1 |
Hi, |
2 |
|
3 |
davecode-Nbr/4NovP0l9pMjJd8zWoA@××××××××××××.org: |
4 |
> > we'd never release ~arch tarballs |
5 |
> I'm talking testing, you're thinking stable... |
6 |
> Testers by definition do not want old stuff. The reason they are |
7 |
> testing is they want new stuff. Yes, they understand the risk. In |
8 |
> Debian they use "unstable" and "experimental" branches. |
9 |
|
10 |
We have to think of people that install the first time or rely on |
11 |
having a stable system, they need arch as it is tested. A fixed and |
12 |
tested status has the pro of having less problems for Newbies. |
13 |
|
14 |
> > I think your idea of how Gentoo releases work is a bit skewed. |
15 |
> > Everything comes from stable. Always. |
16 |
> I'm not clear how I said otherwise? Well, okay, straighten me out ... |
17 |
> My understanding is |
18 |
> * Gentoo has two parallel ongoing branches, arch and ~arch |
19 |
|
20 |
Correct. |
21 |
|
22 |
> * ~arch has more recent packages than arch but less stability |
23 |
|
24 |
Correct. |
25 |
|
26 |
> * both branches keep upgrading over time with bug/security/feature |
27 |
> fixes |
28 |
|
29 |
Correct. This is done by teams per architecture, have a look at |
30 |
<URL:http://devmanual.gentoo.org/keywording/index.html>. |
31 |
|
32 |
> * the only stage3 tarballs that exist are for the previous mega public |
33 |
> release |
34 |
|
35 |
Correct. |
36 |
|
37 |
> * Gentoo releng team plants a pole in the ground and ~arch becomes |
38 |
> arch "beta" |
39 |
|
40 |
No. There is a snapshot taken which contains the packages stable at |
41 |
the moment of shooting. |
42 |
|
43 |
> * arch "beta" quickly turns into arch-stable, while a new ~arch forks |
44 |
> ahead |
45 |
|
46 |
We don't have an immediate switch, the parts of the tree move |
47 |
constantly and at different speeds. |
48 |
|
49 |
> Rolling tarballs for both arch and ~arch together is no more work than |
50 |
> one or the other. It would be the same automated stuff. |
51 |
|
52 |
Problems that occur one day might blow support...so those automated |
53 |
stages need to be unsupported and we would win nothing out of it. |
54 |
|
55 |
V-Li |
56 |
|
57 |
-- |
58 |
Christian Faulhammer, Gentoo Lisp project |
59 |
<URL:http://www.gentoo.org/proj/en/lisp/>, #gentoo-lisp on FreeNode |
60 |
|
61 |
<URL:http://www.faulhammer.org/> |