Revision d3960ee by Sergey Sharybin November 19, 2012, 16:46 (GMT) |
Fix #33223: Instant Crash on Window minimize Issue was caused by manipulating with triple buffers of minimized window. It's more like an Intel driver bug but we can workaround it in quite clear way by skipping draw of minimized windows. |
Revision cfd5d58 by Sergey Sharybin November 19, 2012, 16:42 (GMT) |
Fix #33229: Blender crashes on finishing render |
Revision 1e98e96 by Brecht Van Lommel November 19, 2012, 16:02 (GMT) |
Windows Libs: move OpenEXR headers to same directory structure as other platforms. |
Revision abce630 by Brecht Van Lommel November 19, 2012, 15:50 (GMT) |
Scons: better fix for OSL compiler path, so it also works on OS X. |
Revision 2f50102 by Lukas Toenne November 19, 2012, 15:18 (GMT) |
Implementation of string lookup functions for collections in the C API. These were declared in RNA_blender.h but not actually implemented in the respective *_gen.c files. |
Revision 48639af by Campbell Barton November 19, 2012, 14:58 (GMT) |
use input and output slots for bmesh operators, needed for the python api to get return values. |
Revision 312abf8 by Brecht Van Lommel November 19, 2012, 14:54 (GMT) |
SCons: make OSL linking work on windows, and fix OSL compiler path issue to properly expand to absolute path. |
Revision 9947f70 by Sergey Sharybin November 19, 2012, 13:15 (GMT) |
Localization should not depend on debug locale library Linking happens fine, but blender crashes on startup -- crash with familiar backtrace happens with i18n disabled (in that case it's something to do with OIIO). |
Revision 3b57044 by jens verwiebe November 19, 2012, 11:46 (GMT) |
OSX/scons: refine oslc compile workaround |
Revision 45f75df by jens verwiebe November 19, 2012, 11:45 (GMT) |
OSX/cmake: py 3.3 migration |
Revision 7ecbc39 by Bastien Montagne November 19, 2012, 11:12 (GMT) |
Basic work for an ALL_STATIC option, to enable when you want to build blender in a portable way. Still WIP! |
Revision 063ca9b by Bastien Montagne November 19, 2012, 10:45 (GMT) |
No boost system when we have own compiled version! |
Revision 29fe8ab by jens verwiebe November 19, 2012, 10:14 (GMT) |
OSX/scons: migration to py 3.3 |
Revision 67031c1 by Bastien Montagne November 19, 2012, 08:51 (GMT) |
Fix when statically linking with distro's boost: in this case, we most likely also need to statically link against icu, as most boost packages are built with it. Without that, you get a bunch of errors at link time (when using boost_locale, or, in freestyle branch, boost_regex). So when you enable Boost_USE_STATIC_LIBS, you should also set Boost_USE_ICU to True. Will add a note about that in build doc too. |
Revision 509e178 by Bastien Montagne November 19, 2012, 08:48 (GMT) |
Definitive fix for version_ge with equal versions like "1.0.0" and "1.0". Also fixed DEB boost version checking, own fault. And disabled building ocio's apps, else it would go searching for an oiio lib (and we have not yet built ours) - anyway, if users want them, they can build them on their own! |
Revision 185cf60 by Campbell Barton November 19, 2012, 02:26 (GMT) |
improvements to bevel - the resulting selection is now correct internal details - bev_rebuild_polygon() now only rebuilds polygons that are attached to a bevel vertex (was rebuilding ALL). ... need to take care we don't leave faces pointing to removed geometry, so far this works fine. - bev_rebuild_polygon() uses stack memory for <32 size ngons to reduce allocs. - skip hash lookup when removing bevel verts (use tag instead). |
Revision cdc4037 by Campbell Barton November 19, 2012, 00:54 (GMT) |
bmesh: BM_verts_in_face was using bmesh operator flag which is no longer ensured to be available, use internal apiflag instead, Thanks to Nicholas Bishop for spotting. also quiet some warnings. |
Revision b7cd9ec by Tamito Kajiyama November 19, 2012, 00:41 (GMT) |
Merged changes in the trunk up to revision 52340. Conflicts resolved: source/blender/blenloader/intern/readfile.c |
Revision eb1fccd by Mitchell Stokes November 18, 2012, 20:56 (GMT) |
BGE: The Step Height for a character physics controller is now set to 0.01, which means that Step Height can no longer be 0, which causes issues such as those noted in issue #33221, "KX_CharacterWrapper inconsistencies and error," reported by Angus Hollands (agoose77). I've also added a do_versions() fix that will give objects with 0 Step Height sane defaults. This means old files will now work better with character physics. |
Revision 2bed09e by Mitchell Stokes November 18, 2012, 20:38 (GMT) |
BGE: Changing "transform_uv" -> "transformUV" in some error messages. |
|