1 |
On 21 October 2013 01:09, hasufell <hasufell@g.o> wrote: |
2 |
|
3 |
> what does the games team think of an official games repository? |
4 |
> |
5 |
> These are some arguments for such a repo: |
6 |
> * sunrise does not allow ebuilds when any version is already in the |
7 |
> tree (so no live or beta ebuilds for existing in-tree-games), so that |
8 |
> is not always a real alternative |
9 |
> * easier contribution and reviews |
10 |
> * let people contribute proprietary games ebuilds here which cannot be |
11 |
> added to the tree, because no dev owns the game |
12 |
> * let people contribute alpha/beta/live/experimental ebuilds here |
13 |
> * deprecate repositories like "gamerlay" which do not undergo any kind |
14 |
> of review and have low quality |
15 |
> * some of the work here can directly flow into the tree |
16 |
> * less decentralization of games ebuilds when there is a central |
17 |
> overlay for contribution and testing |
18 |
> * a lot of other herds already practice this workflow with success |
19 |
> |
20 |
> |
21 |
> Yes, overlays suck. But bugzilla as a review platform sucks more and |
22 |
> IRC is also not the first place people look for. |
23 |
> |
24 |
|
25 |
I feel most these concerns can be solved by simply migrating the existing |
26 |
SVN based games overlay to git. |
27 |
|
28 |
Branches are much less troublesome under git, and ebuilds can be segregated |
29 |
into branches by relative quality, with mandatory review to escalate the |
30 |
branch an ebuild is in. |
31 |
|
32 |
IME, gitifying things greatly reduces barriers to contribution, and the |
33 |
games svn repo doesn't have the same burdens to migration that gx86 does. |
34 |
|
35 |
So given that, why not just propose a git migration, and then we can work |
36 |
on improving the workflow afterwards. |
37 |
|
38 |
|
39 |
-- |
40 |
Kent |