[Debichem-devel] gromacs build failures

[Nicholas Breen]

On Wed, Apr 18, 2018 at 11:41:59AM +0000, Mark Abraham wrote:
> The recent build failures for GROMACS 2018.1-1+b1 seem to be only with the
> openmpi configuration (and not mpich? is there a way I can check that more
> efficiently than browsing the log files?), which I suggest we suppress (as
> I think we did for GROMACS 2018).

It's only with OpenMPI - the order of the Makefile first builds the non-MPI
binaries, then MPICH, then OpenMPI, so we always know it's an OpenMPI-specific
fault when it fails in that target.

It *was* building for me when I tested locally on some of the earlier OpenMPI 3
prereleases, and I'm not sure yet what changed.  Let me poke at it a bit.  It
looks like it's a GTest death test issue again.



-- 
Nicholas Breen
nbreen at debian.org