[net.mail.headers] proposed Mail Transfer Protocol

Crispin@SUMEX-AIM.ARPA (Mark Crispin) (09/19/85)

Folks -

     I propose the formation of a committee to specify and design
an Internet Mail Transfer Protocol to extend and ultimately
replace SMTP/RFC822.  I suggest that the various mail system
implementors form a committee to meet at Stanford or ISI and
subsequently carry on the discussion via netmail.

     My thoughts along this line are that a new command, XMTP, be
defined in the SMTP command set.  A successful response to an
XMTP command will indicate the SMTP server's willingness to
become an MTP server.  This is upwards-compatible with the
existing SMTP world, since a server that does not support MTP
will reject the XMTP command and the transaction will proceed
using SMTP.  It is also superior to establishing a separate MTP
service port, which would force every mail sender to try two
ports before giving up.

     The MTP will have two goals:
 . expanded and completely machine-readable envelope
 . support for multi-media mail

     What I wish to do with the envelope is to completely
eliminate the RFC822 header and have all information which was
previously transmitted in the header appear in the envelope that
is transmitted via MTP commands.  This data will be completely
machine-readable (esthetics as to what constitutes a "pretty
header" will NOT be considered).  Compatibility with CCITT and
other mail transmission standards will be a strong consideration.

     Message headers transmitted over the network as we now know
them will cease to exist.  This information would be presented to
users in a way determined by the local mail software.  The local
message system at the receiver's end would be free to write
messages in RFC822 format in the user's mail file for the
convenience of existing software.  Or, it could use some
completely different format; the protocol couldn't care less.

     The entire issue of whether or not mail relays should
"reformat headers" will be laid to rest.  There are no headers to
reformat, and there will be well-defined heuristics on how the
envelope is altered when it passes a relay.

     My ideas for multi-media mail are less well-defined, and
could use input from the MMM folks.

     I am seriously planning on implementing this as a private
protocol between TOPS-20's and other cooperating systems.  I
would like to see this be made an official experiment.  I don't
really want to write up an RFC yet because I feel it should be a
group effort.

     Looking for a positive response!

-- Mark --
-------

bob@basser.oz (Bob Kummerfeld) (09/23/85)

In article <1587@brl-tgr.ARPA> Crispin@SUMEX-AIM.ARPA (Mark Crispin) writes:
> Folks -
> 
>      I propose the formation of a committee to specify and design
> an Internet Mail Transfer Protocol to extend and ultimately
> replace SMTP/RFC822.  I suggest that the various mail system
> implementors form a committee to meet at Stanford or ISI and
> subsequently carry on the discussion via netmail.
> 
> .........
> 
>      The MTP will have two goals:
>  . expanded and completely machine-readable envelope
>  . support for multi-media mail
> 
>      What I wish to do with the envelope is to completely
> eliminate the RFC822 header and have all information which was
> previously transmitted in the header appear in the envelope that
> is transmitted via MTP commands.  This data will be completely
> machine-readable (esthetics as to what constitutes a "pretty
> header" will NOT be considered).  Compatibility with CCITT and
> other mail transmission standards will be a strong consideration.
> 

Why not adopt X.400?

Bob.