Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!clyde.concordia.ca!thunder.mcrcim.mcgill.edu!snorkelwacker.mit.edu!apple!mips!sdd.hp.com!wuarchive!emory!hubcap!Vaughan
From: pratt@cs.stanford.edu (Vaughan Pratt)
Newsgroups: comp.parallel
Subject: Linda lessons
Message-ID: <12495@hubcap.clemson.edu>
Date: 7 Jan 91 13:47:43 GMT
Sender: fpst@hubcap.clemson.edu
Lines: 45
Approved: parallel@hubcap.clemson.edu

Linda has two lessons to teach:

1.  It builds experience and intuition with parallel programming.  I'm
sorry I wasn't more involved with Linda myself to benefit from those
lessons, I've been too involved with other stuff.

2.  The other stuff is what you do after you realize the following.

	Linda went along for years with an early 1960's-style
	"definition" with "obvious" semantics that no one ever bothered
	to write down - until, of course, what became "obvious" was
	that different readers (and different implementors) where
	finding different "obvious" interpretations.)"

The question of the meaning of concurrency hit me hard in January 1980
when I was a naive associate professor at MIT.  At that time I was
satisfied with both the meaning of sequential computation and the
complexity of concurrency, both of which I'd previously written some of
my best papers on.  Michael Dertouzos asked me how I would integrate
workstation software so that it all makes sense; I think he didn't like
the way MIT's Nu Terminal interface felt compared to the Alto's.  I
also think he was hoping for an answer in *well* under ten years, and
I'm *sure* he wasn't thinking about the kind of answer I had in mind.

It's lucky I didn't realize the problem was so hard when I started work
on it, though maybe if I hadn't taken time off to design a logo and
graphics software for a small workstation startup in 1972 I'd be
further along this road.  But ideas are like babies: you can't speed up
their gestation, and by the same token you can't slow it down by
working at a mundane job for a while or abortion clinics would be out
of business.

I think I have a much better answer today than I did three weeks ago.
However it is not *the* answer, though it's a very nice one so far.
More work is needed concerning what concurrency really is.

Meanwhile let advertise again Boole.Stanford.EDU's /pub directory for
anyone who feels like browsing around in an eccentric professor's mind
to see what he thinks true concurrency is in Jan. 1991.  The latest
paper (3 days old or so) is catl.{tex,dvi}.  Catl rustling strongly
encouraged.  I will be adding a few paragraphs on the connection with
linear algebra in a few hours, so if you hurry now you won't have to
have your nice clean copy sullied with such rubbish.

Vaughan Pratt