The `create_liquid_geometry` still used the older attributes
API to write the velocities.
This replaces the use of `BKE_attribute_new` with
the newer attribute API.
No functional changes expected.
Pull Request: https://projects.blender.org/blender/blender/pulls/123854
Fribidi uses Meson, which auto-detects presence of ccache and uses it
when found. This could cause incorrect compilation results when doing
a cross-compilation to different platforms on the same machine.
Pass the typical configuration environment to the Meson setup, which
takes care of specifying deployment targets and architecture as the
compiler flags, preventing ccache from using result from a different
architecture.
This is similar to the Meson setup used for Harfbuzz and Epoxy.
Pull Request: https://projects.blender.org/blender/blender/pulls/123828
It is possible that the file does not have Grease Pencil paint
yet, leading to a crash in the BKE_paint_init().
The simple fix is to swap the order of acquiring the paint
pointer and the code which ensures that this paint exists.
The issue was introduced with f06fd85d97 where the building of the keylist used was
restricted to the visible range of the dope sheet.
The optimization uses `BKE_fcurve_bezt_binarysearch_index` which can only work when
the FCurve is sorted. During transformations in the dope sheet this is not the case as the
sorting only happens when the transformation is applied.
The fix is to do a range check in the for loop instead of a binary search.
Testing the range comes with a performance impact though.
| Before optimization | broken with f06fd85d97 | this PR |
| - | - | - |
| 90ms | 6ms | 20ms |
An alternative solution would be to sort FCurves during transformations of the Dope Sheet.
This is done in the Graph Editor and with the recent speedups introduced there this could be
a viable option. However this is out of scope for this fix.
Pull Request: https://projects.blender.org/blender/blender/pulls/123824
When a buffer resource was read, and after that read from another
shader stage it would not generate the appropriate pipeline barrier.
This is fixed by keeping the last write action around and check
if the shader stage was already included in the previous barriers.
Pull Request: https://projects.blender.org/blender/blender/pulls/123845
The `Stick` drawing mode for armatures was overriding the colors,
meaning it ignored bone colors specified by the user and also didn't use
the correct color for selected vs active.
Pull Request: https://projects.blender.org/blender/blender/pulls/123544
The "verts per grid" calculation was wrong for the non-flat/indexed
vertex buffer layout used for smooth shading. For example, for a
multires subdivision level of 8, each PBVH node's vertex buffers
were 3 times larger than they needed to be.
Pull Request: https://projects.blender.org/blender/blender/pulls/123808
Combined with previous logic, this roughly doubles the performance of
sculpt-mode multires draw vertex buffer extraction in my simple test.
That measurement is for when we initially enter sculpt mode when the
VBO creation is single threaded. Otherwise in my testing the copying
was bound by memory bandwidth and the improvement was smaller.
Draw Manager visibility compute shader triggered a read-after-write
hazard. This root cause was that the incorrect access was read from
the shader interface. It was set to none, skipping the needed pipeline
barrier.
Pull Request: https://projects.blender.org/blender/blender/pulls/123839
This was due to wrong indexing into `num_triangles_per_stroke` and
`num_vertices_per_stroke`. We were using the the index of the stroke
in the drawing instead of the index into the `visible_strokes` mask.
The extensions UI was aligning all buttons then adding a separator.
Simplify the layout by removing align which removes the need for an
explicit separator.
Clarify the tooltip for the half float color depth property for EXR
images to indicate that it doesn't affected data passes and that they
will still be saved as full float.
Prevent potential problems from untrusted XML (typically themes)
traversing into data outside the intended targets.
From what I can tell it's not currently possible but changes to RNA
could allow for this which would likely go by unnoticed.
Further details in code-comments.
This reverts commit cb76781be7, and fixes
the issues from original da05bff96c commit (missing initialization
of a pointer in copy constructor in some cases, and forgot to handle
one allocated string in the move constructor).
Many thanks to @julianeisel for finding the actual issues here.
It was too easy to click on Install by mistake, while trying to click on
the "⌄" menu. This is particulary problematic since we enable add-ons by
default.
Co-authored by: Pablo Vazquez, based on (informal) report by Campbell Barton.
Caused by a6ed013baa
Tags live on the `AssetMetaData` of an `AssetRepresentation` (so
`metadata` in py),
`asset_data`(also pointing to `AssetMetaData`) is a member of
`FileSelectEntry` only [which the above commit tried to get rid off...]
Pull Request: https://projects.blender.org/blender/blender/pulls/123818
This one was a bit more involved than the previous ones, since the
mismatch was intentional here, and happened on a non-trivial type.
It was done because the new object (managed by the `unique_ptr`) steals
the internal data of the original object. Calling `MEM_delete` (and
therefore the destructor of the `AssetMetaData` object) would then lead
to access-after-free and double-freeing errors.
This is addressed by adding two new 'copy' and 'move' constructors to this type.
The copy one ensures that deep-copy of internal data happens as expected, and
allows to simplify greatly the code in `BKE_asset_metadata_copy`, which
becomes a mere wrapper around it.
The move one allows `make_unique` to properly steal (and clear) the internal
data of the source object, which can then safely be deleted.
Pull Request: https://projects.blender.org/blender/blender/pulls/123693
Add utility class to check extension visibility to remove
incomplete logic that was duplicated into operator code.
Also minor refactoring to reduce the number of arguments passed
to internal function.s
