Re: BUG #15437: Segfault during insert into declarative partitioned table with a trigger creating partition

On Tue, Oct 30, 2018 at 01:11:19PM +0900, Michael Paquier wrote:
> That seems to be what we are looking for. Let's wait a couple of days
> and see if anybody else has any input to offer on the matter. I also
> would like to think about it again once.

And so I did.

s/paritioned/partitioned/ in the new comment of tablecmds.c.

The tests could be designed better. We had better not use the same
object names across multiple tests. If for a reason or another both
test suites are moved to the same series when running in parallel, this
can lead to race conditions which would be annoying to debug. Test
cases are not much portable as they rely on default partitions and this
bug happens down to v10, where we need to back-patch. It can be
changed so as we use FOR VALUES (1) or such in the DDL part of the
function.

Those are easy enough to fix and improve, and the patch looks good to
me. Any objections to commit and back-patch before the next point
release?
--
Michael