RE: Techniques for GL 4.2 L1 SC6

What I hear Becky saying is that the intent of this success criterion
falls wholly - not just partially - in the user agent's responsibility.
Some of the other comments in the wiki seem to be asking that too, once
I see things through Becky's lens.

I guess I would say that this is a functional requirement, that we would
hope user agents would handle, meaning no author responsibility, but we
would expect authors to do something in case of any UA problem. In WCAG
1.0 days there were known problems in this area, which is why it
occurred to us to create this SC. If there are no longer such problems,
so we can't think of any necessary authoring techniques, then we might
question its value. We could do a couple things:

* Assume present and future technologies will automatically fulfill the
intent of this, and / or that other specifications, such as UAAG, will
take on the requirement, and remove the SC.

* Keep the SC because we think it's important, mainly as a beacon for
technology developers, even though there are no authoring techniques for
it.

* Come up with specific authoring techniques, or ways to fail, that
demonstrate the need to keep this SC. I guess that was my job on taking
an action item for this and I haven't yet succeeded to everybody's
satisfaction.

With those three possible approaches on the table, does anyone have a
preference for which way we go, or data to support choosing one over the
other? It seems to me the weight of opinion is leaning towards the first
one unless someone makes a strong argument for one of the others.

Michael

-----Original Message-----
From: w3c-wai-gl-request@w3.org [mailto:w3c-wai-gl-request@w3.org] On
Behalf Of Becky_Gibson@notesdev.ibm.com
Sent: October 31, 2005 3:06 PM
To: w3c-wai-gl@w3.org
Subject: RE: Techniques for GL 4.2 L1 SC6



Regarding the techniques for GL 4.2 L1 SC6

I'm not sure what "Providing an event whenever a change to structure, 
selection, focus, attributes, values, state, or relationships occur. " 
means?  As a JavaScript programmer I generally respond to system events 
rather than providing them.  There is a DOM api,
dispatchEvent(someEvent) 
which can be sent to a particular node. Do other technologies such has 
Flash have the concept of providing an event? 

I can see using DHTML Accessibility to satisfy some of this success 
criterion but even then, I'm not sure how to deal with focus?   I can 
programmatically set focus to an object by calling object.focus(). But,
I 
have to first be using an object that can accept focus - either a form 
element or link or I have set the tabindex value on the object.   Is 
calling focus on an object considered programmatically determined since
it 
is in the script?    The user agent has to do something with that method

call (usually by generating a focus event).  As the programmer, I am 
relying on the user agent to generate the focus event for that object
and 
the AT to acknowledge that focus.  Is a success criteria about that
really 
necessary?  Maybe not for HTML but perhaps for other technologies? 


I have the same question about changing values?  In HTMLand JavaScript
I 
can programmatically change the value of an input type=text field. If I 
then set focus to the field, the AT should pick it up.  I'm not sure
there 
are really any specific techniques for this.  If the call to 
inputObject.focus() is in the JavaScript code then isn't it
automatically 
programmatically determined? 

Regarding the script technique:
Writing new content directly to the DOM  - while this is the preferred
way 
to add new content, there is no guarantee that the AT's will pick it up.

For example, if you add the new information at the top of the page and 
then do not set focus to it, the AT user may never know the information 
has been added. But, if focus is set the change in focus may be
unexpected 
and you could be violating GL 3.2 L1 SC1 or L3 SC2. 

Also, using document.write is listed as a common failure. While my
advice 
is to create objects via the DOM,  when done right, document.write can
be 
identified by AT.  I generally don't recommend document.write because it

is not useable in XHTML and thus code will have to eventually be
rewritten 
if migrating from HTML to XHTML. So, perhaps it can stay as a common 
failure but the reasons need to be identified.


I'm not at all sure I see the need for this success criterion.  Is the 
focus really more on future technologies?  It still seems like we are 
specifying requirements on the user agent rather than the web content.

my two cents,
-becky



Becky Gibson
Web Accessibility Architect
                                                       
IBM Emerging Internet Technologies
5 Technology Park Drive
Westford, MA 01886
Voice: 978 399-6101; t/l 333-6101
Email: gibsonb@us.ibm.com

Received on Monday, 31 October 2005 20:31:04 UTC