Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!uwvax!vanvleck!uwmcsd1!marque!doug
From: doug@marque.UUCP (harris)
Newsgroups: comp.sys.att
Subject: Re: PC 7300: Why now?
Message-ID: <1714@marque.UUCP>
Date: Fri, 26-Jun-87 17:44:42 EDT
Article-I.D.: marque.1714
Posted: Fri Jun 26 17:44:42 1987
Date-Received: Sat, 27-Jun-87 11:40:35 EDT
References: <820@sfmin.UUCP>
Reply-To: doug@marque.UUCP (harris, douglas)
Distribution: na
Organization: Marquette University, Milwaukee, WI
Lines: 79

Larry Geary wonders "why now?" and I'm not sure I can answer to that.
We purchased then (no comments about prices please, at least none
that my Dean can hear).  We purchased two 3B2s the day they went on
the market, a 3B5 (high end model) as soon as we could get it, about
three years ago, a number of PC6300s, and about 20 PC7300s.  AT&T
also donated a year and a half ago a pair of 3B5s.  We are very
happily (despite my earlier posting about maintenance :-) using all
of these, the Unix PCs in faculty offices and in a lab for the
introductory course.  Since they are the "hot" item just now I'll
comment only on them.

Originally, the windows were painfully slow, and the machines hung
for all sorts of unexplained/unexplainable reason to the point that
they had to be rebooted.  The disk was slow and noisy.  We were
attracted to the product initially in part because of the GSS drivers,
as well as the hope that an AT&T machine would quickly be provided
with reasonable communications facilities.  The Unix was, and is,
somewhat bastardized, the documentation does not quite describe
accurately the facts (doesn't it usually ;-?), and the window manager
has the strongest "mind of its own" I have ever seen.

With all of the defects and problems (each new release of the OS
solves some, introduces other - is it an IBM product really?), it
is a 68010 box with its own MML (Mickey Mouse logic) memory management,
well done in the "traditional" 68010 manner, good graphics hardware
(720 by 348 with a "rasterop" in the C library), nice bus
architecture (well, I have a Technical Manual, and fortunately,
an ohmeter, so was able to map out the bus architecture from the
conflicting diagrams in the manual), ST506 disk (brain damaged of
course by the fact that only a single disk can be selected, by
hardware choice), nice modem electronically (don't ask about the
library software that sometimes drives it).  So it is an absolutely
marvelous hacker's machine.

To me the recent "sudden interest" is due to the fact that the
$1200 price frame is one that is reasonable for individuals, the
earlier $3000++ price frame (ask what we paid for ours) was not.

As to why University folks did not purchase them, and why many of
those who were given them have stuffed them into a corner and not
used them in a major way, the "hassle factor" has just been too
high compared to other possibilities.  We have SOURCE CODE for
PDP, Vax, 3B5, 3B2, Minix, Xinu, who knows what else, while for
the Unix PC it took years (literally) to get a hardware manual,
which is not accurate, and the bastardized Unix means that you
can't operate from either your BSD experience or your SysV
experience, and you have NO IDEA what the smgr/wmgr daemons will
cook up and decide to carry out, which will lock your machine,
hang your phone, pop up a window, or generally just hassle you.
Having completed a semester using 10 UnixPcs each with a terminal,
connected via Starlan to themselves and a 3B2 (also on our Ethernet),
as "lab stations" for 15 two hour sessions a week with 20 students
per session, I can cite chapter and verse.

I have been told, by AT&T folks at a reasonably high level, that
"we tried to put Unix on the desktop, and failed".  My rather
violent response, which I am entitled to as "the man who owns one",
and who bought them wanting Unix on the desktop, is that they in
fact did not put UNIX on the desktop, they put a bastardized Unix
underneath and a fouled up "user agent" on top, and did not allow
those who knew what to do with Unix to get underneath it to make
it go.  And all of this with lots of hardbound "easy to use" manuals
that told you nothing accurate in many cases.  Nothing would have
improved the chances of that machine more, in my opinion, than to
have totally "cleaned it out", putting real SysV or BSD, with some
low level communications hooks, and high-level OPTIONAL interfaces
to windows and user agents.  This is the approach that I at least
expected from AT&T (and may yet see if their pockets are truly deep).

Sorry for the length of the posting; three years is a long time to
store up opinions.  I would like to see some actual discussion in
this group of 3Bs, 7300s, and the like (yes  I know about U3G and
can't wait for it to start), what folks are doing with them and where
they want to go, and I can't wait for the first of you hackers who
can now afford them to "clean one up" and turn it into a real
"desktop Unix box".  The hardware deserves it.

Doug Harris (Chairman of the Department and resident unik)
doug@mu.edu
...!uwvax!marque!doug