From owner-sc22wg5  Thu Mar 21 21:44:26 2002
Received: from math.jpl.nasa.gov (math.jpl.nasa.gov [137.79.7.57])
	by dkuug.dk (8.9.2/8.9.2) with ESMTP id VAA11378
	for <sc22wg5@dkuug.dk>; Thu, 21 Mar 2002 21:44:21 +0100 (CET)
	(envelope-from vsnyder@math.jpl.nasa.gov)
Received: from math.jpl.nasa.gov (vsnyder@localhost)
	by math.jpl.nasa.gov (8.11.6/8.11.6) with ESMTP id g2LKikw09136
	for <sc22wg5@dkuug.dk>; Thu, 21 Mar 2002 12:44:46 -0800
Message-Id: <200203212044.g2LKikw09136@math.jpl.nasa.gov>
X-Mailer: exmh version 2.4 06/23/2000 with nmh-1.0.4
X-Exmh-Isig-CompType: comp
X-Exmh-Isig-Folder: inbox
To: sc22wg5@dkuug.dk
Subject: Re: Fortran archeology
Mime-Version: 1.0
Content-Type: text/plain
Date: Thu, 21 Mar 2002 12:44:46 -0800
From: Van Snyder <vsnyder@math.jpl.nasa.gov>

Miles Ellis quoted John Backus:

> "My own opinion as to the effect of FORTRAN on later languages and the
> collective impact of such languages on programming generally is not a
> popular opinion.  ...  I now regard all conventional languages (e.g.,
> the FORTRANs, the ALGOLs, their successors and derivatives) as
> increasingly complex elaborations of the style of programming dictated
> by the von Neumann computer.  These "von Neumann languages" create
> enormous, unnecessary intellectual roadblocks in thinking about
> programs and in creating the higher level combining forms required in
> a really powerful programming methodology.  ...  We have come to
> regard the DO, FOR, WHILE statements and the like as powerful tools,
> whereas they are in fact weak palliatives that are necessary to make
> the primitive von Neumann style of programming viable at all.
> 
> ....
> 
> "My point is this: while it was perhaps natural and inevitable that
> languages like Fortran and its successors should have developed out of
> the concept of the von Neumann computer as they did, the fact that
> such languages have dominated the thinking for twenty years is
> unfortunate.  It is unfortunate because their long-standing
> familiarity will make it hard for us to understand and adopt new
> programming styles which one day will offer far greater intellectual
> and computational power."

I tried to interest JPL and NASA in functional languages and data-flow
computers in the late 1980's.  I even went to a one-week class on data-
flow computing, given by Arvind at MIT.  I also went to a meeting on
"High Performance Functional Computing."

Several of the presentations reported on mixed-language programs, in
which the nuts-and-bolts were written in Fortran, and the computational
kernels were written in Sisal. In general, the computational kernels
were written in Sisal because the users got better performance that way.

The nuts-and-bolts parts were written in Fortran because it's difficult
to express "do this, and then do that" in functional languages -- you
use a kludge called a "continuation."  Also, functional languages are
generally nearly brain dead when it comes to input/output.

Dave Cann wrote an article in Communications of the ACM in that era,
reporting that he had converted eight computationally intensive routines
from Fortran to Sisal, and run them on an eight-processor Cray X-MP.  In
general, the Sisal code ran faster.  Given the investment that Cray had
put into optimization, that Sisal was developed and maintained by a
small shoestring-funded team, and that the output of Sisal was C code,
this was a bit of a surprise.

The net result of these successes was that DOE discontinued funding for
Sisal, before Sisal-90 could be completed.

--
What fraction of Americans believe   |  Van Snyder
Wrestling is real and NASA is fake?  |  vsnyder@math.jpl.nasa.gov
Any alleged opinions are my own and have not been approved or disapproved
by JPL, CalTech, NASA, Sean O'Keefe, George Bush, the Pope, or anybody else.