[Dune-devel] dune-istl umfpack support

Martin Nolte nolte at mathematik.uni-freiburg.de
Wed Oct 16 08:50:43 CEST 2013


Hi Christoph, hi all,

Markus is right: dune-istl is optional for dune-fem and it can use umfpack 
without dune-istl being present. Therefore, I am in favor of migrating the 
test to dune-common (to reduce maintenance). We have done so for for other 
tests, like metis / parmetis before.

You can, of course, argue that it would be cleaner to have the test only in 
the modules using it. On the downside duplicating tests not only increases the 
maintenance (you have to maintain each copy), but having different tests might 
also violate the principle of least surprise:
- Why does dune-istl find umfpack while dune-fem does not find it?
- Why is umfpack no longer found by dune-fem once I checkout dune-istl?
You also end up with an authority problem: Which test is correct - the one 
from dune-istl or the one from dune-fem. Migrating the test to dune-common 
settles all the problems once and for good.

Best,

Martin

On 10/15/2013 02:05 PM, Markus Blatt wrote:
> On Tue, Oct 15, 2013 at 01:40:58PM +0200, Christoph Grüninger wrote:
>> wouldn't it be better to leave the test in dune-istl? Dune-fem should
>> anyway rely on dune-istl. In dune-istl is already the superlu and
>> pardiso test. For me solver related tests are supposed to be in dune-istl.
>
> AFAIK dune-istl is optional for dune-fem.
>
> Markus
>
> _______________________________________________
> Dune-devel mailing list
> Dune-devel at dune-project.org
> http://lists.dune-project.org/mailman/listinfo/dune-devel
>

-- 
Dr. Martin Nolte <nolte at mathematik.uni-freiburg.de>

Universität Freiburg                                   phone: +49-761-203-5630
Abteilung für angewandte Mathematik                    fax:   +49-761-203-5632
Hermann-Herder-Straße 10
79104 Freiburg, Germany




More information about the Dune-devel mailing list