It seems like this should be a field in ompi_info output, too... Maybe a component can effect a "license" m4 variable, or somesuch...?
On Oct 5, 2011, at 1:14 PM, hje...@osl.iu.edu wrote: > Author: hjelmn > Date: 2011-10-05 13:14:24 EDT (Wed, 05 Oct 2011) > New Revision: 25234 > URL: https://svn.open-mpi.org/trac/ompi/changeset/25234 > > Log: > first cut at general pmi check > Added: > trunk/orte/config/orte_check_pmi.m4 > > Added: trunk/orte/config/orte_check_pmi.m4 > ============================================================================== > --- (empty file) > +++ trunk/orte/config/orte_check_pmi.m4 2011-10-05 13:14:24 EDT (Wed, > 05 Oct 2011) > @@ -0,0 +1,59 @@ > +# -*- shell-script -*- > +# > +# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana > +# University Research and Technology > +# Corporation. All rights reserved. > +# Copyright (c) 2004-2005 The University of Tennessee and The University > +# of Tennessee Research Foundation. All rights > +# reserved. > +# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, > +# University of Stuttgart. All rights reserved. > +# Copyright (c) 2004-2005 The Regents of the University of California. > +# All rights reserved. > +# Copyright (c) 2009 Cisco Systems, Inc. All rights reserved. > +# Copyright (c) 2011 Los Alamos National Security, LLC. All rights > +# reserved. > +# $COPYRIGHT$ > +# > +# Additional copyrights may follow > +# > +# $HEADER$ > +# > + > +# ORTE_CHECK_PMI(prefix, [action-if-found], [action-if-not-found]) > +# -------------------------------------------------------- > +AC_DEFUN([ORTE_CHECK_PMI],[ > + AC_ARG_WITH([pmi], > + [AC_HELP_STRING([--with-pmi], > + [Build PMI support (default: no)])], > + [], with_pmi=no) > + orte_enable_pmi=0 > + > + AC_MSG_CHECKING([if user requested PMI support]) > + AS_IF([test "$with_pmi" = "no"], [ > + AC_MSG_RESULT([no]) > + orte_want_pmi_support=no],[ > + AC_MSG_RESULT([yes]) > + AS_IF([test ! -z "$with_pmi" -a "$with_pmi" != "yes"],[ > + AS_IF([test -d "$with_pmi/lib64"], LDFLAGS="$LDFLAGS > -L$with_pmi/lib64", > + LDFLAGS="$LDFLAGS -L$with_pmi/lib") > + CFLAGS="$CFLAGS -I$with_pmi/include"],[]) > + > + orte_want_pmi_support=yes > + orte_have_pmi_support=no > + AC_MSG_CHECKING([if PMI support installed]) > + AC_CHECK_HEADERS([pmi.h > slurm/pmi.h],[orte_have_pmi_support=yes] [break],[],[AC_INCLUDES_DEFAULT]) > + AC_CHECK_LIB([pmi],[PMI_Init],[],orte_have_pmi_support=no) > + AS_IF([test "$orte_have_pmi_support" = "yes"], [ > + AC_MSG_RESULT([yes]) > + AC_MSG_WARN([PMI SUPPORT HAS BEEN INCLUDED - RESULTING]) > + AC_MSG_WARN([BINARIES ARE SUBJECT TO ADDITIONAL > LICENSING]) > + AC_MSG_WARN([RESTRICTIONS - SEE THE SLURM LICENSE FOR > INFO]) > + orte_enable_pmi=1] [$2],[ > + AC_MSG_RESULT([no]) > + AC_MSG_WARN([PMI support requested (via --with-pmi) but > not found.]) > + AC_MSG_ERROR([Aborting.])] [$3])]) > + AC_DEFINE_UNQUOTED([WANT_PMI_SUPPORT], > + [$orte_enable_pmi], > + [Whether we want PMI support]) > +]) > _______________________________________________ > svn-full mailing list > svn-f...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/svn-full -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
