yade-dev team mailing list archive

[Bruno Chareyre <bruno.chareyre@xxxxxxxxxxx>]

I'm worrying about that too. It needs some discussion, as we put weights 
a bit differently sometime, when we define what is "better". :)
For instance, I fear oversplitting because it is so hard for somebody to 
read the code.
> PS I think the way both Dem3DofGeom and ScGeom currently work is not
> prospective. ScGeom is dumb and only stores a few numbers inside, making
> Law doing all the computation;
Don't you refer to a months-old version of ScGeom? ScGeom has members!

> As we already discussed, it would be much more efficient to precompute
> normal and shear displacements in the Ig2 functor and store it in the
> *Geom class. Only writing back to *Geom from Law2 (such as upon a
> plastic slip) would need a few virtual methods, which could be
> interface-wise identical for incremental and total formulations BTW.
>
>   

Ok for precomputing un and us, but is there a general reason to put more 
members in a geometry class?
With ScGeom, there is no plastic slip to update in the geometry, I'm not 
sure we need any virtual function in incremental formulation.

Bruno