Re: Commitfest 2023-09 starts soon

On Mon, 4 Sept 2023 at 18:19, Aleksander Alekseev
<aleksander(at)timescale(dot)com> wrote:
>
> Hi Matthias,
>
> > I'm a bit confused about your use of "consensus". True, there was no
> > objection, but it looks like no patch author or reviewer was informed
> > (cc-ed or directly referenced) that the patch was being discussed
> > before achieving this "consensus", and the "consensus" was reached
> > within 4 days, of which 2 weekend, in a thread that has (until now)
> > involved only you and Peter E.
> >
> > Usually, you'd expect discussion about a patch to happen on the
> > patch's thread before any action is taken (or at least a mention on
> > that thread), but quite clearly that hasn't happened here.
> > Are patch authors expected to follow any and all discussion on threads
> > with "Commitfest" in the title?
> > If so, shouldn't the relevant wiki pages be updated, and/or the
> > -hackers community be updated by mail in a new thread about these
> > policy changes?
>
> I understand your disappointment and assure you that no one is acting
> with bad intentions here. Also please note that English is a second
> language for many of us which represents a challenge when it comes to
> expressing thoughts on the mailing list. We have a common goal here,
> to make PostgreSQL an even better system than it is now.
>
> The patches under question were in "Waiting for Author" state for a
> *long* time and the authors were notified about this. We could toss
> such patches from one CF to another month after month or mark as RwF
> and let the author know that no one is going to review that patch
> until the author takes the actions. It's been noted that the letter
> approach is more productive in the long run.

This far I agree - we can't keep patches around with issues if they're
not being worked on. And I do appreciate your work on pruning dead or
stale patches. But:

> The discussion can
> continue in the same thread and the same thread can be registered for
> the upcoming CF.

This is one of my major concerns here: Patch resolution is being
discussed on -hackers, but outside of the thread used to discuss that
patch (as indicated in the CF app), and without apparent author
inclusion.To me, that feels like going behind the author's back, and I
don't think that this should be normalized.

As mentioned in the earlier mail, my other concern is the use of
"consensus" in this context. You link to a message on -hackers, with
no visible agreements. As a patch author myself, if a lack of comments
on my patch in an otherwise unrelated thread is "consensus", then I'll
probably move all patches that have yet to be commented on to RfC, as
there'd be "consensus" that they should be included as-is in
PostgreSQL. But I digress.

I think it would be better to just remove the "consensus" part of your
mail, and just put down the real reason why each patch is being RfC-ed
or rejected. That is, don't imply that there are hackers that OK-ed it
when there are none, and inform patch authors directly about the
reasons why the patch is being revoked; so without "see consensus in
[0]".

Kind regards,

Matthias van de Meent