Re: Erratum notification: XLink, href, escaping

* Chris Lilley wrote:
>> The procedure is applied when passing the URI reference to a URI
>> resolver.
>
>  No, it can't do that because to get a URI reference, I have to have
>  already applied the procedure and thus don't need to apply it again.
>  Also, it doesn't say what to do in the more useful case where the
>  attribute value is not already a URI reference. Its clear enough
>  what was meant:
>
>The procedure is applied, if needed, when passing the attribute value
>to a URI resolver.

Since neither XLink nor any of its normative references define the term
"URI resolver" and the term in practise means anything from resolving a
relative URI reference to an absolute URI reference (which is the most
obvious interpretation as RFC 2396 uses the verb "resolve" only to refer
to just this very provess) to retrieving information identified by the
URI reference, this isn't much clearer to me.

Whatever wording is chosen by XLink and Technical Reports which used the
same text (SVG 1.0 for example) it should ensure that it does not make
any difference whether an element with base URI http://www.example.org/
specified any of

  xlink:href="Björn#Björn"
  xlink:href="Bj%C3%B6rn#Björn"
  xlink:href="Björn#Bj%C3%B6rn"
  xlink:href="Bj%c3%B6rn#Björn"
  xlink:href="Bj%c3%B6rn#Björn"
  xlink:href="Bj%C3%B6rn#Bj%C3%B6rn"
  xlink:href="http://www.example.org/Björn#Björn"
  xlink:href="http://www.example.org/Bj%C3%B6rn#Björn"
  xlink:href="http://www.example.org/Bj%C3%B6rn#Bj%C3%B6rn"
  xlink:href="Bj%c3%b6rn#Björn"
  ...

to any software as far as the disallowed characters are concerned.
The term "URI resolver" should be avoided due to its ambiguity.

Received on Wednesday, 23 July 2003 22:03:20 UTC