Talk:PARC Universal Packet

Influence
In Dealers of Lightning: Xerox PARC and the Dawn of the Computer Age (ISBN:0887308910), Michael Hiltzik quotes John Schoch saying "We were told [by Xerox] to participate," in ARPANET technical meetings, "but we were ordered not to describe what we were doing." (p. 293). Schoch goes on to suggest that PARC's networking people contributed to the development of internetworking protocols by asking "inscrutable but cunningly pointed questions" at the meetings, since they were prohibited by Xerox corporate from actually sharing what they had working in Palo Alto. Hiltzik concludes that significant pieces of the PUP architecture therby made it into the design of TCP/IP.


 * It is probably safe to say influence went both directions. Should probably mention the people involved in the body of the article, since some at least have Wikipedia articles (and they all should eventually). Vint Cerf was on the faculty of Stanford, PARC was located in the Stanford industrial park adjacent to campus, and several of the designers of PUP were grad students at the time. So they certainly were very aware of IP. Should also add the PUP was eventually opened up in 1979, when Xerox gear was donated to Universities, resulting in PUP networks at least in the Stanford University Network and at CMU, for example. By then a few TCP/IP implementations were being tested in various stages, while PUP had been running for a few years. Noel of course knows more, e.g. the MIT deployment and Chaosnet etc. W Nowicki (talk) 00:08, 8 April 2011 (UTC)

Merge
This could do with merging in to Xerox Network Services, since it is XNS was escentialy PUP with a different name. --John R. Barberio talk, contribs 21:31, 29 September 2005 (UTC)


 * Only in the same way that XNS and IP are the same things with different names. Have you actually read any PUP technical documentation? Noel (talk) 00:35, 1 October 2005 (UTC)


 * XNS appears to be a development of PUP into a more robust form, rather than a development across a diferent track. There *are* diferences, but the only major difference appears to be the replacement of Rendezvous and Termination Protocol and Byte Stream Protocol, with that of Sequenced Packet Protocol as a single transmision control.
 * Since these articles are purely of historical interest, it makes more sence to merge them as they are two versions of the same system. --John R. Barberio talk, contribs 10:18, 1 October 2005 (UTC)

That is not the only significant difference. PUP and XNS are two separate things, which had very different roles in terms of how they affected later work. Although there is a familial relationship, there's absolutely no reason to try and cover them in a single article, any more than we cover dozens of other groups of historical topics with a familial relationship in one article (e.g. the PDP-6 and PDP-10, which were so nearly identical that most PDP-6 assembler programs could run on a PDP-10). Noel (talk) 12:23, 1 October 2005 (UTC)

Now that I've thought about it some, and had a chance to review my PUP and XNS documentation, PUP and XNS are quite different. Not only are the transport layers completely different (RTP/BSP being totally unlike SPP), but the applications are even more different; the printing, filing etc protocols of the XNS stack are built on top of Courier. So my characterization of them as "lightly modified" is not really accurate. Noel (talk) 11:40, 2 October 2005 (UTC)


 * Yes, glad this was not done. In particular, the network layer addresses were totally different. In PUP to get from experimental Ethernet address to PUP address you just concatenated, while XNS needed an ARP mechanism. W Nowicki (talk) 00:08, 8 April 2011 (UTC)