[
Lists Home |
Date Index |
Thread Index
]
> >Just so. While the RSS developer community may be a terrifying mess
> >-- stilettos in everyone's back and all that -- RSS as a spec (or
> >family of specs) is one of the few real successes XML can point to:
> >information in XML actually distributed online to lots of users.
>
> Except that RSS isn't XML. It claims to be, but in practice it isn't,
> and many members of the RSS community are radically opposed to making
> it real XML, with draconian error handling and markup encoded as
> markup. If RSS is one of the few real successes we cam point to, then
> something's seriously wrong with XML.
This is a considerable oversimplification. For a start, which version are
you talking about? RSS 1.0 is RDF/XML, so the normal rules don't apply there
anyway...
Jeff Barr has some version usage stats (from Syndic8) [1] :
0.91 : 47%
1.0 : 26%
2.0 : 14%
0.92 : 11%
A moderate proportion of the 0.91/0.92 feeds are iffy XML. The same applies
to RSS 2.0, the version that has just gone to Harvard. This is backwards
compatible with the 0.9x branch. (Incidentally, 2.0 adds an novel <guid>
element which is either a string or a URI depending on the value of one of
its attributes). These formats were marketed as "simple" and so lots of
people implemented based on a casual "View Source". The 0.9x/2.0 specs are
thin and vague, and things like validation (XML Schema, DTD) were swept
under the carpet, although Sam Ruby and Mark Pilgrim went to the trouble of
putting together an online feed validator not too long ago. Sam recently
estimated that a little over 80% of feeds are well-formed. RSS 1.0 feeds can
also be checked using RDF validators, btw.
It all may sound bad, but this material is mostly content, so compare these
figures with the quality of (X)HTML there is around. But the real point is
that there's a lot of good stuff around.
Nearly all RSS feeds (all versions) are machine-generated, so they tend to
be usually right or usually wrong. A useful proportion of these feeds *are*
well-formed XML, a fair proportion can be validated, a fair proportion are
valid RDF/XML.
The proportion of end user tool-generated RSS is increasing all the time
(over hacker's bedsit-generated), this will usually be as good XML as
whichever spec/library has been chosen allows.
But look at what it provides. There are maybe hundreds of thousands of RSS
feeds to choose from, for me to read in my little desktop aggregator. It
works very well, saves me the trouble of browsing to the 100 or so
frequently updated sites I look at most mornings (I simply wouldn't do this
otherwise). Quite often there'll be a glitch in a feed or two, 404 or can't
parse, but this isn't a big problem. If some of the fruit has gone off, you
just stick with the stuff that hasn't.
Another point is that rapid development is taking place - the data can be
used directly as RDF (RSS 1.0) or translated with XSLT or programmatically.
It can and is being used alongside information expressed using other RDF
vocabularies (e.g. FOAF in the new TypePad blog publishing system and the
NewsMonster reader). RSS may sometimes be awful, but most of the time it's
useful XML, it's a good use of the web, and some of it is already part of
the Semantic Web.
It's very useful and it's mostly XML. Sounds a promising start.
Cheers,
Danny.
http://dannyayers.com/index.rdf
[1] http://www.syndic8.com/stats.php?Section=rss
|
