[Bf-committers] BGE instability

[Benoit Bolsee]

Hi,

Back home, I did a full recompile and didn't see any instability
anymore. It looks like the problem came from an incomplete build: I
trusted MVSC to determine what needed to be recompiled after the merge
with the changes from revision 15367 to 15424. Will do more
experimentations but it seems that the problem is gone, sorry for the
noise.
I was having a great variety of crashes, many were actually runtime
breakpoints: in debug mode, MSVC implements many runtime checks on
memory; it slows down execution a bit but helps finding memory
corruptions. I was running with (less) and without (more crash) display
list, on an Intel graphic card under Windows32.

On Tue, 08 Jul 2008 at 15:36:54 +0000 Brecht Van Lommel wrote:

I can't find errors in the code looking over it, and I'm not getting
crashes either. Could you be more specific about which commit broke
things (15264, 15249, 15387)? Also, does this happen on any .blend file,
or just with textures, use blender materials enabled, display list
enabled,.. ? And which operating system and graphics card is it crashing
on? Do the backtraces have any information or are they just random due
to memory corruption?

Cheers,
Brecht.

On Tue, 2008-07-08 at 14:42 +0200, Benoit Bolsee wrote:
Hi,
> 
> Since last merge, the game engine became totally unstable: it crashes 
> very frequently with memory corruption. This has to do with the 
> texture/material rasterizer because the instability disappears when I 
> run the game in wire frame mode. I've seen some update in the 
> rasterizer when I did the merge with my logic updates. Could the 
> author of these changes double check the code for memory corruption 
> problem?
> 
> /benoit