On Mon, 30 Jan 2012 23:34:16 +0100, Gregor Zattler <telegraph@gmx.net> wrote: > Hi Mark, > * Mark Walters <markwalters1009@gmail.com> [30. Jan. 2012]: > > On Mon, 30 Jan 2012 20:04:25 +0100, Gregor Zattler <telegraph@gmx.net> wrote: > >> * Pieter Praet <pieter@praet.org> [30. Jan. 2012]: > >>> On Mon, 30 Jan 2012 00:42:14 +0100, Gregor Zattler <telegraph@gmx.net> wrote: > >>>> * Pieter Praet <pieter@praet.org> [26. Jan. 2012]: > >>>>> Here's another couple of threads squashed into a single one: > >>>>> - [O] [Use Question] Capture and long lines > >>>>> - id:"BANLkTikoF4tXuNLLufRzNSD6k2ZYs7sUcg@mail.gmail.com" > >>>>> - [O] Worg update > >>>>> - id:"m1wrfiz3ch.fsf@tsdye.com" > >>>>> - [O] Table formula to convert hex to dec > >>>>> - id:"20110724080054.GB16388@x201" > >>>>> - [O] ICS import? > >>>>> - id:"20120125173421.GQ3747@x201" > >>>>> > >>>>> > >>>>> AFAICT, none of them share Message-Id's... > >>>> > >>>> Do you consider this a bug? > >>>> > >>> > >>> I do. No idea what causes it or how to fix it though... :) > >> > >> First I thougt it' not a severe bug since one see's more not less > >> messages in notmuch show buffer. But later I realised one also > >> sees less not more threads in notmuch search buffer and might not > >> read certain notmuch threads because of "wrong" $Subject: in > >> notmuch search buffer. > > > I think notmuch links two messages into the same thread if they have an > > in-reply-to or reference header in common: i.e the messages reference a > > common parent message. (See comment in lib/database.cc "Even before a > > message is added, it's pre-allocated thread ID is useful so that all > > descendant messages that reference this common parent can be recognized > > as belonging to the same thread.") > > So in case message a from thread A and message b from B would > name the same Message c in their In-Reoply-To:/References: > headers, while c is not (for some reason) in A or B, notmuch > would assume both threads linked? Makes sense. > > > As far as I can see your grep tests haven't checked for that. > > True. > > > Also, could you email me the mbox you had (I think you said that it was > > a mailing list so all public) and I will take a look? > > Sure, I do so off-list because of the size of the attachment. Hi I have looked at this and I think this is not notmuch's fault: I think it is a mua doing strange things: One of the mails has an in-reply-to header which looks like In-reply-to: Message from Carsten Dominik <carsten.dominik@gmail.com> of "Tue, 15 Mar 2011 12:18:51 BST." <17242340-A14F-495A-B144-20C96D52B620@gmail.com> and I think notmuch is taking the carsten.dominik@gmail.com as message id. A similar in-reply-to header appears in the other thread so notmuch pairs them up. According to http://www.jwz.org/doc/threading.html this form of header is not allowed under RFC2822 but was allowed under the earlier RFC822. You can see several such messages on the gnu-mailing list site eg ftp://lists.gnu.org/emacs-orgmode/2011-11 search for "in-reply-to: M" but they all appear to be from the same person (running mh-e 8.3 nmh under emacs 24) In my collection from the linux kernel mailing list I get some examples of in-reply-to not just being : <msg-id> but it was only about 200 from 100,000 messages in the second half of 2010 (the most recent archives I have). Best wishes Mark