On Tue, Mar 18, 2008 at 02:45:10PM -0500, Kyle Wheeler wrote: > On Tuesday, March 18 at 06:06 PM, quoth Vincent van Leeuwen: > >All the messages have Message-ID's and References or In-Reply-To > >headers which all seem to be correct. However, the Message-ID's from > >person B do look a bit odd: > > > >Message-ID: <[EMAIL PROTECTED]@example.com> > > Technically, it's not just odd, but an invalid message-id. Message-IDs > may only have a single @ symbol in them. What email client generated > that? >
X-Mailer: Microsoft Office Outlook 12.0 Surprise surprise. > >Can it be that Mutt doesn't recognise that Message-ID as a real one > >and ignores it for threading purposes? > > That's exactly correct. Mutt holds invalid message-id's in dim regard, > because they're so much more likely to be forgeries. Several mail > clients do bad things in this regard, including using email addresses > as the "message-id". Mutt does some gymnastics to try and accept as > many weird variants of message-id's as it can, but there's a limit, > and you've found one: mutt explicitly ignores so-called message-ids > that have more than one @ symbol. So when parsing the References > header, this message-id is ignored. > > Part of the reason for this is the philosophy "be conservative in what > you send, liberal in what you accept." Mutt is trying to avoid putting > invalid message-id's in the "References" header of email that it > sends. > Oh I hadn't realized the part about reusing it in a references header, that's definitely a good reason to ignore it. > >If so, is there anything I can do (except editing the messages by > >hand ofcourse) to make Mutt thread them properly again? > > It depends on what you're trying to achieve. If you want mutt to > display the threads properly, then editing them by hand (or feeding > them to a script) to alter the Message-ID into something that is > valid, and then using the & key to force generate the References > header in the reply is probably the only solution. > > If you want mutt to ignore the fact that the message-id is invalid, > you can edit mutt's source, but I don't recommend that. > Yeah, I'm not the only user using that binary on that server so I'll refrain from doing that ;) I already have a procmail+formail script to add missing message-ID's, seems like I'll need another one to correct this scenario. I'll have to think a bit on that one as it will produce References and I-R-T headers that make no sense. Although I guess I shouldn't be too nice to people who use mailclients who largely ignore those headers and produce invalid Message-ID's :) Thanks for the help and the prompt reply! Regards, Vincent van Leeuwen Media Design - http://www.mediadesign.nl/