Re: [9.4 CF 1] The Commitfest Slacker List - Mailing list pgsql-hackers

From Josh Berkus
Subject Re: [9.4 CF 1] The Commitfest Slacker List
Date
Msg-id 51D47CDA.5010406@agliodbs.com
Whole thread Raw
In response to [9.4 CF 1] The Commitfest Slacker List  (Josh Berkus <josh@agliodbs.com>)
Responses Re: [9.4 CF 1] The Commitfest Slacker List
Re: [9.4 CF 1] The Commitfest Slacker List
List pgsql-hackers
Michael Meskes wrote:
>> So, as an experiment, call it a mixed result.  I would like to have some
>> other way to motivate reviewers than public shame.  I'd like to have
>
> Doesn't shame imply that people knew that were supposed to review patches in
> the first place? An implication that is not true, at least for some on your
> list. I think I better not bring up the word I would describe your email with,
> just for the fear of mistranslating it.

If you didn't feel obligated, you wouldn't be pissed at me.  You'd just
blow it off (like Bruce did).  I think you're angry with me because you
feel guilty.

My *personal* viewpoint is that all committers should feel obligated to
review and commit patches from other contributors.  That's why they're
committers in the first place.  Certainly if a committer looks at the CF
application and notices that 80% of the reviewing and committing is
being done by three people, none of whom have any more "spare time" than
they do, they should feel obligated to help those people out.

We have a problem with patch reviewing and committing in this project;
it's not being done in a timely fashion in general (every CF last year
ended late), and the people who are doing most of the work feel
overworked and frustrated.  This problem is getting worse every year,
and will kill the project if it continues on its current trajectory.

There are *only* three ways out of this hole, all three of which I'm
trying to address:

1) more automation and better tools in order to reduce the total time
required of each reviewer/committer;

2) a program of recruitment of new reviewers, including giving respect
and recognition to people for their reviewing efforts

3) getting most of our existing contributors to shoulder their fair
share of patch review.

(3) is what I'm addressing on this thread.  The reason I volunteered to
be CFM this time was directly because of our discussion in Ottawa of how
the review process wasn't working.  I decided to find out *why* it
wasn't working, and the first obvious thing I ran across was that most
of our current and our long-term contributors weren't doing any patch
review.  For CF1, the number of people submitting patches outnumbered
those who had volunteered for review 2 to 1.  That *is* the review
problem in a nutshell; everybody wants someone else to do the work.

I don't think it's too much to ask people who are listed on the project
developers page as major contributors to review one patch per CommitFest
most of the time.  If they did just *one* it would substantially
decrease the workload on the people who are currently doing the vast
majority of review and commit.

On 07/03/2013 11:24 AM, Cédric Villemain wrote:
> You're looking at a short term, big effect.
> And long term ? Will people listed still be interested to participate
in a
> project which stamps people ?
>
> With or without review, it's a shame if people stop proposing patches
because
> they are not sure to get time to review other things *in time*.

Yes, I am, because the CF is only supposed to be 30 days long, and I
plan to finish it on time.  That's my job as CFM.

Several people on this thread have raised the fear of discouraging patch
submitters, but the consistent evidence is that we have more submissions
than we can currently handle.  I'd rather have half as many submissions,
but do a really good job of reviewing, improving, and integrating those
than the current mess.

Furthermore, there are quite a number of people who are submitting
patches on paid company time.  For those people, "submit one, review
one" has to be an ironclad rule so that they can tell their bosses that
they *have* to spend time on patch review.  Otherwise, the review
doesn't happen.

--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: refresh materialized view concurrently
Next
From: Cédric Villemain
Date:
Subject: Re: [9.4 CF 1] The Commitfest Slacker List