Re: pgsql: Don't Memoize lateral joins with volatile join conditions

David Rowley <dgrowleyml(at)gmail(dot)com> writes:
> On Tue, 8 Aug 2023 at 01:02, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Is this really something to be pushing into stable branches less than
>> 12 hours before a release wrap?

> My apologies. I had been under the impression that the go-no window
> was between stamp and tag and *that* was the window that was used for
> a buildfarm cycle. It sounds like it's not quite as black and white as
> that. I'm not aware of any documentation which gives guidance on
> this.

Hmm. Customarily we avoid pushing stuff into due-for-release branches
beginning 48-72 hours before the wrap, and then continuing until the
tags are applied ~24 hours after wrap. The pause before wrap is to
allow a complete buildfarm test cycle even on slower machines. The
pause after wrap is to wait for feedback from packagers.

> * https://wiki.postgresql.org/wiki/Committing_checklist

I thought we had something written down there after the last
discussion of this point, but maybe it was hanging fire because
we hadn't chosen a specific definition of when to start the pause.

> (I'm assuming backing the patch out now won't improve the situation,
> but please correct me if you think I'm wrong on that)

No, I've already made the tarballs. I did look closely at the patch
and decided it was probably harmless and not in need of reversion.
I'm just whining because this wasn't following process.

regards, tom lane