- Amsterdam, The Netherlands
-
Blender developer, mainly focused on 'Core' module.
- Joined on
2010-10-07 -
-
AFAIK Blender only reports the memory allocated by our own 'Guardedalloc' allocator (call to MEM_get_memory_in_use from ED_info_statusbar_string_ex). So any memory allocated by other…
Moving this to user interface module, as this is a UI issue.
Still not convinced that we should hide these though.
I am not aware of any fix regarding that behavior. I would guess that it got fixed 'by chance' as part of another code change. maybe @JulianEisel or @Harley would know better?
In any case, I…
Thanks for the precisions!
@Sergey afaik we try to avoid using OpenMP in Blender code itself because it can compete with TBB calls, and lead to sub-optimal threads usage? Indeed we cannot do…
I think option 2 also makes the most sense when it comes to required amount of work vs. safe & sound solution. I would rather avoid a global variable there though, I'd expect a mapping per…
@Sean-Kim note that this fix adds an extra memfile undo step actually, recording the change of active object.
AFAICT, the undo step added by https://projects.blender.org/blender/blender/src/br…
My opinion regarding 'sculpt' undo types is that they should go to the canal...
The 'unified' undo stack has never really fully worked when mixing undo steps from different types, we still have…
Fix looks correct to me (in the sense that it does restore the logic from code before commit 23cd299ba7).
Actually too late for 4.3.0 anyway... ideally would love to merge it in 4.3 branch before the end of the week though. :)
@blender-bot build
Luckily the fix is a single line, and add back the creation of an undo step that was removed in breaking commit (9e0b673467). So think it's fairly safe for 4.3.1.
Could…