So the trick for MERGE on equality would be to refactor the btree api
so that you could find the matching leaf page and *keep* that page
locked. Then do an update against the conflicting row found there if
any without ever releasing that page lock. Someone else can come along
and delete the row (or have already deleted it) before the update
locks it but if they do you can insert your row without worrying about
anyone else inserting a conflicting entry since you're still holding a
lock on the page they'll need to insert the btree entry on and have
been holding it continuously for the whole operation.
I think that such a lock would also be useful for improving the FK
deadlock issues we have.
--
-- Josh Berkus
PostgreSQL Experts Inc.
http://www.pgexperts.com
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
