Linus thinks it's a grey area and he's not wrong.

Posted Dec 15, 2006 7:18 UTC (Fri) by xoddam (guest, #2322)
In reply to: Linus thinks it's a grey area and he's not wrong. by drag
Parent article: Binary-only modules not to be banned - this time

> That if
> [it's illegal to link against GPL'd software and not follow the
> terms of that license]
> then
> [adding open source shims and playing games with what is derived
> vs what is not derived is just silly pointless stuff, it's still
> a violation of the copyright license.]

It's a valid syllogism, but I do think that the premise
<it's illegal to link...> is false.

> But I guess you beleive that it's fine to link against GPL software
> and not follow the terms of the GPL software.

Erm ... I believe that it doesn't violate the GPL to develop
something which links with GPL software, and that it still
doesn't violate the *letter* of the GPL to release what you
have developed under the licence of your choice, including a
GPL-incompatible licence, providing you don't distribute
something derived from the GPL software.

There's an argument to be made that it violates the *spirit* of
the GPL. Personally I don't think it does, because the stated
purpose of the GPL is to preserve the freedom of the end-user with
respect to the licenced work. The restrictions on distributors
are to ensure that freedom is preserved for end users, and that
GPL'd code is not swallowed up within a proprietary architecture
without which it is useless. It is up to users to decide what
use to make of this freedom (and how much of it they are willing
to trade for convenience). ATI, nVidia and their customers are
all users in this context.

> So we are at a impasse.

The FSF claims that GPL-licenced libraries are not available for
non-free software to use. This case is one of the weakest I've seen
them make and I'm unconvinced that it has a basis in law (besides
the barrier to entry imposed by integrators and distributors, who are
barred from distributing combined works with incompatible licences).
Normally I'm a strong FSF supporter, but I'm just not convinced in
this instance.

The real reason I weigh in on the subject at all is that people seem
to be adducing reasons for the illegality of binary kernel modules
which haven't actually been made by the people who claim that the
binary modules violate their copyright. The Kororaa shutdown came
down to the fact that their live CD included a kernel image together
with ready-made GPL-incompatible driver modules:

http://kororaa.org/index.php?entry=entry060521-200059

Various people have said words to the effect that the mere existence
of a GPL-incompatible module violates kernel copyright. None of
them has made a convincing case for this, though. The most coherent
argument I've heard involved the complexity of certain kernel header
macros, here on lwn.net when discussing Kororaa:

http://lwn.net/Articles/184885/
http://lwn.net/Articles/185169/

I'm not entirely convinced, and I've never seen any of the actual
copyright holders or lawyers try to argue on this basis.

Linus thinks it's a grey area and he's not wrong.

Posted Dec 15, 2006 11:15 UTC (Fri) by drag (guest, #31333) [Link]

Don't worry. you've given me good things to think about.

It's also worth noting that Kororaa is still distributing it's software, but it is now using open source drivers to do it. They basicly said the same thing you did which is that even though it's gray it's against the spirit and general wishes of the kernel team so they decided to stop shipping binary driveres.

I think that in situations like this intent means a lot. I think that if the copyright owner of the GPL'd work makes it very obvious when questioned that they do not allow linking of propriatory software then it's going to have a different effect then if the author was more permissive.

But I am so far from a lawyer that it's not even funny. So what do I know?
(not a whole lot)