Blender Git Loki

Merging r59529 through r59550 from trunk into soc-2013-depsgraph_mt

Merging r59513 through r59528 from trunk into soc-2013-depsgraph_mt

Fix possible crash when blender exits before task scheduler was initialized

Workaround for other object's DM build from modifier stack

Original issue is caused by dependency graph limitation,
which doesn't always give you other object's DM in such
modifiers as boolean and array. This might easily happen
when you render other scene from compositor,

Ideally, we need to make it so dependency graph is also
used in render database filling routines. This will also
solve issues with preview parameters used for boolean
operand as well.

For now let's use simpler change to make blender behave
stable at least by using CD_MASK_BAREMESH for other
object's derived final.

The thing is, handle_object_update will use baremesh mask
for derived mesh calculation and it seems baremesh is enough
for our purposes.

Mask curve BB code as TODO resolved

Nothing really could be done here to improve anything,
we'll just live with CV min/max + radius.

Remove workaround for dupligroups update

After checking bunch of tube and mango files, it ends up
that using DAG nodes instead of base objects will handle
all needed objects from dupligroups.

Merging r59407 through r59512 from trunk into soc-2013-depsgraph_mt

Merging r59402 through r59406 from trunk into soc-2013-depsgraph_mt

Code cleanup

Merging r59400 through r59401 from trunk into soc-2013-depsgraph_mt

Merging r59379 through r59399 from trunk into soc-2013-depsgraph_mt

Objects from Effector Weight group were missing in dependencies

This lead to situations when particle system modifier was using
other object's data, but because of missing dependency link that
object might have been not updated at the time it's needed.

Merging r59367 through r59378 from trunk into soc-2013-depsgraph_mt

Merging r59359 through r59366 from trunk into soc-2013-depsgraph_mt

Merging r59356 through r59358 from trunk into soc-2013-depsgraph_mt

Merging r59350 through r59355 from trunk into soc-2013-depsgraph_mt

Experiment: get rid of static in_next_object in mballs traversal

This check was used to prevent recursion caused by object_dupilist
calling MBall display list creation.

Solved by using object_duplilist_ex with update flag set to false.

From quick own tests and .blend file form original report seems
everything is fine. But maybe some unforeseen case will break
blender again.

Bad merge happened

Merging r59317 through r59349 from trunk into soc-2013-depsgraph_mt

Remove unused argument