On Apr 15, 2016, at 9:05 AM, DERBEY, NADIA <[email protected]> wrote:
>
> Actually, we have a functional test suite that used to pass for these
> types and it fails now with v2.x. So I thought it was a regression.
Gotcha.
> But from what you're saying, the new standard doesn't have a strong
> requirement on LOGICAL and CHARACTER.
It looks like that definition of "numeric" types for MPI_SIZEOF goes back to
MPI-2.0 (I just checked), which is when MPI_SIZEOF was introduced.
> So I really don't mind, since I
> don't know whether these types are widely used in client applications. I
> guess no?
I doubt that MPI_SIZEOF is used much in client applications, but I don't know
that for a fact.
There is lingering / low priority work in the MPI Forum to get rid of
MPI_SIZEOF altogether because the Fortran "storage_size" intrinsic should
wholly replace this (which didn't exist back when MPI_SIZEOF was created).
...I had to dig to find the old ticket, but it's here:
https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/465
I.e., it didn't migrate over from the Forum old Trac system to the new Forum
Github system, which goes to tell you how much work has occurred on this
ticket... :-)
--
Jeff Squyres
[email protected]
For corporate legal information go to:
http://www.cisco.com/web/about/doing_business/legal/cri/
