yade-dev team mailing list archive

[Václav Šmilauer <eudoxos@xxxxxxxx>]

> 3. Added phase flag "cycle" to Interaction. SpatialQuickSortCollider is modified so interaction's cycle flag is used to mark stale contacts instead isReal flag.
>   
I wanted to address this some time ago. I think though that we could
have a meaningful set of booleans (like isOverlapping, isNew, is*) that
would cover all we need for collisions; otherwise every collider will
have different set of phases of interaction and constitutive laws that
rely on those information will be collider-specific.

I would be also glad if you could document it somehow, since
PersistentSAPCollider would be modified as well, perhaps.

BTW, do you have benchmarks comparing PErsistentSAP and SpationQuickSort?
> 4. Small changed InteractingSphere2InteractingSphere4SpheresContactGeometry and GLDrawElasticContactInteraction. I think, towards an improvement. If there are objections, changes must be reverted.
>   
In GLDrawElasticInteraction, you could use static serializable flags to
control rendering of things (and you would have to make
GLDrawElasticInteraction serializable). Have a look at
GLDrawBrefcomContact, the I use MetaBody::miscParams (which doesn't
serve anything else)

 o.miscParams=[Generic("GLDrawBrefcomContact",{'dmgLabel':False,'epsT':True,'epsTAxes':True,'normal':True})]

to control GL rendering. If you keep the old behavior as default, there
can be no objections ;-)

> 5. pyOmega.save allows recovery 
>   
Good. Although what I want to do is to move virtual time and iteration #
into MetaBody, then there will be no difference of recovery/nonrecovery
save.

BTW this recovery thing is (should be) triggered if you send HUP to yade
- it was meant as a way to save very long simulation when UPS signals
power loss and the system shuts down, so that all what has been
calculated is not lost. Yade should open recovery files automatically at
startup, but this has not been very much tested and probably never used
in real.

Vaclav
_______________________________________________
yade-dev mailing list
yade-dev@xxxxxxxxxxxxxxxx
https://lists.berlios.de/mailman/listinfo/yade-dev