Bruce Momjian <maillist@candle.pha.pa.us> writes:
> I think I have solved the optimizer problem. It appears in samekeys().
> Can someone check that function, and see if you come up with the same
> fix I do (without knowing my fix)?
"member" -> "equal", perhaps?
I looked at that before and thought it was a little strange, but I
didn't and still don't understand the data structures being compared.
I also wondered whether the two lists ought not be required to be
exactly the same length, rather than allowing keys2 to be longer.
> A 9-table join that used to run for minutes and fail now completes in
> seconds!
Pick some smaller joins and see whether the optimizer still finds
the same answer...
regards, tom lane