[Debtags-devel] Ideas that came out of an Aglianico bottle last night

[Enrico Zini]

--LQksG6bCIzRHxTLp
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Thu, Dec 02, 2004 at 02:25:09AM +0100, Erich Schubert wrote:

> Time draws closer for when i depart to berkeley to work on Debtags.

YAY!
I realize I still haven't wrote you how happy I am with that, but I've
been thinking "that's so cool!" constantly since you first announced it!


> Enrico, maybe we should try to get back to the old concept of
> providing "exhaustive, but not too much choice". ;-)
> I wouldn't try to limit the facets shown at the beginning. One time or
> the other you want to start differently.

I would play with algorithms for a while.  One thought is to have some
fixed, and a few ones computed to cover most of those the packages the
fixed ones don't cover.  But Custom Debians at least, will *want* to
have some fixed facet set at the beginning of a search.  Users may want
to select a favourite facet to be in that fixed set, too.

Think of it more like an interface issue rather than a structural issue.


> For the web frontend, my current idea is to show alternating a list of
> facets together with a count of how many packages are tagged and maybe
> the selectivity; then only show contents of that facet to the user,
> maybe sorted by selectivity again.

What do you mean by 'selectivity'?

=20
> OTOH, I'm not yet 100% sure about the tagging thing. IMHO, recent work
> on tags has shown more complex demands.
> For example file type handlings. I love the idea of having tags for
> file format support. That can show so damn useful.
> But sooner or later we are going to need to differentiate between "can
> play mp3" and "can create mp3". using the role tags won't do: for
> example a program may be able to playback mp3 and ogg-vorbis, but only
> create ogg-vorbis.
> So I fear we are going to need some kind of predicates, which means
> making progress towards "reasoning".
> Tags would then look like action::playback(file-format::mp3
> file-format::vorbis), action::create(file-format::vorbis)

Uhm... yes, that would be useful, and sounds very much like RDF can do
it easily.  In fact, the whole tags in the 'use::*' facet can be thought
as these kinds of actions.

I'm not sure how a good interface could be realized to use such a broad
AND specific body of knowledge in a simple way; maybe the one in
segusoland/logic-desktop, but then we would not want to tag packages,
but rather the binaries inside.


> But i'm a bit biased here, since there is a lot of
> reasoning-on-the-web work being done at my institute (and in the
> rewerse project) and i'm considering this area for my diploma thesis.
> So what do you think?

I'm (personally) for thinking about it when we have something based on
the current technology that gets spread and widely used.  But if you
intend to try it out, I strongly support working on a database format
that can store the information you need, so that noone is hindered when
doing research.


Ciao,

Enrico

--
GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini <enrico@enricozini.org>

--LQksG6bCIzRHxTLp
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: Digital signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)

iD8DBQFBrvxq9LSwzHl+v6sRAnUaAJ9YygPDR8bHLDQddmyptE8msBciKACfcw9P
hiip28y9iIX3AlLNh9YPpT8=
=DWZ3
-----END PGP SIGNATURE-----

--LQksG6bCIzRHxTLp--