On 15 October 2008 at 00:43, Sylvestre Ledru wrote: | | > at the top of the Makefile under "A word of explanation"). The fact that the | > shared libraries aren't present, when using "--enable-static" for configure, | > seems to cause problems for dh_install in the dpkg-buildpackage process. | > After the builds are done, via dpkg-buildpackage, the process gets to a | > point where: | > | > dh_install -s --sourcedir=/usr/local/src/openmpi-1.2.7/debian/openmpi | > --list-missing | > | > is executed and it fails with: | > | > cp: cannot stat | > `/usr/local/src/openmpi-1.2.7/debian/openmpi/usr/lib/openmpi/lib/libmca_common_sm.so.0.0.0': | > No such file or directory | What is the behaviour we prefer here ? | | Here is the full comment from the Makefile: | | # This library is linked against various MCA components because all | # shared-memory based components (e.g., mpool, ptl, etc.) need to | # share some common code and data. There's two cases: | # | # 1. libmca_common_sm.la is a shared library. By linking that shared | # library to all components that need it, the OS linker will | # automatically load it into the process as necessary, and there will | # only be one copy (i.e., all the components will share *one* copy of | # the code and data). | # | # 2. libmca_common_sm.la is a static library. In this case, it will | # be rolled up into the top-level libmpi.la. It will also be rolled | # into each component, but then the component will also be rolled up | # into the upper-level libmpi.la. Linkers universally know how to | # "figure this out" so that we end up with only one copy of the code | # and data. | # | # Note that building this common component statically and linking | # against other dynamic components is *not* supported!
Seems sensible enough. | I commited a solution in our SVN, I just remove this library from the | libopenmi1 package and import *.a & *.la. | | As it is said in the Makefile, it will work but we are duplicating this | code in each component. | However, /usr/lib/openmpi/lib/libmca_common_sm.so.0.0.0 is only 6k... | We could go this way. | | Any objections ? | If not, I will upload it tomorrow. I'd say go for it, especially if you can run one set of tests from an Open MPI 'consumer' -- probably scilab in your case ? -- and see that everythings still works. Thanks for doing this, and so quickly to boot! Dirk -- Three out of two people have difficulties with fractions. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
