Re: [hwloc-devel] dynamic cpuset_t?

2009-09-25 Thread Samuel Thibault
Hello, Just to give my personal opinion: having another ABI shouldn't hurt if we put the cpuset size in the soname for instance. In comparison, having to use functions that allocate/free cpusets like I've done for the AIX and OSF is a bit tedious (those don't have the choice, they have to preserv

[hwloc-devel] Fwd: [hwloc-svn] svn:hwloc r1000

2009-09-25 Thread Jeff Squyres
Woo hoo -- r1000! (it's the little things in life...) :-) Begin forwarded message: From: Date: September 25, 2009 11:32:16 AM EDT To: Subject: [hwloc-svn] svn:hwloc r1000 Reply-To: Author: jsquyres Date: 2009-09-25 11:32:16 EDT (Fri, 25 Sep 2009) New Revision: 1000 URL: https://svn.open-

[hwloc-devel] nightly tarballs

2009-09-25 Thread Jeff Squyres
Brice / Samuel -- Can you add mpit...@osl.iu.edu to the hwloc-devel mailing list and set it to "nomail"? This is the address where the hwloc nightly tarball results will be sent from. Nightly tarballs are only made if the r number changes since the last nightly tarball was made. Nightly

Re: [hwloc-devel] nightly tarballs

2009-09-25 Thread Brice Goglin
Jeff Squyres wrote: > Brice / Samuel -- > > Can you add mpit...@osl.iu.edu to the hwloc-devel mailing list and set > it to "nomail"? Done. > This is the address where the hwloc nightly tarball results will be > sent from. Nightly tarballs are only made if the r number changes > since the last ni

Re: [hwloc-devel] nightly tarballs

2009-09-25 Thread Jeff Squyres
On Sep 25, 2009, at 1:09 PM, Brice Goglin wrote: > http://www.open-mpi.org/software/hwloc/nightly/ By the way, there's a typo at the above URL ("develpoment"). I don't see what you're referring to... ;-) (thanks) -- Jeff Squyres jsquy...@cisco.com

[hwloc-devel] Create success (hwloc r0.9.1a1r1006)

2009-09-25 Thread MPI Team
Creating nightly hwloc snapshot SVN tarball was a success. Snapshot: hwloc 0.9.1a1r1006 Start time: Fri Sep 25 13:28:01 EDT 2009 End time: Fri Sep 25 13:29:42 EDT 2009 Your friendly daemon, Cyrador

Re: [hwloc-devel] Create success (hwloc r0.9.1a1r1006)

2009-09-25 Thread Jeff Squyres
Cool. I forced a [somewhat arbitrary] commit and made cron on the build machine run early. Everything appeared to work nicely -- r1006 snapshot tarball is now on the web site. Since the hwloc tarballs are MUCH smaller than the OMPI tarballs ;-), "make distcheck" takes a lot less time, an

Re: [hwloc-devel] dynamic cpuset_t?

2009-09-25 Thread Jeff Squyres
Sorry for not replying earlier. If we anticipate changing the fixed-size items now, how about adding some kind of version control in the type name now? Or -- more specifically -- what are the exact ABI goals? I.e., if we change the size of the type, will we just require a recompile? Or w

[hwloc-devel] Create success (hwloc r0.9.1a1r1008)

2009-09-25 Thread MPI Team
Creating nightly hwloc snapshot SVN tarball was a success. Snapshot: hwloc 0.9.1a1r1008 Start time: Fri Sep 25 21:01:04 EDT 2009 End time: Fri Sep 25 21:02:53 EDT 2009 Your friendly daemon, Cyrador