[ODE] Joint tutorial for ODE?

[nlin@nlin.net]

Hi Russ,

Great to hear you're back from holiday! Welcome back to the list.

I was wondering if you could update the documentation on joints anytime
soon - for instance, a tutorial on how the ball and socket joint constraint
is derived. I _almost_ have it figured out (after finally finding out
which Anitescu/Potra and Trinkle/Stewart velocity-based implicit time stepping
method you are using), but I am still somewhat foggy on EXACTLY what equations
you use to derive the joints. What I'd like to see in the documentation
is something simple like: 

- Ball/socket joint: point-to-point constraint. C(...) = ...
- Derive Jacobian from C as follows ...
- Calculate right hand side as follows ...

I think this sort of documentation would enable others to contribute their
own joints. As it is, without knowing the exact constraint equations ODE
is solving, I can't make too much of the joint code. As I understand it,
the physically-based modeling course notes (which you refer to in the ODE
docs) use a non-time-stepping, instant-to-instant formulation of the rigid
body problem, meaning that the equations derived in the course are similar
in spirit, but not in detail, to ODE's equations. Can you clear this up?

No hurry, but I think this would help enable further ODE development by
others.

Thanks,
-Norman