Re: XLink 1.1: Error handling

On Wed, 15 Feb 2006 18:35:03 +0100, Norman Walsh <Norman.Walsh@Sun.COM>  
wrote:
> The spec now:
>
> 1. Identifies conformant markup.
> 2. Specifies that non-conformant markup has no XLink semantics

It apparently does not define what "no XLink semantics" means, does it? It  
should be made clear that when an _unknown_ attribute in the XLink  
namespace is applied on an element the element can no longer be a link per  
definition.

However, if the WG is still considering this I would like to suggest  
easier error handling where an unknown attribute or attribute with an  
unknown value does not necessarliy effect the other attributes. Basically,  
"MUST ignore":

  1. Attributes not defined in this specification in the XLink
     namespace MUST be ignored.
  2. Attributes defined in this specification having values not
     allowed by this specification MUST be ignored.

I think this would be much easier to implement and I think this would  
satisfy my concern.


> That means that neither of the elements above have XLink semantics,
> they don't function as links. We haven't said anything about what a
> processor must do with these errors, but I think that's ok.
> Applications are free to report them (applications, as I'm fond of
> pointing out, are free to warn about the phase of the moon if they
> wish) but they are not required to take any particular action (which
> might not be appropriate in simple link-harvesting applications
> anyway).

Hmm, I read this later... So this is what I basically disagree with. See  
above.


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

Received on Wednesday, 15 February 2006 23:29:18 UTC