[Xesam] packaging problem

[Ivan Frade]

Hi

 Sorry for this very basic questions but...

what do we need to validate? We have a small-but-useful "validator" in
tracker written in C to verify that all the referenced classes and
properties are defined somewhere, and few easy things like that. (It helps a
lot when adding new stuff in the ontologies). Do we need anything more
complicated? We can reuse that script or rewrite it in python.

why do we need the description of the ontology in the validation? if the
description is in a second file (same name with .description suffix), we
only need to use it for the HTML generation.

 I started to write the HTML generation for the tracker ontologies and i did
the first step (1 ontology -> 1 HTML). My idea (and i guess the nepomuk
scripts are not so different) was: two files (a.ontology, a.description),
and an extra "global" file that contains: all namespaces + publication URL
(i.e. "nie" http://www.semanticdesktop.org/ontologies/nie/
http://www.semanticdesktop.org/ontologies/nie/) . The publication URL is
needed in case you want to generate your doc for an internal server, or the
URL of a concept doesn't match with a real URL.

 I dont see why to use trig or merge description and ontology in one file.
But maybe i am missing something.

 Regards (and thanks for the effort!),

Ivan



On Tue, Jun 30, 2009 at 10:36 PM, Sebastian Trüg <strueg at mandriva.com>wrote:

> On Wednesday 01 July 2009 21:25:46 Evgeny Egorochkin wrote:
> > On Tuesday 30 June 2009 19:07:59 Sebastian Trüg wrote:
> > > Hi guys,
> > >
> > > I am having a little trouble with setting it all up.
> > > Leo pointed out that using trig directly might give us trouble when we
> > > want to verify since there are no tools.
> > >
> > > That means we have 2 possibilities:
> > >
> > > 1.
> > > write our own tools for verification and so on (it would probably be
> > > sufficient to extract the data graph and the metadata graph from each
> > > trig file and then validate them separately.
> > >
> > > 2.
> > > use 2 turtle files and merge them at build time (I see no point in
> having
> > > both the turtle and the trig in svn. It only leads to trouble with
> people
> > > commiting changes to the trig by accident or what not. And it is not
> > > clean). But that means that we need a script that can merge the files
> and
> > > output trig. I don't think we can use the existing tools written in
> java
> > > for that as it would add a compile-time java-dependency. Thus, we need
> a
> > > script in python or perl or something. I looked at python rdf libs,
> there
> > > is no trig serializer support.
> >
> > AFAIK redland/raptor has ruby and python bindings. Also you can easily
> > generate docs using either of these.
>
> sure, I know that. I was rather looking for a decision between solutions 1
> and
> 2.
>
> > > So what should we do? Store trig or 2 turtle files and merge on
> > > build-time?
> > >
> > > BTW: I also think the docs should be merged at build-time. That also
> > > means that the java tools need to be ported to a scripting language. I
> > > could use help with that.
> > >
> > > Cheers,
> > > Sebastian
> > >
> > > _______________________________________________
> > > Xesam mailing list
> > > Xesam at lists.freedesktop.org
> > > http://lists.freedesktop.org/mailman/listinfo/xesam
> >
> > _______________________________________________
> > Xesam mailing list
> > Xesam at lists.freedesktop.org
> > http://lists.freedesktop.org/mailman/listinfo/xesam
>
> _______________________________________________
> Xesam mailing list
> Xesam at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/xesam
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freedesktop.org/archives/xesam/attachments/20090701/6a59d1d9/attachment.html