Re: Comment on WAI-ARIA Role: Mandatory 'describedby' Too Restrictive

Doug,

Briefly:

We have been working on amending the host-language-implementation
section to make it clear that host languages can define implicit ARIA
semantics based on their markup.  You do have to convince your  
processors
to pick these up.  This would mean that the nesting-implied relationship
based on your specified algorithm would be an implicit @aria-describedby
in this case.

In the absence of an explicit @aria-describedby the implicit  
relationship
would be binding on the accessibility-API mappings.  But the author  
could
override the implicit information with explicit ARIA markup.

This will show up in the public editor's draft some time after Michael
Cooper returns from a well-deserved holiday on Monday.

Less briefly:

In particular, I would expect that svg:desc should acquire the
an implied @aria-describedby
emanating from the containing <g> to the <desc> in question.
In which the 'description' role is possible but does not add
anything operationally.

In the case of svg:title, are you anticipating that some would
show tooltip behavior and some not?  Will this be controlled by
scripting?  I would assume that svg:title is appropriate to
get an implicit @aria-labelledby from the container.

And yes, we are aware that there is implicit semantics if the
control is indeed an html:input and the author has used <label>
to label it.

[but still hardly exhaustively...]

Al

On 22 Aug 2008, at 5:58 PM, Doug Schepers wrote:

>
> Hi, ARIA folks-
>
> The SVG WG is discussing the use of a few specific roles for SVG
> elements, 'title' and 'desc'.  We think it would be very useful and
> appropriate to allow the role values 'description' and  
> 'tooltip' (among
> others) on those elements.  However, both of those role values  
> require that:
> [[
>   Objects with this role must be referenced through the use of a
> describedby.
> ]]
>
> In SVG, we are defining an algorithm that describes the relationship
> between the 'title' or 'desc' element and the element or elements they
> are describing, based on a parent-child (or more precisely,
> container-contained) relationship.  Basically, for most uses, the
> 'title' and 'desc' elements describe their parent element (which  
> may in
> turn simply be a group of elements, in which case the general
> description applies to the collection of child elements).
>
> Therefore, the 'describedby' role value is superfluous and possibly
> contradictory to the well-structured solution of SVG; similarly, it is
> redundant with the HTML 'label' element, with its 'for' attribute.   
> For
> both these cases, the 'describedby' role may (and should) be  
> inferred by
> the UA.
>
> Further, requiring this reference may discourage use of roles in these
> circumstances, by introducing arbitrary and unnecessary extra work on
> the part of the author.
>
> Therefore, our suggestion is that you remove this mandatory stricture,
> or soften it to a SHOULD or MAY, or describe a way in which a language
> may describe an implicit value for the 'describedby' role based on the
> host language's semantics.  You do allude to this last option in
> WAI-ARIA: [1]
>
> [[
> Build relationships
>
> ...
>
> Some relationships are determined automatically from the host  
> language,
> such as by using the label tag in HTML.
> ]]
>
> However, you don't seem to describe the manner in which a language
> defines this, and the MUST statement in the roles seems to contradict
> the "Build relationships" wording.
>
>
> [1] http://www.w3.org/TR/wai-aria/#buildingaccessibleapplications
>
> Regards-
> -Doug Schepers
> W3C Team Contact, WebApps, SVG, and CDF
>

Received on Saturday, 23 August 2008 01:32:09 UTC