Re: rel=CURIE in RDFa, but rel=URI in Link:

On Sep 28, 2008, at 1:33 AM, Anne van Kesteren wrote:
> On Sun, 28 Sep 2008 09:41:49 +0200, Julian Reschke  
> <julian.reschke@gmx.de> wrote:
>> I think the other point Roy was making that link relations are  
>> URIs, not IRIs, so new specs that allow non-ASCII characters in  
>> link relations should deal with the information the loss when  
>> transferred as URI.
>
> To my knowledge HTML4 was the first specification to define link  
> relations and they were not URIs and could contain non-ASCII  
> characters if you defined one yourself if you declared a profile:
>
>   http://www.w3.org/TR/html4/types.html#type-links

Link types have been in HTML for as long as the WWW has been an open  
project.
They were defined by the type attribute in

<http://1997.webhistory.org/www.lists/www-talk.1992/0000.html>

as strings. "Strings for types with particular semantics will be
registered by the W3 team."  That registration was in hypertext.

<http://www.w3.org/DesignIssues/LinkTypes.html>
<http://www.w3.org/DesignIssues/Topology.html#4>

and have been defined many times in many places over the years

1993:
  <http://www.w3.org/Protocols/HTTP/Object_Headers.html#link>
  <http://www.w3.org/MarkUp/draft-ietf-iiir-html-01.txt>
  <http://www.w3.org/MarkUp/HTMLPlus/htmlplus_54.html>
1995:
  <http://ftp.ics.uci.edu/pub/ietf/html/draft-ietf-html- 
specv3-00.txt.gz>
  <http://www.w3.org/MarkUp/Resource/Specification>
  <http://ksi.cpsc.ucalgary.ca/archives/HTML-WG/html-wg-95q1.messages/ 
thread.html>
  <http://ksi.cpsc.ucalgary.ca/archives/HTML-WG/html-wg-95q2.messages/ 
thread.html>
  <http://ksi.cpsc.ucalgary.ca/archives/HTML-WG/html-wg-95q2.messages/ 
0277.html>
  <http://ksi.cpsc.ucalgary.ca/archives/HTML-WG/html-wg-95q2.messages/ 
0887.html>
  <http://ksi.cpsc.ucalgary.ca/archives/HTML-WG/html-wg-95q2.messages/ 
1008.html>
1996:
  <http://ftp.ics.uci.edu/pub/ietf/html/draft-ietf-html-relrev-00.txt>
  <http://www.w3.org/Architecture/NOTE-link.html>

> Also, they are case-insensitive and therefore might need to be  
> normalized if you want to "bind" them to a URIs.

The definition has changed so many times that it doesn't really matter
what we do to normalize them.  Only singleton short strings work today.
The decision to lowercase the short string and, if a URI is needed,
treat it as relative to the IANA registry, supports backwards  
compatibility
for singleton relationships while satisfying the crowd that thinks
they need arbitrary URIs.

....Roy

Received on Monday, 29 September 2008 05:42:53 UTC