License Committee Report for July 2007

[John Cowan]

Wilson, Andrew scripsit:

> Perhaps a non-GPL example will help clarify.  As is well known,
> Mac OS X contains great gobs of BSD-licensed code which Apple
> re-releases under their own proprietary license.

You mean the APSL?  Not a proprietary license, and IMHO Apple's right to
hike off the BSD license and put the APSL on the FreeBSD-derived code
instead is pretty shaky.

It's true that Apple uses the BSD-granted right to release some of that
code as proprietary binary bits; hopefully they obey the BSD license
and carry the license text along as well.  

> With BSD, anything that is not forbidden is allowed.  You may apply your
> own additional terms and conditions (e.g. relicense).   [...]
> Just as Apple may add their EULA on top of BSD, you or I or anyone else
> may add GPL on top of BSD-licensed code and distribute the result.

The GPL begins thus:

# 0. This License applies to any program or other work which contains
# a notice placed by the copyright holder saying it may be distributed
# under the terms of this General Public License.

"By the copyright holder".  You aren't the copyright holder.  You can
place such a notice, but I fail to see how it has any legal force at all.

--
The first thing you learn in a lawin' family    John Cowan
is that there ain't no definite answers         cowan@...
to anything.  --Calpurnia in To Kill A Mockingbird

John Cowan wrote:

> Brendan Scott scripsit:
>
>> "But when you distribute the same sections as part of a whole which
>> is a work based on the Program, the distribution of the whole must be
>> on the terms of this License, whose permissions for other licensees
>> extend to the entire whole, *and thus to each and every part regardless
>> of who wrote it.*"
>
> Permissions, yes.  All the parts must offer the permissions of the GPLv2.
> But it is not required that all the parts likewise impose the restrictions
> of the GPLv2, only that the work as a whole do so.

Assuming, for the sake of argument, that it only applies to permissions...

If you concede that the GPLv2 permissions apply to each part, including the BSD part, aren't you also conceding that there is relicensing of the BSD part?


Brendan

Alexander Terekhov wrote:

> On 8/1/07, Brian Behlendorf <brian@...> wrote:
>> On Wed, 1 Aug 2007, Alexander Terekhov wrote:
>>> Yet you believe the claim in
>>> http://www.linuxrising.org/files/licensingfaq.html that:
>>>
>>>       In order to link GStreamer [an LGPLv2 library] to Totem [a GPL
>>>       application], you need to use section 3 of the LGPL to convert
>>>       GStreamer to GPL.
>>>
>>> is simply wrong?
>>>
>>> I can't follow your reasoning.
>> I can.  It's wrong, because you can link an LGPLv2 library against other
>> code licensed under any other license, so long as some specific conditions
>> are met (the LGPL library can be replaced with an updated version, etc).
>> The LGPL code did not need to be "converted", because the GPL effects
>> those LGPL conditions.
>
> As far as I'm concerned, the LGPLv2 could have conditioned its
> "additional permissions" on renaming One Microsoft Way into One FSF
> Way. The proposition preceding my "yet" that you snipped was that
> (quoting John Cowan)
>
> <quote>
>
> Software that your program links to does not count as "other software".
>
> </quote>
>
> regarding the GPL and OSD#9.
>
>> The GPL may cover the work as a whole, but the LGPLv2 still covers the library.
>
> Sorry, but what exactly is "work as a whole", and "the LGPLv2 still
> covers the library" in this context?

I think the issue is drawing a distinction between the library in the abstract and the particular copy of the library which is linked.

Brendan

[Michael Poole]

Brendan Scott writes:

> John Cowan wrote:
>> Brendan Scott scripsit:
>>
>>> "But when you distribute the same sections as part of a whole which
>>> is a work based on the Program, the distribution of the whole must be
>>> on the terms of this License, whose permissions for other licensees
>>> extend to the entire whole, *and thus to each and every part regardless
>>> of who wrote it.*"
>>
>> Permissions, yes.  All the parts must offer the permissions of the GPLv2.
>> But it is not required that all the parts likewise impose the restrictions
>> of the GPLv2, only that the work as a whole do so.
>
> Assuming, for the sake of argument, that it only applies to permissions...
>
> If you concede that the GPLv2 permissions apply to each part, including the BSD part, aren't you also conceding that there is relicensing of the BSD part?

The BSD license grants all the permissions that GPLv2 does, and then
some.  (Another way to look at it is that the BSD license grants the
same permissions, but does not make some of the restrictions, that the
GPL does.)  This is true whether the BSD-licensed bits are part of a
GPLed whole or not; there is no relicensing involved in either case.

Michael Poole

Michael Poole wrote:

> Brendan Scott writes:
>
>> John Cowan wrote:
>>> Brendan Scott scripsit:
>>>
>>>> "But when you distribute the same sections as part of a whole
>>>> which is a work based on the Program, the distribution of the
>>>> whole must be on the terms of this License, whose permissions
>>>> for other licensees extend to the entire whole, *and thus to
>>>> each and every part regardless of who wrote it.*"
>>> Permissions, yes.  All the parts must offer the permissions of
>>> the GPLv2. But it is not required that all the parts likewise
>>> impose the restrictions of the GPLv2, only that the work as a
>>> whole do so.
>> Assuming, for the sake of argument, that it only applies to
>> permissions...
>>
>> If you concede that the GPLv2 permissions apply to each part,
>> including the BSD part, aren't you also conceding that there is
>> relicensing of the BSD part?
>
> The BSD license grants all the permissions that GPLv2 does, and then
> some.  (Another way to look at it is that the BSD license grants the
> same permissions, but does not make some of the restrictions, that
> the GPL does.)  This is true whether the BSD-licensed bits are part
> of a GPLed whole or not; there is no relicensing involved in either
> case.

The wording each license uses is different, so this is not obvious to me (eg: the GPL refers to "copying, distribution and modification" as well as "running" the program.  The BSD talks about use and redistribution and (implicitly) modification).

Can you explain why you think the BSD permissions are (at least) a superset of those in the GPL?

Thanks


Brendan

[Matthew Flaschen]

John Cowan wrote:

> I agree with *that*, I just see no necessity to do so.  Your arguments
> are not addressing the powers granted by section 6 (or 4).

I think what Andrew Wilson is saying is that while LGPLv3 section 4
allows linking the LGPLv3 library with code under GPLv2 (or indeed any
license), GPLv2 /doesn't/ allow that same link.  Obviously, both would
have to allow the combination for it to be legal.

Matt Flaschen

[Michael Poole]

Brendan Scott writes:

> The wording each license uses is different, so this is not obvious to me (eg: the GPL refers to "copying, distribution and modification" as well as "running" the program.  The BSD talks about use and redistribution and (implicitly) modification).
>
> Can you explain why you think the BSD permissions are (at least) a superset of those in the GPL?

In short, plain reading and the common understanding of the two.

Redistribution includes copying and distribution.  "Use", when it
comes to a program, includes running it.  The BSD license permits
"[r]edistribution and use in source and binary forms, with or without
modification" if the copyright, license and disclaimer notices are
preserved.

The GPLv2 license says (in section 0) that it does not restrict
running of the program.  Section 1 allows distribution of unmodified
source forms, if the copyright, disclaimer and license notices are
preserved.  Section 2 allows distribution of modified source forms,
subject to certain copyleft restrictions (plus those of section 1).
Section 3 allows distribution of binary forms, with or without
modification, subject to further copyleft restrictions (plus those of
sections 1 and 2).  Each of those sections are at least as strict, and
in no case broader, than the BSD license.  Sections 4 through 12 do
not grant permissions.

Also, GPLv2's section 6 amplifies that the license to exercise
copyright-reserved rights comes from the original licensor (rather
than from the redistributor).  This is consistent with the idea that
the original license still attaches to the code.

Michael Poole

[David Woolley (E.L)]

Brendan Scott wrote:

> The wording each license uses is different, so this is not obvious to me (eg: the GPL refers to "copying, distribution and modification" as well as "running" the program.  The BSD talks about use and redistribution and (implicitly) modification).
>
> Can you explain why you think the BSD permissions are (at least) a superset of those in the GPL?

One of the reasons goes beyond the wording of the licence.  It is well
known that BSD's intention was that the code be incorporated into
proprietary binaries with typical commercial licences, and it is the
case that Windows, and all modern Unix systems include such code without
providing source.

--
David Woolley
Emails are not formal business letters, whatever businesses may want.
RFC1855 says there should be an address here, but, in a world of spam,
that is no longer good advice, as archive address hiding may not work.

[John Cowan]

Brendan Scott scripsit:

> If you concede that the GPLv2 permissions apply to each part, including
> the BSD part, aren't you also conceding that there is relicensing of
> the BSD part?

No.

--
John Cowan  cowan@...   http://www.ccil.org/~cowan
O beautiful for patriot's dream that sees beyond the years
Thine alabaster cities gleam undimmed by human tears!
America! America!  God mend thine every flaw,
Confirm thy soul in self-control, thy liberty in law!
        -- one of the verses not usually taught in U.S. schools

[David Woolley (E.L)]

John Cowan wrote:
> Brendan Scott scripsit:
>
>> If you concede that the GPLv2 permissions apply to each part, including
>> the BSD part, aren't you also conceding that there is relicensing of
>> the BSD part?
>
> No.
>
Specifically  because the BSD licence already includes all of them.


--
David Woolley
Emails are not formal business letters, whatever businesses may want.
RFC1855 says there should be an address here, but, in a world of spam,
that is no longer good advice, as archive address hiding may not work.

[Wilson, Andrew-2]

[off-topic for purposes of LGPLv3 OSD compliance]

> Matthew Flaschen wrote:
>John Cowan wrote:
>
>> I agree with *that*, I just see no necessity to do so.  Your
arguments
>> are not addressing the powers granted by section 6 (or 4).
>
> I think what Andrew Wilson is saying is that while LGPLv3 section 4
> allows linking the LGPLv3 library with code under GPLv2 (or indeed any

> license), GPLv2 /doesn't/ allow that same link.  Obviously, both would

> have to allow the combination for it to be legal.

Well put.  With no way to retroactively
change GPLv2's copyleft, FSF would have had to special-case
allowing a GPLv2/LPGLv3 combined work by allowing such a
combined work to be governed by GPLv2's copyleft under LGPLv3.

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.

Andy Wilson
Intel open source technology center

[Wilson, Andrew-2]

whoops, fixing a typo.

[off-topic for purposes of LGPLv3 OSD compliance]

> Matthew Flaschen wrote:
>John Cowan wrote:
>
>> I agree with *that*, I just see no necessity to do so.  Your
arguments
>> are not addressing the powers granted by section 6 (or 4).
>
> I think what Andrew Wilson is saying is that while LGPLv3 section 4
> allows linking the LGPLv3 library with code under GPLv2 (or indeed any

> license), GPLv2 /doesn't/ allow that same link.  Obviously, both would

> have to allow the combination for it to be legal.

Well put.  With no way to retroactively
change GPLv2's copyleft, FSF would have had to special-case
allowing a GPLv2/LPGLv3 combined work by allowing such a
combined work to be governed by GPLv2's copyleft under LGPLv3.
                                                 ^^^^^^^^^^^^
                                                 rather than LGPLv3's
copyleft.    

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.

Andy Wilson
Intel open source technology center

[David Woolley (E.L)]

Wilson, Andrew wrote:

> allowing a GPLv2/LPGLv3 combined work by allowing such a
> combined work to be governed by GPLv2's copyleft under LGPLv3.

Which would allow trivial downgrading any V3 licensed code by adding V2
code with no real purpose other than to benefit from the proposed
special case.

--
David Woolley
Emails are not formal business letters, whatever businesses may want.
RFC1855 says there should be an address here, but, in a world of spam,
that is no longer good advice, as archive address hiding may not work.

[Wilson, Andrew-2]

David Woolley wrote:

> Wilson, Andrew wrote:
>
>> allowing a GPLv2/LPGLv3 combined work by allowing such a
>> combined work to be governed by GPLv2's copyleft under LGPLv3.
>
> Which would allow trivial downgrading any V3 licensed code by adding
V2
> code with no real purpose other than to benefit from the proposed
> special case.

This is quite correct.

Two points:

1) There is a great deal of GPLv2 code in the world.  As 'special cases'
   go, this one is extremely likely to occur in practice.

2) This discussion is in the context of /L/GPL.  Presumably, one
   chooses LGPL for one's code rather than GPL prime with the intent
   of enabling its broadest possible use.  See (1) for why preventing
   use with GPLv2 code seems inconsistent with the raison d'etre of
   LGPL.
   
Andy Wilson
Intel open source technology center

[Wilson, Andrew-2]

 
The specific question under discussion is what license applies to
a combination of GPL and non-GPL code.  My assertion is that GPL applies
to the entire combination, and that the creator of such a combination
must place additional terms and conditions, in the form of GPL, on
the underlying non-GPL license.  Downstream distributees of such a
combination
receive a license under GPL from the creator of the combination
and from any subsequent distributors of the combination.  The license
which
subsequently applies to the non-GPL portions is either the union of GPL
and the
original non-GPL license (which obviously must contain no provisions
which
would be in conflict with GPL), or an outright conversion of the
non-GPL licensed portions to GPL (as allowed, for example, by LGPL).  
Put succinctly, relicensing under GPL occurs to the non-GPL code.

John Cowan has disputed this and wrote:

> The GPL begins thus:
>
> # 0. This License applies to any program or other work which contains
> # a notice placed by the copyright holder saying it may be distributed
> # under the terms of this General Public License.
>
> "By the copyright holder".  You aren't the copyright holder.  You can
> place such a notice {on non-GPL code}, but I fail to see how it has
any legal force at all.

I do not believe this section is relevant to the question at hand,
the combination of GPL and non-GPL code, e.g. to a derivative work.  
I believe the relevant section is GPLv2, sec. 2, para (b), which reads

# You must cause any work that you distribute or publish, that in whole
# or part contains or is derived from the Program or any part thereof,
# to be licensed as a whole at no charge to all third parties
# under the terms of this License.

There is no obligation here to be the actual copyright owner of all
code in a derivative work in order to apply GPL terms to the derivative.

Rather, there an affirmative obligation to
license (note the verb) said entire derivative work under GPLv2.

Andy Wilson
Intel open source technology center

[John Cowan]

Wilson, Andrew scripsit:

> The specific question under discussion is what license applies to a
> combination of GPL and non-GPL code.  My assertion is that GPL applies
> to the entire combination,

I agree.

> and that the creator of such a combination must place additional terms
> and conditions, in the form of GPL, on the underlying non-GPL license.

I disagree.

In particular, I hold that although the work as a whole can only be
distributed under the terms of the GPL, any parts that you receive
with BSD (or similar) licenses applying to them (which may not be
removed, by their own terms) may be distributed by you under the
said terms, with no GPL restrictions.

To refute this, you will have to explain to me how the express
terms of the BSD license somehow no longer apply to the file in
which they are physically embedded.

> Downstream distributees of such a combination receive a license
> under GPL from the creator of the combination and from any subsequent
> distributors of the combination.

I agree.

> The license which subsequently applies to the non-GPL portions is either
> the union of GPL and the original non-GPL license (which obviously
> must contain no provisions which would be in conflict with GPL),

I'm not sure what you mean by "union", whether conjunction or disjunction.
In either case, I don't see how the restrictions can possibly apply to
otherwise unrestricted code.  The *permissions* of the GPL must apply
to each and every part of the code, as the GPL  says.

> or an outright conversion of the non-GPL licensed portions to GPL
> (as allowed, for example, by LGPL).

No problem with that, though I deny that it's necessary.

> I do not believe this section is relevant to the question at hand,
> the combination of GPL and non-GPL code, e.g. to a derivative work.  
> I believe the relevant section is GPLv2, sec. 2, para (b), which reads
>
> # You must cause any work that you distribute or publish, that in whole
> # or part contains or is derived from the Program or any part thereof,
> # to be licensed as a whole at no charge to all third parties
> # under the terms of this License.

I agree that the work *as a whole* must be licensed under the GPL.
That is different from claiming that any parts with non-restrictive
licenses of their own inherit the GPL restrictions.

> There is no obligation here to be the actual copyright owner of all
> code in a derivative work in order to apply GPL terms to the derivative.

Quite so.  For example, if you make a derivative work by modifying a
BSD-licensed single-file program, you may license your derivative
under the GPL, since you are the copyright owner of that derivative work.

> Rather, there an affirmative obligation to license (note the verb)
> said entire derivative work under GPLv2.

I agree.

--
Newbies always ask:                             John Cowan
  "Elements or attributes?                      http://www.ccil.org/~cowan
Which will serve me best?"                      cowan@...
  Those who know roar like lions;
  Wise hackers smile like tigers.                   --a tanka, or extended haiku

[Matthew Flaschen]

Wilson, Andrew wrote:

> David Woolley wrote:
>
>> Wilson, Andrew wrote:
>>
>>> allowing a GPLv2/LPGLv3 combined work by allowing such a
>>> combined work to be governed by GPLv2's copyleft under LGPLv3.
>> Which would allow trivial downgrading any V3 licensed code by adding
> V2
>> code with no real purpose other than to benefit from the proposed
>> special case.
>
> This is quite correct.
>
> Two points:
>
> 1) There is a great deal of GPLv2 code in the world.  As 'special cases'
>    go, this one is extremely likely to occur in practice.

I wouldn't say there is a great deal of GPLv2-only code in the world.  I
can think of only a few significant programs with that license.

Matt Flaschen

[Roger Fujii]

John Cowan wrote:

> Wilson, Andrew scripsit:
>>and that the creator of such a combination must place additional terms
>>and conditions, in the form of GPL, on the underlying non-GPL license.
>
> I disagree.
>
> In particular, I hold that although the work as a whole can only be
> distributed under the terms of the GPL, any parts that you receive
> with BSD (or similar) licenses applying to them (which may not be
> removed, by their own terms) may be distributed by you under the
> said terms, with no GPL restrictions.

This is incorrect.  Though in practice, it would be difficult
to prove a violation (assuming all you did was copy the GPL license
on top of the BSD code), or that anyone would really care.

> To refute this, you will have to explain to me how the express
> terms of the BSD license somehow no longer apply to the file in
> which they are physically embedded.

You are missing the mechanics of how licenses work.  These things
are all source/receiver based - that is, the last person in the
chain agrees with the conditions set by the previous person (assuming
that the conditions are legal and is legally allowed to do so).
If user A gives BSD program P to user B, who 'relicenses' it as GPL
and gives it to user C, user C must abide by the GPL for P, since THAT (GPL)
is the condition user C made to user B when he got the program.
If user C wants P under BSD, he must go to user A and get it from there.

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

[Matthew Flaschen]

Roger Fujii wrote:

> John Cowan wrote:
>> Wilson, Andrew scripsit:
>>> and that the creator of such a combination must place additional terms
>>> and conditions, in the form of GPL, on the underlying non-GPL license.
>>
>> I disagree.
>>
>> In particular, I hold that although the work as a whole can only be
>> distributed under the terms of the GPL, any parts that you receive
>> with BSD (or similar) licenses applying to them (which may not be
>> removed, by their own terms) may be distributed by you under the
>> said terms, with no GPL restrictions.
>
> This is incorrect.  Though in practice, it would be difficult
> to prove a violation (assuming all you did was copy the GPL license
> on top of the BSD code), or that anyone would really care.

If there is an unmodified (I'm saying unmodified for simplicity.  But
i's possible modifications are also under BSD) BSD component in a GPL
work, the user of the GPL work can see for themselves that the BSD
author is giving BSD permissions to all third parties.  It doesn't
matter which physical copy of the BSD work they use, or who they got it
from; they have the permissions granted by the copyright holder.

Matt Flaschen

[David Woolley (E.L)]

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".

--
David Woolley
Emails are not formal business letters, whatever businesses may want.
RFC1855 says there should be an address here, but, in a world of spam,
that is no longer good advice, as archive address hiding may not work.