jim@uw-beaver.UUCP (06/09/83)
This problem has been described before. What happens is that some site drops the message-ID. Then some later site regenerates the message-ID from the article-ID. But the article-ID doesn't contain enough information to regenerate the message-ID correctly. This can happen if the site name is longer than 8 characters or if the domain is not .UUCP (both of which are true here). Why does the message-ID get dropped? Most old news systems simply pass on headers they don't recognize, but some drop them. A+ news used to drop headers until a fix came out recently. At least as far back as 2.7, B news was passing unrecognized headers. It would be nice if we could find the offending site and get them to update their software. It has to be one of the sites in the path: Path: hao!seismo!philabs!linus!genrad!decvax!harpo!seismo!rochester!lee Of these, I know that hao, seismo, harpo, and rochester don't drop headers. That leaves as candidates philabs, linus, genrad, and decvax. Can someone eliminate any of those sites? (From the list, not the net.)