W3C home > Mailing lists > Public > wai-xtech@w3.org > July 2007

Re: conflation of issues or convergence of interests?

From: Jason White <jasonw@ariel.its.unimelb.edu.au>
Date: Sat, 28 Jul 2007 20:38:49 +1000
To: wai-xtech@w3.org
Message-ID: <20070728103849.GA6241@jdc.local>

On Sat, Jul 28, 2007 at 05:30:25PM +1000, Lachlan Hunt wrote:
>
> I don't understand that reason.  In particular, "while presenting only a 
> single link to the user interface" doesn't make sense.  Where and how are 
> you imagining the UA to present this single link?

I am imagining that the UA presents a link to the video, traversal of which,
if video playback is enabled, results in retrieval and rendering of the video
resource. If playback is not enabled, or, equivalently, alternative renderings
are enabled, then the alternative content is presented to the user. This
avoids specifying two links in the content itself, the relation between which
(namely that the one is an alternative to the other) may be ambiguous,
depending on whether the HTML document is well or badly designed.

An explicit association removes this ambiguity.
>
> As I see it, the UA is still going to render the whole page.  If a user 
> configures their UA to not download videos by default, then it doesn't have 
> to.  The user can then read the page, and when they reach the link to the 
> alternative content, they can download it.

I would separate case 1 (automatic retrieval) from case 2 (manual retrieval).
In the first case, with an explicit association, the UA can retrieve and
insert the alternative content at the appropriate point in the document, or
otherwise make it available through a user interface mechanism. In the second
case, the UA can provide a single retrieval option which obtains either the
video or its alternative according to the delivery context.
>
>> It would allow testing tools to check that there is an alternative 
>> associated
>> with the video. This is important for the purpose of verifying conformance 
>> to
>> site-wide or organizational policies, WCAG specifications, etc.
>
> There is no requirement for all accessibility features to be testable using 
> automated accessibility testing.  Such a requirement would only place 
> arbitrary restrictions upon the development of accessibility features, 
> limited by the abilities of testing tools.  Accessibility features should 
> be evaluated according to how well they benefit users, not how easily a 
> tool can test for it.

While I agree with this, it is not a good argument against an explicit
association. All that my position requires is that automated testing be
valuable and desirable, where possible. Adding semantics to a markup language
that support, among other things, accessibility-related requirements, has the
additional benefit of making the requirement more amenable to automated
verification.

The larger the Web site, the more important such evaluation becomes in
practice.
>
>> The explicit association would also open up the possibility of special 
>> treatment, in the user interface, of links to alternative content, such as 
>> aural highlighting or access keys provided by the user agent, without 
>> leaving these to be implemented by the content author as an implicit 
>> association would do.
>
> Possibly, but that would need research to determine things like whether 
> users actually want or need that feature, whether or not they would really 
> use it if it were available, and whether UAs, especially assistive 
> technology, could or would implement it.

No, it wouldn't require such research. If there are good a priori reasons to
think that such features would be useful, then providing the necessary
semantics in a widely used markup language is likely to motivate UA and
assistive technology developers to take advantage of the possibility thereby
opened up.

>
> BTW, there was a suggestion for <a rel="longdesc"> a while ago, which, when 
> used within <figure> and <legend>, is a possible solution.

I would favour a general solution such as this to a specific solution for each
affected element. This also simplifies implementation.
> No, the user is the whole point.  If it can be conclusively shown that the 
> user is somehow disadvantaged because their UA doesn't provide some special 
> functionality, then we can look at addressing those problems for the user.  
> But simply assuming that the UA needs to do something special and then 
> jumping to the conclusion that we need to provide for that doesn't help.

To the contrary, if it can be shown that the user (or the evaluator/author)
can be significantly advantaged by implementation of particular features made
possible through the inclusion of proper semantics in the markup language,
then this is a good ground for including those semantics when designing a
language feature. The implication regarding jumping to conclusions, applied to
my arguments, is false, of course.
> Only one of the reasons you gave was a possibility, and even that's not yet 
> conclusive.  Therefore you can't yet claim that implicit association is 
> inadequate.  Likewise, I can't conclusively claim that it is, but that 
> would still be my hypothesis.

I'm unconvinced by any of the supposed counter-arguments.

Other features of HTML that provide for textual or other alternatives make
the association explicit, as in the content model of the OBJECT element, or
indeed the ALT attributes of IMG/AREA. For consistency, the same should be
true in the video case, quite apart from the other considerations presented
above.
Received on Saturday, 28 July 2007 10:39:23 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 13:15:43 GMT