1 |
On Wed, Aug 22, 2018 at 8:02 AM Alarig Le Lay <alarig@××××××××××.fr> wrote: |
2 |
> |
3 |
> I’m a little curious about the way a package is considered as stable or |
4 |
> ~arch. |
5 |
> |
6 |
|
7 |
Packages always start out in ~arch and sometimes become stable. A |
8 |
package version CAN be made stable if: |
9 |
1. It has been in ~arch for 30 days (exceptions made for security fixes) |
10 |
2. It has no major problems |
11 |
3. It works when built/run against stable dependencies. |
12 |
|
13 |
Now, not every package that CAN be made stable actually gets marked |
14 |
stable. Half of this is the same reason that lots of desirable things |
15 |
don't happen - people don't get around to it. The other half are |
16 |
situations where the maintainer doesn't think that it makes sense to |
17 |
stabilize a package, usually for reasons you'd probably agree with. |
18 |
|
19 |
If a package already is stable, then at one point in time it probably |
20 |
worked fine. It would only lose the stable keyword if it had a fairly |
21 |
serious problem and it wasn't likely to get solved. I couldn't really |
22 |
speak to the current state of libreoffice-bin, but for most of its |
23 |
history the binary openoffice packages have been problematic, but of |
24 |
course popular. In some sense stable is a relative term - it may be |
25 |
desirable to offer both a stable and testing version of openoffice-bin |
26 |
so that users who want to use it don't have to run bleeding-edge, even |
27 |
if neither is as stable as the from-source version. |
28 |
|
29 |
Also, a lot of bugs are somewhat situational. Something that you |
30 |
consider critical might not be serious to somebody else. If the |
31 |
stable version works as well as the versions marked as ~arch then |
32 |
there is little benefit to dropping the stable keyword, since users |
33 |
STILL will have to deal with the issue, and now they might have to |
34 |
deal with other issues as well. |
35 |
|
36 |
I guess to sum up you could say that the stable version of a package |
37 |
has received more testing than an unstable version OF THE SAME |
38 |
PACKAGE. There are no promises that a stable version of one package |
39 |
is comparable to a stable version of a different package. |
40 |
|
41 |
Finally, I'll note that if you ask 10 Gentoo users/devs what they |
42 |
think stable ought to be, you'll probably get at least half a dozen |
43 |
answers, so the above is meant more as a description of the status quo |
44 |
than anything else. |
45 |
|
46 |
-- |
47 |
Rich |