dovecot, init script - is this expected or PR worthy?
dar at darklajid.de
Wed Feb 12 08:08:57 UTC 2014
On Wednesday, February 12, 2014 2:56:18 AM CEST, Stephen R Guglielmo wrote:
> On Tue, Feb 11, 2014 at 12:24 PM, Benjamin Podszun <
> benjamin.podszun at gmail.com> wrote:
>> service dovecot start
>> and .. that succeeds always, if dovecot_enable="YES" is missing (which
>> totally is an error on my part, obviously).
>> What I expect though is the typical "Can't do that, Dave. Try onestart
>> perhaps" response - and some exit code that signals failure.
>> Is that a (tiny, minor) bug? Should I submit a PR for that? Or is that
>> totally irrelevant and people out there would never fall for
>> that anyway..?
> I can't reproduce this on my system. I typically run the rc.d/ scripts
> myself, but I did try using the `service` command as well. It works as
> expected: The script returns no output when dovecot_enable is commented
> out or set equal to NO in my /etc/rc.conf.local file, and it does not
> or stop the dovecot processes.
> I'm using dovecot 1.2.17 on 10.0-RELEASE. Are you using dovecot 2?
I .. think my mail was probably too convoluted and messy. My point?
service foo start shouldn't be silent if the service isn't starting. The
default rc functions ensure that, give the 'Need foo_enable or onestart
instead' message. Dovecot's rc script doesn't, because it doesn't even
_invoke_ the rc functions if dovecot_enable != YES.
(dovecot2 for me, btw - but you described seeing exactly what I see:
Nothing, but the service isn't starting either)
The whole point is that this silently fails. Is that okay?
More information about the freebsd-ports