[ODE] Re: triangle collider surface

[Ruud van Gaal]

> More people had this problem. I'm not quite sure why it 
> happens, but it
> appears to come from the class initialisation order.

> > I think I'm down to the last problem now.  Boxes bounce off 
> the terrain as
> > expected, but spheres seem to behave as if the surface was inverted.

Seems like the normal gets inverted. Either in the Box-Tri collisions
function, or the sphere-tri collision function. A matter of defining the
upward tri ordering perhaps (clockwise or counterclockwise).

Ruud