[Mpi-forum] Changelog question
Bronis R. de Supinski
bronis at llnl.gov
Thu Apr 21 04:59:54 CDT 2011
> Normally, the routine names of the new routines should be mentioned.
I do not agree that this rule is appropriate. If broad
concepts are added then you mention the concept, not every
routine that was added for it. As a primary example, if
we adopt the MPIT proposal then we only need to refer to
the concept. We do not need to list each routine of which
the new concept consists (it is obvious when one reviews
the new standard that any MPI_T_ routine was added).
In this specific instance, Jeff's proposed text is certainly
specific enough. I agree with Jeff that we might want to be
less specific; we certainly do not need to add the name of
every new routine that accepts MPI_Count.
> They automatically go also into the Index.
> Taking the last position in the index, normally one can see
> with which MPI version, this routine was introduced,
> if it is newer than MPI-2.0.
> In your case, I would also say a word,
> - that several inquiry functions are modified to return
> MPI_UNDEFINED in the case of too large numbers
> (again the names should be fully listed),
> - and that other the existing routines are
> not modified as long they are internally able to handle
> large messages.
> Please have a look at the existing change-log items.
> ----- Original Message -----
>> From: "Jeff Squyres" <jsquyres at cisco.com>
>> To: "MPI Forum list"
>> Sent: Wednesday, April 20, 2011 11:07:30 PM
>> Subject: [Mpi-forum] Changelog question
>> Rolf --
>> Per your comments, we made up a changelog entry for the MPI_Count
>> ticket. MPI_Count isn't *too* large, but it does touch a lot of
>> individual places.
>> Here's how our changelog entry rendered in PDF; is this generally what
>> you had in mind?
>> Section 2.5.8 on page 16, Section 3.2.2 on page 29, Section 3.3 on
>> page 31, Section 4.1.5 on page 97, Section 4.1.7 on page 99, Section
>> 4.1.8 on page 101, Section 4.1.11 on page 105, Section 5.9.2 on page
>> 172, Section 12.3 on page 412, Section A.1.1 on page 545,
>> Introduce MPI_Count, a type large enough to represent element counts
>> in memory, file views, etc., and associated functions.
>> Or should we be more brief and not list every individual section/page
>> that we touched? I could see this getting problematic for some other
>> tickets (e.g., Fortran).
>> Jeff Squyres
>> jsquyres at cisco.com
>> For corporate legal information go to:
>> mpi-forum mailing list
>> mpi-forum at lists.mpi-forum.org
> Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner at hlrs.de
> High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530
> University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832
> Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner
> Nobelstr. 19, D-70550 Stuttgart, Germany . (Office: Allmandring 30)
> mpi-forum mailing list
> mpi-forum at lists.mpi-forum.org
More information about the mpi-forum