Re: [exim] Is not honoring bounces-to violation of RFC?

Pàgina inicial
Delete this message
Reply to this message
Autor: John C Klensin
Data:  
A: jeffschips
CC: exim-users
Assumpte: Re: [exim] Is not honoring bounces-to violation of RFC?
Since the confusion seems to have continued after the message to
which I responded...

--On Tuesday, June 28, 2016 15:59 -0400 Chip
<jeffschips@???> wrote:

> My mistake NOT "bounces-to" rather "return-path" as in the
> following snippet of campaign emails from Home Depot, Martha
> Stewart and Sears:
>
> From - Mon Jun 20 08:43:03 2016
> X-Account-Key: account15
> X-UIDL: UID1962-1324328699
> X-Mozilla-Status: 0001
> X-Mozilla-Status2: 00000000
> X-Mozilla-Keys:
> Return-path:
> <bounce-21178_HTML-212410161-2947777-10142840-2@???
> otemail.com>


RFC 5321 requires that an MTA create a Return-path: header field
and copy the mailbox argument of the MAIL command into it before
making delivery. If the MTA bounces the message, there will be
no Return-path: header until the bounced message is delivered to
wherever it was sent (see earlier note).

Note that, unlike Envelope-to" and a number of other header
fields that are made up by many MTAs, "Return-path:" and the
conditions under which it is created are quite explicitly
specified by RFC 5321 (and all of its SMTP predecessors) and are
required, not discretionary. (Chris's explanation is otherwise
correct, including the discussion of transitions into other mail
systems (Section 3.7 of RFC 5321 may be further helpful in that
regard.)

--On Tuesday, June 28, 2016 23:03 +0200 Heiko Schlittermann
<hs@???> wrote:

> Maybe this helps.
>
> https://en.wikipedia.org/wiki/Bounce_address


FWIW, that article contains a lot of subtle errors, starting
with 5321 being as clear as we could make it that the names of
the relevant commands are "MAIL" and "RCPT", with "FROM" and
"TO" as arguments. For those who are curious, the distinction
became important when RFC 1425 was published in February 1993.

> And reading the above article I just learned that my
> terminology was wrong too, 'envelope from' I meant where I
> used 'envelope sender'.


There is a lot of sloppy terminology around, including in the
SMTP spec itself. In my experience, unless complete pedantry
has taken over, no one really cares as long as the intent is
clear. If one cares, the precise terms are "mailbox argument to
the MAIL command", as above, "reverse path", or
"backward-pointing address".

best,
john
(for identification, I'm the editor of RFC 5321)