dolfin team mailing list archive

[Johannes Ring <johannr@xxxxxxxxx>]

On Tue, Sep 15, 2009 at 9:16 PM, Anders Logg <logg@xxxxxxxxx> wrote:
> On Tue, Sep 15, 2009 at 08:33:08PM +0200, Johannes Ring wrote:
>> On Tue, Sep 15, 2009 at 7:19 PM, Anders Logg <logg@xxxxxxxxx> wrote:
>> > The buildbot is currently failing on
>> >
>> >  demo/pde/poisson1D/python
>
> Thanks for the output. This should be fixed now.
>
>> Is MUMPS a dependency? Or do we require PETSc to be built with MUMPS
>> now? I thought it was only optional.
>
> Yes and no. It is not required in serial, but it is required for
> running some of the demos in parallel. When running in parallel, some
> demos require a parallel linear solver and MUMPS and SPOOLES are two
> options.
>
> So even if it is not a requirement for DOLFIN, the buildbots need to
> install it so they can run the demos in parallel.
>
> I used the following configuration for PETSc to get MUMPS:
>
> ./config/configure.py COPTFLAGS=-O3 --with-debugging=0
> --with-clanguage=cxx --with-shared=1 --download-hypre=1
> --with-parmetis=1 --with-parmetis-dir=/usr/include/parmetis
> --download-scalapack=1 --download-blacs=1 --download-mumps=1

OK, I will try to rebuild PETSc with MUMPS on the slaves tomorrow.

Johannes