- From: Richard Schwerdtfeger <schwer@us.ibm.com>
- Date: Tue, 29 Mar 2011 12:13:06 -0600
- To: public-html@w3.org, public-canvas-api@w3.org, public-html-a11y@w3.org
- Cc: franko@microsoft.com, chuck@jumis.com, cyns@exchange.microsoft.com, faulkner.steve@gmail.com, david.bolter@gmail.com
- Message-ID: <OF6A58DDBD.13A92692-ON86257861.007D6358-86257862.00641349@us.ibm.com>
As part of the canvas accessibility work we need to provide assistive technologies with bounding rectangles for fallback content that has a 1:1 mapping to visual objects on canvas. This information would be provided to platform accessibility API to support: - Screen Magnifiers to zoom to semantic objects in the fallback content based on their visual rendering in canvas - Screen readers to make decisions on what gets sent to refreshable Braille devices (where the line breaks occur) When standard HTML is used to render content the user agent uses the layout engine to compute this information but because canvas rendering is separated from the actual fallback this cannot be achieved unless there is an association of the drawing space in the canvas UI for the object represented in fallback content. The canvas accessibility working group's solution to the problem is combine what we feel is an improvement on hit testing in canvas by creating a draw path representation of the clickable region for the the fallback content and associating them with each other. This solves two problems: - When you click on the canvas area the mouse click would go to the fallback content associated with the clickable region. This also allow the same fallback node that processes the keyboard input for the canvas rendering to also receive the pointing device input. A common complaint about canvas is that the author had to manage all the hit testing. - The drawing path in canvas will manage any transformations to screen coordinates that you might otherwise need to compute if you were to use say CSS. - Allows the user agent to map the the drawing path for the clickable region to the accessible object This concept is not new and in fact has been around GUIs for decades. For example, in Windows, you create a handle to a device context that has things like: font information, foreground and background color, bounding rectangles, and clipping regions. This HDC is used in hit testing and the bounding rectangle was used to supply the accessibility API bounding rectangle. To do this properly we need to address overlapping clickable regions where the last drawn should get the pointing device input. We could provide an API to do this (something like setLastDrawn(element) but we feel authors will get quickly frustrated at having to manage that. Ideally, we would like to see authors associate each drawing call with a corresponding element in fallback content for which the default could be the canvas element itself. This is of course a change to the way canvas apps. are authored. We would like to hear feedback from the canvas development community. This early draft of the approach highlights what we are considering (see between the ZZZ sections). The setLastDrawn(element) is a placeholder. We would really prefer that the drawing APIs deal with the computation of the lastdrawn element when clickable regions overlap. We should also note that an investigation of the DOM event specification left out any mention as to how the user agent determines if a mouse click occurs within the drawing space of an element. (See attached file: Canvas2DClickableRegion.html) Thank you, Rich Rich Schwerdtfeger CTO Accessibility Software Group
Attachments
- text/html attachment: Canvas2DClickableRegion.html
Received on Tuesday, 29 March 2011 18:17:26 UTC