Re: @required and @disabled - strong or weak ? (was RE: Does the HTML5 required attribute have the same accessibility affect as aria-required for an ARIA defined widget?)

On May 9, 2014, at 8:38 AM, John Foliot <john@foliot.ca> wrote:

> The general rule-set should reflect that in the case of any implied (boolean) attributes, author-supplied aria-attributes are the "stronger" of the 2.

By this I assume you mean "implicit value due to missing boolean attributes."  If that's what the group wants, we need an issue for ARIA 1.1, because the 1.0 spec says the opposite.

>  This applies not only to @required / aria-required, but also to such chestnuts as @checked (or for those old XHTML1 fans checked="checked") versus aria-checked, where the absence of the @checked attribute DOES NOT over-ride the aria-checked="true" author declaration.

Be prepared for formal objections against this one. I'd file one myself if we speced it the way you've phrased it here. I think we can write a clause for the HTML 5 and other host languages to declare implicit strong semantics on some missing attributes, including checked and disabled.


> On May 9, 2014, at 7:38, Richard Schwerdtfeger <schwer@us.ibm.com> wrote:
> 
>> If required is absent then aria-required should be allowed to override it. We also need this for backward compatibility to older browsers which do not support HTML5. I don't want to break existing code because HTML5 decided to enforce a host language semantic where the author *chose to* override it. 

Rich and James N, what do you think about John's example above? <input type="checkbox" aria-checked="true">

This essentially means "only checked for screen reader and other AT users." This seems like a pretty terrible anti-pattern. IMO, even worse than the identical misuse of @required/@aria-required.

James

>> I agree that <input type="text" required aria-required="true:> is redundant but it harms nothing.
>> 
>> The spec. states that  <input type="text" required aria required="false"> is invalid. I agree. required does not exist in HTML4 and this conflicts with the host language semantics and there are no backward compatibility issues with this. This is already covered by the spec. 
>> 
>> So, the net is although HTML5 has some implied semantics wrt the absence of a boolean attribute we have backward compatibility issues we must continue to support. ... and at the end of the day I have NEVER seen this to be an issue.

Received on Friday, 9 May 2014 17:56:38 UTC