xml:id and validation

It seems to me that some of the goals of RDF/A depend on validation ...
I raise this not because I see it as a problem, but because I am aware 
that some people do.
I might be wrong here, it interacts with XHTML modularisation and 
documents from multiple namespaces.

====

Here is my thought process:

RDF/A is explicitly a set of attributes to be used within an XML host 
language, e.g. XHTML.
The attributes do not themselves have qualified names, and so depend on 
the namespace of the element.

Thus if we have svg and xhtml2 mix we may find something like

<svg:svg id="image">
    <xhtml2:meta property="dc:creator" content="an artist"/>
</svg:svg>

The meta element has a property and an object and so takes its subject 
from the parent.
The subject of the parent is either an xml:id or a gensymed blank node, 
depending on whether the id attribute on svg:svg is, or is not, an ID 
attribute. IIRC this can only be resolved by looking at the DTD.

===

I think my implementation would be likely to treat any id as an ID, in 
this sort of case - in practice the cost of rejecting invalid xhtml is 
just too high to require validation.

Jeremy

Received on Tuesday, 26 October 2004 14:39:57 UTC