W3C home > Mailing lists > Public > www-tag@w3.org > February 2010

Re: HTML media type vs. # URIs that do not identify document elements

From: <noah_mendelsohn@us.ibm.com>
Date: Fri, 5 Feb 2010 14:47:58 -0500
To: Jonathan Rees <jar@creativecommons.org>
Cc: Ben Adida <ben@adida.net>, www-tag@w3.org
Message-ID: <OF14E1EB0B.F81C0125-ON852576C1.006C773F-852576C1.006CC444@lotus.com>
Jonathan Rees writes:

> Clearly being out of spec does not seem to be a problem for anyone who
> does this kind of thing, but it is sort of an embarrassment.

I can't dispute that it doesn't "seem" to be, but I think I'm right that 
having HTML fragments used in this way could cause a user agent to do the 
wrong thing, I.e., to attempt to scroll to or otherwise focus on a piece 
of the document with that identifier.  Though most browsers fail silently 
when there's no match on a fragid, I don't think it would be inappropriate 
for a browser or any other software to display some sort of "URI doesn't 
resolve" error message when attempting to dereference such a URI.  So, I 
don't think this usage is entirely benign.

Noah

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------








Jonathan Rees <jar@creativecommons.org>
Sent by: www-tag-request@w3.org
02/05/2010 02:09 PM
 
        To:     www-tag@w3.org, Ben Adida <ben@adida.net>
        cc:     (bcc: Noah Mendelsohn/Cambridge/IBM)
        Subject:        HTML media type vs. # URIs that do not identify 
document elements


http://www.w3.org/TR/swbp-vocab-pub/ advocates providing RDF and HTML
versions of ontologies using content negotiation, and this is a
pattern that is, I believe, widely deployed. The hack is that in the
HTML version you have
  <a name="foo"> ...documentation for http://blah/bar#foo ...
and in the RDF you have
  <... rdf:resource="http://blah/bar#foo" ...> ... properties of
http://blah/bar#foo ...
There is a problem: the media type registration for text/html (also
application/xhtml+xml) says: "For documents labeled as text/html, the
fragment identifier designates the correspondingly named element". So
using the #foo URI to designate anything other than an element, as the
RDF 'representation' does, is out of spec (when there is an HTML
representation).

The same problem can arise with RDFa, even in the absence of content
negotiation.

Clearly being out of spec does not seem to be a problem for anyone who
does this kind of thing, but it is sort of an embarrassment.

Since the text/html media type is under revision, I wonder if anyone
has looked into making it more RDF-friendly, so that this usage
becomes legitimate?

Jonathan
Received on Friday, 5 February 2010 19:48:50 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:48:19 GMT