Re[4]: [ODE] using dCreateGeomClass in the last ode version

[Russ Smith]

> Seems it works all right.

good.

> But one problem still remains for user definite geometry classes.
> If I have some user definite classes it always necessary to support
> right sequence of their creating or colliders would not be set in
> colliders array again. Of course I can register all user classes in
> right order before start my simulation. But if I want to support
> ode stile of creating geometry classes when creating geometries, it is
> very inconvenient.

can you show me a situation that fails? - because i thought i had
allowed for arbitrary creation order. dCreateGeomClass() sets a default
collider (dCollideUserGeomWithGeom) and does not call the supplied
colliders function. it returns the geom class number. the collider
function is only referenced by dCollideUserGeomWithGeom() ... so i don't
see how creation order makes any difference.

> Why not open setCollider function?

this is not a good option, because this function exposes some of the
internals of the way dCollide() works. exposing this will prevent any
further major overhaul of the collision system (not that i have one
planned).

russ.

-- 
Russell Smith
http://www.q12.org