Mail Archives: opendos/2000/10/29/16:24:35
Hi,
We seem to periodically run into the history of operating systems in
this list, and it saddens me that so little is remembered. Perhaps it is
just due to the youth of the contributors (which in itself is a good
thing). So, although this is a bit off topic, I would like to present
the following brief history - as it applies to the PC.
In the beginning there was CP/M. Developed by Gary Kildall in 1974, it
was based on some work he had been doing at Intel for their development
systems (the PL/M language which ran on the Intel o/s called ISIS: I
used it in the late 70's). CP/M was the least operating system you could
have which would allow you to boot from an 8 inch floppy and get a
command prompt. Kildall had also done a lot of work with DEC systems,
and so he borrowed some ideas from them. The most obvious was "PIP"
(Peripheral Interchange Program) which does the job everyone else calls
COPY. Note that here is NO derivation from UNIX here, or even any
indication that Kildall was aware of it. CP/M became hugely popular on
all 8080 and Z80 based personal computers.
Then in 1980 it all started to go wrong. The exact details are shrouded
in the mists of time and the dust of lawyers offices, but the result was
that IBM, unsatisfied with DR's incomplete and behind schedule CPM/86
took up with the company that was supplying their Basic interpreter -
Microsoft. MS purchased rights to a CP/M clone from Seattle Computer
Products for $50,000 and re-worked it along side IBM engineers to
produce PCDOS (the IBM version) and MS-DOS (the generic version). It is
said the MS made nothing much out of the deal with IBM but banked on
there being a hardware clone market to sell to - which was smart or
lucky depending on your world view.
Having entered the o/s market, MS wanted a multiuser offering as well.
In about 1982 (I'm guessing here) they bought a UNIX system 7 license
from Bell Labs and marketed it as Xenix. It wasn't Intel only: I used
Xenix on a PDP-11 in 1984 or there abouts. Here in the UK Xenix was
distributed by Logica. Before long MS decided to get out of that game
and sold the whole thing to SCO who used it to create the x86 port that
we all know. For many years it was practically the only x86 Unix around,
and also had the largest installed base of any Unix flavour. So that is
how the Microsoft copyrights turn up in Xenix code. Just to finish off
this strand quickly, the three main species of Unix: BSD (aka SunOS),
AT&T System V and SCO Xenix were merged together in about 1988. The open
source movement was founded around GNU in the mid 80's and with the
Linux kernel of the 90's provide the GNU/Linux distributions that we
know and love. The latest is that SCO have sold off all the Unixware and
Xenix business to Caldera Systems, so of course there is convergence
between SCO Unix and GNU/Linux.
From 1981 onwards, MS dominated the PC o/s market. Version 1 was shipped
with the original PC. Version 2.0 came with the XT and added a
hierarchical file system which was inspired by Unix, but had no
architectural similarities. It was at this point they chose the "wrong"
sort of slash: '\' instead of '/' which bugs everyone who switches
between both systems to this day. In 1984 came 3.0 to support the PC/AT
and a little later v 3.1 with network support for Microsoft's feeble
first file server. Things start to get interesting again in 1986. In
that year I saw a system called 286DOS (or DOS286 perhaps). MS had
worked out how to switch a 286 from protected mode back to real mode and
so create a system that could run DOS programs as well as new protected
mode programs. They sold it to IBM and in 1987 IBM and MS launched it
together as OS/2. MS was never that committed to OS/2 it now turns out.
In fact I attended a Microsoft briefing in about 1988 when they said
quite plainly that they were going to torpedo OS/2. Which they did in
1990 with Windows 3.0.
Meanwhile, DR's operating system business declined. To compensate they
switched emphasis to GUIs and produced GEM in 1985 ish (this is not my
strong area so I may be out by a year or so). I remember evaluating GEM
and Windows 1.0 in 1986 and thinking that GEM was far better. However DR
managed to snatch defeat from the jaws of victory a second time when
Apple sued them for "look and feel" similarities with the Mac (which was
a bit of a cheek since the Mac is a complete rip off of the Xerox Star).
DR played by the rules, changed the look and feel and lost the market.
MS ignored the rules, won their case against Apple and took the market.
Life is just not fair.
Another episode demonstrates this. The DR operating systems group did
not die (otherwise I would not be writing to this mail list). CP/M 86
became more and more DOS like. Around about 1988 DR created a clone of
MS-DOS 3.3, shipped to a small number of OEM's as DR DOS 3.31 (note no
hyphen; that was added much later by Caldera). Successive OEM versions
were shipped up to v 3.41. Then in 1990 DR made the innovative step of
selling DR DOS 5.0 retail. So far as I know all previous DOS versions
from MS, IBM and DR had been OEM only. MS were caught completely
unawares as people dumped MS-DOS 3 and the horrendous version 4 for DR
DOS 5.0. The motivation was basically the much improved memory
management. It took MS a full year to respond. The MS-DOS development
group had been disbanded and had to be re-built from scratch. However
MS-DOS 5.0 was shipped both retail and OEM, and they began to regain
market share. The unfair part of this story is in the OEM sector, where
DR stood to make a killing: OEM sales are almost pure profit, whereas
with retail you have to ship actual boxes around. MS used a series of
very dodgy practices to block DR from this part of the market (details
too gory to go into here). So for the third time success was snatched
away from them.
Around 1990 Microsoft started work on the other operating system they
always wanted. They recruited the VMS development team from DEC - headed
by Dave Cutler if I remember correctly. This was released in 1993(?) as
"New Technology" Windows, or Windows NT. Hence NT has some similarities
in architecture to VMS, but of course no shared code. NT has been a bit
of a mixed blessing for MS, I think. It has forced them to duplicate a
lot of effort maintaining two o/s strands with a lot of overlap on the
desktop market. And still the DOS based strand refuses to die, because
even Windows ME still has MS-DOS, and therefore a faint echo of CP/M, at
its core.
And that is as much as I am prepared to write in one go. If you managed
to follow me this far, thank you for sticking with it. If there are any
inaccuracies in the above please let me know.
Chris.
- Raw text -