[ih] very early email question

[John Day]

>In message <a06230989c065557c9c1b@[10.0.1.8]>, John Day writes:
>
>>Note that 385 is not an FTP spec.  The title is "Comments on the File
>>Transfer Protocol." Have another cup of coffee, Craig  ;-) (This is
>>back when RFC still meant Request for Comment.) This is Bhushan's
>>proposal for including them.   MAIL and MLFL were not included in the
>>spec until the March 1973 meeting.
>
>Hi John:
>
>Actually my note was the result of two days of research, so perhaps coffee
>*is* in order.

;-)

>Nonetheless if you read RFC 542 (product of the March 1973 meeting) it states
>that it incorporates:
>
>     "a considerable number of changes for the last 'official' version,
>     see RFCs 354, 385."
>
>And RFC 385, while entitled comments, is an odd hybrid to wit:
>
>     "The following comments... include errata, further discussion,
>     emphasis points, and additions to the protocol.  I shall incorporate
>     these comments into the main protocol document after we have had
>     sufficient experience."
>
>I've got email notes from at least one other person from that time
>saying that MLFL and MAIL were widely implemented soon after 385 came out.

Interesting.  I was at the March meeting but not the earlier one. 
Grossman may have been there for us.

>So I could be deeply confused, but there's a strong suggestion that
>someone had a brainwave (Bhushan?) between July and August 1972 and that
>it was promptly implemented, at least in west Cambridge (perhaps east
>Cambridge too longer :-)?).

Good point.  That is interesting because I have this very strong 
recollection of Steve Crocker coming into the 73 FTP meeting near the 
end of the second day and basically saying shouldn't we put mail in. 
There was a discussion and the two commands were agreed to.

>On the other subject...
>
>Yes, SNDMSG was for TENEX.  It was the program Ray Tomlinson created
>in 1971 and was the prototype email system discussed in the April '72 FTP
>meeting (that led to RFC 354).  It remained TENEX's mailer for some time
>and I assume it evolved a lot (I have to talk to Ray Tomlinson some more about
>it).
>
>Have you got a good pointer to the evolution of Multics' mailer?  I've been
>digging through archives trying to figure out how the mailers (MTAs) worked
>and the literature is worse than sparse.  Lots of stuff about user agents
>(RD, BANANARD, MSG, etc) but until Eric Allman appears with delivermail
>(sendmail's predecessor) nothing I can find on MTA development (even though
>clearly a lot had to be going on).

I would guess Pogran or MAP would know.  I remember there was a mail 
daemon fairly soon.  Not surprised there is not much documentation. 
The concept of MTAs and MUAs was still several years away.

Take care,
John