- From: Ian Hickson <ian@hixie.ch>
- Date: Sat, 29 Mar 2008 00:50:43 +0000 (UTC)
- To: public-pfwg-comments@w3.org
Hi, I've been asked to send some comments to this list about the prospects of integrating ARIA in HTML5. I think there are two issues that could prevent HTML5 from adopting ARIA. The first is lack of clear normative conformance criteria. The exact semantics of the ARIA features must be described in detail. It is not sufficient to say that the feature maps to an accessibility API. For example, if ARIA includes a way of defining that an element is a header, the specification must define exactly what it is a header of. HTML5 does this, for instance, when defining in detail the semantics of the headers and sectioning elements: http://www.whatwg.org/specs/web-apps/current-work/multipage/section-sections.html#headings This has to be done for everything. Is something a table? What's the table model? Is something defining the relationship between a control and a label? What is the exact mapping? What happens when there's more than one label, or a label maps to more than one control? Everything must be defined in excruciating detail. The second issue is conflicts with existing features. For example, if ARIA defines a feature to say that something is a header, this will conflict with the HTML5 header algorithm. If something defines that an element is a radio button, but that something is applied to a checkbox control, then we have a problem where some users will get a different behaviour than other users, which fundamentally doesn't solve the accessibility problem we are trying to solve, it just makes it worse. Cheers, -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
Received on Saturday, 29 March 2008 00:51:21 UTC