1 |
On Thu, Dec 14, 2017 at 3:34 PM, Kristian Fiskerstrand <k_f@g.o> |
2 |
wrote: |
3 |
|
4 |
> On 12/14/2017 01:01 PM, Rich Freeman wrote: |
5 |
> > In the beginning the system would be opt-in. Then once we have |
6 |
> > confidence that it is working well the flag could potentially be made |
7 |
> > opt-out. |
8 |
> |
9 |
> The only place I imagine this being a good idea is for the kernel, given |
10 |
> the strict no break of userland policy (but even they fail from time to |
11 |
> time). For rest of the applications, even if we add tools to help |
12 |
> automate part of the stabilization, I'd very much oppose it being |
13 |
> automated without being initiated / acked by the maintainer. |
14 |
> |
15 |
|
16 |
This reminds me a bit of advertising actually: |
17 |
|
18 |
"Nineteenth century Philadelphia retailer John Wanamaker supposedly said |
19 |
“Half the money I spend on advertising is wasted; the trouble is I don't |
20 |
know which half.” " |
21 |
|
22 |
I feel like stabilization are similar; we know many of them are likely safe |
23 |
and don't need humans. But many are unsafe and require validation. |
24 |
Can we tell them apart? Perhaps Rich's flag is sufficient? |
25 |
|
26 |
Its also a risk mitigation problem. Is it better for Gentoo to have more |
27 |
packages marked stable (so that stable is closer to HEAD). Or is it more |
28 |
valauble |
29 |
for Gentoo to have a very old stable (like debian) and then have a split |
30 |
keywords system? |
31 |
|
32 |
I'm skeptical the keywords for most packages matter, particularly on common |
33 |
arches. Remember this is usually software that upstream |
34 |
already tested and released; so most of the bugs would be ebuild / Gentoo |
35 |
related. |
36 |
|
37 |
-A |
38 |
|
39 |
|
40 |
|
41 |
|
42 |
> |
43 |
> -- |
44 |
> Kristian Fiskerstrand |
45 |
> OpenPGP keyblock reachable at hkp://pool.sks-keyservers.net |
46 |
> fpr:94CB AFDD 3034 5109 5618 35AA 0B7F 8B60 E3ED FAE3 |
47 |
> |
48 |
> |