Decision on aria-labelledby/role=presentation

I'm OK with chair's WG Decision on aria-labelledby and 
role=presentation, and don't see need to press for reopening. Agree 
with what Steve said on subgroup meeting today:

>> SF: role="presentation" is in a11y layer; alt="" will be represented 
>> differently in view where images disabled -- if role="presentation" 
>> won't treat same way -- need to treat null alt as role="presentation"

When I read the starter draft letter however, 
(http://www.w3.org/mid/E1QEO8R-0006yO-C2@lisa.w3.org) it prompted me to 
write this:

Re: "The presence of role=presentation does not make missing alt
     conforming."

My reading: 
 * role=presentation doesn't permit author to not omit a
   (supposedly) empty @alt
 * does not (need to) mean that role=presentation in combination
   with non-empty alt is forbidden. Which is OK - because the
   alt text could be decorative  - or perhaps the alt text 
   will be used as a label by some element that links to the
   image via aria-labelledby.
Important:
 * HTML5 encourage browsers to repair for lack of @alt - and 
   typically they use "Image." as repair text. Thus what is 
   suggested in the starter drarft letter means that browser
   will add "Image." for non-AT users = not good.
 * Thus Richard's statement in today's sub group minutes is not 
   something I can subscribe too:
   ]] @alt with role="presentation" eliminates need for alt="" [[
   Clearly he is right, if we see it squarely from the POV of
   a ARIA enabled AT user. I do support the view that it is
   good to encourage to use of role=presentation, though. 
   But it doesn't need a carrot in the form of "you can drop
   the empty alt when role=presentation is present."
   

Re: 'The presence of aria-labelledby does not make missing alt
     conforming.'
  * Seemed like the sub group decided to drop its disagreement.
    Nevertheless, I'll state this:
Important issues:
  * For aria-labelledby, it is the same issue for role=presentation: 
omitting @alt will make non-AT repair for the lack of @alt. HTML5 says 
that one should not rely on the repair that browsers perform for the 
lack of @alt. To not do alt repair when aria-labelledby is present, is 
not an option: the sighted needs to know that there is an image there - 
after all, the AT user is told that there is an image, and so should 
the sighted user who uses the Yahoo e-mail service with image display 
off.
  * Note also that even if at least VoiceOver doesn't support it, the 
ARIA 1.0 spec still says that if aria-labelledby is present, then @alt 
should not be presented to the AT user unless @aria-labelledby also 
includes the img element itself: 

   ]] User agents give precedence to aria-labelledby over 
      aria-label when computing the accessible name property.[[
      http://www.w3.org/TR/wai-aria/complete#aria-labelledby

and

   ]] The aria-labelledby attribute takes precedence as the 
      element's text alternative [ snip ] However, the element's
      aria-labelledby attribute can reference the element's own 
      IDREF in order to concatentate a string provided by the 
      element's aria-label attribute or another feature lower in
      this preference list.[[
      http://www.w3.org/TR/wai-aria/complete#namecalculation
      (And @alt is lower on the feature list.)

  * Thus, from a AT point of view, the use of aria-labelledby should 
hide the alt text.  
  * What perhaps is needed is authoring advice for images with 
@aria-labelledby: perhaps the author can get away with a really one 
word alt text such as "photo" or something like that.
-- 
Leif Halvard Silli

Received on Tuesday, 26 April 2011 01:15:32 UTC