[Dune] [Dune-Commit] dune-grid r6196 - trunk/m4

[Martin Nolte]

These warning messages are to be expected. They origin from the "wrong" test for
geometryInFather. This test assumes that the union of the children of one
element from exactly the father element. But this is not true, if the nodes
created during adaptation are projected to the boundary (which is the case,
here). Should we disable the geometryInFather-check in this special setting?

As for the problems mentioned in the FlySpray: Do they still persist? I usually
only check with Alberta 3.0 (except when I know it might break Alberta 2.0).

Yours,

Martin

Jö Fahlke wrote:
> Am Tue, 19. Jan 2010, 19:49:23 +0100 schrieb joe at dune-project.org:
>> Author: joe
>> Date: 2010-01-19 19:49:23 +0100 (Tue, 19 Jan 2010)
>> New Revision: 6196
> [...]
>> This magically seems to make the test dune/grid/test/test-alberta pass,
>> although there are still messages like
>>
>>   geometryInFather yields different vertex position (son: 1.41421 0, father: 1 0)
> 
> Actually, it doesn't, at least not with Alberta 2.0.1.  The message was
> produced with Alberta 3, and there are more messages like it:
> 
> ======================================================================
> geometryInFather yields different vertex position (son: 1.41421 0, father: 1 0).
> geometryInFather yields different vertex position (son: 1.41421 0, father: 1 0).
> geometryInFather yields different vertex position (son: 0 -1.41421, father: 0 -1).
> geometryInFather yields different vertex position (son: 0 -1.41421, father: 0 -1).
> geometryInFather yields different vertex position (son: -1.41421 0, father: -1 0).
> geometryInFather yields different vertex position (son: -1.41421 0, father: -1 0).
> geometryInFather yields different vertex position (son: 0 1.41421, father: 0 1).
> geometryInFather yields different vertex position (son: 0 1.41421, father: 0 1).
> Warning: geometryInFather yields different vertex positions.
>          This behaviour may be correct if the grid is not nested geometrically.
> ======================================================================
> 
> So does anyone know whether this is expected?
> 
> Bye,
> Jö.
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Dune mailing list
> Dune at dune-project.org
> http://lists.dune-project.org/mailman/listinfo/dune

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

Universität Freiburg                                   phone: +49-761-203-5642
Abteilung für angewandte Mathematik                    fax:   +49-761-203-5632
Hermann-Herder-Straße 10
79104 Freiburg, Germany