Re: Misc web page changes, possible breakage
On Fri, Dec 17, 2004 at 10:26:19AM -0500, Jay Berkenbilt wrote:
>
> > > > > I would like the site much better if the "Applicants waiting for
> > > > > DAM approval" section of the page were shorter. :-)
> > > >
> > > > Indeed... that would be great improvement ;)
> > >
> > > Certainly. This concerns me as well.
> > >
> > > To me it looks like no applicants have been approved since early March
> > > this year. That is almost 10 months! Can this really be?
> >
> > That's not true. 9 applicants have been approved in August.
> > But previous applicants have been proceeded before March as you said.
[...]
> I think many of us who are either waiting for our accounts to be
> created, ftp.debian.org bugs to be handled, or buildd-related issues
> to be resolved are secretly wondering whether we have a problem of
> human bottlenecks, but don't want to say anything about it in public
> for fear of jeopardizing our NM status or otherwise prematurely
> rocking the boat. To me, the fact that several seemingly unrelated
> things are all backlogged in part because of the same people being
> behind is simply an indication that those people are overcommitted.
[...]
Yes, that does seem to be the case.
[...]
> In other areas, such as package maintenance, there are mechanisms in
> place that prevent a single person from becoming a bottleneck.
> Perhaps we need a similar mechanism in the NM process.
Having an NMU-like process for overriding the DAM seems like a
horrifically bad idea.
However, I think reducing the bottleneck is very important. I've been
trying to push for opening up the application review process more so
that more people can be involved. Currently, the process works
something like this:
1. AM submits a short (aka useless) report to debian-newmaint and
submits a full report (including mail logs, advocate and AM comments) to
both the front desk and DAM.
2. The front desk (currently comprised of one busy person) reviews the
report, and if it looks OK, writes a brief summary for the DAM.
3. The DAM (currently comprised of one even busier person) then must
review the whole report and make a judgment on it. Anyone who has gone
through the NM process knows these reports are very long--as you might
guess, they also take a long time to review.
As you can see, there is very little interaction. The onus of reviewing
reports lies solely on the front desk, and then the DAM.
I'd like to see the process more open so that others may participate. I
envision something similar to how licenses are reviewed on
debian-legal--licenses are sent to debian-legal questioning their
freeness, and it is discussed among the participants until a rough
consensus is reached. If any particular developer is too busy to
participate, they simply don't and no harm is done. This sort of system
could be applied to the NM process something like this:
1. The full AM report is sent to the whole NM committee or even the
whole Debian developer population. I don't know if opening it to the
general public would be reasonable.
2. Any developer who has the time and energy may review a report and
offer questions, comments, and critiques. The developer may choose to
only make a partial review of the sections of the report that are most
interesting to him or her, or whatever.
3. Hopefully several people perform their own reviews, some discussion
is generated, and after some time, a consensus is made that either
the report is complete and the NM is ready, the report is incomplete
and more info is needed to come to a judgment, or the report is
complete but the NM simply isn't suitable for Debian.
4. The front desk and DAM may monitor and participate in the discussion,
and ultimately reach their own judgments based on the generated
discussion.
There would be a few advantages that I see.
* The front desk and DAM don't have to perform reviews in isolation.
Other developers can ask the difficult questions for them, so that all
the blanks in the report are filled in. And, it's a look easier to
reject a report if several other people speak out that a report sucks.
* AM reports are shared between AM's. With the current process, AM's
never see each other's reports. So, there is little opportunity to
learn from each other how to compose a thoroughly complete report.
* No structural changes are required. The current front desk and DAM
positions may remain in place. However, hopefully much of the work is
offloaded from them so that both are less prone to becoming
bottlenecks.
* "Many Eyes Make All Bugs Shallow." That philosophy works great in
free software development--couldn't it work for the NM process as
well?
If it doesn't work out as I hope, I don't think any harm would be done.
If anything, it would make the current bottlenecks more
understandable--if no other developers are willing to review reports,
why should we complain if the front desk and DAM are loathe to do it as
well?
--
For every sprinkle I find, I shall kill you!
Reply to: