Re: Moving longdesc forward: Recap, updates, consensus

On Fri, 06 May 2011 22:38:46 +0200, John Foliot <>  

> Benjamin Hawkes-Lewis wrote:
>> I'm wary of constraining the user agent conforming classes to do
>> *anything* with the semantics we are defining in HTML5 ...
> While I can accept that statement on face value, I am not sure how you  
> would see this manifest.... Having spent some time talking with a few  
> engineers who have an
> accessibility bent to their work (including the 2 core developers of
> NVDA) I see a consensus pattern emerging around the contextual menu
> as the means to interact with the longdesc attribute value. Could you
> elaborate more please?

I agree with Ben. Having got the context-menu approach implemented in  
Opera, and then taken a very different approach with TellMeMore I am very  
conscious that what seems like a reasonable plan might not hold up to the  
requirements of the near future, or might be overturned very quickly.

That said, a simple statement "you MUST provide access to the thing the  
href points to" gives at least a minimal requirement. I've often seen the  
argument (from people in this discussion - I have used it myself although  
I think it's fundamentally a failure to have a sensible answer) "it's in  
the DOM and that is the only formal requirement to claim implementation",  
and the like. Something that gives the minimal basis to reject such an  
argument on common sense seems reasonable. Anything more about UI seems  
overly constraining in the HTML spec.

>> I think UAAG 2.0, not HTML5, is the right place to specify
>> accessibility requirements for browsers, such as universal access to
>> text alternatives:
>> alternative-content


>> Irrespective of whether we propose "SHOULD" or "MUST", I think it might
>> be a good idea to provide spec text concisely explain how text
>> alternatives, short and long, can be useful to different user groups.
>> (Otherwise we may end up with implementations optimized for one user
>> group such as users who are blind.)
> I agree with you here.

plus two.

>> If we do shift to "MUST", I think we should talk about "interactive
>> user agents" rather than just "user agents". For a discussion of this
>> distinction, see the conformance classes section of the spec:

Hmm. I'm not so sure... lemme think about that a bit longer.


Charles McCathieNevile  Opera Software, Standards Group
     je parle français -- hablo español -- jeg lærer norsk       Try Opera:

Received on Monday, 9 May 2011 13:52:18 UTC