On Tue, Aug 12, 2025 at 08:58:47AM +0900, Michael Paquier wrote:
> I would suggest waiting a bit for the resolution of the other bug, and
> make sure that we have at least one reference to the trick with the DO
> block. While the way you are writing the test is more readable,
> there's also a case for keeping documented this trick in the tree in
> the shape of a SQL sequence that does not depend on \gset and psql.
I was looking again an this thread, and this argument feels a bit off
because we have a test already posted on the thread, and there would
be some past history. There is also the ease of debugging failures
when the commands fail in the DO block. At the end, applied 0002 on
HEAD.
--
Michael
