Over on the bbDev mailing list, Michael Adams started a conversation about the implementation of meta data (post meta, user meta, etc). The current implementation could be improved in a way that would give us better optimization from the database queries, while (I hope) maintaining backwards compatibility from the meta data API functions. Here's the beginning of the thread, for those interested:
http://comox.textdrive.com/pipermail/bbdev/2006-November/000887.html If you follow the thread, you'll see that we're considering a change to the WordPress database schemas to eliminate the meta_id field, and make post_id+meta_key a unique database index. This, in turn, will require 'smushing' multiple keys into a single key with an array of values. Which, is exactly how the PHP data structure for meta data is implemented anyways, so it will just make the DB mirror the in-memory data. Anyhow, I haven't started working on the actual code changes yet, but I did start a Trac ticket, in case anybody has any comments (pro or con) about this idea: http://trac.wordpress.org/ticket/3364 The Trac ticket is probably the best place to comment further about this, rather than cluttering up this mailing list too much. -- Dougal Campbell <[EMAIL PROTECTED]> http://dougal.gunters.org/ _______________________________________________ wp-testers mailing list [email protected] http://lists.automattic.com/mailman/listinfo/wp-testers
