HTML Accessibility Task Force Teleconference -- 10 Mar 2011

<trackbot> Date: 10 March 2011

<MikeSmith> hmm

<janina> Meeting: HTML-A11Y telecon

<janina> agenda: this

<scribe> scribe: Gregory_Rosmaita

<scribe> scribenick: oedipus

Face to Face

Please Register http://www.w3.org/2002/09/wbs/44061/201103_ftf/

<MichaelC> agenda: http://lists.w3.org/Archives/Public/public-html-a11y/2011Mar/0090.html

Update from Facilitators

JS: in case we use full hour, want to mention that re: longdesc we have been advised there is an industry group that is concerned that it is missing and that will be damaging to their ability to support people with disabilities and to get support from browser vendors and authors they are working with -- letter on way to W3C from industry body by the end of the week expressing concern and...
... need for long description mechanism
... will alert HTML WG about this in HTML WG call at noon

PC: can't see any reason to bother HTML WG until can provide copy of letter

JS: ok, good point

PC: work with team -- get letter on public email list
... getting into public list archive is important -- can work with JS on this

Face2Face and Associated Deadlines

JS: need a head-count, so please register at http://www.w3.org/2002/09/wbs/44061/201103_ftf/ as soon as possible -- which means today --

CS: if you don't register by 11 march 2011, can't provide special lunches as can be requested via registration page

<janina> http://www.w3.org/WAI/PF/HTML/ftf_2011-03.html

JS: draft agenda for F2F at http://www.w3.org/WAI/PF/HTML/ftf_2011-03.html
... specificity -- want to ensure named everything need to look at and allotted appropriate time for work

<JF> +q

MC: text alternatives request from SteveF
... much of agenda split into parallel sessions so that subgroups can work in parallel -- still want to have main body meet together, but media will be breaking out -- text alternatives not on list yet

LW: agree with steve, need text alternatives time slot

JS: anything else?

JF: media going to compose working agenda for media breakout sessions -- feel that have an ordered list of things to be accomplished -- recap on sunday morning -- some dial-in commitments
... couple of other open issues in issue tracker that are a11y related -- need to sit back and look where we are and identify any holes need to address

MC: sunday after lunch session is for that

JF: issue of timing -- push that a bit higher in stack? will lose people on sunday incrememtally

RS: flying back monday

JS: want to be sensitive to who is there while we are there, main reson for f2f is to resolve problems where we are able to discuss issues in person -- overviews not as tricky to do in telecon as is seizing the group dynamic
... text alternative discussion is stalled -- want to come out of F2F with very specific directions for immediate future work -- need big boost that comes from solving issues --
... things we know about, and things we need to organize (entire spec review) -- done once by PF and WAI -- realize need to do this every few months

JF: additional IRC channels?

MC: join new channel and creates one dynamically -- suggest html-a11y-media
... keep in mind, IRC channel name goes into the name of the minutes

JS: would be good to preannounce the IRC channel

MC: makes sense for media, suggest other subgroups use main channel
... almost all of the breakout sessoins include one for media

JF: works for me

Subteam Reports: Canvas; ARIA Mappings; Media; Bug Triage

JS: want to hear what need to address at F2F

RS: canvas update: submitted change proposal -- working on how to get bounding rectangles for each accessible object -- number of things kicked around -- want to impact authors the least while adding value
... one problem is hit-testing in canvas
... canvas today: can draw via canvas, but not great as an environment -- when first started a11y work on windows for MSAA, what happened is when draw to screen have device context (color info, font info, clipping info) -- clipping info used to do hit-testing as well as supplying a11y APIs with a rectangle -- hit-testing very poor in canvas -- suggested that we bind a clipping rectangle...
... which can be used for hit testing AND binding to a11y API
... fallback content becaomes object model bound to Canvas -- for author, clipping region based on transformations already used in canvas, so easier than CSS -- bind to element in shadow dom, provide a11y API computed rectangle AND can do hit-testing (onClick event in shadow DOM that is bound to click on canvas -- right now no way to do
... feedback: google wants examples
... solution to problem going to take more work because have to redefine how hit-testing is done; may not be completely done until LC or after -- asked Maciej for feedback from apple, haven't yet received -- that is next step

PC: ... maciej on 2 weeks of vacation
... away until 20 march 2011

RS: didn't know -- thatnks
... really need to investigate this -- solves VERY big problem for canvas -- allows clickable regions AND operate off of the DOM -- that is a major gain
... opened bug on clarification of fallback content is implemented -- got feedback from google/webkit -- more editorial than anything else

http://www.w3.org/WAI/PF/HTML/wiki/Canvas

http://www.w3.org/WAI/PF/HTML/wiki/Canvas/Meetings

http://www.w3.org/WAI/PF/HTML/wiki/Canvas/Meetings/Minutes

JS: ARIA Integration at F2F?

RS: at f2f need to do lexical processing peice for ARIA attributes in HTML5 -- me, MCooper, SteveF -- has to go into section 3.2.6 in document -- said that is where will go when ARIA went to CR -- told hixie we are discussing at face2face and invited his participation

<JF> +q

RS: do we need to make changes to ARIA for HTML5 -- number of HTML5 elements that don't have default semantic role? do we need longdesc mechanism for ARIA.next?

JF: question came up -- in element such as canvas does multiple things, can one assign multiple roles?

RS: aria allows one to take a collection of roles to apply to any element -- ARIA 1.0 says "take first valid role and map to a11y APIs" -- most a11y apis only take 1 role -- need to investigate how to handle multiple roles

JF: came up in context of another discussion

RS: Gregg Vanderhieden asked same question for kiosks -- wants to use native aria roles, but what if wanted additional one -- ARIA has extensibility capability, can set additional role in list, and when your AT runs in kiosk, can trigger off other role values

CS: precedent in UIA

JF: something to discuss at F2F if have time

JS: media update?

JF: good call yesterday -- right now have an omnibus issue (issue 9) -- was very vague, so letting go -- working on issue 152 -- what to do with multitrack api --associate secondary media files to primary video -- proving to be a tricky proposition -- lot of discussion with engineers, focusing in on few solutions, hoping that F2F is where major progress will take place
... disposed of some bugs associated with Issue-9

issue-9?

<trackbot> ISSUE-9 does not exist

JF: couple of bugs we feel can be closed not in closed state -- who is responsible for closing bugs we didn't specifically open -- there are 2 that still need to be addressed -- all go back to original issue: how to support multiple content selection that is synchronized

JS: still looking at 8 proposals?

JF: still 8 proposals in wiki; discussion focused on 2 or 3 specific; haven't eliminated any "formally"
... discussion focused on 2 or 3 possible solutions

JS: next step: advance the 2 or 3 on the way to working towards 1

JF: procedural -- discussion is onging
... wiki is a placeholder -- silvia shepherd of wiki pages for media

JS: need to synchronize is wider than just a11y -- i18n support requires it as well, needs wider conversation

JF: right, not just captions, but subtitles
... progress IS being made

JS: bug triage?

MK: JF covered media bug clean-up; action 108 can be closed; 3 bugs can be closed, 1 or 2 need further comments from media subteam others discuss at F2F meeting
... Marco and MCooper did usual triage of new a11y bugs this week -- inbox
... "inbox" is empty

MC: did have 1 bug to bring to wider attention of TF

<Marco_Ranon> http://www.w3.org/Bugs/Public/show_bug.cgi?id=8659

MC: Bug 8659 - indicate captions and audio-description

<kliehm> It says "we wait for the browser vendors to catch up before specifying an API".

<kliehm> If the API is imperative for accessibility, it needs to be specified

MC: don't know why flagged

<kliehm> pre-Last Call. We suggest to discuss this in the TF telcon.

JF: this is related to issue-152 again -- questoin is are we actually firing an event? answer is "sort of" -- needs to be discussed at F2F
... not resolved, but once get 152 settled will dispense with this bug

MC: if in results closer to verified and closed -- may need to reopen and assign to yourself

JF: can you do that for me?

MC: can but not sure if right way to go -- "resolve later" can drop off radar

JF: bug not resolved -- final response is hixie has partially accepted point of part of bug -- agrees need API for timed-text data -- don't have for multi-track
... if can associate with issue-152, ok -- have for text alternatives, but not multimedia alternatives

MC: procedural discussion with JF about administrivia

JS: in terms of closing bugs we didn't open ourselves, we empowered bug triage team to handle them

MC: correct
... moving to "verified" which is "closing" we are empowered to do -- chairs will move verified to closed at some point

JS: annotating that subteam looked at bugs being discussed, alert bug triage team to mark "verified" or other --- provide annotation

MC: don't need to wait for bug triage team -- best handled by media subgroup since they have expertise

JF: those we had questions about are "verified"

Actions Review

http://www.w3.org/WAI/PF/HTML/track/actions/open

MC: skipping subteam actions

<kliehm> http://www.w3.org/WAI/PF/HTML/track/actions/90

MC: action-90 on MC -- still open -- related to text alternatives

<kliehm> http://www.w3.org/WAI/PF/HTML/track/actions/101

<kliehm> http://www.w3.org/WAI/PF/HTML/track/actions/105

MC: action-101 on Gregory - modal for ARIA.next -- will have done after CSUN

<MichaelC> action-105 due 15 April

<trackbot> ACTION-105 - review w3c bugzilla installation versus newest version 4, coordinate with those who volunteered, present options and needs due date now 15 April

<kliehm> http://www.w3.org/WAI/PF/HTML/track/actions/106

GJR: please change action 105 to be due in april

<JF> +q

JF: open action page action-109 for sean hayes should be tagged "media"

The <canvas src> Proposal

http://lists.w3.org/Archives/Public/public-html-a11y/2011Mar/0072.html

JS: thought in coordination meeting should discuss on call due to discussion in email
... what do we need to discuss around this proposal? someone introduce in summary form?

RS: haven't really dug into it, but understand request (structural fallback)

JF: TabAtkins made proposal -- his attempt to look at longdesc problem and coming up with a new way to solve verbose descriptor mechanism -- use CANVAS as magic container with src, can put textual equivalent inside of CANVAS tag
... multiple roles on element -- silvia asking about this with regard to this proposal -- if embed image inside of CANVAS, might have confluence of roles that needs to be investigated
... interesting proposal, but not the most logical way to achieve what we need

RS: canvas fallback today -- fallback isn't fallbak in canvas especially on mobile -- has to be a 1 for 1 mapping between what are rendering on content and all in CANVAS sub-DOM

<JF> -data that the chart represents-

RS: longdesc is not a 1-to-1 -- why not just allow img to have subtree for fallback

<Leonie_Watson> +q

RS: this concept is not new -- did this in XHTML2

http://www.w3.org/TR/xhtml2

RS: if container, just put it in -- subtree that hides underneath image specified by canvas -- can also do with object

JF: understand -- this is what is being pitched -- happy that engineers coming forward with alternatives -- need for longdesc seems to be accepted, but trying to do without baggage of longdesc

<Leonie_Watson> +1 to JS

RS: don't want bind to visual rendering -- this is alternative text (can be structured) -- value added: if want to see longdesc, don't need to fetch in separate instance, but can pull forwared

GJR: trying to do for canvas and images what OBJECT provided for free

JF: write once use many times
... interesting proposal
... all for choice -- potentially another tool in toolbox

<paulc> leaving to Chair the WG Weekly meeting

JS: can be done without tying to canvas

RS: worried about it breaking canvas
... navigation disconnect betwen image rendered and longdesc being exposed

LW: doesn't solve problem of longdesc on static image?

RS: depends -- 2 choices: actually have content <IMG src="">Verbose Description not rendered by default</IMG>
... will authors want this on seperate page

MR: read proposal looks to me like we are confusing fallback and alternative -- fallback giving alternative to longdesc -- something UA can provide? how to expose to make aviaiable to those using speech and sight

GJR: worried about use of "fallback" as we corrected use of "fallback" for definition of image http://www.w3.org/html/wg/wiki/ChangeProposals/first_2_paragraphs_of_definition_of_img

JS: no meeting next week
... face2face 19 to 20 march 2011 -- thanks to Cyns and microsoft for hosting

- DRAFT -

HTML Accessibility Task Force Teleconference

10 Mar 2011

Attendees

Contents

Face to Face

Update from Facilitators

Face2Face and Associated Deadlines

Subteam Reports: Canvas; ARIA Mappings; Media; Bug Triage

Actions Review

The <canvas src> Proposal

Summary of Action Items

Scribe.perl diagnostic output