12 Mar
2023
12 Mar
'23
12:39 a.m.
On Sat, Mar 11, 2023 at 9:41 AM Clem Cole <clemc(a)ccc.com> wrote:
I have never figured out who was first (Peter Weiner at ISC or the folks at
Wollongong) or the amount of the fees involved, but at some point, both
managed to negotiate a special license to redistribute UNIX in some manner.
My memory is that the commercial target had to get some sort of license
from AT&T first. My memory of the ISC product was it was the source for
your 11/70 [factiod - the Motorola guys were using it for what would
eventually become the 68000 - Les Crudele told me they had source]. I also
remember that when later Wollongong Vax products appeared, sources were
available, but I've forgotten the details - I was never a customer --
Warner might know more here.
Here's what I know about TWG's products. It's tangentially related to
unix,
and a bit rambly...
After the original Unix port from Wollongong, they branched out. They knew
they couldn't compete with Berkeley sending out tapes from the early 1980s,
so they pursued two niche markets. They got into two niche markets. They
used their Unix license to sell Eunice, which had been developed at
Stanford by David Kashtan. He took BSD Unix and managed to get enough of
the kernel to run as a process (and some device drivers?) under VMS. I
don't know if he started with 4.2 or redid the work later with 4.2, but
that added networking to the VAX, which DEC didn't have at the time. TWG
marketed Eunice for a pretty penny. The emulation wasn't very complete
(though many things just worked) owing mostly to the mismatch between the
VMS process model being super heavyweight and Unix's fork/exec being
lightweight. Plus, the pipe device driver never quite got to complete
compatibility (it lacked the ability to pass fd credentials from process to
process, for example). So it was kinda a mess. Source code was available,
but hella expensive and it was only available so that TWG could sell into
the government market that required it. TWG's
So, v7 was kinda dead, and Eunice was a super-niche thing from the get go,
what did TWG do? Networking. They separated (poorly, imho, but more ports
better than one good port) the networking part of enuice from the rest and
marketed that as a product. It was a total hack job, but for a product in
high demand. That experience, and their relationship with Bell Labs meant
they ported the networking code to System III and newer machines and
marketed it to all of those (so we had several 3Bx systems around running
System Vr2 and newer, though we had some machine that was system III
nominally, though i don't recall those details, but Sony NEWS, SunOS, Sun
road runner, HP running unix and non-unix, IBM maybe and a lot of others
were in the QA lab). My rather simple .cshrc and similar files date from
this time period since we had NFS running on all (many) of them. They also
purchased IP/TCP or hired someone whose name I should remember but don't to
make it good. He optimized the heck out of it to turn it into their
software to compete with FTP Software's offering. Source wasn't available
for any of this. They were going for quantity of ports, not quality of any
individual one. They also had an ISO stack that they sunk a bunch of money
into (port of BSD's to System V), but that didn't go anywhere...
The quality issues is why TGV got started. I have a vague memory that David
Kashtan went to SRI and redid networking for VMS right and spun out TGV so
there was a lot of bad blood between TWG and TGV. Multinet was cool because
it could plug in ISO protocols too, and was a native VMS thing with only
the TCP stack itself being BSD code. It's integration into VMS was quite
good, and they did better at benchmarks than TWG. I have friends still that
used to work there if people are interested in fact checking my maybe not
so great memory here...
I only ever logged into Eunice once or twice. I did a lot of work with
TWG's VMS TCP/IP product in college and went to work for them afterwards
back when I thought VMS would win over Unix (silly me).
Warner