On Sep 18, 2010, at 4:54 PM, Ralph Castain wrote: > Good point - nothing secret about it!
Yes, nothing secret, but it's kinda deep in our ompi_mca.m4 file. I'm not quite sure why Brian removed the AS_IF clause when we did the changes; I don't know if he had a deep reason for it or not. But it seems like we clearly need it because the memcheck framework tries to disable itself by setting $DISABLE_memcheck=1 if you don't --enable-memcheck. The new ompi_mca.m4 always calls MCA_$1_$2_$3_CONFIG (i.e., the component's config macro in its configure.m4 script) -- the old version checked $should_build first before calling it. I put the AS_IF back in the patch below. > Index: config/ompi_mca.m4 > =================================================================== > --- config/ompi_mca.m4 (revision 23770) > +++ config/ompi_mca.m4 (working copy) > @@ -564,10 +564,12 @@ > > # try to configure the component. pay no attention to > # --enable-dist, since we'll always have makefiles. > - m4_ifdef([MCA_$1_$2_$3_CONFIG], > + AS_IF([test "$should_build" = "1"], > + [m4_ifdef([MCA_$1_$2_$3_CONFIG], > [MCA_$1_$2_$3_CONFIG([should_build=$should_build], > [should_build=0])], > [m4_fatal([MCA_$1_$2_$3_CONFIG macro not found])]) > + ]) > > AS_IF([test "$should_build" = "1"], > [MCA_PROCESS_COMPONENT($1, $2, $3, $4, $5, $6, $7, $compile_mode)], > @@ -619,7 +621,7 @@ > if test "$should_build" = "1" ; then > OMPI_CONFIG_SUBDIR([$1/mca/$2/$component], > [$ompi_subdir_args], > - [should_build=1], [should_build=2]) > + [should_build=1], [should_build=0]) This just looked like a typo. -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
