On Fri, 2007-06-15 at 06:03 -0400, Daniel Hazelton wrote:
You're interpreting 'sections' to mean individual linker sections? And
you think it's talking about distributing those 'sections' "as separate
works"? Despite the fact that all the rest of the language in the
document is high-level and doesn't even mention _linking_?
That's... an interesting interpretation.
The GPLv2 doesn't "magically" change the law, and has no need to. The
above is just a demonstration that a licence can have conditions which
involve things _other_ than derived works.
I can release code under a 'viral' licence which requires you to release
_everything_ you write for a whole year under that same licence. You
don't _have_ to obey, and there's no suggestion that your own code would
be 'derived' from mine -- but if you don't follow the conditions of the
licence, then you don't get permission to use my code. There's no
"magical" change to the law required.
It's just a demonstration that a licence _can_ make requirements about
non-derived code. You seemed to be making two bogus claims -- first that
it _can_ not, and then that the GPL _does_ not.
I've dealt with the first; let's look at the second, leaving aside your
weird digression about bison... (hint: it's about the code stubs in the
output which weren't _produced_ mechanically in the first place; they
were only _put_ there mechanically by the software).
Yes. That's why I said 'not necessarily' rather than 'no'. If it just
happens to be on the same hard drive / tape / CD-ROM that's not
important. The important question is whether it's distributed 'as a
separate work' or whether it's part of a larger, coherent whole.
If by 'correctly' you mean I should interpret 'sections' to mean linker
sections, then you're right -- I really can't bring myself to read it
that way. Of course, you're not actually _wrong_ until/unless it's
interpreted in court. But I'm willing to place bets :)
I'm certainly confused as to what you're talking about _now_; it seems
to make little sense. What do you mean by 'this' in the context of 'is
this what the process of making a module does?'? And how is it relevant?
You said that modules aren't derivative works. I said 'Who cares?'
because there are far more obvious reasons why the module would be under
GPL, because of the GPL's requirements about collective works. What part
of that confuses you? You seem to _keep_ going back to talking about
derived works.
The GPL applies if you _distribute_ a work which combines such modules
and the kernel into a larger whole. You can do what you like if you
don't distribute it.
The 'mere aggregation' thing means you can even distribute them together
if they just happen to be side-by-side on a storage medium as if by
coincidence. But putting them together into a product which actually
uses and requires both of them is not 'mere aggregation'. Some people
_were_ bundling the ATI and nVidia modules together with the kernel in a
'product', and they stopped under threat of legal action.
The module on its own is questionable -- it is a matter of opinion
whether that's a derived work or not. But when you ship a product which
combines both kernel and module into a coherent whole, it doesn't
_matter_ if the module is a derived work or not. You are not permitted
to distribute the kernel unless your module is also licensed under the
GPL. Not because it's a derived work, but because the licence of the
kernel says that's the price you pay if you want to distribute the
kernel.
--
dwmw2
-
