Graham, that could have something to do with it. It's surprising though, because postgres 8.3 is smarter in dealing with the LIMIT and avoiding unnecessary sorting etc. Perhaps this is a performance regression associated with that.
On this query I see no fewer than 3 sequential scans: on BugTask, Bug, and SourcePackageName. We may have an incredilicious database setup but it's still a lot to ask for. The SourcePackageName one is cheap though, so jettisoning the prejoins probably doesn't help.
Graham, that could have something to do with it. It's surprising though, because postgres 8.3 is smarter in dealing with the LIMIT and avoiding unnecessary sorting etc. Perhaps this is a performance regression associated with that.
On this query I see no fewer than 3 sequential scans: on BugTask, Bug, and SourcePackageName. We may have an incredilicious database setup but it's still a lot to ask for. The SourcePackageName one is cheap though, so jettisoning the prejoins probably doesn't help.