[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: RDF redux




Sam Ruby wrote:



I see some discussion going on in


http://www.intertwingly.net/wiki/pie/RSSVote

I think such exploration is healthy, but personally would prefer that a proposal more fully flushed out on the mailing list first.

This "RSSVote" suggests "abandons the current Atom syntax" yet "Some Atom syntax elements not already present in RSS 1.0 would be added to RSS 1.0 as a module extension.". Which ones?

I'll offer the following as a guide for an initial checklist of things to explore: http://www.intertwingly.net/slides/2003/xmlconf/ . That doesn't mean that any proposals need to come up with the same solutions, or even accept all the same requirements, but it does provide a basis for comparision.

Sam's right - before we can really consider this proposal, we need some more information about what's actually being suggested.

The good bit of RSS 2.0 is a simple syntax, the good bit of RSS 1.0 is a powerful model. The worst bits of RSS 2.0 are spec ambiguity and insistence on backwards-compatibility (times have changed). The worst bit of RSS 1.0 is the syntax. If Atom is to improve on legacy RSS, shouldn't it just be getting the good bits?

RSS 1.0 syntax is a subset of RDF/XML with some structural constraints. A key advantage of using RDF/XML is that the mechanism for using terms from other namespaces is well-defined. In the context of Atom, this could mean that the elements already defined by Dublin Core could be used directly (see Ian Davis's list [1]).

But if RSS 1.0 was revisited, parts of the syntax that add to its complexity could be dropped: the outer <rdf:RDF> element is no longer required for RDF/XML documents (RDF spec change), the items/Seq elements aren't necessarily desirable. Making these improvements would mean that current plain XML-oriented tools (i.e. most aggregators) would no longer be able to read the syntax. We might as well use the Atom elements we've got now.

So the arguments about complexity of RSS 1.0-style syntax aren't all that valid any more - if you drop the <rdf:RDF> and Seq/items then the only 'complication' remaining is XML namespaces, something that any TAG-fearing format should support these days anyway.

I still believe it would make sense to use full RDF/XML syntax to define Atom - Atom is describing resources after all, exactly what RDF is designed for. But there has been resistance to such a move. Some retooling would be required to get aggregators to handle full RDF/XML, arguably be an upgrade but not everyone sees it that way. However as a fallback position I personally think that Atom defined as a tight, *extensible* XML spec with a mapping to the RDF model would fulfil most requirements of the RDF community, and still benefit Atom. We'd get the model.

So how exactly would the RSS 1.0 syntax be reused? What would be gained?

Cheers,
Danny.

[1] http://internetalchemy.org/2004/03/theNucleusOfAtom

--
----
Raw
http://dannyayers.com