[Mpi-forum] Preprocessor Constants

[N.M. Maclaren]

On Nov 23 2010, Hal Finkel wrote:
>
>> Damn Webmail!  Sorry - please ignore this grobble - it got pasted in by
>> accident, and is not true.  I believe that there are such issues, but
>> they are more subtle than that.
>
>Are you concerned that there might be subtle issues with requiring any
>predefined macro, with _MPI being in the namespace reserved for the
>implementation, or both?

Neither.  And that got pasted in the wrong place, anyway.  I believe that
POSIX does effectively require an implementation to predefine macros that
are not in standard C, but don't have the time to refresh my memory.

In any case, I agree that it's not a problem.  And I agree that there is
no relevance to Fortran.  Whether this scheme is a good idea is less
clear, but I don't see any significant problems with it, in practice.

Regards,
Nick Maclaren.