bug#335: bug tracking system leads to duplicate replies appearing

[Joe Wells <jbw@macs.hw.ac.uk>]

Don Armstrong <don@donarmstrong.com> writes:

> On Thu, 29 May 2008, Joe Wells wrote:
>> I believe this is due to messages being addressed to both
>> bug-gnu-emacs@gnu.org and XYZ@emacsbugs.donarmstrong.com where XYZ
>> is the bug number. When someone follows up, their message gets sent
>> to bug-gnu-emacs@gnu.org twice, because everything sent to
>> XYZ@emacsbugs.donarmstrong.com will also be sent onward to
>> bug-gnu-emacs@gnu.org.
>
> This is because people are using reply-to-all, instead of
> reply-to-list or reply-to. I can resolve this by discarding duplicate
> message ids, but thats more of a sledgehammer. Far better would be for
> people to stop being silly and using reply-to-all, and instead use
> reply-to-list or reply-to, both of which are set properly and behave
> correctly.

Unfortunately, there is no way to know whether all of the recipient
addresses in a message are on the mailing list.  Even if this is so
for one mailing list, it might not be the case for other mailing
lists.  So people can not in general rely on the behavior of any
“reply” function of their mail program to reach the right people.

In this case, a good solution is that the bug tracking software could
make sure to always omit bug-gnu-emacs@gnu.org from the list of
recipients whenever XYZ@emacsbugs.donarmstrong.com is going to be one
of the recipients.  This will work in this case because anything sent
to XYZ@emacsbugs.donarmstrong.com will also get sent to
bug-gnu-emacs@gnu.org.

I notice that you just now have started adding a Mail-Followup-To
header.  This is a good idea.  I notice that don@donarmstrong.com is
not in the Mail-Followup-To header.  I'm curious, how does your
software know to omit this address?  Do you receive a copy of all bug
report traffic regardless?

>> Interestingly, by the time these two copies show up in the
>> gnu.emacs.bug USENET newsgroup, they have the two different message
>> IDs <mailman.12118.1211578329.18990.bug-gnu-emacs@gnu.org> and
>> <mailman.12121.1211578334.18990.bug-gnu-emacs@gnu.org>.
>>
>> By the way, it seems at first glance that none of these are the
>> original message ID of Stefan's message, but instead these are fresh
>> message IDs chosen by the bug tracking software and mailing list
>> software.  At least the two message IDs that begin with “mailman” are
>> definitely fresh.  These pieces of software should *not* be making up
>> fresh message IDs. 
>
> Debbugs has nothing to do with these message IDs; it adds
> Resent-Message-Id: for the messages which it forwards, and retains the
> original Message-Id:. Most likely this is something to do with the
> mailman list-to-news gateway.

Indeed, the “mailman” prefix of the message IDs hints at this.  I
don't know how the mailman software maintainers can justify this.

When I wrote my report, it seemed to me that the message ID
<jwvwslkzpun.fsf-monnier+emacsbugreports@gnu.org> might have been
generated by your software.  I am happy to hear this is not the case.

-- 
Joe


-- 
Heriot-Watt University is a Scottish charity
registered under charity number SC000278.