Hi Nicholas,

> >But... What's difference between hinge to elbow? What do you want?
> 
> oh, just an elbow has a 0 - 180 degree range of motion. So if you an 
> constrain joints then you can make al sorts of 'figures' like 
> marionettes/puppets - which will be 'realistic' if you 'shake'/'perturb' 
> them.

Ah, that is fun. I'm trying to think how it is designed.

> I started reading the ode docs, the api is huge and has a lot of power. Is 
> your goal near total coverage (with a subset only in e-toys)? how close are 
> you to your goal?

ODECo has three layers to deal with ODE.
1) etoy layer:
   No, e-toy is 2-D world, it doesn't need full function of ODE.

2) ODE class library layer:
   Almost yes, but that's not so easy...

3) ODE plugin layer:
   Yes, many APIs are build automatically from C definition.

> I read in their docs they have a 'slider' joint that you can add constraints 
> to, and that the constraint has 'springiness' - but I've not read enough to 
> know what that means exactly.

I will try it.

> > > 4 - what do you think is the limiting factor for number of bodies? ode 
> >or
> > > the squeak graphics?
> >
> >Almost 50 - 100 with small morph. But in many case, performance issue is
> >in drawing TransformationMorph.

Actually, my answer was not exactly. In case of using small morphs,
the limiting of 50 - 100 objects is inside ODE's C++ code (although
just my guess) and it crashes vm. But other case, the limit is in
drawing morph.  Of course, it depend on your machine spec.

> your SAR mentions croquet, this would be fun in a wonderland also.

Basically, ODE is for 3-D simulation. In the fact, it would be easier to
building 3-D simulation with ODE than 2-D like etoys. But I don't have good
idea for 3-D interface for ODE yet.

regards,

Takashi