1 |
On 03/11/16 01:14 PM, Rich Freeman wrote: |
2 |
> On Thu, Nov 3, 2016 at 12:41 PM, Ian Stakenvicius <axs@g.o> wrote: |
3 |
>> |
4 |
>> Yep pretty much. The different in my suggestion would be that |
5 |
>> stabilization commits would be done by those that don't actually test |
6 |
>> said arch -- I think the way it worked before is that the |
7 |
>> stabilization commit was still done by an AT. |
8 |
> |
9 |
> ATs aren't developers, so they can't do commits. The commits were |
10 |
> done by developers who were part of the arch project. |
11 |
> |
12 |
> Unless you're using "AT" to refer to either a non-dev arch tester or a |
13 |
> developer member of the arch project team. It is possible that the |
14 |
> term wasn't used consistently, but on amd64 back when it was active I |
15 |
> don't think we used the term AT for anything other than non-devs. |
16 |
> |
17 |
|
18 |
Ahhh, I wasn't aware of that. I thought the AT (Arch Tester) label |
19 |
included both the dev and the non-dev members of the arch team. |
20 |
|
21 |
|
22 |
>> This lack of on-system |
23 |
>> verification from the committing dev would be why we'd need a higher |
24 |
>> level of trust from the non-dev ATs. |
25 |
> |
26 |
> Generally speaking the arch project member wasn't checking anything |
27 |
> other than that the AT was in fact an AT for their arch. So, it |
28 |
> required the same level of trust. |
29 |
> |
30 |
> However, you're right that the commits could be done by anybody, or as |
31 |
> I suggested they might even be automated. Arch testing really seems |
32 |
> to be the sort of thing that could go into some kind of tool that |
33 |
> tracks all the statuses/etc and as testers check in it just does the |
34 |
> commit behind the scenes. |
35 |
> |
36 |
|
37 |
Yes, I never even considered automation in this case. That could aid |
38 |
things significantly. |