Blender Git Commits

Blender Git "master" branch commits.

Page: 3061 / 5574

October 4, 2013, 01:10 (GMT)
fix [#36942] unfreed memory on trackball widget use
October 3, 2013, 16:39 (GMT)
Typo in variable name from previous commit, sorry.
October 3, 2013, 16:34 (GMT)
set mempool debug in the game-engine-player too. also remove redundant mempool includes.
Revision f372d14 by Sergej Reich
October 3, 2013, 16:12 (GMT)
rigidbody: Flag bodies for validation when changing shape

This fixes a pretty old crash that occurs when changing the collision
shape to a mesh shape during simulation.

Thanks to Thomas Beck (plasmasolutions) for the report.

svn merge -r60310:60311 ^/branches/soc-2013-rigid_body_sim
October 3, 2013, 15:29 (GMT)
fix crash in editmesh edge-dissolve, when 2 faces shared multiple edges, freed edges would be checked.
October 3, 2013, 14:44 (GMT)
freeing mempool elements now fills freed memory with --debug for debug builds.
Revision c819fd4 by Gaia Clary
October 3, 2013, 14:24 (GMT)
Changed some labels from Second Life to OpenSim
October 3, 2013, 13:35 (GMT)
fix for crash dissolving faces, bmesh walkers were using freed mempool elements.

use a convention for freeing walker-states where freed-dangling pointers aren't left assigned to prevent errors like this happening in future.
October 3, 2013, 13:15 (GMT)
Fix crash starting game engine on linux

Issue was caused by bug in mesa #54080 which makes
glXQueryDrawable fail with GLXBadDrawable for any
request with direct context.

Worked around by temporary overriding X error handling
when getting old interval value and disablingintervals
extension if this query fails.

Also added check for glXSwapIntervalEXT which is
apparently NULL here with GLX_EXT_swap_control=1.
October 3, 2013, 12:22 (GMT)
add cmake option WITH_MEM_VALGRIND, helps to track down errors with mempool use which sometimes only show up as bugs in very rare cases (because even though the element is freed, the chunk is still allocated).
October 3, 2013, 10:50 (GMT)
yet another instance dissolve deletes verts it loops over.
October 3, 2013, 10:29 (GMT)
overlapping faces made join-triangles fail, check for this case and silently ignore them.
October 3, 2013, 08:02 (GMT)
addition to r60523 missed using BM_ITER_MESH_MUTABLE in one place.
October 3, 2013, 07:06 (GMT)
fix [#36923] Merge / Delete vertices crashes for some meshes

add BM_ITER_MESH_MUTABLE which steps before entering the for() loop body and prevents the assert from complaining about removing mesh data while iterating as well as the crash.

this was done in quite a few areas, more may turn up.
October 3, 2013, 07:02 (GMT)
remove assert recently added to EDBM_backbuf_check and explain odd logic.
October 3, 2013, 07:01 (GMT)
fix bug in ED_vgroup_subset_from_select_type, setting negative index in boolean array.
also was freeing NULL pointer in vgroup_blend_subset()
October 3, 2013, 06:12 (GMT)
add debug mode assert if mesh-data is removed while iterating over it. (can crash).
October 3, 2013, 05:27 (GMT)
fix [#36924] Calling .define(...) on Macro subclass crashes Blender, if subclass wasn't registered
October 3, 2013, 04:31 (GMT)
console: shift pad-enter now clears the line too.
October 3, 2013, 03:05 (GMT)
adjustments to sculpt cache scaling code, (float/double promotion)
By: Miika HämäläinenLast update: Nov-07-2014 14:18MiikaHweb | 2003-2021