From: "D. Hazelton" <dhazelton@enter.net>
To: Alan <alan@lxorguk.ukuu.org.uk>
Subject: Re: GPL vs non-GPL device drivers
Date: Thu, 22 Feb 2007 16:36:37 -0500
User-Agent: KMail/1.9.5
Cc: "Michael K. Edwards" <medwards.linux@gmail.com>,
       "David Lang" <david.lang@digitalinsight.com>, davids@webmaster.com,
       "v j" <vj.linux@gmail.com>, trent.waddington@gmail.com,
       "Linux-Kernel@Vger. Kernel. Org" <linux-kernel@vger.kernel.org>,
       "Neil Brown" <neilb@suse.de>
References: <3d57814d0702191458l1021caeyaefd7775398c5f2a@mail.gmail.com> <200702212030.58317.dhazelton@enter.net> <20070222141047.425337a0@lxorguk.ukuu.org.uk>
In-Reply-To: <20070222141047.425337a0@lxorguk.ukuu.org.uk>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200702221636.37306.dhazelton@enter.net>
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 6245
Lines: 131

On Thursday 22 February 2007 09:10, Alan wrote:
> > As a side note: The distinct wording of the GPL actually *invalidates*
> > the GNU/FSF claim that dynamically linking a work with, say, the readline
> > library,  means the work is a derivative of said library. The GPL states
> > (in
>
> Not that I can see no, but you could take this to a list with lawyers not
> programmers on and improve life for both parties

See Section/clause 0 of the GPL. 

> > clause 0) that the license only covers copying, modification and
> > distribution. Unless they are confusing "Linking" with "copying" or
> > "creating a derivative work" the claim is invalid - because, as it has
> > been shown, a mechanical process such as compilation or linking *cannot*
> > create a derivative work.
>
> If you take the microsoft windows source code and compile it yourself
> believe me you will get sued if you ship the resulting binaries and you
> will lose in court.

But that's because it is *WINDOWS*, which, unless specifically granted to you, 
does not include a transfer of the right to distribute in *ANY* form. Every 
PC manufacturer that wants to distribute Windows on new machines they produce 
*MUST* sign an agreement with M$. As I have never seen any of those 
agreements I cannot state what the terms are and whether they are different 
for each company holding such a license.

And unless you've signed a licensing agreement over the source code to 
Windows, you're more than likely to have another lawsuit on your hands for 
possessing it. 

<snip>
> I would also note that the FSF makes no claim about dynamic v static
> linking, merely about derivative works - which is the boundary the law is
> interested in. Indeed the GPLv2 was written in the days where dynamic
> linking was quite novel which is one reason the LGPL talks about

Granted.

> "For example, if you distribute copies of the library, whether gratis
>  or for a fee, you must give the recipients all the rights that we gave
>  you.  You must make sure that they, too, receive or can get the source
>  code.  If you link a program with the library, you must provide
>  complete object files to the recipients so that they can relink them
>  with the library, after making changes to the library and recompiling
>  it.  And you must show them these terms so they know their rights."

Eh? Complete *object* files so that after making changes and recompiling they 
can relink it? Umm... I don't know about you, but that makes me laugh. What 
is the purpose of providing "Complete Object Files" to everyone if they are 
just going to recompile and relink the library?

Yes, I know this quite likely refers to any object files (or other binaries) 
that are part of the library but not part of the source. (and *are* required 
for the library to be completely functional)

> and says nothing about dynamic/static linking.
>
> > Related to that... Though a parser generated by Bison and a tokenizer
> > generated by Flex both contain large chunks of GPL'd code, their
> > inclusion in the source file that is to be compiled is mechanical - the
> > true unique work is in writing the file that is processed by the tool to
> > produce the output.
>
> Flex is more complex because the resulting binary contains both compiled
> work of yours and a support library of FSF owned code (-lfl). 

Copyright *doesn't* extend to compiled code. It cannot, because compiled code 
is a machine generated translation. A machine generated translation isn't the 
product of a creative process. And you can also provide all the routines 
normally provided by the support library. This means that the support library 
is *NOT* a necessary part of the system.

> The non 
> computing analogy here is the difference between using a paint program to
> create a work, and using a paint program to create a work but also
> including other artwork (eg clipart). 

Yes, but in both cases the result is *CLEARLY* the result of a creative 
process, and said clipart, unless it is in the form of an entirely machine 
generated image, is a separate work (and one resulting from a creative 
process) that you are using under license. (Unless said clipart was released 
into the public domain)

> The same is true of the GCC compiler 
> as it merges your work with supporting library helper code modules which
> are themselves creative works. 

Again you are confusing a mechanical translation process with a creative 
process. But it doesn't matter, in this case. The binary form of a program 
*technically* falls under the copyright on the source code - a mechanical 
process that translates a copyrighted work into another form *cannot* remove 
the original copyright.

But said modules have clearly defined and limited purposes.

> Clearly you could replace those helper 
> modules with your own work and the result would be different.

Yes.  And you've noted that yes, they can be replaced. Which means that they 
are also not a necessary part of the system. 

Claiming that any library (that can be replaced), either dynamically or 
statically linked to a program, is a requirement isn't smart. That also means 
that claiming a program that uses those libraries is a derivative work is 
bull.

Further, if such a library or module has an interface that must be accessed in 
one specific way, then said interface cannot be copyright. (As has been 
pointed out to me several times in other threads on LKML)

> A better example for your case might be indent where the program
> processes your work mechanically and produces an output that doesn't
> contain any other creative works, or most of intltools which merges
> translations mechanically. (the merge code is sometimes a little creative
> but thats in the sense of being a nuisance not in the legal sense of
> creative work)

True. But in those cases there isn't the gray area that exists surrounding 
claims of a Flex generated scanner or a Bison generated parser being 
derivative works of either program. 

> Alan

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