I said I didn't have the patience to retest until rc2, but a bit of
scripting has changed that.
It not only gave me the patience to re-test, but to expand my testing to
15 non-linux machines.
I've tested the 1.5.5rc1 tarball on a pile of non-linix systems with 2
fixes in place:
+ Added #includes of sys/types.h and sys/wait.h to
ompi/contrib/vt/vt/extlib/otf/tools/otfprofile/clustering.cpp
+ Deleted
opal/mca/hwloc/hwloc122ompi/hwloc/include/hwloc/autogen/config.h
Both have been the subject of previous posts and I'll not elaborate on
them here.
With those in place, I have been able to "(g)make all install check" on
the following:
MacOS 10.4 on PPC
MacOS 10.5 on PPC
MacOS 10.4 on x86-64
MacOS 10.5 on x86-64
MacOS 10.6 on x86-64
MacOS 10.7 on x86-64
FreeBSD 6.3-RELEASE on amd64
FreeBSD 7.2-RELEASE on amd64
FreeBSD 8.2-RELEASE on amd64
FreeBSD 8.2-RELEASE on x86
NetBSD-5.1 on amd64
NetBSD-5.1 on x86
OpenBSD-5.0 on amd64
OpenBSD-5.0 on x86
Solaris-11 on amd64 (incl. OFUD, OPENIB and UDAPL)
Note that the *BSD platforms all required gmake (in place of the system
default Berkeley Make).
Other than --prefix, the only configure argument used was
"--disable-io-romio" on the BSD platforms.
I have slow Solaris10 boxes (1 each x86 and sparc) that I will test as
time allows.
-Paul
--
Paul H. Hargrove phhargr...@lbl.gov
Future Technologies Group
HPC Research Department Tel: +1-510-495-2352
Lawrence Berkeley National Laboratory Fax: +1-510-486-6900
