UNIX Family Tree? - (nf)

[Guy Harris]

	Bell Labs released the early versions of UNIX (System I and II
	internally)...

Actually, those weren't System I and System II.  There were several UNIX
development streams in Bell; "version 6" and "version 7" came from the
"Research" group, so they are referred to as "Research UNIX".  There was
also a UNIX supported by a group within Bell Labs, used for various Bell
System projects, which was called USG UNIX (for Unix Support Group), and
there was a UNIX done for the PWB (Programmer's WorkBench) project called
PWB/UNIX.  PWB/UNIX was based on a version later than version 6 (it had
a C compiler, linker, archiver, Standard I/O library, etc. similar to the
one released with the "Phototypesetter, Version 7" tape, and a kernel similar
to the one made by adding the list of "50 changes" to UNIX that Ken Thomson
sent out to V6).  Later, the PWB/UNIX stuff was added to a version earlier
(I believe) than the released Version 7 (but with the V7 file system, etc.)
which made a system called UNIX/TS 1.0, which was more-or-less System I.
This became one of the two official UNIXes (the other was a descendant of
MERT called UNIX/RT), and the descendants of UNIX/TS 1.0 have been collectively
referred to as USG UNIX.

	with the sixth edition of the manual (So, people named it version
	6).  Berkeley hacked up version 6 and shared it with other institutions.
	Bell labs incorporated some Berkeley features and released the version
	7 ( this is where the compatibility ended).

Well, most of the changes between V6 and V7 were from Bell, not from Berkeley.
Neither "Version" 6 nor "Version" 7 were really "versions" in any formal
sense; I believe "Version 6" literally was "what happened to be on the
Research machine when the 'Version 6' tape was spun."  V7 was just a snapshot
of what the Research people had done to UNIX since V6.

	Bell labs went on with system III internally and Berkeley developed
	BSD4.1.  Eventually, Bell labs decided to support UNIX officially and
	released system III while developing system IV.  Berkeley turned to
	virtual memory with BSD4.2.

Actually, BSD 3.0 was the first release with virtual memory.  I'm not sure
whether the "First Berkeley Software Distribution" was their hacked V6 or
their hacked V7 (for the PDP-11), but the "Second Berkeley Software
Distribution" was a hacked V7 for the PDP-11.  The "Third Berkeley Software
Distribution" was based on UNIX/32V, which was UNIX V7 ported to the VAX-11.
The "Third Berkeley Software Distribution" had virtual memory.  The "Fourth
Berkeley Software Distribution" (frequently called, for obvious reasons,
4BSD) had a lot of new stuff, like job control, etc..  There have actually
been several "Second BSD" and "Fourth BSD" releases; the current "Second BSD"
release is 2.9BSD, and the current "Fourth BSD" release is 4.2BSD.  4.2BSD
is the one with the totally new file system, the totally new IPC/networking
system, the new signal mechanism, etc., etc..

	Bell labs decided to support the same version they sell and use
	(Finally!), but didn't want to use the number 4 or IV.

No, the reason that the version they sold which was supported and up-to-date
with the one they used internally was called System V was that it was
Bell release 5.0; "System V" would have been Bell release 4.0, 4.1, or 4.2.
Why they went with the Roman numerals, I dunno...

Any corrections, Dennis?  (DMR is the closest thing to an official UNIX
historian around; if he says it, you can believe it (but if you tell us
that UNIX was originally written in JOVIAL for the PDP-14, *that* will
change quickly! :-)).)  (Also, the internal history of UNIX within Bell is
even more ramified that my discussion implied; that wouldn't be directly
relevant to versions of UNIX released to the outside world, but would be
a *very* interesting discussion.  Any takers from AT&T Bell Laboratories?)

	Guy Harris
	{seismo,ihnp4,allegra}!rlgvax!guy