From: Rolf Rabenseifner <Rabenseifner@XXXXXXXXXXXXXXXXXXXX>
Subject: Re: MPI-2 standard issues
To: Nicholas.Nevin@XXXXXXXXXXXX
Date: Fri, 9 Feb 2001 16:31:49 +0100 (CET)
Cc: mpi-core@XXXXXXXXXXX
In-Reply-To: <20010209095646.C20056@XXXX>
X-Mailer: ELM [version 2.5 PL3]
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: owner-mpi-core@XXXXXXXXXXXXX
Precedence: bulk

> 2. Why the restriction that types obtained from calls to
> MPI_Type_create_f90_{real,complex,integer} cannot be freed? 

Without commenting your further arguments, I can only tell the
(historic) reason: 
MPI_Type_create_f90_{real,complex,integer} does not create 
an derived datetype. These routines only return a predefined datatype
(like MPI_INT or MPI_INTEGER). 
MPI-2, p.295, line 32 "They cannot be freed;" is based on the idea,
that is not allowed to free basic datatypes.
MPI 1.1, p.8, line 36 "Such objects [predefined opaque objects]
                       may not be destroyed."

This should not say that Bill's comment may be reasonable.
 
Best regards
Rolf 
 
> This restriction doesn't appear to serve any useful purpose and the
> requirement that MPI_Type_get_envelope return the precision/range used
> to create such a type seems to either require an implementation to
> maintain a potentially very large table or allocate dynamic storage
> (which cannot be freed) to store the precision/range.
> 
> 
> Thanks, Nick.
> 



Dr. Rolf Rabenseifner                      High Performance Computing
Parallel Computing                         Center Stuttgart    (HLRS)
Rechenzentrum Universitaet Stuttgart (RUS) Phone:    ++49 711 6855530
Allmandring 30                             FAX:      ++49 711 6787626
D-70550 Stuttgart                   rabenseifner@XXXXXXXXXXXXXXXXXXXX
Germany                        http://www.hlrs.de/people/rabenseifner