I can't explain this one so I'm pretty embarrassed to commit this, but
it crashes (see situation in #5799) on my machine and I can't recreate
it outside this scenario. In any case I think as a general strategy we
should:
1. Minimise reliance on msgbus
2. Msgbus listeners themselves should be wary of invalid state (e.g. if
object is now unlinked)
3. Not clear listeners (don't overreach)
The commit that introduced the problem was trying to optimise rendering
dense meshes with the opening decorator. However, it introduced more
calculation in the initial project load and stored more data on every
object. This resulted in models loading 3x slower for me which is pretty
major. The original problem needs further investigation.
Sometimes we want to fetch selected objects, and that includes the
active object, even if the active object isn't actually highlighted in
the viewport (albiet rare, I think?).
Conversely sometimes we want to get the active object, even if it isn't
actually highlighted. The tool.Blender functions now have kwargs to
distinguish between these.
I find this a bit of a strange behaviour in IFC. Imagine you start with
a Tessellation representation type. You're only allowed meshy things, no
solid extrusion. And vice versa if you have a SweptSolid representation
type.
Now imagine you suddenly create a boolean. Now, you can mix both
tessellations and solids together.
The ZIP installer now contains a `libs/bin` folder that contains
`ifcmerge`. Bonsai on `register()` adds this folder to the system `PATH`
for the current Blender session. Advantage of this is that we don't have
to fiddle with the registry on windows, or install files to
`~/.local/bin` on Linux, it _should_ work on Darwin, updates are
automatic, and we have a mechanism to ship other executables if
required.
(Note that ifcmerge.exe increases the size of the Windows Bonsai ZIP
download by about 7MB)
If Git isn't installed on Windows, the Git panel now offers to install
it from the Windows Package Manager Community Repository using `winget`.
The annotation adding code was a bit of a mess, spread around
create_annotation, create_annotation_occurrence, and
bpy.ops.bim.add_annotation. I've now consolidated it all into
core.add_annotation and updated it to work with the new item editing
mode (basically new objects need to reload their representation to
populate item_ids). Things are still messy, but a bit less now.
Dion Moult
