Minutes: January 25, 2010 HTML 5 Canvas Accessibility Call

So, we had a problem with RRSAgent running that seems to be related to the
fact that a conference matching the time slot was not scheduled. I will
look into it with Mike Smith. We made a lot of progress and I believe we
are getting closer to a middle ground on addressing canvas accessibility. I
will summarize shortly. The text of the minutes from the irc channel is
included here.

richardschwerdtfe: trackbot, start meeting
[1:59pm] • trackbot is starting a teleconference
[1:59pm] trackbot: RRSAgent, make logs world
[1:59pm] trackbot: Zakim, this will be 2119
[1:59pm] Zakim: I do not see a conference matching that name scheduled
within the next hour, trackbot
[1:59pm] trackbot: Meeting: HTML Accessibility Task Force Teleconference
[1:59pm] trackbot: Date: 25 January 2010
[2:00pm] franko joined the chat room.
[2:00pm] davidb: passcode?
[2:01pm] richardschwerdtfe: 92473
[2:01pm] davidb: ty
[2:01pm] richardschwerdtfe: it look like Mike Smith did not set up the call
[2:01pm] richardschwerdtfe: I am not sure why
[2:01pm] richardschwerdtfe: it was a repeating call
[2:01pm] richardschwerdtfe: there you go
[2:01pm] davidb: Zakim, Mozilla is David_Bolter
[2:01pm] Zakim: sorry, davidb, I do not recognize a party named 'Mozilla'
[2:01pm] richardschwerdtfe: ok. it seems to be working
[2:03pm] richardschwerdtfe:   1. HTML 5 Canvas Accessibility Meeting
[2:03pm] richardschwerdtfe:   Monday, 2010-1-25
[2:03pm] richardschwerdtfe:   Time: 3:00pm-4:00pm Boston local
[2:03pm] richardschwerdtfe:   Name: WAI_PFWG(CANVAS)
[2:03pm] richardschwerdtfe:   Code: 92473 ("WAIPF"))
[2:03pm] richardschwerdtfe:   One time
[2:03pm] richardschwerdtfe:   irc channel= #html-a11y
[2:03pm] richardschwerdtfe:   Note: I am sending this out to give people
notice of Monday's repeating
[2:03pm] richardschwerdtfe:   meeting with a preliminary agenda. The agenda
may change based on Friday
[2:03pm] richardschwerdtfe:   HTML5 working group discussions which I have
not yet reviewed due to
[2:03pm] richardschwerdtfe:   travel.
[2:03pm] richardschwerdtfe:   Agenda:
[2:03pm] richardschwerdtfe:   1. Identify Scribe
[2:03pm] richardschwerdtfe:   2. Any issues with time slot
[2:03pm] richardschwerdtfe:   3. Shadow DOM - (what should element should
we call it? <access>?)
[2:03pm] richardschwerdtfe:   4. Media Query application to canvas for
alternative renderings and
[2:03pm] richardschwerdtfe:   fallback renderings for non-supporting
browsers
[2:03pm] richardschwerdtfe:   - use of media types as attributes vs.
elements (HTML 5 uses elements)
[2:03pm] richardschwerdtfe:   - Impact on the fallback DOM
[2:03pm] richardschwerdtfe:   - Impact of having a shadow DOM - How can we
avoid it's being rendered
[2:03pm] richardschwerdtfe:   for non-supporting browsers without a huge
burden on the author?
[2:03pm] richardschwerdtfe:   5. Canvas fall back impact for non-supporting
browsers
[2:03pm] richardschwerdtfe:   6. JavaScript method for focus setting/focus
ring (Steve F. what issues
[2:03pm] richardschwerdtfe:   remain?)
[2:03pm] richardschwerdtfe: scribe: davidb
[2:04pm] davidb: rs: 'shadow dom' - a proposal to use an element called
<access>
[2:05pm] jongunderson joined the chat room.
[2:05pm] davidb: frank: two core design principles 1) don't break any
fallback content out there today - don't break canvas 2) make clear to a11y
tool there is fallback
[2:06pm] davidb: frank: marking it with a certain tag, won't do it.
[2:06pm] davidb: frank: how soon will people upgrade browser to one which
understands <access>
[2:06pm] davidb: frank: if you event magic, you have something quirky
[2:06pm] davidb: frank: so we have two non optimal solutions
[2:07pm] davidb: frank: the easiest and simplest is to not give authors a
way to statically create fallback content - they do it programmatically
[2:07pm] davidb: frank: all named element solutions seem quirky to me tbh
[2:07pm] davidb: rs: are you suggesting fallback also be shadow dom - not
sure will work... bound to canvas
[2:08pm] davidb: frank: obscured by canvas?
[2:08pm] davidb: rs: that and the visual rendering
[2:08pm] davidb: (discussion to get on same page)
[2:09pm] davidb: rich: maybe a strategy approach...
[2:09pm] davidb: rich: suppose no shadow dom, and your fallback content is
accessible, how do you choose it, not knowing the browsers capabilities?
[2:10pm] davidb: rich: and assume no <access> element
[2:10pm] davidb: rich: the only way to select it is that if your browser
doesn't support canvas, then the fallback disappears
[2:10pm] davidb: frank: my preferred approach, you have to put fallback,
and programmatically create accessible content
[2:11pm] davidb: frank: shadow dom always programmatically created
[2:11pm] davidb: frank: probably least painful is to just say, there are
two different ways to markup canvas tag 1) fallback 2) accessible content?
[2:11pm] davidb: frank: diff tags for both
[2:12pm] davidb: rich: not every user wants the same alternative content.
the group doesn't want to introduce media tags within the canvas element.
we want to control the rendering through css, and standardized properties
[2:12pm] davidb: rich: e.g. subway map
[2:13pm] davidb: rich: i want to choose an alt representation, if
someone ???? media query
[2:13pm] davidb: rich: we could define some properties that would cause alt
content to be rendered.
[2:14pm] davidb: db: how about aria-media="aural"?
[2:14pm] davidb: rich: trying to satisfy both groups.
[2:15pm] davidb: frank: you'll have authors who want to add one alt
presentation, not bother with multiple
[2:15pm] davidb: frank: and definitely want a way to specify multiple
encodings of the same canvas content.
[2:15pm] davidb: frank: some way to say, this tree is useful for these
users.
[2:16pm] davidb: frank: you want a comma separated generic list (accessible
tag), specify one or more modalities for that content
[2:16pm] davidb: rich: we could tag the fallback as that alternative
[2:16pm] davidb: frank: sure
[2:17pm] davidb: rich: does group agree metadata is good here, and user can
select through styles
[2:17pm] davidb: db: yep
[2:17pm] davidb: rich: so we do this using properties, try to use fallback,
and inevitable, if people want to make canvas accessible we need to define
an element within the canvas tag.
[2:18pm] davidb: rich: and fallback
[2:18pm] davidb: db: i'm fuzzy on that
[2:18pm] davidb: rich: say i left out <access>, and no binding to canvas,
we could add a11y properties to the container for the fallback content that
is within the canvas and have that rendered instead of the canvas based on
css properties
[2:19pm] davidb: frank: would be useful to have very clear examples,
minimum case to full featured case
[2:19pm] davidb: rich: we'll have to do that. and reserve some properties.
can't use ID (ofcourse)
[2:19pm] davidb: rich: what should we call this 'shadow dom' element
[2:20pm] davidb: frank: something with a11y
[2:20pm] davidb: frank: access, accessible, accessibility
[2:21pm] davidb: rich: accessible (element)
[2:21pm] davidb: rich: we could put aria on that too i suppose
[2:21pm] davidb: rich: they could put it anywhere in the dom, and it would
be hidden
[2:22pm] davidb: rich: we need to come up with a set of properties
[2:22pm] davidb: rich: we should reuse IMS taxonomy
[2:22pm] davidb: frank: agree
[2:23pm] davidb: jg: when you say a css media query, is that your hardware
or does the user say oh i don't want canvas, i want xyz
[2:23pm] davidb: rich: user can specify based on properties
[2:23pm] davidb: rich: can use a style sheet
[2:24pm] davidb: db: you could use jetpack
[2:24pm] davidb: frank: we have a preference for users to specify a style
sheet
[2:24pm] davidb: rich: ok
[2:24pm] davidb: group discusses using "accessible" as element name
[2:25pm] davidb: rich: until browsers support this, we can recommend that
they provide fallback that can be rendered via stylesheet
[2:25pm] davidb: frank: depends on user agent
[2:26pm] davidb: frank: solvable.
[2:27pm] davidb: resolution: "shadow dom" now represented via <accessible>
[2:27pm] davidb: resolution: can select fallback or some alt content for
canvas via a stylesheet like design
[2:28pm] davidb: jg: say the canvas is a chart, you are saying here the
<accessible> would be an alt content, don't use canvas, use something else
that is equiv.
[2:28pm] davidb: rich: right
[2:28pm] davidb: rich: within canvas tags you would have a table (<table>)
[2:28pm] davidb: jg: if i'm using a screen reader, it says i can't use
canvas, so i will look in accessible tag?
[2:28pm] davidb: rich: good q.
[2:29pm] davidb: rich: in terms of binding to a11y API... (thinking)
[2:29pm] davidb: jg: skeptical about creation of alternatives.
[2:30pm] davidb: david: there are problems with all solutions.
[2:31pm] davidb: jg: would you able to use the same techniques you use on
IMG on the canvas, so simply point to a data table
[2:31pm] davidb: jg: use aria-describedby?
[2:32pm] davidb: david: i don't see why not
[2:32pm] davidb: rich: yeah
[2:32pm] davidb: jg: i think that will be the most common use case
[2:32pm] davidb: david: yeah
[2:32pm] davidb: jg: easier to talk about something, when it is on the
screen for everyone
[2:33pm] davidb: rich: can choose via css
[2:34pm] davidb: david: match user properties to what exists in the dom
(available)
[2:34pm] davidb: ?
[2:34pm] davidb: rich: yes
[2:35pm] davidb: david: how many accessible elements allowed? where do
properties go?
[2:35pm] davidb: rich: the tag within canvas would just be for binding what
is in canvas to shadow dom, otherwise use fallback content, via css
[2:36pm] davidb: david: agree examples are good
[2:37pm] davidb: frank: we need an interactive example especially (for more
scrutiny)
[2:37pm] davidb: rich: volunteers for example creation?
[2:37pm] davidb: jg: interested.
[2:37pm] davidb: jg: canvas has js events right? handlers etc?
[2:37pm] davidb: group: yep
[2:38pm] richardschwerdtfe: RRSAgent, make log member
[2:38pm] • davidb richardschwerdtfe i gotta go
[2:39pm] richardschwerdtfe: scribe: Rich
[2:39pm] richardschwerdtfe: Jon: would there be a way to have the controls
in canvas be keyboard accessible
[2:40pm] • davidb cheers all
[2:40pm] richardschwerdtfe: david: Canvas is just a way to draw to the
canvas area
[2:40pm] richardschwerdtfe: jon: do you imagine tools?
[2:40pm] richardschwerdtfe: frank: the people who create the toolkits need
to keep accessibility in mind
[2:41pm] richardschwerdtfe: frank: Once we have a viable solution the next
step is to add accessibility to the toolkit
[2:41pm] richardschwerdtfe: Jon: we need to have an example with the test
suites
[2:41pm] richardschwerdtfe: frank: the tab and focus solution we need to
show. it is not necessarily clean
[2:42pm] richardschwerdtfe: jon: you need to create your own UI manager.
[2:42pm] richardschwerdtfe: frank: yes, there is more work
[2:42pm] richardschwerdtfe: frank: many canvas examples are mouse only
[2:42pm] richardschwerdtfe: jon: does the canvas receive keyboard focus
now?
[2:42pm] richardschwerdtfe: frank: good question
[2:43pm] richardschwerdtfe: frank: I need to look at James' example again.
It is not elegant but doable
[2:44pm] richardschwerdtfe:
http://lists.w3.org/Archives/Public/public-canvas-api/2009OctDec/0026.html

[2:45pm] richardschwerdtfe: ACTION: frank will do multiple controls with
interactivity
[2:45pm] • trackbot noticed an ACTION. Trying to create it.
[2:45pm] trackbot: Created ACTION-15 - Will do multiple controls with
interactivity [on Frank Olivier - due 2010-02-01].
[2:46pm] richardschwerdtfe: ACTION: Jon: Do Chart example with the data
table for alternative selection
[2:46pm] • trackbot noticed an ACTION. Trying to create it.
[2:46pm] trackbot: Created ACTION-16 - Do Chart example with the data table
for alternative selection [on Jon Gunderson - due 2010-02-01].
[2:47pm] richardschwerdtfe: ACTION: Rich define CSS attributes and meta
data for alternative content selection
[2:47pm] • trackbot noticed an ACTION. Trying to create it.
[2:47pm] trackbot: Created ACTION-17 - Define CSS attributes and meta data
for alternative content selection [on Richard Schwerdtfeger - due
2010-02-01].
[2:48pm] richardschwerdtfe: Rich: do you think they are going to have
problems of defining attributtes to allow for selection of alternative
content?
[2:48pm] richardschwerdtfe: Frank: there is no way around this. People will
have to accept it
[2:49pm] richardschwerdtfe: Jon: the discussion related to the table
[2:49pm] richardschwerdtfe: Jon: the discussion seems more like there is a
right and wrong way to choose alternative content
[2:50pm] richardschwerdtfe: Jon: if aria-labelledy by is preferred then we
use that ... otherwise use a table
[2:50pm] richardschwerdtfe: Rich: Whatever is rendered is mapped to the
accessibility APIs
[2:52pm] richardschwerdtfe: Jon: The question seems to center around
marking content for accessibility api mapping
[2:53pm] davidb left the chat room. (Quit: davidb)
[2:54pm] richardschwerdtfe: Jon: Summary is required or not.
[2:54pm] richardschwerdtfe: Rich: It is about implied host language
semantics
[2:55pm] richardschwerdtfe: Jon: right. so, this is about api mappings
[2:56pm] richardschwerdtfe: Rich: We are describing how to allow the author
to meet the user's needs with canvas

Rich Schwerdtfeger
Distinguished Engineer, SWG Accessibility Architect/Strategist