[ODE] iterative solver: testing needed

[David Whittaker]

> Hi,
>    are you still working on the iterative solver?
> i've encountered some problems while playing with it:
>
> in CYGWIN env:
> in test_crash(with the only change being uncommenting #define WALL) upon
>  switching to fast mode it crashes immediately

Are you sure you're switching into fast mode and not back into normal
mode?  I know the wall crashes in normal mode in some environments,
probably due to a stack overflow.  You can tell what mode you're in by the
color of the ball in the sky... red is normal, green is fast.

> in MSVC env:
> joint behavior is very odd in general (a hinge joint seems to come apart
>  like an unrestrained slider)
> presense/use of amotorjoints causes this error:
>
> "TEXTURE! dJointGetAMotorMode + 560 bytes
> TEXTURE! dInternalStepIslandFast(struct dxWorld *,struct dxBody * const
> *,int,struct dxJoint * const *,int,double,int) + 741 bytes"
>
> ode was compiled in DEBUG mode (otherwise the card wouldn't appear)

Haven't seen this one, I'll get back into windows (ugh) and test it there
a bit before I post the next version.  The next version should be a good
bit more stable, and even more memory efficient, but just a tiny bit
slower.  I'll have the autoEnableDepth parameter set up (as a global
variable), but I won't set up autoDisabling until I work out the bugs in
the solver itself.  I'm aiming for another release either tonight or
tomorrow night (if I can keep from mesmerizing myself.... how big a wall
can I take out with one car :).

> any thoughts?
> mike