RE: ACTION-156: Review of

Stuart Williams writes:

> it does remain unclear to me particularly wrt to "markup tag names" 
> whether you are talking of: 1) the name of an occurence of a tag in 
> a document (which might be named by say an xpath based frag id 
> extending a document URI - though formally the FYN story stumbles a 
> little in RFC 3026

No, not what I meant.  I'm not quite sure why the term "markup tag name" 
would be suggestive of the particular occurrence of such a tag in a 
document.  It's the name of the tag we're talking about, e.g. the "dog" in 

> or 2) the name of a kind of element eg. the xhtml html element


> which presumably has to appeal to something like schema component 
> designators. However, I suppose given that I can conceive of way to 
> do either it perhaps doesn't matter what you had intended.

Why are schemas involved at all? I can write a specification for an XML 
vocabulary to be used for, say, printer control.  So, that tag names might 


etc.  In the paper specification I could write something like:  "The W3C 
in its finding on the Self-Describing Web Recommends the assignment of a 
URI for the name of each tag used in a vocabulary like this.  Mindful of 
that advice, we have assigned URIs for tag names such as the ones shown 
above, e.g.:

Why the appeal to schemas or schema component designators?   In fact, SCDs 
do not designate tag names, they designate schema components, some of 
which can be used to declare elements and attributes (and, by the way, 
SCDs are at the moment much more like fragment identifiers than full URIs, 
owing to the lack of a standardized means of providing a URI for the 
schema that results from the composition of, for example, 5 schema 
documents for 5 different namespaces that together are used to validate 
one mixed namespace document).  Ideally, to get the benefit of Web based 
extensibility, that specification should further provide an ability to mix 
in other tag names minted by different organizations, but the means of 
disambiguating them would presumably be specific to that particular 
specification.  In practice, I think that QNames are the most likely means 
of getting distributed extensibility for XML element names, as described 
later in section 4.2.3 of the draft finding.

Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142

Received on Monday, 15 September 2008 18:41:54 UTC