Re: ACTION-79 discussion on URI vs. IRI in the specs

On 5/26/2011 7:49 AM, Manu Sporny wrote:
>
> I agree with Julian on all points. We should be using the IRI
> terminology. Mischa's original comment after his read of the document
> demonstrated that the current language is confusing. Changing it in a
> minor way probably won't change how it reads to someone that has no idea
> about the nuances between all of the documents listed.
>
> While all of what you said is logically sound, Shane - I think people
> are going to become more and more confused if we keep using the term URI
> when we really mean IRI.

Hmm.  Okay.  I am not going to fight about this.  I have several 
concerns though:

   1. We use datatypes that include the characters 'URI' in their
      names.  Some of these are legacy (e.g., @href) and immutable. 
      Using a datatype name like 'SafeCURIEorCURIEorAbsIRI' on one
      attribute and 'URI' on another attribute will lead people to
      believe that one permits internationalized data and the other does
      not.  How do we reconcile this?
   2. We are dependent upon many other specifications, all of which seem
      to use the term 'URI' when they in fact mean 'IRI'.  XHTML
      Modularization is only one such.  You might also look at the RDF
      Recommendation (which uses weasel wording similar to what I
      proposed in its section 6.4).  If we use the term 'IRI' throughout
      our specifications, how to we connect this back to the term 'URI'
      in documents upon which we depend?
   3. In the case of HTML5+RDFa, HTML5 uses the term 'URL' in the way we
      would use the term 'IRI'.  How should we reconcile this?
   4. In the case of XHTML+RDFa, XHTML uses the term 'URI' in the way we
      would use the term 'IRI'.  How should we reconcile this?
   5. Other specifications (The Role Attribute, for example) depend upon
      our definition of CURIE.  If we shift our terminology to say 'IRI'
      when everyone else at the W3C seems to be using the term 'URI' or
      'URL' to mean the same thing, aren't we just introducing a
      potential source of confusion?

> /If/ for some reason we do end up using the term URI when we mean IRI,
> we should probably put some text in the document to this effect:
>
> """
> RDFa has complete support for internationalized characters. This
> includes internationalized characters in the subject, property and
> object location.
> """
>
> I think we should just come out and say something like that very early
> in the document in case there is any doubt. Perhaps we should put an
> example in the RDFa 1.1 Primer to that effect as well - Japanese, Arabic
> or some other language far removed from English may make the best example.

I like this regardless.  It's important, and we should tell people about it.

-- 
Shane McCarron
Managing Director, Applied Testing and Technology, Inc.
+1 763 786 8160 x120