Trimesh-trimesh collider (was Re[2]: [ODE] Triangle-box collidercontribution)

[Adam D. Moss]

Jeffrey Smith wrote:
>>One of the things it's missing in this respect is the
>>updating of the geoms' homogenous transform matrix, which you
>>have indicated improves results -- I ran out of time,
> 
> Oh, jeez.  It won't work at ALL if that's not being passed to dCollideTTL.
> The collision code uses that to determine the velocities of the colliding
> objects, and thus the best directions for collision.  If those transforms
> aren't being passed, then you're going to get some educated guesses, but
> mostly garbage, out.

It worked well enough for me to check that the merge worked and
the code attempted to perform as advertised, that being all I had
time for.  I assumed that using dGeomTriMeshDataSet() would make
things rather better but the proof of that will have to wait
until I have more time or someone submits a patch.  I wanted to
get this overdue submission in the tree...

-- 
Adam D. Moss   . ,,^^   adam at gimp.org   http://www.foxbox.org/   co:3