On Sat, Dec 9, 2017 at 12:00 AM, Jeremy C. Reed <reed@reedmedia.net> wrote:

On Fri, 8 Dec 2017, Warner Losh wrote:

> OK. I'm confused. Maybe people here can help me understand.
> Looking at the V7 sources, it looks for all the world like they were
> released Jan 10, 1979. This release was PDP-11 only. So far, so good.
>
> 32V, a port to the VAX, is listed as 'early 1979'. Dates in the files in the
> archive suggest March 26th, 1979 (though there are dates as late as May 3rd
> and April 30th on two files that are trivial). The tape we have in the
> archive has a date Feb 22, 1980 written on it. Given the dates, that's only
> 3 months after V7 was released. This seems very fast, but maybe it's OK
> since it's a swapping release....
>
> 3BSD, The Berkeley 32V has file dates as late as Mar 22, 1980... This seems
> reasonable for turning V7 from swapping into paging... about a year is fast
> but not crazy fast.
>
> My question is: did these three events really happen in this quick
> succession? Did USDL folks get started with a preliminary V7 for V32 or was
> the port really done in 2 and a half months? Likewise with UCB and 3bsd: did
> they start early?

Bell got their VAX in Feb. 1978.

London and Reiser ported Unix to it and was multi-user in April.

Then spent five months to clean up for internal release in the Fall.

UCB purchased their VAX in the Fall.

Bell sent their prototype 32/V to Berkeley.

Bell's paging work began in late 1978.

Berkeley's own work from swap to paging began in late spring of 1979.
(Joy presented about this PUNIXVAX in June 1979 UNIX Conference.)

32/V did come out a few weeks after System 7.

Yea, the dates on the files in the archives say January 10 -> March 28th in 79... It makes a lot of sense that they started a year or more earlier...

By mid November, the Berkeley virtual memory code was in use
and was expected to be available for others by early December.

Berkeley's "Virtual Vax/UNIX" was available for students after
Christmas (1979) break.

BSD distribution to others began in January 1980.

Looks like 3BSD was cut in March 28th 1980, so that fits. Most of the work appears to end sometime over what would be that Christmas break (judging from file dates), with bug fixes continuing through March...

The Kashtan performance evaluation versus VMS (with benchmark code) was
done very soon after. (And Joy's fixes in March.)

Which matches the 3BSD release....

(I did a few interviews specifically related to this. I have long
chapters about this and I can provide my references/citations when I get
back from a trip I am on.)

That's totally cool! The key missing bits were that 32/V work started before 7th Edition was finalized. This is way more detailed than I expected...

Thank You!

Warner