ietf-smtp
[Top] [All Lists]

Re: Last Call: draft-crocker-email-arch (Internet Mail Architecture) to Proposed Standard

2009-03-05 18:17:15



--On Thursday, March 05, 2009 13:24 -0800 Dave CROCKER
<dcrocker(_at_)bbiw(_dot_)net> wrote:

John C Klensin wrote:
     from a RFC 5321/SMTP perspective, any Message
Submission (RFC 4409) server clearly falls into Dave's
Mediator category. 


Huh?  I don't see that at all.

How can something that defines the submission event "take
delivery", which is one of the defining components of a
Mediator?  MSA's do 'receive' the message from the MUA, but it
ain't 'delivery'.

Depends on whether the MSA "receives" the message via some
private protocol or via SMTP.   If an MUA sends a complete
message to the MSA using SMTP, the submission server looks an
awful lot like an MTA (SMTP-receiver).  It differs from a
"normal" SMTP-receiver in only two respects: the MUA isn't a
full-service SMTP-sender because it doesn't maintain queues or
resolve MXs (but, in the general case, the MSA can't tell) and
it may have the ability to patch up broken/incomplete messages
(but those functions are not exercised if the submitted message
is complete and conforming).  

On the the other hand, it isn't clear to me that something is
"delivered", in the SMTP sense, to a gateway either.  To the
system sending something to a gateway over SMTP, the gateway
looks like a relay and there is no way to tell the difference
unless one deduces it heuristically from funky MX records.   As
with the MSA, the thing that makes a system a gateway (assuming
it speaks SMTP on both its receiving and sending sides) is the
functions it performs, not how the messages come in or go out.
Of course, that changes if either the low-level transports or
the mail-level ones are different, but that is also true of MSAs
in general (as distinct from ones that are strictly
4409-compliant).

   john


<Prev in Thread] Current Thread [Next in Thread>