[LWN Logo]

Date:	Fri, 2 Oct 1998 17:10:04 -0400
From:	"Theodore Y. Ts'o" <tytso@MIT.EDU>
To:	Peter Mutsaers <plm@xs4all.nl>
Subject: Re: Linus is on a powertrip..

   From: Peter Mutsaers <plm@xs4all.nl>
   Date: Fri, 2 Oct 1998 22:09:58 +0200 (MET DST)

   I don't know. Those conflicts were from before my FreeBSD time (I was
   using Linux 4 years ago and had been using it for years back then) but
   since at least 4 years I haven't seen any real conflicts at all.

Some of these conflicts have been more recent than that, but you may not
have been aware of them.  Charles Hannum (Mycroft) hangs around MIT a
lot, and I know a number of the NetBSD core team members, so I got to
see some of the backwash of the NetBSD/OpenBSD conflict --- fortunately
only second-hand!  (And at the last Usenix, I was still hearing
mutterings from the NetBSD folks, so it isn't just in the past.)

It was enough for me to be convinced that a potential risk of a core
team "architecture" is the *BSD factionalization.  It doesn't have to
happen, and there are certainly plenty of examples of projects that have
not had this problem --- especially among application-level projects.
It may be that application-level projects are smaller in scope, and
that's why they work out.  It may be that FreeBSD is working out because
somehow (either by accident or by design) FreeBSD doesn't have some of
the more..., ah, combative... personalities which OpenBSD and NetBSD are
blessed with.

In either case, the essential problem with a core team is specifying how
you decide who is a member of the core team, and who is not (which is an
extremely political act), and how do you resolve disputes within the
core team.  In the Linux model, one of the things which works is that we
don't have the argument of who is on the core team, because there is no
such thing (in all of the *BSD cases, the splits occurred when someone
was thrown out of the core team).  Also, we have a relative simply
dispute resolution mechanism --- Linus decides.  Richard and Larry and
others may argue about scheduling changes, or Richard and I may argue
about the appropriateness of devfs, but ultimately Linus gets to make
the final decision.

This all boils down to the old saying that the benevolent dictator is
the best form of government --- there's only one problem: finding the
benevolent dictator.  Linus has, up till now, served as a very good
benevolent dictator.  It may be that the job has been putting much
pressure on him, and we need to find ways of relieving this pressure, or
otherwise solving the problem.

But I am not convinced that the *BSD model is the only solution, or even
the best solution.  FreeBSD may have worked out well, and I'm happy that
you've found a community where that's worked.  But I also see the NetBSD
and OpenBSD, and the conflicts which produced them, and that's also part
of the *BSD model.  Note that for better or for worse, the organization
of the various *BSD's aren't all that different.  The personalities
involved seem to be what makes the most amount of difference.

						- Ted


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/