*BSD News Article 3632


Return to BSD News archive

Path: sserve!manuel!munnari.oz.au!mips!mips!sdd.hp.com!usc!news
From: merlin@neuro.usc.edu (merlin)
Newsgroups: comp.unix.bsd
Subject: Restrictions on 'free' UNIX / 386BSD (Re: selling 386BSD)
Date: 14 Aug 1992 05:13:00 -0700
Organization: University of Southern California, Los Angeles, CA
Lines: 122
Sender: merlin@neuro.usc.edu (merlin)
Message-ID: <l8n8qcINN2c5@neuro.usc.edu>
References: x
NNTP-Posting-Host: neuro.usc.edu

>>Strangely, Chris Dimetrios (excuse wrong spelling) recently mentioned
>>in a patch he posted that he had put his *own* Copyright note on the patches.
>>And he further writes that if anyone want's to put hist software on CD-ROM 
>>that checking back with him would be required. 

>if someone sells 386bsd and gives ALL the money to the jolitz's, there is 
>nothing wrong with it - i would like some sort of centralized version
>revision management for 386bsd >= 0.2, and in order to do it, 
>they need money !

I personally believe Bill and Lynne Jolitz have done a great service for
the academic community by making a version of 4.3BSD-NET2 'Berkeley UNIX'
available for free which may be redistributed without restriction and may
be used for any purpose -- just like the UC Regents original copyright!!!

Many kudos to Bill and Lynne!

As I understand their request, Bill and Lynne simply want acknowledgement
as the original authors of some modifications to the 4.3BSD-NET2 -- just
like the UC Regents original copyright.  I believe this is just fine.  I
also believe that the obvious quality and steadily increasing popularity
of 386BSD will reap many rewards for Bill and Lynne in the form of their
own textbook sales, consulting revenues, and contract & grant awards.

However, I have a real serious problem with someone like Chris Demitrious
(whatever the spelling) 'contaminating' the UC Regents and Jolitz's free
4.3BSD-NET2 code with further restrictions of his own.  It's bad enough
to have AT&T/USL mucking up the waters with intellectual property claims
without having someone who happened to fix some bugs and/or who wrote a
somewhat better driver code 'contamining' the freely restributable Jolitz
system with their own ad hoc restrictions on what can be done with their
part of the product.

Boo!  Hiss!  cgd !!!

Frankly, I think it is extremely important to refuse to integrate any
such restricted / 'contaminated' code in the purely free redistributable
386BSD system.  Otherwise, five, ten, twenty years from now we are going
have to track down and get permission from dozens of bozos (who probably 
won't be possible to find) to use either their code or code which might 
be n-th generation 'derived' from the work of these bozos to avoid some
damn lawsuit along the lines of the current AT&T/USL action.

The reason the Jolitzs' could release 386BSD is that the UC Regents had
no such proprietary restrictions on the 4.3BSD-NET2 code.  The Jolitzs'
are free to keep, give away, adapt, extend, sell, buy, or whatever they
want to do with 4.3BSD-NET2 derived sources.  The freely redistributable
nature of 4.3BSD-NET2 without any restriction is precisely why we have a
working 386BSD system today.  Not even the Jolitzs' could have written
386BSD from scratch -- they started with a freely redistributable product.
Hopefully, future academic operating system developers will be able to
use the freely redistributable Jolitz code as a basis for their new works.

The whole point of a freely redistributable system is not to have to pay
financial hommage (tithe) to any particular creator (god) be they Regents,
AT&T, USL, Jolitz, or CGD.  If anyone seeks to impose any restrictions on
the use or redistribution for free or for sale of any 4.3BSD+ derived or
related drivers/etc then we might as well be dealing with AT&T/USL.  The
whole point is that the system can be used for anything without any kind
of restriction beyond reasonable acknowledgement of authorship.

If people like cgd really feel obligated to contribute 'contaminated' code
then their contributions should be made available separately from the free
redistributable core -- preferably in a separate 'restricted' distribution
-- and clearly labeled 'infectious waste' -- with ample warnings to future
users of such code about potential intellectual property rights claims and
intellectual health risks.  In my opinion the same label should be applied
to anything carrying the even more restrictive GNU copyleft.  

I don't mean to sound too adamant about this -- but, if people like me are
to consider committing to converting free image processing/reconstruction 
software to your 386BSD -- to make it usable as a base for real scientific
computation -- and to incorporate it as an element of future contract and
grant proposals to government, foundations & industrial research sponsors
then it is very important to preserve the very limited restriction of the
request for acknowledgement of authorship with no other restrictions.  On
the other hand, if you want to play the tithing, petty proprietary rights,
ad hoc restriction -- intellectual property 'contamination' -- game then
we might as well forget about using this system for real work -- and defer
to people who would rather view it as a mere coursework curiosity and/or 
playpen for the home hobbiest.

I really like 386BSD -- and I've put a lot of time into understanding it.
I've read most of the kernel, filesystem, network, startup, and memory
management source codes.  I reread the Leffler 'devil book' several times.
I'm eagerly awaiting the arrival of the announced Jolitz internals book --
and I hope they write a specific 386BSD system management, configuration,
problem solving book.  I've diagnosed the most serious limitation on the
use of 386BSD for scientific computation -- the emulation of the higher
floating point operations by the old libm.a -- and managed to get someone
to post an upgraded libfpu.a which performs as well as the SCO UNIX libm.a.
I've ported the core modules of several commonly available image processing
systems (including some of my own codes) to 386BSD -- to prove feasibility
of completing these ports in a reasonable amount of time -- and I'm in the
process of negotiating for financial support on my campus to do a full port 
of each of these packages.  

However, if it is the concensus of this group that 386BSD should fall into
trap of any further restrictions (like cgd deciding whether his claims of
intellectual property rights in some driver code) which could arbitrarily
interfer with the free redistribution of this code for any purpose -- then
I would drop these projects like a hot potato -- and go back to porting my
stuff to SCO UNIX SYSV/386 3.2r2.0 ODT 1.1 (maybe even invest in ODT 2.0).

Either 386BSD is freely available and redistributable for any purpose --
as in the UC Regents original copyright -- or it will end up a proprietary
bounded system contaminated with intellectual property claims of a large
ad hoc group of difficult to track down contributors -- and as a result it
would probably become a mere intellectual curiosity & hobbiest playpen.
If that's the outcome you want then maybe AT&T/USL should win their case.
At least if AT&T/USL wins there is only one god demanding human sacrifices
in exchange for access to the course codes -- not dozens of demigods/cgd's.

------------------------------------------------------------------------------
Alexander-James Annala
Principal Investigator
Neuroscience Image Analysis Network
HEDCO Neuroscience Building, Fifth Floor
University of Southern California
University Park
Los Angeles, CA 90089-2520
------------------------------------------------------------------------------