First, much thanks to Pat Miller. I was able to get pyMPI going on my first system.
But now I have a second one to worry about. I was able to compile and link, but when I fired up the pyMPI executable, it crashed immediately, telling me that it couldn't find the MPI shared library libmpi.so.1.
Well, it definitely exists, and other codes running MPI on this new platform have not had a problem finding it.
I was able to get past this problem by adding the library location to LD_LIBRARY_PATH. But I don't like doing this. It means asking my users to do one more thing to use this executable, and I understand that it involves some security issues.
Any idea why this would work on one platform and not another similar (both Linux) platform? Most importantly, is there a cleaner solution to this problem than editing LD_LIBRARY_PATH? Thanks!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
First, much thanks to Pat Miller. I was able to get pyMPI going on my first system.
But now I have a second one to worry about. I was able to compile and link, but when I fired up the pyMPI executable, it crashed immediately, telling me that it couldn't find the MPI shared library libmpi.so.1.
Well, it definitely exists, and other codes running MPI on this new platform have not had a problem finding it.
I was able to get past this problem by adding the library location to LD_LIBRARY_PATH. But I don't like doing this. It means asking my users to do one more thing to use this executable, and I understand that it involves some security issues.
Any idea why this would work on one platform and not another similar (both Linux) platform? Most importantly, is there a cleaner solution to this problem than editing LD_LIBRARY_PATH? Thanks!
You can hack the location using -rpath in the final link stage
Something like running ./configure with -with-libs=-Wl,-rpath,/path/to/where/missing/lib/is
Thanks, Pat. Apparently, I'm good to go now. Much obliged!