yade-dev team mailing list archive

[Jérome Duriez <jerome.duriez@xxxxxxxxxxxxxxx>]

The attached file....


Le Vendredi 19 Septembre 2014 00:38 CEST, Jérome Duriez <jerome.duriez@xxxxxxxxxxxxxxx> a écrit:

> Hi,
>
> I'm reading for the first time capillary code (Law2_ScGeom_CapillaryPhys_Capillarity::action). Depending on the contact law used in the simulation, two different interaction physics are handled :
> - either "cundallContactPhysics" (CapillaryPhys type)
> - or "mindlinContactPhysics" (MindlinCapillaryPhys type)
>
> And, in the whole code, "herzOn" bool variable is checked, to know if cundallContactPhysics, or mindlinContactPhysics has to be accessed. I have the feeling that, in the end, we always access variables with same name and meaning (give a look at the attached diff, comparing the attributes list of these two classes), either from cundallContactPhysics, or mindlinContactPhysics. E.g. hertzOn? mindlinContactPhysics->meniscus : cundallContactPhysics->meniscus...
>
> My question is : could we get rid of these two different variables, use only one variable "contactPhys", defined once for all with :
>
> if !hertzOn
>     contactPhysics = static_cast<CapillaryPhys*>(interaction->phys.get());
> else
>     contactPhysics = static_cast<MindlinCapillaryPhys*>(interaction->phys.get())
>
> And, after that, use only contactPhysics->meniscus
>
> A second question would be "is it worth ?', but I could do the task without answering this question...
>
> Jerome

Attachment: diffCS_HM.pdf
Description: Adobe PDF document