[MOAB-dev] [PATCH] Call MPI_Init if using MPI.

Jed Brown jed at 59A2.org
Wed Oct 28 17:23:08 CDT 2009


Jed Brown wrote:
> PETSc's stubs must have been getting included inadvertently, there are
> no symbols because they are just macros.  An alternative is to
> namespace the stubs appropriately.

It's not possible to implement everything as macros.  PETSc only
namespaces the non-macro part of MPIUNI if MPIUNI_AVOID_MPI_NAMESPACE is
defined.  Perhaps that was the issue.

Note that having MPI_Comm in the serial interface is usually okay
because a serial MPI will never reference it (therefore it's okay if the
value gets corrupted because extra bits are ignored).  Of course if you
have a parellel code calling a serial PETSc, you can't rely on the
serial PETSc handing you back valid communicators.

Jed

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 261 bytes
Desc: OpenPGP digital signature
URL: <http://lists.mcs.anl.gov/pipermail/moab-dev/attachments/20091028/294f72ad/attachment-0001.pgp>


More information about the moab-dev mailing list