[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: draft alternative proposal: fix problem at the root



Stefano Zacchiroli <zack@debian.org> writes:

> From those who want to drop the CTTE, I'd like to know what would they
> have done to decide upon the init system for Jessie. It seems to me that
> we have tried not to use the CTTE on that dispute for several years,
> with the net result of raging flamewars and no decision. These days we
> still have (now dissipating, IMO) flamewars, but at least we have a
> decision. And the CTTE seems to have reached that decision after having
> studies very deeply the problem (
> https://lists.debian.org/debian-ctte/2013/12/msg00234.html and similar
> messages come to mind).

> The only alternative to a CTTE seems to be using GRs to settle otherwise
> non-settlable disputes. But that (old arguments follow) seems to go
> against the wisdom of not using GRs for technical decisions, and also
> against the common sense of turning technical disputes into popularity
> contests.

There's another alternative to using the CTTE, and my understanding is
that this was generally the method used prior to the existence of the
CTTE, but I'm not sure it's really any better.

There are specific teams in Debian, generally delegates at this point,
that sit at choke points and can effectively make decisions like this as
part of the course of their duties.  For example, for the init system
decision, I suspect a lot of the pressure would have been on the d-i team
to pick a default.  In the past, the usual victim for many of our disputes
has been ftp-master, since they can block archive uploads or eject things
from the archive.  For others (and I can recall some epic ones), it was
the DAM.

I actually seriously considered proposing eliminating the CTTE a few weeks
ago, rather than simply resigning, because I felt like I was leaving
problems unsolved and just walking away from them and leaving them for
other people to have to deal with.  But after giving it a lot of hard
thought, I couldn't convince myself that the results would be better.
That doesn't mean they wouldn't be, only that I'm personally quite
uncertain.

I don't believe the likely result of eliminating the CTTE would be more
GRs (although the init system debate almost certainly would have gone to a
GR, for better or for worse).  Rather, I think it would be pressure on the
various Debian core teams to make those decisions.  Maybe that's better,
but I'm not sure it is.  It has the possible advantage that those teams
are more accountable to the project in a sense than the CTTE is, since
they're mostly project delegates and the delegation can be revoked
(although that's pretty much the nuclear option, and I'm dubious it would
ever be used).  It has the significant disadvantage that those teams are
trying to get work done that doesn't involve resolving some major dispute,
and may be quite unhappy about the pressure and negative attention that
comes from being thrust in the middle of them.  It's already hard to find
enough people to work on the Debian installer; what would happen if those
people knew they'd be responsible for deciding things like the default
init system too?

Basically, I think eliminating the CTTE would just make the lives of
ftp-master, DAM, the release team, the d-i team, and other similar core
teams miserable.

Making contested decisions, or (better, and what I wish would happen all
the time, but which isn't always possible) finding ways to make the
decision less contested is pretty stressful and difficult and requires a
lot of time and emotional energy.  The advantage of the CTTE structure in
theory is that you can have a group of people who signed up for that work,
not for some other job that happened to cross paths with some hard
decision.  I think this gets much better with term limits, since it
changes the nature of that role.

If we keep the CTTE, I would encourage the project, and the CTTE, to stop
thinking of it as our council of elders, and start thinking of it as jury
duty.  (I know that's not a universal concept among the countries
participating in Debian, but hopefully everyone is familiar enough with it
or similar duties like lay judges to get the reference.)  It's hard and
draining, but worthwhile and something that the project needs to have a
way to do.  But any one person probably shouldn't do it for a really long
time, because it *is* so draining.  Instead, those of us who feel like we
can do that work should consider rotating through it as a way of serving
the project for a while, and then stepping down and going back to doing
what we were doing before.  At least, that's the way that I feel about it
for me.

In that sense, it would be a lot more like being the DPL.  Being the DPL
is a ton of work, people put their other Debian work on hold to do it, and
no one expects to be the DPL for an extended period of time.  You come in,
serve the project in that capacity for a year or two or three, and then
step down and go back to whatever you were doing or pick up something new
and let someone else take the load.

(That said, I feel like I should add here that there are people who *have*
done this role for a long time, and I think very well, and I don't want to
imply something else.  These comments are in the abstract, or in general
terms.)

-- 
Russ Allbery (rra@debian.org)               <http://www.eyrie.org/~eagle/>


Reply to: