Since the wording appears to address me, I will briefly respond, but I see
the possibility that some parts are actually addressed to the forum.
I'm not sure that I would say that MPI1 was not designed correctly. But
when pressed by Richard to point out ways in which it could be made simpler,
I responded with information that is available to me today, rather than
that which was available when MPI1 was in its initial stages of design. In
other words, I was trying to answer the question "How can it be made simpler"
rather than "How could it have been made simpler". I didn't feel it proper to
restrict my answer based on the practicality of having the forum entertain
changes at this point, since I have neither the authority nor the information
to decide this practicality.
I hope it is clear that my comments and suggestions are being made with the
benefit of 20-20 hindsight, and that the efforts of the MPI forum have made
this possible, to a large extent. I hope it is also clear that CDS has not
reached a stage of development yet to assure that it is superior to MPI at
all, though I personally believe that it has some advantages. I was primarily
using it as a data point for comparison and illustration.
Being no longer actively involved in the forum, I cannot be involved in
deciding discussion topics for the forum. I did agree earlier with the
possibility that this may not be fruitful debate for the MPI forum at this
point, when this was suggested by Al Geist, but I have also seen some other
views, so I remain neutral.
Thank you for the response,
-Dave
========================================================================
David C. DiNucci | MRJ, Inc., Rsrch Scntst | NASA Ames Rsrch Ctr
dinucci@nas.nasa.gov| NAS (Num. Aerospace Sim.)| M/S T27A-2
(415)604-4430 | Parallel Tools Team | Moffett Field, CA 94035