Re: [exim] Which behaviour is expected/standard from a retur…

Pàgina inicial
Delete this message
Reply to this message
Autor: Jon Bendtsen
Data:  
A: Tony Marques
CC: exim-users
Assumptes vells: Re: [exim] Which behaviour is expected/standard from a return-path value?
Assumpte: Re: [exim] Which behaviour is expected/standard from a return-pathvalue?
Den tirsdag 4.jul kl. 22:01 skrev Tony Marques:

> If by "i send out emails through exim with a forced return-path to
> root", you mean you insert a "Return-path:" header yourself, then that
> is not RFC compliant and it proper that Exchange or any other MTA or
> wannabe MTA to ignore such headers. The "return path" as opposed to
> "return-path" as in "Return-Path: header" is different.


i let exim set the return path in a transport, i would assume that exim
does it as it should.


> You should be discussing the MAIL FROM: in the SMTP envelope and it's
> value as that is where Exchange should (and usually does) bounce
> messages...


When i use tcpdump on the traffic it appears like my exim sets the
correct
address there and uses that as the mail from address when telling
exchange
who it is from.

Later comes the From address in the headers that shows the users
address.


> ...except if Exchange redirects the message to another SMTP recipient
> as then it does seem to use the From: header when forwarding, rightly
> ignoring a pre-existing Return-Path: header but stupidly ignoring the
> original MAIL FROM: in the SMTP envelope which it may have already
> discarded without storing the value in a new Return-Path: header which
> it should remove again before forwarding.


There are no other return-path set in the headers that are sent out.
i can see that while using tcpdump.


> The "return path" should be exclusively derived from the SMTP
> protocol's MAIL FROM: command while travelling in SMTP space and it
> would be upto Exchange or whatever ultimate recipient to create the
> "Return-Path: header" when the "final delivery" is made and it is
> deposited in a mailbox per RFC822 4.3.1 or otherwise leaves SMTP
> space.  RFC2821 is more explicit on Return-Path: headers and says
> stuff like:
>     A message-originating SMTP system SHOULD NOT send a message
>     that already contains a Return-path header.


I think exim uses the return path setting as the SMTP Mail from.


> So Return-Path: headers should never be found in messages MTA's
> receive or deliver.  It is also wrong for Return-Path: formats to be
> appear like...
>     Return-Path: "John Smith" <some@???>
> as they logically should be derived from the MAIL FROM: in the SMTP
> protocol envelope.


I dont have those.


> If it was up to me I would bounce all messages with Return-Path:
> headers as non-RFC compliant, but that isn't pratical and not RFC
> complaint either (since no inspection is supposed to be made, haha).


what happened to "be strict with what you sent out, be forgiving with
what you receive?"


Output from tcpdump:
exim                            exchange
------------------------------------------------------------------------
EHLO frodo.laerdal.global
                            nosrv003.laerdal.global Hello [192.168.119.133]
MAIL FROM:<root@???> SIZE=2075
RCPT TO:<foobar@???>
                            root@???....Sender OK
                            54 Start mail input; end with <CRLF>.<CRLF>
Received: from dkjbe1 by frodo.laerdal.global with local (Exim 3.36  
#1 (Debian))
         id 1Fy53d-0004PA-00
         for <diffs@???>; Wed, 05 Jul 2006 12:56:37 +0200
To: diffs@???
Subject: CVS: jontest find.etc,1.64,1.65
Message-Id: <E1Fy53d-0004PA-00@???>
From: Bendtsen <dkjbe1@???>
Date: Wed, 05 Jul 2006 12:56:37 +0200


Update of /data/cvs/jontest

...

                            <E1Fy53d-0004PA-00@???> Queued mail for delivery
------------------------------------------------------------------------ 
--------


So i think exim is doing it correct, and that there are no return
path in the headers,
and that SMTP envelope from is correct.




JonB