RE: OT Re: What is the expected behavior of scrollable divs within touch screen devices, and does ARIA apply?

Also, since the next question is likely ‘why’, I figured I would just circumvent it by answering this in advance.

 

Basically scrollable regions are already built into browsers and they work fine from the keyboard when you add tabindex=0 to the scrollable element. All of the visual rendering is handled by the browser as well when offscreen content is recognized and overflow is set using CSS.

 

The problem is that there is no textual equivalent for non-sighted AT users, to convey which area is scrollable or what the platform level commands are for interacting with it.

 

So, if you had an ARIA attribute, such as role=”scrollable”, you could define this functionality and give it a specific name that could be mapped in the accessibility tree via aria-label/aria-labelledby.

 

This would make it possible for platforms to recognize the role, and convey the proper key commands or gestures for AT users when the region receives focus, similar to the way that VoiceOver announces that an element marked up with role=”slider” is a slider, and to swipe up and down to move it.

 

Also, as James N mentioned about voice activation, this would make it possible to use the explicit name of a region to enable scrollability as well, such as saying “scroll viewer down” if aria-label or aria-labelledby referenced the name of the scrollable region as “viewer”, which could be tied into the platform.

 

Obviously there would need to be a way to convey this functionality, like listing scrollable regions so that the names would be recognized, but this too would be doable.

 

 

 

From: Bryan Garaventa [mailto:bryan.garaventa@ssbbartgroup.com] 
Sent: Sunday, April 27, 2014 9:37 AM
To: 'James Craig'; 'Matthew King'
Cc: 'W3C WAI Protocols & Formats'
Subject: RE: OT Re: What is the expected behavior of scrollable divs within touch screen devices, and does ARIA apply?

 

Thanks, I’m not really sure this is off topic though, because my original question still stands, regarding whether there would be benefit in standardizing a method for defining scrollable regions for ATs. I think this is needed, because there is currently no way to do this reliably.

 

This could be done using ARIA, such as creating a role that conveyed this type of functionality, which could be combined with aria-label to set an explicit label for the scrollable region.

 

Also, the compatibility differences we’ve been discussing are relevant for the ARIA support table we were discussing a couple of weeks back, which I’m still planning a template for. I should be able to send through a sample by the end of next week.

 

From: James Craig [mailto:jcraig@apple.com] 
Sent: Saturday, April 26, 2014 12:58 PM
To: Matthew King
Cc: Bryan Garaventa; W3C WAI Protocols & Formats
Subject: OT Re: What is the expected behavior of scrollable divs within touch screen devices, and does ARIA apply?

 

[This thread is going well off topic for ARIA standardization discussion, though it may be interesting to discuss some of the points on the IndieUI list.]

 

Bryan and Matt,

 

Did you file bug or enhancement reports as requested with any AT vendor? If so, privately send me the bug numbers for the Apple-related ones, but don’t forget to file relevant bugs against Android, WP, and FFOS, too.

 

It really highlights some of the downsides of a touch-based screen reader that renders all visual elements. When reading by item, you have to swipe through the entire tree. 

 

As for the screen reader usability of this example, much could be improved for the sighted user interface of this example as well. Desktop design patter paradigms based on keyboard and mouse interaction don’t always transfer well to other modalities, hence the relevance to IndieUI.

 

Be careful to not confuse existing behavior (on any platform or device) with a fundamental incompatibility or limitation. Remember that only a few years ago, most accessibility “experts” assumed touchscreen interfaces would always be completely inaccessible. Never underestimate the power of software and a creative mind.

 

Let’s get back to talking about ARIA 1.1.

 

Cheers,

James

 


On Apr 26, 2014, at 5:09 AM, Matthew King <mattking@us.ibm.com <mailto:mattking@us.ibm.com> > wrote:

On IOS 7.1, the aria-selected="true" is also not revealed. 

You can swipe through the content in the scrollable viewing pane, but even though the content reads, VO does not scroll it. so, if you touch outside the scrollable viewer, you have lost your place. 

If VO focus is inside the viewer, a three finger tap always says page 1 of 1; VO clearly does not recognize the scrollable div for what it is. 

Going off topic .... This is the first time I have looked at an ARIA tree view with VO in Safari. It really highlights some of the downsides of a touch-based screen reader that renders all visual elements. When reading by item, you have to swipe through the entire tree. Unlike VO in OSX or JAWS, you can't just skip past the tree with a single swipe (equivalent of pressing tab. While it is nice to touch anywhere and hear what is visually under your finger, e.g., one of the plus or minus signs for expanding and collapsing, it is not at all nice if you want to navigate by item. In fact, if the scrolled on for a long time, it could be a keyboard usability hell. 

I think it would be really nice if VO would somehow encapsulate composite web widgets like trees, listboxes, grids, and toolbars so there would be some single gesture manner for moving past it to the next element. I suppose the VO OSX interact and stop interacting concept could be used, but that would mean that touching the screen anywhere on that widget would have to speak the same thing instead of revealing what is immediately under the finger. I guess that would be OK; when interacting the standard touch to speak gesture could work. But then there would be the question of whether or not it would be desirable for the rest of the screen to not be interactive when inside the widget -- that could be both good and quite bad..... not an easy set of problems to think out. You wouldn't need any new gestures though -- double tap could start interaction and 2 finger scrub could stop it. Then, after you have made your tree selection, tapping anywhere on the tree would just announce the selected item, tree view, and the tree view name .... could be quite nice. 

Matt King
IBM Senior Technical Staff Member
I/T Chief Accessibility Strategist
IBM BT/CIO - Global Workforce and Web Process Enablement 
Phone: (503) 578-2329, Tie line: 731-7398
mattking@us.ibm.com <mailto:mattking@us.ibm.com>  



From:        "Bryan Garaventa" <bryan.garaventa@ssbbartgroup.com <mailto:bryan.garaventa@ssbbartgroup.com> > 
To:        "'James Craig'" <jcraig@apple.com <mailto:jcraig@apple.com> >, 
Cc:        "'W3C WAI Protocols & Formats'" <public-pfwg@w3.org <mailto:public-pfwg@w3.org> > 
Date:        04/25/2014 08:32 PM 
Subject:        RE: What is the expected behavior of scrollable divs within touch screen  devices, and does ARIA apply? 

  _____  




Alright, how about this one?
 <http://whatsock.com/tsg/Coding%20Arena/ARIA%20Trees/Tree%20(External%20XML)/> http://whatsock.com/tsg/Coding%20Arena/ARIA%20Trees/Tree%20(External%20XML)/
demo2.htm
Imagine an e-book reader, where, on the left, you have a scrollable index of
the chapters that you can scroll through and activate, and to the right of
that, you have a viewer pane where you can read specific chapters or
articles.

E.G The above demo uses an ARIA Tree construct as the index mechanism, and a
keyboard accessible scrollable div pane on the right for readers to view the
desired scene from Hamlet.

The ARIA Tree is surrounded by a scrollable div that includes role="region"
and aria-label="Hamlet, by William Shakespeare - Index", and the scrollable
reader pane on the right includes role="region" and aria-label="Viewer".

So, in theory, it should be possible to view this using a mobile device like
the iPhone/iPad in landscape mode, tap the Scene that you want to load after
expanding the desired Act, then move focus into the scrollable "Viewer" pane
on the right, and swipe up and down to review all of the content displayed
there from top to bottom.

With regard to VoiceOver, this appears to expose several bugs, the first
with ARIA Trees, where the expanded/collapsed state via aria-expanded within
branch nodes is not conveyed, nor is the level information declared via
aria-level, nor is the positioning information via
aria-setsize/aria-posinset.

Additionally, none of the region information is conveyed, such as the
explicit label for each region via role="region" + aria-label to convey its
purpose and where the boundaries are, nor is it possible to swipe up and
down to scroll the viewer content using VoiceOver, at least this isn't
working for me using the iPhone with iOS7.

If anybody would like to try this using an Android with TalkBack to confirm
functionality there too, that would be great.

All the best,
Bryan

-----Original Message-----
From: James Craig [ <mailto:jcraig@apple.com> mailto:jcraig@apple.com] 
Sent: Thursday, April 24, 2014 11:35 PM
To: Bryan Garaventa
Cc: W3C WAI Protocols & Formats
Subject: Re: What is the expected behavior of scrollable divs within touch
screen devices, and does ARIA apply?

On Apr 24, 2014, at 10:33 PM, Bryan Garaventa
<bryan.garaventa@ssbbartgroup.com <mailto:bryan.garaventa@ssbbartgroup.com> > wrote:

> I'm not really sure where this would be handled though, would it be with
the spec, or some spec, identifying a way for browsers to convey a textual
representation exposed by scrollable regions that can be used by ATs, or is
it simply a best practice for labeling and defining a region, and is it then
up to the AT to ensure support for this even though it's not specified
anywhere?

You could try coming up with an real world example where this is
necessary-make sure you don't make it an contrived test case-and submitting
it as a bug report to the various AT vendors. Like all software companies,
AT vendors have a long list of work they intend to complete, and the more
*real* you can make the example, the more likely it is to get a higher
priority fix order. 

James