| 1 |
On Tue, Mar 31, 2015 at 5:21 PM, Alec Warner <antarus@g.o> wrote: |
| 2 |
> |
| 3 |
> |
| 4 |
> On Tue, Mar 31, 2015 at 11:14 AM, Mike Gilbert <floppym@g.o> wrote: |
| 5 |
>> |
| 6 |
>> Hi all, |
| 7 |
>> |
| 8 |
>> I have been bumping heads with Mike Frysinger (vapier) on the topic of |
| 9 |
>> drop-in config files that are utilized by quite a few system services |
| 10 |
>> on Gentoo. For reference, see bug 544150. |
| 11 |
> |
| 12 |
> |
| 13 |
> I am going to the movies with Mike tomorrow, I will be sure to cuddle him on |
| 14 |
> your behalf. |
| 15 |
|
| 16 |
Thanks? ^_^ |
| 17 |
|
| 18 |
>> |
| 19 |
>> |
| 20 |
>> Mike claims that Gentoo has a policy of "not enabling anything by |
| 21 |
>> default", and that this policy applies to both init scripts, and |
| 22 |
>> drop-in configuration files. |
| 23 |
> |
| 24 |
> |
| 25 |
> I would say the policy for *services* is that non-critical services are not |
| 26 |
> enabled by default. I would argue that is a policy decision that is distro |
| 27 |
> wide. |
| 28 |
> Maintainers are of course, at liberty to determine if their service is |
| 29 |
> 'critical' or not. |
| 30 |
|
| 31 |
Right, I agree that this makes sense for services. |
| 32 |
|
| 33 |
But I don't really think the configuration fragments I am referring to |
| 34 |
could really be called "services". However, they do affect the |
| 35 |
operation of services. |
| 36 |
|
| 37 |
Should packages be allow to set/alter the configuration of a system |
| 38 |
service automatically? I would say yes, and it is up to the maintainer |
| 39 |
to decide what is reasonable here. |
| 40 |
|
| 41 |
>> My questions to the community: |
| 42 |
>> |
| 43 |
>> - Do we have a policy regarding enablement of drop-in config files? |
| 44 |
> |
| 45 |
> |
| 46 |
> Maintainers discretion. |
| 47 |
> |
| 48 |
>> |
| 49 |
>> - If so, what is it? Where is it documented? |
| 50 |
> |
| 51 |
> |
| 52 |
> My brain; seriously though, generally undocumented things imply maintainers |
| 53 |
> discretion. |
| 54 |
|
| 55 |
We either have a policy that the maintainer is supposed to follow |
| 56 |
(barring some reasonable exception), or we don't have a policy and the |
| 57 |
maintainer can do what they want. |
| 58 |
|
| 59 |
In the referenced bug, I'm being told that an existing policy applies |
| 60 |
here and that a bunch of existing packages violate this policy; I'm |
| 61 |
trying to verify if that is the case, and if so, what is the policy, |
| 62 |
and how is it applicable? |
Archives