Bug#522163: standard for disabling daemons in /etc/default

[Ansgar Burchardt]

I agree that we should *not* encourage using some `ENABLE={yes,no}` in
/etc/default/* to disable a service.  This should be done via the
regular tools (update-rc.d; systemctl disable; ...).

So +1 for closing this bug as wontfix.

Ansgar

Bug#522163: standard for disabling daemons in /etc/default

[Felipe Sateler]

On Tue, 6 Dec 2016 11:20:57 +0100 Andreas Henriksson  wrote:
> Hello!
>
> On Thu, Oct 04, 2012 at 11:11:20AM +0100, Simon McVittie wrote:
> [...]
> > I don't think packages should do this, except as a way to respect
> > sysadmin configuration done in previous versions. openarena-server used
> > to do this (and be off-by-default), but for wheezy I changed the
> > documented way to disable it:
> [...standard update-rc.d usage snipped...]
>
> I would like to second Simon McVittie.
>
> I propose closing this bug as wontfix.

Yes, please lets not document this anti pattern.

> Alternatively, I'd suggest turning this bug around completely and
> discuss policy discuraging (or even forbidding) reinventing the
> update-rc.d policy layer via /etc/default.
>
> (TTBOMK this anti-pattern surfaced because of noone volunteering to fix
> dh_installinit to avoid starting/enabling on install, but that has since
> long been fixed now.)

Do we have somewhere else to document that ENABLE= is an antipattern?
I'm not sure this belongs in policy. After all, there's an infinite
number of ways to get things wrong ;)


Saludos

Bug#522163: standard for disabling daemons in /etc/default

[Andreas Henriksson]

Hello!

On Thu, Oct 04, 2012 at 11:11:20AM +0100, Simon McVittie wrote:
[...]
> I don't think packages should do this, except as a way to respect
> sysadmin configuration done in previous versions. openarena-server used
> to do this (and be off-by-default), but for wheezy I changed the
> documented way to disable it:
[...standard update-rc.d usage snipped...]

I would like to second Simon McVittie.

I propose closing this bug as wontfix.

Alternatively, I'd suggest turning this bug around completely and
discuss policy discuraging (or even forbidding) reinventing the
update-rc.d policy layer via /etc/default.

(TTBOMK this anti-pattern surfaced because of noone volunteering to fix
dh_installinit to avoid starting/enabling on install, but that has since
long been fixed now.)

Regards,
Andreas Henriksson