Subject: Re: Why did NetBSD and FreeBSD diverge?
To: James Howard <email@example.com>
From: Greg Lehey <firstname.lastname@example.org>
Date: 01/18/2001 01:22:45
On Wednesday, 17 January 2001 at 11:40:57 -0800, James Howard wrote:
> Why did Jolitz pull support from 386BSD? And what was BSDi doing at
> the time?
OK, let's look at some time lines. I'm currently travelling, so I
don't have hard facts to back up all these statements, which are from
my recollection. Feel free to counter them with facts.
pre-1990: Some people at the Computer Sciences Research Group in
Berkeley realized that the days of the CSRG were numbered,
and work on releasing the Berkeley code in unencumbered
form, primarily for people who wanted TCP/IP stacks. The
result was the Berkeley Networking Tape, later called
Net/1. It didn't pretend to be an operating system, but it
was a complete TCP/IP stack.
Still at Berkeley, Bill Jolitz and some others work towards
porting 4.3BSD Reno to the 386, and making the result
unencumbered. They failed, but Bill described the work in a
very detailed series of articles in Dr. Dobbs Journal,
starting (I think) in early 1991.
mid-1991: The CSRG released Net/2, the unfinished attempt at a 4.3BSD
port to the 80386. A large proportion of the CSRG members,
including Mike Karels, Kirk McKusick, Chris Torek and Bill
Jolitz, join up with some others, notably (at a later date;
I think 1 December 1991) Rob Kolstad, to create a company
called Berkeley Software Design Inc. (BSDI) to market this
software. Quite early on people started writing the
abbreviation as "BSDi", but they didn't in fact lower-case
the i until April 2000.
It's not clear what Bill Jolitz thought the goals of BSDI
were. Rob Kolstad told me that he got very upset towards
the end of the year because BSDI wanted to charge money for
the system. It's not clear how he thought they were going
to be viable without doing so, but he left BSDI on 1
December 1991, not before he had destroyed all his work.
Feb 1992: BSDI releases the first Beta versions of their commercial
operating system, BSD/386.
Mar 1992: Bill Jolitz releases the first alpha version (0.0) of his
free operating system, 386BSD.
14 July: Bill Jolitz releases version 0.1 of 386BSD.
At this point, BSD/386 was quite a usable system. I was
running both Interactive UNIX/386, a System V.3.2
derivative, and BSD/386 0.3.3, and the BSD/386 was already
much more polished than Interactive. By all accounts 386BSD
was still a disaster. I once started trying to install it,
but didn't get very far.
Apr 1993: NetBSD 0.8 came out.
Dec 1993: FreeBSD 1.0 came out.
End 1995: Dr. Dobbs markets "386BSD 1.0" on CD-ROM for $99, promising
support. It was a disaster, no support was forthcoming, and
the documentation was in a proprietary Microsoft format. I
don't know that anybody ever got it running: by that time
FreeBSD and NetBSD were just too far ahead, and the CDs were
a lot cheaper.
So why did Bill "pull support"? I don't think he did. He never
offered any support, and much of the ill-feeling came from people who
thought that he should put their patches back into the base. That
would have been a sensible thing to do, of course, but he obviously
didn't want to do it. I suspect that he found the whole thing had
grown over his head. In hindsight, it's surprising that it took so
long for the NetBSD and FreeBSD people to get started. If it had
happened earlier, it's possible that people might have got over their
differences and formed a united BSD project. I don't know if that
would have brought better results.
Finger email@example.com for PGP public key
See complete headers for address and phone numbers