bug#70054: 30.0.50; Unknown ids in References: break threading

[Eric Abrahamsen <eric@ericabrahamsen.net>]

On 04/18/24 12:04 PM, Eli Zaretskii wrote:
> Ping!  Eric, any comments?
>
>> Cc: 70054@debbugs.gnu.org
>> Date: Sat, 06 Apr 2024 12:18:22 +0300
>> From: Eli Zaretskii <eliz@gnu.org>
>> 
>> > Date: Thu, 28 Mar 2024 13:18:25 +0000
>> > From:  "Jose A. Ortega Ruiz" via "Bug reports for GNU Emacs,
>> >  the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>
>> > 
>> > 
>> > This is possibly not a bug per se, but certainly a nuisance. When using
>> > the ProtonMail bridge service to read email in Gnus, some internal headers
>> > are added to the messages, all of them prefixed by X-pm- so that they
>> > pose no problems. But among them is one internal message id (X-pm-id)
>> > that then gets added to the References: header contents as needed. Thus,
>> > instead of seeing the original header, e.g.
>> > 
>> > --8<---------------cut here---------------start------------->8---
>> > References: <1399485980.12016.1.camel@Iris>
>> >  <CAJsg1E_Spqmm+wiwuKj6oeCFtZsyoTX9hRY1wAJ4T=TR2yKSmQ@mail.gmail.com>
>> >  <E1WiXKp-0003iD-2k@fencepost.gnu.org> <nn8uqblav2.fsf@bacon.lysator.liu.se>
>> >  <536C9013.1010806@gnu.org>
>> >  <CACujMDMbDfPP_7Es2Q0umgrgo9GH3_RhTYsfs3JKdmc2hORbRA@mail.gmail.com>
>> >  <E1WjRba-0005Aw-6q@fencepost.gnu.org> <1399810320.5752.0.camel@Iris>
>> >  <20140511122347.GA4055@intra> <1399814588.5752.4.camel@Iris>
>> >  <20140511183915.GA13216@fencepost.gnu.org>
>> > --8<---------------cut here---------------end--------------->8---
>> > 
>> > one gets
>> > 
>> > --8<---------------cut here---------------start------------->8---
>> > References: <1399485980.12016.1.camel@Iris>
>> >  <CAJsg1E_Spqmm+wiwuKj6oeCFtZsyoTX9hRY1wAJ4T=TR2yKSmQ@mail.gmail.com>
>> >  <E1WiXKp-0003iD-2k@fencepost.gnu.org> <nn8uqblav2.fsf@bacon.lysator.liu.se>
>> >  <536C9013.1010806@gnu.org>
>> >  <CACujMDMbDfPP_7Es2Q0umgrgo9GH3_RhTYsfs3JKdmc2hORbRA@mail.gmail.com>
>> >  <E1WjRba-0005Aw-6q@fencepost.gnu.org> <1399810320.5752.0.camel@Iris>
>> >  <20140511122347.GA4055@intra> <1399814588.5752.4.camel@Iris>
>> >  <20140511183915.GA13216@fencepost.gnu.org>
>> >  <c3j0GRUnm7ZRQKF-bDqrBcUaFhUN184gOb3xe4f2rcPF-kJwXJB2clP_k0wkt8BGK_85N2ro4YkuIBmgZC7kOg==@protonmail.internalid>
>> > <t_dhLBndvM2vyK9s5Qx7Njsmmk0oTeVhk-78ViYQwbwRCsok-6SuIxGrG9qKdBdg-ekWFdqwPKOenA86C1Vn9g==@protonmail.internalid>
>> > --8<---------------cut here---------------end--------------->8---
>> > 
>> > Note how two extraneous ids have been added to the list. I've checked
>> > that if I remove them by editing the message threading works again. With
>> > them, all I see is a flat tree where all mails are direct children of
>> > the first one. It's as if having an id in the References that Gnus
>> > cannot find broke proper threading. That's my theory anyway :)
>> 
>> Eric, any comments?

This certainly does sound like a bug. This is a pretty complex part of
Gnus, and it might take a while to track down. Jose, can you tell me
your value of `gnus-build-sparse-threads'? I would think that, if Gnus
*isn't* trying to build sparse threads, these unknown Message-IDs would
simply be treated like messages that haven't been loaded into the
Summary buffer, and ignored. Can you tell me your current value for that
variable, and also if/how changing the value (particularly between 'more
and 't) changes the behavior here?

Thanks!
Eric