[Bf-blender-cvs] SVN commit: /data/svn/bf-blender [36622] branches/bmesh/blender: remove modified guardedalloc/ghost to rebranch from trunk

Joseph Eagar joeedh at gmail.com
Wed May 11 22:36:31 CEST 2011


Revision: 36622
          http://projects.blender.org/scm/viewvc.php?view=rev&root=bf-blender&revision=36622
Author:   joeedh
Date:     2011-05-11 20:36:30 +0000 (Wed, 11 May 2011)
Log Message:
-----------
remove modified guardedalloc/ghost to rebranch from trunk

Modified Paths:
--------------
    branches/bmesh/blender/source/blender/blenlib/BLI_cellalloc.h

Removed Paths:
-------------
    branches/bmesh/blender/intern/ghost/
    branches/bmesh/blender/intern/guardedalloc/

Modified: branches/bmesh/blender/source/blender/blenlib/BLI_cellalloc.h
===================================================================
--- branches/bmesh/blender/source/blender/blenlib/BLI_cellalloc.h	2011-05-11 20:23:55 UTC (rev 36621)
+++ branches/bmesh/blender/source/blender/blenlib/BLI_cellalloc.h	2011-05-11 20:36:30 UTC (rev 36622)
@@ -33,6 +33,18 @@
  is found we can get rid of this code, but until then this is necassary
  (though, disabling it if jedmalloc is in use might be feasible).
  
+ ideas for replacement:
+	ok, mdisps could store a mempool in CustomDataLayer.  there might be
+	one there already?  vgroups, uh. . .not sure what to do with vgroups,
+	they do cause a significant performance problem.
+	
+	it's tempting to split vgroups into lots of little customdata layers,
+	but that would waste a LOT of memory.  ugh.  can we plug in jemalloc
+	to guardedalloc, on all platforms?  that would work.
+	
+	I really hate this little library; it really should be replaced before trunk
+	reintegration.
+	
  - joeedh
 */
 




More information about the Bf-blender-cvs mailing list