On Sun, Jul 22, 2007 at 08:54:42PM -0700, David Boles wrote:
> Can you name me one personnel computer that would run then, or will today,
> a true Unix OS?
Sure. Altos 586. Compaq portable ca. 1984. AT&T 3B1. Heck, Dell was
selling full AT&T System V for Intel PCs on a cart tape in 1985 for
$1100--I know, I bought it. Randy Suess took Chinet (the original,
very first BBS) from CP/M to Unix in 1983 or so, and it always ran on
consumer-grade personal computers.
Heck, there was a slew of Unix distros in the early days--AT&T Unix,
BSD, Xenix, Venix, etc. All of them had distros targeted at Intel
microprocessors--what you'd today call personal computers. Heck,
some didn't even have a MMU, but they'd hacked Unix to run on them
(most totally insecure, of course)--Venix is the prime example of this.
I worked in 1981 or so on a port of Unix 3.0 for the 3B Field Test
Set--the first portable computer running Unix (for classes of portable
weighing in at over 30 lbs. and just *barely* meeting FAA requirements to
fit under-seat).
Unix was a renegade OS from the start--its *purpose* was to get away
from the glass house and high priests and put a working OS in the hands
of researchers and developers; of *course* it was targeted at micros as
soon as possible.
> Oh. And since AT&T, you said this I don't know this - actually don't even
> care - owned Unix would you have expected them to give it away?
Give away? No. I would have hoped they would have understood marketing,
and where the computer industry was going, but didn't expect it at all;
I was there, and *knew* how hopeless AT&T was at marketing.
Unix lost the hearts&minds not because of tech issues, but for two reasons:
1. AT&T fully misunderstood marketing, market share, profit margins and
market penetration strategies in the emerging microcomputer world.
2. Business was *desperate* for some standardization in the microcomputer
market.
Item one meant that AT&T treated each distribution of Unix as a
custom-crafted, very expensive contract. The Dell distribution was
an admitted anomaly--enthusiasts at Dell Computer managed to run it
by Michael Dell, and I really don't think he understood what they were
proposing when he OK'd selling a bare OS without hardware. They bundled
a full distribution of AT&T System V Unix with many of the then-new Gnu
utilities and sold it, no strings, for only $1100--an amazingly low price
back then, and I jumped on it like a shot. It cramped my style to have
to work with a binary-only distribution, but it was a full, real Unix OS.
(Michael finally realized what was going on when the support costs started
to show up on the bottom line, and killed it _muy pronto_. I was mad
at him for a long time about that, but I don't think he noticed.)
Item two can't be overstated. Micros had made significant penetration in
business markets while running CP/M; and there's no question that CP/M was
far more robust and mature than DOS would be for several years after its
release.
HOWEVER. The machine independence of CP/M came at too low a level;
individual distributions had unique disk mappings, meaning you couldn't
interchange floppies with another site unless they either had the same
CP/M distribution as you, or you ran the floppy through a converter (I
still have one with my Osborne-1 machines in the basement.) There was
funky and unique diddling necessary to tune parameters for the BIOS
and BDOS halves of the OS, and this even showed up in setting of
parameters for the then-killer app, WordPerfect. And most of these
distributions were from companies that didn't inspire confidence in
corporate America--either they were too new or too geeky, and failed at the
normal rate fro startups.
Enter IBM with the PC. It was clunky, slow, behind the existing CP/M
machines both in hardware and software capabilities. BUT IT WAS IBM.
AND, in a brilliant move, they published **everything** about it--hardware
schematics, BIOS source listings, detailed DOS interface and internals
specs. So a raging foam of companies surged up almost immediately,
providing add-on hardware, software, OS extensions, etc. CP/M, just
making the transition to MP/M, was doomed.
Enough; this has gone off topic (in a topic which is off-topic itself).
But yes, there is a long history of real Unix on PCs.
--
Dave Ihnat
President, DMINET Consulting, Inc.
dihnat@xxxxxxxxxx
