Pipermail threading and Debian bug 167758

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Pipermail's threadKeys currently assume that the Date header of each mail message sent to the list is unique.
If two messages with the same Date both start new threads, then the thread display in the pipermail archives will erroneously intermingle the two threads. This is http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=167758 .
Debian have applied a patch to fix this, by including Message-IDs in the threadKeys too. Ironically, this actually creates an even worse bug, since they have failed to notice that '-' is a perfectly legal character in a message-id, thus, when pipermail counts the number of '-' characters in the threadKey to determine the nesting depth, it often gets a totally wrong answer.
This happens often, since all it needs is a message-id containing a dash. Exim message-ids always contain at least two, and message-ids often include domain names, which sometimes include dashes.
So, after that little bit of rambling, my essential question is:
Was there ever any previous discussion about this issue, or did the Debian folks not raise the issue upstream?
Max.
-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (Cygwin)
iD8DBQFDivZ5fFNSmcDyxYARAuroAJ9o7fzyy4kBtepLPg9PkEVKp5F+AACghh7f RyhxrGfae810YN1K8zfpGnE= =BOP5 -----END PGP SIGNATURE-----

On Mon, 2005-11-28 at 12:22 +0000, Max Bowsher wrote:
Was there ever any previous discussion about this issue, or did the Debian folks not raise the issue upstream?
I don't remember anything specific, but it's certainly possible that the issue's been raised before.
-Barry

On Mon, 2005-11-28 at 12:22 +0000, Max Bowsher wrote:
Was there ever any previous discussion about this issue, or did the Debian folks not raise the issue upstream?
I don't remember anything specific, but it's certainly possible that the issue's been raised before.
-Barry
participants (2)
-
Barry Warsaw
-
Max Bowsher