[Xesam] An open-source project for desktop ontology maintenance

[Philip Van Hoof]

On Tue, 2009-05-05 at 18:49 +0300, Ivan Frade wrote:
> Hi
> 
>  Why not...
> 
> * XESAM open source project that maintains three pieces
>   - An ontology (based on nepomuk [keeping names etc.], community
      driven).
>   - An API for client applications (give it a proper name)
>   - A query language (very probably sparQL)
> 
>  This makes happy the people in KDE, GNOME, freedesktop and open
> source fans. It is simple and understandable. Am i missing anything?

No you are not. I think some other people might be missing things :)

Hasn't that always been what Xesam 2.0 is (going to be) about?


Now apparently there's a request to have a more community driven project
structure for Xesam 2.0. Which means that for the ontology we need:

- A git repository with some accounts for some people

- A format (I'm assuming .trig or .ttl is perfect for this)

- Scripts for generating the website out of the git repository

- A bug tracker so that we can irritate each other with bugs. Preferably
  a bug tracker that is actually good (at enabling us to irritate each
  other). Not just a @#*& trac website.

- A procedure that we agree with Nepomuk upstream to get our Xesam 2.0
  ontology requirements into upstream Nepomuk. If Nepomuk upstream wants
  this, of course. Else we just fork it. But apparently nobody at
  Nepomuk is against letting Xesam 2.0 use Nepomuk, or against letting
  Xesam 2.0's people help define improvements to upstream Nepomuk?

  That's great. I mean, then let's just make some agreements on how we
  communicate change requests to Nepomuk upstream?!

  What is certain is that several projects, like Tracker, need a very
  fast response time for their ontology changes. Which means < than a
  week. I assume upstream Nepomuk wants to take a more restricted
  approach and wants us to prove that our experiments work before
  accepting said ontology changes?

I personally hope that some time in future we'll just have a package
called "nepomuk-ontology-VERSION.tar.gz" instead of the directory that
we are shipping as part of the Tracker package [1]. Note that we have a
few requirements like having a load-order in the filename (much like
rc.d directories work). But nothing that we can't easily agree on.


Which means that for the DBus API we need:

- Unit tests

- The introspection XML in above mentioned git repo

- Documentation

And for the query language I don't think we need anything specific in
the repository. Except maybe the SPARQL and SPARQL UPDATE specifications
as part of the documentation.


[1] http://git.gnome.org/cgit/tracker/tree/data/ontologies


Looking at the list above, for example freedesktop.org provides all what
we need. We just need to ask. Right?


-- 
Philip Van Hoof, freelance software developer
home: me at pvanhoof dot be 
gnome: pvanhoof at gnome dot org 
http://pvanhoof.be/blog
http://codeminded.be