Blender Git Commits

colormanage_colorspace_get_named() can (and does) return NULL, added checks to prevent null pointer dereference if the named color profile isn't found

Artifacts on splash are resolved.

Correction to commit rev48866 -- convert_tface_mt must happen before BKE_mesh_do_versions_convert_mfaces_to_mpolys

Discovered when were looking into crystal_cube.blend from our regression files
collection. Now it should look the same as in 2.62 release. 2.63 release wouldn't
work correct for this file because of wrong mtface->material conversion after
bmesh merge.

* Fix for the elsyiun.xml theme, Reports in Info Header were not readable (grey on grey). Patch by "ejnersan" in IRC.

Fix #32742: Motion path calculation on linked armatures locks up Blender

Issue was happening when linking armature object and making proxy and was
caused by not copying visualization settings in BKE_pose_copy_data.

This lead to deadlocks in motion path drawing code.

After discussion with Campbell decided it is crucial fix since it fixes
bug appearing in really common scenario of using armatures.

fix [#32743] Freed memory access when freeing materials.

fix [#32739] Glare node does "add" instead of "lighten"

fix for cmake constructing an invalid include path.

Entering Bcon5 stage!

- Bump version to 2.64
- No version char for this cycle
- This is finally release stage!

Splash would be commited in some hours from now.

Thanks everyone!

Added example of KX_GameObject subclassing to game engine docs.

freeing node trees no longer decreases their user counts, this cause causing invalid memory access when freeing the blend file.

revert fix for [#31555] Username with special chars in Windows 7

this breaks and causes bug: [#32720], where sys.stdout becomes invalid and print() does nothing.

On investigation - python is not getting the environment variable from blender (aparently because its a DLL?) so this should be resolved rather then overwriting sys.stdout.

correct spelling error in operator property

* fix for regression test [compo_map_uv.blend]

the wrapping was resized to fit the uv map. this step was wrong and has
been deleted.

* fix for regression file [compo_map_zcombine_cubes.blend]

the alpha mix formula was wrong. updated it.

Be aware that the regression file does not take the alpha into account,
but it should. or at least one z combine should and the other not.

this fails in 2.63a.
- At Mind -

UI:
* Fix RNA name for "turbulence_strength" property.

fix for crash in own recent masking commit with 'flood fill' operator.

fix for mesh_foreachScreenEdge__mapFunc running the callback with V3D_CLIP_TEST_RV3D_CLIPPING'd verts. (used uninitialised stack memory)

correct some include dirs not being included as SYSTEM paths in cmake.

fix for error in mask drawing, was using glVertex3fv on 2d verts. (buffer overrun)