipped the bottom slab face's normal
at low slope angles. The kernel's outward-inference becomes
ambiguous on near-flat geometry once ``remove_doubles`` and
internal-face deletion run, and the early ``recalc_face_normals``
pass at line 389 ran BEFORE the topology was final. A second pass
on the final closed mesh fixes the eave plane (now reliably points
down regardless of slope).
* ``bpypolyskel.polygonize`` can emit a face whose vertex list
contains the same index twice on certain footprint/slope
combinations (a straight-skeleton ridge collapse). ``bm.faces.new``
rejects those with ``found the same (BMVert) used multiple times``,
aborting the whole rebuild. Filter the degenerate faces out so the
rest of the roof renders.
* ``_RoofEditMixin._restore_viewport_after_cancel`` now rebuilds the
bmesh from the just-restored draft via ``update_roof_modifier_bmesh``.
The hook was abstract on ``PathPreservingEditMixin`` and raised
``NotImplementedError`` on cancel-after-edit, leaving the user
stranded.
Also folds in a parallel ``tool/loader.py`` swap from
``tool.Blender.Modifier.is_railing`` to ``tool.Parametric.is_railing``
(consistent with the rest of the loader using ``tool.Parametric.*``).
Verified: headless smoke green, test_parametric_registry.py 8/8,
test_roof_gizmos.py 15/15. ruff + black clean on the touched files.
Generated with the assistance of an AI coding tool.
Three wall-gizmo fixes:
* GizmoWallFilletPreview crashed on every draw_prepare after the
DRY-colors refactor moved decoration lookups onto
self.get_decoration_colors() — that method lives on
BillboardingGizmoGroupMixin / BaseParametricGizmoGroup, but
GizmoWallFilletPreview inherited only from bpy.types.GizmoGroup.
setup() AttributeError'd silently, leaving radius_dim and friends
unset. Add the mixin to the bases; rename _position_gizmos to
position_gizmos so the mixin's refresh/draw_prepare dispatch lands
correctly and drop the now-redundant overrides.
* GizmoWallAddOpening's poll gated on the strict is_wall predicate,
which rejects fillet-corner walls (no LAYER2 usage by IFC spec).
Switch to is_path_connectable_wall on both the active and the
partner-exclusion checks so the add-opening icon surfaces over
curved corners — matching every other wall-state gizmo's host gate.
* Show / hide openings was only available on LAYER2 walls because
GizmoWallEdition's parametric edit pipeline (which carries the
toggle) refuses fillet bodies. Add GizmoWallFilletToggleOpenings,
a dedicated single-icon group that polls on is_fillet_corner_wall
and reuses bim.toggle_wall_openings — the body stays untouched.
Forward-compat AST guards in test_wall_gizmos_forward_compat.py pin
both invariants: every wall GizmoGroup that calls
self.get_decoration_colors() must inherit a mixin that provides it,
and GizmoWallAddOpening.poll must keep using the looser predicate.
Generated with the assistance of an AI coding tool.
tool.Parametric.refresh_post_commit was calling update_bim_tool_props
after every IFC mutation. The function does two things — refresh
read-only header values (extrusion_depth/length/x_angle) and re-target
user-intent enums (ifc_class, relating_type_id) from the active object.
Doing both on the commit path crashed on IfcAnnotation actives (the
type isn't in the bim_tool ifc_class enum) and silently overwrote the
user's "what to build next" choice on every other element.
Split the function: update_bim_tool_props remains selection-driven and
does both halves; new refresh_bim_tool_headers is header-only and is
what refresh_post_commit now calls. Behaviour on selection change is
preserved. Also ports the upstream PR #8136 try/except guard onto the
props.ifc_class write for the selection-driven path. Adds
test_handler_forward_compat.py to pin both contracts via AST.
Generated with the assistance of an AI coding tool.
NSZombieEnabled-lldb on macOS revealed the actual cause of the
"OS_os_log displayLock" / segfault that has been chasing us:
*** -[QMetalLayer displayLock]:
message sent to deallocated instance 0xb5e234e40
Qt's QMetalLayer (its CAMetalLayer subclass installed when surfaceType
== MetalSurface) is dealloc'd while Qt's QCocoaWindow still holds an
internal reference to it. Once wgpu-native bridge-retains the layer in
its Rust surface code and re-publishes the drawable pool from
configureSurface, Qt's QMetalLayer life is implicitly handed to
wgpu-native and Qt's separate ref winds up dangling. The next Qt
expose event sends -displayLock to the dead pointer.
Earlier guesses (Qt-6.11/macOS-26 incompatibility, multi-display) were
both wrong — single-display still crashed, and the Tahoe os_log
selector-cast theory was a red herring; the real error is "deallocated
instance", revealed only by NSZombieEnabled.
# Fix
Set surfaceType to OpenGLSurface on macOS too. On macOS that still
gives us a layer-backed NSView; Qt just doesn't install QMetalLayer.
WgpuMetalSurface_mac.mm's else-branch (the one that always fires when
the existing layer isn't already a CAMetalLayer) now consistently
attaches a vanilla CAMetalLayer we fully own — wgpu-native can do
whatever it wants to the layer's lifetime without stepping on any
Qt-side bookkeeping.
We never bind a real GL context on top of OpenGLSurface — it's just
the most portable "hardware-rendering-ready surface" hint Qt has, and
it's already what Linux and Windows use.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Two bugs in BIM_PT_array:
1. The "is this layer in edit mode" predicate compared a BoolProperty
against an int (props.is_editing == i). Python evaluates False == 0
as True, so layer 0 always rendered the per-layer edit form even
when no edit was active — clicking validate/cancel then dispatched
against a phantom edit state. Switched to
props.editing_item_index == i, which defaults to -1 and matches
exactly one layer when an edit is active.
2. The panel's CHECKMARK and CANCEL buttons called bim.edit_array /
bim.disable_editing_array, a parallel lifecycle that only cleared
editing_item_index. Entering edit mode via the viewport gizmo
(bim.enable_editing_array, the triad enter) sets is_editing=True
and hides array children; the legacy panel exit unwound neither —
so committing or cancelling from the panel left is_editing=True
with children hidden, and the viewport gizmo thought the edit was
still in progress. Re-bound both panel buttons to the canonical
triad operators (bim.finish_editing_array /
bim.cancel_editing_array), which _ArrayEditMixin already owns and
which the viewport gizmo group already uses. Panel and gizmo now
share one exit path.
The three now-unreachable operators are deleted with their
registration entries: EditArray (bim.edit_array), DisableEditingArray
(bim.disable_editing_array), and EnableEditingArrayItem
(bim.enable_editing_array_item, never called from any UI). The two
test/tool/test_model.py sites that drove bim.edit_array as a commit
step are switched to bim.finish_editing_array.
External scripts or user keymaps bound to bim.edit_array /
bim.disable_editing_array will need to update — the replacements are
bim.finish_editing_array and bim.cancel_editing_array, both taking no
parameters (the layer is read from props.editing_item_index).
Partly generated with the assistance of an AI coding tool.
The recent per-gizmo-prefs cleanup left ``update_swing_gizmos`` with a
stale ``prefs`` reference that raised NameError mid-refresh, so the flip
arc's ``matrix_basis`` was never reassigned and the gizmo drifted to the
world origin. SINGLE_SWING_RIGHT also lacked an X-mirror on the primary
arc, so the swing extended past the door's right edge instead of
sweeping back over the panel.
Five related fixes / additions:
* Drop the leftover ``prefs.decorations_colour[:3]`` per-frame colour
override (the setup-time ``decorator_color_special`` is the durable
contract — there's no reason to overwrite it every refresh).
* Add X-mirror to RIGHT-hinged single-panel transforms so the arc
sweeps back over the door rather than past the right edge.
* Treat DOUBLE_DOOR_SINGLE_SWING as a two-panel layout: 4 arcs total
(left + right panels, each with its own Y-mirrored flip) scaled to
``overall_width / 2``.
* Hide all swing arcs for SLIDING_TO_LEFT / SLIDING_TO_RIGHT /
DOUBLE_DOOR_SLIDING — sliding doors don't swing. A slide-direction
indicator is deferred to a separate change.
* Pin ``select_bias = -1000.0`` on every arc gizmo so the big
quarter-arc hit shapes don't steal clicks from the smaller dimension
and edit gizmos drawn on top.
Architectural cleanup driven by the same diff: the imperative
4-create + 50-line update block is replaced by a declarative
``swing_arc_props`` list of ``SwingArcConfig`` entries (mirrors the
existing ``dimension_gizmo_props`` pattern). Setup iterates the list
and creates one (main, flip) pair per entry under
``gizmo_swing_arc_<name>`` / ``gizmo_swing_arc_<name>_flip``; update
iterates the same list and positions each pair via the lambdas. Adding
a hypothetical multi-panel variant becomes a config entry rather than
two more attribute names plus a transform branch.
``ToggleDoorSwing`` gets a ``description`` classmethod that returns
user-facing wording per ``flip_geometry`` branch so the tooltip on
hover stops reading like operator internals.
``test/bim/module/model/test_door_gizmos.py`` (new) pins the
per-door-type contract: 11 cases covering LEFT / RIGHT hinge positions,
DOUBLE_SWING parity with SINGLE_SWING, DOUBLE_DOOR 4-arc layout, the
sliding-types hide invariant, ``is_editing=False`` hide invariant,
flip-arc matrix re-assignment, and world-matrix pre-multiplication.
Verified: ``pytest test/bim/module/model/test_door_gizmos.py`` 11/11
green; combined wall + stair + door gizmo lanes 37/37 green; ruff +
black clean on the three touched files.
Generated with the assistance of an AI coding tool.
Default WGPUPresentMode was a static Mailbox. That works on DX12 and
on Vulkan with most drivers, but the Metal backend in wgpu-native v29
only exposes [Fifo, Immediate], and asking for Mailbox makes
wgpuSurfaceConfigure panic from Rust:
thread '<unnamed>' panicked at src/lib.rs:605:5:
Error in wgpuSurfaceConfigure: Validation Error
Caused by:
Requested present mode Mailbox is not in the list of supported
present modes: [Fifo, Immediate]
fatal runtime error: failed to initiate panic, error 5, aborting
# Fix
Query wgpuSurfaceGetCapabilities and walk a preference list
(Mailbox → FifoRelaxed → Immediate → Fifo), picking the first mode
the surface actually lists. Fifo is the only spec-required mode and
will always be present, so the loop always finds something.
The env override (WGPU_PRESENT_MODE=...) still wins when set, but
also drops back to Fifo if the requested mode isn't supported on the
current backend — no panic.
# Outcomes per backend
DX12 / Vulkan: picks Mailbox (low input lag, our previous default).
Metal (macOS): picks Immediate (Mailbox unavailable). On Metal the
CAMetalLayer presents through CoreAnimation, so the
compositor still vsync-aligns; Immediate is effectively
low-latency-with-no-tearing on macOS.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
BonsaiViewer.app crashed at launch on a fresh macOS arm64 mac with:
Library not loaded: @rpath/libwgpu_native.dylib
Referenced from: /Applications/BonsaiViewer.app/Contents/MacOS/BonsaiViewer
Reason: no LC_RPATH's found
The exe had \`LC_LOAD_DYLIB @rpath/libwgpu_native.dylib\` (CMake baked
that in from the upstream dylib's install_name), but zero \`LC_RPATH\`
entries, so dyld had nowhere to look — and macdeployqt hadn't pulled
the dylib in either, since it sat at \`<install_root>/lib/\` rather
than inside the .app.
Two changes:
- \`src/ifcviewer-wgpu/CMakeLists.txt\`: on macOS, when
BUILD_BONSAIVIEWER is on, install libwgpu_native.dylib straight
into \`BonsaiViewer.app/Contents/Frameworks/\` instead of
\`<prefix>/lib/\`. That matches the standard macOS bundle layout.
- \`src/bonsaiviewer/CMakeLists.txt\`: set
\`INSTALL_RPATH "@executable_path/../Frameworks"\` on the
BonsaiViewer target on Apple. That's where dyld looks at launch,
and where the dylib now lives.
Together the @rpath load resolves at launch without depending on
macdeployqt to follow non-Qt @rpath references (it usually only
chases Qt frameworks).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
# Why
The stage-1 default was Fifo because it is the only present mode
WebGPU spec *requires* every backend to support — conservative,
"always works", lowest power. But on DX12 Fifo maps to a DXGI
flip-discard swap chain whose default \`MaximumFrameLatency\` is 3,
which queues ~50ms of pre-rendered work between submit and display.
Even when the FPS counter shows 60 the cursor-bound interactions
(marquee, pivot, orbit) feel ~3 frames behind because they *are*.
# What
Mailbox: vsync-aligned (no tearing) with a one-frame queue
(last-frame-wins). ~16ms input→display latency. wgpu-native handles
the fallback to Fifo automatically on backends that don't implement
Mailbox (Vulkan + NVIDIA on Linux is the historical one).
# Trade-off
Mailbox uncaps the render loop: with no vsync gating, the
\`requestUpdate()\` → render → present loop spins at whatever Qt's
event loop allows (~600 Hz on an empty scene), and the GPU does
useful-but-discarded work on each redundant frame. On any non-trivial
scene the GPU is the bottleneck and the loop self-paces near the
display rate. The FPS counter measures \`render()\` invocations, not
unique frames the user actually sees — that's expected.
# Override
WGPU_PRESENT_MODE=fifo restores the old behaviour for the rare
laptop-battery / heat-conscious case. fifo_relaxed / immediate also
still work as before.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
props.ifc_class is an EnumProperty whose items list only the element/space
types present in the model. Assigning element_type.is_a() crashed with
`enum "<class>" not found` when the selected element's type wasn't a member
(e.g. a raw IfcTypeProduct, or a stale item list mid-rebuild), aborting the
post-commit refresh.
Wrap the assignment in the same try/except TypeError guard already used for
the sibling relating_type_id assignments (added in 233cc344fa).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a clickable "xN" badge to GizmoStairEdition's edit row, mirroring
the array's popup-input UX: click opens a number dialog (no more
shift+click-into-modal). Text-only — no 2x2 grid glyph.
Structural changes that enable this cleanly:
* IconSlot.placeholder=True: slots reserve an X position in the row
without auto-creating a gizmo. Subclasses resolve the reserved X via
_slot_x_positions()[name] to place their own dynamic gizmos. Drops
the brittle "remember to add extra_gap_before" workaround that would
silently rot on slot reorders.
* Array bug fix: the count badge collided with the "-" icon because
the slot manager placed count_minus at the cycle position (X=0.87)
where ICON_NUMBER_X also lives. Migrating the badge to a placeholder
slot lets the manager allocate the X naturally and the "-" no longer
overlaps. ICON_NUMBER_X constant removed.
* IntegerInputDialogMixin in parametric_lifecycle.py: extracts the
popup-dialog plumbing shared between InputArrayCount and the new
InputStairTreads. Subclasses declare an IntProperty + attr_name +
props_getter; the mixin owns invoke/execute. _resolve_props helper
factors the common obj/props/requires_editing prologue.
Tests: BIM_GT_count_label registration; IconSlot placeholder contract
(no gizmo_idname required; gizmo_attrs() returns empty); the stair
edit-row slot layout reserves the label position between tread_lock
and plus at one ICON_ARRAY_GAP each; visibility propagates from
props.is_editing.
Partly generated with the assistance of an AI coding tool.
The narrow `<AppKit/NSView.h>` header only forward-declares NSWindow,
so `view.window.backingScaleFactor` fails to compile on macOS with:
error: property 'backingScaleFactor' cannot be found in forward
class object 'NSWindow'
Use the AppKit umbrella header so NSWindow's interface (including
backingScaleFactor) is in scope. Same idiomatic include any non-trivial
Cocoa code reaches for; we're already linking -framework Cocoa so
there's no compile-time cost beyond a slightly heavier translation
unit (the .mm is ~30 lines anyway).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Without this, BonsaiViewer.app launched on macOS would show a white
viewport for the same reason Windows did before d4d0c934b — the
`createSurface()` else branch fell through to "wgpu surface creation
not yet wired for this platform", `init()` returned false, and the
Qt window background was all the user saw.
Pieces:
- `WgpuMetalSurface_mac.{h,mm}`: tiny Objective-C++ bridge. Takes the
NSView pointer that Qt's `winId()` returns on macOS, attaches a
CAMetalLayer (using Qt's existing one when surfaceType is
MetalSurface, attaching one ourselves as a defensive fallback),
sets `contentsScale` from the window's backing scale factor so
retina drawables come out at native resolution, and returns the
layer as `void*`. The .mm keeps the Objective-C namespace pollution
out of WgpuViewportWindow.cpp.
- `WgpuViewportWindow` ctor: `setSurfaceType(QSurface::MetalSurface)`
on macOS so Qt backs the NSView with a CAMetalLayer at window
creation; OpenGLSurface elsewhere as before.
- `WgpuViewportWindow::createSurface()`: new `#elif defined(Q_OS_MAC)`
branch that fills a `WGPUSurfaceSourceMetalLayer` with the layer
pointer from the bridge and hands it to `wgpuInstanceCreateSurface`.
- `CMakeLists.txt`: `enable_language(OBJCXX)` + the .mm file added to
the source list on Apple, and links `-framework Cocoa` (NSView) +
`-framework QuartzCore` (CAMetalLayer).
`winId()` on macOS returns the backing NSView*, not the NSWindow* —
that's the layer-bearing host wgpu-native expects.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Without a Windows branch in `WgpuViewportWindow::createSurface()` we
were falling through to the `qWarning() << "wgpu surface creation not
yet wired for this platform"` else clause on Windows runs, causing
`init() -> createSurface()` to return false and the viewport to render
nothing (the user sees the Qt window's background fill — a white
viewport — and the log says "wgpu init failed; viewport will not
render in DebugView").
Add a `#elif defined(Q_OS_WIN)` branch that fills a
`WGPUSurfaceSourceWindowsHWND` chained-struct from
`GetModuleHandleW(nullptr)` (HINSTANCE) and `winId()` (HWND, as a Win32
window handle on the Qt Windows platform plugin), then passes it as
`surface_desc.nextInChain` to `wgpuInstanceCreateSurface`.
`<windows.h>` is pulled in inside the gated block with NOMINMAX and
WIN32_LEAN_AND_MEAN defined first so the preprocessor pollution
(`min`, `max`, etc.) doesn't leak into Eigen / `std::min`,`std::max`
elsewhere in the TU.
Note on the unrelated DXC log line the user also sees:
[wgpu err] DxcCreateInstance failed:
No such interface supported (0x80004002)
That is wgpu-native's DX12 backend probing for a modern
`dxcompiler.dll`. `E_NOINTERFACE` means a *too-old* dxcompiler.dll was
found on the system DLL search path (typical: a stale copy in
System32 / Visual Studio install). wgpu-native then falls back to its
Vulkan backend, so this log line is recoverable on its own — the
fatal failure was the missing surface branch above. If we hit shader
compilation issues after this lands, we can ship a known-good
dxcompiler.dll + dxil.dll alongside wgpu_native.dll separately.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The Windows branch of the wgpu-native FetchContent block was
hardcoding the x86_64 archive name regardless of host arch — the
macOS and Linux branches already switch on
\`CMAKE_SYSTEM_PROCESSOR MATCHES "arm64|aarch64"\`, but Windows
didn't get the same treatment because the wgpu work was done on
x86_64 hosts.
Result on \`windows-11-arm\`: CMake downloaded
\`wgpu-windows-x86_64-msvc-release.zip\`, IfcViewerWgpu linked against
the x86_64 import library, and the final link of
IfcViewerWgpuMinimal/BonsaiViewer emitted ~60 unresolved \`wgpu*\`
externs because the import-lib symbols are x86_64-only.
Upstream wgpu-native v29.0.0.0 already publishes
\`wgpu-windows-aarch64-msvc-release.zip\` — switching on
\`CMAKE_SYSTEM_PROCESSOR\` so the right archive gets fetched is
sufficient.
Also restores the ARM64 row in \`.github/workflows/build_win.yml\`
that the prior commit dropped (the comment there was wrong; upstream
does ship the binary, our CMake just wasn't asking for it).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The existing "Package .zip archives" step only sweeps
\`\$install_root/bin/\` for plain executable files (via \`find -type f
-perm /111\`). That captures \`IfcConvert\` and \`IfcGeomServer\` but
misses macOS app bundles entirely:
- BonsaiViewer.app installs at \`\$install_root/BonsaiViewer.app\`
(BUNDLE DESTINATION ".") — not under bin/, and it's a directory,
not a file.
So the prior bonsai macOS CI run got a green tick but the
ifcopenshell-builds S3 bucket only ended up with IfcConvert +
IfcGeomServer + the python wheel — no BonsaiViewer.
Add a second packaging pass that finds \`*.app\` directories at the
install-prefix root and zips each one as-is. macdeployqt has already
embedded the Qt frameworks inside the bundle during install/strip, so
no extra dependency staging is needed.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
# Why this exists
The bonsai macOS CI (`build_osx.yml`, arm64) currently fails the
`IfcOpenShell-Python` smoke test with:
ImportError: dlopen(.../_ifcopenshell_wrapper.cpython-311-darwin.so,
0x0002):
Library not loaded: @rpath/ifcopenshell.document.rdb.dylib
Reason: tried: '$ORIGIN/ifcopenshell.document.rdb.dylib'
(no such file)
`_ifcopenshell_wrapper.cpython-311-darwin.so` has a hard `LC_LOAD_DYLIB`
of `@rpath/ifcopenshell.document.rdb.dylib` and its only `LC_RPATH` is
`$ORIGIN` (= `site-packages/ifcopenshell/`). The plug-in dylib is not
present at that path on macOS, so the wrapper fails to load and the
build smoke test (`build-all.py: compile_python_wrapper`) errors out.
BonsaiViewer.app builds, installs, and macdeployqt-deploys cleanly
before this point — the failure is downstream and unrelated to wgpu,
BonsaiViewer, or anything else on this branch.
# Where the regression came from
Two commits on the branch line that became `ifcviewer-wgpu`:
b599ee10 "More work on isolating into plug-ins" (2026-04-18, Thomas Krijnen)
b022ca7e7 "Some plug-in work" (2026-04-21, Thomas Krijnen)
`b599ee10` added a hard link dep:
target_link_libraries(ifcopenshell_wrapper PRIVATE document_serializer_rdb)
which bakes `@rpath/ifcopenshell.document.rdb.dylib` into the wrapper's
`LC_LOAD_DYLIB`. `b022ca7e7` added a `if(CREATE_BUNDLE) ...
install(TARGETS ${_ifcopenshell_python_runtime_targets}
LIBRARY DESTINATION "${python_package_dir}/ifcopenshell" ...)` block
that was *intended* to satisfy that link dep by copying plug-ins next
to the wrapper. On macOS arm64 the install rule does not actually
deposit `ifcopenshell.document.rdb.dylib` into
`site-packages/ifcopenshell/`, so the runtime dlopen fails.
# Why 227d85d worked
`227d85d` (2026-05-15) is on the `v0.8.0` line, not on the
`datamodel-v1.0 -> ifcviewer -> ifcviewer-wgpu` line. The merge-base
of `227d85d` and `ifcviewer-wgpu` is `e6258ab4` (2026-04-13). Both
b599ee10 and b022ca7e7 live on the wgpu side of that fork and are not
ancestors of `227d85d`:
$ git merge-base --is-ancestor b599ee10227d85d
[exit 1 — NOT an ancestor]
$ git merge-base --is-ancestor b599ee10 v0.8.0
[exit 1 — NOT an ancestor]
So the macOS Python wheel built fine on `v0.8.0` because that branch
never had the plug-in refactor; it has been broken on our branch line
since 2026-04-21. Nobody noticed because nobody had been firing
`build_osx.yml` against this branch line until this week's bonsai CI
work.
# What this commit does
Adds an `IFCOS_BUILD_PYTHON_WRAPPER` env var to `nix/build-all.py`.
Defaulting to `on` preserves existing behaviour everywhere; setting
it to `off` (or `0`/`false`/`no`) drops `IfcOpenShell-Python` from
the target set so `build-all.py` skips the wrapper build + smoke
test entirely.
`build_osx.yml` sets `IFCOS_BUILD_PYTHON_WRAPPER=off` so the bonsai
macOS CI can complete and upload `BonsaiViewer.app` while the plug-in
install rule is broken.
# What Thomas should do
Once the install rule in `src/ifcwrap/CMakeLists.txt` (the
`if(CREATE_BUNDLE) ... install(TARGETS ${_ifcopenshell_python_runtime_targets}
LIBRARY DESTINATION "${python_package_dir}/ifcopenshell" ...)` block,
added in b022ca7e7) is fixed to actually drop
`ifcopenshell.document.rdb.dylib` next to the wrapper in
site-packages on macOS — this commit can be reverted in its entirety:
the env-gate in `build-all.py` AND the `IFCOS_BUILD_PYTHON_WRAPPER=off`
in `build_osx.yml`. The bonsai macOS workflow will then build the
Python wrapper too.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Linux (Rocky manylinux, GCC 11):
- WgpuAreaMeasurement.h: include <cstddef> directly. GCC 11 does not
transitively pull `size_t` through <vector>, so triangleCount()'s
return type fails to parse.
- WgpuViewportWindow.cpp:meshLocalToGlobal: use static_cast<double>(...)
instead of double(mesh_local[N]) when constructing the Eigen::Vector4d.
The latter triggers GCC 11's most-vexing-parse: it reads
`Vector4d local(double(mesh_local[0]), double(mesh_local[1]), ...)`
as a function declaration of `local` taking parameters
`double mesh_local[0]` etc., colliding with the outer `mesh_local`
parameter and failing with "redefinition of double* mesh_local".
macOS arm64:
- IfcViewerWgpuMinimal + BonsaiViewer install rules: change
`BUNDLE DESTINATION bin` → `BUNDLE DESTINATION .`. Qt's deploy
generator emits `macdeployqt <Target>.app` with no path prefix, which
only resolves when the bundle sits at the install-prefix root.
`BUNDLE DESTINATION bin` put it at `<prefix>/bin/Target.app` and the
install/strip step failed with "Could not find app bundle".
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Three related cleanups in one pass:
* **Per-gizmo preferences removed.** The ``visibility_pref`` field on
IconSlot, the ``prefs.gizmos.<feature>.<icon>`` PropertyGroups, and
the dispatcher that surfaced them in the addon preferences UI are
all gone. ``update_gizmo_visibility`` loses its ``pref_enabled``
parameter — visibility is now driven purely by editing state and
modal gating. bim/ui.py drops ~257 lines of dead PropertyGroup
definitions; bim/__init__.py and tool/parametric.py shed their
matching wiring; door / wall slot declarations stop referencing
the now-nonexistent prefs.
* **Dynamic-wall face normals fixed.** ``regenerate_wall_mesh_from_props``
in wall.py now calls ``bmesh.ops.recalc_face_normals`` before writing
the mesh. Without it, walls regenerated from the parametric edit
draft could ship with inward-facing normals on some faces, which
rendered as visual holes under any backface-cull or normal-aware
shading. ``test/bim/module/model/test_wall_preview_mesh.py`` pins
the invariant (every face's normal points away from the wall centre).
* **Color constants DRY.** ``COLOR_RED`` / ``COLOR_GREEN`` /
``COLOR_BLUE`` / ``COLOR_NEUTRAL`` now live at module scope in
gizmos.py; the BaseParametricGizmoGroup class attributes alias the
same tuples so ``self.COLOR_GREEN`` keeps working. IconSlot
declarations in stair.py (plus / minus) and array.py (count_minus /
count_plus / delete) now reference the named constants instead of
duplicating the RGB tuples inline.
Verified: headless smoke green at 1267 BIM_OT_ classes,
test_parametric_registry.py 8/8, wall lane 31/31 (includes the new
preview-mesh test). ruff + black clean on the touched files.
Generated with the assistance of an AI coding tool.
Hovering a wall-junction link-toggle icon today only swaps the icon
shape — the user doesn't see which wall the click will disconnect from
until after they click. ATPATH (T-junction) configurations especially
make the partner ambiguous when multiple connections sit close together.
On hover, paint a wireframe bbox around the partner wall using the same
shader, constants and color the array module already established for
its layer-children highlight (POLYLINE_UNIFORM_COLOR, decorator_color_special,
line width 1.8, alpha 0.8). The line-width / alpha constants in decorator.py
are renamed from _ARRAY_LAYER_BBOX_LINE_* to _BBOX_HIGHLIGHT_LINE_* and
shared between draw_array_layer_children_bbox and the new
draw_wall_partner_bbox so the two highlights stay in lockstep.
The trigger lives in a new GizmoWallLinkToggle subclass in wall.py
which keeps the base gizmos.GizmoLinkToggle generic (per the
generic-naming convention for shared widgets). The subclass's draw()
calls super().draw(context) then on self.is_highlight outlines its
partner_obj via the shared decorator helper. Same trigger pattern as
GizmoArrayLayerIndicator.
Blender's Gizmo API exposes target_set_operator but no symmetric
getter, so the partner reference can't be read back from the bound
operator handle. Instead GizmoWallUnjoinSingle.position_gizmos
mirrors the resolved partner_obj onto each visible icon every frame
next to the existing other_wall_guid write — the icon's draw() reads
from its own __slots__-declared attribute.
A forward-compat AST test pins the contract: GizmoWallLinkToggle.draw
must reference is_highlight and call draw_wall_partner_bbox. Catches
the regression where someone tidies the draw() override into super()
or replaces the shared helper with an ad-hoc draw call.
Generated with the assistance of an AI coding tool.
The parametric edit toolbar row used to assign each feature icon its
own ICON_<NAME>_X constant, with a separate FEATURE_ICON_MAX_X override
each subclass had to bump whenever a new icon was added. Forgetting the
bump silently collided icons — wall's rotate icon and the array button
both landed at X=1.24 in edit mode.
The new IconSlot dataclass + feature_slots tuple replace the
constants-and-override pattern with order-driven positioning: the
layout manager assigns each slot an X from its tuple index plus a
uniform ICON_ARRAY_GAP. Adding an icon is now a one-line append; the
"forget to bump" failure mode is structurally impossible.
Slot capabilities cover every existing icon-row shape:
* Single icon (wall rotate, array delete).
* N-variant slots — N gizmos at the same X with one visible per frame
via a subclass picker (stair tread-lock open/closed, wall baseline
exterior/center/interior). Pair becomes the N=2 case; triplet the
N=3 case. Variant idnames can be authored either as a tuple of
explicit names or as a string prefix that auto-suffixes _<variant>.
* Visibility prefs gate slot rendering without reflowing the row —
hidden slots still consume their X position.
* Extra per-slot gap before for visual separation (array's delete
trails the routine controls by an extra 0.2 m).
* Operator props forwarded to target_set_operator so adjusters
(+/-, increment) and generic toggles (property_name=...) work.
When the cycle slot is unused, feature slots collapse into the cycle
position so the row stays tight — that's how wall's baseline triplet
sits at X=0.87 without a gap before it.
Three subclasses migrate to the new system:
* wall.py — rotate icon + baseline triplet variants. Drops
ICON_ROTATE_X, _BASELINE_GIZMO_ATTRS, the manual triplet creation
loop, and the matching positioning block in _update_icon_row_extras
(it now just picks variant visibility).
* stair.py — tread_lock pair (open/closed) + plus + minus.
_update_editing_icon_positions reads slot X via _slot_x_positions
instead of three hardcoded constants. Also fixes the standalone
total_length_lock gizmo, which was broken since PR4 split
VIEW3D_GT_lock into open/closed pair (caller wasn't updated).
* array.py — count_minus + count_plus + method + delete (with
extra_gap_before=0.20 to separate the destructive action).
Drops the manual edit-row positioning loop entirely; the base
loop handles it. GizmoArrayChild now inherits BillboardingGizmoGroupMixin
and uses the shared setup_icon_gizmo helper, dropping its
duplicated _make_icon wrapper.
Two helpers added on BillboardingGizmoGroupMixin to fold the duplicated
prefs/color preamble that appeared at the top of six wall gizmo setups
plus the array-child setup:
* get_decoration_colors() — (decorations_colour, decorator_color_selected),
the active-state pair.
* get_unselected_decoration_colors() — (decorator_color_unselected,
decorator_color_selected) for gizmos surfaced on already-selected
geometry that should not pull focus.
Verified: headless smoke green at 1267 BIM_OT_ classes,
test_parametric_registry.py 8/8 pass, wall lane 29/29 pass,
model lane unchanged at 135 pass + 7 pre-existing v0.8.0 failures
(no regressions). ruff + black clean.
Generated with the assistance of an AI coding tool.
Bundles four portability fixes uncovered by manually firing the platform
workflows against this branch:
- WgpuOverlayRenderer.cpp: GCC 11 (Rocky manylinux runner) does not parse
a multi-line raw string inside `#define`. Converted
THICK_LINE_HELPERS_WGSL from a `#define` to a `static const char*` and
switched AXIS_WGSL / SECTION_WGSL / MARQUEE_WGSL to `std::string` so
they can concatenate at static-init time. Three call sites now pass
`.c_str()` to svFromCStr.
- bonsaiviewer/CMakeLists.txt: added BUNDLE DESTINATION to the install
rule (same fix already applied to IfcViewerWgpuMinimal). MACOSX_BUNDLE
targets fail at configure on macOS without it even when nobody runs
`make install`.
- build_osx.yml: dropped the x64 (Intel cross-compile) matrix row. The
runner is arm64 so `brew --prefix qt` returns the arm64 prefix; we'd
need a separate x86_64 Qt install under /usr/local to cross-build
BonsaiViewer. Revisit if Intel-Mac demand resurfaces.
- build_win.yml: dropped the ARM64 matrix row. wgpu-native does not ship
a Windows-ARM64 binary, so IfcViewerWgpu's link step fails with ~60
unresolved wgpu* externs. Re-enable when upstream publishes that
target.
Cherry-pick this commit to v0.8.0 so the workflow_dispatch buttons see
the dropped rows.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The previous single-wall unjoin gizmo used a bracket-pair icon
(VIEW3D_GT_unjoin) that reads as "unjoin" only after you know what
it is, with no clear "linked" inverse — closing the brackets to
suggest the connected state collapses to a hollow square that
doesn't read as a link at all.
Add GizmoLinkToggle (VIEW3D_GT_link_toggle): two filled dots joined
by a horizontal connector in the default state. On hover the two
halves shear vertically apart — left dot+stub slip down as a unit,
right dot+stub slip up — with a horizontal gap at the centre,
signalling that a click will sever the underlying connection. The
glyph lives next to the generic icon classes (GizmoLockOpen/Closed,
GizmoArc) so any path / link / pair-of-connected-items context can
reuse it; it isn't wall-specific despite the first caller.
The class keeps its own per-state GPUBatch cache so the shape swap
on hover doesn't allocate per frame. The hit-shape is sourced from
the broken form (the larger bbox of the two states) so the cursor
doesn't lose hover at the offset dots' outer edges and flicker
between states.
GizmoWallUnjoinSingle.setup() now requests VIEW3D_GT_link_toggle.
The operator binding (bim.unjoin_wall_path_connection), the
POOL_SIZE, and the per-frame partner-GUID write are unchanged.
Generated with the assistance of an AI coding tool.
When the two source walls were placed at different elevations, the
fillet corner wall ended up with sub-degree X and Y Euler rotations
even though both source walls had only a Z rotation.
Cause: _apply_fillet_corner_geometry derived the corner's local X
axis from `chord = tangent_b - tangent_a` (a 3D vector). With walls
at different Z, `chord.z` was non-zero, so `x_dir = chord.normalized()`
inherited that Z component. The Z axis was already hardcoded to world
Z, so x_dir and z_dir were no longer orthogonal — the resulting
matrix_world was non-orthonormal, and Blender's Euler decomposition
surfaced the skew as the visible X/Y rotation drift.
Project the chord to the XY plane before normalising so x_dir is
strictly XY-aligned and orthogonal to z_dir. The corner wall is now
placed at wall A's elevation with a pure Z rotation, which matches
the user's expectation when both inputs are Z-aligned regardless of
their relative elevation.
Generated with the assistance of an AI coding tool.
When validate_type selected a preferred_item from remaining_items
(e.g. the sole IfcBooleanResult in a representation), it left that
item in the list. The subsequent Items filter removed every item,
leaving Items=[] and causing guess_type to return
"MappedRepresentation" — silently corrupting the representation.
Also removes temporary debug print statements added during
investigation of the wall-to-slab extension workflow.
Generated with the assistance of an AI coding tool.
Re-running the operator on the same wall/slab pair created
additional IfcPolygonalFaceSet booleans each time. Now each
wall's existing booleans are removed before re-clipping, and
previously connected slabs are merged with the new selection
so no earlier clips are silently discarded.
Generated with the assistance of an AI coding tool.
When Shift+G is pressed on a LAYER3 element, any LAYER2 walls
connected via IfcRelConnectsElements(TOP) are now re-clipped
to the slab's updated geometry after recalculate_slab runs.
Generated with the assistance of an AI coding tool.
extend_walls_to_underside now accepts multiple slab/roof
objects in a single operation — all selected non-LAYER2 IFC
elements are treated as clip targets, all LAYER2 elements as
walls. Placement sync is done once upfront; each wall is then
clipped against every selected slab before reloading.
Also adds bim.regenerate_wall_to_underside (Shift+G): after
moving a slab, re-clips connected walls using the existing
IfcRelConnectsElements(TOP) relationship. Old booleans are
removed via remove_representation_item before re-clipping.
Generated with the assistance of an AI coding tool.
When extend_walls_to_underside is applied to a wall and the
roof/slab is later moved, pressing Shift+G now re-clips the
wall to the slab's new position.
The IFC relationship created by connect_wall_to_slab
(IfcRelConnectsElements, Description="TOP") is used to look
up which slabs a wall is clipped to. On regeneration, the
existing manual booleans (IfcPolygonalFaceSet operands) are
cleanly removed via remove_representation_item, then
clip_wall_to_slab is re-applied for each connected slab.
Shift+G on a LAYER2 wall that has a TOP connection now calls
bim.regenerate_wall_to_underside; walls without a connection
continue to call bim.recalculate_wall as before.
Generated with the assistance of an AI coding tool.
When the operator was called twice on the same wall for a
ridge roof, the two IfcPolygonalFaceSet clip solids shared
an exact ridge edge (kissing-solid). OCCT produced spurious
extra vertices at the coincident boundary.
Fix by building the clip solid from a rectangle on the slope
plane that extends slightly past the face edge (1 project
unit margin) rather than the exact face footprint. Adjacent
slope solids now volumetrically overlap at the ridge instead
of sharing a boundary face, which OCCT handles correctly.
Generated with the assistance of an AI coding tool.
GizmoWallUnjoinSingle.poll accepts fillet-corner walls via the looser
tool.Parametric.is_path_connectable_wall predicate (fillet corners
have no LAYER2 usage by IFC spec, but they still participate in
IfcRelConnectsPathElements). The partner filter inside
_iter_path_connections used the stricter tool.Blender.Modifier.is_wall
(LAYER2-only), so adjacent LAYER2 walls silently dropped their
fillet-corner partners from the connection list — the unjoin icon
appeared when the fillet wall itself was selected but not on either
of its LAYER2 neighbours.
Switch the partner filter to is_path_connectable_wall so host and
partner predicates match. Add a regression test for the fillet case
and an AST forward-compat guard pinning the predicate symbol so a
future "tidy the imports" can't silently re-introduce the asymmetry.
Generated with the assistance of an AI coding tool.
Ports the array parametric-edit lifecycle, gizmo group, child guard,
per-layer ARRAY entry icons, and the array bbox decorators
(preview + selection highlight + layer-children) from gizmos-8088.
Restores the array_gizmo icon's positioning + visibility in the
framework's parametric edit row.
Registry (tool/parametric.py):
* EDIT_TYPES adds ParametricObject("array", supports_build_edit_lifecycle=True).
_ArrayEditMixin in array.py feeds build_edit_lifecycle which auto-
generates EnableEditingArray / FinishEditingArray / CancelEditingArray
with the conventional bl_idnames the gizmo references.
tool/blender.py:
* Adds is_array predicate wrapper around tool.Parametric.is_array.
The registry contract test test_every_entry_has_modifier_predicate
enforces every EDIT_TYPES entry has a matching is_<name> wrapper on
tool.Blender.Modifier.
array.py (+1130 LOC port from gizmos-8088):
* _ArrayEditMixin(ParametricEditMixinBase) drives the auto-generated
enable / finish / cancel lifecycle.
* GizmoArrayEdition: validate + cancel + count display + +/- adjusters
+ method toggle + delete button + per-layer ARRAY entry icons
(preallocated pool of MAX_LAYER_GIZMOS=8).
* GizmoArrayChild: child-array gizmo for the array-replica case.
* EditArrayFromChild: resolves the spawning layer via
tool.Array.get_child_layer_index so clicking a child's array gizmo
opens the layer that produced that child rather than always layer 0
(the gizmos-8088 source itself hardcoded item=0; HEAD has the helper
to do it right).
* New operators: EnableEditingArrayItem, ArrayParentGizmoClick,
ArrayGizmoClick, ToggleArrayMethod, RemoveArrayLayerFromEdit,
InputArrayCount, AdjustArrayCount.
prop.py: BIMArrayProperties gets per_child_opening BoolProperty
(when the array parent fills a host, give each child its own
opening + filling pair).
Bug fix: guard update_relating_array_from_object against the
cleanup-time None set. _finish_one writes relating_array_object = None
to clear the source-array reference; that fired the update callback,
which dispatched bpy.ops.bim.enable_editing_array(item=self.is_editing).
With is_editing just flipped to False, the bool coerced to 0 and
re-opened layer-0 edit immediately after every validate. The guard
short-circuits on None; item is also fixed to 0 (the bool-as-layer-
index was always meaningless for the legitimate user-pick path).
decorator.py (+312 LOC, all ports from gizmos-8088):
* bbox_world_edges / draw_polyline_segments / _BBOX_EDGES - shared
geometry helpers usable across array decorators.
* draw_array_layer_children_bbox - green wireframe bbox per child of
one array layer, drawn inline from a gizmo's draw() so the highlight
tracks the hover cursor without POST_VIEW lag.
* ArrayPreviewDecorator - faint cyan ghost bboxes at each future
array instance during the edit lifecycle (offset math mirrors
Model.regenerate_array, gated on props.is_editing).
* ArraySelectionHighlightDecorator - bounding-box overlay surfacing
the array family of the selected object. Child selected -> parent
in special color + siblings in unselected color; parent selected
(idle) -> all children in unselected color. TokenCache-backed.
handler.py: imports + uninstall/install the 2 always-on decorators in
_install_viewport_overlays. Both self-poll, so installation has no
cost when no array is selected / in edit mode.
Registration (bim/module/model/__init__.py):
* Adds the 3 lifecycle classes generated by build_edit_lifecycle
(CancelEditingArray, EnableEditingArray, FinishEditingArray) -
they exist as module-level names but are only visible to Blender's
operator registry when included in the classes tuple.
* Adds the 8 new operators + 2 new gizmo groups in alphabetical order.
gizmos.py: restores the array_gizmo icon position + visibility block
in BaseParametricGizmoGroup.update_editing_gizmos. Was force-hidden
in c250b2c1a because no array gizmo existed; the icon's plumbing
comes back online now that GizmoArrayEdition is registered.
Verified by test/bim/test_parametric_registry.py: all 8 tests pass -
enable/finish/cancel ops resolve, PropertyGroup attached, is_array
predicate present, predicate is total on non-matching elements.
Generated with the assistance of an AI coding tool.
ifc.get_object(element) can return None for IFC elements that aren't loaded as Blender objects (e.g., decomposed sub-elements).
The loop now skips those instead of passing None into collector.assign().
Cheers!
The temporary-offset workaround (#7408, commit bd57cc8735) subtracts the
directrix centroid (`mean`) from the curve points before building the
sweep near the origin, then must add it back to restore the original
location. The restore negated the sign — `Move(-mean)` instead of
`Move(+mean)` — placing the swept solid at -mean (mirrored through the
origin) rather than its true position.
Only triggers for polyline directrixes (`is_polyhedron()`) whose centroid
is more than 100 m from the origin (`mean.norm() > 1e2`), so models
centered near the origin are unaffected. Models that keep absolute site
coordinates (e.g. many Revit/ODA IFC exports) render affected swept
solids — reinforcing bars, pipes — at a mirrored phantom location far
from the rest of the model.
PR3 shipped tool.Wall.path_connection_location_world; the local
_path_connection_location_world added in PR4 commit 70845e4dd
duplicated the same logic. The only caller in wall.py already uses
the tool method (line 3687 area), so the local helper has been
dead code since the migration in 7e5e7b8d6 routed _get_wall_geom_cached
to tool.Wall.read_geometry. Drop it.
Generated with the assistance of an AI coding tool.
Pre-existing architectural smell on v0.8.0: core/root.py.copy_class
called a module-level _has_material_styles helper that did
ifcopenshell.util.element.get_materials() directly, bypassing the
Prophecy mock seam that every other branch in copy_class flowed
through. Symptom: test/core/test_root.py::TestCopyClass::
test_AAAAAAAAAAAA passed mock strings into copy_class, the helper
called .is_a() on the string, AttributeError.
Move the check to tool.Root.has_material_styles (paired with
assign_body_styles — they're called in sequence as "is there a
material style? if not, assign body style"). core/root.py now
calls root.has_material_styles(new) like every other dependency,
fixing the test failure and dropping the ifcopenshell.util.element
import that was the only consumer of the ifcopenshell import at
module load in core/root.py.
* core/tool.py: add abstract has_material_styles to Root interface.
* tool/root.py: add concrete classmethod near assign_body_styles.
* core/root.py: replace _has_material_styles helper call site with
root.has_material_styles; drop the local helper and its import.
* test/core/test_root.py: add the new mock expectation
root.has_material_styles("element").will_return(False) before the
existing assign_body_styles expectation.
Generated with the assistance of an AI coding tool.
Linux (build_rocky.yml) and Windows (win/build-all-win.py) already pass
BUILD_BONSAIVIEWER=ON when building from source; macOS was the odd one
out. Three small changes to bring it up to parity:
1. .github/workflows/build_osx.yml — \`brew install qt\` (Qt6 with Svg)
in the Install Dependencies step, then set QT_DIR=\$(brew --prefix
qt) and BUILD_BONSAIVIEWER=ON in the Run Build Script env.
2. nix/build-all.py — install_qt6() now honours a pre-set QT_DIR.
Before this change get_qt6_aqt_config() raised on non-Linux,
blocking bonsai builds on macOS/Windows from ever using a
system-provided Qt6. We now validate that QT_DIR points at a real
Qt6 install (probes lib/cmake/Qt6/Qt6Config.cmake) and skip the aqt
download path if so. Linux flow is unchanged: when QT_DIR is unset
the function falls through to the existing aqtinstall path.
build_osx.yml stays workflow_dispatch-only — slow run (~1h with
ccache, longer cold), so manual fire when wanted. Cherry-pick this
file + nix/build-all.py to v0.8.0 to make the workflow dispatchable
against the ifcviewer-wgpu branch before the squash lands.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Ninja rejects \`-j\` without a numeric argument (unlike make, which
treats bare \`-j\` as unlimited parallelism). Build step exited with
\`ninja: fatal: invalid -j parameter\`. Drop the \`-- -j\` tail
entirely; Ninja already parallelises across available cores by
default.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The first run after BUNDLE DESTINATION fix didn't trigger because that
commit only touched src/ifcviewer-wgpu-minimal/CMakeLists.txt, which
the workflow's paths: list didn't cover. Add the two sibling
subprojects (-minimal and wgpu-mem-probe) since they participate in
the same configure pass.
(Manual workflow_dispatch is still unavailable until this workflow
lands on the default branch — GitHub gates the "Run workflow" button
on the default branch's copy of the file.)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The target has MACOSX_BUNDLE ON, which on Darwin requires
install(TARGETS) to specify a BUNDLE DESTINATION — CMake validates
that at configure time, even when nobody runs `make install`. The
macOS CI configure step failed with:
install TARGETS given no BUNDLE DESTINATION for MACOSX_BUNDLE
executable target "IfcViewerWgpuMinimal".
Set BUNDLE DESTINATION bin alongside RUNTIME DESTINATION bin so both
platforms install into the same spot. No behavioural change on Linux
(no MACOSX_BUNDLE) — verified locally.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Top-level cmake/CMakeLists.txt:328 does an unconditional
find_package(Boost REQUIRED COMPONENTS program_options regex thread
date_time iostreams) before the BUILD_BONSAIVIEWER gate, so we have
to install it on the runner even though IfcViewerWgpu itself doesn't
touch Boost. Removed via task #12 (extract ifcviewer-core) later.
Other unconditional finds in the top-level CMake (manifold,
nlohmann_json, USD, RocksDB, zstd) are already gated on flags that
default to OFF — no action needed for those.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Two iterations on the first CI run:
1. The top-level cmake/CMakeLists.txt unconditionally find_package's
CGAL (line 217) and OpenCASCADE (222) before our BUILD_BONSAIVIEWER
gate kicks in, so configure failed with "Could NOT find CGAL". Pass
BUILD_IFCGEOM=OFF + BUILD_IFCPYTHON=OFF + BUILD_CONVERT=OFF +
BUILD_EXAMPLES=OFF + BUILD_GEOMSERVER=OFF + WITH_OPENCASCADE=OFF +
WITH_CGAL=OFF + COLLADA_SUPPORT=OFF so all the heavy deps stay out
of the configure step. Verified locally on Linux.
2. The ctest -R "wgpu" filter was case-sensitive and the Catch2 test
names begin with capital "Wgpu" (e.g. "WgpuSelectionState starts
empty..."), so it matched zero tests and ctest exited with "No
tests were found". The standalone wgpu config only builds the
wgpu tests anyway, so the filter is unnecessary — drop it.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Lightweight workflow that compiles IfcViewerWgpu (the static lib) on
macOS arm64 + runs the wgpu state tests. Skips IfcViewerWgpuMinimal
because createSurface has no Metal path yet (task #32); the platform
surface blocks in WgpuViewportWindow.cpp are wrapped in
#if defined(Q_OS_LINUX) so the lib itself compiles cleanly on macOS.
Goal: catch portability regressions in the wgpu source on Apple
Silicon without paying for build_osx.yml's full IfcGeom + OCCT +
Python wheel pipeline. ~5 min vs hours.
Triggers on push/PR that touches src/ifcviewer-wgpu, the shared
headers it depends on, the top-level CMake, or this workflow itself.
Also workflow_dispatch for manual runs.
Hoists the Catch2 fetch in cmake/CMakeLists.txt out of the
BUILD_BONSAIVIEWER gate so the standalone wgpu config
(BUILD_BONSAIVIEWER=OFF + BUILD_BONSAIVIEWER_WGPU=ON +
BUILD_BONSAIVIEWER_TESTS=ON) can build tests without dragging the
whole bonsai/IfcGeom tree in. Default remains OFF, so default builds
stay offline-capable.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
LodBuilder.cpp guards its real body behind #ifdef WITH_MESH_OPTIMIZER
(the stub is `return;`). The IfcViewer static lib propagates the
define via target_compile_definitions, but test_lod_builder compiles
LodBuilder.cpp standalone (it doesn't link IfcViewer), so the test
silently exercised the no-op path. summariseLods and buildLods cases
asserted on the post-build state and saw zero LOD1 output.
Pre-existing regression since 884e7ba32 ("Make meshoptim optional");
adds the define to the test target directly so the real build path
runs. 5/5 LOD cases pass after.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Two Tier-1 unit binaries under src/ifcviewer-wgpu/tests/ — same Catch2
+ CTest harness as the surviving GL-side tests, gated by
BUILD_BONSAIVIEWER_TESTS.
- test_wgpu_selection: 17 cases / 71 assertions covering replace, add,
remove, toggle, clear, contains, count, selectionIds, fillFlagsArray,
active-id semantics, dirty-bit, and id == 0 sentinel handling.
- test_wgpu_visibility: 8 cases / 27 assertions covering hide, show,
clear, isHidden, hiddenIds, idempotence, and the 0 sentinel.
The wgpu state classes have a deliberately simpler shape than the GL
ones (no Q_OBJECT, no signals — replaced by a dirty bit; no bulk
set/add/remove methods — bulk behaviour lives in the viewport verbs).
One *intentional* behavioural difference is documented in the test:
add(id) steals active in the wgpu API, where GL's addToSelection kept
the prior active. Each pick should drive the properties panel to the
most recently touched object.
Bulk hide/isolate/show-all semantics live in WgpuViewportWindow, which
composes WgpuVisibilityState + the model instance lists; those are
integration-level, not Tier-1, so they're not covered here.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Bonsai now drives the wgpu viewport for both sidecar and direct-IFC
loads. The GL viewer and its supporting state classes are gone.
SceneLoader rewire:
- Takes WgpuViewportWindow* instead of ViewportWindow*.
- Sidecar path reads metadata only (readSidecarMetadataOnly) and hands
the StreamingSidecar off to the new applyCachedModel. Field accesses
inside applySidecarData go through .meta.
- Direct-IFC path uses the wgpu A-path (upload{Mesh,Instance}Chunk +
finalizeModel). The applyLodExtension call is dropped — wgpu has no
live LOD1 splice; LOD1 still lands in the on-disk sidecar for the
next open.
Bonsai migration:
- ViewportWindow → WgpuViewportWindow across MainWindow, Measurement,
SessionState, and every modules/*/{Commands,Panel,View}.{h,cpp} —
116 sites total. Same s/OverlayRenderer::/WgpuOverlayRenderer::/
rename, 12 sites.
- Includes flipped from ../ifcviewer/ViewportWindow.h to
../ifcviewer-wgpu/WgpuViewportWindow.h. OverlayRenderer.h include
dropped (transitively reached via the viewport header).
- BonsaiViewer links IfcViewerWgpu in addition to IfcViewer for the
duration of the migration; the GL-side IfcViewer also publicly links
IfcViewerWgpu so SceneLoader can resolve WgpuViewportWindow.
GL backend deletion:
- src/ifcviewer/ViewportWindow.{cpp,h}, BvhAccel.*, OverlayRenderer.*,
Selection.*, Visibility.* all gone.
- src/ifcviewer-minimal/ removed entirely (MinimalWindow drove the GL
viewport).
- src/ifcviewer/tests: test_bvh_accel, test_selection, test_visibility
removed. The first has no replacement (wgpu doesn't use a per-instance
BVH); the latter two are ported separately. test_lod_builder,
test_sidecar_cache, test_instanced_geometry, test_federation remain
(backend-agnostic).
- IfcViewer's CMakeLists drops OpenGL, Qt::OpenGL, Qt::Widgets — none
of the surviving translation units reach for them.
Build flag plumbing:
- BUILD_BONSAIVIEWER now auto-enables BUILD_BONSAIVIEWER_WGPU since
SceneLoader requires the wgpu lib for its WgpuViewportWindow* arg.
- The wgpu subprojects add_subdirectory ahead of the GL one so
IfcViewerWgpu exists when IfcViewer's link evaluates.
- src/ifcviewer-minimal subdir reference removed from cmake/CMakeLists.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Make the wgpu viewport ready for bonsai's verb actions, federation
refresh, and tool routing — i.e. callable from an outside host, not
just from the minimal viewer's own hotkeys.
Surface additions on WgpuViewportWindow:
- Qt signals: objectPicked, frameStatsUpdated, surfacePickedInTool,
toolModeChanged, toolBackspacePressed.
- FrameStats struct + rolling 60-sample frame-time window for the
fps field; emit at end of render() so external listeners see fresh
numbers in the same tick.
- InstanceLookup struct + findInstance(object_id, ...) const for the
measurement tools' O(1) object → (model, mesh, placement) resolve.
- Federation hooks (setFederatedFalseOrigin / setModelCoordinateOperation
/ setModelTransformation) + per-model coordinate_operation_meters /
model_transformation_meters fields on WgpuModelGpuData. Implement
composeInstanceFromPlacement + recomposeAndUploadModel so each setter
actually applies — model recompose runs in double, casts to float for
the GPU upload, and refreshes per-chunk world AABBs. meshLocalToGlobal
now composes coordinate_operation · placement properly.
- showModel / hideModel for per-model visibility, plus element-level
verbs (hideSelectedElements / isolateSelectedElements / showAllElements
/ invertElementVisibility) and setSelectedObjectId / cameraState() /
projectionOrtho() / toggle{Area,Length,Volume}Tool wrappers.
- Section-cutting methods (toggleSectionTool / clearSectionPlanes /
sectionToolActive) moved to public so bonsai's Commands.cpp can call.
- QVector3D overload of computeObjectAabb to match the GL signature.
- ToolMode::None → ToolMode::NoTool (X11 macro collision avoidance).
Direct-IFC ingestion (A-path), mirrors the GL streaming push API:
- uploadMeshChunk / uploadInstanceChunk stage into pending_direct_loads_
using the same vertex quantisation as SidecarBuilder so direct-load
and sidecar-load produce byte-identical buffers.
- finalizeModel wraps the staged data in a file-less StreamingSidecar,
routes through the existing applyCachedModel chunk planner, then
gathers per-chunk vertex+index bytes from memory and feeds
applyStreamedChunk synchronously. Every chunk lands is_resident=true
immediately (no disk I/O to defer).
Streaming collapse:
- Delete the applyCachedModel(SidecarData) full-load path entirely.
- Rename applyCachedModelStreaming → applyCachedModel; loadSidecar
always uses the metadata-only reader. Drop the --streaming CLI flag
from IfcViewerWgpuMinimal and the streaming_enabled_ field.
WgpuSelectionState::ids() → selectionIds() so bonsai's
`viewport_->selection().selectionIds()` compiles unchanged.
Eigen3 added as a public dep of IfcViewerWgpu for the federation
matrices.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Three test files covering PR4's new surfaces — preview registry,
wall-gizmo poll behaviour, fillet operator registration. Every test
walks the live registry or class hierarchy instead of hard-coding
preview keys, operator names, or helper function names, so adding a
new preview / wall gizmo group / fillet operator exercises the same
invariants without test edits.
test_preview_base.py (6 tests):
* RegistryContract: every PREVIEW_CANCEL_OPS entry resolves to a
callable cancel operator on bpy.ops.bim.
* GetPreviewPropsTolerance: get_preview_props returns None for
contexts without a scene (regression guard for the SimpleNamespace
bug fixed in commit ee63137c6).
* ActivationCycle (registry-driven loop): any_preview_active toggles
with each registered preview's is_active flag;
discard_pending_previews clears every active flag across every
registered preview.
* SaveOnDiscardWired: locates the bim.save_project operator
dynamically and verifies its execute path references the discard
helper by its actual __name__.
test_wall_gizmo_poll_gate.py (4 tests):
* WallGizmoGroupsHideDuringPreview: walks the wall module for
bpy.types.GizmoGroup subclasses (skips preview-owner exceptions
whose bl_idname contains 'preview'), mocks any_preview_active to
True, and asserts every discovered gizmo's poll returns False.
* BaseParametricGizmoPollHidesDuringPreview: mirrors the test for
the cross-feature parametric framework base class.
test_fillet_operators.py (3 tests):
* FilletOperatorsRegistered: at-least-four-ops + every-discovered-op-
is-callable. Catches accidental deregistration.
* EnableRejectsIneligibleSelection: poll returns False without a
selection so the operator is greyed-out in menus.
State-clearing tests via bpy.ops.bim.cancel_wall_fillet_preview() are
deliberately omitted — the operator early-returns when context.screen
is unattached and prior tests in the model lane can leave the screen
in that state, making the dispatch path inherently flaky. Live testing
covers the behaviour.
Net: 13 tests pass cleanly in both single-file and full model lane.
Generated with the assistance of an AI coding tool.
Fixes two coupled streaming pathologies on working-set > pool scenes:
1) The candidate loop used `break` when a candidate couldn't fit even
after eviction. Comment justified it with "sorted by priority, lower
candidates can't beat it either" — true for *priority* eviction, but
the failure is *size-based fitting*. A 31 MB candidate that doesn't
fit in 24 MB largest-free was starving the entire per-frame budget,
including smaller candidates that would have fit happily. Replaced
with `continue`.
2) Same blocked candidate re-entered the candidate list every frame
forever, spamming `[blocked]` and (worse, on web) paying for the same
byte-range fetch over and over when apply-time OOM happened. Added
`blocked_cooldown_until_frame_idx` on the chunk: when OOM strikes at
enqueue *or* apply, the chunk is skipped from candidate gathering for
~3s. Web-friendly cap of one wasted fetch per 3s per chronic chunk
instead of per-frame. Cooldown expires naturally; if the pool layout
changes within the window (other chunks evicted, fragmentation
coalesces) the chunk re-enters automatically.
Also added eviction-attribution + chunk thrash detection (gated behind
WGPU_STREAM_EVICT_LOG=1) to confirm A→B→A 2-cycles vs simple
sacrificial-victim cycles. Quietened the steady-state stream debug
dump — moved the verbose multi-line "missing/resident/bottom" snapshot
behind WGPU_STREAM_DEEP_DEBUG=1 with a wider 300-frame interval, and
added a single-line `[stream]` health summary every ~5s in interactive
mode. Removed the hardcoded one-off "brace.ifc bracing all
chunks" dump that was investigation scaffolding for a now-closed bug.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Ports Bonsai's LengthMeasurement onto WgpuViewportWindow as a new
WgpuLengthMeasurement class. Each LMB appends a world-space pick point
and the readout adapts to the running count:
1 pt → laser-measure: coplanar-patch BFS on the click's surface
projects every patch vertex into the surface's own tangent
basis to get face extents (X/Y/Z bars dashed in world space),
plus ENH coords for the picked point, plus a vertical
raycast for floor/ceiling distance on horizontal surfaces.
2 pts → distance A→B + axis-coloured ΔX/ΔY/ΔZ stair-step + dashed
perpendicular projection when both picks landed on
near-parallel surfaces.
3 pts → angle at middle vertex + triangle area + perimeter.
4+ pts → polygon area via best-fit-plane shoelace (Jacobi-3x3
eigendecomp inline; no Eigen dep) or fan-triangulated
fallback for non-planar loops, plus closed-loop perimeter.
Backspace / Del removes the last point; Esc / L again exits.
Dependencies layered in:
- pickMeshLocalAt now refines the AABB-coarse pickSurfaceAt hit into a
real triangle hit via Möller-Trumbore against the picked instance's
CPU mesh shadow. Without this the BFS seeds with whatever triangle
is closest to the bounding-box corner — producing patches and
extents shaped like the AABB instead of the surface.
- meshLocalToGlobal: applies the instance's placement_transformation
only (no per-model CoordinateOperation in wgpu yet). ENH equals
IFC-world for non-federated loads, which is what the minimal viewer
handles.
- raycast: brute-force world-AABB cull + Möller-Trumbore over the CPU
mesh shadow. Used by the laser-measure ceiling/floor distance.
- ToolMode gains Length; click handler routes plain/Alt LMB through
onLengthPick, Backspace through onLengthBackspace. Marquee-arm is
gated off in Length mode.
Volume HUD now shows "Volume: 0.0000 m³ (0 objects)" the moment V is
pressed, matching how A primes "Area: 0.0000 m²" — gives the user a
visible cue the tool is active before any selection.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
deepdiff 9.1.0 added cachebox<6,>=5.2 as a direct runtime dep.
cachebox 5.2.3 only publishes macOS x86_64 wheels for macosx_10_12+,
incompatible with the macos py311 build's --platform macosx_10_10_x86_64.
The daily build's linux-wheel safeguard fires when the resulting
cachebox-*-manylinux_*.whl leaks into the macOS / windows wheels folder
(builds run on ubuntu-latest and cross-build via pip download --platform).
Pin deepdiff to <9.1 (resolves to 9.0.0, no cachebox transitive dep) as
the minimal hotfix. Long-term cleanup: bump the macos py311 platform tag
from 10_10 to 10_13 (matching py312/py313) and re-flag this line with the
standard \$(PYPI_PLATFORM) --only-binary=:all: pattern used by brickschema
and python-socketio.
Partly generated with the assistance of an AI coding tool.
Save-path:
* SaveProject._execute (project/operator.py) now calls
preview_base.discard_pending_previews(context.scene) right after
tool.Parametric.commit_pending_edits(). Previews are session-
transient — discard rather than commit. Sibling gizmo polls gate
on each preview's is_active flag; a stuck flag persisted through
the save would silently hide them on reload. Mirrors the pattern
already in gizmos-8088.
Preview-active gate hardening:
* preview_base.get_preview_props tolerates contexts without a
``scene`` attribute. Pre-existing tests use SimpleNamespace mocks
for the context; the previous getattr(context.scene, ...) raised
AttributeError before the inner default kicked in.
Test update:
* test_wall_header_refresh.test_geom_generation_invalidates_wall_geom_cache
patches tool.Wall.read_geometry instead of the now-deleted local
wall._read_wall_geometry (commit 7e5e7b8d6 migrated the call site).
Generated with the assistance of an AI coding tool.
When override_mode_set_edit encounters an unsupported profile (Couldn't
import profile), deselect the object so Tab continues to cycle cleanly.
Also restores the selection that existed before entering aggregate mode
when finally tabbing out, via save/restore_previous_selection().
Following the pattern from 586f9be077, deselect the active object after
exiting item mode so Tab continues to cycle cleanly. Also deselects
parametric LAYER1/LAYER2 items that cannot be edited directly, avoiding
the need to manually deselect before Tab-cycling out of aggregate mode.
Ports Bonsai's AreaMeasurement onto WgpuViewportWindow as a new
WgpuAreaMeasurement class. Each LMB pick resolves to (instance,
triangle), BFS-expands the coplanar patch (dot(normal, seed_normal)
> 0.9999, ~0.81° tolerance), and toggles it in/out of the running set.
Alt+LMB skips BFS for single-triangle accumulate. Connected-components
sweep over the selected set produces one "X.XXXX m²" label per patch
at its area-weighted centroid in world space; HUD shows the running
total + triangle count.
Dependencies layered in:
- WgpuOverlayRenderer.setHighlightTriangles / encodeHighlightTriangles:
translucent world-space triangle list (cyan @ 0.45 alpha), depth-
tested but depth-write off so the corner gizmo + labels still sit
on top.
- WgpuViewportWindow.pickMeshLocalAt: reuses pickSurfaceAt for the
world hit, then inverts the instance's composed transform to express
it in mesh-local space — what the BFS needs. Uses the live map key
(`mid`) rather than InstanceCpu.model_id, which is whatever the GL
streamer wrote at sidecar-write time and goes stale across sessions.
- WgpuViewportWindow.readbackMeshTriangles: CPU mesh shadow lookup.
The shadow itself is populated during the same dequant pass that
computes mesh-local volume — applyCachedModel for full loads and
applyStreamedChunk for streaming, so the BFS has data the moment
the user can pick it.
WgpuModelGpuData gains a MeshTriangles vector indexed by mesh_id;
doubles per-vertex CPU memory (12 B/vert) but skips wgpu mapAsync
plumbing for now. Bounds-check at pick time gracefully no-ops when a
stale sidecar field is out of range.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Ports Bonsai's volumeOfObjects + volumesPerObject onto WgpuViewportWindow.
Mesh-local volumes are precomputed at applyCachedModel via signed-
tetrahedra-from-origin (dequantising positions from the 12 B/vertex GPU
layout); per-instance volume is just the cached local × |det(placement)|.
No GPU readback — measurement is O(K) in the selection size.
Streaming path computes volumes per-chunk as they arrive — fills any
mesh whose chunk just delivered, then re-runs updateVolumeReadout if
the user is staring at a Volume readout while the geometry pages in.
UX matches GL: V toggles, Esc exits, selection-driven (LMB pick / marquee
/ Shift/Ctrl set ops all funnel into updateVolumeReadout). HUD shows
total + count; one overlay label per object at its AABB centre, capped
at 200 to keep the label-texture cache bounded on large marquees.
Side fixes layered on the label overlay:
- O(1) AABB lookup via object_id_to_instance instead of linear-scanning
every model's instance list per selected object.
- Label texture cache evicts entries not touched this frame, so churning
through "X.XXXX m³" strings doesn't pin GPU memory.
- DrawRec stores the WGPUBindGroup handle by value rather than a
LabelTexture* pointer into the QHash — getOrCreateLabelTexture can
rehash the table and invalidate every captured pointer, which crashed
large marquee selections with BindGroup-no-longer-alive.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Two cohesive cleanups in one commit.
A. Migrate wall.py to PR3-absorbed tool methods (fixes bug 4: pen icon
missing on fillet corner walls):
PR3 shipped tool.Wall.read_geometry + tool.Wall.validate_for_parametric_edit
but wall.py kept local duplicates predating that work. The local
_read_wall_geometry guards on tool.Blender.Modifier.is_wall (LAYER2-only)
while the tool method guards on tool.Parametric.is_path_connectable_wall
(LAYER2 OR fillet corner). Consequence: _get_wall_geom_cached → local
_read_wall_geometry returned None for every fillet corner →
GizmoWallFilletReedit.position_gizmos hit `if geom is None: hide` →
pen icon was unreachable for every fillet corner the user created.
Three _read_wall_geometry callers migrated to tool.Wall.read_geometry
(_read_wall_state_into_props, _get_wall_geom_cached,
GizmoWallJoinIntersection.position_gizmos). Two
_validate_wall_for_parametric_edit callers migrated to
tool.Wall.validate_for_parametric_edit (_maybe_resync_wall_props_from_ifc,
EnableEditingWall._execute). Local helpers deleted; docstring references
updated.
B. Drop over-restrictive gizmo gates (fixes bug 1: join icons missing
when walls intersect away from endpoints):
GizmoWallJoinIntersection.position_gizmos no longer hides itself when
the projected intersection lands further than MAX_DISTANCE_TO_ENDPOINT_
FACTOR (0.75 wall lengths) from any endpoint. The remaining
PARALLEL_DOT_THRESHOLD (cos 2°) gate via project_axis_intersection
returns None for near-parallel walls and is the only correctness bound;
distance from endpoints is a UI concern, not a geometric one.
GizmoWallFilletReedit.poll drops the has_a / has_b ConnectedFrom +
ConnectedTo guard — the IsFilletCorner pset is the authoritative signal.
EnableWallFilletPreviewFromCorner.execute already separately validates
both neighbour connections and reports a user-facing error if either
side is disconnected.
Generated with the assistance of an AI coding tool.
Three live-session regressions surfaced after the fillet feature
landed.
Sister gizmos competed with the active preview:
* preview_base.any_preview_active(context): new helper iterates the
PREVIEW_CANCEL_OPS registry and returns True if any preview is open.
Future previews registered there automatically gate sister gizmos.
* BaseParametricGizmoGroup.poll (gizmos.py): short-circuits on
any_preview_active so every parametric gizmo (door/window/stair/
roof/railing/wall edition) hides during ANY preview.
* The 4 wall gizmo groups with explicit polls (GizmoWallAddOpening,
GizmoWallExtendVertically, GizmoWallJoinIntersection,
GizmoWallUnjoinSingle) + GizmoWallFilletReedit gain the same gate.
DRY: extract _wall_gizmo_poll_gate(context):
* 5 wall gizmo polls each duplicated the 2 pre-flight checks
(viewport-gizmos enabled + no preview active). The helper centralises
them — each poll becomes a single short-circuit line followed by its
per-feature selection inspection.
ESC cancels the active preview:
* try_cancel_active_preview already existed in preview_base since PR3
but had no caller. Hooked into OverrideEscape.execute (geometry/
operator.py) as a new elif branch — same keymap that already cancels
pen gizmo edit mode + item mode + edit mode + aggregate mode. Order
in the branch chain matters: try preview cancel before falling back
to try_canceling_editing_modifier_parameters_or_path so the in-
flight preview wins over a stale modifier-edit cancel attempt.
Generated with the assistance of an AI coding tool.
In BIM_PT_type_attributes and BIM_PT_object_attributes (when
the active object is a type), attribute value buttons now use
"type.<Attr>" as the selector key so the operator finds
matching occurrences via their relating type rather than the
occurrence's own (often unset) attributes.
Generated with the assistance of an AI coding tool.
After selecting objects in the same container, copy a `location="Name"`
filter query to the clipboard and report it — consistent with the same
behaviour in SelectSimilarType, SelectSimilarAggregate, SelectIfcClass,
and SelectSimilarMaterial.
Generated with the assistance of an AI coding tool.
Ports GL OverlayRenderer's setOverlayLabels + setHudText to wgpu.
Each unique string is rasterised via QPainter into a QImage (dark-grey
rounded background + white antialiased text) and uploaded as an RGBA8
texture; the cache is keyed by content + font size so identical
strings across frames are texture-free. Per-frame work is projection,
vertex assembly, and one draw per visible label.
Drawn last in the frame on the resolved surface so labels sit on top
of every other overlay (no depth-test). HUD uses pt 11 at top-left
matching GL; world-anchored labels use pt 9 centred at the projected
screen position.
WebGPU has no QOpenGLPaintDevice equivalent — the GL backend's two-
stage GL-rect + QPainter pass becomes one textured quad per item here.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Ports GL OverlayRenderer's setOverlayPoints API to WgpuOverlayRenderer.
Each point becomes a 6-vertex screen-space quad sized to inner_diameter
+ 2*stroke_extra; the fragment reads its per-vertex corner varying
instead of gl_PointCoord (WebGPU has no sized-point primitive). Sharp
inner/stroke transition + AA on the outer edge only, matching GL.
Single uniform slot per set — colors are global to the call, not per
point. Vertex buffer regrows 1.5× on demand so steady-state sets don't
re-allocate.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
When a `+`-separated filter group returns no results, `FacetTransformer.facet_list`
was skipping the reset of `has_additive_facet_in_current_list` because the reset
was inside the `if self.elements:` guard. The stale flag caused the next group's
`add_default_elements()` to bail out early, leaving its element set empty and
silently dropping every subsequent group from the result.
Move the flag reset outside the guard so it always fires regardless of whether
the group produced any results.
In the Placement panel, show Location and Rotation X/Y/Z
each on their own row beneath a header label. When the IFC
file uses imperial units, display a read-only feet-and-inches
label alongside each Location input field.
Generated with the assistance of an AI coding tool.
Display general and calculated stair parameters (Width,
Height, Tread Run, Tread Rise, Length, etc.) formatted
to the IFC file's configured length unit rather than
raw numeric values.
Generated with the assistance of an AI coding tool.
End-to-end fillet flow on top of the helpers + recreate_wall hook
(landed in the previous commit). Users select two LAYER2 walls, click
the fillet entry icon, drag the live radius widget, and validate to
replace the corner with a curved LAYER2 corner wall (banana body).
Operators (5):
* EnableWallFilletPreview: 2-wall selection → validates LAYER2 +
straight axis + zero-slope + intersect-or-joined state → seeds the
preview props with a default radius computed from the shorter
available leg.
* FinishWallFilletPreview: dispatches CreateWallFillet with the tuned
radius; clears preview state on FINISHED, preserves it on failure so
the user can re-tune without re-selecting.
* CancelWallFilletPreview: clears preview state, no IFC mutation.
* EnableWallFilletPreviewFromCorner: pen-icon re-edit on an existing
fillet corner — pre-fills the preview from the corner's BBIM_Wall
pset + walks the inverse graph to recover wall A and wall B.
* CreateWallFillet: deletes any prior corner + A↔B path connection,
shortens A and B to the tangent points, instantiates a corner wall
from A's type, unassigns the swept-layer material/type (the explicit
banana body MUST own its geometry), assigns the dominant material,
rebuilds the body, sets a straight 2-point chord axis, stores
BBIM_Wall.IsFilletCorner+FilletRadius, reconnects A and B to the
corner with NOTDEFINED on the corner's side.
Gizmo groups (2 new + entry icon on existing):
* GizmoWallFilletPreview: visible while a preview is active. Bundles
a radius_dim widget at the arc apex, a trim_dim widget along wall A
expressing the same DOF via the leg setback distance
(trim = |radius| * tan(sweep/2)), and validate / cancel icons
anchored above the apex in screen-up.
* GizmoWallFilletReedit: pen-icon entry on an existing fillet corner
wall (single-selection, BBIM_Wall.IsFilletCorner set, both neighbour
connections present). Mutually exclusive with an active preview.
* GizmoWallJoinIntersection now stacks a fillet entry icon
(VIEW3D_GT_fillet → bim.enable_wall_fillet_preview) above the
existing join/unjoin icon in the joined and intersect state branches.
Property + decorator infrastructure:
* prop.py: BIMWallFilletPreviewProperties (Scene-level draft) +
BIMPreviewProperties umbrella with only the wall_fillet pointer.
The umbrella is the seam preview_base.py (landed in PR3) already
reads via getattr(scene, "BIMPreviewProperties", None).
* decorator.py: _stroke_lines_alpha helper + WallFilletPreviewDecorator.
Polls is_active; renders leg projections + arc + arc-center
construction lines from tool.Wall.compute_wall_fillet_geometry.
* __init__.py: registers operators + gizmo groups + property groups +
wires Scene.BIMPreviewProperties.
* handler.py: WallFilletPreviewDecorator.install/uninstall in
_install_decorators — always installed, self-polls on is_active.
Drive-by: extract gizmo.get_screen_up(billboard_rot) helper —
the local +Y of a billboard rotation is the camera's screen-up world
direction. Replaces 4 inline `billboard_rot @ Vector((0.0, 1.0, 0.0))`
sites added across the fillet feature's gizmo groups.
Generated with the assistance of an AI coding tool.
Ports GL OverlayRenderer's LineGroup API to WgpuOverlayRenderer. Each
group's segments are CPU-expanded into screen-space quads; the WGSL
fragment reproduces the GL pixel-distance stroke pick + arc-length
dash logic. One uniform slot per group, bound via dynamic offset so a
single bind-group services up to N groups.
No caller yet — sets up the API the wgpu measure tools (task #29) will
use. WgpuViewportWindow.setOverlayLines mirrors the GL viewport's
signature so the bonsai Measurement code can target either backend
through one interface.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
WgpuViewportWindow.cpp had ~1100 lines of pipeline/shader/buffer plumbing
for the axis indicator, pivot gizmo, section visualizer, and marquee
drag rect. Mirroring the GL backend's split, that lives in its own class
now; the viewport keeps the camera/cull/draw loop and hands the renderer
a per-frame WgpuOverlayFrame snapshot for each encode call.
No behavioural change — pixel-identical screenshot on basic.ifcview.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Drag LMB in empty space to select every visible object whose pick-pixel
falls inside the rect. Mirrors GL ViewportWindow's marquee.
UI flow
- LMB press in non-tool-consuming context arms the marquee. The
cursor must move past kBoxSelectThresholdPx (5 logical) for it to
become active — until then a release falls through to single-pick,
so an unintentional micro-drag still picks under the cursor.
- Press-time modifiers decide the set op so a mid-drag Shift release
doesn't flip behaviour:
plain → selection.clear() then add every picked id
Shift → add to current selection
Ctrl → remove from current selection
- Section tool intercepts plain LMB first (already wired); the
marquee is mutually exclusive with it.
Rectangle pick
picksInRect(x, y, w, h):
- Render the existing pick pass (R32UInt object_id + RGBA16F normal
MRT) into the persistent pick attachments.
- copyTextureToBuffer the rect region of pick_color_texture_ into
box_pick_staging_buffer_ (regrown 2× on demand to fit the
largest rect we've seen). R32UInt is a color format so partial
sub-rect copies are allowed (unlike Depth32Float).
- Iterate the mapped staging buffer, accumulate unique non-zero ids
into an unordered_set, return.
Visual rect
A new marquee overlay pipeline draws the drag rect on the resolved
surface after the corner gizmo. Two passes per active frame share one
uniform buffer / bind group:
fill — 6-vert unit quad, vs_fill maps (0,1)² to NDC via
rect_min/rect_max, fs_fill outputs color × fill_alpha
outline — 24-vert thick-line quad (4 segs × 6 verts), uses the
shared thick_line_clip + fs_main from THICK_LINE_HELPERS_WGSL
so the rect outline has analytical AA without MSAA
Colour: Bonsai decorator_color_special (0.157, 0.565, 1.000) for the
axis-blue parity the user requested; outline alpha 0.95, fill alpha
0.20 of that so geometry behind the rect still reads.
Closes#41 and #61.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Consolidation
THICK_LINE_HELPERS_WGSL — a macro that both AXIS_WGSL and SECTION_WGSL
prefix via adjacent string-literal pasting — holds:
- VsOut: clip_pos + rgba colour + side_t for AA
- thick_line_clip(p_start, p_end, t, side, viewport, line_width):
the screen-space quad expansion with consistent perpendicular so
the quad never collapses into a bowtie
- fs_main: |side_t| + fwidth() coverage smoothstep — analytical
1-pixel AA regardless of MSAA
Each gizmo shader now only declares its uniform struct + a 10-line
vertex shader. C++ side gains thickLineVertexLayout(attribs[5]) so
both call sites set up the 5-attribute layout in one call instead of
20+ lines each. Net diff is -28 lines on this commit and roughly -60
relative to the unconsolidated section commit; the next thick-line
gizmo (measure tool, selection outline, …) starts from ~30 lines of
WGSL + a vertex buffer.
Bonsai decorator palette
All overlay colours now come from src/bonsai/bonsai/bim/ui.py's
decorator_color_* defaults so they match Bonsai's Blender add-on:
decorator_color_error = (1.000, 0.200, 0.322) red → +X axis, section gizmo
decorator_color_selected = (0.545, 0.863, 0.000) green → +Y axis
decorator_color_special = (0.157, 0.565, 1.000) blue → +Z axis
Section gizmo polish
- Entire gizmo (quad outline + arrow shaft + arrow head) goes red.
GL's white quad + yellow arrow disappeared against light surfaces;
one saturated red reads against any background and identifies the
geometry as a tool overlay.
- Line width bumped to 5 logical px and the per-vertex tint dropped
to (1, 1, 1, 1) so the tint multiplier stays available for a future
"selected" state without changing the base colour.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Mirrors GL ViewportWindow's section tool end-to-end.
Hotkeys
K toggle the tool active
Shift+K clearSectionPlanes
Esc deactivate the tool
Del/Bksp remove the most recently added plane (tool-active only)
LMB click pickSurfaceAt → addSectionPlaneAtSurface (no modifier)
LMB drag on the arrow gizmo: slide the plane along its normal
State
FrameUniforms grows by clip_count (i32) + clip_planes[6] (vec4). The
six-plane cap matches GL's MaxSectionPlanes. WGSL pads via three
scalar i32s instead of a vec3<i32> so the array starts at offset 144
to match the tightly-packed C++ struct (240 B) — vec3 would have
forced clip_planes to 160 and broken the binding-size match.
is_section_clipped(world) in WGSL evaluates all active planes and
returns true if any signals "on the positive side". Both main and
pick fragments discard with it so cuts are visible AND selection is
consistent — you can't pick something the user can't see.
Surface pick
Pick pass now emits 2 color targets: R32UInt object_id at @location(0)
and RGBA16F packed world-space normal at @location(1). Normal is
packed × 0.5 + 0.5 so unsigned-ish halfs keep the sign. pickSurfaceAt
reads both via 1×1 texel copies (RGBA16F is a color format with no
full-mip restriction, unlike Depth32Float). World position comes from
ray-AABB intersection against the picked instance's AABB — equally
accurate for "drop a plane where I clicked" and dodges the Depth32Float
copy-extent rule entirely. The pick normal is decoded into the
per-fragment surface normal so the plane lands perpendicular to the
actual triangle (not the AABB face).
Plane gizmo
Identical geometry to GL's renderSectionPlanes: 2×2 m quad outline
(white) + 1 m arrow shaft along +n (yellow-orange) + 4 arrow-head
diagonals. Drawn inside the main MSAA pass with depth LessEqual + no
depth write. Lines are rendered as screen-space-expanded thick quads
with fwidth-based AA, same technique the axis indicator uses, so the
gizmo reads against busy BIM geometry rather than disappearing as
1-px hairlines.
Drag
mousePressEvent claims a plain-LMB press if it hits an arrow gizmo
(12 logical-px grab radius, distance to the (origin, origin+n)
screen-space segment). The drag handler projects the cursor delta
onto the screen-space axis and converts to metres via
delta·axis / |axis|² — same formula GL uses. Mid-drag camera moves
keep working because the projection re-runs every frame against the
press-time origin.
Closes the click-to-add + drag halves of #30 / #60.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Both overlays draw the same three positive-axis rays (origin → +X / +Y
/ +Z) — one screen-space-thick-line shader, one vertex buffer, one
bind group layout. The vertex stage transforms each vertex as
`mvp * (origin + position * arm)` so the same primitive serves both
modes:
corner: viewport set to a 110×110 px box in the bottom-left,
camera-orientation ortho MVP, origin=0, arm=1.
pivot : full viewport, main view-proj, origin=camera_target,
arm = 30 logical px in world units.
Pipelines
axis_pivot_pipeline_ — MSAA + depth LessEqual (α=1)
axis_pivot_xray_pipeline_ — MSAA + depth GreaterEqual (α=0.30)
axis_corner_pipeline_ — resolved surface, no depth, sampleCount=1
Pivot renders inside the main MSAA pass after geometry (depth
interaction); corner renders on the resolved surface after the edge
silhouette pass so the laplacian can't darken its lines. The pivot's
two passes — x-ray first then visible — give an occluded-side hint
matching GL's renderPivotIndicator.
Screen-space thick lines
WebGPU has no lineWidth, so each axis is a 2-triangle quad expanded
by `line_width / 2` pixels along the screen-space perpendicular in
the vertex shader. Every vertex carries BOTH endpoints (start, end)
plus `t ∈ {0,1}` and `side ∈ {-1,+1}` so the direction is computed
consistently as `s_end - s_start` regardless of which end the vertex
sits at — an earlier "this vertex vs the other end" formulation
flipped sign at the end vertex and produced a bowtie.
Analytical AA
|side_t| ∈ [0,1] is the perpendicular distance from the line centre.
`smoothstep(1-fwidth, 1, |side_t|)` gives a 1-pixel coverage falloff
at the long edges — gizmos read cleanly even on the resolved-surface
corner pass which has no MSAA.
Pivot visibility
- orbit / pan drag press → on, release → off
- wheel zoom → on with 600 ms afterglow via QTimer
Pole fallback for the corner gizmo's lookAt mirrors buildViewProj's
identical fix (swap Y-up when |pitch| ≥ 89°), so top/bottom standard
views don't degenerate.
Tracked under task #59.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Removed the WGPU_SPATIAL_BUCKETS=1 octree-style instance planner
(planSpatialChunks, SpatialPlan, the env-var pair, the field, the
load-time branch). Was an opt-in prototype kept in tree as a possible
acceleration for "find the right instance chunk", but:
- Benchmarked slower than mesh-keyed (~24-26 ms vs ~19.7 ms) — the
higher chunk count's per-chunk bind-group + draw overhead more than
ate the tight-AABB win on this dataset.
- Not load-bearing for the brace correctness fix — that turned out to
be the AABB-projected screen-rect priority metric (commit 6200ab9fa),
which works on either chunk topology.
- The "find the right chunk" hypothesis is moot: instance_chunk_idx[]
is precomputed at load time, so cull has nothing to look up.
Git log preserves the implementation if it's ever revisited.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Three correctness bugs found and fixed, plus an unrelated fly-mode
deadlock surfaced along the way.
HiZ false-rejection at the bottom of the screen
------------------------------------------------
The mip-pyramid sizing floored when halving — for a 256×70 mip 0 the
level-3 mip is 32×8, but mip-0 row 69 maps to ly = 69>>3 = 8, which is
out of bounds for an 8-row mip. ly1 then clamps down to 7 while ly0
stays at 8, the sampling loop runs zero times, max_d retains its
initial 0.0, and `min_z > 0` rejects every AABB whose projected y
range touches the bottom row. Same class for the right edge on very
wide viewports.
Fix: ceil rather than floor when halving mip dimensions so every
parent row has a covering child texel, plus std::clamp on both lookup
endpoints as belt-and-suspenders for any future mip-sizing change.
Surfaced after the user added more sidecars and saw "anything near the
bottom of the screen, no matter close or far" disappear ~0.5 s after
camera stops — that delay was the strict-VP gate + readback latency
opening the HiZ window. Found via WGPU_HIZ_TRACE rejection logs that
showed every rejection had `max_d=0` and `ly0 > ly1`.
Streaming priority lets the ocean starve out the bracing
---------------------------------------------------------
Per-instance projected screen footprint was estimated as bounding-
sphere radius squared. BIM geometry is overwhelmingly thin-in-one-axis
(slabs, pipes, columns, windows) and a flat ocean plane viewed nearly
edge-on gets a sphere projection ~250× larger than its actual screen
rect. Its chunk dominated the priority ranking and evicted the brace
chunks despite the braces being one of the closest visible things.
Fix: per-instance priority is now the screen-space AABB rectangle area
(world AABB extents projected onto the camera right/up basis vectors,
divided by view-z). Sphere radius is retained for the contribution
cull and LOD pick because conservative-over is the right failure mode
there.
Stale-VP HiZ gate
-----------------
HiZ resolves into an async ping-pong of staging buffers, so the
pyramid resident at cull time was typically captured one or two
frames ago. During camera motion the captured VP differs from
vp_this_frame and AABBs end up sampling depth taken for what was at
slightly-different screen positions in the old view. Strict by
default now: HiZ engages only when hiz_vp_ == vp_this_frame.
WGPU_HIZ_MOTION=1 trusts the stale pyramid (matches GL's default).
Fly mode Shift+Q deadlock
--------------------------
keyPressEvent requested a redraw only on the first key of a new held
set (was_empty). Pressing Shift first then Q never satisfied that
condition because Shift had already populated the set, so the render
loop never ticked. Now every relevant keypress calls requestUpdate
unconditionally.
HiZ stays opt-in behind WGPU_HIZ=1 for one release while the fix
bakes; WGPU_HIZ_TRACE=1 keeps the per-rejection diagnostic available
for future bugs.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
DecorationsHandler now caches dissolved edges (mesh-keyed), world-space
draw payload, and GPUBatch objects with per-object epoch invalidation —
moving one wall doesn't wipe 50 opening caches. Object-mode dissolve
removes triangulation noise; 2-pass depth-test split dims occluded lines
instead of hiding them. Edit-mode behavior unchanged.
Also: disable viewport shadows for IfcFeatureElementSubtraction objects,
and wire DecorationsHandler.uninstall() into the model module's
unregister() so the new persistent handlers don't leak on addon disable.
Generated with the assistance of an AI coding tool.
WGPU_SPATIAL_BUCKETS=1 swaps the applyCachedModelStreaming planner from
mesh-keyed Morton+greedy to octree-style instance bucketing. Default
behaviour unchanged (env var unset → mesh-keyed planner runs).
Phase 1 of #55 / #56. The mesh-keyed planner produces chunks whose
AABBs are the union of all instances of the chunk's meshes — for
heavily-deduplicated IFC meshes (a "standard floor tile" used 800
times across a federation) the mesh's "centroid" is a mean of scattered
instance positions and the chunk's AABB ends up spanning the entire
model. Symptom: chunk-level frustum cull rarely fires (AABB always
intersects view), and the screen-area priority metric under-rates
big-AABB chunks because their corners straddle the near plane. Visible
objects pop in/out as the camera tilts, even though they're fully on
screen.
The spatial planner bucketises INSTANCES directly. Each leaf bucket
contains its instance list + the unique mesh data those instances
reference. A mesh whose instances scatter into multiple buckets gets
its vertex/index data uploaded into multiple pool slices — duplication
is the cost for tight bucket AABBs. For IFC this is acceptable:
heavily-shared meshes tend to be small (fittings, fasteners), so
per-bucket duplication adds tens-of-MB not GB.
Octree implementation (planSpatialChunks):
- work-stack subdivision: for each (instance subset, AABB), split into
8 octants around centre and recurse
- stop conditions: bucket fits WGPU_CHUNK_VERTEX_BYTES_LIMIT for
union vertex bytes AND ≤ spatial_max_instances_ instances; OR single
instance left; OR every instance falls into the same octant
(pathological — emit as leaf rather than infinite recurse)
- spatial_max_instances_ default 5000, overridable via
WGPU_SPATIAL_BUCKET_MAX_INSTS env var so the prototype can be
swept without rebuilding
Data-model adjustment beyond what dc2927997 prepared:
- Per-chunk per-mesh chunk-local offset table (chunk_mesh_offsets)
built during the chunk-construction loop. The mesh-keyed per-mesh
global arrays (mesh_chunk_idx etc.) still get populated for
legacy reads, but under spatial bucketing they're overwritten when
the same mesh appears in multiple chunks — harmless because cull
reads the per-instance arrays exclusively (per dc2927997).
- Post-construction, per-instance arrays are populated from
chunk_mesh_offsets via (instance_to_chunk[i], inst.mesh_id) lookup.
Mesh-keyed planner derives identical values to before
(pixel-identical); spatial planner now writes the correct
per-bucket offsets even when the mesh appears in multiple chunks.
basic.ifc parity on all three paths confirmed (non-streaming
mesh-keyed, streaming mesh-keyed, streaming spatial all produce 0
pixel diff vs the reference). Spatial planner produced 1 bucket on
basic.ifc (3 instances, well under thresholds) as expected.
Non-streaming applyCachedModel left unchanged — the prototype targets
the streaming path which is where the federation-scale missing-objects
issue lives.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Preparatory refactor for spatial instance bucketing (#55). Cull
previously routed through per-mesh tables (mesh_chunk_idx,
mesh_chunk_local_base_vertex, _ebo_first_u32, _lod1_first_u32) to
find the chunk and chunk-local offsets for each instance. That
assumes a mesh lives in EXACTLY ONE chunk — the assumption holds
under the current mesh-keyed planner but breaks under spatial
bucketing, where the same mesh can be duplicated across multiple
buckets if its instances are scattered.
Adds four per-instance arrays (instance_chunk_idx,
instance_base_vertex, instance_ebo_first_u32, instance_lod1_first_u32)
populated at planning time. The current mesh-keyed planner derives
them by translation:
instance_chunk_idx[i] = mesh_chunk_idx[instances[i].mesh_id]
The spatial-bucket planner (next commit) will populate them directly,
allowing the same mesh_id to map to different chunks for different
instances.
cullModelCpuCompute now reads the per-instance arrays:
- frustum_visible_count uses chunks[instance_chunk_idx[i]]
- VisibleDrawGpu uses instance_base_vertex / instance_ebo_first_u32
/ instance_lod1_first_u32
- LOD-select branch and use_lod1 logic unchanged.
Per-mesh tables stay (used by makeChunkRequest, debug logs, the
planner itself). Memory cost: 16 bytes × N instances ≈ 16 MB on a
1M-instance scene. Pixel-identical on basic.ifc in both non-streaming
and streaming modes.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Brings the wgpu viewport's keyboard + mouse into line with GL ViewportWindow
+ Bonsai's MainWindow shortcut table, lands fly-mode, swaps in three
diagnostic env vars, and fixes a chunk-priority bug exposed by the
diagnostics.
Keyboard parity with GL + Bonsai:
P — toggle perspective / ortho projection
X / Shift+X — front / back view (eye on ±X, pitch 0)
Y / Shift+Y — right / left view (eye on ±Y, pitch 0)
Z / Shift+Z — top / bottom view (pitch ±90°)
F — focus camera on currently selected object
Home — frame entire scene
C — print --camera CLI args for current view
H — hide selected
Shift+H — isolate selected (hide everything not in selection)
Alt+H — show all (clear hidden set)
Shift+F — enter fly mode (matches BonsaiViewer)
Escape (fly) — exit fly mode
WASD/QE/Shift — fly movement (when in fly mode)
The previous H/Shift+H/I assignments were wrong vs Bonsai (Shift+H went
to show-all, I to isolate); both are fixed.
Fly mode:
- GL-style absolute m/s base speed (default 5.0), Shift = 5×, scrollwheel
adjusts ×1.25/×0.8 per notch (Blender convention). Scrollwheel does
NOT zoom in fly mode; that interfered with speed when speed was
distance-scaled (it was, briefly; replaced with absolute m/s).
- Mouse-look pins eye: yaw/pitch update first, then target is re-derived
so orbitEye(target, dist, new_yaw, new_pitch) == old eye. Result:
camera rotates in place (FPS) rather than orbiting the pivot.
- dt ceiling clamp at 100ms (matches GL fps_move_speed_) so a stall
doesn't warp the camera.
- Pitch sign matches non-inverted FPS convention (mouse-up = look up).
Mouse-nav presets (WGPU_NAV_PRESET=blender|rhino|revit, default blender):
Blender — Orbit MMB, Pan Shift+MMB
Rhino — Orbit RMB, Pan Shift+RMB
Revit — Orbit Shift+MMB, Pan MMB
LMB stays free for selection in every preset. Nav-drag kind is captured
at press time so a mid-drag Shift release doesn't flip orbit↔pan. The
pan up-vector switches to world-Y at near-vertical pitch so panning
still works in top/bottom view (would otherwise NaN at pitch=±90°).
Camera-math refactor:
buildViewProj(view, proj) centralises perspective↔ortho selection and
the near-vertical up-vector switch. Four open-coded copies of the
view/proj build (cull, debug, streaming priority, render uniforms) now
call it instead, ensuring projection mode + up-vector switch land
identically everywhere. basic.ifc pixel-diff is 0 (refactor confirmed
output-equivalent on the path with no ortho / no near-vertical pitch).
WGPU_PRESENT_MODE=fifo|fifo_relaxed|mailbox|immediate (default fifo):
Diagnostic toggle for stutter analysis. fifo_relaxed gave the tightest
per-frame dt distribution on a federated bench scene; immediate gave
uncapped throughput at the cost of tearing. Mailbox not supported on
Vulkan + NVIDIA Linux but kept as an option for other backends.
WGPU_FLY_DEBUG=1: per-frame [fly] log printing dt, render gap, key
count, speed, position delta. Confirmed render_gap == dt to four
decimal places — fpsIntegrate runs exactly once per render, no
double-tick. Cull cost (~14-20 ms) is the dominant frame variance and
the eventual fix is task #49 (sub-model parallel cull) — fly-mode
stutter on slow scenes is a downstream symptom of cull cost, not a
fly-mode bug.
WGPU_STREAM_DEBUG=1: per-frame [stream-debug] log with cands / enq /
drained / ev_lru / ev_pri / blocked / resident / cycled / max_load.
Confirms thrash / pool-bound / load-budget cases on big scenes.
Pick-and-track diagnostic: clicking an object enumerates every chunk
holding instances of that object (an IFC object can split across
representations / chunks), printing each chunk's AABB + instance AABB +
residency. If any tracked chunk's is_resident flips true → false in
driveStreamingLoads, an EVICTED dump prints with the chunk's AABB,
priority, pool state, this-frame eviction counts, and the top-5
candidates that displaced it. Surfaces exactly why an object disappeared.
chunkScreenAreaPx fix (uses diagnostic to confirm the bug):
A chunk's AABB is the union of every instance's world AABB in the
chunk. On a federated IFC the camera commonly sits INSIDE that AABB
(e.g. inside a 263×30×15 m floor-area bounding box). Previously the
8-corner projection silently dropped corners with clip.w <= 1e-3
(behind near plane), so the projected bbox of the surviving in-front
corners was a tiny fraction of the chunk's true on-screen footprint.
Result: big-AABB chunks lost every eviction fight, visible objects
popped out as the camera tilted. Fix: short-circuit to full-viewport
area when (a) eye is inside the chunk AABB (mirrors GL's
contributionPasses camera-inside short-circuit), or (b) any AABB
corner sits behind the near plane (AABB straddles → 8 corners cannot
honestly measure footprint; conservatively over-prioritise).
The fix is a workaround for the deeper chunking issue — chunks are
mesh-keyed (group of meshes), and a mesh's AABB used in chunking is
the mean of its instances' positions, which is meaningless for
heavily-deduplicated meshes scattered across the scene. The root fix
is task #55 (spatial instance bucketing, runtime prototype) + #56
(sidecar v15 instance-keyed format). chunkScreenAreaPx fix unblocks
the user-visible "missing objects" issue while those land.
Extracted chunkScreenAreaPx from a driveStreamingLoads-local lambda
to a private member so the disappear-diagnostic and any future call
sites can use it consistently.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
EditAssignedMaterial propagated layer set usage attributes
to all selected objects but skipped this loop for profile
set usage. Add the same loop so CardinalPoint and
ReferenceExtent are copied to each selected object's
IfcMaterialProfileSetUsage on save.
Generated with the assistance of an AI coding tool.
Adds three knobs and three new heartbeat numbers to support the
ongoing perf-parity work. None affect behaviour in default runs.
cull[wall|compute|upload] split timer
The existing cull_timer wrapped both the parallel std::async dispatch
and the sequential cullModelCpuUpload loop (queueWriteBuffer × 3 per
resident chunk × ~120 chunks ≈ 360 wgpu calls per frame). Splitting
them ruled upload out as the bottleneck on a 51-model federation
scene: compute ≈ 16-17 ms, upload ≈ 1 ms.
WGPU_CULL_THREADS=0 — force sequential cull
std::async-per-model was already in place; this env var disables it
so we can compare wall time vs sequential and confirm parallelism is
working. On the federation scene with 52 models: sequential 74 ms vs
parallel 17 ms = 4.4× speedup. Confirmed; the 17 ms floor is not a
parallelism failure, it's the cost of culling the largest single
model (model 43, 114k instances) ÷ no parallelism within that model.
WGPU_STREAM_DEBUG=1 — per-frame [stream-debug] log
Surfaces cands/enq/drained/ev_lru/ev_pri/blocked/resident/cycled/
max_load each frame from driveStreamingLoads. The "cycled" /
"max_load" pair makes thrash vs eviction-churn vs just-loading
distinguishable. Off by default; opt-in via the env var.
Bench-warm timeout dump
When [bench warm] times out (600 frames without 0-loads streak),
prints a structured summary: resident/missing/total chunks,
cycled count, pool usage, largest free run, avg missing chunk
size, and an auto-classifier diagnosis (POOL FRAGMENTED vs
WORKING SET > POOL vs FEW-CHUNK CYCLE vs still-loading). Caught
a real fragmentation pattern (18 MB largest free run vs ~100 MB
typical chunk) on a 51-model run where the dumb classifier
would have called it a load-budget problem.
LOD1 firing counter
"lod1 X/Y (saved Z tris, N no-lod1)" suffix on the [frame] log.
X = LOD1-selected this frame, Y = LOD1-eligible, Z = tris not
drawn vs always-LOD0, N = visible instances with no baked LOD1
(mesh below IFC_LOD_MIN_TRIS). Confirmed LOD1 path is genuinely
firing post the per-chunk LOD1-storage commit, and exposed that
~90% of instances in real scenes are no-lod1 meshes — relevant
to the future LOD-tier-residency design.
Chunk.load_count + Chunk.lod0/1 layout bookkeeping
Per-chunk reload counter for the thrash detector. lod0/1
layout_count fields prep the data model for distance-tiered
residency (Phase B of #31) but aren't acted on yet.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
wgpu computes projected_px from view-Z distance (forward·(centre-eye)),
which is the perspective-divide-correct denominator: an instance's
on-screen radius really is world_radius * focal / z_view. GL computes
the same value but with euclidean distance (sqrt(dx²+dy²+dz²)) — for
off-axis instances euclidean > z_view, so GL underestimates screen
size and culls more aggressively at the same numeric threshold.
Concretely on the federation scene at the test camera, wgpu was
drawing ~3× the instances GL drew despite identical 2/10 thresholds:
wgpu obj 11067, hiz_rej 16532 vs GL obj 2310, hiz_rej 6823. Same
fps (vsync-pinned), but ~30% more cull work for raster output that
the user already wasn't seeing because GL had been quietly dropping it.
Keeping wgpu's view-Z formula (more physically correct) and bumping
the thresholds to 3.0 / 15.0 to match GL's effective drop rate. On
the federation scene this lands obj/tri counts within ~10% of GL
across the orbit, and shaves cull from 3.44 ms to 2.61 ms avg.
basic.ifc parity is unchanged (its 3 instances clear 3 px easily).
WGPU_MIN_PX / WGPU_MIN_PX_MOTION env vars added so the thresholds
can be swept without rebuilding — needed while we visually
confirm the new defaults across more scenes / cameras.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Sidecar LOD1 is per-mesh, index-only — meshoptimizer-baked decimated
index slices that share the LOD0 VBO. Before this change the wgpu
chunk path force-disabled LOD1 (effective_lod1 = false; tagged in a
comment as "until per-chunk LOD1 storage lands"), so it had to walk
every visible instance at full LOD0 even when the per-instance LOD
selection said the projected radius was below the LOD1 threshold.
Per-chunk layout: append LOD1 indices for the chunk's meshes after
all LOD0 indices in the same pool slice. A new
mesh_chunk_local_lod1_first_u32 array records each mesh's LOD1
starting offset (in u32s) within the chunk's index slice; LOD0
offsets stay where they were. The cull's emit then sets
VisibleDrawGpu.ebo_first_u32 to whichever side matches the per-
instance use_lod1 decision the prior #8 commit already computed.
Vertex pulling is oblivious to the LOD split — it just reads the
indices the cull pointed it at.
c.index_count is repurposed as the LOD0+LOD1 total so the pool
allocation, eviction's pool-fit check, and the VRAM accounting all
scale automatically. c.lod1_index_count exposes the LOD1 portion
for stats.
makeChunkRequest appends LOD1 byte ranges to req.i_ranges in the
same per-mesh order; the streaming worker concatenates ranges in
order, so the assembled idx blob lands LOD0-first / LOD1-second
which matches the chunk-local packing.
On a 10-model regen with meshoptimizer-baked sidecars, the bench
[frame] heartbeat shows lod1 firing on ~80-90% of LOD1-eligible
instances and saving 10-30M tris per frame versus the prior
LOD0-only ceiling. Same camera/scene on basic.ifc is still
pixel-identical (no mesh in basic.ifc is large enough to bake a
LOD1, so the cull just follows the LOD0 path it always did).
Temporary debug counters (lod1_dbg_count_ et al.) print
"lod1 X/Y (saved Z tris, N no-lod1)" in both interactive and
benchmark [frame] heartbeats — kept on while LOD1 correctness gets
confirmed across more scenes, will come out once trust is built.
The non-streaming applyCachedModel path runs the same LOD1 plumbing
but no longer fits the full federation scene in pool (extra index
bytes push past the 2 GB single-buffer cap); that mode was
already streaming-only on that scene before.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
When the user enters `-0' - 10"`, Python parses feet as -0.0.
The check `feet < 0` is False for negative zero, so the sign was
silently dropped. Use math.copysign to detect it correctly.
Generated with the assistance of an AI coding tool.
Eleven module-level helpers in wall.py that the upcoming wall-fillet
operators + gizmo groups depend on. Each is self-contained or
references only helpers earlier in the file; the operators and
gizmos themselves land in follow-up commits.
* _wall_fillet_props / _wall_fillet_preview_active /
_wall_fillet_preview_walls: thin read-side accessors over the
BIMPreviewProperties.wall_fillet pointer (added with the
operators commit). Safe today: get_preview_props returns None
until the pointer is attached.
* _walls_have_zero_slope_for_fillet: validates that input walls
are vertical (x_angle ~ 0); slanted-extrusion fillets require
swept-along-curve geometry the banana profile builder doesn't
support.
* _build_curved_corner_body_representation: builds the banana
(annular sector) IfcExtrudedAreaSolid as a polyline-tessellated
IfcIndexedPolyCurve.
* _apply_fillet_corner_geometry: positions the corner wall at
tangent_a and rebuilds its body. Shared by the creation operator
and the regenerate path.
* _resolve_two_walls: pulls (active, other) from a 2-wall
selection, validates both as LAYER2 + straight-axis + not-already-
a-fillet-corner.
* _pick_dominant_wall_material: returns the thickest layer's
material from an element's IfcMaterialLayerSet / Usage.
* regenerate_fillet_corner_wall: re-runs the geometry build from
BBIM_Wall.FilletRadius + current neighbour layer parameters.
Called by tool.Model.recreate_wall when the IsFilletCorner pset
is set; the FIXME(PR4) placeholder in recreate_wall is dropped.
* _wall_fillet_gizmo_x_matrix: 4x4 placement matrix with local +X
aligned to a world-space direction; used by the fillet preview
gizmo group.
Centralises the IsFilletCorner pset read as
tool.Parametric.is_fillet_corner_wall — replaces 3 inline
get_pset(element, "BBIM_Wall", "IsFilletCorner") sites
(tool.Model.recreate_wall, tool.Model.recalculate_walls,
tool.Parametric.is_path_connectable_wall) plus the new
_resolve_two_walls call.
Generated with the assistance of an AI coding tool.
The framework's parametric-edit icon row currently binds an array
icon to bim.add_array_from_feature_edit, but the supporting per-
feature add-array flow and gizmo positioning haven't fully landed.
Showing the icon today lets the user click it and trigger a half-
wired flow.
Force the icon hidden inside the props.is_editing branch of
BaseParametricGizmoGroup.update_editing_gizmos. The else-branch
(not editing) already hides it, so this just mirrors that behavior
during edit mode. Drop this gate when array integration completes
to re-enable the icon position + visibility plumbing.
Generated with the assistance of an AI coding tool.
The extend-X / extend-Z icons in GizmoWallEdition's cursor row are
billboarded toward the camera; without orientation polish they
always point in the same screen-space direction regardless of which
wall endpoint the click will move (or whether the cursor sits above
or below the wall top). New helper mirrors the icon's local-X (extend-X)
or local-Y (extend-Z) axis so each arrow points toward the end it
will move:
* Extend-X: walk wall midpoint to figure out which endpoint stays
fixed (cursor past midpoint → ATSTART stays; cursor before midpoint
→ ATEND stays). Project the fixed endpoint into screen-space and
flip the arrow when the gizmo's anchor sits on the same side.
* Extend-Z: flip when the cursor is below the wall top (within
EXTEND_FLIP_EPSILON tolerance).
Called once per resolved cursor gizmo from
``GizmoWallEdition._update_cursor_gizmos``, after the gizmo's
``matrix_basis`` is set by ``gizmo.billboarded_at``. Reuses
``gizmo.should_flip_extend_arrow`` + ``EXTEND_FLIP_MIRROR_X/Y`` +
``EXTEND_FLIP_EPSILON`` already on tool.
Generated with the assistance of an AI coding tool.
GizmoWallJoinIntersection's unjoin only fires when exactly two walls
are selected and surfaces one icon at their shared corner — useless
when the wall has 3+ joins and the user wants to disconnect just one.
* UnjoinWallPathConnection: surgical counterpart to UnjoinWalls.
Disconnects the active wall from a single partner wall identified
by IFC GlobalId (invariant under Blender-object renames + file
save/reload + undo). Walks both inverse arrays of the active wall
for the specific IfcRelConnectsPathElements joining the pair —
matches DumbWallJoiner.split's pattern and avoids disconnect_path's
direction-sensitivity. Resyncs both walls' draft props after the
recreate_wall pass.
* GizmoWallUnjoinSingle: activates on exactly-one selected
LAYER2 wall. Preallocates a pool of 16 unjoin icons (Blender forbids
gizmo allocation outside setup(); ATSTART + ATEND + ATPATH rels are
rarely more than a handful). Per-frame, iterates _iter_path_connections,
positions one billboarded icon at each join via
tool.Wall.path_connection_location_world, and hides the rest. Each
visible icon's bound operator carries the partner GlobalId, so a
click removes only that one rel.
* model/__init__.py: register both classes alphabetically.
Mutually exclusive with GizmoWallJoinIntersection via poll() — that
group requires len(selected) == 2; this one requires 1.
Generated with the assistance of an AI coding tool.
The single-wall unjoin gizmo needs to enumerate every
IfcRelConnectsPathElements a wall participates in, regardless of which
side of the rel the wall was authored on, and place an icon at each
join's physical location. Two helpers carry that work:
_path_connection_location_world wraps core.compute_path_connection_location
at the Vector boundary. _iter_path_connections walks ConnectedTo +
ConnectedFrom, normalises orientation to (other, self_ct, other_ct),
and filters non-wall partners + None refs so per-frame gizmo positioning
survives malformed IFC.
Generated with the assistance of an AI coding tool.
The chunk priority metric is now the 2D projected pixel area of the
chunk's AABB on screen — 8 corners projected through view-projection,
2D axis-aligned bbox of the projected points, clamped to viewport.
This replaces the prior bounding-sphere-radius² metric, which was a
3D approximation: it treated a 322 × 55 × 5 m slab as a 163 m sphere,
giving it the same huge priority face-on or edge-on. The new metric
genuinely answers "what would this chunk's AABB cover if rendered
solid given the current camera and viewport."
Newly-loaded chunks get a 30-frame grace period at full priority
(visibility_history floor temporarily forced to 1.0). Without it,
just-loaded chunks crashed to history=0 → effective priority = pri ×
0.05 → immediately reverse-swapped by the chunk they displaced.
Cycle starved the per-frame load budget so candidates ranked below
the cyclers never got attempted. 30 frames = HISTORY_ALPHA's time
constant — enough for visibility_history to develop meaningfully.
EVICT_PRIORITY_RATIO bumped 1.21 → 2.0 to suppress more swap noise
between similar-priority chunks.
Interactive heartbeat log added: every render in non-bench mode prints
[frame] with fps, ms, obj, sub_draws, hiz_rej, cull, stream, chunks
breakdown (resident/frustum/total + missing count), VRAM, model count.
Every 30 frames when something's missing, also dumps:
- top 8 models by missing chunk count
- top 20 missing chunks by priority (with AABBs)
- bottom 5 residents by effective priority
- all chunks of brace.ifc (one-off diagnostic, hardcoded
for the brace-visibility investigation)
The heartbeat made the streaming bug visible: a brace model that
isolation-loads correctly is missing in the full set because slabs
covering more pixels win the priority contest. Per-model fairness or
manual pinning are the remaining options if pixel-area + grace +
hysteresis isn't enough — left for follow-up so the user can decide
based on real testing.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
After a one-shot wall IFC mutation (unjoin / split / merge / extend /
join-at-corner …) the always-visible gizmos on the OTHER side of the
join can be left reading stale ``BIMWallProperties`` — the IFC
geometry moved but the draft props that drive the gizmo handles still
point at the pre-mutation numbers, so a subsequent edit-mode enter
shows the wall at its old length / position.
* New ``_maybe_resync_wall_props_from_ifc(obj)``: re-primes a single
wall's draft props from current IFC, with guards for non-walls,
non-parametric walls, and walls in an active draft session (the
draft is then the source of truth, not IFC). Must run from an
operator ``_execute`` — ID writes from gizmo refresh raise.
* New ``_resync_walls_after_mutation(objs)``: iterates the above
across a selection.
* Six existing mutation operators gain a resync call after their
``core.*`` / ``DumbWallJoiner`` mutation completes:
UnjoinWalls, ExtendWallsToUnderside, ExtendWallsToWall, SplitWall,
MergeWall, JoinWallsIntersection. MergeWall resyncs only the
surviving wall — the active wall is the deletion target.
Generated with the assistance of an AI coding tool.
Bundle of bugs surfaced when exercising the new gizmo framework
end-to-end in a live Blender session after the
bim/module/drawing/gizmos.py refactor + TypeAccessor/CycleType/PickType
mixins landed.
Register / annotation resolution
* parametric_lifecycle.py: hoist `entity_instance` import out of
TYPE_CHECKING so typing.get_type_hints resolves the
Callable[[entity_instance], bool] annotation at operator registration
(CycleDoorType, CycleWindowType, CycleStairType failed with NameError).
Clarify the INTERFACE return contract on the picker entry-point so
readers see why the gizmo step stays off the undo stack.
Framework callable contracts
* model/wall.py, door.py, window.py, stair.py: migrate `props_getter`
and `element_checker` from bl_idname strings to bound classmethods
on tool.Model / tool.Parametric. BaseParametricGizmoGroup.get_props
expects a callable; the string form raised TypeError on first
gizmo poll.
* model/door.py, model/stair.py: drop the dead `prop_path=` operator
kwarg from create_arc_gizmo / create_icon_gizmo call sites. The
framework helper blindly setattrs every kwarg onto the operator's
OperatorProperties, but ToggleDoorSwing / ToggleStairProperty don't
declare prop_path — the setattr raised mid-setup_element_specific_gizmos,
so self.gizmo_door_type / self.lock_gizmo never got assigned and
every subsequent draw_prepare tornadoed AttributeError. Nothing
reads op.prop_path anywhere; the kwarg was dead data.
Dispatcher operators
* model/array.py: add EnableEditingParametric (the framework pen-icon
dispatcher that routes to a per-feature edit operator by bl_idname
string) and AddArrayFromFeatureEdit (binds the framework's array
icon to bim.add_array on the current parametric draft).
* model/__init__.py: register both new operators.
Per-frame robustness
* drawing/gizmos.py: guard BaseParametricGizmoGroup.draw_prepare with
is_setup_complete() — matches the existing guard in refresh() and
in BaseSchematicGizmoGroup.draw_prepare(). Defense-in-depth: when
any subclass's setup raises mid-way, draw_prepare now no-ops cleanly
instead of per-frame AttributeError-tornadoing on whatever attribute
the failed setup phase was meant to populate.
* model/decorator.py: guard ProfileDecorator.__call__ against
context.active_object is None. The decorator is a per-frame
viewport draw handler; deselecting or deleting the active object
while it's installed crashed on obj.mode access. Treat None the
same as "no longer in edit mode" — uninstall + fire the exit
callback if present.
* geometry/data.py: ViewportData.load() populates `data` before
flipping `is_loaded`, so a raise from cls.mode() no longer leaves
the class flag-set but data-empty for subsequent reads.
Generated with the assistance of an AI coding tool.
The previous chunk-plan sorted meshes by lexicographic (z, y, x)
centroid — effectively a 1D Z-slab traversal. On a typical IFC
building (50 × 50 × 100 m), a 16-MB chunk's 80-ish meshes spanned
roughly 50 × 50 × 0.5 m. On a city federation it was much worse:
the first chunk grouped ground-floor stuff from every building,
spanning the entire scene horizontally. Per-chunk AABBs that wide
make frustum / contribution / HiZ rejection useless (every chunk
"overlaps the frustum" by virtue of spanning the whole scene).
3D Morton (Z-order) interleaves bits of quantised (x, y, z)
centroids, so consecutive items in the sorted order cluster in all
3 axes — chunks become tight 3D voxels of the model. Prerequisite
for the contribution-aware eviction priority (task #25) to actually
discriminate near and far chunks.
21 bits per axis = ~2 M bins per axis, sub-millimetre precision on
a kilometre-scale scene. Both apply paths (streaming and non-
streaming) share the same sortMeshIdsByMorton helper.
Benchmark unchanged (~47 fps avg, 20 ms cull, 0.3 ms stream) — the
distance-based evictor still keys on chunk centres, which moved
slightly under Morton but not enough to materially shift residency.
The user-visible win comes from the next commit, which switches
priority to screen-space contribution × HiZ history — both of which
need today's tight AABBs to mean anything.
Pixel-identical to non-streaming on basic.ifc on both paths.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Many Vulkan drivers cap a single VkDeviceMemory allocation at exactly
maxStorageBufferBindingSize (NVIDIA: 2 GB on consumer GeForce) or
refuse big contiguous allocations once heap is fragmented. The old
addSubBuffer gave up at the first refusal, latching growth_disabled_
— so on a 4 GB GeForce we extracted 2 GB and called it done.
The wgpu-mem-probe tool (feab05650) showed the driver actually grants
~3 GB total across multiple sub-buffers — invariant under allocation
pattern (2+1+small, 3×1 GB, 6×512 MB, 12×256 MB all land at 3 GB).
The cap is the hardware/desktop, not the request size.
addSubBuffer now starts at last_growth_size_ (initially
per_sub_buffer_capacity_, decays as the driver refuses larger sizes)
and halves on failure inside a single call. Stops at a 64 MB floor;
below that the per-sub-buffer bookkeeping cost (free list, bind
groups) isn't worth it. growth_disabled_ now latches only when even
64 MB is refused — a true hardware ceiling, not just "the first
attempt didn't fit."
pool_can_fit gains a next_growth_size_bytes() accessor to stay
honest about how big a future sub-buffer can be after the driver
has refused larger sizes.
Measured (big federation, --streaming, close camera):
pool capacity: 2048 MB → 2688 MB (2 GB + 512 MB + 128 MB)
VRAM resident: 2155 MB → 2800 MB (whole scene fits, no eviction)
avg fps: 42 → 53
stream time: 2.5 ms → 0.1 ms (no churn — working set is stable)
On larger GPUs (8 / 16 / 24 GB workstations) the same code extracts
proportionally more (e.g. 4 × 2 GB on a 10 GB+ card).
The GL backend's higher "4+ GB resident" claim is overcommit into
host RAM — explicit Vulkan/wgpu memory management deliberately
doesn't paper over that, and the wgpu-mem-probe data confirms it
isn't recoverable.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
New headless wgpu probe app — no Qt, no surface, just initializes a
device and stress-tests buffer allocations. Reports:
1. Adapter + device limits (maxBufferSize, maxStorageBufferBindingSize).
2. Single-allocation probe: descending sizes, each released, finds
the largest single buffer the driver will grant.
3. Cumulative probe: halve-on-failure, finds total VRAM the runtime
will let us park behind one device across multiple sub-buffers.
4. Fixed-size cumulative probe: 1 GB / 512 MB / 256 MB uniform sizes,
to detect whether the "big-first" strategy leaves VRAM on the table.
Findings on a GTX 1650 (4 GB physical) + wgpu-native + Vulkan:
- maxStorageBufferBindingSize = 2 GB (driver cap, not wgpu-native).
- Any single storage buffer > 2 GB is REFUSED.
- Total available across N sub-buffers = ~3 GB, INVARIANT under
allocation pattern (2+1+0.06, 3×1 GB, 6×512 MB, 12×256 MB all
reach 3.00 GB). Driver hands out a fixed VRAM slice; pattern
doesn't matter.
- Remaining ~1 GB is held by the desktop compositor + OS.
- GL's higher "4 GB+ resident" claim is overcommit into host RAM,
which wgpu/Vulkan don't do.
The +50% (2 → 3 GB) improvement is real and worth chasing — a
follow-up halve-on-failure addSubBuffer in WgpuBufferPool will
extract that on this card. On 8/16/24 GB GPUs the same code gets
us proportionally more.
Build: ninja -C build-viewer-wgpu WgpuMemProbe
Run: ./build-viewer-wgpu/wgpu-mem-probe/WgpuMemProbe
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The sync chunk-read on the render thread was causing 100-300 ms spikes
during orbit whenever a new chunk needed to scatter-gather its mesh
bytes from disk. p99 was 326 ms on the close-camera benchmark.
New WgpuStreamingThread: one worker thread with a condvar-protected
request/result queue. driveStreamingLoads becomes drain-then-enqueue:
1. Drain any results the worker pushed since last frame. For each,
pool-allocate slices + queueWriteBuffer + build the chunk bind
group (still main-thread because wgpu queue ops aren't thread-safe).
2. Walk visible non-resident chunks (sorted by distance), evict to
make pool room, and enqueue the request. Chunk gains is_loading
flag to prevent re-enqueueing while in flight.
loadChunkBytesAndUploadGpu becomes the sync fallback path, used only
when a screenshot is pending — the deferred-capture wait would
otherwise let the window manager re-layout the window between frames
and the test framework would capture at the wrong size. Normal
streaming always goes through the worker.
Bench warm-gate / requestUpdate gating updated to consider
streaming_thread_.inFlightApprox() so we don't declare "converged"
while a worker read is still in flight, and the render loop stays
alive until the worker queue is empty.
Refactored loadChunkBytesAndUploadGpu into two helpers:
- makeChunkRequest: builds the worker request from chunk metadata
- applyStreamedChunk: pool.alloc + queueWriteBuffer + bind group
Both the sync and async paths share applyStreamedChunk.
Benchmark (big federation, --streaming):
close camera: avg 24 fps p99 47 ms (was 27/326)
default camera: avg 24 fps p99 46 ms (was 31/186)
stream time: ~2 ms (was 8-12)
cull is now the bottleneck (20 ms median) — task #17 (GPU compute
cull) is the next frontier.
Pixel-identical to non-streaming on basic.ifc on both paths.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
cullModelCpuCompute previously had two paths: a flat linear scan over
all instances (default), or a BVH-stack walk (--bvh, gated off because
it regressed on dense scenes — the BVH built per instance but its
interior-node AABBs spanned huge chunks of model so most subtrees
straddled the frustum and the walk overhead beat the rejection win).
With spatial chunk planning (commit 4d3617420) chunks ARE already a
one-level spatial partition of the model, with tight per-chunk AABBs.
So the same wholesale-reject behaviour falls out of just walking
m.chunks: frustum-test each chunk's AABB once, and on hit, iterate
its (new) instance_ids list. No per-node traversal overhead, no
dependency on rebuilding a BVH alongside the chunk plan.
Changes:
- Chunk gains an instance_ids vector, populated in both apply paths
alongside the per-chunk AABB accumulation.
- cullModelCpuCompute drops the if-bvh / else-linear-scan dichotomy
in favour of `for chunk: frustum-test then iterate c.instance_ids`.
- Per-model ModelBvh field, buildModelBvhOne call sites, BvhAccel.cpp
in CMakeLists, bvh_enabled_ field, and --bvh CLI flag all removed —
dead code now that chunk-cull subsumes them.
- BvhAccel.{h,cpp} stay in src/ifcviewer for the GL backend's use.
Benchmark (big federation, --streaming, close camera): avg 37 fps
(was 36) / median 53 (was 53). Same order on the metric — the
parallelism across models was already amortising frustum-check cost,
so the per-chunk early-out saves only fragments of cull wall time.
Real cull-perf win will come from chunk-level HiZ (potentially) or
GPU compute cull (task #17). What this commit really delivers is
architectural simplification + removal of a dead-but-not-dropped
code path.
Pixel-identical to non-streaming on basic.ifc.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Chunks are now grouped by world-space centroid instead of mesh-id
range, so each chunk's AABB tightly bounds its geometry instead of
spanning the whole model. Distance-based eviction can finally
distinguish the near corner of a skyscraper from the far corner.
Algorithm:
1. Compute each mesh's centroid = mean of its instances' world AABB
centres.
2. Sort mesh indices lexicographically by (z, y, x) centroid. Stable
sort keeps mesh-id order as tiebreaker for instanced repeats.
3. Greedy-pack sorted meshes into chunks ≤ WGPU_CHUNK_VERTEX_BYTES_LIMIT.
4. Each Chunk stores its mesh_ids list; the per-mesh layout (chunk_local
base_vertex / ebo_first_u32) is computed by walking the list at plan
time.
Loader: chunk vertex/index bytes are no longer file-contiguous, so
streaming uses new multi-range read paths
(readSidecarVertexRanges / readSidecarIndexRanges). Each range list
is sorted by file offset and adjacent ranges coalesced with a 64 KB
gap tolerance — on the close-camera benchmark this brings the
per-chunk seek count back down to ~mesh-id-grouping levels, so the
spatial sort costs ~nothing on I/O while delivering tighter AABBs.
Non-streaming applyCachedModel mirrors the spatial plan but gathers
from in-memory data.vertices / data.indices via per-mesh
queueWriteBuffer calls at chunk-local offsets.
Chunk struct drops vertex_byte_offset and index_first_u32 (no longer
meaningful — each chunk is N scattered ranges). vertex_byte_size and
index_count stay as aggregates for pool sizing + eviction math.
Tuning: kept WGPU_CHUNK_VERTEX_BYTES_LIMIT at 128 MB. Tried 8 MB and
32 MB; both gave tighter AABBs but the scatter-gather I/O cost blew
up because the per-frame load count grows linearly as chunks shrink
(orbit shifts the working set faster across finer chunks). 128 MB +
coalescing is the empirical sweet spot pre-v14. Once sidecar v14
re-orders bytes on disk to match spatial chunks, we can drop the
limit to ~8 MB for sharp eviction without re-paying the seek cost.
Benchmarks (big federation, --streaming):
close camera: avg 36 fps median 53 (was 35/49) — parity
default camera: avg 33 fps median 47 (was 40/49) — small regression
likely from increased coalesce overhead on more-
scattered orbit traversals; will resolve with v14.
Pixel-identical to non-streaming on basic.ifc on both paths.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Five interlocking fixes that take --streaming on the big federation
scene from "5 fps + endless flicker + infinite cold-load" to a
stable 35-49 fps with a converged working set.
1. Multi-sub-buffer WgpuBufferPool. Pool now grows lazily by adding
sub-buffers of per_sub_buffer_capacity_ when alloc demand exceeds
existing free runs. Each Slice carries (buffer, offset, size,
sub_idx). On driver refusal of addSubBuffer, growth_disabled_
latches so subsequent allocs don't keep retrying and log-spamming.
pool_can_fit consults can_grow() to know when growth could rescue
a candidate vs when eviction is the only path.
2. Split cull / stream benchmark timers. The previous "cull[wall]"
metric was actually cull + driveStreamingLoads, blaming the wrong
subsystem (~170 ms of "cull" was synchronous disk I/O).
3. frustum_visible_count on Chunk, populated in cullModelCpuCompute
right after the per-instance aabbInFrustum check. driveStreamingLoads
now keys residency on this instead of total_visible_draws (which
includes contribution + HiZ). HiZ visibility flips frame-to-frame
as occluders shift; using it for residency caused chunks to be
evicted then immediately re-loaded, every frame, even with a
stationary camera — both the perf cliff and the visible flicker.
4. Distance-sorted candidates in driveStreamingLoads. Walk the
non-resident frustum-visible chunks in distance order (closest
first). With sorted processing, evict_farthest_than converges
monotonically: each swap replaces a far resident with a closer
candidate; once the next candidate is farther than every
remaining resident, the loop exits. Without sorting the loader
visited candidates in model/chunk-id order, swapping random
chunks every frame without ever converging.
5. 10% eviction hysteresis (EVICT_DIST2_RATIO = 1.21). On scenes
where many chunks are clustered at similar distance from the
camera (e.g. several chunks all ~370 m away), naive
"evict any resident strictly farther than candidate" triggers
sub-meter swaps every frame, never resting. Requiring the victim
to be 10% farther in linear distance kills these cycles while
still allowing genuine "much closer" candidates to evict.
Plus: latched bench_warm_done_ on the cold-load gate, with a
5-frames-of-zero-loads convergence test (default-camera big scene
converges in 20 frames) and a 600-frame timeout fallback that prints
exactly once.
Measured on the test federation (111 sidecars, ~3 GB raw, 1 M
instances) with the user's close-in camera:
- avg 35 fps (was 5), median 49 fps (was 7)
- cull 19 ms (now the bottleneck), stream 5-8 ms (was 172)
- p99 184 ms — occasional big-chunk load on the render thread;
background-thread I/O would smooth that out as a follow-up.
With the default wide camera:
- avg 40 fps, converges in 20 frames, residency grows naturally
from 59 → 76 chunks as orbit shifts the frustum.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Drops the per-machine "guess the OOM ceiling" budget knob in favour of
a single buffer pool whose capacity is *probed* at device-init time.
The runtime answers the question: descend from min(maxBufferSize, 4 GB)
through OOM error scopes, accept the largest size that allocates
cleanly. On a desktop wgpu-native v29 box this lands at 2 GB; on
browser-class platforms it'll land at 256 MB – 1 GB depending on the
implementation. Same code path either way.
Architecture:
- WgpuBufferPool (new): single WGPUBuffer + free-list sub-allocator
with adjacent-range coalescing and first-fit. 256 B alignment for
storage-binding offsets.
- Chunks now hold (pool_vertex_offset, pool_vertex_size) and
(pool_index_offset, pool_index_size) instead of per-chunk WGPUBuffer
handles. Load = pool.alloc + queueWriteBuffer. Unload = pool.free.
- Bind groups bind pool_.buffer() at the chunk's specific (offset, size)
for both the vertex and index storage bindings.
- Eviction queries pool.largest_free_run_bytes() instead of a tracked
budget; the two-phase LRU/distance evictor's policy is unchanged.
What this fixes:
- No more gpu-alloc-rs fragmentation OOM: one VkDeviceMemory block
instead of N per-chunk blocks with rounding overhead. On the test
dataset (~3 GB on disk, 562 k visible instances) the wgpu backend
now runs through to render without OOM at any point.
- No --streaming-vram-mb knob, no hardcoded budget constant, no
per-machine calibration. The pool size adapts to whatever the
runtime grants.
Notes:
- Error scope probing: wgpu-native v29 classifies "Not enough memory
left" as WGPUErrorType_Validation, not OutOfMemory. We push both
filters (nested) and treat either firing as probe failure.
- The 4 GB probe cap is principled, not magic: above that, wgpu-native's
advertised maxBufferSize is sometimes a sentinel (1 TB) that just
forces wasteful halving steps. 4 GB is the largest buffer any
realistic WebGPU implementation will grant a single allocation today.
- Pool destroy()/release happens after model release in shutdown() so
the underlying buffer outlives every bind group that references it.
Follow-ups: spatial chunking (task #22) for finer eviction granularity;
cull perf needs work at 100+ models / 1M+ instances (separate from
streaming concerns).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Defers index buffers per-chunk (alongside vertex bytes) so streaming
fully delivers on its "don't load until visible" contract — the previous
per-model index buffer was upfront-loaded and tipped scenes >~1.5 GB into
allocator OOM at frame 1.
Adds residency tracking + a two-phase evictor: (1) drop LRU non-visible
chunks first, (2) if everything resident is visible-this-frame, drop the
farthest-from-eye chunk only when the candidate to load is closer. This
gives monotonic convergence to "closest visible chunks fit the budget"
instead of "first 4 win, rest never load."
Default budget set to 1 GB — explicitly a stopgap, documented inline.
The per-machine OOM ceiling on wgpu-native (caused by allocator
fragmentation from one VkDeviceMemory per createBuffer call) cannot be
solved by tuning this knob. The proper fix is a probed single-pool
buffer with sub-allocation, tracked under task #16.
Caveat: LOD1 indices are now force-disabled when chunking — per-chunk
buffers only carry LOD0. Re-enabling needs LOD1 to participate in the
chunk plan.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The OOM fix for vertex storage. With --streaming, chunks now load on
demand:
- After cull determines which chunks have visible draws, driveStreamingLoads
walks non-resident chunks with total_visible_draws > 0 and brings up
to MAX_STREAMING_LOADS_PER_FRAME (currently 4) into residency.
- Each load: readSidecarVertexChunk → createBufferWithData →
buildChunkBindGroup → is_resident = true. Same frame's draw loop
picks up the newly-built bind_group and renders the chunk.
- If more non-resident-but-visible chunks remain, requestUpdate is
called so the load loop keeps running until the visible set is fully
resident.
Per-chunk bind group construction refactored out of buildModelBindGroup
into a buildChunkBindGroup(m, chunk_idx) helper so the streaming loader
can build one chunk at a time as it arrives.
4 chunks/frame × 60 fps = 240 chunks/sec ingestion. A 200-chunk scene
fully resides in ~1 second of motion. Off-screen chunks never become
resident, never pay vertex-storage VRAM — that's where most of the OOM
fix lands.
Verified on basic.ifc: pixel-identical to non-streaming. On the user's
real 111-model / 1M-instance scene: all metadata loads succeed (was
OOM before), then loader runs but **indices are still loaded upfront
(1.5 GB!) so OOM still hits when vertex chunks start adding on top.**
Per-chunk index deferral is the next commit.
Eager-no-evict policy (per the design conversation): chunks stay
resident once loaded. LRU eviction lands in a follow-up if a workload
proves it necessary.
This completes the 4-commit stage-1 series for task #16. Stage-2:
defer indices, deferred mesh/instance storage if needed, async worker
thread.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Wires the metadata-only reader (commit 1) through a parallel streaming
load path. With --streaming on:
- loadSidecar routes through readSidecarMetadataOnly: reads header +
mesh dict + instance dict + georef + elements upfront. Skips
vertex bytes entirely.
- applyCachedModelStreaming computes the same chunk plan as the
non-streaming path, allocates the small per-chunk buffers
(visible_draws + prefix_sums + per_chunk_uniform), allocates the
model-shared mesh + instance + index buffers, but leaves each
chunk's vertex_storage NULL and is_resident=false.
- Stores streaming_file_path + vertex_section_offset on the model so
the per-frame loader can range-read chunks later.
- Computes per-chunk world AABB by walking instances → mesh → chunk;
used by both cull (chunk-level frustum reject, future) and the
streaming loader (proximity-prioritised fetch, future).
Index buffer is still loaded upfront in stage 1 (small relative to
vertex data: ~1/2 of vertex bytes on real scenes). Stage 2 may defer
it too if measurements suggest it's worth the extra plumbing.
Render + pick already gate on c.bind_group (null when non-resident),
so the existing guards correctly skip non-resident chunks without
further changes.
With this commit alone, --streaming mode shows an EMPTY scene (just
background colour) because no chunk ever becomes resident. Commit 4
adds the per-frame loader that triggers chunk load when cull marks
them visible — that's the commit where rendering kicks in and the
OOM fix actually lands.
Default behaviour (no --streaming): legacy synchronous full-load.
Pixel-identical to the prior commit on basic.ifc.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Foundation for streaming. Adds to each Chunk:
- is_resident (default true; streaming flips false initially)
- vertex_byte_offset / vertex_byte_size in the sidecar file
- aabb_min / aabb_max world-space chunk bounds (used by future cull
and streaming priority)
Plus on the model:
- streaming_file_path (non-empty = streaming path was used)
- streaming_vertex_section_offset (where the chunks live in the file)
All fields default to backward-compatible values: is_resident=true,
streaming_file_path empty. The existing non-streaming applyCachedModel
sets up a Chunk with is_resident=true (implicit) and ignores the
streaming fields, so no behaviour changes yet.
Commit 3/4 wires the metadata-only reader from (1/4) through a new
applyCachedModelStreaming path that flips is_resident=false initially;
commit 4/4 adds the per-frame loader that brings chunks resident on
demand. This commit is verified pixel-identical to the previous render
on basic.ifc.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
First foundational piece for task #16. WgpuStreamingLoader exposes:
- readSidecarMetadataOnly(path): reads v13 header + mesh dict + instance
dict + georef + elements + string table from disk. Skips the bulky
vertex and index byte sections, recording their on-disk offsets so
they can be range-read later (per-chunk, on demand). The file handle
is closed before return.
- readSidecarVertexChunk / readSidecarIndexChunk: open + fseek + fread
for a byte range. Synchronous; intended to be called from a worker
thread for true async streaming or the main thread for stage-1
on-demand load.
No format change yet — operates on existing v13 sidecars. v14 with an
explicit per-chunk TOC arrives in a follow-up; this layer abstracts
the chunk boundaries so the upgrade stays internal.
No integration with existing applyCachedModel — that's commit 3/4.
Build verifies the API compiles and links into IfcViewerWgpu.
Commits in this series:
1/4: metadata-only reader (THIS)
2/4: per-chunk residency state on WgpuModelGpuData
3/4: --streaming opt-in path through applyCachedModel
4/4: per-frame chunk-on-visible loader (the OOM fix)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Three concerns bundled into one cohesive refactor of gizmos.py
(splitting them surgically requires intermediate commits with
duplicate same-named classes that Python can't parse):
1. Framework primitives — StaticTrisGizmoMixin + TexturedQuadGizmoMixin
replace the older TrisGizmoMixin. New module-level helpers:
_get_static_tris_shader / _get_static_tris_batch / clear_static_
tris_cache for cached GPU batch reuse, _draw_outline_and_body for
the shared outline-then-body render path, draw_tris_with_outline
as the public wrapper. billboarded_at(world_pos, billboard_rot,
scale) is the canonical billboard-matrix helper; should_flip_extend_
arrow encapsulates the view-aware mirror decision for extend
gizmos; get_warning_color_from_prefs reads the user's warning
color.
2. Config classes — BaseValueGizmoConfig (shared visibility + dimension-
text contract), CountGizmoConfig (array N indicator),
DimensionGizmoConfig (length / height / depth labels), IconActionConfig
(icon-only gizmos that invoke an operator on click). DimensionRenderer
draws the actual numeric label using BLF.
3. Icon classes — each rewritten on StaticTrisGizmoMixin so they share
the cached GPU batch + outline-then-body render path:
GizmoLockOpen / GizmoLockClosed (replacing the single-state
GizmoLock), GizmoArc, GizmoFillet, GizmoWallCornerIcon,
GizmoWallTeeIcon, GizmoPen / GizmoValidate / GizmoCancel (the
parametric-edit triad), GizmoPlus / GizmoMinus / GizmoTrash,
GizmoArrayParent / GizmoArrayAll / GizmoArrayLayerIndicator (array
context indicators with a small digit-rendering helper for the "xN"
count label), GizmoMerge / GizmoSplit / GizmoUnjoin (wall-join
icons), and GizmoMenu (textured-quad icon-action menu trigger).
The legacy TrisGizmoMixin, GizmoLock, and DimensionDrawConfig are
removed; downstream callers in subsequent PR4 commits swap to the
new mixin and config classes when their feature operators land.
CycleTypeMixin / PickTypeMixin / TypeAccessorBase live in
bim.parametric_lifecycle (previous commit). The three mixins are
re-exported from gizmos.py here so feature-module access via
``gizmo.<MixinName>`` keeps working until PR5 cleanup drops the
re-exports.
bim/module/drawing/__init__.py is updated in the same commit to
register the 11 new gizmo classes (GizmoLockOpen / GizmoLockClosed /
GizmoFillet / GizmoWallCornerIcon / GizmoWallTeeIcon / GizmoTrash /
GizmoArrayParent / GizmoArrayAll / GizmoArrayLayerIndicator /
GizmoUnjoin / GizmoMenu) — without that, the new classes exist in
gizmos.py but aren't usable as bpy gizmo types.
Generated with the assistance of an AI coding tool.
Three operator mixins for type-selection ops on parametric features
(door type-cycle, window type-pick, stair type-cycle, railing
type-pick, roof type-cycle, etc.). Each shares the same contract:
* ``element_checker`` validates the active object is the expected
IFC type
* ``props_getter`` resolves the BIM<Name>Properties group
* ``type_literal`` is the Literal type whose args drive the enum
* ``type_attr`` is the PropertyGroup field to read/write
* ``skip_element_check=True`` bypasses element validation (for
operators that target a non-IFC context)
CycleTypeMixin shift-click reverses direction (forward by default).
PickTypeMixin opens a popup menu and routes the picked value
through execute() so F6 redo / EXEC_DEFAULT reach the apply path.
The PickType modal-handler dance waits for LEFTMOUSE release before
opening the menu when invoked mid-click (e.g. from a gizmo's
target_set_operator) so Blender's drag-through-pick gesture doesn't
commit an accidental item.
Ships standalone — the next commit's gizmos.py framework refactor
re-exports these names from bonsai.bim.parametric_lifecycle so
gizmo modules can spell ``gizmo.CycleTypeMixin`` / ``gizmo.PickTypeMixin``.
Concrete operator subclasses land in subsequent PR4 commits per
feature (door / window / stair / railing / roof).
Generated with the assistance of an AI coding tool.
Two paired test files for the framework infrastructure landed
earlier in this PR.
test_decorator_cache.py (11 tests):
* The 4-hook invalidation list (depsgraph_update_post + undo_post +
redo_post + load_post) is symmetrically managed by
install_decorator_cache_handlers / uninstall_decorator_cache_handlers.
A future edit that drops a hook from one side without the other
would land as a Blender segfault when a cached bpy.types.Object
ref outlives its underlying ID block — the regression must surface
as a test failure first.
* install is idempotent (calling twice doesn't double-register).
* uninstall when not installed doesn't raise.
* The bump handler accepts Blender's variadic args.
* The depsgraph predicate gates correctly: bumps on Object geometry
or transform updates, silently skips on Material / NodeTree / Image
updates (which would otherwise rebuild every cache on every node
edit).
* TokenCache.get_or_compute short-circuits on key+token match and
recomputes when the token bumps.
test_undo_resync_parametric_drafts.py (3 tests):
* UNDO_REGENERATORS keys must all be in tool.Parametric.EDIT_TYPES.
A typo would silently no-op on Ctrl+Z, restoring the desync the
helper is meant to prevent.
* The dispatcher skips objects with no active parametric edit
(undo_post fires for every undo, most of which touch zero drafts).
* The dispatcher silently skips parametric types that have no
UNDO_REGENERATORS entry (door / window / array are IFC-derived
with no draft preview mesh — they don't need a regenerator).
Mocks use spec=bpy.types.Depsgraph / spec=bpy.types.DepsgraphUpdate
/ spec=tool.parametric.ParametricObject so typos in mocked-attribute
access fail loudly (CLAUDE.md test discipline).
Generated with the assistance of an AI coding tool.
Splitting a wall through a door orphaned the door (door.FillsVoids
became empty). The fill rel was being reassigned by setting its
RelatedBuildingElement slot — schema-wise that's the filling slot, not
the wall slot — so when remove_feature deleted the old opening it
also cascade-removed the rel. Transferring via RelatingOpeningElement
keeps the rel pointing at the new opening so the door stays
associated. Pre-existing bug from 5a6476a57, surfaced by ef144dce2.
Generated with the assistance of an AI coding tool.
DumbWallJoiner.split assigned openings by projecting the opening's
centre-point onto the wall axis, so any opening whose footprint
straddled the cut was silently dropped from whichever wall its centre
missed. Now the full axis-projected extent (via ifcopenshell.geom.
create_shape) drives the assignment; for filled openings whose void
straddles the cut, a pure-void copy is added back to the neighbour
wall so its body is also cut.
Generated with the assistance of an AI coding tool.
Five inline copies of the same defensive pattern lived across
``tool/parametric.py``, ``bim/parametric_lifecycle.py``,
``bim/module/model/preview_base.py`` (twice), and as a near-twin
in ``tool/blender.py:update_all_viewports`` itself.
``tool.Blender.update_all_viewports`` already covered the
``tag_redraw`` job but used an ``assert context.screen`` that would
raise during background-mode operators or early-load_post calls
where ``screen`` legitimately is None. Relax to a defensive
``getattr(context, "screen", None)`` + silent return so the helper
fits every caller's needs, then collapse the 4 inline copies to
single calls.
Net -9 LOC. The helper now describes its contract ("silent no-op
when no screen attached") rather than naming specific callers, so
moving a caller doesn't rot the docstring.
Generated with the assistance of an AI coding tool.
Ctrl+Z / Ctrl+Shift+Z on an in-progress parametric draft (wall /
stair / roof) used to leave the preview mesh frozen in its
pre-undo shape — the IFC mutation rolls back but the bmesh built
from draft props doesn't repaint.
Add a registry of per-type regenerator functions
(``UNDO_REGENERATORS``) that re-build each type's preview mesh
from its current props. The dispatcher
``resync_parametric_drafts_after_undo`` walks all objects, skips
any without an active parametric edit, looks up the regenerator
by feature name, and calls it. Tagged 3D viewports for redraw.
Types without an entry (door / window / railing / etc.) are
intentionally absent — they're IFC-derived, so the undo's
representation rollback + next-frame refresh already repaints
correctly without a draft-side regenerator.
Undo/redo wiring is self-installed by
``bonsai.bim.parametric_lifecycle``: a ``@persistent``
``_resync_on_undo`` callback dispatches into the registry, and
``install_parametric_lifecycle_handlers()`` /
``uninstall_parametric_lifecycle_handlers()`` append/remove it
from ``bpy.app.handlers.undo_post`` and ``redo_post``.
``bim/__init__.py``'s ``register()`` calls the install function
*after* the central ``handler.undo_post`` / ``redo_post`` appends
so the regenerators see restored IFC state — ``bpy.app.handlers``
fire in append order. ``handler.py`` itself stays ignorant of the
parametric subsystem. The lazy function-local imports in each
regenerator break the addon-load cycle —
``bonsai.bim.parametric_lifecycle`` loads before
``bim/module/model/*``.
Generated with the assistance of an AI coding tool.
Three changes to the shared Enable/Finish/Cancel mixins:
1. Always-on drift triad on ParametricEditMixinBase. The base now
provides ``_handle_drift_on_enable`` / ``_handle_drift_on_finish``
/ ``_handle_drift_on_cancel`` classmethods, called from the
per-mixin ``_enable_one`` / ``_finish_one`` / ``_cancel_one``.
Pre-edit Blender-side translations commit to IFC on Enable
(apply_scale=False — only translation/rotation, not the user's
accidental scale), in-edit drag commits on Finish (apply_scale=True),
and Cancel restores the committed IFC placement via
``restore_or_rebaseline_placement``. Prevents the
"uncommitted drag disappears on Finish" and "preview snaps back
on Cancel" UX bugs.
2. ``_ParametricEditMixinBase`` renamed to ``ParametricEditMixinBase``
(public). Per-feature mixins that need to subclass directly
(e.g., when neither FeatureModifier nor PathPreserving fits)
can do so without reaching into a private name.
3. ``_update_modifier_bmesh`` (PathPreserving) renamed to
``_restore_viewport_after_cancel``. The old name was inaccurate
for subclasses that load a different IFC representation on
Cancel rather than rebuilding a bmesh preview from props.
Plus two polish changes:
* ``_mark_type_thumbnail_dirty`` helper on the base centralises the
``ifcopenshell.util.element.get_type`` + thumbnail-mark pattern
that both mixins repeated inline.
* ``FeatureModifierEditMixin._cancel_one`` and
``PathPreservingEditMixin._cancel_one`` wrap the restore in
``try/finally`` so ``props.is_editing = False`` flips even on
partial restore failure. Without this, a Cancel that raised
mid-restore would leave the user locked out of the edit lifecycle.
* ``PathPreservingEditMixin._finish_one`` / ``_cancel_one`` skip the
pset commit + viewport rebuild when the draft equals the stored
pset (no-op Enable→Finish round-trip should not pollute the
representation list or burn an undo entry).
``FeatureModifierEditMixin._finish_one`` now routes the pset commit
through ``tool.Pset.write_bbim_data`` instead of inlining the
``createIfcText(json.dumps(...))`` + ``ifcopenshell.api.pset.edit_pset``
dance. Two test assertions updated to match.
Generated with the assistance of an AI coding tool.
Three concerns folded into ``load_post`` argue for separation:
1. Save-file invariants every load must re-establish (msgbus
subscription, owner-settings, thumbnail cache, draft-flag healing,
blend-warning flag, H5 lock probe).
2. User-preference-driven UI setup (toolbar, workspace, viewport
shading, panel hijack, snap defaults).
3. Viewport overlay sync (every decorator's install/uninstall).
Pull each into its own function (``_apply_save_file_invariants`` /
``_apply_user_preferences`` / ``_install_viewport_overlays``). The
``load_post`` callback becomes a 3-line orchestrator. Each phase
is independently call-able from tests and from PR4 features that
need to re-trigger one phase without the others.
Two new hooks land with the decompose:
* ``tool.Parametric.heal_stale_edit_flags()`` + ``discard_pending_previews(scene)``
fire in ``_apply_save_file_invariants``. The first clears
object-level ``BIM<Name>Properties.is_editing`` flags that lost
their backing IFC element across a load; the second clears
scene-level ``BIMPreviewProperties.<x>.is_active`` so saved
preview state never resurfaces with no UI to interact with it.
* ``install_decorator_cache_handlers`` / ``uninstall_decorator_cache_handlers``
wrap the decorator install/install pass in
``_install_viewport_overlays``. The bump handlers append to
``depsgraph_update_post`` + ``undo_post`` + ``redo_post`` +
``load_post`` so the previous commit's ``TokenCache`` in
``tool.System.get_decoration_data`` finally invalidates on
structural scene changes.
Generated with the assistance of an AI coding tool.
Shared helpers for Bonsai's Scene-level parametric preview flows.
Two PR4 features will consume this — MEP bend preview and wall
fillet preview — both following the same shape:
Enable<X>Preview — populates draft on Scene.BIMPreviewProperties.<x>
Gizmo<X>Preview — polls on is_active, surfaces tunable widgets
<X>PreviewDecorator — GPU lines while is_active is True
Finish<X>Preview — bpy.ops.bim.<verb>(...) with draft kwargs
Cancel<X>Preview — pure state reset
The module hosts the cross-cutting accessors (``get_preview_props``,
``is_preview_active``), lazy-closure factories for gizmo dimension
callbacks (``make_props_callback`` / ``make_dim_getter`` /
``make_dim_setter`` — defensive against missing scene / freed RNA
struct on file open / undo), the Enable-time IFC-placement sync
(``sync_uncommitted_moves``), and the Esc + load_post discard
machinery (``PREVIEW_CANCEL_OPS`` registry, ``try_cancel_active_preview``,
``discard_pending_previews``).
Ships standalone — the consumer features land in PR4 (preview
PropertyGroups, Enable/Finish/Cancel operators, gizmo groups,
decorators, Esc keymap binding). All accessors are defensive
against missing PropertyGroups / operators on v0.8.0 — calling
``discard_pending_previews(scene)`` from the next commit's
load_post hook is a no-op until PR4 attaches BIMPreviewProperties.
Generated with the assistance of an AI coding tool.
System decoration draws on every viewport refresh — the
``_build_decoration_data`` body walks every distribution element,
resolves connected ports, builds the vert/edge arrays for the GPU
batch. A bare call per frame burns time on an unchanged scene.
Add a single-entry cache keyed on ``(decorator_cache_token,
id(decorated_elements_set))``. Reads short-circuit when neither
component moved:
* ``decorator_cache_token`` from ``bim.decorator_cache`` invalidates
on depsgraph / undo / redo / load via the bump handler.
* ``id(decorated_elements_set)`` invalidates when
``SystemDecorationData.load()`` reassigns the set (e.g. when the
user changes the set of decorated systems via the panel).
The handler that bumps the token is installed in the next commit
(bim/handler.py decompose). Until then the token stays at 0, so
the cache only hits when ``id()`` also matches — degraded behaviour
during the bisect window but not incorrect.
Generated with the assistance of an AI coding tool.
New helper module for POST_VIEW decorators. Exports:
* ``get_decorator_cache_token()`` — global int counter consumers
include in their cache key so the value invalidates on structural
scene changes.
* ``_bump_decorator_cache_token()`` — ``@bpy.app.handlers.persistent``
callback that increments the token. Gates on the depsgraph payload
so animation playback / driver evaluation doesn't churn the token.
* ``install_decorator_cache_handlers`` / ``uninstall_…`` — idempotent
append / remove against depsgraph_update_post + undo_post + redo_post
+ load_post. Called once from ``bim.register`` / ``unregister``.
* ``TokenCache[T]`` — single-entry memoiser keyed on ``(caller_key,
token)``. Cached ``bpy.types.Object`` references can't outlive the
underlying ID blocks because any depsgraph / undo / load bumps the
token and forces a recompute.
This commit ships the module standalone. The next commits in this
PR wire it: tool/system.py adds the cache wrap on get_decoration_data
and bim/handler.py installs the bump callbacks. Until both land,
the module is intentionally dead code — keeps the diff narrow and
the commit history bisectable.
Generated with the assistance of an AI coding tool.
Five code paths in slim PR2 referenced symbols that don't exist in
v0.8.0's bim layer, raising at first call. Plus three type
annotations that ty flagged as unresolved.
1. tool/system.py:get_decoration_data — drop the cache layer that
keyed on a token from a bim/decorator_cache.py module. The cache
is dead-or-broken in slim: the depsgraph bump handler that would
invalidate the token lives in PR3's bim/handler.py decompose, so
the token stays at 0 forever. Either the cache never hits
(decorated_elements rebuilt → new id() per call) or returns
stale data (list reused). Revert to direct
`_build_decoration_data()` calls. PR3 reintroduces the cache
atomically: decorator_cache module + handler install + cache
wrap + tests. Keeps `_build_decoration_data` extraction
(cleaner than v0.8.0's monolithic version regardless of cache).
2. tool/spatial.py — add `get_host_element` + `get_host_wall`.
The interface stubs in `core/tool.py:1037-1038` were declared
but never implemented. `tool/duplicate.py:99` (object duplication
with fills) and `tool/model.py:1260` (array per-child opening
mirror) call these and would raise AttributeError.
3. tool/model.py:recreate_wall — drop the fillet-corner branch
that function-locally imports `regenerate_fillet_corner_wall`
from `bim/module/model/wall`. The function lands with PR4; fall
through to the straight-extrusion path preserves v0.8.0
behaviour for fillet walls until then. Tag FIXME(PR4).
4. tool/model.py — drop `get_pipe_segment_props` /
`get_duct_segment_props` accessors. Their return types reference
`BIMPipeSegmentProperties` / `BIMDuctSegmentProperties` which
land with PR4's prop.py; calling either accessor on v0.8.0 would
AttributeError on `obj.BIM<X>SegmentProperties`. Zero callers in
slim — PR4 reintroduces both accessors together with the
PropertyGroups they wrap. Also drops the matching TYPE_CHECKING
imports.
5. tool/blender.py:557 — `Mapping[type[ViewportDecorator], bool]`
needs the qualified `Blender.ViewportDecorator` because the
annotation is on a method INSIDE the same nested class; the
bare name doesn't resolve at type-check time.
6. core/tool.py Surveyor — drop the `obj: "bpy.types.Object"` /
`z: float` / `-> float` / `-> None` annotations on
`get_z_rotation` / `set_z_rotation`. The `@interface` decorator
wraps each method as `classmethod(abstractmethod(...))` at
import time, but ty doesn't track the wrap and flags every
call site as `missing-argument` plus the `pass` body as
`empty-body` against the declared return type, plus the
`bpy.types.Object` forward-ref as `unresolved-reference`.
Reverting to v0.8.0's untyped style (matching the sibling
`get_absolute_matrix(cls, obj)` stub) clears six ty errors at
the cost of zero runtime semantics — the abstract stubs only
serve as registry markers, concrete `tool.Surveyor.*` carries
the real signatures.
Generated with the assistance of an AI coding tool.
Stage 15 implementation lands but doesn't pay off as default-on. On a
562k-instance / 18-model scene with a centred camera, the BVH walk
adds ~10 ms of cull cost without rejecting enough subtrees to
compensate — every interior node's AABB straddles the frustum, so
descents go all the way to leaves anyway. Linear scan beats it by
that 10 ms.
GL's BVH works better mainly because they do full cull (frustum + HiZ
+ contribution) at every node — their per-test cost is lower (likely
SIMD-vectorised) and they get more subtree rejections. My current
impl does frustum-only at interior nodes (HiZ there cost more than
it saved on the smaller dataset).
For now, gate the whole BVH walk behind --bvh, default off. The
infrastructure (BvhAccel build at applyCachedModel, walk in cull,
release) stays in place so it's a one-flag toggle to measure either
side. Real default-on requires further tuning — see updated task #15.
Measured on 562k-instance scene:
--bvh on → 25.9ms total (cull 25.4ms)
--bvh off → 15.4ms total (cull 14.5ms) ← default
For comparison, GL on the same scene + camera:
GL → 18.2ms total (cull 8.5ms wall, multi-threaded BVH)
Net: wgpu beats GL by ~3ms total despite slower cull, because the
GPU side (no edge-pass cost, async HiZ readback, lean main pipeline)
gives back more than the cull deficit.
Also added task #17 (GPU compute-shader cull) as the asymptotic
answer — both backends hit CPU cull as the ceiling on ≥500k scenes;
moving it to a compute shader drops it to sub-ms regardless.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Surfaces any regression in:
* the modules dict in bim/__init__.py (added a folder, forgot the entry)
* PointerProperty wiring on bpy.types.{Scene,Object,...}
* registry-driven GizmoPreferences<Name> auto-registration in
tool.Parametric.iter_gizmo_preference_classes
* bpy.app.handlers append/remove balance
* every register()/unregister() across the 45+ feature modules
as a single PASSED/FAILED test instead of the silent "addon failed to
enable" users encounter in a fresh Blender. Paired with the existing
test_parametric_registry.py contract tests, this catches both the
registry-shape regressions (operators/PropertyGroups/predicates) and
the registration-mechanics regressions (PointerProperty types not
registered before their owners).
Generated with the assistance of an AI coding tool.
Three corrective fixes folded into one commit. All surface as
addon-load / save-time exceptions on v0.8.0's bim layer because
PR2's tool.Parametric refactor over-committed to the PR4 contract.
1. iter_gizmo_preference_classes — the previous implementation
returned only the shared GizmoPreferencesFeature class. v0.8.0's
bim/ui.py declares PointerProperty fields ('door', 'window', ...)
on GizmoPreferences that point at per-feature
GizmoPreferences<Name> classes; those must be registered BEFORE
GizmoPreferences itself. The shared-class-only return broke
addon registration with:
'door' PointerProperty could not register (see previous error)
Restore the v0.8.0 per-feature lookup (iterate EDIT_TYPES, look
up each GizmoPreferences<Capitalize(name)> on ui_module) and
keep the shared-class lookup as forward-compat. Tag FIXME(PR5).
2. EDIT_TYPES — drop the array / pipe_segment / duct_segment
entries from the registry. Their bim.finish_editing_<name>
operators land with PR4. Registering them in PR2's EDIT_TYPES
without the operators makes auto-commit-on-save dispatch a
non-existent finish_op for any object whose
BIM<Name>Properties.is_editing flag is True, raising:
RuntimeError: 'bim.finish_editing_array' must be a registered
tool.Ifc.Operator subclass for undo-safe IFC mutation
PR4 re-adds the three entries together with their operators.
Tag FIXME(PR4).
3. tool.Blender.Modifier shim block — upgrade the prose comment to
a formal FIXME(PR5) marker so the PR5 cleanup sweep finds it via
grep alongside every other tagged shim site.
Generated with the assistance of an AI coding tool.
Two pieces:
1. Per-chunk vertex storage (stage 13)
WebGPU mandates maxStorageBufferBindingSize ≥ 128 MB. Real BIM models
routinely exceed that (one of yours is 139 MB vertex). Without
chunking, every browser load would fail with
"exceeds max_storage_buffer_binding_size".
Strategy: each model's vertex data is split into ≤ 128 MB chunks at
applyCachedModel time. Each chunk gets its own vertex_storage buffer,
visible_draws / prefix_sums buffers, per_chunk_uniform, and bind group.
Index buffer, instance storage, and mesh storage stay single-per-model
(they fit well under the cap on every scene we've seen). Mesh-to-chunk
assignment is bake-time-deterministic (walks meshes in order, opens a
new chunk when adding the next would overflow).
Cull buckets visible instances by their mesh's chunk; render issues
one drawcall per non-empty chunk per model. WGSL is unchanged — the
binary-search vertex pulling works identically per chunk because
base_vertex is now CHUNK-LOCAL (the chunk's bind group binds its own
vertex_storage).
Single code path: chunking is ALWAYS on at 128 MB regardless of
target. Cost on desktop is a handful of extra drawcalls per frame
(1 per non-empty chunk; typical models = 1-3 chunks). Negligible.
A mesh whose vertex range is itself > 128 MB can't fit in any chunk
and would need splitting — typical IFC meshes are nowhere near that
(hundreds of verts), and applyCachedModel warns loudly if one ever
appears.
--web-limits CLI flag requests the WebGPU mandatory floor limits
(128 MB max storage binding, 256 MB max buffer) instead of the
adapter's actual max. Used to verify chunking actually fits through
browser constraints — turns "trust me, web will work" into a hard
test. The 139 MB scene loads cleanly with --web-limits.
2. Settle frame after motion (bug fix)
Reported regression: after orbiting, sub-pixel instances dropped by
motion-mode contribution culling stayed missing after the camera
stopped. Event-driven rendering means no frame is scheduled after
mouse-up, so the cull never re-ran at the still threshold.
Fix: track last_cull_was_motion_. If this frame used the motion
threshold, requestUpdate() after present to schedule one settle
frame. Next frame: camera_moved = false → still threshold → small
instances reappear. Matches GL's last_cull_was_motion_ behaviour.
Verified pixel-identical on basic.ifc; loads the user's dense scene
successfully under --web-limits (chunks=2 on the 139 MB model,
chunks=1 on the others).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Closes the other half of task #10. The wgpu minimal already wrote PNGs
via wgpuCommandEncoderCopyTextureToBuffer + mapAsync; the GL backend
now has the equivalent via glReadPixels on the back buffer just before
swapBuffers.
- ViewportWindow::captureNextFrameToPng(path, quit_after=true) queues
a one-shot capture. render() reads the default framebuffer at full
pixel size (width * devicePixelRatio), flips bottom-up → top-down
into a QImage::Format_RGBA8888, saves PNG, and optionally
QCoreApplication::quit. Synchronous glReadPixels is fine here —
pick is interactive and rare; not used per-frame.
- ifcviewer-minimal --screenshot PATH wires through MinimalWindow
just like --camera / --benchmark. Honoured after all loads complete
(applyPendingBenchmark also drains pending_screenshot_).
Lets a parity script do:
IfcViewerMinimal foo.ifc --camera A,B,C,D,E,F --screenshot gl.png
IfcViewerWgpuMinimal foo.ifcview --camera A,B,C,D,E,F --screenshot wgpu.png
# then pixel-diff with whatever (ImageMagick, PIL, etc.)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Closes the interactive selection loop. After this commit you can:
- LMB-click an object → highlight (selection)
- Press H → hide all selected
- Press Shift+H → show all (clear hidden set)
- Press I → isolate selected (hide everything else)
WgpuVisibilityState (new header) is a plain unordered_set<uint32_t> of
hidden object_ids — mirrors src/ifcviewer/Visibility.h's shape but
stays Qt-free for the ifcviewer-core extract later.
cullModelCpuCompute consults visibility_.isHidden(inst.object_id)
before the frustum test — hidden instances cost nothing on every axis
(no draw, no depth contribution, no pick hit). The CPU vector is
read concurrently by the parallel cull workers, which is safe because
mutations only happen between renders (handlers requestUpdate after
mutating; render reads).
Hiding deselects (matches GL behaviour: H clears the now-invisible
selection rather than leaving phantom selected-but-invisible ids).
Stage 5's last piece — clip planes — is deferred. Adding the uniform
array + WGSL discard is mechanical, but the section-tool UI that
drives them isn't ported yet (minimal viewer has no way to place a
clip plane), so it'd ship as empty plumbing. Will land alongside the
section-tool port.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Closes the loop on stage 4 (pick): clicking an object now highlights it
on screen. Plus the prerequisite plumbing for selection to behave
correctly across multi-sidecar loads.
Pieces:
1. WgpuSelectionState (new header)
CPU-side multi-set + active-id, mirroring the GL Selection.h shape
but pure stdlib (no Qt deps) so it can move into ifcviewer-core
later without dragging Qt across. clear/replace/add/remove/toggle
APIs + a fillFlagsArray helper that packs (selected, active) into
a u32 bitmap indexed by object_id.
2. selection_flags storage buffer + frame_bgl bump to 2 entries
Indexed by object_id, bit 0 = selected, bit 1 = active. Lives in
the frame bind group (group=0 binding=1) because object_ids are
globally unique — making it model-scoped would be the wrong cut.
ensureSelectionFlagsBuffer grows geometrically (64 → 128 → … u32)
as new models push next_object_id_ up, rebuilds the frame bind
group when it does.
3. Global object_id rebase in applyCachedModel
Each sidecar's local ids start from 1 and collide across files;
pick was previously ambiguous on multi-model loads. We now add
next_object_id_ as a base offset, rewrite InstanceCpu.object_id
(CPU mirror stays consistent) + InstanceGpu.object_id (what pick
reads back), and bump next_object_id_ by the model's max + 1.
4. WGSL main fragment reads sel_flags
Vertex shader passes inst.object_id through to fragment as
@interpolate(flat). Fragment reads sel_flags[object_id], mixes
(0.2, 0.6, 1.0) at 0.45 for in-selection and (0.4, 0.8, 1.0) at
0.40 on top for active. Same constants as the GL main shader.
5. Mouse → selection
LMB-click-without-drag pick result feeds the selection:
no modifier → replace
Shift → add
Ctrl → remove (active migrates to another id in the set)
miss + no modifier → clear
uploadSelectionFlagsIfDirty repacks + writes the GPU bitmap at
the top of the next render(); no upload on still frames.
Pick pipeline is unchanged — it already outputs the per-instance
object_id, and that's what the selection storage indexes.
Visibility + clip planes are pending follow-ups in stage 5 (mostly
small, share the same buffer-lifecycle pattern). Edge silhouette
(stage 9 partial), --screenshot diff harness (stage 10 partial),
ifcviewer-core extract (stage 12), and web chunking (stage 13) all
still pending.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Stage 4 of the wgpu port. LMB-click-without-drag now resolves the
object_id under the cursor by running a dedicated pick render and
copying back the single texel at the click position.
- Pick pipeline reuses the existing pipeline_layout_ (same bindings
as main: frame uniform at group=0, per-model storages at group=1).
Different vs / fs entry points (vs_pick / fs_pick) in the main
WGSL module — the vertex pulling logic is duplicated for now but
the bind group layout match means no pipeline_layout rebuild and
pickObjectAt can reuse the current frame's already-uploaded
visible_draws + per-model bind groups.
- Pick FBO: surface-sized R32UInt color attachment + Depth32Float
depth, both single-sample (no MSAA — pick needs exact texel
access). CopySrc on the color so we can copyTextureToBuffer the
1×1 click region. Recreated on surface resize.
- pickObjectAt: encodes a one-shot pick pass + a single texel copy
into a 256-byte staging buffer, submits, mapAsync, sync-spins
processEvents until ready. Synchronous wait is fine here — pick
runs on click, not per-frame, so a sub-ms stall is invisible.
- Mouse integration: existing LMB drag-orbit preserved. A 3-pixel
threshold promotes drag (set nav_dragged_); release without
dragging triggers pickObjectAt at the release coords (logical
Qt → physical pixels via devicePixelRatio). object_id is logged;
selection state to consume the id arrives with stage 5.
Object_id 0 means miss (clear value); the pick attachment is cleared
to 0 before each pass and the fragment writes the instance's
object_id, so any non-zero result is a real hit on a drawn instance.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
First piece of stage 9 — the dark outlines BonsaiViewer / the GL backend
draw at depth discontinuities. Ported the GL renderEdgePass algorithm
verbatim, including the three things my earlier attempt missed:
1. Linearise depth to view-space metres before the Laplacian. Raw
[0,1] clip-z is heavily non-linear so a fixed-threshold edge
detector only caught near-camera silhouettes. Now reverses the
wgpu z-remap (z * 2 - 1 back to GL NDC) then standard reverse-
perspective to view-z.
2. Threshold scales with depth: t = EDGE_THRESHOLD * c. A 4 mm gap
between two surfaces reads the same whether it's 0.5 m or 50 m
away from the camera.
3. Multiplicative blend (Dst, Zero) with fragment output of
vec3(1 - edge). Strictly darkens, never brightens. Matches GL's
(GL_DST_COLOR, GL_ZERO) blend.
Constants EDGE_SCALE=6.0 / EDGE_THRESHOLD=0.004 are GL's tuned values.
Camera near/far hard-coded to 0.1 / 10000 (the viewport defaults);
they'll move to a small uniform when AppSettings ports across.
Pipeline state: depth-attachment-less, sample count 1, blend on, no
cull. Reuses depth_texture_'s TextureBinding usage that HiZ added.
Render pass loads the resolved main-pass colour (LoadOp_Load) and
writes back through the multiplicative blend; encoded between the main
pass and the HiZ resolve so HiZ uses the same MSAA depth that produced
the edges. edge_bind_group_ rebuilds lazily when depth_view_ is
replaced (mirrors the HiZ bind group lifecycle).
Perf cost on the 10-sidecar / 380k-instance benchmark: 0.1 ms (11.5 →
11.6 ms). Fullscreen depth-laplacian is essentially free on this GPU.
Remaining stage 9 work: HUD/labels/lines/points overlay primitives,
which need the QPainter-→-texture path. Lower visual priority than
edges; handled in a follow-up.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Closes the perf gap to the GL backend on real BIM benchmarks. On a 10-
sidecar / 380k-instance corpus at a fixed --camera the wgpu binary went
from 110.6 ms to 11.6 ms (vs GL's 23 ms — half the frame time, but
note GL is doing extra work the wgpu backend hasn't ported yet; see
the caveats list at the bottom). Bundled because the pieces interlock
and shipping any of them without the others reintroduces the same wall.
1. Cross-mesh vertex pulling (single mega-draw per model)
The previous one-drawIndexed-per-(mesh × LOD-bucket) loop was costing
~13ms on a 27k-mesh scene. CPU now emits a flat visible_draws[]
(16 B per visible (mesh,lod,instance)) plus a prefix_sums[] table.
WGSL binary-searches prefix_sums by @builtin(vertex_index) to find
the entry, then manually fetches the mesh-local index from a
storage-bound indices[] and pulls the packed 12 B vertex. No
setIndexBuffer; the shader reads everything from storage. Bind
group grew from 4 to 7 entries (vertices, meshes, instances,
indices, visible_draws, prefix_sums, per-model uniform) — well
under WebGPU's mandatory 8 storage / 12 uniform floor.
2. Async HiZ readback via ping-pong staging buffers
Sync wait via wgpuInstanceProcessEvents was costing ~37 ms on a
real scene (GPU drain). Two staging slots now ping-pong: frame N
kicks a non-blocking mapAsync on slot K, frame N+1's first action
is one processEvents drain. Pyramid is 1-2 frames stale — matches
the "slightly-stale depth, fine" pattern the GL backend already
documents. encodeHizResolve returns -1 (skip) if both slots are
in flight; cull keeps using the most recent pyramid.
3. Cull reorder: contribution before HiZ
HiZ projection is ~10× more expensive than the contribution
check, yet most contribution-survivors would be HiZ-rejected
anyway on dense scenes. Computing projected_px first lets
contribution short-circuit ~80% of HiZ tests with no rejection-
quality loss. Saved ~34 ms on the dense bench.
4. Motion-mode contribution threshold
AppSettings::motionMinPixelRadius parity. While the camera is
changing (orbit/pan/zoom/--benchmark sweep), drop instances
below 10 px instead of 2 px. Halves visible_objects during
motion with no perceived quality loss.
5. Parallel cull (std::async across models)
Per-model cullModelCpu split into Compute (CPU-only, thread-safe)
+ Upload (main-thread wgpu queue writes). std::async fan-outs the
compute across models; main-thread joins and uploads. Wall-clock
cull on the 10-model corpus drops from ~17 ms single-threaded to
~9 ms across cores.
6. --no-hiz CLI flag + per-phase benchmark timings
Benchmark now also prints "per-frame avg ms: cull=X
hiz_readback=Y" so future regressions can be attributed without
guesswork. --no-hiz toggles the master switch from the CLI.
Honest caveats — wgpu is currently faster mostly because GL is doing
work we haven't ported yet:
- Edge silhouette pass (stage 9) will add ~3-5 ms back to wgpu.
- GL's HiZ uses the BVH so it rejects whole subtrees (1.7k vs
our 358 rejects on the same scene). BVH for HiZ is future work
(task #13 / a new task) — until then we draw more sub-pixel
geometry that's behind closer surfaces. Visually correct, perf
cost paid. Stage 4+5 are unaffected.
Verified pixel-identical on basic.ifc through every change. Real-scene
visual diff against GL pending the --screenshot flag on the GL minimal
(task #10's other half).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Fixes addon-load ImportError that surfaces when tool/geometry.py
and tool/model.py (extracted in C8 / C9) reference symbols that
don't exist on v0.8.0:
* bim/ifc.py: get_cache_or_detect_lock — IfcStore.get_cache
variant that tracks the multi-instance-cache-locked-by-other-
process flag, sets it on PermissionError, clears it (along with
the dismiss flag) on subsequent success. Used by
tool.Geometry.* to gate IFC cache reads without crashing when
another Blender instance holds the cache lock.
* tool/cad.py: WELD_TOLERANCE constant + paired CAD helpers
(auto-detect-curves vertex precision, polyline normal helpers,
etc.) used by tool.Model.* + by the parametric model operators
that land in PR4.
Both modules had zero upstream commits since the gizmos-8088 fork
point — safe bulk extraction. PR4 has no caller-line work for
either file (the additions are pure additions, no existing API
removed); the v0.8.0 callers of get_cache_or_detect_lock and
WELD_TOLERANCE are the PR2-scope files that needed them.
Generated with the assistance of an AI coding tool.
The previous commit moved is_<type> predicates off tool.Blender.Modifier
onto tool.Parametric, and earlier C4 moved the Array helper bag off
tool.Blender.Modifier.Array onto tool.Array. PR4 will migrate every
caller; this commit keeps the OLD entry points alive as thin delegates
so PR2 ships without breaking ~30 caller sites that still spell the
old API in v0.8.0:
* tool.Blender.Modifier.is_door / is_railing / is_roof / is_stair /
is_wall / is_window — delegate to tool.Parametric.is_<type>.
* tool.Blender.Modifier.Array.bake_children_transform / constrain_
children_to_parent / get_all_children_objects / get_all_objects /
get_children_objects / get_modifiers_data / remove_constraints /
set_children_lock_state — delegate to tool.Array.<same name>.
These shims are removed in PR5's cleanup commit once PR4 has rewritten
the call sites in bim/import_ifc.py, bim/module/geometry/operator.py,
bim/module/geometry/data.py, bim/module/model/array.py + the per-feature
operators (door, wall, window, railing, roof, stair, ui).
Generated with the assistance of an AI coding tool.
tool.Parametric becomes the central registry for Bonsai's parametric
features (wall, slab, door, window, railing, roof, stair, plus
mep-segment variants). Each feature registers a ParametricObject spec
declaring its enable/finish/cancel op names, props accessor, regen
callback, and is_element_type predicate.
Public surface:
* tool.Parametric.WALL / SLAB / DOOR / WINDOW / RAILING / ROOF /
STAIR / PIPE_SEGMENT / DUCT_SEGMENT — typed accessors per feature.
* tool.Parametric.is_wall / is_door / is_window / is_railing /
is_roof / is_stair — element-type predicates that move off
tool.Blender.Modifier into the parametric registry. The next
commit adds backward-compat shims on tool.Blender.Modifier so
v0.8.0 callers keep working.
* tool.Parametric.is_object_editing(obj) — returns the registered
feature an object is currently editing, or None.
* tool.Parametric.run_bim_op(op_name) — invoke a parametric op by
bl_idname.
* tool.Parametric.heal_stale_edit_flags — clear is_editing flags
on file load so a saved-mid-edit project doesn't leave gizmos
poll-locked.
* supports_build_edit_lifecycle field on ParametricObject — declares
whether the feature implements the build/edit/cancel triad.
The previous bare `print(f"Bonsai: commit of {obj.name!r} via
{finish_op} failed: {e}")` exception-handler is replaced with
logger.warning(..., exc_info=True). Same channel (Bonsai configures
logging to the Blender console at WARNING level), strictly more
information (full traceback), correct idiom for an error-path
message. A second logger.warning is added for parametric predicate
failures, also exception-handler scope.
Generated with the assistance of an AI coding tool.
Stage 7 of the wgpu port. Per-frame after the main render pass:
1. encodeHizResolve runs a depth-only render pass that samples the
MSAA depth texture (sample 0) and max-reduces it into a small
single-sample Depth32Float target (256 × ~h-aspect). Implemented
as a fullscreen-triangle WGSL pipeline; one nested loop per
output texel over its source rect. WebGPU has no built-in depth
resolve, so this combined resolve+downsample fragment shader is
the way.
2. copyTextureToBuffer writes the small resolved depth into a
CPU-mappable staging buffer (≈ 160 KB at 256×160).
3. readbackAndBuildHizPyramid maps the staging buffer (sync via
wgpuInstanceProcessEvents — small enough that the stall is
well under a millisecond), strips per-row padding, and CPU
max-reduces a full mip pyramid (level 0 → 1×1). Stores the VP
used so the next frame can project AABBs into the same space.
Next frame, cullModelCpu calls aabbOccludedByHiz after the frustum
test: projects all 8 AABB corners through hiz_vp_, computes the
screen-space AABB and the nearest projected z, picks the mip level
where the AABB covers ≤ 2 texels per axis, samples that level's 2×2
window, and culls iff min_z > max_pyramid_depth in [0,1] z.
Plumbing changes:
- depth_texture_ gains TextureBinding usage so the resolve shader
can read it.
- hiz_enabled_ master switch defaults true; mirrors IFC_NO_HIZ in
the GL backend. Disabling skips encode + readback entirely.
- Bench output's "hiz_rej N" field now reflects actual rejections.
Verified: basic.ifc (3 instances, no occluders) renders pixel-
identical to pre-HiZ — proves the test rejects nothing it shouldn't.
Real rejection counts need a dense scene; this should drop visible-
objects count noticeably on real BIM benchmarks where back-of-room
walls hide each other.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Quick win before the proper HiZ stage. Adds a min_pixel_radius
threshold (defaults 2.0 to match AppSettings::minPixelRadius() in GL):
instances whose projected bounding-sphere radius falls below it are
dropped from the per-mesh buckets entirely.
The projected_px math (radius_world * focal_px / view_z) is now
computed once per instance and shared with the LOD pick that uses the
same number. Saves one square root per instance per frame on dense
scenes vs the previous code path that only computed it inside the LOD
branch.
Expected impact on real BIM benchmarks: visible-objects count drops by
roughly 10×, matching the GL backend's number. Without this fix, wgpu
was drawing every frustum-surviving sub-pixel instance — most of the
work and most of the geometry the GL backend wasn't even submitting.
Motion-mode threshold bump (10.0 in GL during camera drag) lands
later when mouse-driven motion tracking is wired up.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Two pieces that block proper side-by-side parity with the GL minimal:
1. --camera tx,ty,tz,dist,yaw,pitch. Same format string as the GL
minimal so a pasted camera arg lands the same view on both backends.
setCamera() also flips initial_view_applied_ = true so the auto-
viewAll-on-first-load doesn't snap away from the script-set position
when the model finishes uploading.
2. Real BIM models exceed the conservative WebGPU defaults at device
create time. A 114k-instance / 19M-index sidecar's vertex storage is
139 MB, which trips wgpu's default 128 MB max_storage_buffer_binding_
size and bind-group creation fails. Now wgpuAdapterGetLimits is
called first and the device is requested at the adapter's full
ceiling — every desktop driver supports multi-GB.
Trade-off worth flagging: web parity will fail here because browsers
cap at the defaults. The eventual fix is to split a model's vertex/
instance storage into ≤128 MB chunks with a small per-frame routing
table, which is a real chunk of work. For now this unblocks all the
native benchmarking the user is actually doing.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Reported regression: --benchmark on the same sidecar visibly rotated
around a different point in the wgpu binary than in IfcViewerMinimal.
Root cause was two camera-convention drifts:
1. orbitEye placed the camera at (sin yaw, -cos yaw) from target;
the GL backend uses (cos yaw, sin yaw). Same target, but the
camera faces a different side of the model at yaw=0, which made
the orbit feel like it pivoted around a different point even
though the actual world-space target was the same. Now exactly
matches GL ViewportWindow::updateCamera:
eye.x = target.x + dist * cos(pitch) * cos(yaw)
eye.y = target.y + dist * cos(pitch) * sin(yaw)
eye.z = target.z + dist * sin(pitch)
2. viewAll's distance was an ad-hoc 0.6 * diag / tan(half_fov);
GL uses frameAabb(mn, mx, 1.10): tan_half = tan(fov/2),
min_aspect = min(aspect, 1), distance = (radius / (tan_half *
min_aspect)) * 1.10. Aspect-aware so portrait windows pull back
enough that the bounding sphere still fits on the tighter axis.
Now ported verbatim.
Also logs the computed target + distance on viewAll so a follow-up
side-by-side run prints both backends' framings and any remaining
discrepancy is easy to spot.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Stage 8 of the wgpu port. cullModelCpu now buckets each visible instance
by (mesh_id, lod) instead of (mesh_id), and emits one MeshDraw record
per non-empty bucket. LOD pick projects the instance's world-space
bounding sphere to pixels via
projected_px = world_radius * focal_px / view_z
where focal_px = viewport_h / (2 * tan(fov_y/2)) and view_z is the
forward·(center-eye) depth. When projected_px < lod1_pixel_threshold_
AND the mesh has a baked LOD1 slice (MeshInfo.lod1_index_count > 0),
the instance draws the LOD1 index range instead of LOD0; baseVertex
and the vertex storage are shared between LODs.
mesh_draws can now grow to up to 2 × meshes.size() per frame (LOD0 + LOD1
slice per mesh). The visible_buffer layout per mesh becomes
[LOD0 instances | LOD1 instances] contiguous, with each MeshDraw
referencing its own firstInstance offset.
lod1_pixel_threshold_ defaults to 30 (mirrors AppSettings::
lod1PixelThreshold() in the GL backend); set to 0 to disable LOD1
entirely (always LOD0). AppSettings port lands in a later commit.
Verified: basic.ifc (3 tiny instances, no LOD1 baked by meshoptimizer
since each mesh is well under the 500-tri threshold) renders pixel-
identical to pre-stage-8 — proves the all-LOD0 path is preserved.
Real LOD switching needs a sidecar where buildLods produced LOD1 slices.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Closes the visible gap to BonsaiViewer down to just the post-process
edge silhouette pass (still pending in task #9). Four changes bundled
because together they bring up the parity story:
- WGSL fragment now applies cavity = clamp(length(fwidth(n))*1.5,
0, 0.35) and multiplies by (1 - cavity). Matches GL shader.
- Lighting constants switched to GL's exact values: key (0.3, 0.5,
0.8), fill (-0.3, -0.5, 0.8), sky tint (0.55, 0.60, 0.70), ground
tint (0.35, 0.32, 0.28). My initial guesses were close but not
identical; matching them means side-by-side diffs only flag actual
pipeline differences, not lighting tweaks.
- 4× MSAA: render pass writes into a MULTISAMPLE color attachment
(surface_format_-matched), resolves into the surface texture for
present. Depth is also 4 samples. Pipeline.multisample.count = 4.
ensureMsaaColorTexture / releaseMsaaColorTexture mirror the depth-
texture lifecycle. Matches GL minimal's QSurfaceFormat::setSamples(4).
- sRGB output fix. wgpu-native's Vulkan swap chain on X11 treats
BGRA8Unorm as sRGB-output (applies linear→sRGB encoding on shader
writes), even though caps.formats[0] reports plain Unorm. The GL
backend writes to a non-sRGB framebuffer with no such conversion,
so a clearValue of (0.125, 0.137, 0.161) lands as bytes (32, 35,
41) on GL but (99, 104, 112) on wgpu — ~3× brighter. Pre-decoding
via srgbToLinear on (a) the clearValue in C++ and (b) the final
fragment colour in WGSL makes wgpu's implicit encode round-trip,
so the final bytes match GL. Verified via screenshot pixel sample:
#202329 background reads as exactly (32, 35, 41).
Remaining visible gap to BonsaiViewer is the dark-line edge silhouettes
(renderEdgePass in GL, depth laplacian → outline). That belongs with
the overlay / post-process work in task #9.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Drag-down was decreasing pitch (camera diving), opposite to the GL
viewport's convention where drag-down increases pitch so the top of
the object rotates toward the viewer. Yaw direction was already
correct. Matches the existing user muscle memory from IfcViewerMinimal.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The single "(file missing, wrong magic, or schema mismatch)" message
was making triage harder than necessary. loadSidecar now expands a
leading ~/ (shells skip it inside double quotes, which trips up paste-
from-launcher), and on failure peeks the file's header itself to
report exactly which check failed:
- "Sidecar not found" — file doesn't exist
- "Sidecar unreadable" — exists but open failed
- "Sidecar truncated" — <12 bytes
- "Sidecar magic mismatch" — wrong magic, reports got vs expected
- "Sidecar schema mismatch" — wrong version, reports both numbers
and suggests re-baking
- "Sidecar endianness mismatch" — cross-platform load attempt
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Stage 11 of the wgpu port. WgpuViewportWindow gains setBenchmarkFrames(N);
the minimal driver wires it to a --benchmark N flag. Renders N frames
after a 5-frame warmup, yaw-sweeping the camera at 0.5°/frame, captures
per-frame wall time with QElapsedTimer (cull + encode + present), and
prints avg/median/p1/p99 + last-frame stats in the same line format as
IfcViewerMinimal so a script can diff them line for line.
Per-frame stats (visible_objects, visible_triangles, sub_draws) are now
summed in render() from m.mesh_draws. hiz_rej reports 0 until stage 7
adds HiZ occlusion.
Verified on basic.ifc (3 instances): wgpu 11.68 ms avg vs GL 11.75 ms
avg — same scene, same camera sweep, same window size. Noise-level
delta as expected on a tiny scene; the interesting comparison is on
real BIM corpora once you bake them to v13 sidecars.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
LMB drag → orbit (yaw/pitch, pitch clamped to ±89.9° to avoid gimbal
flip at the poles). MMB drag → pan in the camera's screen-space plane,
world-units-per-pixel sized against the view frustum at the pivot depth
so panning feels constant regardless of zoom. Wheel → zoom (12% per
notch, sign matches "wheel up = closer"). LMB is bound to orbit because
selection isn't wired yet; will rebind to selection + nav preset once
AppSettings ports over.
Pure addition to WgpuViewportWindow — overrides four QWindow event
handlers, no changes to render or cull paths. Lets you actually fly
around a loaded sidecar without a screenshot loop.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Stage 6 of the wgpu port. Replaces the one-draw-per-(mesh, instance) loop
with a CPU cull pass that survives one drawIndexed per non-empty mesh
with packed instanceCount.
Adds to WgpuModelGpuData:
- visible_buffer: u32[] storage SSBO, pre-sized to instance_count at
applyCachedModel so the bind group reference never invalidates.
Re-uploaded each frame via wgpuQueueWriteBuffer.
- mesh_draws: per-mesh schedule (first_instance, instance_count,
first_index, base_vertex, index_count). instance_count==0 means the
mesh contributed nothing this frame and the draw is elided entirely.
cullModelCpu per-frame:
- Extract 6 frustum planes from the same VP we write into the uniform.
WebGPU clip-space z is [0, 1], so near plane = matrix row 2 (not
row 3 + row 2 as in GL); rest of the derivation is standard.
- Per-instance AABB-vs-frustum test using the p-vertex shortcut
(cheapest correct early-out for AABBs).
- Bucket survivors by mesh_id; flatten into a contiguous u32 list;
upload via wgpuQueueWriteBuffer. Per-mesh slice is [first_instance,
first_instance + instance_count).
WGSL adds @group(1) @binding(3) var<storage, read> visible: array<u32>
and an extra indirection: instance_idx = visible[iid]; the rest of the
shader is unchanged. firstInstance on each drawIndexed offsets into
visible[], so each mesh reads its own slice.
Verified two ways:
1. basic.ifc (3 instances, all on-screen) renders pixel-identically
to pre-stage-6 — proves cull keeps everything it should.
2. basic.ifc + a synthetic instance placed at (100, 100, 100) is
culled cleanly: only the cube renders, the far quad is rejected
by the frustum test. Proves cull actually rejects out-of-frustum
geometry rather than passing everything through.
Contribution culling, HiZ, and LOD selection arrive in stages 7 and 8;
they all hook into the same cullModelCpu seam.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Pulls the capture half of task #10 forward so we stop flying blind from
stage 3 onward. WgpuViewportWindow gains captureNextFrameToPng(path);
the minimal driver wires it to a --screenshot PATH flag that renders
one frame, copies the surface texture back to host memory, writes a
PNG via QImage, and quits.
CopySrc is added to the surface configuration usage so the surface
texture can be the copy source. The texel-to-buffer copy honours
WebGPU's 256-byte bytes-per-row alignment by padding rows and stripping
the padding when assembling the QImage. Surface format 28 (BGRA8Unorm)
is byte-swapped to RGBA on the way into QImage::Format_RGBA8888;
RGBA8 surface formats are memcpy'd straight through.
Verified end-to-end on /tmp/basic.ifcview: 3 cube meshes/instances
render with depth, back-face cull, and the hemisphere-ambient + key+fill
lighting model — top face reads sky (bright), front faces read mid-tone,
exactly as the WGSL shading intended. The pixel-diff half of task #10
(comparing against a GL baseline) lands later when the GL minimal binary
gets an equivalent flag.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Stage 3 of the wgpu port. Replaces the clear-only render loop with the
full main shading pass:
- WGSL port of the GL main shader. Vertex-pulling: the vertex storage
buffer is read as array<u32> in the shader, with pos/normal/color
decoded manually per vertex. baseVertex (set per draw to mesh's
vertex offset) folds into @builtin(vertex_index) automatically;
firstInstance carries the instance slot for @builtin(instance_index).
No vertex-input layout — vertex pulling means no IA bindings.
- Render pipeline bound to depth-32-float (write-on, less compare),
back-face cull, CCW front face. Pre-multiplies a [-1,1]→[0,1] z-remap
matrix onto Qt's projection so WebGPU's clip-z convention is met.
- Two bind groups: group=0 per-frame (uniform with view-proj + key/fill
light + hemisphere ambient), group=1 per-model (three read-only
storage buffers: vertices, mesh quant, instances).
- Depth texture is created lazily and recreated on surface resize.
- Orbit camera state on WgpuViewportWindow with viewAll() that frames
the union of all loaded models' world AABBs after the first load.
Mouse navigation lands later.
- Draw loop: one drawIndexed per (mesh, instance) pair per model. This
is correct but CPU-heavy on dense scenes; stage 6 introduces the cull
+ compacted visible list that lets multiple instances of one mesh
collapse to a single call, and the eventual GPU-driven cull (post
sunset of the GL backend) goes further.
Verified on /tmp/quad_v13.ifcview (1 mesh, 1 instance) and on a real v13
sidecar baked from basic.ifc via the GL minimal viewer (3 meshes,
3 instances, 864 B verts). No wgpu validation errors fire across pipeline
creation, depth attachment, bind groups, or the draw loop on either.
Visual confirmation deferred until --screenshot lands (task #10) which
is being pulled forward next so we don't keep flying blind.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Stage 2 of the wgpu port. WgpuViewportWindow gains a queueLoadSidecar
API (called from the minimal driver before init) and an applyCachedModel
that runs after init: reads via SidecarCache::readSidecar, allocates
four wgpu buffers per model (vertex storage, index, mesh-quant storage,
instance storage), uploads via wgpuQueueWriteBuffer, retains a CPU
mirror of the MeshInfo/InstanceCpu arrays for the cull and picking
paths that arrive in later stages.
MeshGpu (the per-mesh quantization basis) is derived from MeshInfo on
the fly; InstanceGpu (transform + ids) is derived from InstanceCpu and
uses the cached float transform — composing from placement_transformation
against federation-stage matrices lands when stage 5 wires those.
SidecarCache.cpp is compiled into IfcViewerWgpu directly: it's pure
C++ with no Qt/OCCT/IFC-parse deps, so dragging in the IfcViewer
static lib for one source file would be wasteful. This duplication
goes away once src/ifcviewer-core/ is extracted (task #12).
Verified on a synthesised v13 sidecar (4 verts, 6 indices, 1 mesh,
1 instance) and a multi-sidecar load that assigns successive model_ids.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds src/ifcviewer-wgpu/ and src/ifcviewer-wgpu-minimal/ behind a new
BUILD_BONSAIVIEWER_WGPU option (default OFF), gated independently of
BUILD_BONSAIVIEWER. Stage 1 brings up a Qt window with a wgpu-native
v29 surface (X11) and clears to the background colour — no rendering
beyond that yet. Mirrors the lifecycle of the GL ViewportWindow so
subsequent stages (vertex-pulling renderer, pick, cull, HiZ, overlay)
slot in without restructuring the host.
wgpu-native is fetched as a pre-built binary release via FetchContent;
its .so SONAME is patched in at configure time so dependents get a
clean DT_NEEDED. The X11 native handle is obtained via the public
QNativeInterface::QX11Application API; Wayland and macOS/Windows
surface creation are stubbed with explicit "not wired yet" warnings.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
tool.Model gains:
* get_pipe_segment_props / get_duct_segment_props — typed prop accessors
for the MEP-segment edit lifecycle.
* resolve_active_props_for_edit — picks the right BIM*Properties to
drive a parametric edit triad based on the active object's IFC class.
* mirror_parent_void_fillings_to_children — when an array parent has
hosted fillings (door/window in a wall), replicate the same fill
rels onto each array child. Uses tool.Array.get_parametric_propagation_
targets so the propagation stays within the array family (the old
get_all_element_occurrences over-propagated to standalone occurrences
of the same type, which silently mutated unrelated arrays).
* unshare_opening_representation — fork a shared IfcShapeRepresentation
so editing one opening doesn't mutate its array sibling.
* duplicate_ifc_objects gains a post-condition select-restore on the
array parent so callers don't get a deselected parent for N>=2 arrays.
sync_object_ifc_position is kept as a thin delegate to
tool.Geometry.commit_placement_if_moved (the new home, added in C8) so
the 6 v0.8.0 callers in mep / product / system don't AttributeError;
PR4 migrates each caller and removes the delegate.
tool.Pset gains:
* upsert_pset — get-or-add-or-edit in one call.
* write_bbim_data — JSON-encode + write BBIM_* metadata in one call.
tool.Slab is new — slab-specific reads (active extrusion, axis
direction) used by the slab gizmos, pure-IFC, no PropertyGroup mutation.
Generated with the assistance of an AI coding tool.
Adds:
* get_body_representation(element) — DRY of the repeated
ifcopenshell.util.representation.get_representation(element, "Model",
"Body", "MODEL_VIEW") call across slab / wall / opening / stair /
roof / door / window / mep. One central place to read the body rep;
every caller stops re-spelling the four magic strings.
* has_axis_representation(element) — predicate for elements with a
GRAPH_VIEW Axis representation. Used by the wall/MEP path decorators
to skip elements without an unambiguous 1D path.
* has_material_styles(element) — predicate for whether the element
carries IfcStyledItem material assignments.
* restore_placement_from_ifc(obj, element) — snap obj.matrix_world back
to element's committed IFC placement + rebaseline the drift checksum.
* restore_or_rebaseline_placement(obj, element) — Cancel-flow helper:
restores if ObjectPlacement exists, just rebaselines the checksum if
not.
* detach_representation(product) — remove the active representation
from a product without deleting the entity (used by parametric
rebuilds that wipe + re-add).
commit_placement_if_moved docstring expanded with a "drop-in scope"
note so callers don't redundantly wrap it in an is_moved check that
the helper already does.
Switches the duplicate-aware helper calls (formerly tool.Root.*) to
tool.Duplicate.* now that the service exists (C6).
Generated with the assistance of an AI coding tool.
Adds:
* ViewportDecorator base class — install/uninstall/draw lifecycle for
3D viewport gpu overlays, with handler-rollback-on-failure so a
partial install can't leave dangling draw handlers.
* sync_all classmethod — drive each listed ViewportDecorator subclass
to its desired install state in one call.
* is_view_top_down + top_down_factor — viewport-camera orientation
predicates used by gizmo billboarding and decorator layout.
* get_screen_up_world — screen-up vector in world space for gizmo
text orientation.
* are_viewport_gizmos_enabled — central gate for the global
draw_gizmos_in_3d_viewport pref, replacing duplicated prefs reads.
* DecoratorColors NamedTuple + get_decorator_colors — single source
for the colour palette every viewport decorator binds.
Preserves Ryan Schultz's add_layout_hotkey_operator polish (719309571,
2026-05-25): the row-position move + separator(factor=1) between the
modifier and key icons stay intact in this extraction.
Generated with the assistance of an AI coding tool.
Extract the duplicate-aware relationship-walk + restoration logic
(get_decomposition_relationships, get_connection_relationships,
get_port_connection_relationships, recreate_decompositions,
recreate_connections, recreate_port_connections, consume_warnings)
out of tool.Root into its own service.
tool.Root's responsibility is identity and addressing of IFC roots;
the duplicate-aware bookkeeping of "before duplication, what relations
did this graph have, and how do I restore them on the new copies?"
deserves its own home. The split was already declared on core/tool.py
(C2); this commit lands the concrete tool.Duplicate implementation.
tool.Root keeps its own copies of the methods on v0.8.0's tool/root.py
during this PR so callers in bim/module/spatial/operator.py keep
working at runtime; the Root cleanup lands in PR4 alongside the
caller updates.
Generated with the assistance of an AI coding tool.
Adds:
* direction_from_port_pair(port_a, port_b) — derive the connect_port
direction kwarg from each port's FlowDirection (NOTDEFINED for
non-canonical pairs). Centralises a pattern that callers were
inlining inconsistently.
* tool.System.walk_connected_mep_elements — BFS over connected MEP
flow elements via IfcRelConnectsPorts.
* tool.System.get_port_world_position — port placement → world-space
Vector, used by the MEP path decorator.
* tool.System._build_decoration_data — cached decoration metadata
for the MEP system-path overlay.
Plus a get_port_relating_element return-type tightening (Union with
None) and a partial-init cycle workaround on bim.module.system.data
imports (now function-local — top-level import triggered the cycle
through tool.Ifc.Operator).
Generated with the assistance of an AI coding tool.
Top-level array-domain service extracted out of tool.Blender.Modifier.Array.
Owns the BBIM_Array pset graph navigation (constrain_children_to_parent,
remove_constraints, get_modifiers_data, get_children_objects,
get_all_children_objects, get_child_layer_index, bake_children_transform),
plus the Blender-side CHILD_OF constraint lifecycle that ties each child
replica to its parent's transform.
Array's own module gives the parent/child semantics a clean home — array
behaviour was previously scattered between tool.Blender.Modifier and ad-hoc
helpers in bim/module/model/array.py. The relocation eliminates the inline
duplication and gives Bonsai callers a single import surface.
Generated with the assistance of an AI coding tool.
Bpy-permitted wall reads — get_axis_local_extent, get_length_and_height,
get_x_angle, get_path_connection_location, walk_connected_walls — used
by gizmo lambdas that need wall dimensions and join topology without
the side effect of loading the wall's draft BIMWallProperties (the
loader mutates PropertyGroup state and would clobber the wall's own
gizmo state when both the wall and a hosted filling are selected).
All reads go through ifcopenshell.util.representation / .util.element
so the IFC graph stays the source of truth. tool.Wall consumes
core.model's PARALLEL_DOT_THRESHOLD + collinearity helpers (no inline
magic numbers).
Generated with the assistance of an AI coding tool.
Declares the bpy-free contract for tool services landing in subsequent
commits — tool.Wall, tool.Array, tool.System, tool.Duplicate (extracted
from tool.Root), tool.Parametric, plus minor additions on existing
interfaces (tool.Spatial.get_host_element / get_host_wall,
tool.Geometry.has_axis_representation / has_material_styles,
tool.Surveyor.get_z_rotation / set_z_rotation).
The @interface declarations are empty-bodied; concrete implementations
land in the per-service tool/* commits below. Keeping the contract in
core lets core/* helpers and tests reference the surface without
importing the concrete tool modules.
Moves get_decomposition_relationships + recreate_decompositions off
tool.Root onto the new tool.Duplicate (extraction of duplicate-aware
behaviour into its own service).
Generated with the assistance of an AI coding tool.
core/model.py gains:
* Three calibrated dot-product / distance thresholds — PARALLEL_DOT_THRESHOLD
(~2° from parallel, cos(2°) ≈ 0.9994), COLLINEAR_LINE_TOLERANCE (50mm
perpendicular distance for two parallel wall axes to share a line),
BASELINE_OFFSET_TOLERANCE — replacing inline magic numbers that the
wall-join classifier, fillet-state machine, and gizmo preview decorator
all read from.
* Pure wall-join geometry helpers (project_axis_intersection,
are_axes_collinear, classify_wall_join_state, wall_join_preview_lines,
resolve_extend_walls_target, extrusion_depth_from_vertical_height,
length_and_height_from_extrusion). They take primitive tuples + floats,
no bpy, no ifcopenshell — testable in the core lane.
core/product.py is new — pure-Python aggregate-walk helpers (resolve_host_
of_product, collect_decomposed_products) that downstream tool/spatial and
tool/aggregate consumers can call without importing ifcopenshell at module
load.
Generated with the assistance of an AI coding tool.
add_railing_representation now factors into two parts:
* compute_wall_mounted_handrail_geometry returns a pure-geometry
WallMountedHandrailGeometry dataclass (handrail polyline + support
list + terminal caps), no IFC mutation.
* add_railing_representation wraps that dataclass into an
IfcShapeRepresentation as before.
Downstream consumers that want the same math without round-tripping
through an IFC file (Blender gizmo previews, viewport drafts) now
drive compute_X directly. Future add_X_representation work in the
geometry API is encouraged to follow the same shape — a sibling
compute_X function + thin IFC wrapper.
The railing_type parameter is dropped from the signature — only
WALL_MOUNTED_HANDRAIL was ever supported, so the kwarg was dead.
The Bonsai railing-modifier caller is updated in the same commit
to stop passing it; without that update Bonsai's
finish_editing_railing_path raises TypeError on the first edit.
RailingSupport and WallMountedHandrailGeometry use @dataclass(slots=True)
— they're constructed N-per-cap during arc sampling, so the per-instance
overhead matters.
Public symbols (RailingSupport, TERMINAL_TYPE,
WallMountedHandrailGeometry, compute_wall_mounted_handrail_geometry,
add_railing_representation) re-exported from ifcopenshell.api.geometry.
New test/api/geometry/test_add_railing_representation.py covers the
compute/wrap contract.
Generated with the assistance of an AI coding tool.
Drops the module-local ``mm()`` helper in favour of the centralised
``ifcopenshell.util.unit.mm_to_m`` (added earlier in this PR). The
``as mm`` import alias preserves the existing call sites' readability.
Generated with the assistance of an AI coding tool.
Drops the module-local ``mm()`` helper in favour of the centralised
``ifcopenshell.util.unit.mm_to_m`` (added earlier in this PR). The
``as mm`` import alias preserves the existing call sites' readability.
Generated with the assistance of an AI coding tool.
ShapeBuilder gains module-level NP_X / NP_Y / NP_Z / NP_XY / NP_XZ /
NP_YZ / NP_YX axis-index constants. Downstream geometry builders had
been redefining local copies for indexing np.ndarray vectors of shape
(3,) or (N, 3); centralising removes the duplication.
mep_transition_length and mep_transition_calculate verbose default
flipped from True to False. The prints are diagnostic-only output;
True-by-default spammed the console on every transition computation,
which fires per-fitting on IFC load.
Generated with the assistance of an AI coding tool.
Centralises the millimetre-to-metre conversion shortcut that
add_door_representation and add_window_representation each defined
locally. Subsequent commits in this PR switch both call sites to
import this from util.unit, removing the duplicate definitions.
Generated with the assistance of an AI coding tool.
Use Qt's system fixed font for viewer overlay text instead of the generic monospace family, avoiding the Windows font-resolution delay seen during measurement overlays.
Generated with the assistance of an AI coding tool.
Use add_layout_hotkey_operator for draw_regen_operations so the Regen
button shows text and shortcut icons in the sidebar like all other
panel buttons. Add a separator between modifier and key icons for
readability.
Build the PyInstaller Autodesk connector without UPX compression. UPX-packed launchers are more likely to trigger enterprise Windows security scanning, and the connector is distributed as a fresh unsigned artifact for each build.
Generated with the assistance of an AI coding tool.
Update the overlay renderer's dynamic VBO uploads to bind the buffer and use glBufferData/glBufferSubData instead of direct-state glNamedBufferData/glNamedBufferSubData.
This avoids Windows/NVIDIA driver corruption seen with overlay axes, pick markers, HUD rects, and marquee rectangles while keeping the same overlay geometry and draw paths.
Generated with the assistance of an AI coding tool.
Build Rocky artifacts with shared IfcOpenShell libraries and keep geometry writer plugins out of executable packages while preserving them for Python packages.
Generated with the assistance of an AI coding tool.
The progress dialog was the only connector window not driven by a Tk
event loop: the handler created it, then blocked inline in httpx I/O.
On Windows CTkToplevel withdraws itself at construction and re-shows via
a delayed after() callback, which never fires without a running loop, so
the progress window stayed invisible for the whole transfer.
Add run_with_progress(): the blocking work runs on a daemon thread while
the main thread pumps the Tk loop and shows the dialog. Progress reports
are coalesced and marshalled back to the UI thread via _ProgressBridge,
and worker exceptions are re-raised on the main thread, preserving the
JSON-RPC error path. All eight upload/download handlers converted.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
test_instanced_geometry previously re-implemented vertex quantization
inline, with a stale comment claiming the helpers still lived in
ViewportWindow.cpp. They now live in VertexQuantization.h, so route the
test through the real quantizeVertex/octEncodeNormal and add coverage
for the degenerate-axis path, octahedral normal round-trip, the i8
normal error bound (~0.78 deg worst observed), and color passthrough.
Add test_visibility and test_selection: Tier-1 coverage of the two
per-object viewport state machines. Both are QObjects for their
changed() signal but touch no GL on the construction/mutation path, so
the tests exercise the pure CPU logic without a context.
Suite goes from 39 to 61 cases.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Introduce pytest coverage for the previously untested connector — rpc,
cache, settings, autodesk (auth + APS client) and connector handlers —
94 tests, runnable via the new `test` optional-dependency extra.
To make HTTP, time and the OAuth redirect testable without a network or
real sockets, add dependency-injection seams to autodesk.py:
AuthSessionService and ApsClient accept an optional httpx transport;
AuthSessionService accepts an injectable clock and callback_waiter; and
_wait_for_callback is extracted to the module-level wait_for_oauth_callback.
All seams default to the previous behaviour.
Remove the APS_CLIENT_ID environment-variable override: the client id now
comes solely from settings.json, collapsing settings.load_client_id and
simplifying the settings dialog.
CI: the build-bonsaiviewer-autodesk workflow gains a `test` job
(Python 3.11 + 3.13) that gates the build matrix.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Compile the Bonsai Viewer as part of the Linux and Windows binary builds,
and ship the Autodesk connector alongside the viewer executable.
Qt6 dependencies:
- The viewer links Qt6::Svg for runtime icon tinting. Svg is a separate
base-Qt archive, so aqt now installs "qtbase qtsvg" (plus icu on Linux)
rather than qtbase alone, on both Linux and Windows.
- Qt6::CorePrivate is exposed differently across Qt versions: Qt 6.8 ships
the target inside Qt6Core, while Qt 6.10 provides it only as a separate
CorePrivate config package. The viewer CMakeLists requests it via
OPTIONAL_COMPONENTS so it resolves on both.
- When cross-compiling Windows ARM64, windeployqt runs from the host x64
Qt, so qtsvg is installed into the host Qt as well.
Windows build:
- build-all-win.py passed -DBUILD_IFCVIEWER, a flag since renamed to
BUILD_BONSAIVIEWER, so the Windows build compiled no viewer at all. It
now passes -DBUILD_BONSAIVIEWER.
- The Autodesk connector is bundled under connectors/ next to
BonsaiViewer.exe in the packaged archive, mirroring the Linux builds.
- The Windows workflow builds the connector (PyInstaller) before the main
build so it is available to bundle.
Connector bundling:
- The Linux rocky workflows build the connector and bundle it into the
BonsaiViewer archive; the Windows build now does the same.
Generated with the assistance of an AI coding tool.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Connector discovery scanned a per-user data directory
(QStandardPaths::GenericDataLocation -> ~/.local/share/IfcOpenShell/
BonsaiViewer/connectors and the macOS/Windows equivalents). Connectors
are now meant to ship with the application, so there is no reason to
look outside the install tree.
Replace userConnectorsDir() with bundledConnectorsDir(), which returns
QCoreApplication::applicationDirPath() + "/connectors". discoverConnectors()
scans only that path; its first-wins / malformed-manifest handling is
unchanged.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Models panel: replace the manual resizeEvent column-sizing hack with
QHeaderView Stretch/Fixed modes, re-applied via sectionCountChanged so
they survive the model rebuilds that QHeaderView resets them on.
Dialog: only wrap the body in a QScrollArea when scrollable, mirroring
Panel. The scroll area caps its sizeHint at 36x24 cells, which turned
wide fixed-size dialog content into spurious scrollbars.
Add Model dialog: reserve a stable, font-metrics-measured height for the
hover description so longer text never reflows the buttons; regroup the
buttons into LOCAL / CLOUD / TOOLS.
Buttons: move the trailing-separator decision out of makeButtonGroup
into a new addButtonGroups row builder, so the last group in a row
never draws a dangling divider.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Loading a federated project (.ifcfed) with several models segfaults
non-deterministically on a fresh start. The viewer's SceneLoader spawns
one background std::thread per model in startDataSourceLoad() to
construct an ifcopenshell::file; with cached sidecars all models reach
that point near-simultaneously, so multiple threads parse different IFC
files at once. Parsing touches the process-wide schema singleton, which
was not thread-safe in two places.
Race 1 — concurrent schema population
-------------------------------------
schema_registry::get() lazily runs the schema's get_() function (e.g.
Ifc4::get_schema() -> IFC4_populate_schema()) and mutates entries_ with
no lock. Two threads calling schema_by_name("IFC4") at once both run
IFC4_populate_schema() concurrently, which fills global arrays
(IFC4_types[], strings[]). One thread reads a slot the other is still
writing.
Core-dump evidence (gdb thread apply all bt):
Thread 1 SIGSEGV in IFC4_populate_schema Ifc4-schema.cpp:1989
<- Ifc4::get_schema
<- schema_registry::get schema.cpp:241
<- schema_by_name("IFC4")
<- ifcopenshell::file::file (NWCH-PIR-SS...ifc)
<- SceneLoader::startDataSourceLoad lambda SceneLoader.cpp:315
Thread 3 also in IFC4_populate_schema (entity ctor for
"IfcMaterialProfileSetUsageTapering")
<- Ifc4::get_schema
<- schema_registry::get schema.cpp:241
<- ifcopenshell::file::file (NWCH-PIR-PT...ifc)
<- SceneLoader::startDataSourceLoad lambda
Two threads inside IFC4_populate_schema() at the same time is the race.
Fix: guard schema_registry's bind()/get()/names()/clear() with a
recursive_mutex (recursive because get() re-enters bind() via
load_schema_plugin(), and a freshly populated schema registers itself
through register_schema()). get() is serialized, so only the first
thread populates the schema; the rest block briefly and then observe
the finished result. Returned schema pointers are stable for the
process lifetime, so holding the lock only across get() is sufficient.
Race 2 — lazy all_attributes_ cache filled during parsing
---------------------------------------------------------
entity::all_attributes() lazily fills a `mutable` optional cache on the
shared schema entity the first time it is accessed — and that first
access happens during parsing (parse_context::construct), not during
schema population. With race 1 fixed, two parser threads still raced
here: both saw the cache empty, both did all_attributes_.emplace() and
std::copy() into it, corrupting the vector.
Core-dump evidence after the race-1 fix:
Thread 1 SIGSEGV in attribute::type_of_attribute (this=0xe130...55c)
<- std::transform(first=0x4, last=0xb0d1...) <-- garbage
iterators into a corrupt std::vector
<- parse_context::construct over
decl->as_entity()->all_attributes() file.cpp:249
<- instance_streamer::read_instance
<- ifcopenshell::file::file (NWCH-PIR-PT...ifc)
<- SceneLoader::startDataSourceLoad lambda
The begin pointer 0x4 is a half-written vector being read mid-resize by
another thread.
Fix: force every entity's all_attributes_ cache in the
schema_definition constructor, while construction is still
single-threaded. The schema is then genuinely immutable after
construction, so concurrent parsing needs no hot-path lock.
Both crashes reproduce reliably on a fresh start at native speed but
vanish under gdb (which serializes thread scheduling) — the classic
signature of a data race. With both fixes the federated load completes
cleanly.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Create a standalone Sphinx docs tree for Bonsai Viewer and migrate the Autodesk connector Markdown documentation into RST.\n\nGenerated with the assistance of an AI coding tool.
Replace the old IFC viewer build switch with BUILD_BONSAIVIEWER in CMake, the Linux workflows, and the nix build script.
Generated with the assistance of an AI coding tool.
Replace the "Open Recent coming soon" placeholder with a working
most-recently-used project list. RecentProjects persists .ifcfed paths
via QSettings, capped and pruned to existing files. The Open Recent
ribbon button now shows a popup menu of recent projects; every
successful open or save (local or cloud) records an entry.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Sort hubs, projects, folders and files alphabetically. Allow
multi-select when adding models so several can be pulled at once.
Rework the progress dialog into a fixed-shape two-line layout that
shows percent and byte counts, middle-eliding long filenames so the
window never reflows.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Persist the OAuth callback port in connector settings, expose it in the settings dialog, and use it when constructing the localhost callback URL.
Generated with the assistance of an AI coding tool.
Rename the user-facing serialisers page to formats and document .rdbview as a Bonsai Viewer package.\n\nGenerated with the assistance of an AI coding tool.
Follows the host viewer's rename to Bonsai Viewer: directory, Python
package, entry point, PyInstaller spec, keyring service, and on-disk
config/cache paths all use the bonsaiviewer-autodesk name. CI workflow
filename and path filters updated to match.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Directory src/ifcviewer-full -> src/bonsaiviewer, CMake target
IfcViewerFull -> BonsaiViewer, namespace ifcviewerfull -> bonsaiviewer,
QApplication / window titles / connector path now use the new brand.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Build the ifcviewer-autodesk connector bundle on push/PR/dispatch for the four supported targets: linux-x86_64 (ubuntu-22.04, oldest reasonable glibc), macos-arm64, macos-x86_64, and windows-x86_64. Each job runs packaging/build.py and uploads the resulting autodesk-<os>-<arch>.zip as an artifact.
Generated with the assistance of an AI coding tool.
Implements the viewer side of CLOUD_SYNC_PROTOCOL.md: connector
discovery, JSON-RPC stdio host, and Open/Save/Sync/Add cloud workflows
wired through the ribbon, Models panel right-click, and Settings tab.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Initial implementation of the ifcviewer-autodesk connector — a separate process that bridges the IfcViewer to Autodesk APS (BIM 360 / ACC). Speaks JSON-RPC 2.0 over stdio per CLOUD_SYNC_PROTOCOL.md (also added). PKCE OAuth with keyring-backed token storage, customtkinter browse/picker UI, and PyInstaller packaging.
Implements both interactive and non-interactive variants of each push/pull (pull_ifcfed[_interactive], pull_models[_interactive], push_ifcfed[_interactive], push_model[_interactive]) so the viewer can offer both "Save"/"Open from Cloud" and "Save As"/"Add Model from Cloud" entry points. File transfers report progress through a dialog with per-byte updates; pull_models shows "(i/N)" for batches.
Generated with the assistance of an AI coding tool.
Use IfcMapConversion.Scale as the source of truth for converting map coordinates to metres, instead of deriving that scale from IfcProjectedCRS.MapUnit. Bump the sidecar version because cached georef matrices and unit scales may differ under the new interpretation.
Generated with the assistance of an AI coding tool.
Keep placement transformations in double precision through streaming, sidecar caching, and viewport recomposition so large coordinates can be cancelled before the final GPU float upload.
Generated with the assistance of an AI coding tool.
Renamed HeadlessSidecarBuilder to SidecarBuilder and reused it for live
loads. SceneLoader now constructs one per stream load, forwards meshReady
/instanceReady chunks alongside the viewport upload, and finalizes +
writes the sidecar at onStreamerFinished — no more GPU readback path
via ViewportWindow::snapshotModel (removed). Same code path now produces
sidecars for both live loads and the .rdbview offline export.
Sidecar use is opt-in per direction via SceneLoader::setShouldReadSidecar
and setShouldWriteSidecar; both default off so embedders that don't want
caching get a pure-streaming loader. ifcviewer-full and ifcviewer-minimal
opt in.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Wire a new "Export Geometry Database" tool button in AddModelDialog,
adjacent to "Convert IFC File to Database", to produce a zipped
read-only artifact combining a lossy RDB (with IfcRepresentationItem
stripped) and a .ifcview geometry sidecar. Intended for cloud
coordination workflows where parametric geometry editing is not needed.
Pipeline changes to support this:
- document_serializer_context gains a `skip_supertypes` field; the
rdb plugin forwards it to RocksDbSerializer so the same registry
path produces full or lossy RDBs.
- Vertex quantization helpers (octEncodeNormal + quantizeVertex) move
out of ViewportWindow.cpp into a shared header so the sidecar's
byte layout stays identical regardless of whether it came from a
GPU readback or a CPU pipeline.
- New HeadlessSidecarBuilder runs a GeometryStreamer on the calling
thread, captures MeshChunk/InstanceChunk into a SidecarData on the
CPU, then computes georef + packed elements + LODs and writes the
.ifcview — no ViewportWindow or GL context required.
The Controller's export flow runs RDB conversion + sidecar build +
QZipWriter packaging on a background QThread, writing through
`<dest>.tmp` then renaming for atomic appearance in cloud-sync
folders. ifcviewer-full now links Qt6::CorePrivate for QZipWriter.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Wire the AddModelDialog "Convert IFC File to Database" button to a new
ConvertToDatabase source mode handled by ModelsPanelController, which
prompts for an .ifc input and .rdb output then runs the existing
document_serializer_registry "rdb" plugin on a background QThread with
a modal progress dialog.
Build the src/serializers subdir for BUILD_IFCVIEWER so the rdb plugin
is produced, and align serializer plugin runtime output with the
kernel/mapping plugins by writing them into $<TARGET_FILE_DIR:IfcGeom>
so default plugin discovery finds them in both dev and install layouts.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Rename leftover interface-era namespaces, settings, and resource identifiers inside the IfcViewerFull source tree without changing the public target name.
Generated with the assistance of an AI coding tool.
Replace the old IfcViewerFull application tree with the interface-based viewer while preserving the IfcViewerFull target and build workflow.
Generated with the assistance of an AI coding tool.
Move model georeferencing state and rendering into a dedicated settings view, and show live IFC coordinate operation and unit data in the dialog.
Generated with the assistance of an AI coding tool.
Style hidden interface models with disabled text and move the first length-tool pick coordinates to the HUD as ENH in the global georeferenced frame.
Generated with the assistance of an AI coding tool.
Show a real status-bar progress bar for interface model loads by wiring the shell window to SceneLoader progress and completion signals.
Generated with the assistance of an AI coding tool.
Add group rename and reparenting, model-to-group moves, drag-and-drop reassignment, and clearer group creation actions in the interface models panel.
Generated with the assistance of an AI coding tool.
Port the streamed-model sidecar writeback path into the interface, including packed element metadata, georef persistence, LOD generation, and viewport LOD application.
Generated with the assistance of an AI coding tool.
Keep the IfcViewer test CMake in sync with the optional meshoptimizer dependency so test_lod_builder is only added when the package is enabled.
Generated with the assistance of an AI coding tool.
Adds VisibilityState, a CPU-only sibling to SelectionState. It owns
the canonical hidden-id set plus a flat per-object_id byte vector that
the cull's hot path queries inline (bounds check + byte load + compare
per surviving instance). Hidden elements never reach the visible[]
SSBO so they don't draw or pick — matching Blender/CAD convention.
ViewportWindow registers every streamed and sidecar-cached object_id
with the new state, resets it on clearScene, and connects the changed
signal to invalidate cached cull state. Three convenience verbs:
hideSelectedElements (union into hidden), isolateSelectedElements
(replace hidden with live-object_ids minus selection, skipping
model-hidden models so element-hide doesn't pile on top of model-hide),
and showAllElements (clears the override; model-hidden models stay
hidden, per the user's spec).
Bound in the View menu: H hide, Shift+H isolate, Alt+H show all.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Promotes five env-var-driven knobs to AppSettings + the settings dialog
(min pixel radius, motion min pixel radius, LOD1 pixel threshold, HiZ
resolution, HiZ on/off). Defaults: motion min pixel radius is now 10
(was 0/disabled) and IFC_HIZ_MOTION is on by default — the strict
view-projection gate reverts via env var =0 when chasing HiZ
correctness bugs. ViewportWindow connects each *Changed signal so
changes invalidate cached cull state and take effect on the next
frame.
Removes "Load Property Data Source" and "Apply Coordinate Operation"
from the settings dialog: both are now hardcoded on. The basic-info
property fallback (used when there's no live IFC source for an object,
e.g. .ifcview without a sibling) now triggers organically when
ElementRegistry::findEntity returns null instead of being gated on a
user toggle. Federation::guessFederatedFalseOrigin lost its
apply_coordinate_operation parameter and now uses
georef.has_coordinate_operation directly.
src/ifcviewer/settings.rst documents the remaining diagnostic env vars
(IFC_HIZ_MOTION, IFC_CULL_THREADS, IFC_SKIP_MDI, IFC_MAX_SUBDRAWS,
IFC_FPS_HITCH_MS, IFC_SUBDRAW_DIAG, IFC_LOD_*) plus a cross-walk from
the old promoted-knob env-var names to their new QSettings keys.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Mirrors the Area tool's display: HUD shows total volume + object count,
each selected object gets a label at its world-AABB centroid showing
its individual volume. Gated behind ToolMode::Volume (Ctrl+Shift+V) so
it stays out of the way until invoked.
Volume is a passive tool — selection works as in None (multi-select,
modifier toggle, box-select all keep working). Area / Length still
intercept clicks through surfacePickedInTool.
Adds volumesPerObject() reusing the same mesh-cached readback path as
volumeOfObjects, so the per-object split costs no extra GL readbacks.
computeObjectAabb is promoted to public for the centroid lookup.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
SelectionState (new) owns the multi-set, the "active" id (last single-
clicked), and a per-object_id flags SSBO bound at binding=3. Main
shader reads sel_flags[v_object_id] for the in-set tint and a separate
u_active_id uniform for a stronger tint on the active.
Click semantics: plain replaces, Shift/Ctrl toggles. LMB-drag past 5px
boxes the rect through a pick-pass readback — plain replaces, Shift
adds, Ctrl removes; box-select preserves the active. Drag promotes
regardless of start point so a press on geometry doesn't disqualify it.
Sidecar fast-path bulk-loads instances, so noteObjectId is also called
from the apply path — without it the flags buffer was sized to 1 slot
while object_ids were in the 100k+ range and the in-set bit was lost.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Add the federation/model settings dialog and related interface wiring for model coordinate configuration.
Generated with the assistance of an AI coding tool.
Move interface features from panels into modules, move AddModelDialog into the models module, and rename module Widget surfaces to Panel.
Generated with the assistance of an AI coding tool.
The streamer carved [0,100] evenly across N prioritised contexts (and
again across the net/gross passes). In practice nearly every element
yields from the first (Body) context, so smooth progress only ever
filled range/n of the bar — typically ~20% — before snapping forward.
Drive progress directly from yielded element count over total instead.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Area mode now drops a label at every connected coplanar patch — a
BFS sweep over selected_ restricted to each mesh's edge adjacency
identifies the components, then each component gets one label at
its area-weighted centroid in world space. Two clicks on
different walls now show two distinct numbers; a single BFS-grown
wall face stays one number across all its triangles.
The 2-pt length perpendicular line is now omitted when |perp|
matches any of ΔX/ΔY/ΔZ within 1mm — the surface-aligned-with-
axis case where the perpendicular is already shown by one of the
RGB legs. Avoids redundant double-readout on axis-aligned walls.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Length tool's 1-pt laser is now hybrid:
- On any surface, a coplanar BFS finds the connected face patch
around the click and projects its vertices into the surface
tangent basis to get an exact bounding-box extent. Stops at
the face edge by construction — no overshoot into adjacent
geometry like the previous tangent-raycast did.
- On near-horizontal surfaces (|n.z| > 0.85, i.e. floors and
ceilings) it additionally fires one raycast in +n to the
opposing surface — so a single floor click reports X extent +
Y extent + ceiling height.
- Bars are labelled by their dominant world axis (X/Y/Z) instead
of "vertical/horizontal", which reads cleanly on either kind
of surface.
The 2-pt readout now draws the world-space XYZ stair-step (red ΔX,
green ΔY, blue ΔZ) with each leg labelled, and a dashed
perpendicular line whenever the two picks landed on near-parallel
surfaces — useful for measuring across walls.
To support multiple line styles per frame, OverlayRenderer's
setOverlayLines takes std::vector<LineGroup> instead of a single
inline style; each group has its own color/halo/width and an
optional dash period. The line shader gained v_along_px +
u_dash_period uniforms (screen-space dashes), and both line and
point shaders now use a sharp step() for the inner→stroke
transition with AA only on the outer halo edge — much crisper than
the previous soft band. Default visual style trimmed: 1.5px lines
(0.5px halo), 6px dots (1px halo), opaque black halo.
Also adds ViewportWindow::raycast(origin, dir, RaycastHit&) — CPU
ray traversal of each model's per-instance BVH followed by
Möller-Trumbore against the candidate meshes' triangles (lazily
read back, cached per call). Used by the floor/ceiling laser path
today and reusable for any future raycast-based feature.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Treat .ifcview sources as geometry-only cache inputs, stop guessing sibling data paths, and only start data-source loading for real model sources after a sidecar hit.
Generated with the assistance of an AI coding tool.
ViewportWindow trades the area_tool_active_ bool for an enum ToolMode
{None, Area, Length}; the existing surfacePickedInTool signal carries
both, the app dispatches on toolMode(). Esc exits any active tool;
Backspace/Delete in length mode emits toolBackspacePressed which the
length tool uses to remove the last point.
LengthMeasurement collects clicked world-space points and adapts the
readout: 2pt → distance + axis-aligned ΔX/ΔY/ΔZ, 3pt → angle at the
middle vertex + triangle area, 4+pt → best-fit-plane PCA + shoelace
when planar (RMS plane distance / bbox diag < 1e-3) else fan
triangulation, with the chosen method labelled in the readout. Per-
segment lengths float at each midpoint.
OverlayRenderer grows three new pipelines to support this:
- point sprite shader: gl_PointCoord-based outlined disc with
fwidth-smoothed inner/stroke bands, a single draw call.
- line shader: CPU-expand each segment to 6 verts carrying both
endpoints + (side, along) corner index; vertex shader computes
the screen-space perpendicular and offsets accordingly. Real
outlined lines independent of the driver's glLineWidth clamp.
- screen-space rect shader: HUD + label backgrounds drawn as raw
GL quads in NDC. QPainter::fillRect on QOpenGLPaintDevice was
silently dropping fills across drivers; bypassing it entirely
via this shader makes backgrounds reliable. Cull-face is also
explicitly disabled here — GL_TRIANGLES respects it but the
line/point primitives don't, so this was the one path needing
the fix.
setOverlayLines / setOverlayPoints take an inner color, an outline
color, and an extra-pixels-per-side stroke amount. Lines + points
draw with GL_ALWAYS so measurement annotations stay visible through
geometry; highlight tris stay depth-aware (GL_LEQUAL) so area
shading still tints the surface in place.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
test_parametric_lifecycle.py covers the door/window/railing/roof
state-transition contracts (enable/finish/cancel; no-op on
non-matching elements; draft preserved on finish-time failure)
that the registry smoke test never exercised.
test_parametric_registry.py gains a check that every is_<name>
predicate stays total (never raises on a non-matching IFC entity)
— a raising predicate would break the save path for unrelated
types. Also rewrites the gizmo-prefs check to read __annotations__
instead of hasattr, which depended on Blender registration timing.
Generated with the assistance of an AI coding tool.
Aligns with the test/bim/ convention: heavy imports go inside test
functions so the autouse _require_real_bpy fixture skips cleanly
when bpy is mocked, rather than module-level imports failing at
collection time and erroring out the whole file.
Generated with the assistance of an AI coding tool.
PowerShell and some wrapper scripts on Windows occasionally strip
or reorder the `--` separator before Blender sees it, dropping the
pytest args into Blender's positional file-load slot ("File format
is not supported"). The env var carries the same args via a
shell-evaluation-free channel. Default `--` path is byte-identical
to the pre-change behaviour.
Generated with the assistance of an AI coding tool.
Docstrings naming sibling methods, private helpers, test files, or
historical symbols silently go wrong on rename. Strip Sphinx :meth:
/ :class: / :func: / :attr: markup that mostly added noise (no
Sphinx in this project), and rewrite five docstrings that cited
specific test paths or private hooks to describe the behaviour
instead.
Generated with the assistance of an AI coding tool.
The dialog's only outcomes were "Apply & Save" (same as silent save)
or "Cancel" (same as not saving) — net friction with no actual choice.
Auto-commit stays as the safety net; the count now suffixes the
existing save-success report so it isn't immediately overwritten.
Generated with the assistance of an AI coding tool.
When two GizmoDimension hit regions overlap (a short dimension
nested inside a longer one along the same axis), the larger one
used to win because hit boxes are scaled by world-space length —
the long box fully contains the short one, leaving the short
gizmo unreachable. The larger gizmo stays clickable at its
exposed ends, so smaller-wins is the right UX default.
Sets self.select_bias = -self._dimension_length inside
GizmoDimension.set_dimension_length. The smaller gizmo writes a
less-negative depth value in the GPU select buffer and wins the
tie-break. select_bias is unused elsewhere in the codebase, so
icon and arrow gizmos keep bias=0 and are unaffected (icons
correctly still win against dimensions, since 0 > -length).
Adds test/bim/module/drawing/test_dimension_gizmo_priority.py
with 5 cases: direct ordering, monotonicity across length ranges,
abs() handling for signed dimensions, and NaN/Inf safety.
Generated with the assistance of an AI coding tool.
set_icon_gizmo_position computed
``mw @ (Translation @ billboard_rot @ Scale)`` — the object's world
matrix was applied AFTER the billboard rotation, so any non-trivial
object rotation (e.g. a wall rotated in plan, a stair rotated to
match a corridor) carried over into the icon's transform and tilted
it edge-on to the camera instead of facing it.
Switch to ``billboarded_at(world_pos, billboard_rot, scale)`` where
``world_pos = mw @ local_pos``: translate to world space first, then
apply the billboard rotation independently of the object's rotation.
This matches the manual pattern the base class's
``update_editing_gizmos`` already uses for validate/cancel/cycle for
exactly this reason.
Drops the now-stale workaround docstring on
``GizmoWallEdition._update_icon_row_extras`` that documented why it
bypassed ``set_icon_gizmo_position`` — the helper does the right
thing now.
Adds ``test/bim/module/model/test_stair_gizmos.py`` as the regression
guard: parametrised over six rotation angles, asserts that the rotation
part of the resulting matrix equals ``billboard_rot`` (no contribution
from ``mw``'s rotation) and that the translation lands at
``world_pos``. Also exercises ``set_icon_gizmo_position`` end-to-end via
a stub gizmo to catch the exact shape of the previously-broken call
site.
Generated with the assistance of an AI coding tool.
Walls gain in-viewport parametric editing matching the door/window/stair
UX: drag handles for length, height, slope (x-angle), layer baseline
cycle, plus cursor-anchored quality-of-life operators (split at cursor,
extend to cursor, extend height, rotate 90, toggle openings) and
two-object state-machine gizmos (unjoin / merge / join-corner /
extend-to-wall / extend-vertically / add-opening).
Wall enters tool.Parametric.EDIT_TYPES, so save-time auto-commit,
GizmoPreferencesWall registration, and the in-progress-edit predicates
all light up automatically through the registry plumbing landed two
commits back.
The three-layer commit model (drag -> BIMWallProperties -> bmesh
preview -> Finish -> single ifc.run) means dragging a handle through
hundreds of intermediate values produces zero extra IFC entities. A
no-op enable->finish round-trip is byte-identical. The snapshot diff
in FinishEditingWall skips unchanged params.
_commit_active_wall_edit_if_any ensures cursor-anchored operators see
committed geometry, not the draft preview box.
Also lands the `prompt_auto_commit_parametric_edits` BoolProperty on
BIM_ADDON_preferences (consumed by the auto-commit dialog landed in
the framework commit) and refactors
`draw_{door,window,stair}_gizmo_parameters` into a shared
`_draw_parametric_gizmo_parameters` helper that the new
`draw_wall_gizmo_parameters` reuses. This commit and the framework
commit are stacked - the framework commit references the BoolProperty
defined here, so they must land together.
Tests cover pure math (core/test_model.py), DimensionGizmoConfig text
formatter, GizmoWallExtendVertically.poll() preconditions, and the
refresh_post_commit cache-invalidation regression. BDD scenarios in
model.feature cover the edit triad, auto-commit on save, and the
two-object gizmos. Documentation added to creating_walls.rst.
Generated with the assistance of an AI coding tool.
Three small post-landing cleanups against the parametric framework commit:
* core/model.py had `are_axes_collinear` and `closest_endpoint_midpoint`
each defined twice — Python silently kept the second copy, the first
was dead code. Removed the dead copies; runtime behavior unchanged
(the live versions were already the kept ones).
* bim/__init__.py's `_parametric_gizmo_preference_classes` docstring
named the wrong link in the import chain (`tool.blender → bim.ifc`).
The real chain is `tool/ifc.py` (and ~6 other tool/* modules) which
import `from bonsai.bim.ifc import IfcStore` at module load. Updated
docstring to cite that root cause and the architectural fix (move
`IfcStore` out of `bim/`).
* tool/blender.py's `from bonsai.bim.ifc import IFC_CONNECTED_TYPE`
carried a 5-line comment claiming it was "lazy" to avoid a circular
load. The import sits inside an `if TYPE_CHECKING:` block with
`from __future__ import annotations` — it never runs at runtime
regardless. Comment removed; the TYPE_CHECKING guard is
self-explanatory.
Generated with the assistance of an AI coding tool.
Establish a single source of truth for parametric element types (door,
window, stair, railing, roof). tool.Parametric.EDIT_TYPES drives:
- BIM<Name>Properties PointerProperty attachment via the registry
- GizmoPreferences<Name> class registration in bim/__init__.py
- save-time auto-commit of pending draft edits
- the refresh_post_commit epilogue called from IfcStore after every IFC
mutation, which fixes the stale-header bug where in-place hotkey
mutations (S_E / C_E) left BIMModelProperties and the gizmo cache
pointing at obsolete values.
Refactors door/window/railing/roof onto shared mixins from
bim/parametric_lifecycle.py (FeatureModifierEditMixin and
PathPreservingEditMixin); stair gets the lock-gizmo refactor and
frame-cache integration. Behavior preserved.
Adds BaseParametricGizmoGroup._prime_frame_caches so the parametric
gizmos stop re-deriving preferences, view direction, and billboard
rotation per frame; reorders poll() to short-circuit on the cheapest
predicate first. Adds the icon library + BillboardingGizmoGroupMixin
that the wall feature in the next commit will consume.
Generated with the assistance of an AI coding tool.
When a project has a ifc file associated, selecting non-ifc objects and duplicating them with SHIFT + D now correctly both duplicate them, keep the new objects selected and starts the transform modal. IFC objects behaviour is unaffected.
New OverlayRenderer module owns every client-supplied overlay primitive
drawn after the main pass: tinted, depth-aware highlight triangles via
its own GL shader, and top-left HUD text via QPainter on a
QOpenGLPaintDevice. Public surface on ViewportWindow is just two
forwarders (setHighlightTriangles, setHudText).
ViewportWindow's MeshLocalPick now exposes the instance's composed
transform so consumers can map mesh-local geometry back to world
space without re-querying. AreaMeasurement uses both: its selection
key is now (object_id, tri) so per-instance highlighting works for
two distinct walls sharing a mesh, and on every pick it rebuilds the
world-space tri list and the HUD readout.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds a click-to-measure area mode triggered by Ctrl+Shift+A. Each LMB
click expands the picked triangle into its connected coplanar patch
(BFS over shared edges, dot(normal, seed) > 0.9999); re-clicking
removes that patch; Alt+LMB skips expansion for a single triangle.
Picks across different meshes accumulate as separate patches.
ViewportWindow gains pickMeshLocalAt (screen pick → mesh-local hit
via inverse composed transform) and a tool-mode pattern mirroring
the section tool (toggleAreaTool, surfacePickedInTool signal,
areaToolToggled signal, Esc to exit). Per-mesh adjacency is built
lazily on first pick of each mesh and dropped on tool toggle.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds neutral primitives on ViewportWindow (readbackMeshTriangles,
findInstance) so consumers can compute per-object geometry queries
without the library retaining a CPU triangle copy. Measurement.cpp in
ifcviewer-full uses them to sum signed-tetrahedra in mesh-local space,
weighted by |det(placement_3x3)| per instance for mapped-item scaling.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Read-only handles reject Flush/CompactRange, so the destructor's status
assertion always fired on shutdown when the streamer's sidecar was
opened with read_only=true. Track the flag and skip the write path; also
guard against a null db when the initial open failed.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Federation gains a nested Group tree (id, display_name, visible,
children); models reference a single group via Model::group_id.
Visibility cascades: a model is effectively visible only when its own
flag is on and every ancestor group is visible. Persistence nests
groups directly in the JSON — no parent_id field.
ifcviewer-full surfaces this in the element tree with right-click
menus to create / rename / move / remove groups, move models between
groups, and toggle group visibility.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Right-click a model root in the Elements tree to get Hide/Show and
Remove. Hide flips the federation's per-model visible flag (already
round-tripped to .ifcfed), pushes ViewportWindow::hideModel/showModel,
and italicises + greys the tree root as a visual cue. Remove drops
the model from the viewport, the SceneLoader (streamer + caches), the
MainWindow UI maps and tree, and the Federation — disabled while the
model is the active load.
Visibility is reapplied on each model's load completion (sidecar or
stream), so a federation saved with hidden models opens with them
hidden. clearScene() now also drops SceneLoader state so streamers
no longer leak across federation transitions.
API additions:
- Federation::setModelVisible + modelVisibilityChanged signal
- SceneLoader::removeModel + isLoadingModel
Tests cover the setter (dirty + signal + idempotence + unknown id);
extends the existing round-trip test to actually exercise the
visibility load/save it always claimed to.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Commit f7add7f4 split getAxis2Placement out of an anonymous helper in
Geolocation.cpp into a shared Placement.{h,cpp}, but the test_federation
target's source list wasn't updated. The test binary failed to link
with `undefined reference to getAxis2Placement(express::Base const&)`
from Geolocation::getWcs. Add Placement.cpp to the explicit-source
list — it has no Qt dependency, only ifcparse.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
When the user adds a model into a fresh, untitled federation that still
has the default (0,0,0, no rotation) FederatedFalseOrigin, derive an
origin from the first instance's placement_transformation (lifted
through CoordinateOperation when enabled) and the helmert grid-north
baked into ModelGeoref::coordinate_operation_meters. Multi-file batches
naturally settle: whichever load finishes first anchors the federation,
the rest see a non-default origin and skip. Saved .ifcfeds keep their
authoritative origin.
Adds Placement.{h,cpp} (port of util/placement.py — a2p,
get_axis2placement, get_local_placement) so Geolocation no longer needs
its own anonymous getAxis2Placement, and xaxis2angleDeg in Geolocation
mirroring util/geolocation.xaxis2angle.
SceneLoader captures the first instance's placement_transformation from
either the sidecar's InstanceCpu[0] or the streamer's first
InstanceChunk, so the guess works on both load paths without re-reading
the IFC.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Previously, applyCoordinateOperationToViewport — which pushes both
CoordinateOperation and ModelTransformation — was only called on
paths that required the IFC source to be loaded
(onLoadedFromStream and onDataSourceReady). Sidecar-only loads
(loadDataSource off, or no .ifc/.rdb sibling) silently lost both
stages.
Cache the per-model georef + unit scales in the sidecar itself so
the IFC source isn't needed to apply them:
SidecarData gains
coordinate_operation_meters[16] // column-major
project_length_to_meters
map_unit_to_meters
has_coordinate_operation
148 B fixed block written/read between instances and elements.
SIDECAR_VERSION 10 -> 11; existing sidecars rebuild on next load.
MainWindow::writeSidecarForModel populates the block from
loader_->modelGeoref(mid) before writeSidecar.
SceneLoader::applySidecarData restores it into the model's
ModelGeoref + sets has_georef = true, so subsequent
loader_->modelGeoref(mid) calls return the cached data without
needing the IFC.
MainWindow::onLoadedFromSidecar now calls
applyCoordinateOperationToViewport(mid) directly — both
CoordinateOperation and ModelTransformation land at sidecar-load
time, no longer waiting on a possibly-never-arriving data source.
Edits to the IFC's IfcMapConversion don't invalidate the cache —
delete the .ifcview manually if the source's georef changes. This
matches the existing cache-invalidation contract.
Tests: round-trip the new fields through the existing sidecar
fixture; assert SIDECAR_VERSION == 11.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Tree -> viewport selection was already wired (onTreeSelectionChanged
calls setSelectedObjectId), but pressing F afterwards routed to the
focused tree widget rather than the viewport, so framing didn't fire.
Add a window-level View > Frame Selected QAction with Qt::Key_F that
delegates to ViewportWindow::focusOnSelectedObject — works regardless
of which child widget has focus. The viewport's own F handler stays
in place for when the viewport itself owns focus.
For debugging coordinate problems, add View > Print Selected Coords
(Ctrl+Shift+P) -> ViewportWindow::printSelectedObjectCoords, which
qInfo's:
- a sample vertex (first vertex of the selected mesh, decoded on
demand from the quantised VBO so no extra CPU storage is needed);
- placement_transformation (the per-instance matrix that maps the
sample vertex from mesh-local into the model's pre-georef frame);
- global = CoordinateOperation . placement_transformation (where
the IFC's own IfcCoordinateOperation has been folded in);
- the sample vertex transformed through both matrices.
The print is a no-op when nothing is selected or GL hasn't initialised.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
ModelTransformationDialog edits one federation model at a time. Top
combobox picks the model; below it the form covers the four pieces
of authoring intent:
- AFrame radio: ModelLocal vs ModelGlobal
- Point A: 3 doubles, label switches between "model project length
unit" and "model map unit" with the radio
- Point B: 3 doubles in federation units (label reflects current
FederationConfig.unit_*)
- Rotation: rx/ry/rz in degrees, intrinsic XYZ
- Pivot: 3 doubles in federation units
Switching models discards unsaved form edits — Ok saves the
currently-visible model, Cancel discards. On Ok calls
Federation::setModelTransformation, which fires
modelTransformationChanged → MainWindow recomposes that model in
the viewport.
Reachable from File > Model Transformations.
End-to-end is now editable: open a federation, edit federation unit /
false origin from one dialog, edit any model's transformation from
the other, watch the viewport recompose live. Visual verification
on a real model still pending.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
New FederationSettingsDialog edits the federation-wide unit and the
FederatedFalseOrigin (XYZ + Z-rotation in that unit). On Ok it calls
Federation::setConfig + setFederatedFalseOrigin, which fire the
granular Federation signals MainWindow listens to → viewport
recomposes immediately.
Reachable from File > Federation Settings. Unit picker is a fixed
combobox of common length units (metres / mm / cm / km / ft / in /
yd / mi); each item carries (prefix, name) in itemData so saving
round-trips correctly. Per-model ModelTransformation editor still
to come — that's a per-model dialog reachable from the model entry,
not the federation-wide settings.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Federation grows three granular signals so consumers can recompose only
what's affected:
- configChanged() — federation unit changed
- federatedFalseOriginChanged() — stage 3 changed
- modelTransformationChanged(fed_id) — stage 4 changed for one model
Emitted from setConfig / setFederatedFalseOrigin / setModelTransformation
in addition to the existing dirtyChanged.
MainWindow gains applyFederatedFalseOriginToViewport and
applyModelTransformationToViewport helpers. Each composes the matrix
from the current federation state (using composeFederatedFalseOrigin /
composeModelTransformation, which already exist on Federation.h) and
pushes to the viewport's setFederatedFalseOrigin /
setModelTransformation. ModelTransformation reads ModelUnits and the
active CoordinateOperation matrix from SceneLoader::modelGeoref so
ModelLocal-frame `a` lifts correctly through stage 2 when authored.
Wiring:
- federation.federatedFalseOriginChanged -> applyFederatedFalseOriginToViewport
- federation.configChanged -> stage 3 + walk all models for stage 4
- federation.modelTransformationChanged -> stage 4 for that one model
- applyCoordinateOperationToViewport now also re-pushes stage 4 (the
compose result depends on the active stage 2 when a_frame is ModelLocal)
- openFederation() pushes the loaded FederatedFalseOrigin once load
completes; per-model stage 4 falls out of the existing
onLoadedFromStream / onDataSourceReady path.
End-to-end pipeline is now active under the AppSettings toggle: edit
the federation in memory and the viewport recomposes immediately. UI
for editing (form-based dialog) still pending.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AppSettings.applyCoordinateOperation (default false, persisted via
QSettings) controls whether each loaded model's IfcCoordinateOperation
is applied at upload time. Off keeps models in their local engineering
frame (current behaviour). On lifts each model into map coordinates
via the stage-2 georef matrix cached on SceneLoader.
MainWindow:
- applyCoordinateOperationToViewport(mid) reads the toggle, fetches
the model's ModelGeoref, and pushes either the
coordinate_operation_meters matrix or identity to the viewport.
- Called from onLoadedFromStream (streamer path) and onDataSourceReady
(sidecar-hit path, where the IFC arrives asynchronously).
- Subscribed to AppSettings::applyCoordinateOperationChanged: a
runtime toggle walks every loaded model and re-applies, so users
can flip georef on/off without reloading.
SettingsWindow gains a "Apply Coordinate Operation" checkbox alongside
the existing per-load toggles.
Default-off so the change is opt-in — users with georeferenced models
(UTM coords etc.) can flip the toggle to see them in their map frame
once they're ready. Visual verification on a real georeferenced
model still pending.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
InstanceCpu now carries both placement_transformation (raw streamer
output, the iterator's per-shape transform with vertex-rebasing offset
folded in) and transform (the composed FederatedFalseOrigin ·
ModelTransformation · CoordinateOperation · placement_transformation
result that lands in the SSBO). World AABBs are recomputed from the
composed transform — frustum/BVH culling sees the actual rendered
position regardless of stage state.
ViewportWindow gains:
- ModelGpuData::coordinate_operation_meters / model_transformation_meters
- federated_false_origin_meters_ (federation-wide member)
- composeInstanceFromPlacement / recomposeAndUploadModel helpers
- public setFederatedFalseOrigin / setModelCoordinateOperation /
setModelTransformation
Each setter rewrites the affected model's SSBO, refreshes the
reflection flags, and rebuilds the BVH. Defaults are identity, so
behaviour is unchanged until something wires a setter up — that's
the next commit (MainWindow listening to Federation::dirtyChanged
and SceneLoader::modelGeoref ready signals).
Sidecar bumped 9 -> 10: InstanceCpu grew 104 B -> 168 B. Existing
sidecars rebuild on next load. v10 sidecars store
placement_transformation, so they remain reusable across .ifcfeds —
the composed transform on disk is overwritten with the right one
on load.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Per-mesh, when the iterator's first source vertex is more than 1 km
from origin (matching bonsai's distance_limit default), pick that
vertex as a rebase offset. buildMeshChunk subtracts the offset from
every emitted vertex (in double precision, narrowed to float at the
end), and each instance's placement matrix is post-multiplied by
T(+offset) so world position is preserved by construction:
T(+offset) · (verts - offset) ≡ T · verts
The offset is stored on the per-mesh MeshAabb so all instances of the
same mesh apply the same compensation. When the mesh's first vert is
near origin (the common case), offset is zero and the work is a no-op
beyond a couple of FP ops per vertex.
Improves float32 precision in the vertex buffer for georeferenced
models (UTM coords etc.) where verts would otherwise have to encode
million-metre magnitudes directly — at 1e6 m, float32 resolves about
6 cm, ruining sub-millimetre detail in the buildings themselves.
Visual verification on a real UTM-coords model still pending — the
math preserves world position by construction but precision claims
warrant a hand-test in the viewer.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds ModelGeoref { ModelUnits units; Eigen::Matrix4d stage2_meters; bool
has_stage2; } and computeModelGeoref(file*) in Federation.{h,cpp}. The
helper reads the project length unit, IfcProjectedCRS.MapUnit, helmert
parameters and WCS, and reduces them to a metres-in/metres-out stage 2
matrix using the existing Geolocation + Unit primitives. When the model
has no IfcMapConversion it returns an identity stage_2 with has_stage2
== false, so the upload pipeline can branch cheaply.
SceneLoader::Model gains a cached ModelGeoref; SceneLoader::modelGeoref
(uint32_t mid) computes lazily on first call (returns nullptr when the
IFC file isn't available yet — happens on the sidecar-hit path before
the data-source thread populates the streamer) and serves from cache
afterwards.
Not yet consumed by the upload pipeline; that's the next commit.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds the structs that were briefly in src/ifcviewer/Federation.{h,cpp}
two commits ago, now folded into the merged Federation alongside the
file persistence layer:
- FederationConfig: federation-wide unit ({prefix, name}). Default
METRE; one-of an IfcSIUnit name with optional prefix or an
IfcConversionBasedUnit name.
- FederationOrigin: stage 3 — XYZ in federation unit + Z-rot.
Composes to R_z · T(-xyz_meters), nominating a point as origin.
- AFrame + ModelTransform: stage 4 intent — A (model project or
map unit, per a_frame), B and pivot (federation unit), full
intrinsic-XYZ Euler rotation in degrees.
- ModelUnits: per-model project_length_to_meters / map_unit_to_meters
cached at load time.
Free functions composeFederationOrigin and composeModelTransform
return Eigen::Matrix4d in metres. composeModelTransform takes the
model's stage-2 georef matrix so it can lift `a` into metres when
authored in ModelLocal.
Federation gains config_, origin_ members + setters that emit
dirtyChanged. Each Model carries a transform_intent. JSON I/O
emits config / origin always; transform_intent only when non-default.
Schema stays "ifcfed/1" — additive, optional, sane defaults.
Five new tests: round-trip of the new fields, default-omission
behaviour, two compose smoke tests for FederationOrigin, and one
verifying the "pivot at B preserves A→B" invariant of
composeModelTransform. All 36 ctest cases pass.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Move src/ifcviewer-full/Federation.{h,cpp} (and its tests) into
src/ifcviewer/ so the lib stays the single source of truth for the
federation data model. Restores the original "agnostic lib usable
from ifcviewer-full and ifcviewer-minimal alike" framing.
Drop the unused per-model transform[16] / has_transform field — it
was round-trip-only with no UI to author it, and is being replaced
by an intent-based ModelTransform in the next commit. No real
.ifcfed in the wild populated this field; old files still load
(unknown JSON keys ignored), they just lose the unused transform.
Replaces the pure-data-model Federation.{h,cpp} that was added a
few commits earlier — that file's structs and compose helpers
return as part of the merged Federation in commit 6.
ifcviewer-full's per-app tests dir is removed (test_federation was
the only one); BUILD_IFCVIEWER_TESTS now wires test_federation in
under src/ifcviewer/tests/, with the Qt6::Core/Gui/Test dependency
declared inline since unlike the other Tier-1 tests it has to pull
Qt in. All 31 tests pass.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
FederationConfig holds the federation-wide display unit (defaults to
METRE; on load the first model's MapUnit becomes the default).
FederationOrigin captures stage 3 — XYZ in federation unit + Z-rot —
and composes to R_z · T(-xyz_meters), nominating a point as the new
origin and rotating around it. ModelTransform captures stage 4 —
A in model project or map unit (per AFrame), B and pivot in
federation unit, full intrinsic-XYZ Euler rotation — and composes to
T(B - R_pivot · A) · R_pivot, rotating first then translating so the
rotated A lands at B.
ModelUnits caches per-model project/map unit-to-metres scales so the
compose helpers don't need to re-read the IFC each call.
All composed matrices are in metres; user-typed numbers are stored
in source units to round-trip without precision loss, and converted
on compose via Unit.h.
Not yet wired into the streamer or .ifcfed I/O — pure data model and
maths, integrated in subsequent commits.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
helmertMetersFromParameters builds the helmert transformation as a
meter-input/meter-output 4x4 directly from parsed parameters, bypassing
autoLocal2Global's normalisation step. This preserves
IfcMapConversionScaled.FactorX/Y/Z in the rotation block so the factor
applies to placement translations when the matrix is precomputed
per-model and composed with placements at upload time. For ordinary
IfcMapConversion (factor = 1) this is bit-identical to
autoLocal2Global; only diverges on rare surveyed models with non-unit
factors, where it is the only correct behaviour.
getMapUnit returns IfcCoordinateOperation.TargetCRS.MapUnit so callers
can resolve the unit-to-metres scale via Unit.h's siScaleFromNamedUnit.
autoLocal2Global is unchanged — kept as a clean port of the python
ifcopenshell.util.geolocation reference impl for one-shot
project-units-in / map-units-out callers.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Mirrors selected helpers from ifcopenshell.util.unit: SI prefix
multipliers, the conversion-based-unit table (foot/inch/etc -> SI
metres), siScaleFromNamedUnit (walks IfcConversionBasedUnit chains
down to IfcSIUnit), getUnitAssignment / getProjectUnit /
calculateUnitScale, and convert / convertUnit. Lives in
src/ifcviewer/ for now alongside Geolocation; will move out when
ifcopenshell.util is ported to C++.
Needed by upcoming Geolocation fix (e/n/h on IfcMapConversion are
in MapUnit, must be converted to metres for the meter-by-default
iterator output) and by the federation module (display-unit
conversion when the user changes the federation unit).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Mirrors ifcopenshell.util.geolocation: HelmertTransformation parameters
(IfcMapConversion / IfcMapConversionScaled / IfcRigidOperation, plus
IFC2X3 ePSet_MapConversion), get_wcs from IfcGeometricRepresentationContext,
local2global, and auto_local2global. Lives in src/ifcviewer/ for now;
will move out when ifcopenshell.util is ported to C++.
Not yet wired into the streamer. A subsequent commit fixes the
unit handling for the iterator's meter-by-default output.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The x-angle transformation for LAYER3 slabs assumed SweptArea
is always IfcArbitraryClosedProfileDef (which has OuterCurve),
but composite profiles use IfcCompositeProfileDef instead.
Apply the coord scaling to each sub-profile individually.
Generated with the assistance of an AI coding tool.
generate_section_reference_points had no handler for
MODEL_VIEW target view, causing it to silently return
None. Add MODEL_VIEW branch that clips the section line
to XY camera bounds while preserving the Z coordinate
for correct 3D placement.
Generated with the assistance of an AI coding tool.
Guard the int() cast on CardinalPoint in
BIM_OT_edit_assigned_material so a None value (no cardinal
point set) no longer raises a TypeError.
Generated with the assistance of an AI coding tool.
The previous projection-toggle commit short-circuited contribution
culling when projection_ortho_ was set — the formula
r_px = focal_px * r / dist looks like it depends on per-instance
distance, which doesn't apply in ortho. Result: every frustum-
visible object drew, including sub-pixel ones, and FPS tanked on
top-down plan views.
In ortho the projected pixel size of a bounding sphere is constant:
r_px = pixels_per_world * r, where pixels_per_world equals the
existing focal_px / camera_distance_ (the ortho box was sized to
match perspective at the pivot's distance). So the same formula
gives the right answer if we replace per-instance dist with
camera_distance_.
cullModelCpu now does that substitution for both contributionPasses
and pixelRadius (the latter feeds LOD1 selection too — sub-pixel
objects pick LOD1 in ortho the same way they do in perspective).
The "camera inside AABB" early-return is kept; it only fires in
perspective where dist→0 would otherwise blow up r_px, and is
harmless in ortho.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds a per-frame depth-laplacian pass that darkens pixels at sharp
depth discontinuities — silhouettes, overlapping-surface boundaries,
section-cut edges. Catches the wall-against-wall and slab-against-
ceiling cases that the cavity hint in the lighting shader misses.
Implementation:
- New edge_depth_fbo_ / edge_depth_tex_ — single-sample D24S8 the
size of the window. After the main draw, blit the default FB
depth into it (handles MSAA resolve in the same call).
- Fullscreen triangle generated from gl_VertexID, samples four
cardinal neighbours, computes |4c - n - s - e - w| on linearized
depth. Linearization branches between perspective and ortho via
u_is_ortho. Threshold scales with depth so distant edges still
register.
- Output is multiplicatively blended (GL_DST_COLOR, GL_ZERO) so
colours just darken; no separate composite step.
- Runs before the pivot/section/axis gizmos so they aren't outlined
themselves. HiZ pyramid build still runs after, unchanged.
Per-frame cost is one MSAA depth blit + one fullscreen pass with
five depth samples. Sub-millisecond at 1080p on a mid GPU.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Replaces the flat 0.25 ambient + single-Lambert key with three cheap
shape-readability tricks, all in the fragment shader:
- Hemisphere ambient (sky/ground tint mixed by n.z) so floors,
ceilings, and walls get visibly different ambient colour even when
shadowed. +Z is world-up.
- Secondary fill light at 35% intensity from roughly the opposite
horizontal direction so backs of objects are not pitch black.
- Cavity hint: clamp(length(fwidth(n)) * 1.5, 0, 0.35) darkens
fragments where adjacent normals diverge sharply. Catches
wall-floor seams, column-slab joints, and stair edges as faint
dark lines without any post-process.
Total cost: ~8 extra ALU ops per fragment, no extra passes, no extra
buffers. No change to cull/HiZ/MDI.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- P toggles ortho/perspective. The ortho box is sized so the
visible rectangle at the pivot's distance matches what the
perspective camera would show — toggling at any zoom keeps the
framing identical, and the wheel keeps working by rescaling the
box. Contribution culling is disabled in ortho since its
r_px = focal_px * r / dist formula assumes perspective; frustum
and HiZ culling still run.
- X / Y / Z snap the camera to look from +X / +Y / +Z; Shift+X /
Y / Z snap to the negative side. Yaw and pitch are set
directly so top/bottom land on exactly ±90°.
- updateCamera() picks the lookAt up vector dynamically: world +Z
except within 1° of the pole, where it switches to world +Y.
That keeps lookAt well-conditioned at the poles and gives top
views the architectural "Y as north" screen orientation.
- Pan now derives screen-right / screen-up from the real camera
basis instead of from yaw/pitch alone — the old derivation
assumed up = world +Z and silently inverted at top/bottom.
- Standard views preserve target and distance — rotate only.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Convenient escape hatch when the user has stacked several cuts
and wants to start over without exiting the tool first. Also
resets the selection and drag state.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Wires up the user-facing section-cut tool on top of the clipping
plumbing landed in the previous commit.
- K toggles the tool.
- LMB while the tool is active:
* On an existing plane's arrow gizmo (screen-space line-segment
hit test, 12 px grab radius) → select + start drag.
* Otherwise on geometry → pickSurfaceAt + addSectionPlaneAt-
Surface, select the new plane.
* Otherwise → deselect.
- LMB drag updates the plane's origin by projecting the cursor
delta onto the screen-space normal axis and converting back to
metres. d is rederived from the new origin each frame.
- Delete removes the selected plane; Esc exits the tool.
- Each plane renders a 2x2 m quad outline plus a yellow arrow
along +n at its origin. Selected plane draws cyan and
thicker.
LMB object-pick is suppressed while the tool is active so plane
creation does not also change selection.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds a clip-plane pipeline used by the upcoming section tool:
- Up to 8 SectionPlane{n, d} entries, AND-combined as
fragment-shader discard against world position. Main and pick
fragment shaders both honour the planes, so cut areas are
neither drawn nor selectable.
- Main vertex shader now passes v_world_pos through.
- Pick FBO grows two attachments (RGB32F world position, RGB16F
world normal) and the pick shader writes both alongside the
object id. pickSurfaceAt() does a single readback of all
three. Existing pickObjectAt() still works unchanged for
callers that just want the id.
- addSectionPlaneAtSurface(point, normal) auto-flips the normal
toward the camera so the first click immediately cuts the
camera-facing half.
No UI yet — that's the next commit (gizmo, drag, K shortcut).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
F (no modifier) re-aims the orbit camera at the selected object's
world AABB centroid and dollies camera_distance_ so the bounding
sphere fits the current viewport. Home does the same for the union
of all finalized models. Both preserve yaw/pitch so the user keeps
their orientation; both no-op in FPS mode.
Scene AABB prefers the per-model BVH root when available and falls
back to walking InstanceCpu world AABBs. Object AABB unions every
matching instance. Distance accounts for portrait windows by using
the tighter of the horizontal and vertical FOV constraints.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
A small RGB axis cross is rendered at camera_target_ while the user is
orbiting, panning, or has just zoomed. Visibility toggles on
middle-mouse press/release; the wheel arms a single-shot QTimer that
hides it 750 ms after the last notch.
Drawn in two passes: GL_GREATER at 30% alpha for the occluded portion
(X-ray cue) and GL_LEQUAL at full alpha for the visible portion. Arm
length is computed from camera_distance_, fovy, and viewport height so
the cross stays ~30 px on screen across zoom levels.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Port get_prioritised_contexts from ifcopenshell.util.representation to
C++ and have GeometryStreamer iterate one context at a time, mirroring
bonsai's create_generic_element loop. Each pass sets context-ids to a
single context id; elements that yield geometry are dropped from the
include set so lower-priority contexts only pick up leftovers.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Match bonsai's process_element_filter for the no-filter branch:
IfcSpatialStructureElement on IFC2X3, IfcSpatialElement otherwise.
They flow through the same net/gross split as IfcElement.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Mirror bonsai's IfcImporter.process_element_filter so the streamer
walks only IfcElement (plus IfcProxy on IFC2X3/IFC4), drops
IfcFeatureElement except IfcSurfaceFeature, and routes elements
with more openings than the configurable void limit through a
second iterator pass with disable-opening-subtractions=true.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Covers the pure-logic modules with no Qt event loop or GL context: BVH
build, LOD decimation, sidecar round-trip, instanced-geometry layout
constants, and Federation save/load + relative-path policy. Each test
binary compiles only the production source(s) under test, so the unit
tier doesn't pull Qt/OpenCASCADE/IfcGeom into the test build.
Gated behind BUILD_IFCVIEWER_TESTS=OFF; default builds remain offline.
Catch2 v3.5.4 is fetched on demand via FetchContent.
The reserved uint32_t was always written as 0 and never inspected on
read. Removing it shrinks the header from 16 to 12 bytes; the version
bump makes pre-existing sidecars fail the version check cleanly rather
than misreading by 4 bytes.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The struct holds per-model bookkeeping; Model describes its contents
rather than its container relationship.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Federation (JSON) tracks an ordered list of model sources plus an
optional home-view camera state. Sources are stored relative when
under the federation file's directory, absolute otherwise.
File menu now exposes New / Open / Save / Save As; Add Files moves
to Ctrl+Shift+O. View menu gains Set/Go to Home View. Window title
binds to dirty state via setWindowModified, and the close-window
prompt offers Save/Discard/Cancel.
Per-model transform (4x4 column-major) and visible round-trip
through load/save but are not yet applied at the viewport — the
georeferencing work uses them.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
WASD strafe, Q/E down/up, mouse-look (cursor hidden + recentered),
Shift to sprint, scrollwheel scales speed, click or Esc returns to
orbit. Exiting drops back to the same viewpoint because rotation
re-pins camera_target_ to keep camera_eye_ stationary.
Movement integrates wall-clock dt inside render() and the next frame
self-schedules via requestUpdate() while any key is held. A QTimer
would fight Qt's event loop during long swapBuffers blocks and produce
"camera pauses one frame" stalls; render-driven integration keeps
movement phase-locked to vsync and absorbs slow frames in a single
catch-up step.
IFC_FPS_HITCH_MS=<n> logs frames slower than n ms while in fly mode.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Sidecar hits skipped opening the underlying .rdb/.ifc, so ifcFile() was
null and the property panel only showed cached name/type/guid. Now, after
a sidecar hit, a background thread opens <stem>.rdb (preferred) or
<stem>.ifc and hands the file to GeometryStreamer via setIfcFile(), with
a dataSourceReady signal so the UI refreshes the current selection.
Gated behind a new AppSettings::loadDataSource toggle (default on) so
users can opt into geometry-only viewing; when off, the sidecar-hit
thread is skipped and the stream-path ifc_file_ is released after
the sidecar write completes.
Also adds *.ifcview to the Add Files dialog filter so a cache can be
opened directly without its source file present.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Two upstream commits on this branch landed without updating all their
callers, leaving `import ifcopenshell.geom` unusable:
89c66f62b "Python import fixes: import from wrapper now which
inherits from mixins" moved the `file` class out of
ifcopenshell/file.py into ifcopenshell_wrapper, but missed
geom/main.py and stream.py which still did `from ..file import file`.
b022ca7e7 "Some plug-in work" dropped the SWIG exports for
`serialise`, `tesselate`, `XmlSerializer` (and other serializers)
with a `// @todo bring back serialization` marker, but left
geom/main.py referencing them at module-load time.
Fix the `file` imports to come from ifcopenshell_wrapper, and guard
the removed-serializer references behind `hasattr`, matching the
pattern already in use for the other optional serializers (gltf, hdf5,
collada, json, ttl). Revert once upstream fixes this.
The viewer can now open a .rdb directory (as produced by
RocksDbSerializer / convert_path_to_rocksdb) anywhere it accepts an
.ifc file. The full GUI gets an "Add Database..." File menu entry
that opens a directory chooser; the streamer lets the file
constructor autodetect the format and opens the store read-only so
multiple viewers can share a database without taking the exclusive
RocksDB lock.
Parallel mapping on RocksDB-backed files still produces
non-deterministic shape counts (the race is outside the instance
cache), so force num_threads=1 for the iterator when the storage is
RocksDB. Serial RocksDB (~2.6s) and parallel SPF (~0.7s) both
produce 107 shapes on AC20-FZK-Haus; @todo in-source points at the
remaining thread-safety work.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Previously readSidecar/writeSidecar were keyed on (path, file_size) with
staleness rejected at read time. Switch to pure path-stem keying: foo.ifc
and foo.ifcdb/ both resolve to foo.ifcview, so the same cache serves either
source format. Staleness is user-managed (delete the sidecar to force a
rebuild), which also lets sidecars be copied or moved independently of the
source.
v8 header drops the source_file_size field.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Buffer viewport model mutations until the OpenGL context is initialized so loads that start before first exposure do not silently drop geometry or model state.
Generated with the assistance of an AI coding tool.
Handle streamer success, failure, and cancellation as distinct terminal states so failed or cancelled loads do not finalize as successful models. Clean up partial model/UI state in the full and minimal viewer apps when a load is cancelled or fails.
Generated with the assistance of an AI coding tool.
MainWindow and MinimalWindow each carried ~150 lines of mirrored
load-queue, sidecar-thread, streamer-wiring, and ID-rebase code. Lift
all of it into a SceneLoader QObject in the library; both apps now
consume it via signals. Sidecar writes stay on the full-app side since
they need the consumer's element metadata strings.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Turn src/ifcviewer into libIfcViewer.so holding the rendering engine +
geometry pipeline (ViewportWindow, GeometryStreamer, BvhAccel,
InstancedGeometry, SidecarCache, LodBuilder, AppSettings). Move the
existing UI shell (MainWindow, SettingsWindow, main.cpp) into
src/ifcviewer-full as the IfcViewerFull executable. Add a new
src/ifcviewer-minimal target with a MinimalWindow that hosts only the
viewport and reuses the sidecar fast-path for benchmark/debug runs.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Edge-collapse decimation (meshopt_simplify) returns BIM meshes unchanged
due to per-triangle vertex duplication and non-manifold topology. The
sloppy voxel-clustering decimator is faster, needs no shadow index
welding, and produces good results at the sub-30px LOD1 threshold.
Remove the non-sloppy branch, shadow buffer, IFC_LOD_SLOPPY and
IFC_LOD_LOCK_BORDER env vars.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace i16x2 octahedral normals with i8x2, filling the 2-byte padding
after position and saving 4 bytes per vertex. int8 gives ~1.4 deg
worst-case angular error — invisible for BIM geometry which is
overwhelmingly axis-aligned. 25% VBO reduction; sidecar files shrink
~15% overall (5.4 GB -> 4.6 GB on a 111-model test scene). Bumps
sidecar format to v7.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Use $<TARGET_FILE_DIR:IfcGeom> instead of hardcoded
${CMAKE_BINARY_DIR}/ifcgeom/$<CONFIG> for plugin runtime dirs — the
old path was wrong on non-MSVC generators where $<CONFIG> expands
empty. Add explicit add_dependencies for kernel/mapping plugins so
IfcViewer waits for them to build, and drop the redundant direct link
against ${kernel_libraries}.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Benchmarks showed negligible gain (52 vs 51 fps) — the CPU BVH path
already culls efficiently, and the GPU path still read back to CPU for
LOD/winding/HiZ. Removes ~570 lines of dead weight: compute shader,
async readback, one-frame-late consume, per-model AABB SSBOs, and
profiling counters.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add --camera tx,ty,tz,dist,yaw,pitch and --benchmark N CLI args for
reproducible performance measurement. The benchmark orbits the camera
(0.5°/frame yaw) for N frames after a 5-frame warmup, prints
avg/median/p1/p99 frame times, then exits. Press C during interactive
use to print the current camera as a --camera argument.
Fix settle recull to fire after ANY camera motion (not just when
IFC_MIN_PX_MOTION is set), ensuring HiZ artifacts from motion frames
are always cleared when the camera stops.
Document Phase 3G (motion-adaptive culling + HiZ during motion) in
README with benchmark results from 1.06M-instance scene:
- Baseline: 16.3 fps
- IFC_MIN_PX_MOTION=10: 26.5 fps (1.6x)
- IFC_HIZ_MOTION=1: 46.6 fps (2.9x)
- Both combined: 51.0 fps (3.1x)
- + GPU_CULL: 52.0 fps (3.2x, negligible gain)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
During camera motion, use a larger pixel-radius threshold (IFC_MIN_PX_MOTION)
to aggressively cull small objects, dramatically reducing sub_draws and
improving orbit fps (e.g. 29→67 fps on 1M-instance scene). When the camera
stops, automatically re-cull at the base threshold to restore full detail.
Key behaviors:
- IFC_MIN_PX_MOTION=N sets the motion threshold (0 = disabled)
- Settle recull fires on the first still frame after motion
- HiZ pyramid invalidated on settle (stale from sparse motion frame)
- GPU cull results skipped on settle (dispatched at motion threshold)
- requestUpdate() ensures the settle frame actually runs
Also adds IFC_SUBDRAW_DIAG=1 diagnostic for sub-draw composition analysis
and documents Phase 3E/3F experiment results in README.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The HiZ pipeline had two bugs causing false occlusions:
1. The scaling depth blit (glBlitFramebuffer from window-size to HiZ-size)
produced GL_INVALID_VALUE on some drivers. Replace with a fullscreen-
triangle shader that samples the resolved depth and writes gl_FragDepth.
2. The resolve texture used GL_DEPTH_COMPONENT24 but Qt's default FBO uses
D24S8 (depth+stencil). Mismatched formats cause the MSAA resolve blit
to fail. Fix by using GL_DEPTH24_STENCIL8 for the resolve texture.
Additionally, the occlusion test was too aggressive for scenes with
compressed depth ranges (entire scene in 0.99-1.0). Change from
"max over coarse mip texels" to "reject only if ALL fine-mip texels
agree the AABB is behind them", with early-out on first non-occluding
texel and a 64-sample cap.
Also fix IFC_HIZ_MOTION=0 being treated as enabled (checked env var
existence, not value).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Only clear and emit mesh buckets that received survivors in the previous
frame, converting both phases from O(total_meshes) to O(active_meshes).
Adds per-sub-phase timing (bin/clr/class/emit) to the stats line.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace the CPU BVH traversal + frustum + contribution stages with a
GPU compute path (IFC_GPU_CULL=1). A single scene-wide dispatch tests
all instances against frustum planes and screen-space contribution
threshold, compacting survivors into a flat uint32 buffer via atomicAdd.
Uses one-frame-late async readback: frame N dispatches and fences,
frame N+1 polls the fence (non-blocking) and reads the persistent-
mapped result buffer with zero GPU sync cost. CPU still handles HiZ,
LOD selection, winding bucketing, and indirect command generation from
the compact survivor list; draw path is unchanged.
On a 1M-instance / 111-model scene (GTX 1650):
GPU dispatch: 0.70 ms (frustum + contribution, brute-force)
Readback: 0.00 ms (fence already signaled, persistent map)
CPU consume: 5.7–6.7 ms (parallel emit across models)
Cull wall: 5.8–6.9 ms (vs 9.6–15.2 ms CPU-only path)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Pack compute shader compacts non-empty indirect commands into
contiguous fwd/rev ranges, eliminating ~690k empty sub-draws that
dominated command-processor overhead. GL 4.6 entrypoint loaded via
getProcAddress with ARB fallback; graceful degradation to uncompacted
MDI when unavailable.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Two-phase compute-cull dispatch when IFC_GPU_CULL=1:
Phase 1 frustum + contribution + LOD, no HiZ → survivors
Depth render survivors depth-only into half-viewport FBO
Build GPU compute max-reduce depth → R32F mip pyramid
Phase 2 same cull + HiZ test → final survivors
Color render final survivors
The compact shader's new hizOccluded() projects 8 AABB corners to
screen space, picks the mip level where the covered rect fits in ≤2×2
texels, and rejects when the AABB's near-depth exceeds the pyramid's
max depth.
New GPU resources (per-window):
hiz_gpu_fbo_ / hiz_gpu_depth_tex_ — depth-only FBO at half viewport
hiz_gpu_pyramid_tex_ — R32F mipmapped pyramid
hiz_gpu_copy_prog_ — compute: depth → pyramid L0
hiz_gpu_reduce_prog_ — compute: max-reduce L(n-1)→L(n)
hiz_gpu_depth_prog_ — vertex + trivial fragment
On a dense 18-model BIM dataset:
survivors: 140k → 65k (HiZ rejects ~50%)
triangles: 22M → 13M
gpu_cull: 0.06ms → 22.5ms (depth pre-pass CP overhead)
The depth pre-pass suffers the same empty-sub-draws CP overhead as the
color pass (690k commands, most with instanceCount=0). Once MDI
compaction lands, both passes will be fast. For now, net FPS is flat
(savings on color ≈ cost of depth pre-pass).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The compact shader now computes per-instance pixel radius and routes
survivors to LOD1 buckets when the projected sphere falls below the
LOD1 threshold (default 30 px, same as CPU path, tunable via
IFC_LOD1_PX).
Layout expanded from 2 to 4 buckets per mesh:
[0..M) fwd_lod0 [M..2M) fwd_lod1
[2M..3M) rev_lod0 [3M..4M) rev_lod1
Two MDIs per model: CCW for [0..2M), CW for [2M..4M). Per-mesh
has_lod1 flags live in a new gpu_mesh_flags_ssbo (binding 4).
Contribution cull refactored: the compact shader now computes
pixelRadius() once and uses it for both the min_pixel_radius rejection
and LOD routing, matching the CPU path's logic.
Visible-buffer worst case is 2 × total_instances (each LOD bucket
reserves the full fwd/rev capacity per mesh, since LOD selection is
dynamic).
Tri count drops ~60% on the test dataset (53M → 22M) thanks to LOD1
decimated meshes. FPS recovers from 16 to 36 despite 690k sub_draws
(4M layout). MDI compaction remains the final perf fix.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Extend the GPU-cull indirect buffer from M to 2M commands: the first M
are the forward (non-reflected, CCW) bucket, the second M are the
reverse (reflected, CW) bucket. The compact shader reads flags bit 0
from the AABB SSBO and routes each survivor to the appropriate bucket
via bucket = reflected ? mesh_id + M : mesh_id.
uploadGpuCullStaticBuffers() now precomputes exact per-mesh fwd/rev
instance counts so each bucket reserves only the slots it needs
(total visible_ssbo size unchanged — sum of fwd + rev = total).
Draw loop issues two MDIs per model under IFC_GPU_CULL: first M
commands CCW, next M commands CW.
Sub-draws doubled (172k → 345k) which further regresses FPS due to
command-processor overhead from zero-instance sub-draws — the same
issue noted in 3a. MDI compaction remains the fix.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Promote the compute cull from a validation shader to the actual draw
driver. With the gate on, the CPU cull fan-out is skipped and MDI
consumes gpu_indirect_buffer / gpu_visible_ssbo directly.
- uploadGpuCullStaticBuffers() pre-fills per-mesh DrawElementsIndirect
commands and a mesh_base prefix sum so the compact shader can scatter
survivors into a fixed per-mesh range. Instance count for each
command is zeroed by a tiny reset dispatch, then the compact shader
atomically writes survivors and increments instanceCount.
- Draw loop branches on the gate: single CCW MDI with all mesh
commands. Fwd/rev winding split, LOD selection, and HiZ are still
CPU-path-only; reflected instances render with wrong winding under
this gate (step 3b).
- Once-per-second readback of each model's indirect buffer populates
the survivor / visible-object / visible-triangle stats so the
[frame] line reflects what the GPU actually drew.
Known regression: sub_draws is the full mesh count per model (~172k on
the test dataset) vs the handful of non-empty commands the CPU path
produces. Command-processor overhead from zero-instance sub-draws is
what drives the FPS drop, not the cull itself (0.05 ms). Compacting
non-empty commands requires glMultiDrawElementsIndirectCount, a GL 4.6
entrypoint not exposed by Qt's QOpenGLFunctions_4_5_Core; deferring to
3a-followup so we don't bolt a getProcAddress loader into the renderer
mid-restructure.
IFC_GPU_CULL is off by default, so this does not affect normal runs.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
First Phase 3E milestone: a compute shader that reads the per-instance
world-AABB SSBO added in the last commit, tests each instance against
the 6 frustum planes, and atomicAdds a global counter. No visible list
or indirect-buffer writes yet — the output is just a survivor count,
cross-checked each frame against the CPU cull's numbers in the stats
line (`gpu_cull[Xms in=A surv=B]`) so we can verify the plumbing end-
to-end before we hand the GPU responsibility for the actual render data.
Dispatched from render() after the CPU cull completes, only when
IFC_GPU_CULL=1 and the camera moved (the skipped-cull still-frame path
doesn't re-check either). The readback is synchronous — that's fine
for a validation path; it'll go away once the GPU writes indirect
commands directly.
Expected invariant: gpu_cull.surv >= cpu_cull.visible_objects, since
the GPU path does frustum-only and CPU adds contribution + HiZ cuts on
top. A large mismatch (orders of magnitude, or surv < visible) means
the SSBO upload or shader logic is wrong.
No shader/buffer bindings overlap with the draw path (compute uses
bindings 0/1, restored before drawing; draw programs rebind 0/1/2).
Scaffolding for Phase 3E (GPU compute cull). After finalizeModel /
applyCachedModel, pack each InstanceCpu's world AABB + mesh_id +
reflection bit into a std430-friendly 32 B record and push it to a
per-model aabb_ssbo. No consumer yet — the CPU cull still drives
rendering — but the next commits will point a compute shader at this
buffer and have it produce the visible list + indirect commands
directly on the GPU.
Cost: 32 B per instance, ~18 MB for the 569 k-instance test scene.
One-shot upload at finalize time; streaming-time appends aren't
mirrored (the CPU cull doesn't need the SSBO, and finalizeModel
rebuilds the whole thing in one go).
The 'Known caveats' bullet still described the old 1-frame-stale
behavior. Since 6b496d802 the cull compares hiz_vp_ to the current VP
and drops HiZ rejection whenever they differ, so HiZ only helps on
still frames — orbiting gets no benefit. Call out the tradeoff and
the planned same-frame-depth-pre-pass fix slated for Phase 3E.
Two stability bugs:
1. Clicking an object left the scene with wrong shading until the camera
moved. The pick pass re-culls every model with its own parameters
(min_pixel_radius=0, no HiZ) and overwrites each model's visible_ssbo
and indirect buffer. The next render() saw an unchanged camera,
skipped the cull via the have_cached_cull_ shortcut, and drew the
stale pick-pass buffers. Fix: invalidate have_cached_cull_ at the
end of pickObjectAt().
2. Loading two sidecar-cached models made the second model's picked
properties resolve to the first model's elements. Sidecars store raw
object_id / model_id values from the session that wrote them, and
both files start at object_id=1, so element_map_ entries collided.
Fix: on load, rebase every PackedElementInfo and InstanceCpu by
(next_object_id_ - min_id_in_sidecar) and overwrite model_id with
the freshly-assigned handle before the elements hit element_map_.
Also document both in the README — the pick-pass note under 3A
contribution culling, the sidecar rebase under the sidecar format
section.
HiZ from last frame encodes depth from last frame's viewpoint. When
the camera moves, projecting a current-frame AABB through the stored
VP answers 'was this occluded last frame?' rather than 'is it occluded
now?' — a self-reinforcing feedback loop where objects culled in
prior frames never appear in any depth buffer and stay permanently
hidden at certain camera angles.
Fix: require hiz_vp_ == current VP for the HiZ test to apply. HiZ
still helps static views (kicks in one frame after camera stops) but
no longer produces false occlusions during orbit. The correct fix for
orbit coverage is a depth pre-pass feeding fresh HiZ — planned as
part of Phase 3E GPU compute cull.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add the parallel cull bullet to the feature list, a Phase 3D section
explaining the fan-out / scratch-ownership design + measured 4x
speedup, and renumber the planned GPU compute cull to Phase 3E so it
can cite 3D as the CPU algorithm being ported.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Split cullAndUploadVisible into cullModelCpu (CPU-only, thread-safe) and
uploadCullResults (GL-only, main thread). render() fans the per-model
culls out via std::async and joins before the serial upload pass.
The cull scratch (vis_fwd/rev_lod0/1, visible_flat, indirect_scratch)
moved onto ModelGpuData so each worker owns its output buffers. Phase
timers and hiz_reject_count_ are atomic since workers fetch_add into
them. A new wall-clock timer around the dispatch block reports the
actual frame-time contribution; the existing clr/trv/emt counters are
now documented as per-thread sums.
Measured on the 18-model / 569k-instance test scene: wall-clock cull
dropped from ~25 ms to ~5 ms while the aggregate CPU work (trv) stayed
~30 ms. Frame time 34 ms -> 19 ms. IFC_CULL_THREADS=0 forces the
single-threaded fallback.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add the event-driven rendering bullet (zero idle cost, in-render frame
timing) and roadmap entries for VBO quantization and event-driven
rendering.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Position now u16x3 normalized against each mesh's local AABB; normal
oct-encoded to i16x2; RGBA8 colour unchanged. Per-mesh dequant basis
lives in a new MeshGpu SSBO at binding 2; both main and pick shaders
mix() against it before applying the instance transform.
Drops VBO and sidecar size by ~43 % (28 -> 16 B/vert), which matters
mostly for warm-load downloads of precomputed sidecars and steady-state
VRAM. LodBuilder dequantizes positions into a scratch buffer before
calling meshopt, since meshoptimizer needs float positions.
Also fixes a streaming-time crash in cullAndUploadVisible: bvh_items
was only populated at finalize, but the linear fallback indexes it
during streaming. Mirror BvhItem appends in uploadInstanceChunk so the
hot path stays valid before the BVH is built.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replaced the 16ms QTimer with QEvent::UpdateRequest delivered via
requestUpdate(), posted from every state mutator (mouse/wheel, model
lifecycle, selection, visibility, resize). A static BIM scene — the
common case for a viewer — now does no work at all between user actions.
FPS is now measured as time spent inside render() rather than wall-clock
gap between frames, so idle gaps don't pollute the 1-second window and
the headline number reflects real render throughput. Headline fps still
caps at vsync; sub-vsync profiling lives in the cull[...] phase timers.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
render() was re-running the full cull every 16 ms timer tick even when
nothing had changed — the camera matrices, scene state, and therefore
visible set were all identical to the previous frame's. The GPU was
still happy to redraw from the cached indirect buffer, but the CPU was
burning 21 ms/frame rebuilding the same visible list.
Detect the no-op case by comparing view/proj against last_cull_view_ /
last_cull_proj_ and checking a scene-dirty flag (have_cached_cull_)
that every mutator on models_gpu_ invalidates — finalizeModel,
applyCachedModel, applyLodExtension, hide/show/remove/reset, and
uploadInstanceChunk. When the check passes we skip both
cullAndUploadVisible and buildHizPyramid (the depth buffer is
bit-identical, so re-reading it produces the same pyramid).
Per-model visible_objects / visible_triangles stats now live on
ModelGpuData so the stats line reports correct numbers on skipped
frames instead of reading from a stale indirect_scratch_.
Measured on a 569k-object overview: still frames go 22 fps → 62 fps;
orbiting goes 23 fps → ~30-50 fps depending on how hard you move the
mouse (the cull only pays its full cost on the ~25 % of frames where
the camera actually moved). The stats line gains a "skipped N/M"
field so you can see the ratio live.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
cullAndUploadVisible was reading each instance's AABB through
m.instances[idx] — a 104-byte InstanceCpu struct — for the frustum /
contribution / HiZ tests. Only 24 of those bytes (the two float[3]
AABBs) are actually used by the tests; the rest (4×4 transform +
header) is pure cache-line waste, and with 569k instances the array
is 59 MB, well past any cache.
bvh_items[idx] already stores a 1:1 compact 28-byte record with the
same AABB, built unconditionally in buildBvhForModel(). Switch the
hot test path to read from it, and only touch InstanceCpu once an
instance has passed all three tests (for mesh_id). Modest ~20 %
drop in cull-traverse time on a 569k-object overview (26 ms → 21 ms).
Also add four cull-phase timers (clr / trv / emt / upl) to the
per-second stats line so future optimisation work has concrete
numbers to chase. Confirmed via these timers that bucket clears,
emit and GPU upload are all <1 ms combined; traversal is where the
remaining CPU cost lives.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
After the main draw, blit the MSAA default-framebuffer depth to a
single-sample 256×128 depth texture, read it back, and build a CPU
max-reduced mip pyramid. Next frame's cullAndUploadVisible projects
each BVH node / instance AABB through the previous frame's VP and
compares the AABB's nearest depth against the pyramid's deepest value
at the matching mip level; strictly-beyond AABBs are rejected.
Conservative direction (aabb_near > hiz_max) — never wrongly rejects a
visible instance, so no flicker. BVH subtree-level test lets a single
8-corner projection reject up to a leaf's worth of instances.
Tuning knobs: IFC_NO_HIZ=1 disables; IFC_HIZ_SIZE overrides base width.
New stats counter hiz_rej shows rejects/frame.
Measured: big win on interior views (GPU-bound), roughly zero net
effect on exterior overviews (CPU-bound on cull traversal, so the
saved GPU work is masked). Tried a 3-deep PBO ring for async readback
and reverted — the extra frame of staleness produced visible flicker
on fast orbit, and the synchronous readback wasn't actually a measured
bottleneck at 256×128.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Decimate each unique mesh once at sidecar-build time and swap to the
reduced index slice per-instance per-frame when projected sphere radius
drops below IFC_LOD1_PX (default 30). Same VBO, same SSBO, just a
different firstIndex/count in the indirect command.
Extends MeshInfo (48→56 B) with lod1_ebo_byte_offset + lod1_index_count
and bumps the sidecar to v5. buildLods() runs inside
onStreamingFinished, appends decimated indices to sd.indices,
applyLodExtension pushes the EBO suffix to the live GPU state, and the
sidecar is written with LOD1 baked in.
simplifySloppy (voxel clustering) is used instead of the default
edge-collapse meshopt_simplify because BIM brep output is per-triangle-
unwelded and non-manifold after welding — simplify returned the input
unchanged for every mesh tested. Sloppy ignores topology. Knobs
(IFC_LOD_SLOPPY, IFC_LOD_ERROR, IFC_LOD_RATIO, IFC_LOD_MIN_SAVINGS,
IFC_LOD_LOCK_BORDER, IFC_LOD_DEBUG) are available for A/B tuning.
Result on the 128M-tri 10-model test scene (GTX 1650, 2px contribution
cull): 20.2 → 43.2 fps, 40M → 14M visible triangles, no change in
object count. LOD build adds 100–600 ms per model on first open,
cached thereafter.
README Phase 3B section is now a full writeup of pipeline, selection,
decimator-choice rationale, env vars, and measured numbers.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Reject frustum-visible objects whose bounding sphere projects below a
pixel-radius threshold. Applied at both BVH-node level (whole subtrees
pruned) and per-instance level; short-circuits when the camera is
inside the AABB so nothing-you're-standing-next-to is ever lost.
Pick pass passes threshold 0 so sub-pixel objects stay clickable.
Threshold defaults to 2 px (radius), overridable via IFC_MIN_PX env
var. Measured on the 128 M-tri test scene (GTX 1650):
0 px (off): 6.7 fps, 128 M tris
2 px: 20.2 fps, 40 M tris (31%)
4 px: 30.3 fps, 15 M tris (12%)
The metric is sphere-based (cheap: one sqrt per test) rather than
AABB-corner projection; loses a little precision on very elongated
bounds but costs ~5x less per test and the BVH-node pre-cull means
the long-tail-of-small-things case is already handled by subtree
pruning before we touch individual instances.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Earlier probes pointed at per-frame glNamedBufferSubData uploads as the
bottleneck (60 fps when those two calls were commented out). That was a
false reading — zeroing the uploads also emptied the indirect buffer, so
MDI drew nothing. "No upload" and "no draw" were indistinguishable.
Two new diagnostic env vars in render() isolate the real costs:
IFC_SKIP_MDI=1 keep cull + upload + binds, skip only the MDI
draws. Gives 62 fps with everything else running,
confirming the non-draw path fits in ~16 ms.
IFC_MAX_SUBDRAWS=N cap each MDI's drawcount. 67k -> 30k sub-draws
saves 0 ms, confirming sub-draw count itself is
not the bottleneck; the long tail of sub-draws
carries ~no triangles.
On a GTX 1650 with 128 M triangles in view, nvidia-smi sits at 95 %
GPU util and FPS scales with triangle work, not sub-draw count. The
card is simply rasterising at ~850 M tri/s. No CPU-side or upload
trick recovers it.
Revised Phase 3 is therefore shedding triangles, not bytes:
3A screen-space contribution culling (next)
3B LOD
3C HiZ occlusion
3D GPU-side compute culling
README Phase 3 section rewritten around the diagnosis, including the
false lead, so future work doesn't re-tread the upload path. The
aborted staging+resident ring-buffer implementation was reverted (the
uncommitted working tree is gone — pure glNamedBufferSubData retained
for the visible + indirect buffers, which we now know is fine).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The previous README described a pre-instancing world (32-byte world-
coord vertices with per-vertex object_id, ObjectDrawInfo structs, EBO
reordering after BVH build, and a Phase 3 plan built around moving
draw submission to the GPU). Most of that is either gone or already
solved:
- Vertices are now 28 B local-coord; per-instance transforms live
in an SSBO read through a visible-index SSBO and gl_BaseInstanceARB.
- ObjectDrawInfo is replaced by MeshInfo + InstanceCpu + InstanceGpu.
- No EBO reorder on BVH build — the BVH is over instance AABBs and
the mesh/EBO layout is orthogonal.
- Draw-call submission is already one glMultiDrawElementsIndirect
per model; the old Phase 3 goal is met.
New content worth keeping:
- GPU instancing section documents the mesh/instance/visible/indirect
buffer contract the whole renderer hangs off of.
- Reflection-aware two-pass draw is documented (det<0 placements,
forward/reverse slice split, glFrontFace toggle).
- reorient-shells and backface culling are called out as correctness
+ perf levers with their tradeoffs.
- Phase 3 is rewritten around the actual bottleneck surfaced by
profiling: per-frame glNamedBufferSubData stalls on the visible
and indirect buffers. Includes the diagnostic methodology (empty-
screen jump to 60 fps, window/MSAA invariance, upload-comment-out
experiment) so future-me remembers why this is the next step.
- 3A (persistent mapped ring buffers, near-term) and 3B (GPU-side
compute cull, longer-term) split out with scope estimates.
- Roadmap updated: instancing / MDI / reflections / reorient-shells
/ backface cull all ticked; 3A surfaced as the next open item.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Enables GL_CULL_FACE by default (user-toggleable in Settings) so
closed solids skip shading their back halves. The catch is that
IFC placements can contain reflections (mat4 with det<0 — mirrored
families, symmetric instances). Naively culling would make every
mirrored instance vanish because the rasterizer sees its screen-space
winding as backwards.
Fix: detect reflections at upload time via determinant sign, bucket
visible instances into forward (det>=0) and reverse (det<0) per mesh
during culling, and issue two glMultiDrawElementsIndirect calls per
model with glFrontFace toggled CCW/CW between them. The indirect
buffer is still one buffer — just split into a forward slice followed
by a reverse slice, with m.indirect_forward_count recording the split.
Vertex shader flips the normal when the transform has negative
determinant, keeping lighting correct on mirrored instances. The
fragment shader keeps the gl_FrontFacing fallback as a safety net
when culling is disabled (e.g. for files with open shells).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
IFC files routinely have IfcConnectedFaceSets whose faces point
inconsistently within the same shell — the result under per-vertex
normals is dark inside-out patches, and under GL_CULL_FACE it's
swiss-cheese. reorient-shells fixes the face winding at geometry
generation time, which is the only place it can be fixed correctly;
no shader trick can recover from a mesh whose triangles disagree
among themselves.
Off by default in IfcOpenShell because it adds iterator time, but
we cache the result in the sidecar so it's a one-shot cost per file.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Two bugs conflated as "weird colors":
1. Two-sided lighting. IFC placements often embed reflection
matrices (mirrored families). Transforming a_normal by
mat3(inst.transform) produces a normal pointing the wrong way
on those instances, and max(n·L, 0) then clamps the surface to
pure ambient — reads as dark / washed out. Use gl_FrontFacing
to flip n in the fragment shader so both winding orientations
shade correctly. The proper fix (ship an inverse-transpose
normal matrix or a det-sign bit per instance) is still owed;
that would unlock re-enabling GL_CULL_FACE for a big fragment-
work win on closed solids.
2. Stats label "inst_draws" was counting indirect sub-draws, not
actual GL draw calls — misleading since MDI collapses N sub-
draws into one glMultiDrawElementsIndirect. Split into
gl_draw_calls (real GL calls, = drawn-model count) and
indirect_sub_draws (packed sub-commands). For a BIM model
with 47k unique meshes at full view this now correctly reads
"1 gl_draws (47092 sub)" rather than suggesting 47k driver
dispatches.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Each visible model now issues a single glMultiDrawElementsIndirect
call instead of one glDrawElementsInstancedBaseVertex per mesh. The
CPU BVH cull populates an array of DrawElementsIndirectCommand
records plus the flat visible-instance list, uploads both, and draws
the whole model in one GL call.
Vertex shaders switch from a uniform u_instance_offset to
gl_BaseInstanceARB (ARB_shader_draw_parameters), so per-draw offset
comes from the indirect command's baseInstance field.
Draw-call counts for BIM scenes with hundreds of unique meshes drop
from hundreds-per-frame to one-per-model, cutting driver overhead.
This also sets up the plumbing for the follow-up compute-shader cull
that will populate the indirect buffer entirely on-GPU.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Pre-allocate the instance SSBO on model creation (4 MB, grow-on-demand)
and append each arriving InstanceChunk directly to the GPU-side
InstanceGpu array in uploadInstanceChunk. This makes a model drawable
as soon as its first mesh + first instance chunk land, rather than
waiting for finalizeModel.
The visible-list architecture already decouples SSBO order from the
draw path, so appending in insertion order is correct — no sorting
required. finalizeModel collapses to:
- compute per-mesh instance counts (for stats + sidecar round-trip)
- build the per-model BVH over instance world AABBs
Render / pick loops now gate on ssbo_instance_count > 0 rather than
the finalized flag. Stats include in-progress models in totals
(excluding only hidden).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Re-wires the BVH acceleration structure on top of the new instanced
renderer. Per model, build a BVH over per-instance world AABBs at
finalize (and on sidecar apply). Each frame, traverse the BVH against
the camera frustum to produce a visible-instance index list, bucket by
mesh_id, and upload to a per-model SSBO at binding=1. The main and
pick vertex shaders do a double-indirection
`instances[visible[u_offset + gl_InstanceID]]` so draws only touch
instances that passed the frustum test.
Models with fewer than BVH_MIN_OBJECTS instances skip the BVH build
and fall back to a linear per-instance frustum test.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Commit B of the instancing migration. The sidecar on-disk format is
reintroduced at version 4 with MeshInfo + InstanceCpu sections in place
of v3's flat per-object draw-info array.
After streaming finishes, MainWindow asks the viewport for a post-
finalise snapshot (VBO + EBO are read back from the GPU, meshes and
instances come from the CPU-side arrays) and writes it alongside
PackedElementInfo + the string table. On a subsequent load,
readSidecar rehydrates the whole struct and ViewportWindow::
applyCachedModel uploads VBO/EBO/SSBO in a single step, bypassing the
iterator entirely.
Staleness check is still by source file size.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Commit A of the instancing migration (Phase 3a). The streamer now runs
the iterator with use-world-coords=false and dedupes by the geometry's
representation id, emitting a MeshChunk once per unique geometry and an
InstanceChunk per placement. The viewport keeps geometry in local
coordinates (28 B/vertex, down from 32) and applies the per-instance
transform in the vertex shader via an std430 SSBO indexed by
gl_InstanceID + a per-draw uniform offset. After streaming finishes
finalizeModel() stable-sorts instances by mesh_id, assigns each mesh a
contiguous range, and uploads the SSBO; render then issues one
glDrawElementsInstancedBaseVertex per mesh.
BvhAccel is reshaped to operate on a generic BvhItem (world AABB +
model_id) so it can drive instance-level culling, but the path is not
wired in yet -- every instance is drawn every frame in this commit.
Progressive-during-streaming rendering is likewise disabled: a model
appears when its SSBO is uploaded, not incrementally. Sidecar cache
is stubbed (reads miss, writes are no-ops); the v4 on-disk format with
MeshInfo + InstanceGpu sections lands in Commit B.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When a BVH leaf passes the frustum test, emit a single glMultiDrawElements
record covering the leaf's entire index range instead of one per object.
Leaves are contiguous in the EBO after reorderEbo, so the range is just
[first_object.index_offset, sum(index_count)]. Cuts draw calls by ~8x
(BVH_MAX_LEAF_SIZE) and shifts the bottleneck from CPU/driver per-draw
overhead toward GPU vertex throughput.
Per-object features (selection highlight, per-vertex color, object_id
picking) are unchanged — they operate on vertex attributes, not draw
state. Future per-object hide/override will use SSBO lookups sampled
by object_id in the fragment shader.
Slight overdraw from skipping per-object frustum tests within a leaf is
negligible given median-split BVH tightness and spare tri throughput.
Also adds visible_objects_ counter so stats still report true object
counts (not leaf counts), plus leaf_draws/model_draws breakdown in the
per-second frame log.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Per-second frame log reports fps/ms, visible/total object & triangle
ratios, VRAM breakdown (VBO+EBO), model count, and pending uploads.
Upload-complete log includes per-model VBO/EBO MB and scene total VRAM.
Streamer runs an instancing analysis keyed on geom.id(): total shapes,
unique representations, dedup ratio, theoretical VBO/EBO/SSBO sizes if
instanced, potential savings, and top-5 most-duplicated representations.
Used to validate whether GPU instancing is worth the architectural
rewrite for a given dataset.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Phase 2 performance: BVH acceleration with median-split build, per-model
trees, and EBO re-sorting for GPU cache coherence. Raw binary .ifcview
sidecar stores full geometry + BVH for instant subsequent loads (skip
tessellation entirely).
Per-model GPU buffers (VAO/VBO/EBO per model) eliminate cross-model buffer
copies on growth. Sidecar reads happen on a background thread. Bulk GPU
uploads are progressive (48 MB/frame chunks) so the viewport stays
interactive while multi-GB models stream in.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Reflect current architecture: per-model streamers, glMultiDrawElements
with frustum culling, 32-byte vertex format with color, multiselect
file picker, settings/stats files.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Introduce ModelHandle and per-model GeometryStreamers so multiple IFC
files can be loaded simultaneously. Object IDs are globally unique
(monotonically increasing across models). File picker is now multiselect.
Each model gets a top-level tree node. Property lookup uses the correct
model's ifcopenshell::file. ViewportWindow supports hide/show/remove
per model via model_id filtering in the frustum cull pass.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Show FPS, frame time, visible/total objects, and visible/total
triangles in the status bar. Toggled via Settings > Show Performance
Stats, persisted in app settings.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Track per-object AABB and index range during upload. Each frame,
extract frustum planes from the view-projection matrix and cull
objects whose AABB is entirely outside any plane. Draw only visible
objects via glMultiDrawElements. Document the three-phase rendering
performance strategy in README.md.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Extracted from the font file like so:
python3 -c "
from fontTools.ttLib import TTFont
tt = TTFont('src/bonsai/bonsai/bim/data/fonts/OpenGost Type B TT.ttf')
for record in tt['name'].names:
if record.nameID == 13:
print(record.toUnicode())
"
The https://github.com/IfcOpenShell/website repo already has bonsai-docs.yml workflow that does the same thing - builds Bonsai docs from the main repo and deploys to bonsaibim_org_docs, so this workflow is redundant and confusing.
Don't leave a broken repo if ifcmerge is misinstalled.
Fix bug where only local branches could be merged.
Fix gitch where merge commits were not considered relevant.
Add update callbacks to the relating_object and related_object
PointerProperties so that selecting an object via the eyedropper
in BIM_PT_aggregate immediately calls aggregate_assign_object
and closes the editing panel, removing the need to click the
checkmark button manually.
Generated with the assistance of an AI coding tool.
Adds `remove_coplanar_boundary_lines()` to operator.py (Bonsai uses this
path, not draw.py's main()). After `merge_linework_and_add_metadata()`
assigns material CSS classes, this post-processes the SVG to delete
projection line segments that appear in two or more adjacent, coplanar
elements with the same material and presentation style.
Key design decisions:
- Material identity: compared via sorted IFC material ID tuples from
`get_materials()`, not CSS class names — avoids false matches between
unrelated `material-null` elements.
- Presentation style identity: compared via IFC IfcPresentationStyle IDs
from `StyledByItem` on geometry representation items — handles elements
with no material but distinct visual styles.
- Physical adjacency: confirmed by a 3D shared-vertex test (tol=0.01 m)
after a quick AABB guard, rejecting elements whose 2D projections
overlap but sit at different depths.
- Coplanarity: determined by the dominant (largest-area) face normal of
each Blender mesh object — area-weighted averages are unreliable for
slabs whose equal top/bottom faces cancel out. Folded walls sharing an
edge but meeting at an angle are correctly rejected (normal dot ≪ 1.0).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add Anthropic prompt caching (cache_control on system prompt and
tools) to reduce repeated token costs by ~90%
- Truncate large tool results in conversation history (2000 char cap)
to prevent context bloat from ifc_tree/ifc_select responses
- Add sliding window (40 messages) on conversation history, trimming
at user message boundaries to avoid breaking tool-call sequences
- Default "New IFC" button to IFC4X3 schema instead of IFC4
- Constrain ifc_new schema parameter with enum to prevent invalid
schema strings like "IFC4X3ADD2"
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Parse ifcmerge JSON output and display a per-conflict breakdown in the
panel when merge fails. Ctrl+click on the Merge button previews
conflicts without committing. Add SelectConflictEntity operator to
select and frame the conflicting object in the 3D viewport.
Generated with the assistance of an AI coding tool.
A degenerate edge (zero-length segment) caused an early `return`
of a tuple instead of continuing the loop, resulting in a
TypeError when snap.py iterated the result and tried to assign
`point["group"]` on a float.
Generated with the assistance of an AI coding tool.
Add a provider selector (OpenAI / Anthropic) to the ifcchat web UI,
allowing users to use their Anthropic API key with Claude models
instead of only OpenAI.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When a remote branch tip is checked out (resulting in detached HEAD),
the new-branch name field is now pre-filled with the local equivalent
of the remote branch name (generating a unique suffix if that name is
already taken), so the commit button is immediately usable.
See #7580
Generated with the assistance of an AI coding tool.
This version has some functional differences:
- Structured JSON error message instead of free text (on STDOUT not STDERR)
- New --prioritise-local flag to control which side wins in merge conflicts (not used by Bonsai yet)
- IfcLocalPlacement conflicts now auto-resolve instead of failing the merge (partial solution to #6885)
- Float values are normalised when comparing entities (workaround for #7696)
Include bonsai_git_branch in get_debug_info(). For dev environments
using the GitPython-based update_commit_data() path, the branch is
read from repo.active_branch.name. For built extensions, a 7777777
placeholder is replaced at build time via the Makefile, matching the
existing pattern for bonsai_commit_hash and bonsai_commit_date.
Generated with the assistance of an AI coding tool.
Move all business logic into bonsai core and tool. Performance fixes to
minimise file IO, various minor bug fixes and tests.
Generated with the assistance of an AI coding tool.
In order to fix error of the type:
| point, _ = cls.intersect_edge_region_border(
| ^^^^^^^^
| TypeError: cannot unpack non-iterable NoneType object
a tuple is expected.
Two bugs introduced in 31b571322:
- SnapObj assumed obj.data is always a Mesh; non-mesh
objects (empties, lights, etc.) have obj.data = None,
causing an AttributeError on obj.data.edges.
- view3d_utils was used but never imported.
Generated with the assistance of an AI coding tool.
Add --format ids to the ifcquery.rst format description and a new
"Scripting with ifcedit" section showing composition examples. Add
the foreach subcommand to ifcedit.rst with usage examples.
Add ifcquery, ifcedit and ifcmcp to the README contents table, the
Sphinx docs toctree and introduction utilities table. Add new .rst
pages for each package documenting subcommands, installation, usage,
and parameter types. Fix plot and render CLI examples in ifcquery
README to use -o/--out-format flags. Update ifcmcp README to use the
installed ifcmcp command rather than python3 -m ifcmcp.
Generated with the assistance of an AI coding tool.
These three packages were added to src/ but lacked the Makefile needed
by common.mk to build distribution wheels, and the GitHub Actions
workflow to publish them to PyPI.
Adds make dist / make test / make qa targets and ci-*-pypi.yaml
workflows matching the pattern used by ifcpatch, ifcclash, etc.
IfcRelConnectsPathElements has an optional ConnectionGeometry attribute for
recording the geometric cut-plane between adjacent elements, but there was
no way to set it via the API.
Generated with the assistance of an AI coding tool.
assign_product creates IfcRelAssignsToProduct linking a structural member to
a physical building element. assign_to_building creates IfcRelServicesBuildings
linking a structural analysis model to a building. add_topology_representation
creates IfcTopologyRepresentation for structural elements, inferring the
representation type from the item class.
Generated with the assistance of an AI coding tool.
Both attributes are required by the IFC schema but were not settable via
the API function. Add physical_or_virtual and internal_or_external parameters
with "NOTDEFINED" defaults for backward compatibility. Update Bonsai boundary
panel to expose both fields in the editor.
Generated with the assistance of an AI coding tool.
Add --format ids to ifcquery to output step IDs suitable for piping into
ifcedit parameters. Add ifcedit foreach to apply an operation to every
element in a query result. Extend clash and relations output so --format ids
extracts all involved element IDs, enabling one-liners like clash detection
piped directly into render.
Generated with the assistance of an AI coding tool.
Enhances the performance of mouse intersection checks for wireframe objects.
Details:
- Calculated the intersection with the mouse in 2D pixels first.
- Converted objects to a BVH Tree to reduce the number of edges checked against the mouse position.
mcp is an optional dependency so that the embedded API (embedded.py) can
be used from Pyodide without pulling in pydantic-core and the rest of the
MCP protocol stack, which may not be available in all WASM environments.
Fix#7034
bpy.ops.bim.extend_to_underside doesn't exist - the correct operator
name is bim.extend_walls_to_underside. The AttributeError killed the
entire panel draw, hiding mirror, align, aggregation, and QTO buttons.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
If a file had an IfcProjectedCRS without an IfcCoordinateOperation (or
vice versa), add_georeferencing would return early without creating the
missing entity. This caused edit_georeferencing to crash with IndexError.
Now detects the inconsistent state, cleans up, and recreates both.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
ifcmcp is a new Model Context Protocol server that wraps ifcquery and ifcedit, holding an IFC model in memory across tool calls. It is the preferred way to interact with IFC models from AI assistants and MCP-compatible clients.
Setup:
claude mcp add --transport stdio ifc -- python3 -m ifcmcp
Session tools: ifc_load, ifc_save
Query tools: ifc_summary, ifc_tree, ifc_info, ifc_select, ifc_relations, ifc_clash, ifc_validate, ifc_schedule, ifc_cost, ifc_schema, ifc_contexts, ifc_materials, ifc_plot, ifc_render, ifc_shape, ifc_shape_list, ifc_shape_docs
Edit discovery: ifc_list, ifc_docs
Edit execution: ifc_edit, ifc_quantify
The model stays in memory between calls - ifc_edit does not auto-save; call ifc_save explicitly when done.
Depends on both ifcquery and ifcedit
Generated with the assistance of an AI coding tool.
ifcedit is a new command-line tool for executing ifcopenshell.api mutations from the shell. It wraps the entire API surface — any function callable via ifcopenshell.api can be invoked without writing Python.
Subcommands:
list [module] — list all API modules, or functions within a module
docs <module.function> — full documentation (params, types, descriptions)
run <file> <module.function> [--param value ...] — execute a mutation; overwrites input file by default, or use -o <output> to write elsewhere; --dry-run validates without executing
quantify list — list available QTO rules
quantify run <file> <rule> — run quantity take-off, writing IfcElementQuantity psets back to the file
Parameter coercion: entity references can be passed as step IDs (strings); lists, dicts, booleans, and None are handled automatically.
Usage:
python3 -m ifcedit run model.ifc root.remove_product --product 42
python3 -m ifcedit docs geometry.edit_object_placement
Generated with the assistance of an AI coding tool.
ifcquery is a new command-line tool for querying and inspecting IFC models. All output is JSON.
Subcommands:
summary — schema version, entity counts, project metadata
tree — full spatial hierarchy (Project → Site → Building → Storeys → Spaces → Elements)
info <id> — deep inspection of any entity by step ID (attributes, psets, placement matrix, type, material)
select <query> — filter elements using ifcopenshell selector syntax
relations <id> — relationships for an element; --traverse up walks to IfcProject
clash <id> — geometric intersection and clearance detection
validate — schema/constraint validation; --rules adds EXPRESS checks
schedule — work schedules with nested task trees
cost — cost schedules with nested cost item trees
schema <class> — IFC class documentation from the model's schema version
plot — SVG plan drawing
render — 3D geometry rendering
contexts — geometric representation contexts
materials — material assignments
Usage:
python3 -m ifcquery <file.ifc> <subcommand> [args]
Generated with the assistance of an AI coding tool.
* Add geometry.clip_solid API
* Add geometry.clip_solid_bounded API
* Add geometry.copy_representation API
Deep-copies the named representation from a source element to a target
element.
Generated with the assistance of an AI coding tool.
* Doc clarification for api.sequence.assign_process
* Doc clarification for api.geometry.edit_object_placement
* Doc clarification for api.feature.remove_feature
* Doc clarification for api.geometry.add_wall_representation clippings normal
* regenerate_wall_representation: document BBIM_Boolean preservation requirement
Generated with the assistance of an AI coding tool.
* shape_builder: complete docstrings and return type annotations
* shape_builder: warn about mixed item types in get_representation
* shape_builder: fix half_space_solid agreement_flag docstring
Generated with the assistance of an AI coding tool.
Add tests for all four covering generation operators: flooring/ceiling
from walls and flooring/ceiling from cursor. Previously only flooring
from walls was tested.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The covering tool used bmesh as an intermediate and relied on
type.assign_type post-listeners (removed in 44a52863a) to generate
the IfcExtrudedAreaSolid body. With those listeners gone, coverings
had no body representation and assign_swept_area_outer_curve crashed.
Build covering representations from scratch using ShapeBuilder, reading
the extrusion depth from the type's IfcMaterialLayerSet. Also replace
bpy.ops.bim.assign_class with bonsai.core.root.assign_class using
should_add_representation=False, consistent with the space fix.
Refactored shared coordinate-conversion and extrusion-building logic
into get_2d_vertices_from_polygon and set_extrusion_representation_from_polygon,
used by both space and covering code paths. Removed all bmesh-dependent
dead code from the spatial tool.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Space regeneration was only updating the Blender mesh and marking the
object as edited, but the IFC representation was never synced on save.
Replace the bmesh-based approach with ShapeBuilder to write geometry
directly to IFC as an IfcExtrudedAreaSolid, then reload via
switch_representation. This applies to both new space creation and
existing space regeneration.
Also changes assign_ifcspace_class_to_obj to call
bonsai.core.root.assign_class directly with
should_add_representation=False instead of bpy.ops.bim.assign_class.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The nest assign_object API now removes existing spatial containment and
aggregate relationships before creating the nest, matching the behavior
documented in its docstring and consistent with aggregate.assign_object.
Fix#7248
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Walk up the full hierarchy via get_parent() in can_nest() and
can_aggregate() to reject assignments that would create a cycle.
Also reject self-assignment.
Fix#7248
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
After assigning an IFC class and undoing, msgbus subscriptions registered
with the old Python object wrapper survived (PERSISTENT flag) but could
not be cleared because: (1) rollback_link_element looked up objects by
their post-link name which no longer exists after undo, and (2) the
per-object clear_by_owner calls in rebuild_element_maps used new Python
wrappers that didn't match the old subscription owners.
Fix by using a dedicated stable object (object_subscription_owner) as
the msgbus owner for all per-object subscriptions, allowing
rebuild_element_maps to clear all stale subscriptions in one call
regardless of Python wrapper identity changes during undo/redo.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The `is not ...` was being captured by the walrus assignment due to
missing parentheses, causing the condition to always evaluate incorrectly.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
remove_representation_item now returns early if removing the item would
leave Items empty. edit_text_literals returns early on empty attributes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
add_boolean was removing second operands from ALL IfcShapeRepresentations
that referenced them, which could corrupt unrelated shapes and leave
representations with empty Items (bug #7803).
The API no longer modifies Items — callers manage this explicitly.
validate_type and Bonsai's AddBoolean operator now handle their own
item removal scoped to the correct representation.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
remove_deep is deprecated and can silently delete elements still in use.
remove_deep2 requires zero inverses before removal, making it safer.
Also fixes a double-removal bug in remove_grid_axis and prevents
removing the last prop template from a pset template.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Tests using custom first/last tread runs were not setting
custom_tread_lock=False, so the custom values were silently ignored
since 8f7cf76d9 introduced the lock gate in the calculation.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
SyntaxWarning: invalid escape sequence '\/' at line 312.
`\/` in a plain string is treated as `/` by accident; replaced with raw string r"..." to be explicit.
- `function_item`, `tags` added in df7318973
- MakeVolume added in c385b93, ignore as all other conversion settings
- moved `SeparateZUpNode` ignore to the other geom serializer settings
Mentioned in https://github.com/IfcOpenShell/IfcOpenShell/issues/7667#issuecomment-4076645173
Traceback:
```
Traceback (most recent call last):
File "\bonsai\bim\module\patch\prop.py", line 55, in get_ifcpatch_recipes
docs = ifcpatch.extract_docs(f, "Patcher", "__init__", ("src", "file", "logger", "args"))
File "\ifcpatch\__init__.py", line 168, in extract_docs
spec.loader.exec_module(submodule)
~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^
File "<frozen importlib._bootstrap_external>", line 1027, in exec_module
File "<frozen importlib._bootstrap>", line 488, in _call_with_frames_removed
File "\ifcpatch/recipes/FixRevit2025TINs.py", line 31, in <module>
class Patcher:
...<509 lines>...
return co / self.unit_scale
File "\ifcpatch/recipes/FixRevit2025TINs.py", line 168, in Patcher
def create_edges(self, obj: bpy.types.Object) -> None:
^^^
NameError: name 'bpy' is not defined
File "\bonsai\bim\module\patch\prop.py", line 43, in get_ifcpatch_recipes
```
To fix warnings below:
```
<python-input-1>:1: DeprecationWarning: datetime.datetime.utcnow() is deprecated and scheduled for removal in a future version. Use timezone-aware objects to represent datetimes in UTC: datetime.datetime.now(datetime.UTC).
```
Closes#7787: Previously bim.assign_selected_as_product required exactly
2 objects. With multiple annotations referencing the same
product, users had to repeat the operation once per
annotation. Now any number of IfcAnnotations can be selected
alongside a single product object and all are assigned in
one operation and one undo step.
Generated with the assistance of an AI coding tool.
* buildingSMART Data Dictionary module: added textfield to change data dictionary url
* moved change of bsdd baseurl change to addon settings
* Receiving Psets from other dictionary sources has been made available by dynamizing the identifier_url using the client baseurl
* Remove unnecessary blank lines in prop.py
* Remove unused import of bsdd module
* Fix#7712 - global alignment controls now affects all literals
* Fix#7760 - goodbye 3x3 box alignment
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Previous logic always skipped the first point. Instead, it should only
skip when actually closing a loop (i.e. >= 3 points).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Since we added more data to debug info in fcf5614 Fatal Error itself started to fail and was never displayed due some props being inaccessible during load, should be fixed now.
Possible error that were fixed:
```
File "\Blender\5.1\extensions\raw_githubusercontent_com\bonsai\__init__.py", line 328, in <module>
print(format_debug_info(get_debug_info()))
~~~~~~~~~~~~~~^^
File "\Blender\5.1\extensions\raw_githubusercontent_com\bonsai\__init__.py", line 117, in get_debug_info
if bpy.data.is_saved:
^^^^^^^^^^^^^^^^^
AttributeError: '_RestrictData' object has no attribute 'is_saved'
Traceback (most recent call last):
File "\Blender\5.1\extensions\raw_githubusercontent_com\bonsai\__init__.py", line 366, in draw
info = get_debug_info()
File "\Blender\5.1\extensions\raw_githubusercontent_com\bonsai\__init__.py", line 152, in get_debug_info
bim_props = tool.Blender.get_bim_props()
^^^^
NameError: name 'tool' is not defined. Did you mean: 'bool'?
Traceback (most recent call last):
File "\Blender\5.1\extensions\raw_githubusercontent_com\bonsai\__init__.py", line 366, in draw
info = get_debug_info()
File "\Blender\5.1\extensions\raw_githubusercontent_com\bonsai\__init__.py", line 141, in get_debug_info
import bonsai.tool as tool
File "\Blender\5.1\extensions\.local\lib\python3.13\site-packages\bonsai\__init__.py", line 355, in <module>
print(format_debug_info(get_debug_info()))
~~~~~~~~~~~~~~^^
File "\Blender\5.1\extensions\.local\lib\python3.13\site-packages\bonsai\__init__.py", line 141, in get_debug_info
import bonsai.tool as tool
File "\Blender\5.1\extensions\.local\lib\python3.13\site-packages\bonsai\tool\__init__.py", line 23, in <module>
from bonsai.tool.attribute import Attribute
File "\Blender\5.1\extensions\.local\lib\python3.13\site-packages\bonsai\tool\attribute.py", line 31, in <module>
import bonsai.bim.helper as helper
File "\Blender\5.1\extensions\.local\lib\python3.13\site-packages\bonsai\bim\__init__.py", line 28, in <module>
from . import handler, operator, prop, ui
File "\Blender\5.1\extensions\.local\lib\python3.13\site-packages\bonsai\bim\handler.py", line 36, in <module>
from bonsai.bim.module.aggregate.decorator import AggregateDecorator
File "\Blender\5.1\extensions\.local\lib\python3.13\site-packages\bonsai\bim\module\aggregate\__init__.py", line 21, in <module>
from . import operator, prop, ui
File "\Blender\5.1\extensions\.local\lib\python3.13\site-packages\bonsai\bim\module\aggregate\operator.py", line 32, in <module>
class BIM_OT_aggregate_assign_object(bpy.types.Operator, tool.Ifc.Operator):
^^^^^^^^
AttributeError: partially initialized module 'bonsai.tool' from '\Blender\5.1\extensions\.local\lib\python3.13\site-packages\bonsai\tool\__init__.py' has no attribute 'Ifc' (most likely due to a circular import)
```
Blender doesn't have it's own quick favorites manager and working with them can be not very flexible - you can add them in context menu and remove them from Quick Favorites menu. But you can't reorder them, you can't rename them and you can't even add a new button to favorites if it's not added by some addon in the UI.
Have been stumbling upon this for awhile and decided to create an experimental manager UI for this. Things it can do:
- help user create a button with any operator in Blender and properties they prefer to then save it Quick Favorites. Which seems can be very useful in Bonsai, since you can create separate buttons for all kinds of selectors expressions, class assignment or other operators.
- it can import quick favorites from user's actual current quick favorites, so they can just modify them a bit, reorder, rename and then add them again.
- Since quick favorites are not exposed to Python API in Blender, we're using a very hacky way to retrieve them from Blender and don't provide our own buttons for adding and removing quick favorites, as it may be dangerous and even more hacky in implementation. So the workflow for user is to either generate some buttons and add them to quick favorites using Manager or to import it's own quick favorites, then change them how they like, then remove quick favorites using usual quick favorites menu and then add new button one by one.
Small demo - https://files.catbox.moe/vyffp6.mp4
Refactored methods for accessing objects in linked models and added a simple operator to select object in linked model by providing guid.
A quick demo - https://files.catbox.moe/sjjw37.mp4
* Fix#7718: Fix FallDecorator label calculation for all slope annotation types
- Fix wrong dict key type in decoration.py: DecoratorData.data["fall"] is
keyed by obj.name (str) but was looked up with obj (Object), causing
object_type to always be None
- Apply obj.matrix_world transform to spline points before computing rise/run
in both decoration.py and svgwriter.py; local coordinates have Z=0 for flat
annotations, world coordinates correctly reflect elevation change
- Use hypotenuse (segment_length) instead of run as the denominator for
SLOPE_FRACTION label display
Generated with the assistance of an AI coding tool.
* docs: add BonsaiPR bleeding edge installation section
Add new section to installation.rst documenting the BonsaiPR
community build, including why it exists, how the automated
PR-merging system works, installation steps with automated
updates, manual installation, and the PR workflow for
contributors.
Generated with the assistance of an AI coding tool.
* whoops
Linking flags were missing for `MODULE` type libraries, example warning: `IfcPythonPYTHON_wrap.obj : MSIL .netmodule or module compiled with /GL found; restarting link with /LTCG; add /LTCG to the link command line to improve linker performance`
Which is useful when debugging and calling tools.ps1 directly - less thing to modify to make it work.
Also replaced `exit 0` with `return`, so it would be possible to reuse functions inside `tools.ps1`
To avoid running in a situation when some builds are using one tag and some are using another and then unstable repo script fails to find builds for some platforms.
Objects with hide_select=True could not be selected during
isolation, causing hide_view_set to incorrectly hide them.
Objects with hide_viewport=True had their H-key hide state
modified as a side effect of hide_view_clear/hide_view_set.
Both are now left unaffected by bim.activate_drawing.
Generated with the assistance of an AI coding tool.
Use EPset_Parametric.LayerSetDirection exclusively to
determine horizontal vs vertical layer rendering in type
thumbnails, rather than hardcoding IfcSlabType checks.
Also fix line drawing to use the is_horizontal flag
consistently.
Generated with the assistance of an AI coding tool.
Fix Sphinx docs: replace csv-table with list-table for formatting functions
The documentation table of formatting/query functions was not rendering
because `.. csv-table::` requires strict RFC4180 CSV escaping. The table
contains nested quotes, inch marks (e.g. `3' - 0"`), backticks, and code
examples, which cause the CSV parser in docutils to treat rows as malformed
and drop the entire directive.
Replaced the directive with `.. list-table::`, which parses reStructuredText
instead of CSV and safely supports inline code, quotes, and multi-line cells.
Also moved the examples text outside the directive block and ensured a blank
line after the table so Sphinx does not interpret following paragraphs as
table rows.
No content changes — documentation now renders correctly.
Generated with the assistance of an AI coding tool.
* Enhance AddReferenceImage operator to use file browser instead of independent popup dialogue
* Fix dimensions assertion in TestAddReferenceImage
* Remove error in return in _execute (it is not execute)
* Add IFC2X3 support to AddReferenceImage
* Adde unit="LENGTH" to the x/y properties (every length dimension everywhere in the UI is in project length units. No need to say it explicitly)
* Manually create the texture always, not just for IFC2X3
* Add poll method to AddReferenceImage operator to check for loaded IFC project
* Refactor AddReferenceImage to add representation manually following pattern in root/operator.py's bim.add_element
* Improve File explorer options between new and select from existing project Ifc Reference Images
* Refactor get_existing_reference_images to use selector for filtering image annotations
* No extra args needed after should_add_representation is False
* Doing clean=True deletes everything
* Don't manually add geometry and materials, don't call bpy.ops. Only create IFC data, then use preexisting loading functions to create geometry.
* Black formatting, also now we can start to remove this operator as it becomes obsolete
* Consolidate duplicate UV generation into Loader.load_generated_uv_map
Replace 3 identical XY-UV baking blocks (create_object IMAGE,
bm_add_image_plane, ImageScalingTool) with a single reusable
classmethod in tool.Loader.
* Fix IFC4 texture display in Solid viewport Texture mode
IFC4 IfcTextureCoordinateGenerator Mode=COORD is used, load_texture_maps
falls back to load_generated_uv_map to bake XY-UV data onto the mesh.
* Fix IFC2X3 texture display
* This looks wrong
* Remove legacy override image feature, because we now have a proper styles and texture manager
* Remove legacy override existing image element, because we now have a dedicated styles texture manager
* Remove unnecessary roundtrip to bmesh and mesh
---------
Co-authored-by: Dion Moult <dion@thinkmoult.com>
Guidelines for external contributors using AI coding tools,
covering licensing, AI disclosure requirements, PR scope,
commit style, code formatting, and testing expectations.
Generated with the assistance of an AI coding tool.
Previously, there was a dance between invoke, execute, and draw. This
can probably be resolved, but is a high-risk for undo bugs. This
simplifies the logic flow to just a traditional _invoke -> _execute.
I add a new feature test to at least make sure it does something, and
this also fixes the segfault in tool tests as it no longer requires the
launching of the file browser.
In my previous commit, I mistakenly believed that there was an API change from
snap_angle_increment to snap_angle_increment_3d
But since the feature was introduced in blender 4.2 the setting is called
snap_angle_increment_3d.
Previously it was implemented inline. This now implements it as a
tool.Blender function with tests. Also the previous tests didn't
actually run and weren't actually testing any tools despite being in a
tool tests.
IfcSectionedSurface and IfcSectionedSolidHorizontal both of CrossSectionPositions attributes which are lists of IfcAxis2PlacementLinear. The implementation of each class used its own bespoke mapping of IfcAxis2PlacementLinear, which were identical to each other and slightly different than IfcAxis2PlacementLinear. Now the two sectioned classes use the one and only mapping for IfcAxis2PlacementLinear
* Linked IFC projects enhancement (multiple links to same project file)
- Implement link management system using UUIDs as identifiers to support multiple links to the same IFC file
- Add georeferencing compatibility detection and UI display (NONE, NOT_COMPATIBLE, PARTIAL_COMPATIBLE, FULL_COMPATIBLE)
- Support for duplicate link creation with Shift+D shortcut and automatic position offset
- Add false origin and project north calculation from 3D cursor for MANUAL mode
- Only store one cache per file, regardless of the amount of links
- Prevent duplicate links based on filepath and position comparison
- Improve error handling for missing files and loading failures
- Update tests
* Remove duplicate georef UI
I try to avoid duplicate UI (especially for one that can be as
sophisticated as georef - e.g. missing is WCS) as it means double the
code, double the tests, potential user confusion. BTW the note about
vertical datum isn't quite accurate as it may be included in the CRS
definition so vertical datum is optional.
* Remove depsgraph_update_post handler for update_link_ui_on_transform as per core developer feedback
* Move get_projected_crs to geolocation module
* Refactor get_projected_crs to simplify as per core developer feedback
* Remove unused import of bonsai.tool from project module
* Use IfcDocumentInformation per linked file and IfcDocumentReference for locaiton information
* Refactor SaveBlendMetadataFile operator to remove try-except blocks and remove linked projects collections since they are recreated by bonsai
* Cleanup removing empty collection instances for linked models in metadata.blend file and call determine_georeferencing_compatibility on link reload
* Add locking mechanism for linked models and update UI to reflect lock status
* Update logic that track IFC to execute_ifc_duplicate_operator instead of having it in execute() which does not track IFC undo/redo
* Refactor link handling to use get_link_empty_handle and set_link_empty_handle methods which in turn use the standard blender-ifc integrations patters (tool.Ifc.get_object(doc_reference) and tool.Ifc.link(doc_reference, empty_handle)
* remove operator.DuplicateLink and move it to tool.Project.duplicate_link()
* Refactor link handling to use sequential identifiers (no need for STEP ID DocRef)
* Refactor IFC linking logic to handle cases without a parent IFC file loaded. Firts link flase origin becomes parent origin
* Lock should not affect selection.
This makes it consistent with grid / spatial lock, and also toggle
selectability is already implemented.
* Remove unnecessary check for loaded library as Blender seems to do this internally already
* Rename util to get_crs because in IFC4X3 you can also have geographic CRS not just projected
* Remove unnecessary call to determine_georeferencing_compatibility
This function is already always called prior to calculate_link_position
so shouldn't be called here. It's also a very expensive function: as it
currently stands, just to link a single IFC, ifcopenshell.open() is
called 3 times. This reduces it to 2.
* Store CRS as metadata for linked models, and compare metadata when indicating georeferencing compatibility
Previously, to check georeferencing compatibility, ifcopenshell.open()
was used. When linking large models, this adds considerable time and
memory usage. This instead captures the georef as standard metadata in
our .cache.json. This now reduces the ifcopenshell.open() calls back
down to only 1 as necessary (see previous commit).
* Use link index instead of link name to fetch link collection item
Link name runs into issues with name uniqueness. This is why you created
a function for "get next link ID". After this refactoring, we can no
longer worry about uniqueness and that function may be removed.
* Simplify reloadlink into just unload and reload (with cache disabled)
This function should not be responsible for editing any data.
* Remove unnecessary get_next_link_id as names no longer need uniqueness
This now frees up the name variable to track a more meaningful, human
name like IfcDocumentInformation's Name attribute.
* Rewrite get / set link_empty_handle to just use the link directly
This prevents needless logic to fetch the link and also removes issues
related to duplicate names.
* Temporarily remove logic in prop callback
Right now, pretty much all the logic is done in a prop callback. In
general logic in prop callbacks should be minimised, since it's hard to
test and easily triggered as a domino effect of another change, and may
also impact undo/redo.
* Remove code that unnecessarily removes cache
This code removes cache, which means any project unlinking an IFC auto
clears the cache for any other project which doesn't make sense, and
also breaks the ability to readd it quickly.
* Rewrite link, unlink, load, and unload IFC
There were a few issues tackled here:
- Operators that change any IFC data must use tool.Ifc.Operator and
_execute, otherwise undo/redo will break. That's one of the risks of
using prop callbacks, as it is not explicit when an IFC edit happens.
- The usage of IfcDocumentReference was not correct. The Location
should store the URL, _not_ the position. The position should be in the
Identification attribute.
- The URL was stored in IfcDocumentInformation location, which does not
work in IFC2X3. There are a few changes here to make it IFC2X3
compatible.
- Generally move logic in operators, not prop callback.
* Remove restriction around manual mode.
Users should be able to use manual mode if they want.
* Restore AUTOMATIC mode to identical behaviour to file open
This is the first step to reusing cache files agnostic of the host.
* Revert tests for a fresh start for updating tests
* Revert "test_feature - clean up .ifc.cache. files after test was executed"
This reverts commit 99ae768ddf.
* Update tests and reimplement calculations for matrix of empty handle
Previously, the empty would always be placed at the origin, unless a
"position" offset was present. This is a problem, because the "position"
is simply a local offset relative to the Blender cache! If the cache was
regenerated, the offsets would be outdated. Also, the cache appeared in
different locations depending on the false origin mode, so the offset
would mean different things to different people.
Instead, a more robust method is:
1. When you link a file, a Blender cache is generated. The Blender
origin of this cache is arbitrary! It depends on the user's false origin
mode and is purely a Blender session specific thing.
2. When you load a link, a link is _always_ loaded into the correct
location with regards to IFC global coordinates. All math is done from
the perspective of IFC.
3. If you choose to transform (move / rotate / scale!?) this link from
its correct location, that gets recorded as a 4x4 transformation matrix.
Note: I haven't implemented this properly yet.
Tests all pass, with a minor modification to the new behaviour that
false origin mode now won't affect the location it ends up in, only the
generation of the cache.
* Remove arbitrary convention around display name
Not needed anymore now that A/M/D is a detail and not significant on
actual coordinates, and also that the UUID is no longer needed.
* Simplify implementation of loading linked models when opening an IFC
* Move link matrix calculation from operator to tool for reuse
* Implement editing link location and calculation of transformation matrix
I changed my mind on the is_locked thing, since it isn't clear to the
user that locking need to be done to save changes.
* Remove old is_locked, prop update callback no longer needed (dedicated operator instead), remove old calculation code
* Simplify code related to placed_as_per_georef
* For now, simple skip for duplicate / delete
IMO duplicate / delete / move a link are very rare and explicit
operations.
* Update tests
* Remove host_model coordinate data as cache is no longer host model dependent
* Move icons outside list because there are too many
* Minor tweaks
---------
Co-authored-by: Dion Moult <dionmoult@gmail.com>
Co-authored-by: Dion Moult <dion@thinkmoult.com>
IfcOpenshell read file, and write file without changes. This round trip introduces truncation noise. It should not hurt to increase the precision to keep this clean.
Replace hardcoded 5-degree angle snapping with Blender's
snap_angle_increment setting in create_wall_from_2_points()
and create_profile_from_2_points().
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Replace hardcoded 5-degree angle snapping with Blender's
snap_angle_increment setting in handle_lock_axis() for:
- Initial angle rounding when locking axis (A key)
- Angle rounding and increments on Shift+Wheel scroll
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Replace hardcoded 5-degree angle snapping with Blender's
snap_angle_increment setting in calculate_distance_and_angle().
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This function retrieves the angle snap increment from Blender's
tool_settings.snap_angle_increment property, which was added in
Blender 4.2. This allows users to configure the angle snap value
through Blender's native UI instead of using hardcoded values.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
| [ifcconvert](https://docs.ifcopenshell.org/ifcconvert.html) | CLI app to convert IFC to many other formats | LGPL-3.0-or-later\* | [](https://docs.ifcopenshell.org/ifcconvert/installation.html) [](https://github.com/IfcOpenShell/IfcOpenShell/releases?q=ifcconvert&expanded=true)
| [ifccsv](https://docs.ifcopenshell.org/ifccsv.html) | Library and CLI app to export and import schedules from IFC | LGPL-3.0-or-later | [](https://pypi.org/project/ifccsv/) |
| [ifcedit](https://docs.ifcopenshell.org/ifcedit.html) | CLI wrapper for ifcopenshell.api IFC model mutation functions | LGPL-3.0-or-later | [](https://pypi.org/project/ifcedit/) |
| [ifcfm](https://docs.ifcopenshell.org/ifcfm.html) | Extract IFC data for FM handover requirements | LGPL-3.0-or-later | [](https://pypi.org/project/ifcfm/) |
| [ifcmax](https://docs.ifcopenshell.org/ifcmax.html) | Historic extension for IFC support in 3DS Max | LGPL-3.0-or-later\* | [](https://docs.ifcopenshell.org/ifcmax.html)
| [ifcmcp](https://docs.ifcopenshell.org/ifcmcp.html) | MCP server for querying and editing IFC building models | LGPL-3.0-or-later | [](https://pypi.org/project/ifcopenshell-mcp/) |
# Build fix: remove `boost_system` from CMake components
`Boost.System` became header-only in Boost 1.69. Boost 1.90.0 no longer ships a compiled library or CMake config for it, so `find_package(Boost REQUIRED COMPONENTS system ...)` fails.
## Fix
`cmake/CMakeLists.txt`:
```diff
- set(BOOST_COMPONENTS system program_options regex thread date_time iostreams)
The headers are still available; no linking is needed.
# Build fix: add `template` keyword for dependent template member calls
Calling a template member function through a dependent expression (e.g. `storage->has_attribute_value<T>(...)` where `storage`'s type depends on a template parameter) requires the `template` keyword to disambiguate from a less-than comparison.
## Error
```
src/ifcparse/IfcParse.cpp:1856:67: error: expected primary-expression before '>' token
1856 | if (storage->has_attribute_value<express::Base>(attr_index)) {
| ^
```
Six identical errors at lines 1856, 1865, 1896, 1905, 1934, 1943.
Applied at all six call sites in `in_memory_file_storage::read_from_stream`.
# Linker fix: missing explicit template instantiations for `InstanceStreamer`
`InstanceStreamer` is a class template with methods defined in `IfcParse.cpp`, not the header. Without explicit instantiations, the linker can't find the symbols when the SWIG wrapper loads.
Cannot use `template class InstanceStreamer<...>` because some constructors have `static_assert` guards that reject certain reader types. Instead, instantiate each member function individually per reader type, only including the constructors valid for that type.
`src/ifcparse/IfcParse.cpp` (after the last `InstanceStreamer` method definition):
SWIG's `stream_from_string` calls `InstanceStreamer<FileReader<FullBufferImpl>>(void*, int, IfcFile*)`, but the `(void*, int)` constructor previously hit a `static_assert` for `FullBufferImpl` — it only allowed `PushedSequentialImpl`.
# Runtime fix: segfault in `parse_context::push()` due to vector reallocation
`parse_context_pool` stores nodes in a `std::vector<parse_context>`. During parsing, `load()` takes a `parse_context&` parameter and calls `context.push()`, which calls `pool_->make()`. If the pool's vector reallocates (via `emplace_back`), all existing references into the vector — including the `context` reference held by the caller — become dangling. Subsequent access through the dangling reference causes a segfault.
Triggered by larger IFC files (e.g. `ISSUE_159_kleine_Wohnung_R22.ifc`, 9.5 MB) that cause enough pool growth to trigger reallocation.
## Error
```
Thread 1 received signal SIGSEGV, Segmentation fault.
0x... in IfcParse::parse_context::push()
#1 in_memory_file_storage::load(...) // context& is dangling after reallocation
`src/ifcparse/storage.h` — change the pool container from `std::vector` to `std::deque`, which does not invalidate references on `push_back`/`emplace_back`:
```diff
+#include <deque>
struct parse_context_pool {
- std::vector<parse_context> nodes_;
+ std::deque<parse_context> nodes_;
```
# Runtime fix: `express::Base` comparison operators throw on null/expired instances
`express::Base::operator<` and `operator==` called `data()`, which throws `std::runtime_error("Trying to access deleted instance reference")` when the internal `weak_ptr` is expired. A default-constructed `express::Base` (the value-type equivalent of a null pointer) always has an expired `weak_ptr`.
## Why this model triggers it
The bug requires two conditions to coincide:
1. A representation is shared by **more than one product** (via `IfcRepresentationMap` / `IfcMappedItem`).
2. At least one of those products has **no material association**, so `get_single_material_association()` returns `express::Base{}` (the null equivalent).
In `advanced_model.ifc`, Body representations like `#449` (Body/Brep) have a single `IfcRepresentationMap` (`#453`) with 13 `IfcMappedItem` usages, meaning 13 products share the geometry. Some of those products (e.g. `IfcFlowTerminal` instances) have no `IfcRelAssociatesMaterial`, so `get_single_material_association` returns `express::Base{}`.
Smaller or simpler models don't hit this because either:
- Every representation maps to only 1 product → `reuse_ok_` short-circuits at `products.size() == 1` before reaching the material check.
- Every product has a material association → no null `express::Base` is ever inserted into the set.
→ reps contains only the 143 Axis tasks created before the throw
→ all 143 Axis reps have Curve2D geometry → map(representation) returns null
→ no valid elements produced → initialize() returns false
```
In the old pointer-based code, `reuse_ok_` used `std::set<const IfcUtil::IfcBaseEntity*>` and `get_single_material_association` returned `nullptr`. Inserting `nullptr` into a `std::set<T*>` is a plain pointer comparison — no dereference, no throw. The refactoring to `std::set<express::Base>` changed the comparison from pointer comparison to `express::Base::operator<`, which unconditionally dereferences through `data()`.
## Error
```
[Error] Trying to access deleted instance reference
[Notice] Created 143 tasks for 143 products ← only Axis reps; all Body reps lost
initialize() returned: False
```
## Fix
`src/ifcparse/express.h` — use `weak_ptr::lock().get()` instead of `data()` so that expired pointers compare as `nullptr` (matching old raw-pointer semantics):
```diff
bool operator<(const Base& other) const {
- return data() < other.data();
+ auto a = data_.lock();
+ auto b = other.data_.lock();
+ return a.get() < b.get();
}
bool operator==(const Base& other) const {
- return data() == other.data();
+ auto a = data_.lock();
+ auto b = other.data_.lock();
+ return a.get() == b.get();
}
```
# Runtime fix: `entity_instance` missing `get_inverse` due to SWIG `%rename` collision
Accessing inverse attributes (e.g. `element.IsDecomposedBy`) on any entity raises `AttributeError: entity instance of type 'IFC2X3.IfcProject' has no attribute 'get_inverse'`.
## Why
`entity_instance_mixin.__getattr__` (line 106 of `entity_instance.py`) calls `self.get_inverse(name)` when it detects an inverse attribute. Since the mixin inherits into the SWIG-generated `entity_instance` class (via the `object = custom_base` hack in `IfcParseWrapper.i:936`), `self.get_inverse` must resolve to a method on the SWIG class.
However, `IfcParseWrapper.i:70` has a global rename:
This was intended for `ifcopenshell::file::get_inverse` (which takes an entity + declaration and returns instances by reference), but SWIG `%rename` is global — it also renames the `%extend express::Base` method `get_inverse(const std::string& a)` at line 551. So the Python-side `entity_instance` class exposes the method as `get_inverses_by_declaration`, not `get_inverse`.
The old code (`v0.8.0`) didn't hit this because `__getattr__` called `self.wrapped_data.get_inverse(name)` on an inner `ifcopenshell_wrapper.entity_instance` object — but in that old layout, the inner object was constructed differently and the rename didn't apply the same way (or the method had a different path). In the new mixin approach, `self`**is** the SWIG object, so the rename is directly visible.
## Fix
`src/ifcwrap/IfcParseWrapper.i` — override the global rename specifically for `express::Base::get_inverse`, restoring the original name on entity instances:
Add this line **before** the global rename (or anywhere before the `%extend express::Base` block). This scoped rename takes precedence for `express::Base`, so:
-`entity_instance.get_inverse(name)` works as the mixin expects
-`file.get_inverses_by_declaration(...)` keeps its intended name
## Python-side workaround
`entity_instance.py:106` — call the method by its SWIG-renamed name:
```diff
- vs = self.get_inverse(name)
+ vs = self.get_inverses_by_declaration(name)
```
# Runtime fix: `entity_instance` class no longer importable from `entity_instance` module
The class rename from `entity_instance` to `entity_instance_mixin` broke external code that does `from ifcopenshell.entity_instance import entity_instance`.
## Error
```
ImportError: cannot import name 'entity_instance' from 'ifcopenshell.entity_instance'
```
Triggered at import time via `ifcopenshell.util.pset` (and likely other modules).
## Fix
`src/ifcopenshell-python/ifcopenshell/entity_instance.py` — add a backwards-compatible alias at the bottom of the module:
# Reregistering test is not added to the standard test suite because during unregister
# Blender removes all Bonsai dependencies breaking dev-environment symlinks.
.PHONY:test-reregister
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
CyrilWaechter
