Re: Query plan degradation 8.2 --> 8.3

Greg,

> How recently did you check out your 8.3 tree?

It's the snapshot from 5/28, which means it was pulled from CVS on 5/27.
So, recent.

> When I run it I get a bitmap index scan which I think might mean you're
> suffering from the same problem Tom found and fixed a few days ago. The
> planner is finding the bitmap index scan with the sort is the best
> possible plan but then discarding that option later leaving it with a
> suboptimal choice.

Apparently. I'll do another build and check.

> It does the right thing if t_s_symb is declared as text instead of
> varchar. When it's varchar, even setting enable_sort off won't make
> it pick the right plan, which suggests that it fails to recognize that
> the index can match the query's ORDER BY. I'm guessing I overlooked
> a binary-compatibility case when I rejiggered the handling of PathKeys
> in connection with the NULLS FIRST/LAST stuff. No time to look deeper
> right now.

Yeah, that looks like the case. We'll move it to TEXT for the tests right
now, but I'll make sure we don't forget this bug during beta. Thanks!

--
--Josh

Josh Berkus
PostgreSQL @ Sun
San Francisco