On Mon, 2010-01-04 at 20:21 +0100, Lucas Nussbaum wrote: > On 04/01/10 at 13:25 -0500, Adam C Powell IV wrote: > > I propose libmpif77.so and libmpif90.so slaves to the mpi alternatives > > symlink which link to libfmpich.so, libmpi_f77.so, libmpi_f90.so etc. > > Isn't it more appropriate to use mpif77/mpif90 (which are managed using > alternatives) to compile, instead of specifying the library to link to > manually? This should work for Fortran (and probably C) objects. But if they are mixed Fortran/C++/C, will all mpif77/mpif90 implementations inspect the object files and properly link to all of the necessary language libraries? I'm pretty sure they don't, which is why many packages have MPI_LIBS make variables to include them. If you're sure that all mpif77/mpiCC/etc. implementations do this right, then feel free to close this bug. -Adam -- GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6 Engineering consulting with open source tools http://www.opennovation.com/
Attachment:
signature.asc
Description: This is a digitally signed message part