[Dune] boundary id

Christian Engwer christi at uni-hd.de
Tue Dec 12 14:43:25 CET 2006


Dear all,

> >2) boundary ids are a bit different since we have no iterators and
> >  no index sets for boundaries - in fact no concept of boundaries in
> >  Dune. It would be nice to at least be able to identify boundaries
> >  without having to rely on world coordinate information. This is
> >  why we introduce boundaryIDs for the boundary intersection. The
> >  semantics is - as I understand it:
> >  For each boundary intersection on the macro grid we can define an
> >  id (>0?). During refinement we have: if a boundary intersection I
> >  lies in a boundary intersection I_0 then it has the same boundary id
> >  as id...
> 
> I agree that it would be nice to have proper boundary handling.  But
> I don't think that it is a good idea to add more intermediate 
> functionality two weeks before creating a release branch.  Also,
> personally, I do not have the resources to do the necessariy
> adaptations in UGGrid.

Here I have to back Andreas. This is not an idea arising 2 weeks
before release, but we agreed on that issue about one year ago.

The specification of this is not very strict. It is just said that the
grid should return an id. This id must not be user configurable. As it
was said already yaspgrid supports this feature only in a very course
manner. UG some how supports it aswell... you get an id... but both of
these implementations require some extra work if you really want to
use them.

I would suggest that we stop discussion about whether the feature is
usefull or not [remember we wanted to freeze the API ;-)], but we
should try to figure out how much work a better implementaiton would
require.

Christian




More information about the Dune mailing list