[Mpi-21] Relevance of C++ MPI binding of www.boost.org

[Jeff Squyres]

On Apr 3, 2008, at 12:09 PM, Rolf Rabenseifner wrote:

> Is there any knowledge of whether in real C++ programming, whether
> the official C++ MPI binding or the C++ boost MPI binding is
> more often used in the C++ community?

Certainly, the boost bindings offer a much richer interface.

However, I've certainly seen at least some users use the C++ bindings,  
if for nothing other than being able to use return values, such as:

   int rank = COMM_WORLD.Get_rank();

I've typically seen "C+" styles -- mixing the simple C++ constructs  
with C (or mostly C) applications.  There are a few other advantages  
to the C++ bindings, too (e.g., references), but nothing that screams  
"you must use me!" to C++ programmers.  That was actually the Forum's  
specific intent.

Please remember (as has been mentioned several times already): the  
Forum explicitly decided not to standardize a rich C++ class library.   
The decision was specifically made to have C++ bindings that were a  
[mostly] 1-to-1 relationship with the language neutral bindings.  If  
we had standardized a class library, it would have different semantics  
than the language neutral bindings, and therefore effectively be a  
whole second MPI standard.

> For me, the answer may have implications on how separate or  
> integrated additional bindings should be integrated into the  
> language independent text of the MPI standard.

I don't quite understand.  All officially-supported language bindings  
should be listed consistently in the standard.  In MPI-2.1, for  
example, that means alongside the language neutral bindings in the  
text and in Annex A.


-- 
Jeff Squyres
Cisco Systems