> > Sylvain Hellegouarch wrote: >> Brian Smith a écrit : >> > Since this is a simple extension element (by syntax), it must be >> > metadata about the entry (by definition). Any Atom+RDF tool >> > should be able to create RDF triples out of that without issues. > >> Out of what? You have defined an element with a different >> namespace but how does it convey any of the information an >> atom:category conveys? I'd be happy to try to give an example >> in comparison to yours but I don't understand how you provide >> the same amount of information here. > > What information does atom:category convey? What is a category, in Atom, > really? It is whatever it is defined to be. Simple extension elements are > the same way. What is the difference in meaning between these two > elements?: > > <atom:category scheme='http://scheme.example.org/' term='fact'/> > <fact xmlns='http://scheme.example.org/'/> > > You can't tell what either of them means without knowing something about > "http://scheme.example.org" and its "fact" component. They might even mean > exactly the same thing. > >> Maybe I'm misunderstanding your entire point, in that case, my apology. > > My point is that atom:category is underspecified, so it is better to just > let end-users do whatever they want with it, and use something else for > things that are supposed to have precise, computer-actionable semantics. >
But you can't unless, as you suggested, you know the meaning of said namespace. So in any case, consumers that automate handling of your entry will need to be programmed in order to either understand that namespace or leave it alone. I can appreciate your point of not stuffing everything into an element that can mean anything and probably that could hurt the end users who consumes entries via a regular feed reader. That's fair in that particular context. But then again, without any further specification, you jump from underspecified to overly restrictive behavior of the atom:category element. I don't really find that attractive as a generalized best practice. You may have services that will not be for public usage and I find using atom:category in that context entirely natural and safe. In regards to using RDF, now that I understand your point I can see why it didn't answer your initial statement. - Sylvain -- Sylvain Hellegouarch http://www.defuze.org
