[petsc-users] How can PETSc configure with MSMPI

Jed Brown jed at jedbrown.org
Sun Feb 2 18:37:13 CST 2014

Satish Balay <balay at mcs.anl.gov> writes:

> On Sun, 2 Feb 2014, Jed Brown wrote:
>> Satish Balay <balay at mcs.anl.gov> writes:
>> > We are not bundling it. We are using it for our nightly tests.
>> But are we recommending that people use it?
> On windows I recommend based upon user requirements - and my
> familiarity in that situation.
> With this old version of mpich - if it simply works when other
> alternatives have issues - then I'll recommend users to use it.

That's reasonable as a stop-gap measure, but I don't think we want to be
in the business of recommending that people build stacks out of
unsupported software.

> Yes - we should fix petsc configure to work better with MSMPI
> - rework all MPI tests to always use mpi.h
> - rework package.py to use 'Inc' as opposed to 'include' for include dir.
> [even then - if one is using old MS compilers on windows - the old
> mpich binary might be easier to use]

A stop-gap, but we should find something supported to recommend,
otherwise we're implicitly taking responsibility when it doesn't work.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140202/d6f6895b/attachment-0001.pgp>

More information about the petsc-users mailing list