1 |
On Wed, Jun 8, 2011 at 7:17 AM, Duncan <1i5t5.duncan@×××.net> wrote: |
2 |
> He didn't say he was going to defy council, rather, that he simply |
3 |
> wouldn't be removing ebuilds /at/ /all/ until either the changelog is auto- |
4 |
> generated (making the case moot) or the council changes policy. |
5 |
> |
6 |
> That means they'll either fall to someone else to do, or will simply |
7 |
> remain there, but either way, it's quite different from directly defying |
8 |
> the council decision. |
9 |
> |
10 |
|
11 |
As long as all versions in the tree compile cleanly and are free from |
12 |
security issues, I don't see any issue with keeping older ebuilds |
13 |
around. If anything I think that some packages are too quick to |
14 |
remove ~arch versions. I run stable but accept the odd ~arch package. |
15 |
When I do accept a ~arch package I only accept one version of it with |
16 |
the goal of going stable once whatever drove me to accept ~arch gets |
17 |
there. When the ~arch package disappears I just have to re-evaluate |
18 |
my new options and try again, and sometimes it feels like I never end |
19 |
up in stable. (I do realize that a few types of packages will |
20 |
probably never be stable by their nature, and that is fine.) |
21 |
|
22 |
If old versions become QA issues then we already have processes to |
23 |
deal with that. It is the duty of maintainers to deal with such |
24 |
problems. |
25 |
|
26 |
In any case, the rule is simple - if you remove an ebuild you have to |
27 |
include a note in the Changelog. That could change, or it might not, |
28 |
or perhaps it will become automated, but either way it is the rule |
29 |
right now. |
30 |
|
31 |
One thing I will say is that I appreciate the civility in this thread |
32 |
so far. I think everybody on both sides of the issue realizes that |
33 |
this is contentious, and I think everybody would be open to a better |
34 |
solution. |
35 |
|
36 |
Rich |