Re: HTML 4 Profile for RDFa

Philip Taylor wrote:
>> Shane McCarron wrote:
>>> [...]
>>> http://www3.aptest.com/standards/rdfa-html/
> [...]
> How are xmlns:* attributes meant to be processed?

To elaborate: Specifications exist for XML Namespaces, but as far as I'm 
aware nothing exists to define the meaning or processing of similar 
syntax in text/html. So I'd want something somewhere to define some 
processing requirements vaguely like:

   The <def>pseudo-namespace mapping context</def> is a map from prefix
   strings to namespace strings, and is defined as follows:

     The pseudo-namespace mapping context of a Document node is the empty
     map.

     The pseudo-namespace mapping context of a DocumentFragment node is
     the empty map.

     The pseudo-namespace mapping context of an Element 'e' is computed
     as follows:

       * If the parent node of 'e' is null, let 'm' be the empty map.
         Otherwise, let 'm' be the pseudo-namespace mapping context of
         the parent node of 'e'.

       * For each attribute in 'e' that is in no namespace, and whose
         local name consists of the case-sensitive string "xmlns:"
         followed by a non-zero-length string of characters 'p', and
         whose value is not the empty string:

         * Update 'm' so that 'p' is mapped onto the value of the
           attribute. (This might replace an existing map entry.)

       * For each attribute in 'e' that is in the namespace
         "http://www.w3.org/2000/xmlns/":

         * Update 'm' so that the attribute's local name is mapped onto
           the value of the attribute. (This might replace an existing
           map entry.)

           (Note: This assumes XML parsers will insert xmlns:* attributes
           into the DOM, which probably isn't guaranteed.)

       * The pseudo-namespace mapping context of 'e' is 'm'.

   Where [RDFaSYNTAX] currently says:

     "2. Using the prefix and the current in-scope mappings, obtain the
     URI that the prefix maps to."

   replace the step with:

     "2. Let 'm' be the pseudo-namespace mapping context of the element
     with which this CURIE is associated. If the prefix is not in 'm'
     (case-sensitively), abort these steps and do some kind of error
     handling that's not defined yet. Otherwise, obtain the URI that the
     prefix maps to."

I just made this up now so it's probably not especially correct or 
well-written (e.g. maybe the definition should be turned inside out), 
but I think it defines enough detail to determine the output of the 
markup examples I gave in my previous email and it generally agrees with 
the very rough testing I performed on some JS-based and html5lib-based 
RDFa implementations.

-- 
Philip Taylor
pjt47@cam.ac.uk

Received on Thursday, 14 May 2009 21:09:28 UTC