[TUHS] Why Linux not another PC/UNIX [was Mach for i386 ...]

arnold at skeeve.com arnold at skeeve.com
Thu Feb 23 06:18:30 AEST 2017

I'll add my two cents. My experience was that source was hard
to get to. My undergrad school had IS/1 - v6 on a PDP-11 - and source
was definitely not available to us seniors. One person who actually
worked for the computer center did have access. This was 1980-1981.

In grad school at Georgia Tech, there was no access to students to
the vax, and even after I went to work there I had to sign something
first before seeing source.

Later I was a sysadmin at Emory U and so I had source but it wasn't
widely open.  We ran Mt. Xinu 4.3 + BSD on vaxen, so there source was

By the mid-80s, even if you had AT&T and BSD licenses, it didn't help,
as there were lots of vendors NOT giving out source (Sun, Pyramid, DEC,
Gould, DG, you name it).  This was pretty much OK; things worked
fairly well and you didn't need to fix drivers and recompile the
kernel and so on on those machines.

Vaxen were aging, BSD didn't support 8500s, and so if you wanted BSD
you pretty much had to go to one of the vendors offering it. Sun
was probably the most popular.

At a startup company we ran ESIX, SVR3 (and later SVR4) based on 386s
for our product alongside a few Sun boxen.  It was a good Unix, but
again no source, but no real need for it either. This was ~ 1990-1991.

W.R.T. personal machines, I shelled out $$ to buy an AT&T 3B1 (68010
based, System V kernel, SVR2 user land + vi) which I used happily
for many years. Later I bought a Sun IPC. (More $$.) Also used happily
for quite a while.

I only got into Intel land for myself when I moved to Israel, buying a laptop
and running Linux on it.  I had played some with Linux on Sparc and
was fairly impressed with it. This was circa 1997.

Linux generally "just works" out of the box on PC hardware, and
with Debian/Ubuntu, software updates are a breeze.  I spend almost no
time having to be a sysadmin for myself, which is wonderful.

I mostly watched the whole AT&T/BSD lawsuit stuff from the side. At
one USENIX I remember talking to Keith Bostic about it, and understanding
that it was trade secret, but I asked him "Given the book by Maurice
Bach on how UNIX works, how can they still think it's trade secret?"
He just sorta nodded and said "yep" or something equivalent.

But yes, the sense while it was going on was definitely that BSD
was risky and problematic. And that the whole lawsuit thing was
really, REALLY stupid.



More information about the TUHS mailing list