On Tue, 2004-11-23 at 23:39, Josh Berkus wrote:
> I'm on vacation, so I don't have the ability to test the actual SQL.
> The
> article text is fine. You might want to, in future articles, suggest
> that
> people run EXPLAIN ANALYZE and not just EXPLAIN in order to see when
> the
> query planner is estimating wrong.
Ah, good idea, thanks.
> BTW, the queries that GForge uses to generate the category counts for
> the
> trove map are egregiously bad and generate wrong counts to boot.
> They're
> basically a loop of queries, one *per category*, when the whole thing
> could
> be done in a single query.
Heh... I bet that's not the only set of hideous queries in there. Ah
well... onwards and upwards...
> I've been meaning to fix this for some
> time,
> but I got hung up on getting GForge installed on my dev machine. Maybe
> later ...
Maybe I can catch up with you on #postgresql sometime and we can sort of
some of those problems...
Yours,
Tom
