Re: [Tracker] PATCH: Allow specifying the ontology directory in DConf

[Jonatan Pålsson <jonatan palsson pelagicore com>]

Hi!

On 12 August 2013 11:19, Martyn Russell <martyn lanedo com> wrote:
> On 12/08/13 07:53, Jonatan Pålsson wrote:
> > Hi!
>
>
> Hi :)

> Using a config does limit you to some extent, because each time you run your
> binary, you're using that config and would have to change that config
> otherwise. Whereas, here I can comfortably run separate instances of Tracker
> (at the same time) with different environment variables and have them
> talking to different databases. There's just more freedom and flexibility
> IMO.
This is a very good point. Changing the config is definitely more of a
hassle than setting a variable.

> And as said, I think user config should be for things *users* may want to
> change and unfortunately, I don't consider the location of the DBs something
> a user would be interested in :)

OK. :)

> > > Perhaps if you can explain why you need such radical changes to the
> > > locations outside of are standardised areas, that could help us
> > > understand
> > > the situation a little better :)
> > The reason for modifying the ontologies path is basically that I have
> > different sets of ontologies (the original Tracker ones, and a more
> > light-weight set of ontologies). I would like a non-root user to be
> > able to modify her ontologies, and configure Tracker to use these.
>
>
> So are you using system-wide instance of Tracker and expecting a non-root
> user to use a different Tracker set up if they prefer?
>
> Interesting :) can you elaborate?

Yep. Well, it's not anything special. The way I see it, the Tracker
binaries are capable of using other ontologies than the ones currently
shipped with Tracker, so these should be configurable. Since the
database is stored in each user's $HOME, I believe the ontologies
should also be configurable, so that the database schemas can be
modified. Again, this is something the average user probably doesn't
want to change.

In the same way, it would be interesting to allow users to specify
their own extractor modules. This could be likened to the way a web
browser software works, where the core of the program is often
installed in the system, but plugins can be added by users in order to
modify the browser behavior.

The obvious remark here is that the user can simply install her own
Tracker, and keep everything local, and this would work out nicely,
but if Tracker is installed in the system, an administrator can ensure
the core software is kept up to date while users are responsible for
ontologies.

In the end, it is a matter of taste, and I accept if this is not a
desired path for Tracker! :)

> --
> Regards,
> Martyn
>
> Founder & Director @ Lanedo GmbH.
> http://www.linkedin.com/in/martynrussell



-- 
Regards,
Jonatan Pålsson

Pelagicore AB
Ekelundsgatan 4, 6th floor, SE-411 18 Gothenburg, Sweden