W3C home > Mailing lists > Public > public-html-a11y@w3.org > August 2012

Re: Other problems, @hidden (ISSUE-204 aria-hidden)

From: Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>
Date: Wed, 15 Aug 2012 19:07:28 +0200
To: Steve Faulkner <faulkner.steve@gmail.com>
Cc: Benjamin Hawkes-Lewis <bhawkeslewis@googlemail.com>, Maciej Stachowiak <mjs@apple.com>, Janina Sajka <janina@rednote.net>, John Foliot <john@foliot.ca>, Sam Ruby <rubys@intertwingly.net>, public-html@w3.org, HTML Accessibility Task Force <public-html-a11y@w3.org>
Message-ID: <20120815190728383729.2e334714@xn--mlform-iua.no>
Steve Faulkner, Wed, 15 Aug 2012 13:45:59 +0100:
>> And my assumption was - and is - that this is because they are hard to
>> document - perhaps because there is nothing to document. It seems to me
>> that, at least for now, @hidden simply maps to style="display:none",
>> and nothing more. Do you have data for any other conclusion?
> 
> your assumption is incorrect, stuff has not yet been documented in 
> the most part because we have not got to it yet. 

When ATs read <p hidden style=display:block>Lorem</p> the same way as 
if you had removed @hidden, then it is impossible that @hidden is 
mapped to aria-hidden=true. So it shall be interesting to see how it 
maps - not to aria-hidden=true, at least.

>> We have heard that Firefox has started to implement what ISSUE-204 is
>> about. So what does Firefox do if the @aria-describedby points to a
>> hidden section which *also* has aria-hidden="true" set? I have not
>> tested that exact example, but based on the tests I have don, I don't
>> think it works. If so, then HTML5 is wrong - or at least not exact
>> enough.
> 
> works fine with aria-hidden+display:none or aria-hidden or 
> aria-hidden+hidden
> test page: https://dl.dropbox.com/u/377471/tests/firefox-describedby.html
> 
> the default action (showLongdesc) on the image is exposed regardless.

Please inform us about the AT and Firefox version you used! And please 
edit your test (see below)!

My results are different from what you have told us. In my results, the 
aria-hidden link does not get announced unless it, per CSS, is visible. 

My testpage: http://www.malform.no/testing/html5/hidden/

I tested with Firefox 14 (and Firefox Nightly) + Jaws 13 on XP and 
Windows 7, with my own variants of your code. (I also tried to test 
with Firefox Nightly for Mac with VoiceOver, but it turned out that 
that combination did not have any support aria-hidden=true what so ever 
- thus it would always work, you could say.)

So for the following example, then Firefox tells me that the page has 
no links. But when it presents the image, then it informs me that I can 
open a description link:

  <DOCTYPE html><title>test 1</title>
  <img aria-describedby=longdesc src=file alt=txt >
  <p aria-hidden=true><a id=longdesc href=link>link</a>

But for this example, which has @hidden, then it does not inform me 
about the link. (However pressing Enter when on the image may still 
open the link):

  <DOCTYPE html><title>test 2</title>
  <img aria-describedby=longdesc src=file alt=txt >
  <p hidden aria-hidden=true><a id=longdesc href=link>link</a>

Prediction: If you remove the first example (with the visible link) 
from your test page, then you will perceive it as if the link in the 
second example has stopped working. This is why: You should use 
different URLs in the two examples, so that you can verify that you 
opened the correct link. Because, when I ran your and mine tests, with 
more than one link on the page, then the link that opened was often not 
the one that I thought would open ...

My test pages follow those 'guidelines'. PS: I did not use <figure>, 
but just <p>, to avoid possible side effects.

The interesting issue here is that Firefox seems to to have implemented 
something for aria-hidden=true. But I can't spot that it has done 
anything with hidden.
-- 
Leif Halvard Silli
Received on Wednesday, 15 August 2012 17:08:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 15 August 2012 17:08:05 GMT