Re: Canvas accessibility draft proposal

Instead of coming up with a third method for alternate media selection, you should try rolling this in line with David Singer's accessible media query proposal, which has initial acceptance from the CSS working group, or Sylvia Pfeiffer's itext proposal, which is being discussed by the HTML 5 working group. 

Portions of this proposal are vague or subjective, and most of the language is not spec ready. For example, anytime you use RFC-2119 keywords (MAY, SHOULD, MUST, SHOULD NOT, MUST NOT, etc.) you should always phrase the requirements in active voice with a clear subject and action. In other words, don't just specify WHAT is required, but also WHO is required to do it, and WHEN. Much of the language below is vague on all three points.

I think all of us have slightly different ideas of what we mean by the "shadow DOM." Instead of concentrating your work on a spec proposal that includes media selection and details about bounding boxes (which may not even be necessary), try to explain in plain language your understanding of how the "shadow DOM" focus model is supposed to work. Before anything is put to spec, we should all have the same basic understanding of how that's intended to work. It's clear to me that no one agrees on that yet so, while I appreciate and admire your eagerness to get something in writing, I feel the result is premature.



On Nov 17, 2009, at 8:30 AM, Richard Schwerdtfeger wrote:

> So, I am listening to all this and would like to make a draft proposal for the HTML working group meeting this week (Paul Cotton made a request):
> 
> 1. Modify the canvas element to support aria-activedescendant where the id refers to a valid id in a shadow DOM.
> 2. The canvas element shall support zero or more child access elements representative of alternative views. The access element shall have a mode attribute. Whose values are based on a subset of the IMS Global Consortium Access for All Meta Data V3 default. The default mode would be representative of the shadow DOM. In many instances a more usable solution may require an alternative.
> 3. The mode content attribute values for access would be: 
> - default - representing the shadow DOM for the complex visual rendering of <canvas>
> - textual - Here we could place a long description or an alternative accessible UI renderable using HTML 5 standard markup - including ARIA.
> - auditory - an audio alternative. Some users may prefer this
> - visual - This is really an alternative for audio but it would address things like signLanguage
> 
> Note: access for all also supports "tactile" and "olfactory". I would recommend ignoring those for now. These could added in future releases of HTML. 
> 
> 4. The access element would include a type attribute for which we should take a subset of the list (from AccessForAll): audioDescription, caption, e-book, signLanguage, highContrast, transcript, alternativeText, longDescription, haptic}
> 
> The e-book may be rendered with a plug-in. We can choose a subset of these if necessary. At this point this is for discussion purposes.
> 
> 5. View Selection should be based on a user preference in the browser based on the mode and type preference supported in HTML 5. If nothing is set, the shadow DOM is used for the accessibility mapping when provided. The user agent should select the first view that fits the user's request. Platform accessibility API may also be used to configure the view selection. Note: is no alternative view is provided to meet the preferences the view reverts back to the canvas + shadow DOM view. The shadow DOM is also optional.
> 
> 6. The canvas tag shall include a script method to set the bounding rectangle for a given valid id in the shadow DOM. This would allow
> the browser to provide the bounding rectangle for the id in the shadow DOM when it receives focus. The rectangle should be relative to the location of the canvas element and would be converted to screen coordinates in the accessibility API and would allow the canvas to move without changing the location of the shadow DOM element. This will allow screen magnifiers to determine the visible focus. (Ideally it would be nice to also do this via XPath statements vs. id)
> 7. All allowable HTML elements in the shadow DOM shall support the tabindex content attribute and ARIA attributes.
> 8. The shadow DOM should include all renderable HTML elements save standard input controls. The reasoning being is these
> controls are user agent managed in terms of the keyboard and would likely conflict with canvas keyboard management.
> 
> 
> The format would be along the following lines:
> <canvas aria-activedescendant="foo">
>   <access mode="default">  /*Since canvas is visual this is the default mode*/
>      <div role="toolbar">
>         <div role="tab" tabindex="-1" id="foo">
>         </div>
>      </div> 
>   </access>
>   <access mode="textual"> /*Here we could place a long description or an alternative aria-enabled widget*/
>   </access>
>   <access mode="auditory">
>   </access>
>   <access mode="tactile"> /*This is in access for all and we could lose it for now for obvious reasons*/
>   </access>
>   <access mode="olfactory"> /*This is in access for all and we could lose it for now for obvious reasons*/
>   </access>
>   <access mode="visual" type="signlanguage">
>   </access>
> </canvas>
> 
> 
> Rich Schwerdtfeger
> Distinguished Engineer, SWG Accessibility Architect/Strategist
>