W3C home > Mailing lists > Public > wai-xtech@w3.org > September 2012

Re: suggest deprecating alertdialog role in ARIA 1.1

From: Greg Kraus <greg_kraus@ncsu.edu>
Date: Tue, 11 Sep 2012 14:07:56 -0400
Message-ID: <CAJL_eQBONF+bgNbb46sgLczggvEaHo-ZzOq7FOykw=cBc0h1kg@mail.gmail.com>
To: "Schnabel, Stefan" <stefan.schnabel@sap.com>
Cc: Alexander Surkov <surkov.alexander@gmail.com>, Joshue O Connor <joshue.oconnor@cfit.ie>, Steve Faulkner <faulkner.steve@gmail.com>, W3C WAI-XTECH <wai-xtech@w3.org>, Richard Schwerdtfeger <schwer@us.ibm.com>, Marco Zehe <marco.zehe@googlemail.com>, James Teh <jamie@nvaccess.org>
FWIW, in some recent testing I have done with the dialog and
alertdialog roles, I have noticed a difference in implementation in
JAWS in IE. JAWS in IE does not automatically announce values of text
input elements that are children of an element with role="dialog" when
the input element receives focus. This is true for pre-populated
values and values the user enters into the fields. If the input
elements are children of an element with role="alertdialog" everything
is read as expected. I think this is probably a bug in JAWS and I have
reported it to Freedom Scientific. Since the issue of the
appropriateness of alertdialog came up I just wanted to put this out

Here is a sample page demonstrating it.



Greg Kraus
University IT Accessibility Coordinator
NC State University

On Fri, Sep 7, 2012 at 7:47 AM, Schnabel, Stefan
<stefan.schnabel@sap.com> wrote:
> Alex,
> maps perfectly for me:
> additional" aria-dialogtype=Error/Warning/Info " ->  STATE_SYSTEM_ALERT_HIGH/MEDIUM/LOW
> I remember seeing those using Inspect32 with standard windows dialogs in the states enum listed, so I don't know what you mean with " but they weren't ever implemented."
> Stefan
> -----Original Message-----
> From: Alexander Surkov [mailto:surkov.alexander@gmail.com]
> Sent: Freitag, 7. September 2012 11:39
> To: Schnabel, Stefan
> Cc: Joshue O Connor; Steve Faulkner; W3C WAI-XTECH; Richard Schwerdtfeger; Marco Zehe; James Teh
> Subject: Re: suggest deprecating alertdialog role in ARIA 1.1
> As far as I know there's no difference. MSAA has states like
> STATE_SYSTEM_ALERT_LOW/MEDIUM/HIGH but they weren't ever implemented.
> I don't recall anything similar in other standards. Anyway many AT
> APIs are extensible so it shouldn't be a problem.
> Alex.
> On Fri, Sep 7, 2012 at 6:23 PM, Schnabel, Stefan
> <stefan.schnabel@sap.com> wrote:
>> Hi Josh, Alexander
>> good point having an additional" type=Error/Warning/Info " parameter in ARIA 1.1 for dialog.
>> But then we should go overboard and declare also modality (modal/non-modal dialog).
>> Regarding API mappings, how is the situation in Windows MSAA and let's say typed MessageBoxes ?
>> Isn't there a standard way to map these additional properties without borrowing from aria-label?
>> Regards
>> Stefan
>> -----Original Message-----
>> From: Joshue O Connor [mailto:joshue.oconnor@cfit.ie]
>> Sent: Freitag, 7. September 2012 11:16
>> To: Schnabel, Stefan
>> Cc: Steve Faulkner; W3C WAI-XTECH; Richard Schwerdtfeger; Marco Zehe; James Teh
>> Subject: Re: suggest deprecating alertdialog role in ARIA 1.1
>> Hi Stefan,
>>> It is not the fault of the ARIA spec that alertdialog is poorly supported in AT.
>> I completely agree but devs need practical solutions right now.. not
>> around the corner and to my mind anything that makes the practical
>> accessible web development easier - is a good thing.
>>> Role "alertdialog" has been invented to indicate simple ok/yes/no/cancel messageboxes in contrast to more
>>> elaborated complex dialogs with role "dialog".
>>> Therefore, I recommend *NOT* to remove the role entirely, in contrary, it should even be extended to reflect the
>>> *type* of "alert" dialog Error/Warning/Info etc. and its usage according to this should be reflected in the AP Guidelines documentation.
>> Not a bad idea. However, I suggest removing it and actually using the
>> role="dialog" as a generic role and allowing it to take the argument
>> dialog Error/Warning/Info etc.
>> Cheers
>> Josh
Received on Tuesday, 11 September 2012 18:09:10 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:51:46 UTC