yade-dev team mailing list archive

[Chia Weng Boon <chiaweng@xxxxxxxxx>]

Bruno,

You are absolutely right in all those cases.   ScGeom can make use of a
boolean for the twist.  Yes, Cundall's paper is almost 30 years old.  He
may have changed his mind too.  I remember the code crashed when I was
using the new updateProperties.  I am not a good debugger.   And the
Clump::del should work.

In fact, there had been some changes in YADE's main program that I had to
modify my old code, to make it work.  I mentioned those few regressions,
 so that active developers are able to identify them and fix it.  I decided
that if I were to keep the code and try to update it nicely before sharing
it out, YADE may get updated again, before I can catch up.  I better share
it when it is working.  This is my advice for other researchers.  Don't
keep your code, because you will likely have to keep up with the
developments in YADE. And you won't have time to play/maintain with your
own code anyway when you move on with life.   Now that I am in construction
and not in a university, it is very difficult to have the luxury of finding
time for this hobby.



Yours,
Boon

On Thu, Mar 17, 2016 at 11:01 PM, Bruno Chareyre <
bruno.chareyre@xxxxxxxxxxxxxxx> wrote:

> Hi Boon,
> Thank you for progress report!
>
> On 16/03/16 16:37, Chia Weng Boon wrote:
> > I have also changed some of the existing files to make it compatible with
> > my old files.  Please update and modify it accordingly if you think there
> > is a better way to do it.
> > ScGeom:  I took away the twist update because for polygons this is not
> used
> > in Cundall's paper.  Please modify accordingly if there is some
> theoretical
> > discovery.
> Does it mean that the twist will be also disabled if someone has, let's
> say, a sphere-sphere contact?
> And did you find a reason why the twist should be defined differently
> (or even undefined) in the case of polygons?
> In general Yade is not meant to reproduce Cundall's equations on every
> aspect.
>
> > Clump:  The Update Properties looked different from the old one, and I
> had
> > problems with clumps earlier on when updating the code.  Not sure if this
> > is the problem, but i used the old one anyway.
> We must make sure that adding some code does not imply regressions as
> side effects.
> Do you have any clue why the polyhedra code needs an older version of
> updateProperties? What sort of problem do you find?
> Rolling back such a function is most likely breaking, or at least
> changing the behavior, of other parts of the code.
>
> > YadeWrapper:  Added deleteClumpMembers function, since it was in my old
> > version.
> Is Clump::del not doing the same thing? Do you need such a function?
>
> As far as I understand, you are half-way to a compatible code overall.
> It will be really compatible when no changes will be required in other
> files.
> If somene else will do the remaining work, it would be helpful to him to
> have more detailed explanations of why you had to roll back some changes
> (which really means something to fix in your code, not something to roll
> back in trunk).
>
> Cheers
>
> Bruno
>
>
>
> _______________________________________________
> Mailing list: https://launchpad.net/~yade-dev
> Post to     : yade-dev@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~yade-dev
> More help   : https://help.launchpad.net/ListHelp
>