Robert Haas <robertmhaas@gmail.com> writes:
> I'm also not sure how skillful the constraint exclusion logic is at
> proving theorems when CASE statements are involved.
It's not at all, and unlikely to become so. However, I think worrying
about that might be focusing on the wrong thing. If this patch expects
us to still use theorem proving to handle partition exclusion, it's
going at things in the wrong way from the get-go. The partitioning
rules should be simple enough that they can easily be applied at runtime
to determine which partition to look in.
Which leads me to the same conclusion: anything as complicated as CASE
is the wrong design. But perhaps for slightly different reasons.
regards, tom lane