1 |
On Wed, 13 Sep 2017 09:00:06 +0200 |
2 |
Ulrich Mueller <ulm@g.o> wrote: |
3 |
|
4 |
> >>>>> On Tue, 12 Sep 2017, Matt Turner wrote: |
5 |
> |
6 |
> > I suggested that when security bugs are complete, that if there are |
7 |
> > exp architectures still Cc'd, that security simply reassign to the |
8 |
> > maintainer and let the bug continue as a regular stabilization bug. |
9 |
> |
10 |
> > Unfortunately Aaron says that this is far too much work -- the hassle |
11 |
> > of reassigning a bug and all. |
12 |
> |
13 |
> Let's look at the security team's own policy on that (thanks to K_F |
14 |
> for pointing me to it): |
15 |
> https://wiki.gentoo.org/wiki/Project:Security/GLSA_Coordinator_Guide#Bugs_in_.5Bstable.5D_status |
16 |
> |
17 |
> | All arches (including "unsupported" arches) must be called. But note |
18 |
> | that only "supported" arches (as defined in the policy) are needed |
19 |
> | before the bug can advance to [glsa] status |
20 |
> |
21 |
> Note that it says "unsupported arches", not "unsupported arches with a |
22 |
> stable profile". In fact, the whole guide doesn't mention profiles at |
23 |
> all. |
24 |
> |
25 |
> The alternative scenario would be only to add supported arches to the |
26 |
> security bug. This would mean that the maintainer had to open a second |
27 |
> bug for stabilisation on unsupported arches (which includes not only |
28 |
> arches with experimental profiles, but also stable ones like arm). |
29 |
> Maybe that would take away some hassle from the security team, but it |
30 |
> would certainly mean more work for both maintainers and arch teams. |
31 |
|
32 |
Thanks for spelling the question out! |
33 |
|
34 |
[ CC security@, CC bman@ explicitly ] |
35 |
|
36 |
Aaron, can you clarify on it how you perceive the rules on security side? |
37 |
|
38 |
It's very hard to get a coherent picture from short sentences on IRC, |
39 |
bugs and email. Here is what information I see: |
40 |
|
41 |
[irc/#gentoo-council]: 02:08:42 <+b-man> slyfox: security bugs do not |
42 |
require cc'ing unstable arches or non-security supported arches |
43 |
[bug/630680#c7]: No, it is not longer security supported and is not a |
44 |
stable arch. |
45 |
[mail] : You're right. Fixed. |
46 |
|
47 |
and I can't infer anything at all from it! |
48 |
|
49 |
Does it mean normal STABLEREQ for exp arches should never be reassigned |
50 |
to security bug of because their notion of exp arch is different from arch |
51 |
team's? |
52 |
|
53 |
If it's a documented rule link would help here. My intention to post |
54 |
to -dev@ was specifically to clarify the rules for everyone to decrease |
55 |
hassle and misunderstanding. Not to increase it. |
56 |
|
57 |
https://bugs.gentoo.org/show_bug.cgi?id=630680#c7 is an example of |
58 |
incomplete answer that does not give any more information to me. |
59 |
|
60 |
The comments above imply sparc@ does not care about stable keywords. |
61 |
sparc@ does care about stable keywords but does not want to make it a |
62 |
burden on other teams. |
63 |
|
64 |
Why CC clarity is important here? |
65 |
|
66 |
Understanding the security workflow would help here: |
67 |
|
68 |
Do you never close any security bug that has any arch CCed? |
69 |
(Is there a policy around that?) |
70 |
|
71 |
Do you never proceed with GLSA if there is any arch CCed? |
72 |
(Stable or not) |
73 |
|
74 |
What do you do if there is not only arches in CC but normal people |
75 |
or other projects? Does it impede the process? |
76 |
|
77 |
Thanks! |
78 |
|
79 |
-- |
80 |
|
81 |
Sergei |