Re: pgsql: Remove unused wait events.

On Mon, Oct 25, 2021 at 12:38 PM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Um ... the removed symbols were at the end of the WaitEventIO enum,
> so is there really an ABI break? I suppose if an extension contains
> actual references to the removed symbols, it would fail to recompile,
> which'd be annoying for a released branch.

I think that you're right. I believe one of the two extensions I know
about hopes that values won't be renumbered or become invalid across
minor releases, and the other contains specific references to these
particular constants.

Now of course it is always arguable whether or not anything that some
extension is doing ought to be deemed an acceptable use of the
facilities provided by core, and how much stability ought to be
guaranteed. But while I agree it's good to remove unused stuff in the
master, it doesn't seem like we really need to back-patch it.

--
Robert Haas
EDB: http://www.enterprisedb.com