[Kolab-devel] Defaults files: kolabd and kolab-server
paul at boddie.org.uk
Tue Nov 12 19:18:29 CET 2013
On Tuesday 12. November 2013 18.43.23 Jeroen van Meeuwen (Kolab Systems)
> On 2013-11-12 17:05, Paul Klos wrote:
> > Jeroen van Meeuwen (Kolab Systems) schreef op 2013-11-12 16:46:
> >> It's a thing that is a difference between what is allowed in RPM
> >> packaging (ship kolab-server for the kolabd service), and what is not
> >> allowed in Debian (the package needs to be called after the service).
> > I think it's the other way around: the kolab-server script actually
> > provides a daemon called kolabd, using the "Provides" tag inside the
> > init script, IIRC.
Yes, I'm not sure that Debian mandates identical naming of the service and the
daemon, or at least I couldn't find any policy about that.
> >> Subsequently, I suppose, the init script names (but not the
> >> /etc/defaults/ file?) was changed to match the kolab-server package
> >> name.
> > You can name your defaults file pretty much anything you like, because
> > if you want to use one, you have to source it inside the init script
> > yourself.
> > It's just a matter of convention to give these related files the same
> > name, and the confusion stems from not adhering to the convention.
Right. In this case, the messages from the init script refer to files that
will not exist when people are told about them.
> > Anyway, It's probably a good idea to rename the files and stick to the
> > convention. We'll just have to take care of deleting any leftovers, so
> > this does imply some work in terms of maintainer scripts, and also
> > package relationships. I'd have to look this up, but I believe that if
> > kolabd is going to replace kolab-server, it should declare that it
> > breaks, or conflicts with kolab-server, to ensure a smooth transition.
> I think that's what I just said, but perhaps not.
We can easily get very confused here. ;-)
What I intend to do is to commit a change that separates the service name from
the daemon name, and in fact this is already done by the wallace init script
(debian/wallace.init). So, by making both init scripts work in the same way,
both having services with one name and daemons with another, there should be
less confusion in the future. Everything will remain the same apart from the
messages in the kolab-server init script.
As I said before, this is trivial stuff but it lets me go through the workflow
that Jeroen and Timotheus have described before considering more challenging
More information about the devel