[Mailman-Users] No Date: line in digest header

[John W. Baxter]

On 9/1/2004 13:12, "Todd K. Watson" <tkw at southwestern.edu> wrote:

> 
> Thanks Brad,
> 
> This makes sense to me, and I was afraid that was going to be the
> answer.  I'll take a look at qmail-based solutions.  However...
> 
> Not to instigate a "who should do what", but this doesn't sit well for
> me.  It seems to me that the MTA should not inject any date except in a
> "Received:" line -- which qmail does.  Since the origination-date date
> line is specified as one of the only required header lines for a message
> in RFC-2822, it would seem to be something that should be fixed on
> Mailman.  I understand that it's up to me to have my MTA allow or deny
> messages without it, but based on my relatively under-educated (I'm new
> to running Mailman) opinion I would rather have the fix on the Mailman
> side.
> 
> Given that, would you recommend that I report this in bugtrack?
> 
> Also, one thing that did turn up in my investigation into this was in
> the release notes (the NEWS file in the root level of the 2.5.1 tarball)
> the following highlight is mentioned in the section for
> Mailman-2.1-Beta-1 (line 591):
> 
> "o Always add an RFC 2822 Date: header if missing, since
> not all MTAs insert one automatically."
> 
> This leads me to think that this was (at least at one time) addressed.
> I can't find anything to indicate that was backed-out.  Also, it doesn't
> specify if it was applied to both digesting and regular messages.  I'm
> assuming only the latter.
> 
> Again, thanks for your response Brad!
> 
> Todd
> 
> 
> Brad Knowles wrote:
>> At 1:53 PM -0500 2004-09-01, Todd K. Watson wrote:
>> 
>>>  Most MTA's and MUA's inject a Date line, but I'm using Qmail as an MTA
>>>  -- which doesn't inject one.  It seems that it's the job of the MUA
>>>  (Mailman in this case) to create the Date: line according to RFC-2822.
>> 
>> 
>>     The MUA should include a Date: header, that's true.  However, it's
>> up to the injecting MTA to either make sure that the minimum required
>> headers are present in the message when it accepts it, or to put the
>> headers on there.  Dan will violently disagree, but qmail is at fault here.
>> 
>>     There are plenty of programs available for qmail users who are in
>> this boat and need to have their MTA add headers by default, which qmail
>> otherwise would not do.  However, you need to use qmail resources to
>> find those programs.

I believe Mailman should generate the required headers, including Date:,
From: (which it does I think), and Message-Id: on digests (where it is
pretty clearly acting as a mail user agent).  Now that I know it doesn't,
I'll configure Exim on our Mailman machine to add the headers if missing
when--some future time--we upgrade Exim on the Mailman machine.

There was a recent thread in the Exim mailing list on this general topic
(although in this case the question was about Bcc: header lines).  And that
thread led to a query on the ietf-822 mailing list, which had discussed the
Bcc: issue at length in the past.

For the Exim developer's summary of the ietf-822 discussion, see
<http://www.exim.org/mail-archives/exim-users/Week-of-Mon-20040816/037254.ht
ml>

The arguments should be similar regarding MTA's adding missing required
headers (Date:, From:, Message-Id:) as for removing Bcc: headers.
Basically, it's the MUA's job...when the MTA does it, it is acting like an
MUA and should only do it if it can be configured to know *when* it is
acting like an MUA.  (Exim does it if called as from the command line with
the -t option.)

Exim's most recent versions have a mechanism for saying essentially "this
message is a submission and you should add missing headers".  (However, this
control does not set up removal of Bcc: header lines...the topic of the ietf
discussion.)

  --John