[Info-vax] What is a "real" Unix ?

[Dan Cross]

In article <ud4ldd$1g3dd$1 at dont-email.me>,
Simon Clubley  <clubley at remove_me.eisner.decus.org-Earth.UFP> wrote:
>[snip]
>> Now that's an interesting question.  4.2BSD is Unix, because it incorporates
>> code from AT&T v7.  It comes with both an AT&T and a Berkeley license.
>>
>> But the latest version of OpenBSD no longer has any AT&T code in it, so
>> strictly speaking it's not really Unix.
>
>I didn't know that about OpenBSD, but in every way that matters, it most
>certainly still is Unix IMHO and it's an excellent example of the point
>I am making above.

Starting in the 1980s, the CSRG at UCB, start an effort, mostly
headed by Keith Bostic, to purge BSD of AT&T proprietary code
and produce an unencumbered BSD distribution: that is, the whole
OS, freely redistributable, as open source, without license from
AT&T.

The first iteration of this was the Net/1 release, which was
incomplete, and came after 4.3BSD-Reno and before 4.4BSD.  Net/2
came shortly thereafter and was mostly complete, but found to
contain some encumbered AT&T code.  4.4BSD came next, and had
(essentially) three releases:

1. 4.4BSD-Encumbered, which contained a very small amount of
   AT&T source in, I think, 7 source files?  Something like
   that...  Anyway, it was a full, bootable distribution.
2. 4.4BSD-Lite, which was basically -Encumbered minus the AT&T
   code and probably missing some cryptographic code as well.
   In that era, there were still restrictions on exporting
   cryptographic software out of the United States.
3. 4.4BSD-Lite2, which was the final official release, came a
   few months or a year or something after -Lite, and contained
   mostly bug fixes and ongoing work done at Berkeley.

Most of the "free" BSDs started with 386BSD, which was a port of
Net/2 to the 386 done by Bill and Lynne Jolitz (and documented
in a series of articles published in Dr Dobbs Journal, and later
collected into a book).  This generated a lot of excitement, as
people could get "real" BSD on PC-class hardware, but the Jolitz
crew could be "challening" to work with and was slow in
producing bug fixes, etc.  Groups started formating ersatz
distributions that brought together sets of patches, coalescing
into the NetBSD and FreeBSD projects (more or less started
concurrently, more or less at the same time).  NetBSD came to be
distinguished by an emphasis on portability, while FreeBSD
started with a focus on, "turning PCs into workstations" and
then focusing on server applications on the x86 ("the power to
serve").  Eventually, OpenBSD forked off of NetBSD after Theo
de Raadt was ejected from that project for abusive behavior.  A
few years later, FreeBSD got a little wild with the cheese whiz
in trying to implement a complex M:N threading model and Matt
Dillon forked off DragonFly BSD.  Those are the four biggest BSD
distributions today; of those, FreeBSD has continued to focus on
server applications and is widely used (e.g., Netflix, Yahoo,
etc) while OpenBSD focuses on security and "correctness" and is
probably better known for its subprojects, in particular
OpenSSH.

Eventually, all of the BSDs effectively rebased themselves onto
4.4BSD-Lite2.

Linux is a totally separate codebase, developed independently
initially by Linus Torvalds, and then by a cast of many
thousands.  Torvalds was working with Minix as a pedagogical
system, but found it unsatisfactory for workaday use, so wrote
his own imitation of Unix.  It is now arguably the most
successful and important operating system in history.

	- Dan C.