Path: utzoo!utgpu!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!samsung!sdd.hp.com!zaphod.mps.ohio-state.edu!mips!daver!ditka!qiclab!m2xenix!randy
From: randy@m2xenix.psg.com (Randy Bush)
Newsgroups: comp.lang.pascal
Subject: Re: UCSD P-system
Message-ID: <1990May27.034028.27700@m2xenix.psg.com>
Date: 27 May 90 03:40:28 GMT
References: <33@usblues.UUCP> <9110017@hpfcso.HP.COM> <milne.642840843@ics.uci.edu> <1990May20.004202.18407@m2xenix.psg.com> <2658603C.12374@paris.ics.uci.edu>
Organization: Pacific Systems Group, Portland Oregon US
Lines: 54

milne@ics.uci.edu (Alastair Milne) writes:
( some smarmy remarks which we can hope were made in jest :-)

Yes, Version IV was mostly SofTech's, though done by many of the same folk
from KB's group.  I have no idea what Eli and the crew in Brooklyn have done
in the interim, though; and would love to hear about changes to the pmachine
architecture if folk on the net happen to know.

The canonical floating point was done at SofTech, to my recollection, and the
byte-sex independence definitely was, and came in with IV.0.  The portability
these provided never was useful in the marketplace, and had a distinct cost in
performance; a lesson there.

I have a rather extensive collection of the architecture and internal
documentation for SCUD [1.3..IV.1].  The history and changes are rather
interesting.

What I would rather have is a copy of Jerry Barron's "Pascal: the Language and
its Implementation", which I did not manage to snag before it went out of
print.  [John Gough, who is visiting, just had the nerve to tell me he picked
up a copy on remainder for $4, the cad!]  If anyone sees a spare copy of this
gem, I would really love to get it.

> SYSTEM.PASCAL has some host-dependent units, like REALOPS, and perhaps one
> or two others I don't remember.

I think you're correct there.  As SofTech's San Diego operation faltered, I
was drifting onward through Violation Systems's implementations which were
more based on enhanced (for Modula-2) II.1 architecture, a la Apple Pascal.

> The fun we could have discussing different p-System versions!  Anybody
> want to mention version III?

For those having Pascal MicroEngines (the III.0 system) in the closet, I just
happen to have the hacks and software to actually make that a usable machine.
Darn shame that WD bungling killed it; for its day it was quite useful.  It
was the first SCUD with tasks, semaphores, and events.  And I think you are
correct in that the interpreter had to handle the ready and semaphore queues,
though the code is not too large (just hard to debug).  The real turkeys in
SCUD interpreter hacking are sets, floating point emulation, and I/O.

Yes, Urs Ammann did the Zuerich P2 compiler.  But the P2 p-code assembler and
interpreter has Kathy Jensen's and NW's names on it.  I actually have source
for all this stuff as well as a collection of the related papers of the time
(all in hardcopy unfortunately).  But I believe that Barron's book is by far
the best overview.

Nice to see actual discussion of that stuff here as opposed to how to get
turbowhatzit (tm Pat Terry) to handle Int42.  I am told that turbowhatzit has
finally discovered UNITs, a decade later, wow.

randy
-- 
..!{uunet,qiclab,intelhf}!m2xenix!randy   randy@psg.com   randy@m2xenix.uucp