Re: [NVDA] #2390: NVDA doesn't handle role=dialog with supporting ARIA attributes according to spec

I understand and agree with your statement

"Assistive technologies like screen readers differentiate themselves by 
their behavior, for better or worse, and it's outside the scope of the W3C 
to dictate how they work."

The thing to keep in mind is that if such differences become overly 
pronounced, nobody wins, not the developers, not the screen reader users, 
and not the AT manufacturers. It's in everyone's interest to reach an 
agreement about fundamental behaviors such as these while we still have time 
to do so.

----- Original Message ----- 
From: "James Craig" <jcraig@apple.com>
To: "Bryan Garaventa" <bryan.garaventa@whatsock.com>
Cc: "James Nurthen" <james@nurthen.com>; <w3c-wai-ig@w3.org>; "James Teh" 
<jamie@nvaccess.org>
Sent: Wednesday, May 30, 2012 11:01 PM
Subject: Re: [NVDA] #2390: NVDA doesn't handle role=dialog with supporting 
ARIA attributes according to spec


Copying Jamie.

On May 30, 2012, at 10:27 PM, Bryan Garaventa <bryan.garaventa@whatsock.com> 
wrote:

> I understand your point, but this isn't what I'm referring to. I'll 
> forward the message I sent to the ARIA group for clarity.

For what it's worth, I agree with your point that NVDA should trust the 
author to define dialog descriptions where needed. I think you're right that 
it's not implemented according to the intention of the spec, as it's 
treating dialog description the same way as alertdialog…

That said, I also see Jamie's point about trying to do what's best for his 
users, and the ARIA spec only makes RFC-2119 "MUST" and "MUST NOT" 
requirements for Authors and User Agents, not for Assistive Technologies, so 
there is no place in the spec that it says AT must not do this… Even the 
statement in alertdialog isn't a normative requirement for AT; it's an 
normative author requirement followed by an informative statement about what 
assistive technologies can or might do. We have a lot of SHOULD and MAY 
statements in the spec for AT, but no MUST statements.

Assistive technologies like screen readers differentiate themselves by their 
behavior, for better or worse, and it's outside the scope of the W3C to 
dictate how they work. The closest thing ARIA and the ARIA-UAIG can do is 
define how authors code it, how user agents expose it to APIs (even this is 
debatable at times), and make recommendations to screen readers and other 
assistive technology.

I think your best bet with this "enhancement request" is to win Jamie and 
Mick over with persuasion rather than by telling them it's a bug. You may be 
able to find a good compromise.

Good luck,
James

Received on Thursday, 31 May 2012 18:56:23 UTC