On Wed, Jan 22, 2020 at 11:21 AM Clem Cole <clemc@ccc.com> wrote:


On Wed, Jan 22, 2020 at 12:55 PM Warner Losh <imp@bsdimp.com> wrote:
I  thought the answer was "ARPANET" since we had a NCP on 4th edition Unix in late 1974 or early 1975 from the University of Illinois dating from that time (the code in TUHS appears to be based on V6 + a number of patches).
Because we can't ask Greg sadly, I think the Holmgren is the last around that would know definitively and I've personally lost track of him.

The publication date for the ACM paper is November 1975. I think I misspoke and you are right. The 4th edition was Nov 73, and 5th Edition was June 74 (6th was June 75). In order to meet deadlines for ACM publication, it most likely was 5th Edition, but there's also earlier references to it.

That said, I don't think UofI had anything earlier than 5th edition (I fairly sure that there were very few copies of 4th edition distributed outside of the Bell: i.e. Columbia, NYU and I thought Harvard; but I don't think too many more than that).  Lou Katz would be a better source than I, but I was always under the impression that the number 5th editions, the count was also a smaller 2 digit integer.  6th was where Unix began to 'spread' and by 7th, 'go viral.' 

Berkeley's license was executed in January 74, so it might be on the list, unless there was a big delay.

And to be honest, I personally thought that Steve and Greg did the ArpaNet NCP work on V6, but it might have been v5th I suppose.  I did not know about it until the 6th edition work.  But, they were fairly early.  BTW: I thought the Rand PIPE code was also developed on 6th, but those also might have been 5th.

The code we have is from the 6th edition (judging by diffs, though there's some weird quirks between it and Dennis_v6, as well as a number of local patches).

In addition to the Nov 1975 CACM paper, there's CAC 155, published by the University of Illinois on 3/15/75 which pre-dates the 6th edition by a few months. You can read it here https://www.ideals.illinois.edu/bitstream/handle/2142/32547/networkunixsyste155holm.pdf?sequence=2&isAllowed=y if you'd like. There's also CAC 177, which covers the period of Jan 1, 1974 to Dec 31, 1974 which references that Unix had been enhanced to add this after the contract periopd. This this report may have been issued after CAC 155 and may not be proof of an earlier date (though the issue date in its metadata in https://www.ideals.illinois.edu/handle/2142/34150 is 31 DEC 1974, it also reports activity through the end of the year so likely was written later). There's also CAC 162 dated May 15, 1975 (https://www.ideals.illinois.edu/handle/2142/32293) that references UNIX talking to the mulix machine via the ARPANET protocols.

RFC 681, dated March 18th, 1975, is another instance of an edited CAC 155 report  (it seems, I've not looked at them exactly, just a quick glance) that talks about this work. It's the earliest mention of Unix in an RFC (the next one isn't until 2 years later for an email address for Dave Crocker DCrocker@Rand-Unix in RFC 724 in May 1977 after which it explodes in references).

Warner