Forum OpenACS Development: Re: Fixing a very slow query

Posted by Don Baccus on
Statistics-based optimization will always be fooled at times, unfortunately.

Regarding "exists" vs. "in", you're probably remembering advice from long ago, as the PG people figured out how to implement "in" as a special kind of join-like operation some years ago, thus making it as efficient in PG as it had been in Oracle.

In the past "in" was always extremely slow.

"exists" should still be faster, though.