[ODE] Linux Trimesh problems

[Timothy J. Wood]

On Sunday, July 6, 2003, at 5:27 AM, nlin@nlin.net wrote:
>>  - Spheres are rolling back and forth never stopping
>> at one place (perpetrum mobila at last?) - I guess
>> this is due to no friction in spheres

   I'm running into this on Mac OS X with gcc 3.3 AND 2.95.2.  Granted, 
Apple's GCC builds have some extensions / mods over FSF GCC, but this 
makes it seem a little less like a compiler issue.

   (The behavior I'm seeing is that the sphere does slow down, but once 
it is down in the 'pit' it occasionally gains energy and spirals out of 
the pit more forcefully)

On Sunday, July 6, 2003, at 12:52 AM, Russ Smith wrote:
> if the behavior is different between compilers, that suggests there
> is be some kind of "accessing invalid memory" bug in the trimesh
> code.

   It might also be a precision issue -- what do you get if you use 
-ffloat-store (I think this helps avoid extra precision on x86)?

-tim