Re: Explicit psqlrc - Mailing list pgsql-hackers

From Robert Haas
Subject Re: Explicit psqlrc
Date
Msg-id AANLkTilzgsTG3rF-yyOosMV0yf0sFsLHdQmSNbXKXgrZ@mail.gmail.com
Whole thread Raw
In response to Re: Explicit psqlrc  (Simon Riggs <simon@2ndQuadrant.com>)
Responses Re: Explicit psqlrc
Re: Explicit psqlrc
List pgsql-hackers
On Tue, Jul 20, 2010 at 3:20 AM, Simon Riggs <simon@2ndquadrant.com> wrote:
> On Mon, 2010-07-19 at 23:40 -0400, Robert Haas wrote:
>
>> Since it has been over a month since this review was posted and no new
>> version of the patch has appeared, I think we should mark this patch
>> as Returned with Feedback.
>
> Mark posted a new patch 6 days ago, AFAICS.

Hmm.  I can't find it in my mail, in my archives, or linked off the
CommitFest application.  Was it posted under a different subject line?Do you have a link to the archives?

> Not sure I see any benefit in doing as you propose anyway, or at least
> not without warning since it just confuses authors who may believe they
> have time while the commitfest is still on.
>
> Commitfests were created to help authors. We must continue to remember
> that 99% of patch authors are not full time and likely to find smooth
> responsiveness difficult.
>
> We should say "Will there be any further action on this patch?"

It isn't the purpose of CommitFests to provide patch authors with an
unlimited right to have their patches reviewed.  They exist, on the
one hand, to make sure that patches don't linger forever without
getting a review; and on the other hand, to create a defined time for
each member of the community to set aside their own work and
review/commit other people's patches.  It is important that we have
them, and it is also important that they end, so that reviews,
committers, commitfest managers, etc. can stop working on the CF at
some point and get back to their own work.  In other words,
CommitFests need to represent a balance between the needs of authors
and the needs of patch reviewers and committers.

Of course, anyone is always welcome to review a patch that has been
posted, and a committer can decide to work on a patch at any time
also.  But if patch authors are entitled to resubmit a previously
reviewed patch up until the very last CommitFest are *guaranteed* a
possible review and commit even then, then the CommitFest will not end
on time.  Even if the CommitFest does end on time, more than 50% of
the time between now and 9.1 feature freeze is CF-time - that is, time
I'm supposed to be putting aside the work I'd like to get done to help
other people get the work they'd like to do get done.  I'm not really
willing to increase that percentage much further.  I feel it's
important that we give everyone a fair shake, and I have devoted and
will continue to devote a LOT of time to making sure that happens -
but I want (and if I care to still be employed, need) some time to do
my own work, too.

To me, the definition of a fair shake is that people get 4-5 days to
respond to review comments.  This patch has had 33.  It's not unfair
to anyone to say, you know, since you didn't get around to updating
this patch for over a month, you'll need to resubmit the updated
version to the next CommitFest.  If we have the resources to review
and commit a late resubmission of some particular patch, that is
great.  But as of today, we still have 32 patches that need to be
reviewed, many of which do not have a reviewer assigned or which have
a reviewer assigned but have not yet had an initial review.  Since
there are 26 days left in the CommitFest and many of those patches
will need multiple rounds of review and much discussion before we
decide whether to commit them, send them back for rework, or reject
them outright, that's pretty scary.  To me, that's where we should be
focusing our effort.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company


pgsql-hackers by date:

Previous
From: Marko Kreen
Date:
Subject: Re: standard_conforming_strings
Next
From: Pavel Stehule
Date:
Subject: Re: patch: preload dictionary new version