On Apr 23, 2020, at 8:50 AM, Patrick Bégou <patrick.be...@legi.grenoble-inp.fr> wrote: > > As we say in french "dans le mille!" you were right. > I'm not the admin of these servers and a "mpirun not found" was sufficient in > my mind. It wasn't. > > As I had deployed OpenMPI 4.0.2 I launch a new build after setting my > LD_LIBRARY_PATH to reach OpenMPI4.0.2 installed libs before all other > locations and all tests were successfull.
Sweet! Glad you got it worked out. > I think that this should be modified in the test script as we usually run > "make check" before "make install". Setting properly LD_LIBRARY_PATH to reach > first the temporary directory were the libs are built before launching the > test would be enought to avoid this wrong behavior. It's a little hard for us to do this, for a few reasons: - we can't just set LD_LIBRARY_PATH to empty, because we can't know what else is needed from LD_LIBRARY_PATH to run the tests (e.g., network stack libraries, other support libraries / dependencies) - the "make check" infrastructure is wholly provided by GNU Automake. It takes care of setting LD_LIBRARY_PATH to the Right Places in the build tree to point to the just-built libraries. I don't know how we'd hijack / patch that. - it's even fairly tricky to write a negative test (e.g., check to see if the "wrong" libmpi/etc. will be found), because which library(ies) are used change depending on which test is used, and re-creating the .so version number rules in a "look for a match" script could be kinda difficult (and OS-specific). -- Jeff Squyres jsquy...@cisco.com
