Hi!
> Thanks for looking at this. This query seems to be the same as the
> first query in my initial report (with a different targetlist) and has
> already been covered in the test cases of v1 patch.
Sorry for the confusion. Yes, it's the same query as yours. I just
wanted to note that without asserts it silently returns wrong results
(1 row instead of 4) rather than raising an error, in case that's
useful for the back-patch decision.
Regards,
Tatsuya Kawata