[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#681544: apache2-dev: provide dh_apache2 option to avoid enabling module by default



Arno Töll <arno@debian.org> writes:
> On 14.07.2012 05:06, Russ Allbery wrote:

>> A nicer mechanism would be to allow the *.apache2 configuration file
>> have an option for mod *.load lines saying not to enable the module by
>> default.

> I think, we should rather stick with a specialized dh_apache command
> line argument, say dh_apache2 --no-start-by-default or whatever instead
> of adding even more complexity to .load files. Does that sound
> acceptable to you?

Sure, yes, that would be fine.  I'm not sure what I was thinking of when
it came to the *.load configuration.  There's no need to change the *.load
files, of course; all this would affect is whether the module was enabled
by default.

> Moreover, we planned to let the maintainer give a local policy on that
> regard. I could imagine a variable in /etc/default/apache2 determining
> the web server reload behavior. In fact, that's quite the point to
> abstract the module load behavior through our wrapper script, although
> such policies are not implemented yet.

I think this is something different, no?  This isn't about whether to
reload Apache, but rather about whether to enable a module by default.
Currently, if one enables mod_webauth by default, it breaks Apache,
because there is mandatory site-local configuration with no meaningful
defaults.

(I consider this a minor upstream bug, and a later version will just warn
and do nothing if minimum configuration is not available.)

>> If the module is enabled, then Apache should still be restarted on
>> upgrade (or other configuration actions).  If the module is not enabled
>> or wasn't previously installed, nothing should be done by default in
>> postinst.  The postrm handling would remain the same.

> We do so, don't we? At least we wrote code which should do right that,
> but it might have bugs of course :>

The restart issue appears to be something different, so let me open a
separate bug about that to keep distinct issues distinct.  :)

-- 
Russ Allbery (rra@debian.org)               <http://www.eyrie.org/~eagle/>


Reply to: