On Mon, 2008-06-09 at 17:33 -0400, Alan Cox wrote:
I do not believe they are "part of a whole which is a work based
this case I believe they are independant works that are merely aggregated.
To be clear, I also believe them to be independent works -- I don't
think there's any debate about that.
But they are independent works which are being combined into a larger,
coherent whole and distributed that way. I do not believe that they are
"merely aggregated on a volume of a storage or distribution medium" like
a magazine cover CD. And thus, I believe that the GPL extends "to each
and every part, regardless of who wrote it".
_That_ is the 'alternate position'; not this:
The alternate position would be that a general interface between two
independent works somehow made them one.
Not at all. That would be silly, and is clearly _not_ my position.
The general interface between the firmware and the kernel exists even
when you distribute the firmware as a separate work. And the GPL clearly
doesn't apply to it when you distribute it as a separate work.
It is the _distribution_ of a collective whole based on both firmware
and kernel together, which makes the difference under the GPL.
That would just as equally say that firefox and the web are one work.
That's a complete non-sequitur -- nobody would ever attempt to
_distribute_ an aggregation of "firefox" and "the web" together as a
single collective whole. You really do seem to have missed the point
that it's about _distribution_, of both derived and _collective_ works.
I believe that if we follow your logic, we should also be able to
distribute GPL'd code linked against proprietary libraries. Yes, we've
combined them together into one executable -- but evidently we can call
it "merely aggregation" and get away with anything. There's a general
interface between the independent parts, which you seem to believe
excuses the combination, yes?
We should be able to distribute binary-only drivers actually linked into
the kernel too. If we accept that they are independent works in the
first place¹, then 'merely aggregating' them into the vmlinux should be
Now admittedly, there _is_ a grey area here -- it's not particularly
clear-cut where "collective works based on the Program" ends, and where
"mere aggregation on a volume of a storage or distribution medium"
But even though there's a grey area, I don't think it's at all realistic
to stretch that boundary as far as you seem to be claiming we should.
You seem to be arguing as if you believe that incorporation of any
'independent' work automatically means that we should consider the
combined whole to be 'mere aggregation on a volume of a storage or
distribution medium', rather than a collective work with associated
requirements under the GPL.
Yet the GPL _explicitly_ states that it refers to such collective works
-- with explicit reference to sections of that whole which, when
distributed separately, would be considered independent and separate
works in themselves. Your viewpoint would render the latter half of §2 a
verbose no-op, and is clearly at odds with the intent, as well as the
letter, of the GPL.
But as ever, there is no definitive answer to that until/unless it's
ruled in court.
¹ which admittedly is disputed by some, but not the point here.