Blender Git Loki

Add a 'lnor spaceset clear' func too (avoids freeing the memarena).

Not crucial currently, but could be later, and for sake of consistency...

Cleanup: rename 'spaces' -> 'spaceset'

Not a good idea to make names like MLoopNorSpace/MLoopNorSpaces...

Experimental: random colors for child hulls.

Revert "Squashed commit of temp_custom_loop_normals branch."

This reverts commit 622e3f15c7d034f7d1b1f419c5ed39339eb2ade8.

Squashed commit of temp_custom_loop_normals branch.

Merge branch 'master' into experimental-build

Merge branch 'master' into temp_custom_loop_normals

Cleanup, and do not execute the modifier when 'autosmooth' is disabled.

Fix for invalid cache index -1 in some cases, like virtual parents.

The code was assuming that child 0 is always valid, but in fact has to
be treated like all others that can be disable due to preview percentage,
virtual parents, etc.

Use the particle system material for OpenGL drawing.

This is probably not correct yet, the code is terribly confusing ...

Depsgraph: Remove the block of unneeded code

Depsgraph: Fix for bones parented to bones in IK chains

Looks like indeed the common-root lookups are needed here. Restoring inline
(instead of requiring a separate loop).

Implemented caps for the convex hair bundle hulls.

The current algorithm generates a triangle strip to fill the end of
each child bundle. This is not quite ideal, because even though the
base child groups are constructed to be convex, the resulting child
paths can become concave again by clumping, roughness, hair dynamics
etc.. This can create face inversion as well as ugly shading artifacts,
so a nicer algorithm may be desirable.

Depsgraph: Knock out debug prints now that they're no longer needed

Depsgraph: Switched RootPChanMap to using good old GHash and GSet

Now the problems with having an one bone referring to the results of another
bone in the same IK chain causing lockups should be solved. At least for the
test file I've been using, it is now solved at last.

Depsgraph: WIP attempt to fix ik-same-chain-constrained lockup issues

The idea is that when determining which opnode from the constraint target should
get used, we use READY when they are part of the same IK chain and DONE otherwise.
This lets us get the best of both worlds for these two situations.

Currently this isn't working in this commit yet, as the root_map (being C++ std stuff)
ends up trying to compare the identifiers by pointer instead of content, thus
breaking whenever we actually try to use this. All the additional debug prints added
were just for figuring out this point.

Depsgraph: Removed last of bone_transforms_key() references

We may have to reinstate an equivalent of this (for wrapping the rootmap checks)
later, but the old one is now redundant with the BONE_READY node

WIP - Noting insights

Merge branch 'master' into temp_custom_loop_normals

Depsgraph: imat is needed by the solvers and need to be initialized

That used to be initialized in legacy bone_where_is and was lost at some point.