Re: [PATCH] RFC: new_oid_hook, to let an extension propose catalog OIDs

On Wed, Oct 28, 2026 at 01:25:03PM +0500, RekGRpth wrote:
> - Is a hook here acceptable at all, or would you rather extend the
> binary-upgrade mechanism ("the next OID for this catalog is X") to
> normal operation?

I don't like the idea of a hook for that. That seems clumsy as it
requires the hook to be loaded on the target server.

If you wish to fix an OID in time for a custom catalog, cannot it be
something that could also be done on the extension side? Or is that
really a case common enough out there that it justifies an in-core
facility to support in core? If that were common enough, not having
to duplicate the same facilities across a few dozen extension could
make sense, yes, though I doubt that a hook is the adapted interface
for that. Your idea of a binary-upgrade-like-mechanism similar in
some way to the existing object dumping for extensions would make more
sense, the idea being to ease the move of the extension object *across
major versions*.

> Co-Authored-By: Claude Opus 5.5 <noreply(at)anthropic(dot)com>

Okay....
--
Michael