On Tue, 10 Sept 2024 at 22:46, Michael Banck <mbanck@gmx.net> wrote:
> How about the indislive flag instead? I haven't looked at the code, but
> from the documentation ("If false, the index is in process of being
> dropped, and
> should be ignored for all purposes") it sounds like we made be able to
> piggy-back on that instead?
Doing that could cause an UPDATE which would ordinarily not be
eligible for a HOT-update to become a HOT-update. That would cause
issues if the index is enabled again as the index wouldn't have been
updated during the UPDATE.
I don't see the big deal with adding a new flag. There's even a free
padding byte to put this flag in after indisreplident, so we don't
have to worry about using more memory.
David
