Re: Damn slow query

Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> wrote:
>
> We already have a TODO item:
>
> * Allow Subplans to use efficient joins(hash, merge) with upper
> variable

Cool.
One thing to note here is that the JOIN query that Joe suggested is both
faster than the subselect thing (no suprise) but also don't care if
z2test has an index on it or not.
The subselect query started taking huge amount of time again if i
dropped the z2test_x_idx ...

So if the todo could somehow figure out that that subselect should be an
JOIN instead of an NOT EXISTS query, that would be great, because the
index on z2test isn't that super-obvious (i think, because i know the
data is tiny).

Magnus

--
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Programmer/Networker [|] Magnus Naeslund
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-