1 |
On Thu, 2007-05-17 at 20:33 +0200, Rémi Cardona wrote: |
2 |
> Chris Gianelloni wrote: |
3 |
> > It's simple. You mask expat-2.0.0 on all the current profiles, we mark |
4 |
> > it stable in the snapshot and don't have it masked in the 2007.1 |
5 |
> > profile. When we release (actually right before), we mark the package |
6 |
> > stable in the tree. We document the expat upgrade as part of the |
7 |
> > profile upgrade guide, and we're done. Users using a <=2007.0 profile |
8 |
> > never see the upgrade. New users use the new expat. Users changing to |
9 |
> > the 2007.1 profile run revdep-rebuild. |
10 |
> |
11 |
> +1 |
12 |
> |
13 |
> Now, how can we do this? Could we start changing the profiles right now? |
14 |
|
15 |
Considering we already have a 2.0.0 ebuild, we do the following: |
16 |
|
17 |
- Mask >=2.0.0-r9 (this allows for security bumps, if necessary, number |
18 |
can be adjusted) |
19 |
- Copy 2.0.0 to 2.0.0-r9 |
20 |
- When we make a new 2007.1 profile, don't mask >=2.0.0-r9 |
21 |
- Stable 2.0.0-r9 in the 2007.1 snapshot and mark it stable in the tree |
22 |
with the release |
23 |
- ??? |
24 |
- Profit! |
25 |
|
26 |
> (I guess people on ~arch will need to unmask it to not downgrade). |
27 |
|
28 |
Well, with what I have said, there's room for version bumps, if |
29 |
required. It also means ~arch people don't have to do anything. There |
30 |
won't be any downgrade and we simply never mark anything below 2.0.0-r9 |
31 |
stable to keep stable users safe. |
32 |
|
33 |
> Should this be brought to the next council meeting? |
34 |
|
35 |
Is that really necessary? What can the Council do that we cannot agree |
36 |
upon here as civil adults? I think we can agree to do this ourselves. |
37 |
I can definitely agree to it from a Release Engineering standpoint. It |
38 |
would work quite well and is beneficial to our users. |
39 |
|
40 |
> Chris, I could write a small paragraph for whatever GWN explaining what |
41 |
> stable and unstable users will have to do if you want. |
42 |
|
43 |
Sure. However, if we did follow my draft plan above, there would be no |
44 |
need. Users running ~arch have probably hit this already by now, so I |
45 |
don't think we would be informing too many people. That being said, it |
46 |
would make a cool article. Even if just to show that, yes, we really do |
47 |
care for our users and think about ways to reduce the impact on their |
48 |
systems. |
49 |
|
50 |
-- |
51 |
Chris Gianelloni |
52 |
Release Engineering Strategic Lead |
53 |
Alpha/AMD64/x86 Architecture Teams |
54 |
Games Developer/Council Member/Foundation Trustee |
55 |
Gentoo Foundation |