From: Paul J Stevens <dbmail-dev@dbmail.org> <snip> > What I like about the yukatan approach is that it tries to make searches on very common headers as cheap and > fast as possible. It does this by going one step beyond separate header storage: preparse certain headers for > common attributes: the in-reply-to and references headers will be used for threading. These headers contain > one or more message-id header values, which are stored separately.
Isn't this exactly what the header/message cache will do, as well? With a nice, generic setup, I had hoped to do exactly that - cache the In-Reply-To and References headers to enable quick/easy message threading in weDBmail. I guess I was just wondering if this whole thread was in reguard to simultaneously implimenting per-message caching as well as improved header storage, or if there're two means to the same end being developed. -- Jesse Norell jesse @ kci.net