1 |
On Friday 23 September 2005 22:28, Chris Gianelloni wrote: |
2 |
> On Fri, 2005-09-23 at 10:38 +0900, Jason Stubbs wrote: |
3 |
> > On Friday 23 September 2005 06:09, Chris Gianelloni wrote: |
4 |
> > > it would be a good idea to give the user some way of knowing that a |
5 |
> > > package requires some additional purchased (or otherwise obtained) |
6 |
> > > portion that is not a distfile/tarball. |
7 |
> > |
8 |
> > It would be a good idea, indeed. RESTRICT="purchase" or somesuch |
9 |
> > parallel to RESTRICT="fetch" would solve this just as well. However, |
10 |
> > whenever adding new stuff like this, as a portage developer, I always |
11 |
> > ask what use of it can be made by portage? I can't see anything other |
12 |
> > than passing the information to a user interface to blink some text at |
13 |
> > the user... |
14 |
> |
15 |
> The problem stems from a fetch restriction (of any kind) not being |
16 |
> sufficient to be this marker. Take my classic doom3 example. There is |
17 |
> *nothing* restricted, download-wise. It just requires you have the data |
18 |
> files available, either via a CD, or copied to disk somewhere. It |
19 |
> *also* requires a CD key. I don't see where any kind of fetch |
20 |
> restriction-like device would accommodate this, as the restrictions are |
21 |
> on installation/execution, not on fetching. |
22 |
|
23 |
*Relax!* ;) |
24 |
|
25 |
I meant extending the fetch-restriction concept to include all cases where |
26 |
an ebuild is not fully self-contained; that is, cases where the ebuild is |
27 |
not capable of obtaining all necessary components to get a package to a |
28 |
fully-functional state. |
29 |
|
30 |
> Also, my proposal was based on several user requests to *have* some kind |
31 |
> of information showing that the package requires a purchase/license. |
32 |
|
33 |
This, I understood. |
34 |
|
35 |
> Since it seems that many are confusing this with the nature of the |
36 |
> package (check out the sun-jdk question), it leads me to believe that |
37 |
> this is best left alone and to simply wait for GLEP23 to be wide-spread, |
38 |
> then to revisit this, since LICENSE definitely isn't the place for it, |
39 |
> and neither is any kind of RESTRICT, as far as I can tell. |
40 |
|
41 |
Deferring discussion about it.. well, it scares me to be completely honest. |
42 |
It becomes another one of those "is this going to change the requirements?" |
43 |
that's always lurking. |
44 |
|
45 |
> I think I was looking for something where a user could tell before they |
46 |
> even decided to merge the package, via emerge -S or packages.gentoo.org, |
47 |
> or preferably both. |
48 |
|
49 |
Is that the only requirement? Just a boolean attribute that says the user |
50 |
will have to pay money in order to make use of the package? If so, it would |
51 |
seem that metadata.xml would be the perfect place, but I'll think on that |
52 |
and come back to it. |
53 |
|
54 |
> > So, if RESTRICT="fetch" were to be overloaded, there is the issue of |
55 |
> > both fetch-restricted and non-fetch-restricted downloads in the one |
56 |
> > package. I would think this issue exists already for some packages. How |
57 |
> > is it dealt with at the moment? |
58 |
> |
59 |
> Currently, we just restrict the whole thing, as we have no other choice. |
60 |
> On some packages, it sucks pretty badly if there's only a single |
61 |
> restricted tarball and many unrestricted tarballs. |
62 |
|
63 |
Perhaps leave this one for another time? :) |
64 |
|
65 |
-- |
66 |
Jason Stubbs |