1 |
On Tue, Oct 25, 2016 at 8:54 AM, Ulrich Mueller <ulm@g.o> wrote: |
2 |
> |
3 |
> Also, calling eclass functions could be considered linking. It is not |
4 |
> entirely clear to me if e.g. a binpkg built with a CDDL licensed |
5 |
> ebuild calling GPL licensed eclasses would be distributable at all. |
6 |
|
7 |
Honestly, I think the GPL linking argument is a difficult one at best, |
8 |
but setting that aside I think it is even harder to consider calling a |
9 |
function in an interpreted language "linking." Is it a violation of |
10 |
the GPL to execute a GPL binary from a bash script that is |
11 |
GPL-incompatible? Heck, is it a violation of the other license for |
12 |
the GPL bash interpreter to read and execute the non-GPL lines in the |
13 |
script? |
14 |
|
15 |
To me linking and word processing are actually on a continuum and I |
16 |
think it is hard to draw a line and say that the GPL prohibits one and |
17 |
not the other, but if you are going to try to draw a line I think |
18 |
interpreted languages are going to fall on the safe side of it. |
19 |
|
20 |
I guess it comes down to what are the essential elements of linking |
21 |
that leads one to believe that it constitutes a violation of copyright |
22 |
to do it without explicit permission? If there is agreement on that |
23 |
(which I think is harder to achieve than some seem to think), then the |
24 |
question becomes whether calling a function in an interpreted language |
25 |
contains those elements. |
26 |
|
27 |
> |
28 |
> So can we be strict there, please? Contributed ebuilds should have our |
29 |
> standard copyright header, or they will be rejected. |
30 |
> |
31 |
|
32 |
Certainly this is the current policy. The draft policy envisions a |
33 |
table of licenses for each project, and we of course can make that |
34 |
table as restrictive or free as desired. I do think it makes sense to |
35 |
whitelist licenses individually by project for the very sorts of |
36 |
reasons that you bring up. |
37 |
|
38 |
-- |
39 |
Rich |