From: m...@gnu.ai.mit.edu (Michael I. Bushnell, p/BSG)
Subject: A short tale
Date: 1996/05/31
Message-ID: < MIB.96May31174422@gnu.ai.mit.edu>#1/1
X-Deja-AN: 157796263
organization: Touring Consulting Services
followup-to: gnu.misc.discuss
newsgroups: gnu.misc.discuss



Many years ago, a man named Simon M. Richardson (smr) had an idea.  He
was going to write a complete operating system, called "UNG".  He
enlisted the help of many volunteers who wanted to cooperating in the
writing of UNG.  Over time, he and ther wrote and released many parts
of UNG, though the complete UNG system had not yet appeared.  

In fact, it could have appeared much more soon, but it turned out that
the pieces of UNG (not unnaturally called `UNG software') were useful
on many other systems, not just the UNG system, and that the task of
making them portable and runnable on many other systems was
important, and a very helpful way of motivating volunteers to the
ongoing task of completing the UNG system.

Many years later, a young Italian, by the name of Tomaso Lupero,
wanted to learn more about how 680x0 processors did context switching,
and began writing a little kernel.  He plugged this kernel into some
components written by a poser named Trunk Applebunk.  Applebunk had
previously written a toy, and called it a "teaching tool".  Tomaso
Lupero found Applebunk's toy system to be a useful way to start
testing and playing with his new little kernel.

Work expanded, and one day Tomaso posted a message about his new
kernel (which he called Tonux).  People liked the idea of Tonux,
because they could play with it, and extend it.  It was a lot more fun
than Applebunk's toy, and Tomaso hadn't attached obnoxious
restrictions to the use and distribution of the code.  In fact,
because he had used some UNG tools in its development, he decided to
discribute the Tonux kernel under the same license as the UNG tools.

Some people decided that it would be a good thing if the Tonux kernel
were used as part of a complete operating system.  So they put one
together, using all of the UNG tools, and the Tonux kernel.  Because
UNG wasn't complete, they also got tools from other sources,
principally from the W Window System and things from the popular VSD
(Virginia Software Distribution) Unix developed at Raleigh-Durham.

Simon Richardson was busy building UNG, so he didn't pay much
attention.  But this new phenomenon, of systems built around the Tonux
kernel, was catching like wildfire.  So smr looked at these systems,
and discovered (mirabile visu!) that these systems were exactly what
he had been thinking of as UNG systems.  In fact, Richardson had
already decided to use exactly the same other tools (like W Windows,
and VSD software) as important parts of UNG.  So it was quite
striking--here these people had, in fact, helped to complete the UNG
system.

Today, we have the following situation.  UNG is nearly complete now,
and the pieces of it are, in fact, nearly identical to the ones in the
systems using Tonux kernels.  And Richardson said "I think these
systems should be named Tonux/UNG, because the systems look just like
UNG systems, except that they use the Tonux kernel".  And some people
say "we were not creating UNG systems, we were creating Tonux
systems".  But that's wrong--they were creating UNG systems, the UNG
system had been designed and what these people had put together was,
in fact, the very UNG system that Richardson announced many years
ago--except, of course, that it used the Tonux kernel, rather than the
UNG kernel.

Michael

From: c...@loiosh.kei.com (Christopher Davis)
Subject: Re: A short tale
Date: 1996/06/01
Message-ID: < w4ennz8kpn.fsf@loiosh.kei.com>#1/1
X-Deja-AN: 157936062
sender: c...@loiosh.kei.com
references: < MIB.96May31174422@gnu.ai.mit.edu>
content-type: text/plain; charset=US-ASCII
organization: Hellspark Trading Association
x-attribution: ckd
mime-version: 1.0 (generated by tm-edit 7.66)
newsgroups: gnu.misc.discuss


Is my (fallible) memory failing me, or wasn't there some discussion here
on gnu.misc.discuss a while ago about "Why the FSF is not going to make a
GNU system based on the Linux kernel"?

(IIRC, it boiled down to "The HURD will be ready Real Soon Now, so we see
no need to do it.")

From: m...@gnu.ai.mit.edu (Michael I. Bushnell, p/BSG)
Subject: Re: A short tale
Date: 1996/06/01
Message-ID: < MIB.96Jun1162149@gnu.ai.mit.edu>#1/1
X-Deja-AN: 157958139
references: < MIB.96May31174422@gnu.ai.mit.edu> < w4ennz8kpn.fsf@loiosh.kei.com>
organization: Free Software Foundation, Cambridge, MA
newsgroups: gnu.misc.discuss


In article < w4ennz8kpn....@loiosh.kei.com> c...@loiosh.kei.com (Christopher Davis) 
writes:

   Is my (fallible) memory failing me, or wasn't there some discussion here
   on gnu.misc.discuss a while ago about "Why the FSF is not going to make a
   GNU system based on the Linux kernel"?

   (IIRC, it boiled down to "The HURD will be ready Real Soon Now, so we see
   no need to do it.")

We were going to release such a system in cooperating with the Debian
project, and we spent some money paying Ian Murdock to help do that.
But he left Debian, and the remaining people didn't want to work
together with the FSF.

We wanted to acquire experience with doing such distributions first
hand this way.  The value of the project would have been that we would
learn how to do such projects well.  So doing it now, by ourselves,
wouldn't help anything--we would not be learning from direct
collaboration with some of the people who have a good understanding of
the problem.

Michael

From: b...@pixar.com (Bruce Perens)
Subject: Why Debian split up with FSF
Date: 1996/06/04
Message-ID: <4p2ch0$155@pixar.com>#1/1
X-Deja-AN: 158462167
references: < MIB.96May31174422@gnu.ai.mit.edu> < w4ennz8kpn.fsf@loiosh.kei.com> 
< MIB.96Jun1162149@gnu.ai.mit.edu>
organization: Pixar - Pt. Richmond, CA USA
newsgroups: gnu.misc.discuss


From: m...@gnu.ai.mit.edu (Michael I. Bushnell, p/BSG)
> We were going to release such a system in cooperating with the Debian
> project, and we spent some money paying Ian Murdock to help do that.
> But he left Debian, and the remaining people didn't want to work
> together with the FSF.

It's not that we did not want to work _together_ with FSF.

The design of the Debian system is a result of discussion and consensus
among the developers, a network collaboration of about 80 people all
around the world who, somewhat surprisingly, work very well together and
get a lot done. The technical direction that was coming from FSF was
concerned with shaping Debian into the eventual GNU system. After 12
years of thinking about it, Richard had very specific ideas of what the
GNU system was supposed to look like, and we had little choice but to go
with his design, overriding the consensus of our own developers, if
we were to have FSF sponsorship.

We were out to build a good Linux system. We had not volunteered to
implement the GNU system according to someone else's plans. For this
reason, we offered the FSF the opportunity to take an active role as a
_peer_ with all of the other developers, rather than a
manager-from-above. Richard was not willing to take part in building
anything but the GNU system that he designed, and to do that he needed
to be able to override our design decisions. I understand his position,
and he's not wrong - he simply was trying to use other people to build
his dream when those other people wanted to build something else. That
was the impasse. We decided to come to an amicable split, and did so.
We still suggest in our FAQ that people make donations to FSF. We still
offer the system for FSF to sell as a fund-raiser. We still encourage
people to GPL their software. We share FSF's goals, but we are building
Debian.

> We wanted to acquire experience with doing such distributions first
> hand this way.  The value of the project would have been that we would
> learn how to do such projects well. So doing it now, by ourselves,
> wouldn't help anything--we would not be learning from direct
> collaboration with some of the people who have a good understanding of
> the problem.

Collaboration would be very nice. We would be happy to have FSF staff
monitor the Debian development. Anyone can get full developer status by
volunteering to maintain a Debian package, and we'd be happy to have
some FSF staff as package maintainers.

In this final paragraph I'm not speaking for Debian: this is my own
opinion. I'm afraid that some of the problem is that Richard, while he
is technically brilliant, isn't a "people person". One symptom of this
is that he can say things that turn important friends into enemies. A
"people person" would consider that keeping a friend is often more
important than being right. There is a way that FSF could address this
problem. I suggest that Richard take the _technical_ leadership
position with FSF, and let someone who could handle people better be
FSF's president. I think that would go a long way towards salvaging
FSF's public perception, and would put them on the track for future
success.

	Respectfully Submitted

	Bruce Perens
	Debian Project Leader
-- 
--
Pixar's "Toy Story": Over 1/3 Billion dollars world box office so far.

Bruce Perens AB6YM          B...@Pixar.com            http://www.hams.com/