Re: [OMPI devel] RFC: Add SunStudio/Libtool helper script for post-configure
* Ethan Mallove wrote on Mon, Nov 24, 2008 at 09:27:28PM CET: > On Sun, Nov/23/2008 09:19:12AM, Ralf Wildenhues wrote: > > > > While I suppose your patch works, I think in similar situations, OpenMPI > > has resorted to patching input files to configure (like aclocal.m4 or > > ltmain.sh). Search autogen.sh for instances of 'patch'. > > I think I got it. I patch libtool.m4 in autogen.sh after libtoolize > creates libtool.m4. How is the patch now? This is what I had in mind, yes. > > (This is merely an observation; I am not an OpenMPI developer.) Cheers, Ralf
Re: [OMPI devel] gather-bug reminder/re-post
On Nov 24, 2008, at 1:34 PM, Peter Kjellstrom wrote: I originally posted this to the users list maybe it should have gone here. I talked to Jeff Squyres about this bug at SC08 and he thought it would be nice to have it fixed before 1.3. The bug probably is in ompi/datatype/ dt_copy.c:ompi_ddt_copy_content_same_ddt unless I misunderstood Jeff :-) Original post: http://www.open-mpi.org/community/lists/users/2008/11/7318.php Sorry for the delays; almost no real work got done by many of us last week because of SC08. But I spoke to George about this issue last Thursday afternoon, and now I see that George committed r20034 that looks like it's a fix for this issue: https://svn.open-mpi.org/trac/ompi/changeset/20034 Can you give the OMPI trunk a whirl and see if this fixes the problem for you? -- Jeff Squyres Cisco Systems
Re: [OMPI devel] gather-bug reminder/re-post
On Tuesday 25 November 2008, Jeff Squyres wrote: > On Nov 24, 2008, at 1:34 PM, Peter Kjellstrom wrote: > > I originally posted this to the users list maybe it should have gone > > here. I > > talked to Jeff Squyres about this bug at SC08 and he thought it > > would be nice > > to have it fixed before 1.3. > > > > The bug probably is in ompi/datatype/ > > dt_copy.c:ompi_ddt_copy_content_same_ddt > > unless I misunderstood Jeff :-) > > > > Original post: > > http://www.open-mpi.org/community/lists/users/2008/11/7318.php > > Sorry for the delays; almost no real work got done by many of us last > week because of SC08. No worries, I'm in no hurry I just wanted to make sure it didn't get lost in the post-SC noise. > But I spoke to George about this issue last > Thursday afternoon, and now I see that George committed r20034 that > looks like it's a fix for this issue: > > https://svn.open-mpi.org/trac/ompi/changeset/20034 > > Can you give the OMPI trunk a whirl and see if this fixes the problem > for you? I didn't have the toolchain for an autogen so I took the 20034 patch and applied it to 1.3b2. With the patch 1.3b2 works and without it it fails (ran the exact same commands and only "patch ; make ; make install" in between). So, thanks Jeff and George the bug appears gone as of r20034. Don't forget to push that to relevant branches. /Peter signature.asc Description: This is a digitally signed message part.
Re: [OMPI devel] IOF round 2
Ralph, I've just tried this out in my environment where I was seeing issues (as reported in #1661) and it looks good to me. Now I'm not seeing any drops. Thanks, --brad On Thu, Nov 20, 2008 at 2:25 PM, Ralph Castain wrote: > Hi all > > I believe I have fixed the IOF problems reported by Tim M a week ago > (11/13). The fixes are in a separate Hg branch: > > http://www.open-mpi.org/hg/hgwebdir.cgi/rhc/iof/ > > Before I bring these to the trunk - and eventually migrate them to the 1.3 > branch - would some of you like to run an MTT pass across the branch? I have > run it through some of the tests that reported issues before, and it seems > to be working just fine - but I wanted to offer a chance for people to test > it prior to bringing it back into the trunk. > > Please let me know if/when you can do this so I can plan on when to bring > this over to the trunk. If I don't hear from folks by Dec 1, I'll just bring > it into the trunk so it can get tested there. > > Thanks > Ralph > > ___ > devel mailing list > de...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/devel >
Re: [OMPI devel] RFC: merge windows branch into trunk
Hi Ralf, Sorry for replying late. what keeps you from using the autotools-based build system with MSVC? All you should need is a wrapper like cccl. I didn't know this 'cccl' before, but I did give a try last weekend with Cygwin. Unfortunately, I couldn't make it work for Open MPI (latest trunk). I think cccl might work for small projects, but for large and complicated project, it couldn't work easily. By simply replacing compiler name and flags cannot make it go trough compilation and build phase, something has to be done manually with some other scripts. Maybe there's some good way to cope it with Open MPI, if anyone has succeeded with that, please let me know about it, it's very interesting for me anyway. :-) On the other hand, there are more advantages to use CMake on Windows, e.g. easy to use with a GUI, better integration with Visual Studio IDE, automatically produce installer(tarballs) with CPACK, make tests with CTEST, etc. Regards, Shiqing
Re: [OMPI devel] RFC: merge windows branch into trunk
Shiqing, Don't waste your time. While the idea behind cccl is nice, the overhead is unbelievably expensive. As a comparison it took 2 hours to compile Open MPI on Windows using cccl and makefile, while it takes less than 4 minutes to compile exactly the same set of functionalities using the project generated by cmake and the native cl compiler. Using cccl was the original approach that I envisioned for the Windows port. Unfortunately, due to long compilation times, I give up this approach and move to using the native tools (project + microsoft compiler). george. On Nov 25, 2008, at 14:18 , Shiqing Fan wrote: Hi Ralf, Sorry for replying late. what keeps you from using the autotools-based build system with MSVC? All you should need is a wrapper like cccl. I didn't know this 'cccl' before, but I did give a try last weekend with Cygwin. Unfortunately, I couldn't make it work for Open MPI (latest trunk). I think cccl might work for small projects, but for large and complicated project, it couldn't work easily. By simply replacing compiler name and flags cannot make it go trough compilation and build phase, something has to be done manually with some other scripts. Maybe there's some good way to cope it with Open MPI, if anyone has succeeded with that, please let me know about it, it's very interesting for me anyway. :-) On the other hand, there are more advantages to use CMake on Windows, e.g. easy to use with a GUI, better integration with Visual Studio IDE, automatically produce installer(tarballs) with CPACK, make tests with CTEST, etc. Regards, Shiqing ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel
