1 |
On Sat, 11 Apr 2009 14:40:27 +0200 |
2 |
Maciej Mrozowski <reavertm@××××××.fm> wrote: |
3 |
|
4 |
> Let's look at tree - one thing can be said about each package - it belongs to |
5 |
> some herd or (doesn't, and it's with status maintainer wanted or maintained by |
6 |
> individual developers). |
7 |
> So creating separate repository for each herd is the most obvious (and naive) |
8 |
> idea. |
9 |
|
10 |
While this may seem like a good idea for those projects and herds that are |
11 |
well defined, well-oiled machines, I can count the number of these projects |
12 |
in Gentoo on one hand. |
13 |
|
14 |
> Pros are the following: |
15 |
> - project members taking care of some herd (or belonging to herd?) receive |
16 |
> (and have access) (only) to repository they are interested in, resulting in |
17 |
> smaller pulls/pushes |
18 |
|
19 |
...and isolating them from the rest of the tree, as well as the rest of |
20 |
developers from them. |
21 |
|
22 |
> - some level of isolation - gives possibility to restrict access (for example: |
23 |
> "only toolchain and arch teams allowed here") |
24 |
|
25 |
This is counter to the ideals behind Gentoo, blah blah. In reality, if |
26 |
someone screws up toolchain stuff, they only do it once. |
27 |
|
28 |
> - profile can be (should be actually) separated in another repository and |
29 |
> developed easier |
30 |
|
31 |
Repositories, when split, are best split into atomic units. Separating |
32 |
profiles from packages, given their dependence on each other, is a |
33 |
non-starter. |
34 |
|
35 |
I'm not sure what we'd do with eclasses either. Obviously they |
36 |
can't be split into herds, but they need to be updated with the ebuilds |
37 |
themselves. |
38 |
|
39 |
In Gentoo development the atomic unit is the repo itself, for better or worse. |
40 |
|
41 |
> Some cons: |
42 |
> - projects are now more dependant on other projects and its responsiveness, |
43 |
> unless access is granted to all repositories for every developer |
44 |
|
45 |
Which is one of the biggest problems we already have IMHO. |
46 |
|
47 |
> - not settled yet what to do with orphaned/proxy maintained packages and herd- |
48 |
> switching |
49 |
|
50 |
We have ~800 maintainer-needed packages right now. We have multitudes more |
51 |
that belong to herds with no real people in them, or have unresponsive |
52 |
maintainers. Cutting them off from the rest of the tree by making them |
53 |
harder to get to isn't going to improve things. |
54 |
|
55 |
|
56 |
|
57 |
-- |
58 |
gcc-porting, by design, by neglect |
59 |
treecleaner, for a fact or just for effect |
60 |
wxwidgets @ gentoo EFFD 380E 047A 4B51 D2BD C64F 8AA8 8346 F9A4 0662 |