[MPI3 Fortran] What if Fortran is the primary interface?

[Jeff Squyres]

On Mar 17, 2008, at 4:03 PM, Aleksandar Donev wrote:

>> FWIW, I don't think the 2-level indirection is necessary.
> Depends on the details of the MPI implementation, and the ABI for  
> passing
> arguments.

My point (poorly stated) was that the MPI standard only needs to  
mandate that the type MPI_COMM needs to exist (for example).  It  
doesn't need to standardize what it is in it, nor whether a 2nd level  
of indirection is necessary.  Some MPI's may want/need that 2nd level,  
others may not -- that should be an implementation-level decision.

That being said, ABI adds a whole new level of complexity here.  For  
ABI reasons, you are likely correct that we *will* need to mandate  
what is in the Fortran MPI_COMM.

But the ABI committee has purposefully excluded the Fortran bindings  
in its first round of discussions (which I personally think is a  
mistake ;-) ).  So unless something changes there, it's probably not  
worthwhile to discuss ABI ramifications on the Fortran bindings.   
Indeed, the semantics of which approach are going to be used for the  
ABI have not yet been decided, so what we decide today (for Fortran  
ABI) may be irrelevant tomorrow.  Just my $0.02...

-- 
Jeff Squyres
Cisco Systems