Simon Riggs <simon(at)2ndquadrant(dot)com> writes:
> OK, that sounds good. Are you also working on transforming NOT IN into
> different form? Or is that the same thing as (1)?
I'm not currently thinking about NOT IN. It could be transformed to
an antijoin if we could prove that no nulls are involved, but that
seems less than trivial as I noted earlier.
regards, tom lane