Minutes: HTML Accessibility Task Force Canvas Accessibility subteam February 11, 2010

The meeting minutes  were merged with the normal task force minutes:

http://www.w3.org/2010/02/11-html-a11y-minutes.html


Thank you Madeleine for attending.

Here is the text version of the chat log:

Minutes: HTML Accessibility Task Force Canvas Accessibility subteam:


[1:01pm] • trackbot is preparing a teleconference
[1:01pm] trackbot: RRSAgent, make logs world
[1:01pm] RRSAgent: I have made the request, trackbot
[1:01pm] trackbot: Zakim, this will be 2119
[1:01pm] Zakim: I do not see a conference matching that name scheduled
within the next hour, trackbot
[1:01pm] trackbot: Meeting: HTML Accessibility Task Force Teleconference
[1:01pm] trackbot: Date: 11 February 2010
[1:01pm] richardschwerdtfe: chair: Rich
[1:02pm] richardschwerdtfe: Meeting: HTML Accessibility Task Force Canvas
Accessibility subteam
[1:02pm] richardschwerdtfe:
http://lists.w3.org/Archives/Public/public-canvas-api/2010JanMar/0153.html

[1:04pm] • davidb richardschwerdtfe got it
[1:04pm] cshelly joined the chat room.
[1:04pm] frankolivier joined the chat room..
[1:04pm] • davidb hi cshelly frankolivier
[1:05pm] cshelly left the chat room. (Quit: CGI:IRC)
[1:05pm] cyns joined the chat room..
[1:05pm] richardschwerdtfe:
http://lists.w3.org/Archives/Public/public-canvas-api/2010JanMar/0153.html

[1:05pm] richardschwerdtfe:
http://lists.w3.org/Archives/Public/public-canvas-api/2010JanMar/0155.html

[1:06pm] • davidb
http://lists.w3.org/Archives/Public/public-canvas-api/2010JanMar/att-0155/afamq.html

[1:06pm] jongunderson joined the chat room.
[1:06pm] richardschwerdtfe: scribe: davidb.
[1:06pm] davidb: scribe: David_Bolter
[1:07pm] davidb: RS: Access4All describes a meta data model and preferences
model, so that they can be matched
[1:07pm] davidb: RS: for individual users
[1:07pm] davidb: RS: trying to bring this to a more mainstream opportunity
[1:08pm] davidb: RS: HTML5 has a facility called media queries
[1:08pm] davidb: RS: you can have style tags around your document, we can
take content in html5 and selectively rendering it through a style sheet
media query
[1:09pm] davidb: RS: I have taken the Access4All metadata and defined it in
terms of a media query-able thing
[1:09pm] davidb: RS: taking pieces of A4All that work for us
[1:10pm] davidb: RS: [giving examples]
[1:11pm] davidb: RS: adaptation types]
[1:11pm] davidb: RS: can say if something is AT interoperable
[1:11pm] davidb: RS: relevant to canvas, currently this is a boolean but we
could specify what APIs this has been tested with per specific browser
[1:12pm] davidb: RS: e.g. firefox with iaccesible2
[1:12pm] davidb: RS: but would like it to be extensible
[1:13pm] davidb: RS: a user can request for example 'full keyboard access'
[1:14pm] davidb: RS: resources should be able to specify multiple supports,
users shouldn't to avoid conflict
[1:14pm] davidb: RS: display options (...)
[1:15pm] • davidb wonders if user can say "no ads please"
[1:15pm] davidb: RS: struggling with 'education level' settings
[1:15pm] davidb: RS: thinking of saying 'simplified' vs 'complex' content
[1:16pm] davidb: RS: language of adaptation... addresses things like
[1:16pm] davidb: RS: say I have an audio object in my page, and I want a
french text version
[1:16pm] davidb: RS: you can specify in media query
[1:17pm] davidb: RS: current html5 spec supports media queries, but don't
specify the format
[1:18pm] davidb: RS: this gives us the flexbility to selectively include
content, gives user to control what is rendered
[1:18pm] davidb: RS: David Singer expresses a similar approach, but hasn't
yet aligned it with this, what has been specified in the education space .
[1:18pm] davidb: RS: I am proposing to go forward with this approach to
allow user and content to be matched
[1:19pm] davidb: RS: comments?
[1:19pm] davidb: FO: does make sense... (garbled)
[1:19pm] davidb: FO: looks good
[1:19pm] davidb: RS: we wouldn't have to modify html5 spec for this
[1:20pm] davidb: SF: no comments, yeah makes sense, it is not what i would
consider my priority in this
[1:20pm] davidb: CS: this is valuable and it is useful to have a mechanism
for content negotiation
[1:20pm] davidb: CS: but don't want to use this as an excuse to not
implement full a11y
[1:21pm] davidb: RS: we want direct a11y if possible, there are issues to
be ironed out with html wg
[1:21pm] davidb: RS: I want to work this solution (media q) out with David
Singer etc, and move forward.. ok?
[1:21pm] davidb: general agreement
[1:21pm] davidb: no negative comments
[1:22pm] davidb: RS: looking toward CSS media query specification now,
looking to resolve for 25th
[1:23pm] davidb: RS: Next issue, making canvas directly accessible
[1:23pm] davidb: RS: If we just use davidb technique, there is no way to
validate that someone made it accessible
[1:23pm] Stevef: q+:
[1:23pm] • Zakim sees Stevef on the speaker queue
[1:23pm] richardschwerdtfe: q?
[1:24pm] • Zakim sees Stevef on the speaker queue
[1:24pm] richardschwerdtfe: david: For me I had not thought about the
validation piece.
[1:25pm] richardschwerdtfe: Stevef: I mentioned in the meeting today that I
like David's technique but unless we provide a method to make canvas
directly accessible interactively then it is a waste of time.
[1:25pm] davidb: SF: I like David's technique. Unless we provide a method
to make interactive canvas elements accessible, then it won't get done most
of the time
[1:26pm] davidb: oops two scribes
[1:26pm] richardschwerdtfe: Stevef: I don't want a situation where it is
not done most of the time.
[1:26pm] richardschwerdtfe: Cynthia: if the <accessible> tree is not in the
content then the author won't do the work.
[1:27pm] richardschwerdtfe: davidb: I want to say that my technique is
built from Steve's and Frank's techniques I am sort of wrapping around
<canvas>
[1:27pm] richardschwerdtfe: davidb: Not being to validate is a negative.
[1:28pm] davidb: RS: so one of the problems we get Cynthia, is that canvas
needs to support 'fallback' content
[1:29pm] davidb: RS: if we put an a11y tree inside canvas element, and
canvas goes away, a11y tree gets erroneously rendered
[1:29pm] davidb: RS: david was suggesting wrap a11y tree around canvas, but
fallback could invade that (scribe's words)
[1:30pm] davidb: RS: if didn't support canvas, make a11y tree go away, and
display fallback
[1:30pm] davidb: CS: can we do visibility hidden
[1:31pm] davidb: RS: we could add an attr to canvas that actually points to
the a11y dom
[1:31pm] davidb: CS: point by ID? seems like it could [satisfy validation
req]
[1:32pm] richardschwerdtfe: david: I was talking to someone about this
on/off approach for canvas. We run into a problem if javascript is not
supported either
[1:32pm] davidb: sorry
[1:34pm] davidb: FO: looked at canvas code today, you can sniff for canvas
by asked for canvaselement.context2d
[1:34pm] richardschwerdtfe: frank: if you convas.getContext2D() and you
don't get an object you don't have canvas
[1:34pm] davidb: FO: that is the technique used today [db: feature
detection]
[1:34pm] Stevef: q+:
[1:34pm] • Zakim sees Stevef on the speaker queue
[1:35pm] richardschwerdtfe: Stevef: It seems like we are stepping away from
what is currently in the spec.
[1:35pm] davidb: SF: it seems like we are stepping away from what is
currently in the spec, with the canvas sub dom being navigable via keyboard
[1:36pm] richardschwerdtfe: Frank: No
[1:37pm] • davidb thinks steve was speaking generally
[1:40pm] davidb: discussion of what's in spec now
[1:41pm] davidb: SF: the issue i see is that what is in the fallback is
really not 'fallback'
[1:41pm] davidb: SF: i think that's a good thing
[1:41pm] davidb: SF: having a separate a11y tree that does not interact is
not good
[1:42pm] davidb: RS: i'm confused here, how are you going to assume that
fallback is rendered on the canvas?
[1:42pm] davidb: RS: you're saying we don't need another a11y tree that is
bound to canvas
[1:43pm] davidb: SF: take a canvas that has some focusable areas, there
will be four links in the canvas subtree, mapped to the focusable areas,
when a user tabs into the first link
[1:43pm] davidb: SF: the first focus rect shows
[1:43pm] davidb: RS: what about things like bespin?
[1:43pm] davidb: RS: rich text, toolbars... and you need structure
[1:44pm] davidb: SF: what I would like to see at least an easy method that
will be used, to cover the vast majority of (simple) cases
[1:44pm] davidb: SF: not saying there should be a way to solve more comples
cases
[1:44pm] davidb: RS: ok
[1:45pm] davidb: [discussion of image map approach being too basic]
[1:47pm] davidb: SF: we don't want to call the content inside canvas
fallback, because it doesn't describe what i'm describing)
[1:48pm] • davidb can someone else scribe?
[1:49pm] davidb: RS: there are a number of ways we can have an accessible
binding to canvas
[1:52pm] davidb: SF: we need to provide things through spec as opposed to
hacks
[1:53pm] davidb: RS: how we bind the a11y tree is just impl detail
[1:53pm] davidb: RS: we can have fallback, we can have an a11y element, or
we can have an attribute on canvas that points to a11y content (for
validator)
[1:53pm] davidb: RS: then how do we get each one programmatically
'rendered'
[1:54pm] kford left the chat room. (Quit: ChatZilla 0.9.86 [Firefox
3.6/20100115144158])
[1:55pm] davidb: SF: we have a process we can use in html wg, so that the
a11y dom inside canvas is traversable by default
[1:58pm] davidb: [discussion about definition of 'fallback'],
[1:59pm] Stevef: <canvas>
[1:59pm] Stevef: <a href="#">normal contentlink</a>
[1:59pm] Stevef: <fallback>get firefox canvas not supported</fallback>
[1:59pm] Stevef: </canvas>>
[2:00pm] davidb: SF: we should protect what is in the spec, and rename it
since it isn't 'fallback'
[2:00pm] davidb: RS: now you are mixing them up
[2:01pm] davidb: RS: fallback is rendered when canvas is gone
[2:01pm] davidb: SF: yes.
[2:01pm] davidb: SF: the spec says you can interact with focusable element
when canvas is there, but spec calls any content within canvas tags
'fallback'
[2:03pm] davidb: SF: when scripting is not supported, then 'fallback' is
shown
[2:03pm] davidb: RS: declaritive markup
[2:03pm] davidb: CS: does the spec say what to do if script is not
supported
[2:03pm] davidb: SF: think so
[2:04pm] davidb: SF: FF and Chrome show fallback, Opera doesn't IIRC
[2:04pm] davidb: RS: please attend the call Monday
[2:05pm] davidb: RS: everyone
[2:05pm] davidb: RS: haven't addressed tracking a caret
[2:05pm] • davidb ducks
[2:06pm] davidb: RS: caret is bound to a focussed object, and map...
(scribe fail)
[2:06pm] davidb: RS: also selected text. actually shouldn't be too bad.
[2:07pm] davidb: DB: do we have a wiki explaining our approach
[2:08pm] davidb: SF: I'll look up Laura's link and send to list
[2:08pm] davidb: DB: thanks
[2:09pm] davidb: RS: steve was saying, within the canvas element we can
have regular html, however to specify 'fallback' content it needs to be
inside fallback start and end tags.
[2:09pm] davidb: RS: fallback is ingored when canvas is rendered
[2:09pm] davidb: CS: other markup inside is used for a11y content
[2:09pm] davidb: CS: makes sense to me, i like that
[2:10pm] Stevef:
http://dev.w3.org/html5/spec/the-canvas-element.html#the-canvas-element

[2:10pm] Stevef: In non-interactive, static, visual media, if the canvas
element has been previously painted on (e.g. if the page was viewed in an
interactive visual medium and is now being printed, or if some script that
ran during the page layout process painted on the element), then the canvas
element represents embedded content with the current image and size.
Otherwise, the element represents its...
[2:10pm] Stevef: ...fallback content instead.
[2:11pm] davidb: DB: every approach has problems, i can get behind this
one.
[2:13pm] richardschwerdtfe: RRSAgent, make log member
[2:13pm] RRSAgent: I have made the request, richardschwerdtfe
[2:13pm] richardschwerdtfe: Zakim, bye
[2:13pm] Zakim left the chat room.
[2:14pm] richardschwerdtfe: RRSAgent, draft minutes

Rich Schwerdtfeger
Distinguished Engineer, SWG Accessibility Architect/Strategist