progress bars, live region?

[David Bolter]

In dijit we implemented progress bars such that the element that held
the aria-valuenow was in the tab order (tabindex='0')... so that
screen readers would work.

Is live region support there for this now? Should we implement, for
example, progress bars as live regions? I'm wondering, since I want to
tackle jquery's progress bar, and am unsure which route to take.

Thoughts?

[David Bolter]

Perhaps a better question is: are widgets ever good candidates for
aria live region markup? I'm sort of embarrassed I don't know the
answer. I'll go through the docs again unless someone answers this
first...

D

[Aaron Leventhal]

You shouldn't use either live region markup or tabindex for a
progressbar. They're never focusable normally and shouldn't be on the
web either.
Also, just the fact that they are a progressbar with a value is enough
semantics to do the right thing.
Just use:
<span role="progressbar" aria-valuemin="0" aria-valuemax="100"
aria-valuenow="[xxx]">Progress bar stuff here</span>

- Aaron

[David Bolter]

Thanks Aaron, I was hoping you would answer :)

How does a screen reader user discover them?

D

[James Teh]

On 21/10/2008 5:22 AM, David Bolter wrote:
> How does a screen reader user discover them?

NVDA reacts to value change events on progress bars. Do value change
events get fired correctly for progress bars when the aria value changes?

NVDA is also an interesting use case because we generally beep on
progress bar updates rather than speaking them, although speaking is an
option. Thus, a progress bar as a live region would be a very bad thing
for most NVDA users.

Jamie

--
James Teh
Email: ja...@jantrid.net
WWW: http://www.jantrid.net/
MSN Messenger: ja...@jantrid.net
Jabber: jt...@jabber.org
Yahoo: jcs_teh

[Aaron Leventhal]

Hi David, from the value change events.

- Aaron

[David Bolter]

Thanks James (and Aaron).

I'm so pleased every time I hear of NVDA's features. You guys are
doing an awesome job.

D

> Jabber: j...@jabber.org
> Yahoo: jcs_teh

[Pratik Patel]

-----Original Message-----
From: free...@googlegroups.com [mailto:free...@googlegroups.com] On
Behalf Of James Teh
Sent: Monday, October 20, 2008 4:44 PM
To: free...@googlegroups.com
Subject: Re: progress bars, live region?

NVDA is also an interesting use case because we generally beep on
progress bar updates rather than speaking them, although speaking is an
option. Thus, a progress bar as a live region would be a very bad thing
for most NVDA users.

From the usability perspective, I think the progress tone is a better
implementation than the incremental announcement. Progressbar as a live
region would be a very bad thing for most screen reader users.

Pratik

[David Bolter]

Thanks Pratik.

I agree that a sonifying a progress bar is the way to go. What does
NVDA do with an indeterminate progress bar? There are some interesting
examples of this... like when there is a little region of the screen
that says "loading...", or "sending...". They seem to me to be sort of
like live regions, but are sort of like indeterminate progress bars.

I'll go with progress bar, thanks everyone!

cheers,
David

[James Teh]

On 21/10/2008 11:56 PM, David Bolter wrote:
> What does
> NVDA do with an indeterminate progress bar? There are some interesting
> examples of this... like when there is a little region of the screen
> that says "loading...", or "sending...".

Are these actually classed as progress bars? I thought a progress bar
was always a moving bar, but I guess there are some that just have an
indeterminate animation which indicates they are working. NVDA doesn't
currently handle this; we always assume a progress bar will give us a
percentage value. Are there use cases for this?

> They seem to me to be sort of
> like live regions, but are sort of like indeterminate progress bars.

Live regions would certainly be one way to do this.

--
James Teh
Email/MSN Messenger/Jabber: ja...@jantrid.net
Web site: http://www.jantrid.net/

[Aaron Leventhal]

Yes. inderterminate progress bars are common. A typical example would be
when you can't yet know the size of the data being loaded over the
network You just know that progress is being made. Perhaps after
connection and data starts to transfer, the size is determined and it's
no longer indeterminate.

We need to handle this case. It's part of the ARIA spec that
aria-valuenow is not required for a progress meter. It will be missing
if the progress is indeterminate. Live regions aren't the intended way
to do it, so if the ARIA spec doesn't currently do a good job here we
need to look at that.

- Aaron

[David Bolter]

Hi James,

There are examples of indeterminate progress bars here:
http://archive.dojotoolkit.org/nightly/dojotoolkit/dijit/tests/test_ProgressBar.html

As Aaron says, the lack of a value-now signifies indeterminate-ness
(for now).

cheers,
David

[Scott González]

A few more examples: http://jqueryui.pbwiki.com/ProgressIndicators

I think it makes sense to not use live regions since it's possible (and common) for indeterminate progress indicators to be used as loading screens for content that will be completely static after it loads.