1 |
On Sun, Mar 22, 2015 at 4:21 AM, Ulrich Mueller <ulm@g.o> wrote: |
2 |
>>>>>> On Sun, 22 Mar 2015, Jeroen Roovers wrote: |
3 |
> |
4 |
>> Better yet, you can claim copyright on a "compilation" which is |
5 |
>> probably what is effectively being done here. This is how people get |
6 |
>> away with defending their copyright on publications of (slightly |
7 |
>> modified/ abridged/ annotated, if at all) "compilations" of |
8 |
>> centuries old works. Because copyright law. |
9 |
> |
10 |
> It doesn't work like this in our case. The Portage tree is licensed |
11 |
> under the GPL-2. If someone takes a subset of ebuilds from it, it will |
12 |
> be a "derivative work" and it cannot be distributed under any license |
13 |
> other than the GPL-2. |
14 |
|
15 |
That is true for the ebuilds that were copied/modified. It isn't |
16 |
necessarily true for other stuff stuck in the same tarball. |
17 |
|
18 |
> |
19 |
> It is even doubtful if any third-party ebuilds added to such a tree |
20 |
> could be under a different license. If such ebuilds inherit from a GPL |
21 |
> licensed eclass or depend on (or are depended on by) other ebuilds, |
22 |
> they cannot "be reasonably considered independent and separate works |
23 |
> in themselves". |
24 |
> |
25 |
|
26 |
That has been a longstanding argument with the GPL (you can link with |
27 |
the LGPL under a non-free license, but not with the GPL). However, it |
28 |
has never actually been tested in court and I'm not convinced that it |
29 |
will actually stand up. The only content of the library/eclass/etc |
30 |
that is present in the "derived work" are the names of symbols and the |
31 |
name of the eclass and copyrighting those seems a lot like SCO arguing |
32 |
that they can copyright signal enums. They don't even get commingled |
33 |
when you run the program - each gets loaded into its own set of memory |
34 |
and the linker just fills in some memory addresses. |
35 |
|
36 |
I'd think a judge would be as likely as not to say that the GPL would |
37 |
only apply to the library itself, and that you could link whatever you |
38 |
want to that library under any license at all, and that effectively |
39 |
there is no distinction between the GPL and LGPL. That is, if you get |
40 |
the sort of judge who wouldn't give the win to SCO. |
41 |
|
42 |
-- |
43 |
Rich |