kernel-libre (hopefully 100% Free) for Fedora 8 and rawhide

[Les Mikesell]

David Woodhouse wrote:
>
>> Do you think it is part of the kernel-as-a-whole when it loads/runs 
>> completely separately on some other component? 
> 
> Another strange question. Interestingly, the GPL says nothing about
> running the code. In fact, it doesn't even mandate that you should be
> permitted to run the code. It explicitly limits its scope to copying,
> modification, and distribution.
> 
> So it doesn't matter very much where the code runs, if you can run it at
> all -- the important part is whether we _distribute_ the parts together
> as part of a coherent whole; comprised of both the firmware and the
> GPL'd Linux-side driver (and, of course, the rest of the kernel).

Every interpretation I've seen of the GPL had the work-as-a-whole 
defined as the running program and had nothing to do with how the bits 
were shipped.

> When we do _that_, the distribution of the whole must be on the terms of
> the GPL, whose permissions extend to the ENTIRE WHOLE -- to EACH AND
> EVERY PART REGARDLESS OF WHO WROTE IT.
> 
>> I don't see how you can have any interpretation other than it being an
>> unrelated chunk that is conveniently aggregated with a loader to make
>> a piece of hardware behave better. 
> 
> The GPL clarifies that it doesn't mean to cover "mere aggregation on a
> volume of a storage medium" -- that is; you don't infect other code with
> the GPL merely by storing it on the same hard drive or CD as GPL'd code.
 >
> It's very hard to argue that including it directly in the kernel, where
> it forms a coherent and necessary part of the overall result and is
> simultaneously not usable for any _other_ purpose, is 'mere aggregation
> on a storage medium'.

No, it's not hard at all.  They don't have much relationship except for 
the bits being conveniently stored together for a while.

 > Perhaps I was overstating the case slightly when I
> said it was "completely crazy", but it's certainly quite far-fetched.

If you accept the premise that shipping a shared library and a program 
that uses it separately still makes the running program covered by the 
GPL if either part was, then you have to concede that how the bits are 
aggregated for transport is irrelevant.

>>   Loading it as the kernel loads doesn't make it any more or less a
>> part of the kernel work-as-a-whole.
> 
> Again, loading is irrelevant. Distributing it as part of the kernel
> makes it... well, distributed as part of a whole which is a work based
> on the Program. With all that that implies; even for sections of that
> whole which are clearly not derived from the Program.

No, it makes it mere aggregation, which implies nothing else.

-- 
   Les Mikesell
    lesmikesell at gmail.com