1 |
Patrick Lauer posted on Thu, 05 May 2016 09:32:01 +0200 as excerpted: |
2 |
|
3 |
> So you're saying that a Gentoo-specific change in Gentoo happens because |
4 |
> the Gentoo maintainer doesn't care about Gentoo? ;) |
5 |
|
6 |
I'm saying that big-picture, there's more than one distro, and once a |
7 |
particular package graduates beyond a single distro, as openrc has, |
8 |
there's likely to be some more or less disruptive changes. |
9 |
|
10 |
Meanwhile, seems there was another package with a runscript executable |
11 |
that debian happens to package, proving my point about namespace |
12 |
collision. Today that's a problem for debian; tomorrow it could well be |
13 |
a problem for a would-be gentoo packager (dev or user) of that same |
14 |
package... if some gentooer out there isn't /already/ having to deal with |
15 |
the problem. So it's not just debian openrc is helping here, it's the |
16 |
entire floss community that may at some point be interested in software |
17 |
with that same name, including gentoo. |
18 |
|
19 |
> Somehow I still don't see a *problem* being solved, and the runscript |
20 |
> binary/symlink pretty much has to stay there indefinitely unless you |
21 |
> want to make life exciting for people that have their own or adapted |
22 |
> init scripts. |
23 |
|
24 |
Many gentoo precedents define your "indefinitely" as "one year at |
25 |
minimum". Beyond that, it's (gentoo) maintainer's preference, taking |
26 |
account of how much of the rest of the tree it still effects, getting |
27 |
someone to take care of the lagging packages if necessary, etc. But |
28 |
while we don't /try/ to break stuff out of the tree, if it's out of tree |
29 |
and particularly if it's in some user's non-public/non-layman overlay or |
30 |
simply a script hacked up on their system, we prioritize accordingly, and |
31 |
yes, we recognize that sometimes that stuff breaks with such changes, but |
32 |
that's held to be a case of "if it breaks, you get to keep the pieces". |
33 |
|
34 |
Despite all that, I expect it'll be more like two years' worth as |
35 |
deprecated but still there, simply for practicality reasons. Meanwhile, |
36 |
once the deprecation warning goes in, a year or more is plenty of time to |
37 |
change things so they'll still work after the deprecation period, and |
38 |
like I said, while we recognize that some users may not upgrade in in |
39 |
over a year, that really has been held to be their problem and |
40 |
responsibility at that point, including if they entirely missed the |
41 |
deprecation warnings as a result of not upgrading in over a year. |
42 |
|
43 |
-- |
44 |
Duncan - List replies preferred. No HTML msgs. |
45 |
"Every nonfree program has a lord, a master -- |
46 |
and if you use the program, he is your master." Richard Stallman |