On Sun, Jul 17, 2022 at 10:58:26AM +0100, Simon Riggs wrote:
> Sounds great, looks fine. Thanks for your review.
Ok, cool. At the end, I have decided to split the tests and the main
patch into two different commits, as each is useful on its own. Doing
so also helps in seeing the difference of behavior when issuing a
REINDEX DATABASE. Another thing that was itching me with the test is
that it was not possible to make the difference between the toast
index of the catalog and of the user table, so I have added the parent
table name as an extra thing stored in the table storing the
relfilenodes.
--
Michael
