[ODE] Sphere sinking through trimesh

[Patrick McColgan]

Hello

I've found that when I place a sphere on a trimesh it sinks into the 
trimesh, until it sinks deeper than its radius and fall through the 
level completely.  I've used spheres and trimeshes before and never 
noticed it before and there is no record of it on the mailing list so it 
could be something I'm doing.

The sphere is of no great mass (10 units) or radius (0.5) and changing 
these values makes no difference.  When the sphere is stationary on the 
trimesh it slowly sinks into the mesh, this doesnt happen with capsules 
or boxes (possibly because they generate more than one contact) and the 
sphere does generate one contact with the mesh.  When its moving it 
seems to be fine.

I found I could correct this behaviour by changing the world ERP from 
0.1 to 0.5, the sphere stops sinking at 0.5 not 0.45, I use a world CFM 
of 0.0 and a contact surface layer of 0.001.

I was using hard contacts (not setting any of the contact surface 
parameter modes) but I now set the soft erp flag and set it to 0.5 (I 
don't want the world ERP to be 0.5 just to fix this).

Whilst I have 'fixed' it I would like to know if this is standard 
behaviour for spheres and trimeshes or if there are other changes I 
could make?

Thanks