Re: Content positioning in Canvas fallback content feedback requested

Ideally, it would be best to define a collection of drawing calls and bind
that to an elements fallback content. When the drawing object changes
position or dimensions it would be reflected in the accessible object bound
to the drawing object.

We think coords would work but it is really heavy weight.

Rich Schwerdtfeger
CTO Accessibility Software Group



From:	Richard Schwerdtfeger/Austin/IBM@IBMUS
To:	Charles Pritchard <chuck@jumis.com>
Cc:	public-html@w3.org, public-html-a11y@w3.org, chaals@opera.com
Date:	02/28/2011 01:18 PM
Subject:	Re: Content positioning in Canvas fallback content feedback
            requested
Sent by:	public-html-request@w3.org



Yes. All platform accessibility API provide a bounding rectangle for an
Accessible Object. That is why CSS positioning does not quite give us what
we want. Consequently, that is why I am suggesting we include allow
coordinates to be passed in the canvas subtree.

I would like to hear others thoughts on this. Shape is not essential.


Rich Schwerdtfeger
CTO Accessibility Software Group

Inactive hide details for Charles Pritchard ---02/25/2011 05:09:06 PM---On
2/25/2011 2:23 PM, Richard Schwerdtfeger wrote: >Charles Pritchard
---02/25/2011 05:09:06 PM---On 2/25/2011 2:23 PM, Richard Schwerdtfeger
wrote: >

From: Charles Pritchard <chuck@jumis.com>
To: Richard Schwerdtfeger/Austin/IBM@IBMUS
Cc: public-html@w3.org, public-html-a11y@w3.org
Date: 02/25/2011 05:09 PM
Subject: Re: Content positioning in Canvas fallback content feedback
requested



On 2/25/2011 2:23 PM, Richard Schwerdtfeger wrote:

      One of the features that platform accessibility API have is bounding
      rectangles associated with accessible objects. Screen magnifiers use
      this information to allow the user to zoom to an arbitrary, visual
      object in view. AI Squared has asked that we provide this feature for
      the fallback content such that fallback content objects can be
      positioned, with dimensions relative to where they are drawn on the
      physical <canvas>.

      There are a couple of ways we see this can be done:

      - apply shape and coords similar to image maps in fallback content
      - Parially, using CSS positioning.




Sounds like they want scrollIntoView and getClientRects to work:

https://developer.mozilla.org/en/DOM/element.scrollIntoView
https://developer.mozilla.org/en/DOM/element.getClientRects

Are those the methods they are currently looking at when they receive focus
events?
Do they set focus prior to zooming into particular objects?

...

We use CSS positioning to work with focus management on the last generation
of browsers,
and pass around getClientRects in our objects (so that we can use DOM or
simple JS objects).
SVG paths might work in current browsers. Some talk has been going on in
the lists about SVG a11y.

Partial CSS positioning might violate the notion that the fallback content
can be viewed independently of the canvas content.
Bounding rectangles would be too small to display text.

It's possible that dataTransfer operations might rely on
management of fallback content in the future.