Convex Hull collision (was Re: [ODE] PosR - a better way?)

[Megan Fox]

I had asked about this some time ago with no real response, and
assumed either it was impractical, or too hard, or that convex
hull-type collisions weren't a good solution... etc... but since you
mentioned it -

Has anyone tried integrating convex hull collision into ODE?  Say,
using qhull, etc?  I keep running up against cases which would be
ridiculously easier to represent via convex hull or trimesh, but I'm
not willing to touch trimesh-trimesh with a 10-foot pole.

-Megan Fox

> That may be true. I haven't found trimesh-trimesh to be stable
> enough to use, so I'm not using it. Unless that got fixed, too :-)
> (I think ODE would need a "convex hull" geom type to do this
> well)
> 
> Cheers,
> 
>                        / h+