Re: XLink 1.1: Xlink vs "legacy" linking

On Mon, 23 Jan 2006 22:31:20 +0100, Bjoern Hoehrmann <derhoermi@gmx.net>  
wrote:
>> The Core WG has considered this question again. As far as we can see,
>> the XHTML specification, if it chose to adopt XLink, could say what a
>> UA should do. The XLink specification describes precisely what XLink
>> semantics the element has. There's nothing more that the XLink
>> specification *can* say. An XHTML UA that encounters this (invalid)
>> XHTML element is free to do whatever the XHTML speicification says it
>> can do. If it's also an XLink application, then it can implement the
>> XLink semantics if it wants (or at user request).
>
> What do you mean by "can"? Can I make a fully conforming XHTML+XLink
> user agent that does not observe any of the XLink processor require-
> ments? All I want to do is to make a XHTML+XLink user agent that fully
> complies with all XLink and XHTML requirements, I think this is not
> currently possible, so either this is made possible or the draft needs
> to point out that making such a user agent is a bad idea. I'd think it's
> a bad idea. I'm happy with this beeing a bad idea, but it needs to be
> clear from the draft.

This question arises when a UA wants to implement both XHTML and XLink.  
Even if you implement them to be used apart when someone does create a  
document that uses elements and attributes from both dialects you somehow,  
as a UA, have to resolve that situation. Given that XLink affects all  
dialects that have some type of linking build in already addressing this  
in the XLink specification seems appropriate.

I've already suggested that "XLink attributes MUST be ignored when  
attributes with equivalent semantics are specified on the same element."  
could be used at some point. Perhaps with examples as given in other  
threads.


-- 
Anne van Kesteren
<http://annevankesteren.nl/>
<http://www.opera.com/>

Received on Monday, 23 January 2006 21:44:09 UTC