Bug 84894 - [F2018] provide iso_fortran_binding.h

[Jeff Hammond]

On Thu, Mar 15, 2018 at 9:31 PM, Steve Kargl <
sgk@troutmask.apl.washington.edu> wrote:

> On Thu, Mar 15, 2018 at 07:45:24PM -0700, Jeff Hammond wrote:
> >
> > Is it your intent to filibuster this feature and prevent
> > gfortran from supporting MPI-3 or do you want to have a
> > constructive conversation where we talk about things that
> > benefit users?
>
> I don't speak for gfortran/gcc development or developers.
>
> I am not preventing a feature.  Anyone, including you, are
> more than welcomed to submit a patch.  iso_fortran_binding.h
> won't be in gcc 8.1 as it is too late in the development
> cycle.  I also predict, that unless Anyone, including you,
> step up to plate that this header won't appear for a very
> long time in gfortran.  It's quite simple.  Too many bugs
> and features to implement, too few developers.
>

Have you analyzed Tobias' patch from 2013 that implements this?  What
issues did you find?

https://gcc.gnu.org/viewcvs/gcc?view=revision&revision=198447


> I simply find it odd that MPI 3.1 implemented a module
> named mpi_f08 that uses features not included in the
> F2008 Standard.
>

ISO Fortran committee members were intimately involved in the development
of the mpi_f08 module.  You are welcome to question them about this.  I
honestly don't remember any debate about the name.  The MPI Forum does not
bikeshed names.

I fix bugs in gfortran as a hobby.  29 bugs since the
> beginning of the year.  The oldest was 10 years old.  The
> newest bug was 9 days old.  I've reviewed a number of
> patches from other contributors as they have reviewed
> my patches.
>

I sincerely appreciate your effort.  While I'm no longer dependent on
gfortran, many users of the applications I develop are, and it is essential
to the HPC enterprise that we have a good free Fortran compiler.


> As to filibustering, I have no intension of implementing
> iso_fortran_binding.h.   If Anyone, including you or
> someone associated with OpenCoarray, implements
> iso_fortran_binding.h before close of stage 4 for gcc
> 8.1, I will review the patch and help shepard it into
> the tree.
>

I'll likely contribute to testing it.  I'm hoping the OpenCoarrays team
will do the real work of implementing it.


> The ball is in your court.
>

A link to a six year-old implementation of the desired feature was included
in the first email in this thread.  The next step in the process is for
people to review that and describe the changes required for it to be
merged.  I can only assume that there was a reason why it was never merged,
but I don't have that information.

Jeff

-- 
Jeff Hammond
jeff.science@gmail.com
http://jeffhammond.github.io/