People kept on getting confused with the align tool thinking it changed
the baseline to reference line offset whereas it actually just aligned
the object bodies. This tool now does exactly that.
Also start refactoring the "DumbWall" classes into the tools so we can
test them properly.
Use explicit element argument instead of props.representation_obj as representation_obj may not be there (e.g. removing rep items without item mode) or some other object might be in item mode and then booleans won't be unmarked.
Found the issue because of the failing test 🥳
1) avoid calling geometry.xxx multiple times as each time it creates a new copy
2) use numpy arrays with correct data type, so Blender could perform buffer copy significantly reducing overhead
The fix in 0899a2a8 ensured that when you switched geometry, all other
elements (sharing the same type or representation) would also switch
alongside it. However it would lead to mesh invalidation which broke
some assumptions in the style code. The style code also repeated a lot
of the logic of switch representation (i.e. finding shared
representations) so this seems simpler.
In general the whole representation part of the code is messy and
hopefully over time it'll get better.
The same issue of handling relative paths, doing as_posix, checking
drives, etc was solved again and again in slightly different ways. This
centralises it.
Previously cache_dir was located under site-packages, which is managed
by blender and could be read-only. Now location defaults to eg.
~/.cache/bonsai or equivalent location on other platforms. See #5564
Note this adds a dependency on the platformdirs python module
When loading an ifc project it will search it for bsdd ifcclassifications and will set one as active if finds it, so now it will be a bit more convenient working with bsdd classification assignment as it active bsdd previously was always getting reset when you reload ifc file.
Example - https://imgur.com/a/stlJqwm
Andrej730
