On Mon, Jul 02, 2018 at 02:07:37PM -0400, Robert Haas wrote:
> I'd rather keep an elog(ERROR) than completely remove the check.
+1.
> Also, for the record, I think the subject line of Michael's commit
> message was pretty unclear about what it was actually doing.
How would you formulate it? Perhaps the error message did not emphasize
enough on the fast that it actually blocked a behavior, say "Block mix
of temporary and permanent relations in partition trees" or such?
--
Michael