Fedora Freedom and linux-libre

[Les Mikesell]

Alexandre Oliva wrote:
> 
>> If it's silly, then you shouldn't have any problem showing where it is
>>>> executed as part of "the Program",
>>> Execution is irrelevant, because copyright law doesn't apply to this
>>> act.
> 
>> Does that mean you disagree with the FSF claim that it is not
>> permitted to distribute non-GPL'd software that dynamically links to
>> unique libraries only available under the GPL?
> 
> I agree that the GPL doesn't grant permission for the distribution of
> such derived works under terms others than those specified in the GPL.

OK, then the relevant question becomes whether the kernel or the 
firmware is a derived work, not whether they are distributed together or 
not.

> 
>> And the GPL's 'work as a whole' concept
> 
> I don't know what you're talking about.  "work as a whole" appears
> only once in the license, and there 'work' is not a verb, but rather
> part of the "modified work" phrase.

Once is enough - and it is the definitive requirement:
"These requirements apply to the modified work as a whole. If 
identifiable sections of that work are not derived from the Program, and 
can be reasonably considered independent and separate works in 
themselves, then this License, and its terms, do not apply to those 
sections when you distribute them as separate works."

The question is, does taking some code and an opaque binary blob and 
sticking them in the same file make a 'work as a whole' or is it 
identifiable sections of code and separate data that are not derived 
from the Program?

>>> It is *distributed* as part of the program, and that's what copyright
>>> law restricts by default. 
> 
>> Which is irrelevant if you have permission to distribute all parts.
> 
> If they form a coherent whole to the point of being regarded by a
> court, according to copyright law, as a derived/collective work, then
> it is relevant.  And then, this is not the case we're talking about
> here.

How is it different?  The question is whether firmware and microcode are 
part of the kernel 'work as a whole', which, I don't think depends on 
how they are distributed at all.

> We're talking of a case in which one part became an integral,
> essential and inseparable part from the other, so the distinction is
> definitely no longer irrelevant, because it's a different case.

Are you sure that the opaque blob of firmware can't be replaced by any 
other opaque blob of bits without affecting the other part?   In which 
case it's just aggregated and along for the ride.

>> I don't see how you can say it is inseparable when the firmware
>> downloader understands the separation perfectly
> 
> You got your facts wrong, I'm afraid.  The code in question doesn't
> even use the firmware downloader.  It absolutely requires the code to
> be part of the source code.

Would the code continue to work if you replace those bits with something 
else that would work in the hardware it loads?

>> it's a part of the program since it gets
>> installed on different hardware - except maybe for the CPU microcode.
> 
> So, when you use say Google Docs, just because some part of the
> program is shipped to your computer while another keeps on running on
> Google's servers, they're not part of the same program named "Google
> Docs"?

I wouldn't assume that. But why not stick to the CPU microcode example? 
  I think it is the purest form of the issue in question.

> Just because a program uses Java RMI to transfer classes from one
> process to another (that may be running on a different machine), the
> classes are not part of that program?

Unless it is covered by the GPL with the 'work as a whole' restriction, 
the question doesn't make much sense, since a program can otherwise have 
any number of components under different licenses running together.  And 
  in the GPL case, I haven't seen even the FSF try to claim that things 
running on other machines or connected via pipes or sockets are covered 
parts of a 'work as a whole'.

> Do you even have an argument here, or are you just saying it, to fill
> in the void? :-)

I would like to see a definitive answer to delimit the 'work as a whole' 
where the GPL restrictions of any component apply, particularly in cases 
like microcode downloads and how shared libs would differ from code in 
ROM or firmware in flash.  I think your interpretation is wrong, and 
mine might be - but I suppose we aren't going to get one.

-- 
   Les Mikesell
    lesmikesell at gmail.com