[Dune] dunecontrol make test

[Mazaheri, Alireza (LARC-D305)]

I forgot …. the Intel MPI is only experimental (I had to get approval to use it; anything but MPT seems to break our InfiniBand) to check that I have the right environment and the code produces expected results.
Eventually I would like to have the fix with the anonymous MPI_COMM_WORLD.

On Jul 3, 2017, at 5:21 PM, Mazaheri, Alireza (LARC-D305) <ali.r.mazaheri at nasa.gov<mailto:ali.r.mazaheri at nasa.gov>> wrote:

Hello,

On Jun 29, 2017, at 5:30 AM, Markus Blatt <markus at dr-blatt.de<mailto:markus at dr-blatt.de>> wrote:

Weiredly you MPI_COMM_WORLD is an anonymous enum. Not sure whether this means that you are using the C++ bindings of MPI.

The SGI MPI defines MPI_COMM_WORLD as an anonymous enum (see https://gitlab.dune-project.org/core/dune-common/issues/76) and that breaks the test.
So I am now using the latest Intel’s MPI, which is the only other available MPI in our system. With this, I was able to pass this issue and go farther.
But I am encountering some failures in building the 'dune-fem' tests (see attached log file). Does this constitute a potential bug or something that I have missed to set?

I am using gcc/6.2 + intel's MPI 2017.3.196 + modules from the master branch.
<dunefem-test.log.zip>
Regards,

Alireza


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.dune-project.org/pipermail/dune/attachments/20170703/6f7fe811/attachment.htm>