[ODE] 0.8.2 candidate

[Jean-Sébastien Guay]

Hello Bram and all contributors,

> Others, could you please test Jason's upload, or alternatively, just fetch
> the current trunk from svn, and test that?

I'm testing from an SVN checkout updated as of a few minutes ago.  
Windows XP SP2, Visual C++ 2005 Express Edition on an Athlon 64 X2  
4200+, 2GB RAM, nVidia GeForce 7900GTX 512MB. Settings as per  
config-default.h, except I wanted to try the new trimesh collider  
(without much success, see below):

#define dSINGLE

#define dTRIMESH_ENABLED 1
#define dTRIMESH_OPCODE 1

#define dTRIMESH_OPCODE_USE_NEW_TRIMESH_TRIMESH_COLLIDER 1

The build is ok, though I get a lot of warnings. Is the policy to  
ignore them or should I post them or fix them and submit a patch? In  
general a simple typecast would fix them, and in my own projects I  
like to keep warnings at zero to minimize noise in the build, but I  
know some people don't agree with that stance and ignore warnings to  
keep the code itself less noisy. Both approaches have merit, it's a  
matter of preference.

I haven't tested all demos, here is what I did test:
- demo_boxstack works well.
- in demo_heightfield, If I drop 4 boxes or more in rapid succession  
(press 'b' 4 times quickly) when the program just started, I get an  
"Access violation reading location 0xbaadf011" (which I think is a  
code for some bad pointer access in Visual Studio) and it stops at the  
dCollide line in demo_heightfield.cpp (line 1568). If I drop less  
boxes (up to 3) it works fine, and even if I drop one, wait for it to  
collide, and then drop 4 or more, it works fine. It's just if I drop 4  
or more as the first collision when the program starts that it crashes.
- in demo_moving_trimesh, when dropping a trimesh with 'm', I get  
"assertion "dFabs(dVector3Length((dVector3&)*normal) - REAL(1.0)) <  
dEpsilon" failed in  
..\..\ode\src\collision_trimesh_trimesh_new.cpp:328". Is this expected  
with the new trimesh collider? With the old collider (#define  
dTRIMESH_OPCODE_USE_NEW_TRIMESH_TRIMESH_COLLIDER 0) I don't get the  
assertion failure, but the dropped bunny never comes to a rest. It  
bounces around (hehe, a bouncing bunny) and penetrates a bit, and  
eventually bounces high enough that it falls beside the other bunnies.  
But I think that's how it's been for a while, and why you generally  
don't recommend to use trimesh/trimesh collision.

Any other more specific things that it would be useful to test?

Good work to all, lots of good stuff in this (potential) release. I  
look forward to using ODE in a real project in the near future (as  
opposed to just reading the mailing list and testing builds once in a  
while... :-) ).

J-S
-- 
______________________________________________________
Jean-Sebastien Guay     jean-sebastien.guay at polymtl.ca
                         http://whitestar02.webhop.org/

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.