On 22 Jun 2008, at 11:52, Simon Reinhardt wrote:
I think it would be best to implement the mechanism described here:
http://www.w3.org/TR/cooluris/#hashuri
This would mean:
<b00b07kw#episode> is the thing
<b00b07kw.rdf> is the RDF variant
<b00b07kw.html> is the HTML variant
<b00b07kw> is a generic, content-negotiated document; it serves the
right variant directly, without any redirect, and gives the URI of
the selected variant in the Content-Location header.
I suppose it's not a popular view anymore because that document has
an official status now, but there's still the problem with #episode
denoting an element in the HTML variant.
We discussed this quite a bit with the TAG before finalizing the Cool
URIs document, and the TAG insisted it's the right thing to do.
Either you do have an element with that ID there, then #episode
denotes both a thing and an HTML element.
Which is bad and should be avoided. Don't put an id="episode" into the
HTML.
Or you don't have it, then this would be regarded as broken HTML.
Well, not really. The HTML is not broken, it's perfectly fine. The
perceived issue is that *the RDF* references a fragment that is not
defined *in the HTML*.
Maybe that's just a theoretical issue which can simply be ignored
for pragmatism.
I think so. Although some of the relevant specs (e.g. the URI RFC and
the HTML MIME type registration) might benefit from some clarifications.
Best,
Richard
Simon
