W3C home > Mailing lists > Public > public-html@w3.org > April 2011

Re: False aria-describedby expectations in ARIA Authoring Practices (longdesc)

From: Steve Faulkner <faulkner.steve@gmail.com>
Date: Fri, 22 Apr 2011 20:13:42 +0100
Message-ID: <BANLkTimHPQ4-Jp64L9BkbspHMbMwSaQxMw@mail.gmail.com>
To: Jonas Sicking <jonas@sicking.cc>
Cc: Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>, Michael Cooper <cooper@w3.org>, Joseph Scheuhammer <clown@alum.mit.edu>, W3C WAI-XTECH <wai-xtech@w3.org>, HTMLwg <public-html@w3.org>, Richard Schwerdtfeger <schwer@us.ibm.com>, David Bolter <dbolter@mozilla.com>
hi jonas,
adding rich and d bolter as they may have some useful input.

I suppose it is not so much that people want to restrict it, but how
all browsers (as in firefox etc) have implemented describedby.

i.e describedby doesn't work the way it is wirtten in the author
guide. the way it is implemented is that the text content of any
elements referenced are used as the value for the acc description
property in accessibility APIs this is a property that can only take
plain text (as far as i know) and the content of which can be accssed
by AT via the API is just that plain text.

the emerging consensus on providing an aria feature like longdesc (for
example) is that it will need to be a new feature, not a modified
describedby.

regards
stevef


On 22 April 2011 19:55, Jonas Sicking <jonas@sicking.cc> wrote:
> On Thu, Apr 21, 2011 at 7:26 PM, Leif Halvard Silli
> <xn--mlform-iua@xn--mlform-iua.no> wrote:
>> Michael and Joseph,
>>
>> Where do we file bugs against your 'WAI-ARIA 1.0 Authoring Practices'
>> editors draft?
>>
>> There is a much erroneous and misleading description of the
>> possibilities of aria-describedby is in the draft. The draft states
>> that it is possible to "follow" what you refer to as an
>> aria-describedby "arc" from an img element to an anchor element - et
>> cetera: [1]
>>
>> ]] ... different from the HTML longdesc attribute ... But if you wish
>> to reference an external resource with aria-describedby, you can
>> reference a link that in turn references the external resource. This
>> requires the user to follow two steps, first following the
>> aria-describedby arc, then following the link, but does address the use
>> case. [[
>>
>> However, this is an incorrect description of the facts:
>> aria-describedby doesn't allow you to "follow" the aria-describedby as
>> if it was a link. Besides that it is technically wrong, the example you
>> show leads to repetition for the user: first the textual content of
>> that anchor element is read to the user when he/she reads the img
>> element (without being told that it is a link nor be given opportunity
>> to link to it). And when the screenreader user reaches that link,
>> he/she gets to listen to the same link text again.
>>
>> Your claims in the draft is already being repeated in the wild. [2]
>> Which seems entirely useless: In the HTMLwg, we've established (for the
>> last time) one month before your last edition of the draft that
>> aria-describedby does not have any such functionality. (I provide two
>> of several possible links. [3][4]) The only thing that happens if one
>> points aria-describedby to an anchor element is that the user is served
>> the textual content of that link, without being whether taken to that
>> link or being told that he/she listens to the text of the link.
>>
>> Please remove those false claims ASAP.
>>
>> [1] http://www.w3.org/WAI/PF/aria-practices/#Descriptions_external
>> [2] http://rebuildingtheweb.com/en/longdesc-replacement/#c20110420054615
>> [3] http://www.w3.org/Bugs/Public/show_bug.cgi?id=12243
>> [4] http://lists.w3.org/Archives/Public/public-html-a11y/2011Mar/0166
>
> Actually, I don't think we have established that ARIA can't be used in
> this way. Back when I wrote my comment in bug 12243 I had taken at
> face value several comments on this list the claim that
> aria-describedby only represents a reference to the textual contents
> of whatever elements it points to.
>
> However once I looked at the spec I found no such requirements of
> limiting the reference to the textual contents. It seems to me that
> ARIA simply says that a description of the element can be found in the
> referenced elements. I see neither any reason nor requirements to
> remove the semantics of those referenced elements when reading the
> contents to the user.
>
> I've since asked for why people feel like there is a need to restrict
> to textual contents when following aria-describedby, but so far have
> not received an answer. So consider this me asking again :)
>
> / Jonas
>
>



-- 
with regards

Steve Faulkner
Technical Director - TPG

www.paciellogroup.com | www.HTML5accessibility.com |
www.twitter.com/stevefaulkner
HTML5: Techniques for providing useful text alternatives -
dev.w3.org/html5/alt-techniques/
Web Accessibility Toolbar - www.paciellogroup.com/resources/wat-ie-about.html
Received on Friday, 22 April 2011 19:14:30 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 May 2012 00:17:28 GMT