Re: [OMPI devel] configure --with paths don't allow for absolute path specification

2017-09-02 Thread Phil K via devel
Gilles, Yes, that works.  i see the m4 code that checks for it in config/opal_check_pmi.m4. It's a little deceptive because you will only see the check for /usr/include/slurm belowafter the slurm headers are installed and only if you supply no parameter.  If you do supply a path to --with-pmi

Re: [OMPI devel] configure --with paths don't allow for absolute path specification

2017-09-02 Thread Gilles Gouaillardet
Phil, Did you simply try `--with-pmi` ? According to the configury macros, we look for /usr/include/pmi.h and then /usr/include/slurm/pmi.h, so that should work just fine. If not, can you please compress and post your config.log ? Cheers, Gilles Phil K via devel wrote: >Exactly.  Sorry if m

Re: [OMPI devel] configure --with paths don't allow for absolute path specification

2017-09-02 Thread Phil K via devel
Exactly.  Sorry if my opener was long-winded and obscure.  You can see below OMPI finds the pmi libraries, but misses the headersbecause it's adding "/include" to my path which I don't want it to do. ./configure --with-pmi=/usr/include/slurm --with-pmi-libdir=/usr checking if user requested PMI

Re: [OMPI devel] configure --with paths don't allow for absolute path specification

2017-09-02 Thread r...@open-mpi.org
Okay, so this has nothing to do with the internal pmix or the pmi-1/2 headers it provides, which is what confused me. You are building the SLURM pmi support for OMPI, which does indeed use the slurm-provided headers and pmi libraries. Someone can take a look at that as it should check first in t

Re: [OMPI devel] configure --with paths don't allow for absolute path specification

2017-09-02 Thread Phil K via devel
The issue is getting through the OMPI configure without error which you   _cannot_ when using --with-pmi=/usr/include/slurm if pmi.h and pmi2.h are installed *only* in /usr/include/slurm. On Saturday, September 2, 2017 9:55 AM, "r...@open-mpi.org" wrote: I’m honestly confused by this

Re: [OMPI devel] configure --with paths don't allow for absolute path specification

2017-09-02 Thread r...@open-mpi.org
I’m honestly confused by this as I don’t understand what you are trying to accomplish. Neither OMPI nor PMIx uses those headers. PMIx provides them just as a convenience for anyone wanting to compile a PMI based code, and so that we could internally write functions that translate from PMI to the

Re: [OMPI devel] Open MPI 2.1.2rc3 available for testing

2017-09-02 Thread Gilles Gouaillardet
Marco, can you please detail how you built Open MPI ? i guess you downloaded a tarball and built from that. in this case, there is no need to run autogen.pl --force and unless something is wrong with the timestamps of the tarball, autoreconf should never be invoked. Cheers, Gilles On Sat, Se

Re: [OMPI devel] Open MPI 2.1.2rc3 available for testing

2017-09-02 Thread Marco Atzeri
On 30/08/2017 22:48, Howard Pritchard wrote: Hi Folks, Open MPI 2.1.2rc3 tarballs are available for testing at the usual place: https://www.open-mpi.org/software/ompi/v2.1/ Fixes since rc2: Issue #4122: CMA compilation error in SM BTL.    Thanks to Paul Hargrove for catching this. Issue #4