| 1 |
On 04/11/2015 01:54 PM, Rich Freeman wrote: |
| 2 |
> |
| 3 |
> 1. What does proxy-maintainers lack in comparison to sunrise |
| 4 |
> exclusively. The immediate question is whether sunrise should be |
| 5 |
> migrated to proxy-maintainers, so this specific comparison is |
| 6 |
> important. |
| 7 |
> |
| 8 |
|
| 9 |
proxy-maintainers lack: |
| 10 |
1. a repository with a usable VCS |
| 11 |
2. an actual review workflow... @proxy-maintainers are just some sort of |
| 12 |
backup committers. it's not a hub for contributors to gather, discuss, |
| 13 |
get reviews and improve skills |
| 14 |
3. means to ensure the tree doesn't break |
| 15 |
4. actively look for and educate potential developers, even before the |
| 16 |
recruitment process |
| 17 |
|
| 18 |
So it should, if at all, be the other way around: dissolve |
| 19 |
proxy-maintainers, fix the sunrise workflow and make it the contribution |
| 20 |
hub again it once was. But I'm not actually advocating for that. I think |
| 21 |
the sunrise concept doesn't work anymore. |
| 22 |
|
| 23 |
What we need is a said hub, but not under the dominance of one |
| 24 |
project/repository. It should rather be a meeting point for various |
| 25 |
high-quality overlay developers. One problem with sunrise often was: no |
| 26 |
one really knew about e.g. java ebuilds, so we ended up telling people |
| 27 |
with those ebuilds to get a review elsewhere. |
| 28 |
That said, it makes more sense to directly contribute haskell ebuilds to |
| 29 |
the haskell overlay where people have the knowledge about these things |
| 30 |
instead of dumping everything in one "user" repository. |
| 31 |
But in order to know where to go, there should be some sort of central |
| 32 |
meeting point. |
| 33 |
|
| 34 |
> 2. What does proxy-maintainers lack in comparison to overlays in |
| 35 |
> general, beyond QA standards? (By QA standards I'm more concerned |
| 36 |
> with our QA goals such as not having security-vulnerable packages in |
| 37 |
> the tree, having consistent depgraphs, having PMS-compliant ebuilds, |
| 38 |
> etc. I'd rather not discuss changes to these, unless there really is |
| 39 |
> something most of us don't think is necessary. How we go about |
| 40 |
> achieving those goals is fair game. ie, one "benefit" of overlays is |
| 41 |
> that you can commit an ebuild that contains only line noise, and I'm |
| 42 |
> not so interested in that. However, maybe another benefit of overlays |
| 43 |
> is that you can go about quality in a different way that makes it |
| 44 |
> easier, ultimately reaching a level of quality comparable to the |
| 45 |
> gentoo repository.) |
| 46 |
> |
| 47 |
|
| 48 |
Pretty much the same as already said above. I'd also like to note that |
| 49 |
the gentoo tree isn't really the nonplusultra in terms of ebuild |
| 50 |
quality, because a lot of people commit stuff without really knowing |
| 51 |
what they are doing. So, dedicated overlays sometimes have higher |
| 52 |
quality. There's just no patrick or diego running global checks. But |
| 53 |
that problem has been solved by exherbo for years now, with far less |
| 54 |
manpower. I'll cut the infra-rant. Maybe mgorny will get us there. |
| 55 |
Pretty much the only one left who's actively trying to improve things. |
| 56 |
And that is both sad and dangerous. |
Archives