This reverts commit 6207708607.
It broke a GDScript test (which didn't exist back when the PR was made,
so was missed prior to the merge).
It choked on:
```
prints("a", test_instance.a, test_instance.a == Named.VALUE_A)
```
With:
```
Invalid operands "VALUE_A (enum value)" and "int" for "==" operator.
```
Stop include Bullet headers using `-isystem` for GCC/Clang as it misleads
SCons into not properly rebuilding all files when headers change.
This means we also need to make sure Bullet builds without warning, and
current version fares fairly well, there were just a couple to fix (patch
included).
Increase minimum version for distro packages to 2.90 (this was never released
as the "next" version after 2.89 was 3.05... but that covers it too).
Fixes#43868.
(cherry picked from commit b7901c773c)
Fixes some issues found by UBSAN and other misc things:
* Fixed memory leak on exit.
* Properly align ray packet buffer to 64 bytes.
* Added some compiler flags from Embree's build system.
* Fixed ray masks.
Create GLTFSkeleton at the same time we create GLTFNode objects.
Create GLTFSkin at the same time we export MeshInstance3D
Fixes export of blend shape arrays for meshes with multiple surfaces.
Fixes array indexing issues in export of glTF morph target animations.
Converts BoneAttachment3D nodes during normal node creation: this avoids
special cases during mesh export, and especially exporting skeletons or meshes
which are children of BoneAttachment3D.
Co-authored-by: K. S. Ernest (iFire) Lee <ernest.lee@chibifire.com>
The GDVIRTUAL_NATIVE_PTR did not declare the correct GDNativeConstPtr
template, resulting in "void*" being used as it's type info in both the
documentation and the extension API dump.
* Fixed LODs for shadow meshes.
* Added a merging step before simplification. This helps with tesselated
meshes that were previously left untouched. The angle difference at
wich edges ar considered "hard" can be tweaked as an import setting.
* LODs will now start with the highest decimation possible and keep
doubling (approximately) the number of triangles from there. This
makes sure that very low triangle counts are included when possible.
* Given more weight to normal preservation.
* Modified MeshOptimizer to report distance-based error instead of
including attributes in the reported metrics.
* Added attribute transference between the original mesh and the
various LODs. Right now only normals are taken into account,
but it could be expanded to other attributes in the future.
