On Thu, 2002-06-20 at 01:23, Colin Walters wrote: > The only reliable way to implement duplicate suppression is to keep a > hash of the content of the mail, somehow. But that gets very tricky, > considering that MTAs along the way can do things like QP->8bit > conversion, etc.
...and I forgot to mention that even decoding and canonicalizing the message body before taking a (say) md5 or sha1 hash will still fail, because Mailman for example adds footers to the end of the message. And come to think of it, that's just the beginning. Some mailing list software strips the text/html part out from multipart/alternative messages. Other mailing list software might add an attachment (e.g. a legal disclaimer), or even conceivably convert text/plain messages into text/html (evil as that would be). And consider the case where the mailing list is bilingual, and volunteer moderators add a translation of each message before sending it on to the list. Or, more insidious, suppose that a mailing list moderator removes swear words from a message. All in all, duplicate suppression is a Hard Problem, and I personally would be very reluctant to trust my email to a program which claimed to solve it reliably. Again, we should just do the Right Thing and not send duplicate messages in the first place (when they're not wanted). _______________________________________________ evolution-hackers maillist - [EMAIL PROTECTED] http://lists.ximian.com/mailman/listinfo/evolution-hackers
