Re: Proposal: Provide better advice on constraint validation instructions

Steve Faulkner, Thu, 14 Mar 2013 16:45:22 +0000:
>>> I do hover not trust you on your browser testing. That is to say: You
>>> want to fix an interoperability problem, and then it does’t help that
>>> it works in an inspector if it doesn't work in browser.
>> 
>> ??? the inspector shows what the browser is exposing via the accessibility
>> API in this case the Axhelp property value.
> 
>> Let's say that description of the AX inspector is true: Still the AX
>> inspector could have bugs.
>> 
> 
> this was coupled with it being announced by voiceover as expected
> identified as a help text...

Excuse me, but how did you verify that VoiceOver announced the help 
text in question "as a help_ text"? (Versus as other kinds of text?)

Like I said: I have tested in the same version of Safari as you, on the 
same systems. And VoiceOver did indeed announce that help text. 
However, further analysis, including comparison with how it works in 
Webkit nightly and iCab, showed that it was only the normal source 
order that caused it to be announced. Thus: You can just remove the 
aria-describedby, and it will work just as the same way. (However, if 
you get a copy of Webkit where aria-describedby works, then the help 
text will be read twice to the user.)

To hopefully clear up exactly the issue above, it would be helpful if 
you could bother yourself to run the test that I advised you to take 3 
messages ago,[1]

[1] 
http://www.w3.org/mid/20130314003355646983.6180c28d@xn--mlform-iua.no

-- 
leif halvard silli

Received on Thursday, 14 March 2013 17:00:52 UTC