Mostly some minor renaming of parameters and variables, adding a few
comments...
The only actual behavior change is a more correct handling of the
`LIB_TAG_KEEP_ON_UNDO` on ID tag. This should not have any consequences
in current code though.
So no actual change in behavior is expected from this commit.
Due a limitation in the precission of the thickness
calculation, the result could be wrong.
As the code tried to use the real thickness, in some
cases this was wrong.
Related to old fix for #103061
Pull Request: blender/blender#107835
GCC documented FILENAME_MAX is the maximum length of a file that can
be opened, not to be used for allocation as it may be INT_MAX for e.g.
Even though this wasn't the case on supported systems,
prefer FILE_MAX which is used for full paths in most places.
Regression in [0] which caused the window to be NULL when the load_post
handler was called. While this can be worked around using
context.temp_override, keep the previous behavior as this
change wasn't intentional.
[0]: 46be42f6b1
The file path was 256 bytes, passing in a buffer size of 1792
for string join & making the path absolute could also overflow
as it assumes a size of 1024.
Fix an issue when drawing node links that caused gaps, when sockets were
horizontally or vertically aligned.
When the inner and outer points of the node link's bezier control
polygon overlap, the link's tangent can be a zero vector. The shader
didn't check for this before normalizing the tangent leading to an
undefined vector which prevented the outermost segments of the link
from being drawn.
Since this only happens when the end points of the node link are
horizontally or vertically aligned, we can use the vector between the
link's end points instead of the tangent.
Fix#106929Fix#89282
Pull Request: blender/blender#107636
Making the custom data API const correct in 3a3d9488a1 caused
the issue, because it previously relied on non-threadsafe behavior. As a
workaround, just restore the thread-unsafety, and note the issue in a
new comment.
There was even one case in BLI ListBase tests!
Order was opposite of the one expected by the BLI ListBase code... Not
sure how, but this did not cause any issue apparently? But would expect
it to at least affect the order in which items in the list would be
iterated.
The order of geometry components should match between
the Separate Components node and the spreadsheet editor.
This switches the Curve and Point Cloud outputs.
Resolves#107837
Pull Request: blender/blender#107868
`BKE_library_id_can_use_filter_id` would not handle properly cases where
the ID is a liboverride - it's hierarchy root pointer can point back to
virtually any kind of ID (although typically a Collection or Object).
This was breaking liboverride resync process in some rare cases, by
failing to remap the 'hierarchy root' pointer of some liboverrides.
Generally render engines can do subframe mixing themselves, but the
purpose of subframe mixing in the simulation output node is to support
higher quality motion blur with bakes when there are topology-changing
operations after the simulation output node. Linear mixing can fill the
gaps while maintaining lower memory usage.
All point/instance domain attributes are mixed, but mixing is only
supported when the domain size is unchanged or when an `id` attribute
gives a mapping between elements. Theoretically it may be possible, but
nested instance geometry is not mixed in this commit due to the
difficulty of finding matching geometries across arbitrary instance
hierarchy changes. Attributes that are completely unchanged are ignored
using implicit sharing for better performance.
Pull Request: blender/blender#107599
This reverts commit 95a114f3ca.
This broke bl_rigging_symmetrize, it seems likely the window was not
NULL on load in this case, setting it to NULL caused the the active
object not to be recognized.
This needs further investigation, reverting for now.
This makes the Blender binary 350 KB smaller. The largest change comes
from using `FunctionRef` instead of a template when gathering indices to
mix in the extrude node (which has no performance cost). The rest of the
change comes from consolidating uses of code generation for all
attribute types. This brings us a bit further in the direction of
unifying attribute propagation.
Pull Request: blender/blender#107823
Previously, it printed the elements of `PropertyScaleType` as floats which does
not make sense. It also resulted in compile errors when attempting to compile
the generated code as c++ code.
Pull Request: blender/blender#107724
VKVertexAttributeObject was both defined as a class and a struct.
Settled on being a class.
Also removed an assert that is currently failing as workbench has
more attributes defined than actually used.
Pull Request: blender/blender#107874
After uploading the vertex buffer the state got corrupted.
Reason of this corruption is an assign-compare operation that should
just be an compare operation.
Pull Request: blender/blender#107875
Remove some nested `if` statements to improve readablilty. My main
intention has been to improve readability and therefore making it easier
to implement changes in the future.
Seperating the symmetrize operator selection logic from the temp pointer
logic may reduces performance but it seems acceptable in this case. The
change allows focussing on the selection logic without taking care of
the temp pointers at the same time.
No functional changes.
Pull Request: blender/blender#106487
Add support for index buffers as subrange. When an index buffer
is a subrange it should bind the source buffer and use the index start
as an offset in the buffer.
Pull Request: blender/blender#107871
Dummy implementation of the VKDrawList that isn't caching the draw calls
but just execute them when they are appended to the list.
Pull Request: blender/blender#107873
When the the armature doesn't have an active pose bone yet so
`ED_object_pose_constraint_list` returns NULL unhandled during
`edit_constraint_property_get`, now fixed.
Pull Request: blender/blender#107864
Regression in [0] which caused the window to be NULL when the load_post
handler was called. While this can be worked around using
context.temp_override, keep the previous behavior as this
change wasn't intentional.
[0]: 46be42f6b1
