[ODE] ccylinder/box collision problem

[Jon Watte (ODE)]

There's one problem where ccylinder/box (or sphere/box) may cause 
denormal numbers, that then get inverted and cause numerical 
instability. I submitted a patch a year ago, but it was rejected because 
it used specific epsilons (derived from actual numerical type limits).

Cheers,

			/ h+


Brad wrote:
> I've got a simple program where I've been doing box/box and box/plane 
> collisions for a while now, and the other day I added ccylinders. The 
> ccylinders collide with other ccylinders, rays, and planes fine, but 
> box/ccylinder collisions seem really buggy. If the ccylinder hits the 
> box with more than just a little bit of force the ccylinder penetrates 
> the box, and then starts dancing around, unable to resolve the 
> penetration. The ccylinder even disappears inside the box, occasionally 
> peeping out.
> 
> Anyone else had similar trouble?
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode
> 
>