[Dune] dune-grid fails at vertexordertest

[Christian Engwer]

On Fri, May 11, 2012 at 11:36:56AM +0200, Christoph Grüninger wrote:
> Hi Jö,
> thanks for your investigations.
> I don't think that my AluGrid is stale, its a 1.50 from the webpage.

I experienced alugrid segfaults in my own apps in places, where I
wouldn't have expected these. The problem seems to be related to the
new Geometry semantic, but I didn't manage yet to boil down the
problem.

Do you have a callstack?

Christian

> And dunecontrol all was run for all core-modules. The segmentation
> fault occurs both with and without Valgrind.
> I am using openSuse 12.2 Milestone 3 with i386. Due to the virtual
> machine I cannot redo the check with x64.
> 
> Maybe I should try with UG...
> 
> make  check-TESTS
> make[1]: Entering directory
> `/home/gruenich/dune/dune-grid/dune/grid/utility/test'
> WARNING: 2d cube grids not tested because no suitable grid
> implementation is available!
> WARNING: 2d simplicial grids not tested because no suitable grid
> implementation is available!
> WARNING: 3d cube grids not tested because no suitable grid
> implementation is available!
> WARNING: 3d simplicial grids not tested because no suitable grid
> implementation is available!
> PASS: structuredgridfactorytest
> = Testding 1D
> == Testing OneDGrid
> = Testding 2D
> == Testing ALUConformGrid<2,2> with simplices
> Created serial ALUConformGrid<2,2>.
> == Testing ALUConformGrid<2,3> with simplices
> Created serial ALUConformGrid<2,3>.
> == Testing ALUCubeGrid<2,2> with cubes
> Created serial ALUCubeGrid<2,2>.
> == Testing ALUCubeGrid<2,3> with cubes
> Created serial ALUCubeGrid<2,3>.
> == Testing ALUSimplexGrid<2,2> with simplices
> Created serial ALUSimplexGrid<2,2>.
> == Testing ALUSimplexGrid<2,3> with simplices
> Created serial ALUSimplexGrid<2,3>.
> = Testding 3D
> == Testing ALUCubeGrid<3,3> with cubes
> /bin/sh: Zeile 5: 11527 Speicherzugriffsfehler  ${dir}$tst
> FAIL: vertexordertest
> 
> 
> Ciao
> Christoph
> 
> -- 
> People who are more than casually interested in computers should have
> at least some idea of what the underlying hardware is like. Otherwise
> the programs they write will be pretty weird.         -- Donald Knuth
> 
> _______________________________________________
> Dune mailing list
> Dune at dune-project.org
> http://lists.dune-project.org/mailman/listinfo/dune
>