yade-dev team mailing list archive

[Václav Šmilauer <eudoxos@xxxxxxxx>]

> > inertia and mass : no.
> > The rest should be in "state" since it changes with time.
> >
> Oh wait!
> 
> Reading this in the blueprint description rings a bell for me :
> "Refactor PhysicalParameters into constant part (Material) and variable 
> part (BodyState)"
> 
> 1/ Mass is usually constant but it is not a material parameter.
> 2/ What will happen if one needs to modify the mass of some body during 
> a simulation?

> 
> There is in fact no strict equivalence between "constant" and 
> "material", or between "body" and "variable".

That is right and I apologize for not being very precise on that. The
distinction should run (what I think at least) between what is per-body
and what is common.

If material is to be shared amongst bodies (which is the plan), then
perhaps _density_ could be part of it, but not mass, which depends on
body's geometry and is part of "State" in that sense (for lack of better
words, I called it State).

If I understand correctly, all of use seem to agree that having
positional things and rotational things in the same bag is the way to
go, right?:

* Positional: 3 vars (position, velocity, accel), 1 constant (mass)
* Orientational: 3 vars (orientation, ang. vel, ang. accel), 1 constant
(intertia vector).

BTW should we keep the somewhat obscure Se3r class or separate that in
Vector3r + Quaternionr?

Thanks, v.