X-Mailer: QUALCOMM Windows Eudora Version 5.1
Date: Wed, 24 Oct 2001 13:40:45 -0500
To: mpi-21@XXXXXXXXXXXXX
From: William Gropp <gropp@XXXXXXXXXXX>
Subject: Re: next ballot
Cc: mpi-21@XXXXXXXXXXXXX
In-Reply-To: <20011023202231.A453536@XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX>
References: <200110072007.PAA45502@XXXXXXXXXXX>
 <200110072007.PAA45502@XXXXXXXXXXX>Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"; format=flowed
Sender: owner-mpi-21@XXXXXXXXXXXXX
Precedence: bulk
Reply-To: mpi-21@XXXXXXXXXXXXX

At 08:22 PM 10/23/2001 +0200, Jesper Larsson Traeff wrote:

Has the voting started? If so (and even if not):

---

NEC agrees to all clarifications of Oct 7, 2001 (1-17), but with
the following suggestion/question:

Item 7:

Drop rationale (or at least the last sentence, "Extentding it...")

I'd rather leave in the rationale (more information is better than less), though I'm open to alternate wordings.


Drop the preceeding lines 6-10 on p. 120 (they are confusing, and seem
to say that with MPI_Accumulate the predefined op's can also be used with
certain user-defined datatypes. This is probably not intended!?)

It was intended that MPI_Accumulate be used with predefined ops on derived datatypes.  For example, an indexed datatype with MPI_REPLACE allows MPI_Accumulate to perform a scatter operation (with different atomicity than MPI_PUT).  Dropping lines 6-10 would significantly change the standard as written.

Bill