1 |
Eric Sammer wrote: |
2 |
> The "stable" or "frozen" tree effort was not exactly abandoned so much |
3 |
> as placed on a back burner to under go further research. There were some |
4 |
> concerns from some of the devs as to how much additional work would be |
5 |
> placed on them and their archs (some of the non-x86 archs have few |
6 |
> developers and certainly don't need extra work piled on). Overall, there |
7 |
> were a number of issues that needed to be worked out prior to any kind |
8 |
> of attempt at starting such a project (officially). |
9 |
> |
10 |
> The original GLEP by Kurt Lieber remains up for review at |
11 |
> http://www.gentoo.org/proj/en/glep/glep-0019.html for those who are |
12 |
> interested. |
13 |
|
14 |
I came up with an idea while driving some time last week, so naturally I |
15 |
forgot to write it down next time I pulled over. :P |
16 |
|
17 |
What if, instead of a stable tree, we introduce a new KEYWORDS modifier? |
18 |
Currently we have -arch, ~arch, and plain 'ol arch. What if we added |
19 |
"+arch" to mean "Tested stable for production deployment." and impose a |
20 |
set of guidelines regarding how long such an ebuild should remain |
21 |
available / in the tree. Server systems could continue using the "main" |
22 |
tree and set ACCEPT_KEYWORDS to +arch to only receive the most tested |
23 |
stable applications. |
24 |
|
25 |
No tree duplication, multiple CVS ./ rsync respositories, and such a |
26 |
small change I'd bet it could be implemeneted in minutes / hours rather |
27 |
than days / weeks. |
28 |
|
29 |
-- |
30 |
Stewart Honsberger |
31 |
Gentoo Developer |
32 |
http://www.snerk.org/ |