It is because MPICH auto-generates manpage's explanation based on error codes tagged on each MPI routine. A generic explanation is applied to all routines with the same error code. So you can even see "May be an uncommitted MPI_Datatype (see MPI_Type_commit)." in manpage of MPI_Type_commit". --Junchao Zhang On Fri, Dec 5, 2014 at 9:16 AM, Junchao Zhang <[email protected]> wrote:

I think it is an oversight. MPI_Type_free in MPICH is ok with uncommitted datatypes. I will correct it. Thanks.

--Junchao Zhang

On Fri, Dec 5, 2014 at 8:04 AM, Marc-Andre Hermanns < [email protected]> wrote:

...

Dear MPICH team,

I just recently stumbled over the MPICH manpage for MPI_Type_commit and MPI_Type_free.

In the description of MPI_ERR_TYPE it states:

"MPI_ERR_TYPE Invalid datatype argument. May be an uncommitted MPI_Datatype (see MPI_Type_commit). "

What is meant by an uncommitted MPI_Datatype? As far as I know, MPI_Type_free should be callable on uncommitted datatypes. So why is there this reference to the commit state?

Cheers, Marc-Andre -- Marc-Andre Hermanns Jülich Aachen Research Alliance, High Performance Computing (JARA-HPC) German Research School for Simulation Sciences GmbH

Schinkelstrasse 2 52062 Aachen Germany

Phone: +49 241 80 99753 Fax: +49 241 80 6 99753 www.grs-sim.de/parallel email: [email protected]

_______________________________________________ discuss mailing list [email protected] To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

_______________________________________________ discuss mailing list [email protected] To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss