1 |
On Thu, Aug 29, 2013 at 12:06 PM, Andreas K. Huettel |
2 |
<dilfridge@g.o> wrote: |
3 |
> |
4 |
> Sounds reasonable, but then you get just abused by Mr_Bones for breaking the |
5 |
> deptree. |
6 |
|
7 |
The solution to that is to either ignore the abuse, or drop the stable |
8 |
keywords on all reverse deps as well. |
9 |
|
10 |
Dropping individual package keywords shouldn't happen much, since the |
11 |
fact that it is happening is a good sign that the whole arch should be |
12 |
dropped. If an arch team can't keep up with a single package they |
13 |
should remove the stable keywords themselves so that this can be done |
14 |
in an orderly manner. |
15 |
|
16 |
> |
17 |
> Anyway, how about when the arch has not even responded to the (much older) |
18 |
> KEYWORDREQ for that package? :| |
19 |
|
20 |
I see this as less of an issue - maybe the maintainer just drops |
21 |
themselves from the bug to be added-back by the arch team only if |
22 |
needed. If the maintainer logged it they can also cancel it if they |
23 |
don't care to see it open. |
24 |
|
25 |
The only people affected by a pending KEYWORDREQ are the users of that |
26 |
arch. They're at the mercy of the arch team no matter what - if the |
27 |
package matters that much to them they should step up to arch test. |
28 |
|
29 |
Please don't interpret anything I'm saying as a lack of care for small |
30 |
archs. I'd just rather see them define their goals in terms of things |
31 |
they can actually accomplish and then accomplish those. Having a |
32 |
bunch of ancient stable versions that are just weighing down the rest |
33 |
of the distro isn't really a good measure of success. That was why I |
34 |
suggested the possible @system-only compromise - it might be a way to |
35 |
capture some of the value of having stable keywords while greatly |
36 |
reducing the amount of work involved. Better to do less but do it |
37 |
properly. |
38 |
|
39 |
That said, I'd also encourage maintainers to leave around old versions |
40 |
as long as they aren't actually maintenance burdens. If they become |
41 |
such, well, time to prune if the STABLEREQ is stale. |
42 |
|
43 |
Rich |