1 |
On Mon, 24 Feb 2014 22:43:41 +0100 |
2 |
Chí-Thanh Christopher Nguyễn <chithanh@g.o> wrote: |
3 |
|
4 |
> Tom Wijsman schrieb: |
5 |
> >> Thanks, I would like to phrase the question a little more |
6 |
> >> precisely: |
7 |
> >> |
8 |
> >> I would like to ask council to state whether the QA team has the |
9 |
> >> authority to mandate a policy when there is neither general |
10 |
> >> agreement that this policy is a good thing, nor the policy would |
11 |
> >> avert any kind of immediate serious problem for users. |
12 |
> > That can be reduced to the question whether QA is considered |
13 |
> > serious. |
14 |
> |
15 |
> I don't think so. If the council answers this question in the |
16 |
> negative, QA still has its emergency powers and can enforce the |
17 |
> policy set by the council. |
18 |
> QA can also propose any new policy they consider useful to the |
19 |
> council, which can then hear the opposing sides and make a decision |
20 |
> based on that. |
21 |
|
22 |
You describe a Gentoo Developer, apart from the small difference that |
23 |
they have this emergency power; when it comes to it, that power is |
24 |
quite similar to having a Gentoo Developer revert according to policy |
25 |
and that deemed as OK (under the assumption of there being no QA team). |
26 |
|
27 |
> > (There is a majority that agrees as well as a serious problem to |
28 |
> > users in the GTK+ USE flag situation; the former can be deduced |
29 |
> > from the mailing list, the latter can be realized when controlling |
30 |
> > the USE flag among a lot of GTK+ packages. It is something that can |
31 |
> > be done later; on the other hand, if we categorize everything that |
32 |
> > way it halts progress) |
33 |
> |
34 |
> The gtk flag naming problem is not serious in the sense that it will |
35 |
> break user systems. |
36 |
|
37 |
It depends on how you define the words; but I find it a serious issue |
38 |
if you have an USE flag that works one way in a set of packages and |
39 |
works the other way in another set of packages, it is broken in the |
40 |
sense that the user now has to use package.use if he has a lot of |
41 |
packages that have a gtk flag and needs that flag to behave. |
42 |
|
43 |
> Additionally, it has been known and discussed for a long time. |
44 |
|
45 |
If it comes up again and again on ML, upsets people in bugs, have people |
46 |
bring it up independently to both QA and Council; it is of much more |
47 |
concern than for example a thread about small bug wrangling details. |
48 |
|
49 |
> Establishing a gtk USE flag policy did not become suddenly so urgent |
50 |
> that next council meeting was too far in the future. |
51 |
|
52 |
The discussions, pings and mails act as a nagging reminder. |
53 |
|
54 |
> > The GLEP states that we can |
55 |
> > |
56 |
> > 1) maintain a list of current "QA Standards"; |
57 |
> > |
58 |
> > 2) ensure all developer tools are in line with the current QA |
59 |
> > standards; |
60 |
> > |
61 |
> > 3) and apply it as per "In the event that a developer still insists |
62 |
> > that a package does not break QA standards, an appeal can be made |
63 |
> > at the next council meeting. The package should be dealt with per |
64 |
> > QA's request until such a time that a decision is made by the |
65 |
> > council." |
66 |
> > |
67 |
> > which effectively makes the "QA Standards" act as policy. |
68 |
> |
69 |
> I don't think that it is necessarily the case. "QA standard" could |
70 |
> also mean telling developers to either use gtk USE flags in the QA |
71 |
> team recommended way, or ensuring through other means that users to |
72 |
> not encounter unexpected breakage. |
73 |
|
74 |
The GNOME team's internal policy was a recommendation; as can be seen |
75 |
from the current state, that doesn't ensure anything. What other means |
76 |
would ensure that users do not have to micro manage with package.use? |
77 |
|
78 |
> Another possible interpretation is that QA standards codify |
79 |
> established good practices which are largely non-controversial (it |
80 |
> was attempted with saying that live ebuilds should be unkeyworded or |
81 |
> p.mask'ed IIRC).. |
82 |
|
83 |
It also depends on who discovers the new problems, who forms the |
84 |
explanations, who decides how to fix the problem; ... |
85 |
|
86 |
For that, you can look at mentions like "This is done primarily by |
87 |
finding and pointing out issues to maintainers and, where necessary, |
88 |
taking direct action.". But then you once again need to know whether |
89 |
"issues" are meant as "policy violation", or as "breakage", I think we |
90 |
can go on for a while here... |
91 |
|
92 |
We need to remind ourselves that a GLEP is a "Gentoo Linux Enhancement |
93 |
Proposals" and that it is nowhere a picture perfect policy; so, yes, I |
94 |
hope the seriousness (and associated consequences) is clarified. |
95 |
|
96 |
GLEP 48 addresses controversy with "In the event that a developer still |
97 |
insists that a package does not break QA standards, an appeal can be |
98 |
made at the next council meeting. The package should be dealt with per |
99 |
QA's request until such a time that a decision is made by the council.". |
100 |
|
101 |
> > As "QA Standards" under this interpretation are needed to raise |
102 |
> > quality; the council should clarify whether we are able to raise the |
103 |
> > quality level, or rather are restricted to what exists and expect |
104 |
> > the council to raise the quality instead (by having the council do |
105 |
> > that). |
106 |
> |
107 |
> "raise the quality" is very subjective. This would not restrict QA |
108 |
> powers at all, because almost any policy can claim that as its aim. |
109 |
|
110 |
Policy formulation is a power that has a near direct effect on quality. |
111 |
|
112 |
-- |
113 |
With kind regards, |
114 |
|
115 |
Tom Wijsman (TomWij) |
116 |
Gentoo Developer |
117 |
|
118 |
E-mail address : TomWij@g.o |
119 |
GPG Public Key : 6D34E57D |
120 |
GPG Fingerprint : C165 AF18 AB4C 400B C3D2 ABF0 95B2 1FCD 6D34 E57D |