I'm in the process of building a tree-structured message base, or rather porting software I wrote back in mid-80s to the web. This isn't the ubiquitous "threaded" BBS software. It's tree-structured. This means that every message on the message base is related through a single ancestor, Msg #0. I've written and re-written the software many times using ISAMs and custom lookup-type indices, usually just keying a parent message pointer in every message. However this is the first time I've done it with SQL. Naturally, the procedural approaches I've used in the past won't do with a result set architecture of a SQL database. To prove it, I tried it. Performance, as expected, was miserable. Is there a way to build table is such a way that a classic Knuth, pre-order tree traversal is possible with a single SQL call? I mean, besides this one: http://www.dbmsmag.com/9603d06.html. That method looked very promising until I realized that adding a new leaf node would mean potentially having to update the right/left counters for every record in the message base. Since a tree can easily grow to 100k messages or more, I don't think that's feasible. -----------------------[ http://www.magpie.com ]-------=o&>o------- Steve Manes Brooklyn, N'Yawk --------------------------------------------------------------------- Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail <[EMAIL PROTECTED]> To unsubscribe, e-mail <[EMAIL PROTECTED]> Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
