Re: [gentoo-dev] Proposal: sys-pam category

[Jan Jitse Venselaar <janjitse@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2377 bytes --]

On Sunday 05 June 2005 23:55, Ned Ludd wrote:
> On Sun, 2005-06-05 at 16:57 -0400, Nathan L. Adams wrote:
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> >
> >
> > Well obviously there needs to be a consensus on *how* to logically
> > organize things before anyone goes willy nilly changing stuff. Do you
> > group by what the package is used for (email vs. game vs. web browser)
> > or by what it is built from (PERL stuff, Gnome apps, KDE apps). It
> > appears that currently its a mix. Is that documented anywhere?
>
> You raise a good point and sadly that is the unfortunate thing here..
> There is no clear consensus right now and we have yet to really have a
> fruitful thread on the subject. I not aware of any intelligent
> documentation on this subject either.
>
> > I personally think the organization should be from an end-user
> > perspective as much as possible. Imagine for a moment that you are a
> > Genewbie (new Gentoo user). You have a new minimal installation and you
> > want to add some applications. How do you know what your choices are for
> > an email client, for instance? You could find most things here:
> >
> >
> > Again, I think better
> > organization and improved tools are both worth while.
>
> I fully agree with you on improved tools and would rather see us go
> this route before we end up with >300 top level categories.
>
Recently I spoke with Enrico Zini (Debian Developer). Debian has the same 
problems in this area, and he is working on Debtags, which is designed to 
solve this exact problem. Basically what it does is take a few different 
classifications (like, purpose, environment, language it was written in 
etc.), and make it configurable what to put in the top level, what to put 
underneath the top level, etc. This is called faceted classification, and was 
invented by the Indian librarian and classificationist S.R. Ranganathan in 
the early 1930s.
The system is expandable, users can add their own classifications and 
categories. 
The main problem is performance, but I understood that that is mainly due to 
lack of focus on that area. 
For more information see the Debtags website, 
http://debtags.alioth.debian.org/
It may not be the ideal solution yet, but what I understood of it, it is a 
very powerful system, and quite intuitive for end-users.

Jan Jitse

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]