Jim Henderson <hende...@gmail.com> posted ghucvl$p0...@ger.gmane.org, excerpted below, on Fri, 12 Dec 2008 19:10:45 +0000:
> IIRC, isn't RFC-compliant usage of the References header to use the root > message in the thread (as best the client can determine), the message > being replied to, and the parent to that? Now to correctly answer that, I'd have to go look up the RFC, wouldn't I? That sounds like work! =:^) Regardless of RFC, tho, it's common sense. But what I suspect is happening here is the server is using a Bayesian filter on either all headers or the entire message, and keying in on some portion of the message-ID. Get too many references with that pattern in them, and it rejects the message. It probably doesn't help that the pattern is likely in the current message-id and possibly the xtrace or some other header as well. It's not, I'd guess, a specific limit on references length or the number of references, but rather a limit on the number of times a particular string, which happens to be part of the message-ids in question, occurs. If we can change the message-id to avoid triggering the bayesian, we should be fine. Meanwhile, reducing the number of message-ids in the references header has the desired effect as well, but is way too cumbersome to do manually as a long term fix. -- Duncan - List replies preferred. No HTML msgs. "Every nonfree program has a lord, a master -- and if you use the program, he is your master." Richard Stallman _______________________________________________ Pan-users mailing list Pan-users@nongnu.org http://lists.nongnu.org/mailman/listinfo/pan-users
