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

Re: Some proposals

On Wed, Feb 26, 2003 at 11:54:48PM -0500, Jaldhar H. Vyas wrote:
> Seeing as everyone else is throwing in their two currency units, here are
> some ideas i've been kicking around.

[ snip ]

> Proposal: put all Debian code in a repository and make all maintainers
> responsible for all packages.
> No more waiting for absentee maintainers to fix trivial bugs in their
> packages.  The BSDs do it and manage to put out regular releases.  There
> may need to be levels of privilege i.e a core team but I would argue there
> is a de facto core Debian team right now anyway.  As long as there is a
> good communication and a clear audit trail for when thngs go wrong I think
> it could work.

Most folks here probably know I rather like a bunch of the stuff the BSD
folks do. Regular (time-based) release cycles, tight code, and lots of
other stuff. Sure. Good. Keen.

The 'core team' that is responsible for what is formally released is
relatively small, *and* it has a relatively small amount of stuff to take
care of. Combine the kernel and GNU libc teams, maybe throw in about half
the current X Strike Force (biased towards the upper end). That's it - and
that's all the software they actively maintain and promise to try to make

The entirety of /usr/ports on FreeBSD or /usr/pkgsrc on NetBSD is done
by the non-core team, and, as a whole, is not in any better state than
Debian's 'testing' usually is (barring artifacts caused by our promotion
scheme from unstable). Sometimes it's closer to unstable itself. And any
given port is, in fact, frequently only looked after by one or two people.

I'm all for consider questions such as "should Debian have an official core
team, and some official packages like dpkg or apt that are shared among
all of them". I don't claim to know the answer, nor even that the current
situation is broken, but it might be a reasonable question. But what you're
proposing, overall, is not actually how the BSDs appear to work, and for
that matter, runs afoul of some fairly basic observations of how humans
tend to operate en masse.

(When there is no responsibility, things either never get fixed, or they
get fixed in broken ways that nobody can track down the cause of, for
one example - Developers put their reputation on the line in doing their
packages, and for most of them, that *is* motivation to try to do work that
Doesn't Suck - and, for that matter, to do work at all).
Joel Baker <fenton@debian.org>

Attachment: pgpCXvNaM96Zr.pgp
Description: PGP signature

Reply to: