- From: William F Hammond <hammond@csc.albany.edu>
- Date: 13 Jun 2003 12:58:14 -0400
- To: W3C HTML Specification Discussion <www-html@w3.org>
Ian Hickson <ian@hixie.ch> writes: > On Thu, 10 Apr 2003, Masayasu Ishikawa wrote: > >> > >> In any case, the effect of nested elements with > >> href attributes is something that does need to be addressed in the > >> XHTML2 standard. > > > > Indeed. That was one of the topics at a meeting with the WAI User > > Agent Accessibility Guidelines Working Group, and something we need > > to address from accessibility point of view as well. > > Hopefully such clarifications in the XHTML2 spec will not conflict with > requirements of event propagation in the DOM specs. What is the significance of href "propogation" to a child element? Isn't it a mistake to have the hypertext attribute collection be so universal? For example, in the 20030506 draft at section 6.5 one finds <abbr href="http://www.w3.org/" title="World Wide Web">WWW</abbr> . How is this different from <a href="http://www.w3.org/" title="World Wide Web"><abbr>WWW</abbr></a> ? Is the former just intended as a form of markup minimization? Or is href with abbr intended to be something like a rel attribute? If the latter, reluri would be a clearer name, but that interpretation does not justify having the hypertext collection, including "target", be universal. And (not that markup should be described in terms of actions) why might an author want something like <body href="http://www.w3.org/"> ? Given that what might one imagine "happening" to a user who clicks on the middle of a page with the intention of giving platform focus to his user agent? What is a user agent like "lynx" or "w3m" supposed to do when every element either has its own href or an inherited one? (Don't we all use ultra-safe user agents for checking out mail of unknown origin that is likely to have "src" attributes wired to http head collectors for delivery verification? This is not just a WAI issue.) > In particular, > implementations that are not XHTML-specific implementations but generic > XML+DOM+CSS UAs, using UA stylesheets and scripting bound through UA > stylesheets (using technologies like XBL or HTCs) should hopefully end up > automatically complying to the XHTML specs. To the extent that user agents are just XML+CSS or XML+DOM+CSS or XML+DOM+Events+CSS, isn't specifying XHTML a waste of time? I'd suggest that XHTML documents should have connotations of "safety" and "sanity" in regard to use of client platform resources not necessarily available in the more general framework. In particular, it should be difficult for an XHTML author to produce a document that is not WAI compliant. -- Bill
Received on Friday, 13 June 2003 12:58:18 UTC