License Committee Report for July 2007

[Alexander Terekhov-3]

On 8/2/07, Wilson, Andrew <andrew.wilson@...> wrote:
[...]
> My hope would be that, as license stewards for all GPL versions,
> FSF might yet be able to issue clarifications or
> guidelines to address this important
> license combination.

Six or seven years ago Shlomi Fish published wonderful April 1st story.

Here it is:

----
A recent press conference of the Free Software Foundation confirmed
the rumors that the GNU General Public License was found to be
incompatible with itself. This newly discovered fact may actually
cause a lot of disorder in the free software world in which most
programs and libraries are licensed under this license.

Richard Stallman, chairman of the FSF, called upon developers to
immediately exempt GPL-licensed software from the GPL, as far as
linking them with GPL programs is concerned. "We have already made
sure all GNU software and every other software that is licensed to
the Free Software Foundation would be ad-hoc compatible with itself.
However we need other developers to do the same for their software",
Stallman said.

Eben Moglen, the FSF's attorney outlined the subsequent steps that
his organization will take to overcome this crisis. The first step
would be releasing a Modified General Public License (or MGPL for
short) that will be compatible with the GPL and with itself as well
as with all other licenses that the GPL is already compatible with.
It will be labeled the GPL version 2.1, thus allowing developers to
convert their software to it. He noted that care would be taken to
make sure the upcoming GPL version 3.0 will be compatible with
itself, as well as the MGPL.

For the time being, though, there is an explosion of commentary,
confusion and otherwise bad temper about the newly formed situation.
Eric S. Raymond, the famous Open Source Guru notes: "This is one of
the greatest blows to the Open Source world, I have yet encountered.
I have already exempted all of my own software from the GPL in this
regard, but there is a lot of other software out there, and many of
its authors are not very communicative.

Bill Gates, Microsoft's co-founder, on the other hand, seems to
find the situation very amusing: "I said times and again, that
viral licenses such as the GPL are a bad idea, and many open-source
advocates disagreed. Now they see that even making sure one's
license is compatible with itself, is hard to do when you open that
can of worms."

The integrity of many software projects whose license is the GPL and
yet contain works licensed by several developers is in jeopardy. The
Linux kernel is a prominent example of such a case. In a post to its
mailing list, Linus Torvalds commented that, in their case, it was
not an issue. "My interpretation of the GPL is already quite unusual,
so I'll simply rule that I also interpret the GPL as compatible with
itself."
----

:-)

regards,
alexander.

[Walter van Holst-2]

> -----Oorspronkelijk bericht-----
> Van: David Woolley [mailto:forums@...]
> Verzonden: donderdag 2 augustus 2007 8:57
> Aan: license-discuss@...
> Onderwerp: Re: Combining GPL and non-GPL code

> That conflicts with a basic concept in the GPL, the "public".
>  The concept is that indirect recipients also receive a valid
> licence.  There may be belt and braces wording, but,
> basically the theory is that the GPL is a bare licence, not a
> contract, so doesn't require offer and acceptance at each step.

That is a lovely notion and I'd like to agree with you here. I've even defended this notion in a peer reviewed journal, but I have seen the error of my ways. As long as the GPL requires the end-user to surrender its rights to compensation for damages caused by the software and not to expect any warranty or guarantee for the software to actually do something useful, it cannot be a unilateral 'licence'. By its very nature it becomes a multilateral set of obligations. No matter how much you will be laughed out of court in practice when you start suing for damages caused by gratis software, the fact that it expects you to surrender your right for compensation makes it a contract and not a unilateral licence.

At least in the Netherlands and I daresay the rest of continental Europe this theory does not apply.

Regards,

 Walter

[Wilson, Andrew-2]

 
David Woolley wrote:

> Roger Fujii wrote:
>
>> You are missing the mechanics of how licenses work.  These things
>> are all source/receiver based - that is, the last person in the
>
> That conflicts with a basic concept in the GPL, the "public".  The
> concept is that indirect recipients also receive a valid licence.
There
> may be belt and braces wording, but, basically the theory is that the
> GPL is a bare licence, not a contract, so doesn't require offer and
> acceptance at each step.
>
> I don't think the FSF would intend that other bare licences for parts
of
> GPLed code should not be "public".

... and then this thread went off into the weeds on the ever-popular
argument over whether GPL is a license or a contract without addressing
Roger's basic point.  To wit, following the chain of licensing for a
derivative
work which contains non-GPL and GPL code inevitably leads to a scenario
where a user only has a license under GPL.  For illustrative purposes,
let's look at a chain

BSD -> GPL+BSD -> GPL -> end user

Roger's argument, with which I complete concur, is that the end user
who has received a copy from a distributor under GPL does not
have sufficient rights to strip off GPL and revert to the original
BSD.  Arguments which say it is ever valid to strip off GPL from
a derivative, under any circumstances,
will no doubt encounter stiff counter-arguments from GPL advocates.

From a pragmatic point of view, although GPL requires you to mark source
code changes, in the real world this doesn't always happen.  The user
who has received a copy of a derivative should see the original
BSD copyrights intact in the sources, but he/she is probably ill-advised
to assume
the original code is also intact.  If the code does contain any
modifications,
if those modifications are made to a GPL'd derivative they
are automatically GPL.  If this user really wants the original
BSD code under a BSD license then he/she would be well advised, IMO,
to find a pristine original copy which has not passed through a
distribution
chain under GPL.

Andy Wilson
Intel open source technology center

[Walter van Holst-2]

> -----Oorspronkelijk bericht-----
> Van: Wilson, Andrew [mailto:andrew.wilson@...]
> Verzonden: donderdag 2 augustus 2007 20:16
> Aan: David Woolley; license-discuss@...
> Onderwerp: RE: Combining GPL and non-GPL code


> Roger's argument, with which I complete concur, is that the
> end user who has received a copy from a distributor under GPL
> does not have sufficient rights to strip off GPL and revert
> to the original BSD.

In case the code was otherwise unaltered, I would lean towards the opinion that anyone can strip off GPL and revert to the original BSD.

>  Arguments which say it is ever valid to
> strip off GPL from a derivative, under any circumstances,
> will no doubt encounter stiff counter-arguments from GPL advocates.

Yup :-)

> > From a pragmatic point of view, although GPL requires you to
> mark source code changes, in the real world this doesn't
> always happen.  The user who has received a copy of a
> derivative should see the original BSD copyrights intact in
> the sources, but he/she is probably ill-advised to assume the
> original code is also intact.  If the code does contain any
> modifications, if those modifications are made to a GPL'd
> derivative they are automatically GPL.  If this user really
> wants the original BSD code under a BSD license then he/she
> would be well advised, IMO, to find a pristine original copy
> which has not passed through a distribution chain under GPL.

Which is precisely what I would advice any client of mine. In case the client doesn't mind to be in court, he could consider the option of holding the opinion that since no changes were marked he had a reasonable expectation of the code being unaltered and therefore having the right to strip the GPL from it. I wouldn't try that at home.

Regards,

 Walter

[John Cowan]

Walter van Holst scripsit:

> As long as the GPL requires the end-user to surrender
> its rights to compensation for damages caused by the software and not
> to expect any warranty or guarantee for the software to actually do
> something useful, it cannot be a unilateral 'licence'.

Well, I see this through a common-law prism, of course, but it seems
to me that the GPL is simply telling you that the licensor is not
giving you certain rights.  It also notifies you that this may be
overridden by operation of law: your legal system may give you
rights (warrantability and fitness, e.g.) that cannot be disclaimed
in this way.  It does not require you to surrender anything you had
in the first place, and as such remains unilateral.

> At least in the Netherlands and I daresay the rest of continental
> Europe this theory does not apply.

Fair enough.

--
Is a chair finely made tragic or comic? Is the          John Cowan
portrait of Mona Lisa good if I desire to see           cowan@...
it? Is the bust of Sir Philip Crampton lyrical,         http://ccil.org/~cowan
epical or dramatic?  If a man hacking in fury
at a block of wood make there an image of a cow,
is that image a work of art? If not, why not?               --Stephen Dedalus

[John Cowan]

Wilson, Andrew scripsit:

> From a pragmatic point of view, although GPL requires you to mark
> source code changes, in the real world this doesn't always happen.
> The user who has received a copy of a derivative should see the
> original BSD copyrights intact in the sources, but he/she is probably
> ill-advised to assume the original code is also intact.  If the code
> does contain any modifications, if those modifications are made to
> a GPL'd derivative they are automatically GPL.  If this user really
> wants the original BSD code under a BSD license then he/she would
> be well advised, IMO, to find a pristine original copy which has not
> passed through a distribution chain under GPL.

I can't argue with that, pragmatically speaking: it's true that it's
not always obvious when looking at a BSDed file that arrives as part
of a GPLed program whether or not it's been modified, and therefore a
derivative work as such.

Let's consider a hypo in a different domain.  I get a book with
the ordinary copyright license: All Rights Reserved, specifically
copying is forbidden.  However, one of the book's chapters is in fact a
verbatim article published by the ACM, with the standard ACM copyright
notice intact:

        Copyright© 20XX by the Association for Computing Machinery,
        Inc. Permission to make digital or hard copies of part of this
        work for personal or classroom use is granted without fee provided
        that copies are not made or distributed for profit or commercial
        advantage and that copies bear this notice and the full citation
        on the first page or intial screen of the document. Copyrights
        for components of this work owned by others than ACM must be
        honored. Abstracting with credit is permitted. To copy otherwise,
        to republish, to post on servers, or to redistribute to lists,
        requires prior specific permission and/or a fee. Request
        permissions from Publications Dept., ACM Inc., fax +1 (212)
        869-0481, or permissions@....

Do you argue that I must not (not should not pragmatically, but must not)
exercise the rights granted there with respect to that part of the book,
and make a copy of it on a copier machine for personal use etc. etc.?
Is this "stripping" the book publisher's global copyright notice, or is it
just recognizing that the article continues to have its permissive license
even when it's part of the book?

--
I suggest you call for help,                    John Cowan
or learn the difficult art of mud-breathing.    cowan@...
        --Great-Souled Sam                      http://www.ccil.org/~cowan

[Wilson, Andrew-2]

 
John Cowan wrote:

> Let's consider a hypo in a different domain.  I get a book with
> the ordinary copyright license: All Rights Reserved, specifically
> copying is forbidden.  However, one of the book's chapters is in fact a
> verbatim article published by the ACM, with the standard ACM copyright
> notice intact:
>
> Copyright© 20XX by the Association for Computing Machinery,
> Inc. Permission to make digital or hard copies of part of this
>     ....
>
> Do you argue that I must not (not should not pragmatically, but must not)
> exercise the rights granted there with respect to that part of the book,
> and make a copy of it on a copier machine for personal use etc. etc.?
> Is this "stripping" the book publisher's global copyright notice, or is it
> just recognizing that the article continues to have its permissive license
> even when it's part of the book?

This one's actually pretty easy.  The book publisher has an anthology
copyright on the compilation.  You can still make fair use
of individual articles.  However, the creator of the derivative
program we were discussing in the GPL scenario, which is itself a
work of authorship in that it combines GPL and BSD code
to form a new expression, has a no-adjective
copyright in the new expression, not an anthology copyright.
I don't see the argument that you can reach into a copyrighted
work and extract subsections without following the copyrighted work's license.
Again, in the GPL hypo, the only license you have from the
creator of the derivative and from any subsequent distributors is GPL.

IANAL, TINLA, etc.

Andy Wilson
Intel open source technology center

[John Cowan]

Wilson, Andrew scripsit:

> This one's actually pretty easy.  The book publisher has an anthology
> copyright on the compilation.  You can still make fair use
> of individual articles.

In this case, licensed use rather than fair use as such.

> However, the creator of the derivative
> program we were discussing in the GPL scenario, which is itself a
> work of authorship in that it combines GPL and BSD code
> to form a new expression, has a no-adjective
> copyright in the new expression, not an anthology copyright.

Well, a derived work copyright, but that doesn't affect your point.

The catch, however, is that the typical tarball in which a work in
source code form is delivered is a lot like an anthology, only even
thinner because ordering is alphabetical.  That is, you have a bunch
of files in it, each under its own license.  If you compile and
link them into a binary and there are GPLed components, you must
follow the GPL with respect to that binary.  But just taking a file
out of the tarball, that's much more like copying the chapter as
described above.

> I don't see the argument that you can reach into a copyrighted
> work and extract subsections without following the copyrighted work's license.

If the work is simply assembled from those subsections, and each subsection
has its own license, the case begins to look different.

> Again, in the GPL hypo, the only license you have from the
> creator of the derivative and from any subsequent distributors is GPL.

True, but you have licenses from the creators of the parts as well.
The BSD license applies to anyone into whose hands the source code comes.

> IANAL, TINLA, etc.

/me too.

--
There are three kinds of people in the world:   John Cowan
those who can count,                            cowan@...
and those who can't.

[Wilson, Andrew-2]

 
John Cowan wrote:
> the typical tarball in which a work in
> source code form is delivered is a lot like an anthology, only even
> thinner because ordering is alphabetical.  That is, you have a bunch
> of files in it, each under its own license.  If you compile and
> link them into a binary and there are GPLed components, you must
> follow the GPL with respect to that binary.  But just taking a file
> out of the tarball, that's much more like copying the chapter as
> described above.

Ah, but we weren't talking about extracting an identifiable
individual file from an archive; the GPL hypo was reaching
inside a single GPL-licensed file and extracting a subsection
which is marked as being originally BSD-licensed.

Andy Wilson
Intel open source technology center

[Walter van Holst-2]

________________________________________
Van: John Cowan [cowan@...]

>Well, I see this through a common-law prism, of course, but it seems
>to me that the GPL is simply telling you that the licensor is not
>giving you certain rights.

The problem is that it is not the author that grants me the right to sue him in case the software is defective, I already have that right, even if he or she provided me the software for free.

>  It also notifies you that this may be
>overridden by operation of law: your legal system may give you
>rights (warrantability and fitness, e.g.) that cannot be disclaimed
>in this way.  It does not require you to surrender anything you had
>in the first place, and as such remains unilateral.

The problem is that the lack of warrantees can be disclaimed, only not unilaterally. I'd be rather hesitant to use this as an argument in favour of the GPL being unilateral, how much I'd like it to be.

Anyway, this horse has been flogged quite extensively past its untimely demise in the legal literature surrounding FOSS licenses and so far none that I am aware of is considers it of a problem with more than just academic value.

Regards,

 Walter

[John Cowan]

Wilson, Andrew scripsit:

> Ah, but we weren't talking about extracting an identifiable
> individual file from an archive; the GPL hypo was reaching
> inside a single GPL-licensed file and extracting a subsection
> which is marked as being originally BSD-licensed.

Well, I wasn't; I was talking about the identifiable-individual-file
case.  I agree that if a mere subsection is marked BSD, then it would
be dangerous to rely on that.

Are we in agreement now?

--
The experiences of the past show                John Cowan
that there has always been a discrepancy        cowan@...
between plans and performance.                  http://www.ccil.org/~cowan
        --Emperor Hirohito, August 1945

[Wilson, Andrew-2]

John Cowan wrote:

>> Wilson, Andrew scripsit:
>>
>> Ah, but we weren't talking about extracting an identifiable
>> individual file from an archive; the GPL hypo was reaching
>> inside a single GPL-licensed file and extracting a subsection
>> which is marked as being originally BSD-licensed.
>
> Well, I wasn't; I was talking about the identifiable-individual-file
> case.  I agree that if a mere subsection is marked BSD, then it would
> be dangerous to rely on that.
>
> Are we in agreement now?

I believe we are!  Thanks for the examples, they
helped define the problem and brought clarity.

Andy Wilson
Intel open source technology center

[Roger Fujii]

Walter van Holst <w.van.holst@...> wrote:
>> Roger's argument, with which I complete concur, is that the
>> end user who has received a copy from a distributor under GPL
>> does not have sufficient rights to strip off GPL and revert
>> to the original BSD.
>
> In case the code was otherwise unaltered, I would lean towards the opinion that anyone
> can strip off GPL and revert to the original BSD.

sigh.  Let's try this simple case:

I write program A.  I put BSD license headers on it, intending to release it under BSD.
I then change my mind, and I place the GPL on top of program A, not removing the BSD
parts because I'm lazy.  I give you program A under the GPL.

I will testify in court that
   A) There is no physical difference between what was originally BSD version and the GPL
      version minus any textual difference for the licensing.
   B) I gave you the program under the conditions of the GPL.

Do you honestly think you would win this argument that you can strip off the GPL just
because a BSD version exists somewhere?  I could even give someone else Program A under
BSD.  As long as I can prove that you didn't get it from them, you are bound by GPL.
Andrew Wilson's original statement is correct.   You have to right to strip off the GPL.
Another analogous examples would be if you distribute GPL program B.   Even if program B was
written by someone else, and can be freely gotten on sourceforge, you are still
REQUIRED to give someone who got it from yopu the source if they ask you for
it (at least for GPL2).   Same principle.  Path matters.

-r
--
Roger Fujii <rmf@...>

[Matthew Flaschen]

Roger Fujii wrote:

> I write program A.  I put BSD license headers on it, intending to
> release it under BSD.
> I then change my mind, and I place the GPL on top of program A, not
> removing the BSD
> parts because I'm lazy.  I give you program A under the GPL.

If the BSD headers were still there, I got it under both BSD and GPL.

>   B) I gave you the program under the conditions of the GPL.

You gave me it under my choice of license.

 > I could even give someone else Program A under
> BSD.  As long as I can prove that you didn't get it from them, you are
> bound by GPL.

I am not a lawyer, but I think the BSD is meant to be a license for all
third parties, regardless of where they get the code.  This seems like a
meaningless dispute, though.  If you release one copy under the BSD, why
try to keep other identical copies proprietary or more restricted?

> Another analogous examples would be if you distribute GPL program B.  
> Even if program B was written by someone else, and can be freely gotten on sourceforge, you
> are still REQUIRED to give someone who got it from yopu the source if they ask you
> for it (at least for GPL2).   Same principle.  Path matters.

This is an invalid analogy.  GPL specifically imposes the source
requirement on each /distributor/.

Matt Flaschen

[Matthew Flaschen]

Wilson, Andrew wrote:

>>From a pragmatic point of view, although GPL requires you to mark source
> code changes, in the real world this doesn't always happen.  The user
> who has received a copy of a derivative should see the original
> BSD copyrights intact in the sources, but he/she is probably ill-advised
> to assume
> the original code is also intact.

Okay, this is a valid point.  But /if/ the original BSD code is intact,
I think the user can use it (and only it) under that license.

Matt Flaschen

[Matthew Flaschen]

John Cowan wrote:

> Wilson, Andrew scripsit:
>
>> Ah, but we weren't talking about extracting an identifiable
>> individual file from an archive; the GPL hypo was reaching
>> inside a single GPL-licensed file and extracting a subsection
>> which is marked as being originally BSD-licensed.
>
> Well, I wasn't; I was talking about the identifiable-individual-file
> case.  I agree that if a mere subsection is marked BSD, then it would
> be dangerous to rely on that.

I was assuming the whole file (possibly multiple whole files) was BSD too.

Matt Flaschen