1 |
Am Sonntag, 16. Juni 2019, 20:09:38 CEST schrieb Michał Górny: |
2 |
> |
3 |
> Some developers were recently complaining that we're turning Gentoo into |
4 |
> a hobbyist distro and that's apparently bad. |
5 |
> |
6 |
> Do you think Gentoo should allow for experimental and unstable features, |
7 |
> and possibly breaking changes that make Gentoo more interesting for |
8 |
> hobbyists? Or should we block breaking changes and become more |
9 |
> conservative for users who prefer stable distribution with minimal |
10 |
> maintenance burden? |
11 |
|
12 |
We're torn between extremes here, and I think nothing we do will make |
13 |
*everyone* happy. That said, we do have the duality stable/~arch, and that |
14 |
should already answer a large part of that question. In an ideal world, stable |
15 |
should be rock-solid and production worthy, and ~arch should be bleeding edge |
16 |
with occasional bugs and compile failures that come with that. |
17 |
|
18 |
Where that doesnt work (say, profiles) ... |
19 |
|
20 |
Let's start with the following assumption: Our main objective should be that |
21 |
Gentoo has a vivid, productive, and growing developer community. (If you think |
22 |
that doesnt sound right, you can try replacing "developer" with "productive |
23 |
contributor".) |
24 |
|
25 |
Developers exist in two overlapping types, broadly speaking: |
26 |
1) hobbyists who do something because it's interesting and cool |
27 |
2) employees who are paid to do something because it's useful |
28 |
|
29 |
So, we need to find a compromise between these two, with weight on the group |
30 |
that contributes to Gentoo most. |
31 |
|
32 |
Right now my personal feeling is that we're trying to set long deprecation |
33 |
times when something is going away, and that nevertheless migration to, say, a |
34 |
new profile tends to only *start* when we threaten that the deprecation time |
35 |
will be over soon and the old one will go away. |
36 |
|
37 |
We can't be stuck at the same level forever. Occasionally we will have to |
38 |
change something that requires manual intervention. Hey, if you have a big |
39 |
server farm, that's what all these horrible automation tools (puppet, rex, |
40 |
...) are good for. I think. Let's do it carefully, announce it, announce |
41 |
deprecation times, and give people time to do it. And then move on. |
42 |
|
43 |
|
44 |
-- |
45 |
Andreas K. Hüttel |
46 |
dilfridge@g.o |
47 |
Gentoo Linux developer |
48 |
(council, toolchain, base-system, perl, libreoffice) |