[ODE] Angular-velocity autodisabling considered broken

[Jon Watte]

> I think that the problem that needs to be addressed is the jittering 
> itself. Using the auto-disabling mechanism to reduce the effects may work 

Because ODE is a penetration based system, I'm not sure you 
can get rid of the jitter. You can minimize it by using a 
constant time step, but you can't make it go away, just 
because of the way the system works. Things aren't "glued" 
when they're resting, they're constantly inter-penetrating 
and having forces applied to cancel that out.

Now, a bunch of people are using variable time steps, usually 
because it's easiest and they don't know better. Could we 
make it harder to use a variable time step, say by having an 
API call dWorldSetTimeStep( dReal ), and not take a time step 
size in dWorldStep() and friends? You still COULD do the 
variable time step if you wanted, by updating it every frame, 
but it would send a not-so-subtle clue that that's not the 
ideal thing to do.

Cheers,

			/ h+