From Keith.Bierman@eng.sun.com Mon Dec 2 21:11:50 1991 Received: from Sun.COM by dkuug.dk via EUnet with SMTP (5.64+/8+bit/IDA-1.2.8) id AA18844; Mon, 2 Dec 91 21:11:50 +0100 Received: from Eng.Sun.COM (zigzag-bb.Corp.Sun.COM) by Sun.COM (4.1/SMI-4.1) id AA26256; Mon, 2 Dec 91 12:08:54 PST Received: from chiba.Eng.Sun.COM by Eng.Sun.COM (4.1/SMI-4.1) id AA07266; Mon, 2 Dec 91 12:06:35 PST Received: from localhost by chiba.Eng.Sun.COM (4.1/SMI-4.1) id AA17399; Mon, 2 Dec 91 12:06:30 PST Message-Id: <9112022006.AA17399@chiba.Eng.Sun.COM> To: Lawrie Schonfelder Cc: SC22/WG5 members Subject: Re: (SC22WG5.27) PCF Parallel extensions In-Reply-To: Your message of "Fri, 29 Nov 91 15:58:46 GMT." <9111291737.AA20579@danpost2.uni-c.dk> Date: Mon, 02 Dec 91 12:06:30 PST From: Keith.Bierman@eng.sun.com X-Charset: ASCII X-Char-Esc: 29 >it is done by the mechanism of proposal to WG5 and is processed vis the It is my understanding that the formation of X3H5 was quite controversial even within ANSI. However, must all national bodies do work for international standards only? How can we expect to stop people from doing things like MIL STD 1753? Whether or not ANSI acts as parent, if enough folks want such a thing, they shall meet and declare it. If enough users want it, vendors will build it. Let's not try to play King Canute ;> While I argued against the language extension solution at X3H5; X3J3 discussed this at two meetings (at least) and both times the consensus was that X3H5 was right, and that language semantics was a sensible approach (not judging the actual content of the X3H5 work as of yet; it changes a lot from the start of a meeting to the end). >proposed. There is a a strong embedding of a particular architectural model >into the thinking of most of these suggestions, viz. a shared-memory For better or worse, this is actually quite explicit. X3H5 has three documents being worked on in parallel: 1) Model document 2) Fortran binding to the model 3) C binding to the model The majority of X3H5 seem to feel that it is important to have a conceptual model for the programmer to "hang a hat on". X3H5 seems to feel that the model (shared memory systems) is commerically important enough to fret about. Some vendors have expressed a counter-concern, namely that most of the benefit can be had via advances in compilation technology, so that making all the nuts and bolts programmer visible is probably a mistake. However, X3H5 has not yet abandoned the effort. >type into Fortran. We should be trying to reduce the degree of architecture >lock-in in the language. The good bits of F90 were an attempt to reduce the In general, I agree. However, the oft repeated counter argument is that performance is very important to a large (enough?) group of users and that these people will code to the machine. Therefore we would be better off with one standard way to doing (arguably silly) things than many. The next X3H5 meeting takes place tomorrow morning. I shall place your comments "on the table" (unless you would prefer otherwise). If there is interest, I can post an informal meeting report upon my return. Best Regards, khb