what should happen when mda sees multiple List-ID headers?

what should happen when mda sees multiple List-ID headers?

[Eric Wong]

Given my recent traumatic experience[*] around multiple
From/To/Cc/Subject headers; I guess we should prepare for the
possibility of multiple List-ID headers showing up in -mda.

Right now, we handle the first one (and I'm updating -learn to
support List-ID, too); but it's possible that multiple List-ID
header can map to different inboxes, too...

[*] https://public-inbox.org/meta/20191016211415.GA6084@dcvr/

Re: what should happen when mda sees multiple List-ID headers?

[Eric W. Biederman]

Re: what should happen when mda sees multiple List-ID headers?

[Eric Wong]

"Eric W. Biederman" <ebiederm@xmission.com> wrote:

... nothing?  Just checked my mail server and it's not out of
space and I'm not seeing any errors in logs.  Anyways I'm
offline for a bit and will be back (hopefully :x)

Re: what should happen when mda sees multiple List-ID headers?

[Eric W. Biederman]

Eric Wong <e@80x24.org> writes:

> "Eric W. Biederman" <ebiederm@xmission.com> wrote:
>
> ... nothing?  Just checked my mail server and it's not out of
> space and I'm not seeing any errors in logs.  Anyways I'm
> offline for a bit and will be back (hopefully :x)

Apologies, I accidentally hit send before I wrote my reply.

Eric

Re: what should happen when mda sees multiple List-ID headers?

[Eric W. Biederman]

Eric Wong <e@80x24.org> writes:

> Given my recent traumatic experience[*] around multiple
> From/To/Cc/Subject headers; I guess we should prepare for the
> possibility of multiple List-ID headers showing up in -mda.
>
> Right now, we handle the first one (and I'm updating -learn to
> support List-ID, too); but it's possible that multiple List-ID
> header can map to different inboxes, too...

So I looked at RFC2919 and it had something interesting to say.

- That when you receive a message from a mailing list only the
  List-ID set by the mailling list software should be present.

I think in practice that means if everything is RFC conformant
and working correctly you should never see a message with
multiple List-IDs.

Which leads us to what should we do with potentially buggy software.

Or possibly a creative list processor that realizes that handles
multiple lists, sees that a message is cross-posted and sends
out a single copy of it to the addresses that are subscribed to
both lists.



There are two reasonable things that can be done, and I suggest
we do them both.
- Print a warning. (To be deleted if this case turns out to be common).
- Deliver to all of the lists you have mailboxes for the List-IDs.

I think that would be the most robust response long term, and it
is roughly what my code in import_imap_mailbox so I don't see it
causing any problems at present.

In particular the delivery logic I have been running with for a long
time now is:

		my @dests = email_dest($config, $mime);
		for my $ibx (@dests) {
                	my $im = PublicInbox::InboxWritable->new($ibx);
                        die "no im" unless defined($im);
			$im->add($mime);
		}

So I can say at worst it is harmless to have code that handles multiple
List-ID headers.

Eric

Re: what should happen when mda sees multiple List-ID headers?

[Eric Wong]

"Eric W. Biederman" <ebiederm@xmission.com> wrote:
> There are two reasonable things that can be done, and I suggest
> we do them both.
> - Print a warning. (To be deleted if this case turns out to be common).
> - Deliver to all of the lists you have mailboxes for the List-IDs.

Thanks, I was leaning towards delivering to all matches, too.

I didn't think to print a warning; but I guess -mda and -learn
should have -v and -q switches to be more/less quiet...

> I think that would be the most robust response long term, and it
> is roughly what my code in import_imap_mailbox so I don't see it
> causing any problems at present.

<snip>

> So I can say at worst it is harmless to have code that handles multiple
> List-ID headers.

Agreed.

Re: what should happen when mda sees multiple List-ID headers?

[Eric W. Biederman]

Eric Wong <e@80x24.org> writes:

> "Eric W. Biederman" <ebiederm@xmission.com> wrote:
>> There are two reasonable things that can be done, and I suggest
>> we do them both.
>> - Print a warning. (To be deleted if this case turns out to be common).
>> - Deliver to all of the lists you have mailboxes for the List-IDs.
>
> Thanks, I was leaning towards delivering to all matches, too.
>
> I didn't think to print a warning; but I guess -mda and -learn
> should have -v and -q switches to be more/less quiet...

I was thinking a warning in the sense of.  We don't think this should
happen please let the developers know if it does so we can confirm
that this case is being handled correctly.

>> I think that would be the most robust response long term, and it
>> is roughly what my code in import_imap_mailbox so I don't see it
>> causing any problems at present.
>
> <snip>
>
>> So I can say at worst it is harmless to have code that handles multiple
>> List-ID headers.
>
> Agreed.

Eric