geronimo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Greg Stein <gst...@apache.org>
Subject committer process
Date Thu, 11 Sep 2003 23:41:10 GMT
On Wed, Sep 10, 2003 at 06:02:51PM -0000, geronimo-dev-digest-help@incubator.apache.org wrote:
>...
> From: "Jeremy Boynes" <jeremy@coredevelopers.net>
> Subject: [vote] Process for adding committers
> To: <geronimo-dev@incubator.apache.org>
> Cc: <dims@yahoo.com>
> Date: Wed, 10 Sep 2003 10:21:21 -0700
> 
> With two options on the table, I think we need to put this to bed quickly so
> I am calling for a vote between the two following options:
> 
> Option #1 from Davanum Srinivas:
>...
> Option #2 from Ryan Ackley:
>...

I believe both of these are incorrect forms for voting in new committers.
My original statement was:

> I would recommend coming up with, say, a list of four people and submit
> that list to pmc@incubator for consideration. How about by end of week?

The implied part here is that you're sending a list to the PMC. The PMC
then decides *in private* on whether to add the new committers, and whom.

Currently, there are two general forms of adding committers at the ASF. I
believe one works, and the other fails.

1) httpd/apr style: somebody on the PMC sends a nomination to the PMC list
   and the PMC votes privately.

2) jakarta style: a committer sends a nomination/vote to the public dev
   list and the other committers vote publicly.

There are two primary differences (and problems) here:
a) public vs private voting
b) committers vs PMC members

With regard to (a), the biggest issue is that you will *very* rarely see
any -1 votes. People just don't want to do that in public. Especially in a
healthy and cooperative community. It is very hard for somebody to say,
"yes, they're good, but they just don't "get it" very well." What'll
happen is that they'll simply abstain. Or maybe vote, but without comment.
And dropping the comments means that other people don't get a chance to
consider those words and how they feel about them, and whether they may
resonate with those thoughts and want to change their own vote.

Public votes about *people* just don't work. That needs to be done
privately for it to be successful. This is exactly why most countries have
secret ballots: to avoid repercussions against how people vote.

With regard to (b), the set of committers does not always match the set of
people who are *responsible*. The Jakarta project is pretty bad about
this: the set of committers is way larger than those responsible (the
PMC). Within the Incubator, the separation is actually quite prevalent:
the committers are generally not on the PMC, yet the PMC is the group that
is responsible for this project. By moving the vote explicitly onto the
PMC's private mailing list, you are putting the vote in front of the
people who are ultimately responsible.

In my experience, I have seen *many* private votes raise concerns which
are very valid. The usual result from these "edge" cases is that the group
simply decides to wait for a while and see where things go (with some
coaching for the person to help them remedy the concern). But I have never
seen a public vote raise the same kinds of concerns and the resulting
discussion. Mostly out of consideration for the feelings of the nominee,
but also so that the "contrary" position-holder is not embarrassed to post
their concerns. Net result: I see the private form succeed, and the public
one fail.

Consider the situation where somebody who gets voted in as a committer via
the public mechanism. Since there isn't any real strong way to vote
*against* a person, then they will usually end up as a committer. Almost
by default. All you need is a *single* person to put a nomination out
there, and "ta-da!" you're in. Now what happens when you discover that it
really was a mistake. That you didn't have an opportunity to discuss the
problems with that committer. Now you have the unfortunate task of
figuring out how to fix that committer, or to remove their access. And
trust me: you *REALLY* don't want to go down that path. It is so
incredibly painful, that you want to have a very high confidence that when
you make somebody a committer, that they will make for a great committer.
While the "six month rule" doesn't truly exist, it is a very good
benchmark for being able to see how somebody interacts with the community
over a long period of time, to see whether they are dedicated, and to
get a good look at the quality of their code.

The process of voting in new committers is a choice of the PMC. In this
case, that is the Incubator PMC. *However*, the intent is that the
Geronimo project will receive its own PMC which means that it will be able
to define whatever process it would like. Thus, I think the Incubator PMC
will defer to the community to choose the process that it would liek to
use. They will certainly help the community to choose a process that fits
in with the meritocratic principles of the ASF, and to show some of the
forms in use at the ASF.

I would request that the community also considers and votes on my proposal
to use a private voting system. The mechanics of that would be a private
email to the PMC to nominate somebody for commit access.

Thanks,
-g

-- 
gstein@apache.org ... ASF Chairman ... http://www.apache.org/

Mime
View raw message