- From: Al Gilman <Alfred.S.Gilman@IEEE.org>
- Date: Mon, 21 Aug 2006 21:41:33 -0400
- To: Shane McCarron <shane@aptest.com>, Karl Dubost <karl@w3.org>
- Cc: XHTML-Liste <www-html@w3.org>
At 4:32 PM -0500 8/21/06, Shane McCarron wrote: >Karl Dubost wrote: >>It is not a *defined* mechanism. >>The whole point of QNames is disambiguation, not interoperability. >I maintain that the requirements for the interpretation of QName >namespaces and their associated taxonomies is beyond the scope of >XHTML. It would be great if some group were to somehow instrument >the Internet so that people could define the semantics of their role >/ property / whatever attributes in a machine interpretable way. >Sadly, the mechanism for this does not exist. At the end of the >day, all you are doing is adding more and more layers of abstraction >- the taxonomy interpreter, whatever it is, still requires arcane >knowledge of *something* in order for the interpretation to take >place. Some of each. The fact that @role takes a QName value - is more defining than the extensibility of @class in HTML4 [but just by a bit.] There is a rather popular way of introducing definitions for extension properties annotated in the @class attribute, commonly known as "microformats." Even if you don't use topic maps or RDF there is a lot one can do. But let me set that aside for now, and talk about our PFWG work where we do apply RDF to giving the kind of machinable backup to QNames that Karl seeks. Coming (with any kind of luck, soon) to a TR Page near you will be a draft specification on "Roles for Accessible Rich Internet Applications." This specification will use RDF to associate semantics with URIs that can be referenced in brief by QNames. This is a conscious, designed partner to the XHTML Role Attribute Module. We in the PFWG believe that this mechanism allows us to publish responsible, machinable concept definitions and have web authors employ them as role definitions that they can cite without a great deal of hypertext bloat. User agents can call on them for augmented user coaching, and author tools can call on them as well to construct prompts and menus for authors. The QName nature of the possible values of @role does not guarantee you a *uniform* manner of defining the role concepts cited. But it does *encourage* the use of well-defined concepts such as we believe we will be publishing. The value that @role will add [1] to the Pronunciation Lexicon Specification format would not be possible if we get too hidebound about what manner of definition the QName value refers to. The discriminant patters in pronunciation run the gamut from context to concept to speaker group. Ada was the language that worked hardest at making it hard to write programs wrong. The market has opted to leave the quality of programs to the writers, not to the notation. The varying quality is real; but it's in the taxonomies. If you have quality definitions, a QName is fully adequate to invoke one of them. Al [1] http://lists.w3.org/Archives/Public/www-voice/2006JulSep/0029 >If someone tells you otherwise, they are selling something. > >-- >Shane P. McCarron Phone: +1 763 786-8160 x120 >Managing Director Fax: +1 763 786-8180 >ApTest Minnesota Inet: shane@aptest.com
Received on Tuesday, 22 August 2006 01:41:56 UTC