RE: [syndication] Re: PRISM and RSS

[dave.cantrell@gunter.af.mil]

>**1st off, I'd like to say that I only have some basic knowledge of 
>RDF, so if I make any mistakes (and I invariably will), please 
>correct me.  The specs at W3C are pretty harry.  _Thanks_.**

Ditto that.

>> I like PRISM a lot, but RSS 1.0 is simpler.
>
>"Simpler" is subjective... I think there is more to their spec than 
>RSS 1.0, but you don't have to use what you don't need.

Likewise you don't have to use everything in RSS 1.0 (I'm assuming, anyway).

>RSS 1.0 uses "item" tags to describe news items and the "channel" tag 
>to denote the channel.  In PRISM, they are all rdf:Description.  Why 
>doesn't RSS 1.0 use rdf:Description?  Also, I see tags like "title" 
>and "link" in an RSS 1.0 document.  By convention, the rdf:about 
>attribute is "link", so why the redundancy?  Why "title" in the RSS 
>1.0 namespace when there is one of the same name in the Dublin Core 
>namespace?  These things strike me as RSS 1.0 "doing its own thing" 
>and deviating from more standard compliance.

I agree with you on the <title> element. The <channel> is there I presume to
follow the baseline of 0.9, as set forth previously on the list by others. I
for one never understood the need for a <channel> element anyway -- RSS is a
channel, so why the nested redundancy?

As for rdf:about and rdf:Description: I personally almost became ill when I
tried to figure out the OCS spec precisely because it was based on
rdf:abouts. Nothing against the OCS spec itself, mind you, I'm sure it works
great, and I use it myself, but I chose it more as an excercise in playing
with RDF syntax than anything else. Practical concerns should have steered
me into a simpler direction.

This was compounded when I asked about how to extend it to include
categories, since I wanted to show a series of links to channels as a group
with a title, e.g. "Top Stories", etc. I was instructed to use the RSS
Taxonomy module, which is likewise based on rdf:abouts. Since I didn't need
URIs but instead plain-text labels, I created my own ad-hoc namespace with
it's own <category> tag, and solved the problem myself.

Maybe I don't get the Taxonomy module (and to a larger extent, RDF) but why
the need to use URIs as categories instead of plain ol' text strings? I can
see the usefulness in URIs as an identifier, but unless I'm categorizing
something into say a DMOZ category, it doesn't help me one damn bit.

>Given the close similarity between both specs, and the fact that both 
>are basically finished (barring namespaced extentions), I don't think 
>differences in culture is a sufficient argument for RSS 1.0 to 
>exist.

Okay. Try this:

  Given the close similarity between both specs, and the fact 
  that both are basically finished (barring namespaced 
  extentions), I don't think differences in culture is a 
  sufficient argument for PRISM to exist.

It can work both ways. But I do feel PRISM is different. At least from what
I've looked at of the spec, it seems to be more of an "xmlNews-ish"
application geared toward magazine publishers, etc, and specifically at
syndicating the *entire* contents of an article, including binaries (or at
least providing URI pointers to them).

RSS 1.0 is not that. It is headline syndication at it's core. But being
extensible by nature, it can grow to include this. All someone has to do is
come up with a spec for
"mod_full_article_syndication_with_everything_included" or whatever. I
believe this was the intent behind the earlier statements about being able
to place PRISM elements in with RSS 1.0.

RSS 1.0 is designed to be lightweight at it's core, yet extensible enough to
grow as-needed. This is why I find myself liking it more and more. By
contrast, PRISM seems locked into providing one particular type of
information: "large-scale" syndicated content (e.g. full articles). I just
read over the spec briefly, but nothing jumped out saying "you can start
small (e.g. headlines) and add more on later". That's exactly what RSS 1.0
offers. (as I understand it, anyway)

And for the record, my liking it has nothing to do with it being RDF. I
would just as soon have seen it in the earlier idea that was discussed,
regarding namespacing the <rss> element to make it extensible, like so:

	<rss xmlns="http://www.purl.org/rss/1.0";>
		blah blah blah
	</rss>

This would allow us to build modules into it without having to worry about
RDF syntax. But it's already been done, and I can live with that. And even
though I can't see any real-world practical use for RDF yet, I have been
around long enough to never say "never" and just put faith in those smarter
than I.

-dave