Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 29 Jul 1999 07:25:50 -0500 (EST)
From:      "John S. Dyson" <toor@jdyson.com>
To:        rfotescu@idsrom.com (Radu-Cristian FOTESCU)
Cc:        freebsd-advocacy@freebsd.org
Subject:   Re: What to tell to Linux-centric people?!
Message-ID:  <199907291225.HAA08586@jdyson.com>
In-Reply-To: <C1F3617BDC52D21185E40000214C247C1540AA@id-bucharest.idsrom.com> from Radu-Cristian FOTESCU at "Jul 29, 99 01:00:28 pm"

next in thread | previous in thread | raw e-mail | index | archive | help
> > the other predominant FREE U**X clone doesn't effectively
> > have SMP at all yet -- so for that, it is a non-issue.
> 
> No! Linux DOES HAVE SMP! And in kernel 2.2 they say it's better than in 2.0.
> I can't prove it, I don't have multi-processor machines here...
>
But, Linux isn't free, but is under GPL.  Note that I distinguish between
code that is licensed with significant strings attached vs. code that
is free in a non-discriminatory way for almost all legal purposes.  IMO,
there are two reasons why GPLed code is often called free:  1) because
source availability with usage restrictions or control is sometimes
confused with the code being 'free.'  2) Those who advocate GPL licensing
often feel justified in taking advantage of a limited definition of the word
'free' in order to forward their cause.  As a case in point, I used to
normally get the source for the RSX-11M (and RSX-11M+) kernel so that I
could fully resysgen the system, and have all info required to produce
a driver, ACP or somesuch.  Just because I had access to that source,
didn't mean that I have the freedom to redistribute built RSX-11M systems
without paying money or other item in trade.  In the case of DEC, I could
sell my binaries that was based upon RSX-11M code, if I sold an RSX11M license
to my customer (in essense, I had to make a trade for the right to
redistribute binaries of my own work.)  In the case of GPL, I can give
binaries of my work that is effectively a result of merging (linking)
my work into a GPLed work, only if the source code for the entire work
is redistributed.  In both cases, my freedom to make profit from my
value added is constrained by 1) rightfully giving DEC money or 2) rightfully
following the GPL and bona-fide offering to redistribute my hard won source
code.

So, the cost of dealing with DEC is to pay license fees, but I get
to keep my source code, optionally giving it to my customer for some
money over and above the DEC licensing fees (i.e. my source code is
still an object for negotiation.)  In the case of GPLed works, the
significant redistribution cost is to provide a bona-fide offer to
give the source code of my work away, and create another potential
competitor who can easily use my work to compete against me.  It
doesn't make any sense to call either the DEC or GPLed codebase(s) 'free.'

In the case of a BSD licensed work, I don't have to agree to anything,
other than preserve copyrights and credits (which are indeed common
sense minimal etiquette or legal issues) in order to have the right
to redistribute works that are a result of the two works (my work, and
the 'other' codebase).  In essense, it is free for me to redistribute
those merged works, and my own additions to the work aren't encumbered.
With the BSD license, I still have the freedom to negotiate a price for
source code that I invented.

Two predominant free OSes would include FreeBSD and NetBSD.  The
predominant source-available OS, but has "strings attached" is Linux.

> 
> But, stated that this way, Linux weakens Bill Gate's imperium, I don't think
> that these "religious wars" between FreeBSD and Linux advocacies should
> continue.
> 
It is difficult to deal with 'religious wars', however there are
issues of being deceptive (where all camps have their own problems.)
IMO, the issue with Microsoft isn't that they have aggressive business
practices, but it is that they are willing to inform their userbases (and
the mgmt of the userbase and support teams) with very biased and often
deceptive information (IMO.)  Such deceptiveness has even extended to
some interestingly worded (in clinton-esque terms) testimony and bogus
demos.  Likewise, it is often (but not always) deceptive to call software
under GPL restrictions "free."  (Almost any usage of the word 'free'
when used in association with GPL, requires singnificant amount of
qualification to be accurate.)  It is quite clear that most usage of
the term "free" when describing GPL by those who *understand* the GPL
license is for the easy but errant assocation of GPL with 'free', and
the marketing advantage associated with the term 'free.'  (It is
wrong to claim that everyone who uses the term 'free' without
qualification when describing the GPL is being dishonest, because mistakes
are easy to make and the spin associated with certain historically incorrect
association between the terms 'GPL' and 'free' are hard to correct.)

In the case of software licenses, the restrictions to freedom are often
dependent on how the limitations affect the user and developer bases.  By
far, the largest numbers of users and developers who use/utilize the GPLed
compiler suites are not negatively affected by licensing restrictions
associated with simple use of the GPLed compiler.  In the case of OS code,
which is often used and incorporated into a product, the longer term
consequences of basing one's work on GPLed code needs to be seriously
considered.  In the case of a simple end user, the nature of the GPL seldom
has significant negative consequences.  However, it is very likely that most
of the people who are reading the FreeBSD mailing lists (for example) are NOT
simple end users.

One major beauty of source available OSes is that there is alot of freedom
to adapt or modify and include the OS kernel (or other) code into a product.
In the case of an OS kernel, there is alot of opportunity for such
modifications (given different hardware environments, different performance
tradeoffs, etc.)   The GPL causes more than necessary encumberance of new,
potentially novel and potentially expensively innovative code added to
previously GPL licensed works.  It is a fallacy to make the general claim
that work added to a preexisting piece of code is necessarily inferior or
less important to a project.  For example, given a GPLed compiler, those
compiler writers who consider adding an innovative algorithm (or subsystem)
need to weigh the cost of having to choose to give their work away to
potential competitors, EVEN BEFORE THE WORK IS DONE and EVALUATED.  There
are analoguous situations for microprocessor or embedded developers
using kernel code.

IMO, calling BSD licensed software "free" is also probably untrue in an absolute
sense, but in the scheme of things is licensed under very, very free terms.  So,
given the various predominant OSes, where source is normally available, the
BSDL terms provide the most freedom for those who add significant value to the
works themselves.  Until a predominant OS becomes freer than those OSes
that are distributed under the BSD licensing terms, it is probably okay to
call the *BSD OSes 'free.'

John



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-advocacy" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199907291225.HAA08586>