Re: Kudos for Reviewers -- wrapping it up - Mailing list pgsql-hackers

On Wed, Aug  7, 2013 at 12:39:01PM -0700, Josh Berkus wrote:
> Bruce,
> 
> > You are getting into some kind of loop where not wanting to expend
> > unlimited effort on something means, to you, that the person doesn't
> > think the goal is important.   Effort has to be balanced.  This is not
> > the first time I have seen such loops.  And why do you even care about
> > my opinion?
> 
> Aha, OK.  So you're talking about all the different things we might do
> to get more reviewers.  I'm only talking about adding reviewers to the
> bottom of the release notes, which is certainly a bounded activity of
> *very* limited effort.  Which is why I was confused and aghast at your
> talk of "unbounded work".

Well, reviewers on the bottom was just for 9.3 or 9.4, but the final
goal was to get reviewers who modified patches credited with the release
note items.  I asked how that was to be accomplished, and suggested that
the only practical way would be for every committer to check the patch
chain to see who else had modified the patch.  

You suggested something about the commit-fest-manager doing it, and I
couldn't see how that would help because it has to be in the commit
message at the time the release notes are being written.  You said our
release note writing process was not written stone, and that we had to
do whatever it takes to get those names on the items in the release
notes.  At that point I pointed out that there was no consideration of
the effort necessary to accomplish this, and that's how we got here
today.

> To be completely clear: I am talking only about the compromise discussed
> on this thread, namely:
> 
> a) listing reviewers who did "extensive work" as co-authors on the
> patch, and

See above --- I need to know how that is going to get to the release
note items _with_ reasonable effort.

> b) listing other reviewers at the bottom of the release notes.

Yes, that is somewhat easy in that we can get the names from the
commit-fest app, but it doesn't include reviewers who replied via email
but did not record their names on the commit-fest app.  I can tell you
from my release note writing experience that a partial job in this area
is likely to get lots of negative feedback from people who are
excluded.

As an example, I got a pg_upgrade patch fix for 9.2, thought it was
ugly, tried other methods, ended up doing the same thing the original
patch author did, but didn't mention the patch author because I had
forgotten I ended up with the same fix.  When the minor release notes
came out, the person complained on the hackers list, and Simon and Tom
had to apologize as I was away:
http://www.postgresql.org/message-id/CABRT9RAe3zxdins=BBysjqEPA8=NqnxWtA4A0XM01PbdVbmC_A@mail.gmail.com

My point is this has to be done accurately.

> Per earlier discussions.  You started this thread by claiming that
> adding the reviewers to 9.4 would be too hard, and I argued that it
> would not and in fact I'm already working on it.  Nothing I've talked
> about in this thread has been about anything else.

You have to distinguish between names at the end of the release notes,
and names on release note items, and you have to tell us how this going
to happen with reasonable effort.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + It's impossible for everything to be true. +