Re: Sequence Access Methods, round two

On Thu, Jun 20, 2024 at 03:12:32PM +0900, Michael Paquier wrote:
> While on it, I have noticed a couple of conflicts while rebasing, so
> attached is a refreshed patch set.

Please find attached a new patch set for the next commit fest. The
patch has required a bit of work to be able to work on HEAD,
particularly around the fact that pg_sequence_read_tuple() is able to
do the same work as the modifications done for pg_sequence_last_value()
in the previous patch sets. I have modified the patch set to depend
on that, and adapted pg_dump/restore to it. The dump/restore part has
also required some tweaks to make sure that the AM is dumped depending
on if --schema-only and if we care about the values.

Finally, I have been rather annoyed by the addition of log_cnt in the
new function pg_sequence_read_tuple(). This patch set could also
implement a new system function, but it looks like a waste as we don't
care about log_cnt in pg_dump and pg_upgrade on HEAD, so I'm proposing
to remove it on a different thread:
https://www.postgresql.org/message-id/Zsvka3r-y2ZoXAdH%40paquier.xyz
--
Michael