D&C GLug - Home Page

[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]

Re: [LUG] Don't compare physical objects and software.

 

Paul Weaver wrote:
> On Wed, Jul 19, 2006 at 02:04:09PM +0100, Neil Williams wrote:
>> The author gives away "ownership" when using a copyleft licence.
> 
> Not at all, they grant people permission to copy their work according
> to the terms of the GPL. They still own copyright on the work.

Copyright != ownership.

Copyright is undoubtedly in force but it is a mistake to use the word
"own" in that context. Copyright is granted automatically but the
software does not have an owner.

http://www.gnu.org/philosophy/why-free.html

> Once the work has been copied under terms of the GPL, then those
> terms stick.

Of course.

> 
> All the GPL is is an OPTIONAL license that allows you to copy the 
> licensed code without breaking copyright law. You can negotiate a 
> different license with they copyright owner (Trolltech/QT allow this,
>  for example).

For a new licence to be granted, ALL copyright holders must agree. If a
single copyright holder refuses permission to distribute their
contribution under the new licence, that code must be replaced or their
copyright is infringed.

The GPL does NOT allow the copyrighted material to be transferred to a
different licence without permission - to do so would allow a
coach-and-horses through the GPL because anyone could do what the FEMM
author tried to do.

> You could argue that the act of running a computer program creates a 
> copy in memory, and therefore is forbidden without a specific license
> to copy from the copyright holder, but that's a whole other issue.

It's only an extension of the same issue - and yes, certain licences
could be construed that way.

> Originally written by Ron Minnich, his original version is
> copyrighted ("owned" as you will) 

No. The copyright holder does NOT *own* the software. Once released
under the GPL, that software can be "forked" by anyone and they have as
much freedom as anyone else to completely rewrite the entire source code
in ways that the original author would simply not want done. The author
has NO right to prevent this - he does not own the software and cannot
therefore impose ANY restrictions on the fork team. This is specifically
required under the GPL.

> Modifications by Eric Van Hensbergen, those modifications are owned
> by him. 

No, they are not owned at all. His copyright will remain LONG after all
his code has been completely rewritten. After a few code churns, there
may be nothing left of the original code - nothing for him to own - yet
his copyright still remains.

> Ron can't take the modifications and reuse them unless he
> complies with the terms of the GPL.

True - these modifications were submitted under the GPL and therefore
the GPL applies to their use.

> 
>> My GPL code is not mine to do with as I please anymore. Someone 
>> else can easily take a previous / current version and take it in
>> their particular direction.
> 
> They can, but your original code is still yours,

No, it is not.

> you can do whatever
> you want to it, ignoring the GPL.

Only in the exceptional circumstance that 100% of the code within the
package is under my sole copyright. This INCLUDES libraries and macros
from the build system!! Unless you write the entire thing using no GPL
dependencies, then the entire package must be released as GPL AND you
are bound by the GPL when you distribute the code that includes the
other GPL code when it is linked. To change to a new licence, you would
have to ensure that the elements that you pull in from the wider system
are either replaced or already available under a compatible licence.

> You can't revoke the permission you
> have given to someone else (Ron can't revoke Eric's permission to
> modify the file above), but you still have copyright on the original.

True. The GPL is irrevocable. Once code is distributed under the GPL,
no-one can prevent further distribution and modification of that code
provided that such actions are still within the GPL itself.

>> Ownership - in regard to steering, restriction of rights and
>> removal from availability - are NOT within my rights as author once
>> I've released the code as GPL. If the code contains any
>> contributions from other authors, I also cannot relicence that code
>> without their consent.
> 
> That's right, but if you don't have any modifications (the 2003
> version of 9p.c, before Eric's modifications) you can do whatever you
> want, aside from revoke permissions you've already granted.

These corner cases are few and far between. Most packages include code
from multiple authors, most include macros from other GPL packages, most
depend on GPL libraries.

> You can add your own code, and choose not to release the source under
>  the GPL, for example.
> 
> You don't have any control over what other people do with the code 
> (outside of trademark law), but your code is still yours to do with
> as you please.

Only in as far as the code is 100% your own - that is rare.

>> Hence, I attribute some files to the Free Software Foundation so
>> that there can be no doubt that relicensing is not an option.
> 
> Admirable, but not neccersary under the terms of the GPL, 

True.

> you'll find
>  most of the Linux Kernel is copyright of various people (about
> 470ish). If all those people get together and decide that the next
> version of the kernel will not be released under GPL2, then that's
> their choice. 

It is and there is talk that at least some of the kernel may not be
available under GPL3.

> You and I will still have rights under the GPL2 license
> that 2.6.14 (or whatever) is running, but their new version would be
> out of our reach.

It is highly unlikely that the kernel could be licensed in such a way as
to be incompatible with a GPL compliant system, GPL2 or GPL3. GPL2 code
will remain compatible with GPL3, it's just that it could not be
released under GPL3 or later.

> 
>> c.f. Perl - the original author cannot now take Perl away from the 
>> community, it cannot be made proprietary.
> 
> No, but he can make modifications to the code that he originally
> wrote, and release it as proprietry. He wouldn't be infringing
> anyone's rights under copyright.

But what's the point of that? The original core code has been rewritten
dozens of times. It bears no resemblance to the current core. It would
undoubtedly be buggy.

>> It was sufficient to merely demonstrate that the licence could not
>> be changed unilaterally. FEMM has since re-appeared as an open
>> source project. http://femm.neil.williamsleesmill.me.uk/
> 
> In that case that's right. The kernel can't change to GPL3 unless all
>  contributers OK it

Not quite true. Not all kernel contributors have removed the section
about "at your option any later version". Those files that have an
unmodified GPL licence notice can be redistributed under any subsequent
version of the GPL.

>, OR any non-OKing contributer's code is removed
> from the kernel (and independently reimplemented as allowed by
> copyright law)

Yes.

> 
> In the case where somebody reasigns copyright to an organisation like
>  the FSF, the FSF would have to OK it.

Yes.

>> The fact that Robin's code was a small % of the codebase does not
>> matter - he contributed some of the code that constituted the
>> software that was relicenced without his consent. His copyright was
>> infringed.
> 
> Inded, he "owns" that code. 

He holds the copyright on that code. He does not own it. Very subtle,
but extremely important difference.

> If that code were removed, then he
> wouldn't have a case. It wasn't, so he does have a case.

Yes. The problem was that the case is hard to prove because his
contribution is not easily identified.

>> The original author is not above the licence once ANY contribution
>> has been accepted from a third party because that contribution is
>> made under the terms of the *old* licence and the contributed code
>> cannot be put under a different licence without agreement from the
>> copyright holder of that portion of the code. The only option is to
>> rewrite the infringing code - that was not done in the case of
>> FEMM.
> 
> Yes, but the original release didn't have robins code, and was whoely
>  owned by the original author, who can relicense.

Not quite. The author accepted input from others, as well as Robin. We
don't know if those others were approached and consented to the new
licence. All we do know is that Robin did not consent - nor was he even
consulted.

> 
>>> You have copyright over it.
>> True - that is never in doubt. The GPL is governed by copyright law
>> but it makes an explicit emphasis on copyleft too. Anyone who
>> contributes to the software (including documentation) also retains
>> copyright over their contribution.
> 
> Yes, they own their contribution, which is what I said in the first 
> place!

No, you keep equating ownership with holding the copyright.

Example:

#include <stdio.h>

int
main (int argv, char** argv)
{
        printf ("Hello World!\n");
        return 0;
}

The actual snippet is public domain by sheer repetition but imagine it
was a slightly more complex snippet. I am automatically granted
copyright over that code. Acknowledging that copyright does nothing to
change that, except that it makes it clear to others about the copyright
holder for that section.

I now release that code under the GPL. It is distributed and downloaded
by others under that licence. I accept a patch that allows translation
of the string so that French and Russian users can see "Hello World" in
their own languages.

I still hold the copyright in the *modified* version (along with the
copyright holder(s) of the patch) because it is a modified version of my
copyrighted work, but just who is supposed to "own" the modified work?

Ownership of software is a flawed concept and has no place in free
software. Please stop any discussion of ownership, it simply confuses
the issues around copyright.

Copyright != ownership and copyright is all that matters. Ownership is
irrelevant.

It is copyright that dictates whether I can redistribute the original
unpatched code - it also determines whether someone else can
redistribute the unpatched or patched code instead of me, including
against my express wishes. Ownership of the code is irrelevant because
ownership confers no additional benefits on the copyright holder nor
places any additional restrictions on the user than what is already
granted under copyright law and the licence which itself is underwritten
by copyright law, not ownership.

Ownership takes us right back to the original point. You only "own" a
physical object. Something that has an owner is an item of property.
Software is NOT a physical object, it is an abstract concept.

Who "owns" the idea behind a patent? The patent holder only holds the
patent and that can be traded so that the patent holder is not the same
person as the original person with the idea.

Who "owns" E=mc2?

Who "owns" Boyle's Law?

Who "owns" Newton's Laws?

You can own a book that expresses those ideas - you cannot own the ideas
themselves.

You can own the media that stores a copy of the software, you cannot own
the software itself.

>>> You allow others to copy it given the terms in the GPL, you don't
>>>  have to obey those terms as you don't have to accept the GPL,
>> That is ONLY true if you have written 100% of the code yourself in
>> a clean room environment. If you've taken any GPL code into your 
>> programme, your programme also needs to be GPL - you ARE bound by
>> that because YOU didn't write that bit of the code. That is why
>> readline is a GPL library.
> 
> Indeed, although I'm surprised that readline is GPL and not LGPL

It is a deliberate choice made by RMS and the FSF. QOF is also a library
 that is released only under the GPL.

http://www.gnu.org/licenses/why-not-lgpl.html

http://qof-gen.sourceforge.net/#free

Do NOT think of the LGPL as "the GPL for libraries". The LGPL is the
LESSER GPL - it is weaker, it is a compromise and the decision to NOT
put a library under the GPL should be a strategic, practical, decision -
not a blind assumption that it is a library, therefore LGPL.

It is free software, therefore it should be GPL. Only in exceptional
circumstances should any free software be available under the LGPL.

(Naturally, if the package is Perl or Apache module, etc., then a
suitable free software GPL-compatible licence should be used but STILL
the LGPL should be a licence of last resort.)

> Of course you are. Every time I use an ATM I use proprietary
> software, and I haven't agreed to any EULA (I'm agreed to a certain
> implicit contract, same as when I buy a marsbar from a shop, but
> that's not the same thing)

That EULA allows you to do that - another EULA may not.

>> It is within the remit of a proprietary licence, built on the same
>> copyright laws, to prevent anyone from compiling, executing,
>> viewing, copying, distributing and even storing the programme
>> without consent. That is what "All Rights Reserved" means.
> 
> Under what law? Copyright law only covers making a copy

Any operation on software involves copying. Software on a storage medium
is completely inert. You cannot modify it, execute it or view it without
making a copy in memory. The EULA does not necessarily have to allow ANY
copying. It may or may not allow copying into memory for the purposes of
execution, it could only allow copying into memory for the purpose of
display in a read-only mode.

> (permament
> copy I believe, otherwise CD's would infringe when you play them as
> the data is stored in memory temporarilly, as would reading a
> webpage).

CD's are specifically licensed to allow playback but only to a certain
number of people - public broadcast is disallowed in the vast majority
of cases. Read the notice on a music CD carefully - it's often printed
on the actual CD around the edge.

> Running software isn't covered by that.

It can be. Executing software involves making a copy in memory. That can
be specifically prohibited under copyright law - that is one of the
features of the non-disclosure agreements for the Windows source code,
that it may not be made into an executable form. Those agreements are
based on copyright law because no other law applies.

This is not at all uncertain - it is an extremely important point within
the software patent discussions because it is central to whether certain
types of embedded systems can be patented and the point at which a
silicon chip changes from hardcoded ROM to modifiable ROM. Robin or
Aaron may be the best people to take that point on.

> Installing it to
> your computer may be an infringment of copyright law, I can't
> remember offhand. You are allowed to (in the UK of course) make
> back-up copies of software with no license, again under copyright
> law.

No. That is a false assumption. "Fair use" is a concept borrowed from
public libraries of text. It has NOT been accepted in a UK court that it
applies to ANY digital format. "Fair use" actually only applies to text
for the specific and limited purposes of "personal research".

> The vast majority of large software products, however 9p.c in the
> linux kernel could be relicensed with permission of only 2 people.
> The 1998 version of lilo is copyrighted by Werner Almesberger alone.
> Many small GPL projects have no contributions other than that of the
> original author, I've created some myself.

You'd be surprised - look closely at the code and the dependencies.

 > The copyright is held by all contributers, as such the program is
> owned by all contributers.

The programme is not owned by anyone. See above.

I hope I've shown that *holding* copyright is NOT the same as *owning*
anything.

> Your original point was you lose ownership
> of the code when you license under GPL (and somebody accepts that
> license), I guess it depends on your definition of "own". Your code
> is still your own, you don't lose any rights to it by licensing under
> the GPL, you simply grant rights to others.

My point is that this is simply a corner case - the vast majority of
code simply does not consist of code entirely and completely contributed
by one single copyright holder.

>> M$ would disagree. Personally, I cannot afford to test that one in
>> court.
> 
> That's a whole other story, but I haven't agreed to the Windows 2000 
> license, I don't see how I can be held to it's content.

You may not see it but if M$ choose to ensure that you do see how this
is possible, it will be a US courtroom that makes it clear to you.

> Copyright covers copying, not use (unless that use involves copying
> -- running a computer program arguably falls under this, however
> reading a book clearly doesn't, and there is no way that a copyright
> holder of a book would be able to hold anyone to a license
> agreement).

True.

>> The EULA dialogue itself is only one part of the package - the
>> software usually comes with a lengthy paper licence and/or a
>> licence file on the original media.
> 
> I wouldn't know, I don't buy proprietry software. That license grants
>  you rights to copy the work. If you don't accept the license, then 
> copyright law applies and you can't make a copy except for certain 
> execptions like Current Affairs, reviews etc.

NO! If only copyright law applies, the law is:

"All Rights Reserved except by express and written consent of the
copyright holder under any and all conditions required by the copyright
holder for the duration of the copyright period."

>> Ignorance of the law is no defence and in a business setting, not
>> reading an EULA could give rise to a claim for negligence against
>> you by your employer.
> 
> I don't agree to windows EULA's at work,

You agree that you will do nothing to bring your employer into legal
confrontation with a third party - that is part of your employment contract.

Your company HAS accepted the EULA. You work for them. You accept the
EULA because you accept working for an employer who requires that you
operate under the terms and conditions of that employment. One of those
conditions is that you agree to abide by the agreements and contracts
made by the employer in the normal and reasonable practice of their work.

> I don't feel I'm qualified
> to sign a legal contract on behalf of my employer. Other people do,
> but that's their choice.

You don't have to. Your employer has agreed to the terms and that is all
that matters. You agree to accept the conditions of employment and your
company agrees to uphold the EULA. Your employment requires you to abide
by agreements made by the company in so far as those agreements affect
your sphere of work. If your work involves using software controlled by
the EULA to achieve tasks that are required of you by your employment
contract, then you have an obligation to uphold the agreements that the
company has made in good faith concerning the software that was obtained
by the company for that task. If you disagree, your only real option is
to provide, at your own cost, a replacement piece of software that your
employer agrees is comparable and will produce acceptable results.

If a company discovers that an employee has copied 10,000 Windows 2000
CD's and sold them at car boot sales, they have the right to begin
disciplinary proceedings against that employee in order to reduce their
liability should M$ come after them as licence holders of the copied
software.

>> All messages signed with 0x28bcb3e3 or 0xa897fd02 are Copyright
>> Neil Williams <linux@xxxxxxxxxxxxxx>, date as gnupg signature
>> timestamp.
> 
> I take that as implicit consent to reproduce your email on my
> computer (storage), and in this reply. If not, too late, sorry I'm
> infringing, but I doubt any court would find in your favour.

Yes, implicit consent is reasonable in the circumstances that this is a
publicly archived mailing list. You do not need my consent to reply -
the comment was added just to demonstrate a point.

Note that the statement ALSO applies to any private emails sent between
us, if any. There, no such consent can be assumed although, again, it
would be reasonable to assume that the right to copy the original in a
reply is acceptable, especially as the reply would normally include a
statement along the lines that the original (as identified by some kind
of quoting symbol) was attributable to the copyright holder.

See how silly this whole copyright thing becomes in the absence of a
free and open licence!

-- 

Neil Williams
=============
http://www.data-freedom.org/
http://www.nosoftwarepatents.com/
http://www.linux.codehelp.co.uk/


Attachment: signature.asc
Description: OpenPGP digital signature

-- 
The Mailing List for the Devon & Cornwall LUG
http://mailman.dclug.org.uk/listinfo/list
FAQ: http://www.dcglug.org.uk/linux_adm/list-faq.html