Tom Lane, 19.07.2012 16:52:
> If you're using a reasonably recent version of PG, replacing the NOT IN
> by a NOT EXISTS test should also help.
Thanks. I wasn't aware of that (and the NOT EXISTS does indeed produce the same plan as the OUTER JOIN solution)
>> Now I was wondering if a DELETE statement could be rewritten with the same "strategy":
>
> Not at the moment. There have been discussions of allowing the same
> table name to be respecified in USING, but there are complications.
Thanks as well. It's not a big deal for me. I was just curious if I missed something.
Regards
Thomas