Re: ISSUE-636 CTION-1398 Provide spec. text for aria-roledescription

Hi all,

On 2014-03-31, 10:28 PM, James Craig wrote:
> On Mar 31, 2014, at 6:39 PM, Shane McCarron <shane@aptest.com> wrote:
> 
>> On Mon, Mar 31, 2014 at 8:28 PM, Steve Faulkner <faulkner.steve@gmail.com> wrote:
>>> I think there needs to be some form of registry for aria-roledescription  values, otherwise its going to be the wild west. 
> 
> The Web is already the wild west.
> 
>> Steve,   I disagree.  The whole point of having vocabulary spaces is to avoid the need for registration.  Within a vocabulary space, the maintainer(s) can manage their own collection of terms.  We don't need to be gatekeepers here.  In particular since the values for the attribute don't have any semantic meaning to an AT.
> 
> At least two platform APIs already support this feature: AX API as AXRoleDescription and MSAA/UIA as localizedControlType. Neither of those APIs has imploded due to the ability to use localized role descriptions. Custom role descriptions are used successfully in various places throughout the Windows and Mac OS X systems.
> 
> Granted, platform application development has a slightly higher bar to entry than the Web, so there is a possibility that this will be overused in places by new or inexperienced web developers, but I don't think that diminishes the need for it, and I don't think that means we should limit it to a registry. Opening it to free form text is liberating for two reasons: 1) We are willingly admitting that a formal role taxonomy will never be able to match pace with the Web's expansion, but 2) that is a good thing. It frees us to see how these custom role descriptions will be used in the wild, and allows us to expand future versions of ARIA to cover the most useful of those needs.
> 
> As an example let's discuss the following EPUB markup we've previously discussed. 
> 
> <div role="epub-chapter group" aria-roledescription="chapter" id="intro" aria-labelledby="c1h1">
>         <h1 id="c1h1">Introduction</h1>
>         <!-- remaining chapter contents -->
> </div>
> 
> This is already backwards compatible with ARIA 1.0. Many screen readers would currently speak the element as "Introduction, group."
> 
> • The label is "Introduction"
> • The (undefined) aria-roledescription attribute is ignored.
> • The role value "epub-chapter group" is parsed and determined to have a computed role of "group" b/c the first token "epub-chapter" is not a valid ARIA role.
> 
> Some benefits to the the new attribute in lnclude the following:
> 
> 1. It can retain use of a known role (group) which is recognized by user agents, APIs, and assistive technology.
> 2. It can be spoken as "Introduction, chapter" instead of "Introduction, group" solving an author and user need.
> 3. It allows us (Members of PFWG/ARIA) to see how this is being used "in-the-wild", and determine the benefit of adding additional roles that may provide additional benefit in future versions of ARIA. Some uses may be obvious and make it into a spec quickly. Some uses may be so obscure or uncommon that they never make it into a spec. In either case, the author's need is met and the user is served something more usable than the generic "group" role description.
> 
> Eventually I expect we'll bring in many of the EPUB roles, so the example above would change over time:
> 
> From ARIA 1.0:
> 	role="epub-chapter group"
> 
> To ARIA 1.1, once aria-roledescription (tbd) is added:
> 	role="epub-chapter group" aria-roledescription="chapter"
> 
> To a later version, once the chapter role is added, but before all browsers support it completely:
> 	role="chapter group" aria-roledescription="chapter"
> 
> And finally, once all modern browsers support the new role consistently:
> 	role="chapter"
> 

I want to hear from web devs.

As usual I'm worried about complexity. We would present web devs with
these ways of describing an object:

alt
title
aria-describedby
aria-label
aria-labelledby
aria-roledescription

Also note possible additional complexity with the naming and reporting
algorithms. If this attribute is only for role group and region does it
really meet the bar? Why wouldn't aria-label work?

Note I'm all for aria extensibility mechanisms and I just want to make
sure we agree this particular attribute is the right/best way in the
long run. Sorry I can't keep up very well with the meetings and emails,
is the extensibility work happening somwhere?

Cheers,
David

Received on Tuesday, 1 April 2014 15:58:21 UTC