Philip Warner <pjw(at)rhyme(dot)com(dot)au> writes:
> The optimizer seems to want to use sequential scans on inherited tables
> when crossed with another table, as the following seems to demonstrate:
Is it intentional that your test case omits an analyze on t2? Coz when
I add that, I get the same plan you show for 8.0. Without the knowledge
that t2 is small, that plan is not a good choice.
(The larger point that joins of inheritance unions aren't well-planned
is true, but it's always been true...)
regards, tom lane