On Apr 08, 2012, at 01:11 PM, Richard Wackerbarth wrote: >I would propose a slightly different scheme for converting messages to stable >URIs.. > >If we create our ID by concatenation of some hash and a part of the date, >then the mail server need remember only those messages that fall in the same >date-sensitive part of the namespace. It can "forget" about ancient history.
Hi Richard, We had a very lengthy discussion about the hash a year or so ago, when the current algorithm was agreed upon. I'm too swamped at the moment to dig up the links, but look for input from Jeff Breidenbach and Jeff Marshall. The conclusion was that Message-ID was both sufficient and preferable as the sole input into the X-Message-ID-Hash value used for stable URLs. Of course date information could certainly be used to determine expiration from any kind of Message-ID cache for LMTP acceptance purposes. It doesn't have to be part of the hash input for that. Note though that Mailman has long had a feature to "clobber" the Date header when forwarding the message on to the archive. In mm2.1 this was closely tied to Pipermail, but in mm3 this can be enabled for any archiver. The problem was that Date headers can get skewed enough that it would cause threading problems in Pipermail. It's probably true that most bogus Date headers come from spam (trying to get their message at the top or bottom of my date sorted inbox summary). >Further, if we maintain sufficient Hamming distance, we can perform "error >correction" (mapping multiple IDs to the same canonical one)) and, thus >compensate for minor encoding differences caused by timing skew. Hmm, I'm having trouble seeing how useful this would be if the Date is not used to calculate the stable url. -Barry
signature.asc
Description: PGP signature
_______________________________________________ Mailman-Developers mailing list Mailman-Developers@python.org http://mail.python.org/mailman/listinfo/mailman-developers Mailman FAQ: http://wiki.list.org/x/AgA3 Searchable Archives: http://www.mail-archive.com/mailman-developers%40python.org/ Unsubscribe: http://mail.python.org/mailman/options/mailman-developers/archive%40jab.org Security Policy: http://wiki.list.org/x/QIA9
