On Fri, 27 Sep 2024 11:44:25 +0100 Dagfinn Ilmari Mannsåker <ilm...@ilmari.org> wrote:
> Michael Paquier <mich...@paquier.xyz> writes: > > > - New functions introduced during a development cycle should use OIDs in > > the range 8000-9999. See 98eab30b93d5, consisting of running > > ./unused_oids to get a random range. > > There's been seen several fixups of this kind recently. Should we add a > note about this to the comment at the top of all of the pg_*.dat files > that have explicit oid assignments? People might be more likely to > notice that than the the section over in bki.sgml. How about adding more to unused_oids output to explain the reason why patches should use consecutive OIDs in the range 8000-9999 and low OIDs should not be used in patches(that is, this minimizes the risk of OID collisions with other patches) instead of just saying it is the best practise. I think patch authors looking for OIDs they can use will run unused_oids, so more likely notice this. Regards, Yugo Nagata -- Yugo Nagata <nag...@sraoss.co.jp>
