1 |
On Mon, Jul 1, 2013 at 3:08 PM, William Hubbs <williamh@g.o> wrote: |
2 |
> On Mon, Jul 01, 2013 at 08:37:57PM +0200, Chí-Thanh Christopher Nguyễn wrote: |
3 |
>> hasufell schrieb: |
4 |
>> >> They can each have their own ebuild in portage. I do not think that overlays |
5 |
>> >> are the solution here. |
6 |
>> > |
7 |
>> > That idea is so bad I hope we will never see it happen. |
8 |
>> |
9 |
>> That's what GLEP 39 explictly allows. |
10 |
> |
11 |
> I have to agree on this point. glep 39 allows, and should allow |
12 |
> competing projects. |
13 |
|
14 |
GLEP 39 ALLOWS me to make a competing apache ebuild, or a competing |
15 |
amd64 arch. It doesn't FORCE me to do so. |
16 |
|
17 |
If all I want to do is introduce some optional feature distro-wide |
18 |
that doesn't impact anybody who doesn't want to use it (aside from |
19 |
trivial numbers of inodes), then I shouldn't HAVE to fork every |
20 |
package in the tree to do it. |
21 |
|
22 |
This isn't about taking away the freedom of people to fork things, |
23 |
this is about taking away the freedom of Maintainers to force people |
24 |
to fork things. |
25 |
|
26 |
>> > I think we will not improve as a distro if we do not redefine our |
27 |
>> > priorities. |
28 |
>> > I have the feeling that our work is not user-centered anymore, but |
29 |
>> > developer-centered and that concept is simply wrong and no sane business |
30 |
>> > manager would ever disagree. |
31 |
> |
32 |
> We are a group of volunteers, not a business, so I'm not sure how much |
33 |
> the business model can apply to us. |
34 |
|
35 |
Sure, but why would anybody volunteer to work on an unusable distro? |
36 |
Just what is Gentoo's purpose? If all I wanted was a set of packages |
37 |
that only could be reliably used in some particular configuration I |
38 |
could take my pick of binary distros. |
39 |
|
40 |
I'm a developer because I'm also a user. Nobody pays me to work on |
41 |
ebuilds - I'm scratching my itch. Nobody is asking maintainers to |
42 |
scratch somebody else's itch - they just have to stay out of the way |
43 |
when somebody else chooses to do so. |
44 |
|
45 |
> By the nature of being a source based distribution, we can offer a |
46 |
> uniform user experience without being a poor copy of any binary |
47 |
> distribution, and that uniform user experience could be far more |
48 |
> flexable than any binary distribution. |
49 |
|
50 |
Absolutely, but that flexibility depends on a certain amount of |
51 |
standardization. If every package maintainer was free to make up |
52 |
their own arch keywords there would be chaos (x86_64 vs amd64 |
53 |
anyone?). No problem - just fork every package in the tree and users |
54 |
can try to guess whether apache or apache-fixed is the one that will |
55 |
work better with their choice of php or php-improved. Don't like |
56 |
USE=X? Just make your package have IUSE=x - somebody can fork it if |
57 |
they don't like it. |
58 |
|
59 |
If all you want is a bunch of clean upstream tarballs, go download |
60 |
them from upstream and roll your own LFS. If we want to offer choices |
61 |
to users or ourselves in any kind of usable way, then we need to |
62 |
cooperate in their implementation. That means listening to everybody, |
63 |
but ultimately arriving at a decision. |
64 |
|
65 |
Rich |