Blender Git Loki

Removed deprecated old debug data types and per-thread debug data.

The Debug class now uses a static mutex to prevent race conditions on
the debug elements. This is inefficient of course, but for the purpose
of debugging perfectly fine.

Debugging elements for hair are stored in world space and must be drawn
accordingly.

New debug drawing system for the hair solver.

Now storing anonymous "elements" (dots, lines, etc) instead of explicit
elements of the simulation. This makes it much easier to quickly add
debug information where needed without going through all stages of data
conversion. Could add tagging to associate elements with the
various debug settings again.

Another partial fix for bending problems.

Main requirement is that the frame iterator rotates the initial frame
already for the first curve segment.

Split Viewer: 1-line fix, all viewers working now =)

Chrome improvements: shortcuts, trim select/drag, Shift-D, per-trim tessellation resolution, etc.

Add circular trim

Got RNA working and frontend for adding trims

Methods to find flow lines that are near.

Compositor handling all views on update (so you can edit the views
individually and see the updates) and viewer node fully working.

SplitViewer Node on the other hand is not working :| not sure why yet.
But apart from SplitViewer and the FileOutput node (easy fix, I'll work
on it shortly), the compositor should be fully functional for
multiview.

More WIP code for particle initialization.

Only remove solver if it exists.

Fix crash with bullet when we remove a hair system solver.

WIP.

Add stub functions to synchronize data between particle system and new
solver.

Only generate new solver if needed in solver step.

Minor fixes and little code for conversion.

Hook up new solver settings to old particle code.

New settings are now exposed in UI and saved/loaded in blend files.
However the solver is not yet hooked up.

To make this work, all references to the old clothmodifier code have
been ifdefed out. This means that the particle simulation will currently
not do anything.

Merge remote-tracking branch 'origin/master' into multiview

Set clnors code: add option to use existing custom split normals as ref if possible

In case of setting custom loop normals, add option to weight them against existing
custom loop normals if existing, instead of auto loop normals.

Allows to really edit existing clnors, instead of systematically overwritting them.

Partial fix for bending force issues: was missing vector normalization.

Merge branch 'master' into temp_custom_loop_normals

More debugging options for hair display.