Michael Day <blake(at)rcmail(dot)com> writes:
> I was able to reproduce with this set of data.
Ah, thanks for the test case! Looks like the handling of
JOIN_UNIQUE_INNER cases for parallel plans is completely confused ---
it forgot about needing to unique-ify the inner rel, and I'm rather
surprised it didn't trip over some Asserts while at it.
Will fix, thanks.
regards, tom lane