by Darius Kazemi, Feb 6 2019
RFC-37 is the meeting notes from the network meeting held on March 17, 1970. This RFC is dated March 20, three days after. Not a bad turnaround! The RFC is titled “Network Meeting Epilogue, etc.” and I love the et cetera there.
The technical content
The meeting was attended by about 25 people from a variety of institutions. The meeting was held because there was a lot of pushback on RFC-11's protocol, which resulted in the creation of the RFC-33 protocol, and they wanted to get ahead of any criticism by hosting this meeting to put an early draft in front of the group before formalizing it.
It seems the basic protocol involving connections and sockets was well-received at the meeting, with some friction around the dynamic reconnection concept that Crocker introduced in RFC-36.
James Forgie replaces Paul Rovner as the main Lincoln Lab contact. Forgie was an early pioneer in text-to-speech systems and would soon become Lincoln Lab's lead on the ARPA Speech Understanding Research Program (source, p. 109), which funded major speech recognition research from 1971-1976 at a number of US institutions.
Process is defined formally for the first time in an RFC. It is “a program which has an assigned location counter and an address space”. Which translates to my lay definition I provided earlier: it's a program that is currently running on your computer.
I just wanted to note that Crocker is very careful to state that this RFC is his own recollection of events and as such there may be errors.
Also of note is that Crocker discusses the emotional aspect of negotiating technical features at length. He points out a “conflict of gut level feelings” in a conversation between himself, BB&N's William Crowther, and others over the dynamic reconnection concept.
I stumbled across this 597 page behemoth of a Lincoln Laboratory retrospective, produced by the laboratory itself in 2011.
The RFC references the definition of process in “Structure of the Multics Supervisor” a 1965 paper by Vyssotsky, Corbató, and Graham. This further corroborates my suspicion that Multics (the main predecessor to Unix) was the inspiration for a bunch of this now-common terminology.