← Back to team overview

yade-dev team mailing list archive

Re: Dem3DOFGeom : ScGeom

 


What is the reason for merging? I don't see any single benefit in that.

Huh!? I really wasted my time then.

The only would be if you precomputed summary shear displacement rather
than increment, so that it could be stored from Dem3Dof and ScGeom
together, but you do not do that.
Precomputing what is needed for Dem3Dof functors and laws is not my job, is it? Shear displacement is not needed for most functors we have in Yade, and for me it is impossible to define in a general way.

Do you have a design paper/wiki page etc for ScGeom? I cannot read all
that code now. To my taste, it is all too hairy to serve as a good
basis.
Maybe read it before stating how it is... For reference, see PFC manual.
You put twisting/bending inside ScGeom. That is not economical and just
adding that to Dem3DofGeom without any value added... why?!

There is no twisting and bending in ScGeom itself. Again, ScGeom is normal, contact point, r1, r2. Cached values are rotations of the local coordinate system and shear increment, not bending and twisting.
those classes contains mostly duplicated code btw
No more than necessary.

With ScGeom, it is not necessary at all.

The fact of not distinguishing sphere+{sphere,facet,wall} cases makes me
think it cannot work properly, especially when it comes to plasticity
and so on.
It can. A contact point is a point. It doesn't care if the bodies behind are cubes or spheres.
Could you suggest an exemple of where this does not apply?

Sorry to be brief. To summarize: I don't understand the reason for that.
So, I revert and resume working with ScGeom as usual? Not really a problem. I this case, I hope I'll not hear comments on developments being "not-prospective" when energy tracing is implemented or when 6DOF code will be moved from Ip to Ig6DOF (it will be in a class inheriting from ScGeom obviously).

Bruno






Follow ups

References