Blender Git Loki

Transform 2D center was using short. Not good enough when it's it's way off screen. Switched to ints, that fixed the bug in the tracker.

Switching to floats would probably be safer in the long term, but too many things to test to do that now.

ccommented stuff in qhull geom.c with standard C comments not C++ ones

Float precision error could cause RKEY to start with 0.02 degree rotation.
This because the used integer+float division could result in a
9.999999e-01 value, which is for acos() to return 3.452670e-04.

Converted the division to use doubles instead.

new round of warning fixes. we are now down to 24 with Xcode on blender
alone with the following flags :
-Wall -Wno-char-subscripts -Wno-missing-braces.

the only one still worrying me is in rand.c line 57 :

rand.c:57: integer constant is too large for "long" type

but i have no clue about how correct cross-compiler and 32/64 bits friendly

see also my mail to commiter list for signed/unsigned issues

Added threadsafe patch from Martin.
Now envmaps of type "Load" should not give errors. I assume Martin tested!

Bug fix #2651

Area light soft shadow now renders transparant shadow as well.
ALso: found error in correct threadsafe use of soft shadow tables. Caused
dither/noise not to look as well as it could.

Using render option "Crop" with OpenGL render (button in 3d header)
caused a struct Part to be allocated and not freed.

Fix for #2657

Appending from files also set the string for "last loaded file". That's
a real bad one... so you can save over accidentally libraries.

Another fix: appending from files that have dynamic other files linked
with relative paths didn't work. (Yah, now it should all work!)

renderconverter change is just a more clear comment.

This is a dumb little commit to do while things are frozen.
Just updated the comments about the imbuf there were lots of
comments that were /* What is this? */

Kent

Bug fix #2660

When number of OSA samples is larger than number of AO samples, the AO
loop doesn't function well. Since rendering 16 OSA samples with only 9
AO samples isn't very useful, I've made the AO render level to become
at least equal or larger than amount of OSA samples.

Is first bug in 2.37 that worked in 2.36...

Bugfix #2652

Typo in toolbox, editmode curve, "set vector handle" should be hotkey V.

Bugfix #2648

Thread render with using Area lights was not fully thread safe yet. I
thought I had a smart method to trick threads, but apparently it can best
(and only) be done with Mutexes...

Bug fix #2659

The bugfix for linking recursively from files worked, but there was another
bug hidden... only showed up when actually *using* the recursive feature,
and not with only testing the bugfix. :) Famous coder error, but yah, we
got luckily users to test it!

I'v asked the reporter to give it more extensive testing!

Bug #2661

Drawing face stats in editmesh didn't work... a typo caused using the 2nd
point of a face as third one.

Stricter check for setting posemode. It is possible to have armatures
linked around to more scenes, which confuses the POSEMODE object flag.

Now posemode is only entered on file read when it is also the active
object.

Lukep noticed bug in outliner, when an object has a parent that only resides
in another scene, the outliner inserts that child object in the parent's
scene. Although it seems cosmetic, i could create crashes with it... fixed.

- hide button in button window didn't do anything due to some changes in
transform code, fixed now

Since there still seem to be some commits being done, I thought I might
sqeeze in a minor bugfix.
I wouln't be surprised it is too late,
but since it really is sort of minor, it doesn't matter really.
I had mistakenly put the export of dispersion and absorption parameters in the
ray_mirror block instead of ray_transp, which means that both dispersion and
absorption can only be used if both ray mirror and ray transp are enabled.
Anyway, I don't want to make everyone recompile release builds if already
done, it is minor, but users should take this into account at least.

- last time commit. It fixes bug reported by Kent Mein. Polygonisation of Metaball was broken. Test file:

http://www.cs.umn.edu/~mein/blender/blend/metaball.blend
http://www.cs.umn.edu/~mein/blender/blend/metaball.png

- it should fill octal tree much more better ... polygonisation of MetaBalls is faster little bit now too
- stiffness limit is 10.0

No "release ahoy" yet, so it's probably still ok for these last minute updates (excuse me anyway):

Script bug fixes:
- Paths import: SVG update sent by author Jean-Michel Soler;
- Wavefront OBJ import / export updates sent by author Campbell Barton.