W3C home > Mailing lists > Public > www-rdf-interest@w3.org > October 2004

Re: Atom and RDF

From: Danny Ayers <danny.ayers@gmail.com>
Date: Tue, 12 Oct 2004 00:21:03 +0200
Message-ID: <1f2ed5cd04101115212ee4281b@mail.gmail.com>
To: Joshua Allen <joshuaa@microsoft.com>
Cc: rdfweb-dev@vapours.rdfweb.org, www-rdf-interest@w3.org, semanticweb@yahoogroups.com, rss-dev@yahoogroups.com, atom-syntax@imc.org

Thanks guys, I'll collate the posts in a day or two and pass feedback
back to the Atom list.

On Mon, 11 Oct 2004 12:26:08 -0700, Joshua Allen <joshuaa@microsoft.com> wrote:
> > HTTP), so there's clearly some overlap with RDF efforts. RSS
> > 1.0 extensions must follow the constraints of RDF. So RDF is
> 
> Are there actually any RSS 1.0 extensions in popular use?  

In popular terms, the FOAF stuff really seems to be gaining ground,
and I've seen quite a few niche (not /quite/ the right word) things
like Nature's use of the PRISM vocab.

I note that
> RSS 2.0 approach made it very easy for podcasting to catch hold, and
> this may be a consideration when evaluating an extensibility philosophy.

<enclosure> is a core element, so I'm not entirely sure how that
relates to extensibility. Definitely something to learn from though.
As is usually the case I doubt the lessons will be entirely clear cut.
Check out these comments: http://gonze.com/weblog/story/5-17-4

<enclosure> does appear to have significant flaws, and was frozen into
the spec that way. At the time I'm not even sure Userland were using
it, there certainly wasn't what you'd call significant deployment. Yet
now, two or so years later it's suddenly the hottest thing. Ok, maybe
there is a point there in relation to extensibility - had this
appeared in a 3rd-party extension, would it still have been picked up
in the same way? Another question, could it have been better designed
and picked up in the same way?

Speaking philosophically, I'm starting to wonder if the good Mr. Winer
mightn't simply have entered into a pact with some horned deity or
other...

> > reading) support both formats. Most syndication tools only
> > support enough of the RDF model to be able to extract the
> > syndication-domain data (there are notable exceptions, and of
> 
> This is perhaps wishful.  Most syndication tools do not support RDF
> model at all, and simply read the RSS 1.0 file as XML.  Most tools would
> barf given any of a number of isomorphic RDF models which stray even
> slightly from the expected XML syntax.

Ok, it's syntax based, but the format and model is defined through
RDF, not as a direct mapping between characters and syndication-domain
information. Our dog only understands about three words* (only one of
which he generally responds to**), but those words are most definitely
English, not Dog.

> > course virtually all RDF tools can consume & process RSS 1.0
> > out of the box). Developers of syndication tools tend to have
> 
> Again, I would temper this -- they can consume the RDF graph, but not as
> RSS (unless you meant to say "virtually all RDF-based syndication
> tools").

Well no, I meant what I said - they can consume the RDF vocabulary
defined in RSS. They can merge graphs, make inferences based on the
statements in schema and instance data and so on.

> > Why does it matter? I think RDF potentially stands to gain
> > quite a lot from Atom, in the form of a relatively
> > lightweight but versatile and thoroughly spec'ed transport
> > (and editing protocol) which is likely to gain widespread
> 
> How is this use case any different than using RSS (or Atom) as a
> protocol for moving around audio streams, bittorrent pointers to TV
> programs, and other payloads?  

Hmm, I was going to ask then why you distributed your RSS feed as text
rather than as a zipped blob. Then I realised that it probably does
get delivered as a gzipped blob.

The fact that RSS can be useful with a
> wide variety of payloads is evidence IMO that it defines a good
> separation between the protocol and payload, and my suggestion would be
> to treat RDF as an opaque payload exactly like any other.

Perhaps, but I have a strong feeling that such an approach might miss
out on significant potential of the Web. Opaque audio streams are one
thing, they have one primary form of interpretation, but descriptive
data is something else entirely. It's bad enough with documents - the
web is built of HTML, not PDFs. It's clear that descriptive metadata
distributed through syndication channels is useful, thats what RSS and
Atom are. I suspect long term that top-level flexibility will prove
more interesting than efficient tunnelling, though both probably have
their place.

Cheers,
Danny.

* "Basil", "biscuit", "sit"
** "biscuit"

-- 

http://dannyayers.com
Received on Monday, 11 October 2004 22:21:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:10 GMT