datatype: Add support for MPI_LOGICAL16

[dalcinl]

Support for MPI_LOGICAL16.

TODO: Support for 128bit integral C types via GCC/LLVM __int128 extension.

[github-actions]

Hello! The Git Commit Checker CI bot found a few problems with this PR:

7b93c9d: datatype: Add support for MPI_TYPECLASS_LOGICAL

• check_signed_off: does not contain a valid Signed-off-by line

64e5f5f: datatype: Add preliminary support for MPI_LOGICAL1...

• check_signed_off: does not contain a valid Signed-off-by line

Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks!

[dalcinl]

I believe I forgot to bump OMPI_DATATYPE_MAX_PREDEFINED by one.

[dalcinl]

@hppritcha You may want review and eventually merge this little one.

[hppritcha]

thanks I was just working on this one but this saves me some time.

[hppritcha]

I think a new MOOG needs to be added in ompi_datatype_module.c for logical16 set to 77. and in mpif-values.py set to 77,

[dalcinl]

I think a new MOOG needs to be added in ompi_datatype_module.c for logical16 set to 77. and in mpif-values.py set to 77,

You are definitely right. PR updated accordingly.

[bosilca]

Please add a comment to justify the gap. leave room for C/CXX specific datatypes should be enough.

[dalcinl]

Comment added. However, this makes me thing that maybe this script is missing the 16bit floating types in the MPIX_ namespace?

From ompi/datatype/ompi_datatype_module.c

    /* Datatypes proposed to the MPI Forum in June 2017 for proposal in
     * the MPI 4.0 standard. As of February 2019, it is not accepted yet.
     * See https://github.com/mpi-forum/mpi-issues/issues/65 */
    MOOG(short_float, 74);
    MOOG(c_short_float_complex, 75);
    MOOG(cxx_sfltcplex, 76);

[bosilca]

I'm not even sure they are equivalent to Fortran's REAL*2. If you want to add support for MPIX you can do it in another PR, this one looks ready.

[hppritcha]

I agree. that's for another day.