1 |
On Wed, Sep 20, 2006 at 03:36:07PM +0200, Krzysiek Pawlik wrote: |
2 |
> Simon Stelling wrote: |
3 |
> > I would like you to share your comments on the attached GLEP with me. |
4 |
> |
5 |
> I like the idea with one exception: |
6 |
> |
7 |
> > Licenses that need to be explicitly accepted before installation of a package |
8 |
> > (and only these) should be package.masked by default with a header like |
9 |
> > the following: |
10 |
> > |
11 |
> > :: |
12 |
> > |
13 |
> > # Simon Stelling <blubb@g.o (20 Sep 2006) |
14 |
> > # This license needs to be agreed on explicitly to be considered |
15 |
> > # legally binding. |
16 |
> > # By unmasking and installing the package you agree with its terms. |
17 |
> > txt-licenses/wierd-license |
18 |
> |
19 |
> Why not make the ebuild ask for confirmation? Would work with versioned licenses |
20 |
> (for example: txt-licenses/wierd-license-2.1 and |
21 |
> txt-licenses/wierd-license-2.999 - both would require ACK). Breaks portage in a |
22 |
> way it's interactive, but it's already happening in few ebuilds |
23 |
> (eutils.eclass::check_license()). |
24 |
|
25 |
Thats one of the basic flaws with this proposal; that data isn't |
26 |
easily represented to the front end code, thus it makes doing |
27 |
proper eula confirmation that much harder. |
28 |
|
29 |
Also, yes, license is left behind, but hacking up the front end code |
30 |
to walk all deps trying to identify license deps (instead of just |
31 |
using the license metadata key) is pretty fugly in comparison to what |
32 |
glep23 proposed. |
33 |
|
34 |
Further... glep23 is simple, and there already; data is ready to go, |
35 |
all that is required is a patch. |
36 |
|
37 |
So... write a patch (this isn't that hard), or mangle the 24k ebuilds |
38 |
in the tree while losing capabilities... ;) |
39 |
~harring |