W3C

- DRAFT -

Protocols and Formats Working Group Teleconference
12 May 2014

Agenda

See also: IRC log

Attendees

Present
Rich_Schwerdtfeger, janina, +1.603.882.aaaa, joanie, Michael_Cooper, Jon_Gunderson, +49.322.110.8.aabb, Cynthia_Shelly
Regrets
Chair
Rich
Scribe
MichaelC

Contents


<trackbot> Date: 12 May 2014

<richardschwerdtfeger> meeting: W3C WAI-PF ARIA Caucus

<richardschwerdtfeger> the agenda is here: http://lists.w3.org/Archives/Public/public-pfwg/2014May/0045.html

<scribe> scribe: MichaelC

WAI-ARIA 1.1 Heartbeat Draft

rs: how are we to do a heartbeat publication?

jc: in addition to IndieuI?

js: yes we should

jc: been focusing on IndieUI, not a lot done on ARIA

mc: UAIG is more ready for publication

FPWD under its new name

ARIA heartbeat either later or at same time

js: would be good to exercise the new process with GitHub

rs: status on IndieUI?

jc: User Context pretty much ready for FPWD, about a week to ready for Events

js: planning a CfC on that this week

rs: so can we publish ARIA at end of month?

jc: sure, we could

lots of formatting changes

though nothing to prevent publication

mc: publishing moratorium 6 - 11 June 2014

otherwise we can target Tue / Thur with a few days advance notice

rs: so end of May?

mc: ok for me

jc: what are you looking for in it?

rs: e.g., role=none

jc: have mostly put in editorial changes

but could throw in some more stuff by end of month

mc: substantive edits need time for group review

plan for that in timeline, or defer them to next heartbeat

mc: June 5 last pub day before moratorium, need doc ready by June 2

(May 29 for UAIG FPWD)

need at least a week review time, so edits done by May 26

so two weeks from today

jc: some edits, @@

some would be good to get in, others can wait

think IndieUI is focus for this week

rs: also setsize

jc: that one is just editorial I think

rs: can pass on table for now

<jcraig> rs: posinset and setsize on tabs ACTION-1352

<jcraig> action-1352?

<trackbot> action-1352 -- James Craig to Patch issue-576: add aria-posinset, and aria-setsize to tab role -- due 2014-01-31 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1352

role=none and action-1352 priorities

hint?

jc: not critical

none is a bigger edit than it seems

jc: would like to migrate issues to bugzilla so can process

mc: we already have in theory for public comments, just haven´t been comments yet

haven´t decided what to do for tracker

jc: action prioritization would be easier

need some components

mc: didn´t realize that, send me what you need set up

ss: can use role=none yet?

rs: need UA implementation first

jn: does IE ??

cs: IE takes first role

jn: IE won´t backport, right?

jc: @@

cs: so role cascade is high priority feature

jc: yes

starting to do stuff that need that

cs: will pass that along

didn´t know it was a priority, may surprise IE team

jc: priority because other browsers calculate

cs: didn´t know was used in the wild

jc: not yet but we set up for that in 1.0 and expect to use soon

<jcraig> role="foo link"

<jcraig> role="specialbutton button"

ls: there was a proposal to allow roles to concurrently hold multiple roles

jc: this is just about supporting the first recognized role

cs: multiple inheritance propose patterns in ARIA 2

Issue 587 Request to have aria-selected on more roles and also to have

an aria-active state

<jcraig> http://www.w3.org/WAI/PF/aria/complete#host_general_role

<richardschwerdtfeger> https://www.w3.org/WAI/PF/Group/track/issues/587

issue-587?

<trackbot> issue-587 -- Consider allowing the aria-selected state on any focusable element, or add a new attr like aria-active or aria-current -- open

<trackbot> https://www.w3.org/WAI/PF/Group/track/issues/587

<jcraig> The first name literal of a non-abstract WAI-ARIA role in the list of tokens in the role attribute defines the role according to which the user agent MUST process the element. User Agent processing for roles is defined in the WAI-ARIA User Agent Implementation Guide [ARIA-IMPLEMENTATION].

rs: issue is whether to allow aria-selected on any focusable element

have determined can´t do on radio button, will need to be sure to exclude

on tabs, is there active or selected?

cs: there is focused, selected

what is active?

rs: e.g., in tablist with active tab, and you go into panel

jc: applies e.g., to page you´re currently on

cs: don´t have now, will suggest it

concerned ¨active¨ will be confused with CSS pseudoclass

<jcraig> +1 to "current"

maybe ¨current¨

<jcraig> "Current thread is suggesting aria-current over aria-active. James Craig, 24 Apr 2014, 08:00:44"

<jcraig> FWIW, I'm worried about overloading "selected" which currently implies user-selectability to mean something else on readonly or static interface elements like links. The term "selected" is already misused or overused. It's commonly confused with other similar terms like focused, activated, etc. ARIA 1.0 at least uses aria-selected consistently. If we change that, I think we'd b introducing more confusion for authors.

<jcraig> There is also PFWG-ISSUE-504 to consider, where where we want to change the taxonomy so that aria-selected is NOT allowed on radio buttons to avoid author confusion. Radio buttons currently inherit aria-selected from the option role and inherit aria-checked from the checkbox role, so authors frequently use aria-selected="true" on radios when they intend to use aria-checked="true".

<jcraig> Despite my objections to using aria-selected on *any* focusable element, I think the idea of indicating which link points to the current page is sound, and am generally supportive of adding either a new attribute, or allowing aria-selected on more roles such as link.

<jcraig> James Craig, 20 Jun 2013, 17:27:37

jg: current sounds good to me

jc: ^

<Zakim> jcraig, you wanted to mention my comment listed in the issue

<jcraig> issue-504?

<trackbot> issue-504 -- radio shouldn't have aria-selected -- closed

<trackbot> https://www.w3.org/WAI/PF/Group/track/issues/504

<jcraig> action-1338

<trackbot> action-1338 -- James Craig to Patch issue-504, then assign to cooper for the taxonomy doc -- due 2014-01-30 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1338

rs: would like to remove aria-selected from radio; requires taxonomy change

jc: we already resolved that

jg: thought it was the visible tab panel is selected

what if something is ¨current¨ but not visible? error?

ls: tab panels work well with current semantics

aria-hidden works well

do we need to add semantics that require authors to know fine details?

jc: this is not so much for tab panels in spite of example above

<richardschwerdtfeger> “In either case, authors SHOULD ensure that a selected tab has its aria-selected attribute set to true, that inactive tab elements have their aria-selected attribute set to false, and that the currently selected tab provides a visual indication that it is selected. In the absence of an aria-selected attribute on the current tab, user agents SHOULD indicate to assistive technologies through the platform accessibility API that the currently focus[CUT]

steps in a process

<richardschwerdtfeger> is selected."

ls: aria-selected used now

jc: right; have to be clear were we need this and where we don´t

we don´t need it on tabs

ls: ok

<richardschwerdtfeger> http://www.w3.org/TR/wai-aria/roles#role_definitions

rs: <brainstorms roles it might go on>

jg: what would AT do?

<jcraig> https://www.yahoo.com

<jcraig> home, mail, news

<jcraig> <a href="/">Home</a>

<jcraig> <a href="/" aria-current="true">Home</a>

jg: don´t need an attribute to know that

jc: for regular HTML, true

<jcraig> <button role="link" aria-current="true">Home</button>

<andrewlarkin> +q

but for roles like above, needed

al: this is important for single-page web apps, or anything using frameworks like backbone where all links link to the same page but load specific content that represent a single "page"

link to same page and load additional content

bg: in a process, a CSS class indicates which step is current

no accessible mechanism except off-screen text

an ARIA attribute that applies to focusable and non-focusable elements would

ls: ??

bg: @@

AT read label of focusable elements, but not all elements

jc: whether text explications get read can depend on situations

this indicator would be in addition to label

bg: yes

<jcraig> <ul>

rs: issues around aria-label anyways

cs: also have graphical indicators of current step

<andrewlarkin> +q

don´t want users to have to jump to the image to find it out

having the property would help out

might modify focus order etc. in response

<jcraig> <ul>

<jcraig> <li>Step 1: Login</li>

<jcraig> <li aria-current="true">Step 2: Add Payment Information</li>

<jcraig> <li>Step 3: Profit</li>

<jcraig> </ul>

<jamesn> we have something we call a train for this http://www.oracle.com/webfolder/ux/applications/fusiongps/patterns/content/guidedprocesses/singletask/nonsequentialhoriz/index.htm

jg: what is the problem we´re solving?

alternative to title?

@@

rs: when you´re in a context

(not selecting for an operation)

know where you are

jg: so user wants to query what the current step is, regardless of where they´re on the page?

or be told current location when landing on a widget?

rs: yes

current step, or current location within context

it´s like a point of regard that you can refer to

jc: though not user´s point of regard, which is like focus

this is more of a progress indicator

cs: this is indicated visually all the time

jn: we call them trains

cs: or breadcrumbs

jc: e.g., steps 1 - 5, which of those are you on

jg: what will user do with this?

query for ¨what is current¨?

AT announces to you at some point?

cs: suggest a specialized interaction

jg: how will user know?

cs: by reading documentation

ls: if aria-label isn´t working well, maybe we need to fix support for that rather than mint something new

don´t make new semantics when there are existing semantics not supported well

that´s a difficult path

is this part of the breadcrumbs role?

rs: it´s like it

cs: more general

ls: you´re still missing context

being told what is current, there is still visually available information you´re not getting

rs: it´s not @@

jc: this is different from breadcrumbs, an overly used term

this is more of a status of where you´re at in a process

though has more general uses

al: in one use case, a link to a page indicates it´s current

in another, @@

if aria-current goes on one element, how will users reference it?

can there be a controls-like relationship to indicate what element is related?

cs: <talked over>

jc: one that changes over time?

al: there can be steps that are visible but non-active ones displayed differently

<Zakim> jcraig, you wanted to indicate wre and to indicate we're not deciding UI, that would be determined by AT vendors. We're just defining the API. It could be that this would be

jc: sounds like progress bar

al: when user navigates to page, how can they figure out the current part from the rest?

right now have to hide the content

will current help?

jc: back to user expectation - this is new, so there aren´t current expectations

need AAPI to communicate it, and allow AT to solve the UI issue

jg: sure, but we need to understand what happens when current not present

right now I hear ¨AT users have to learn another command¨

cs: there´s no way to do it at all now

jg: can be put in the title, users are used to that

cs: if I´m anywhere in page, hard to get back to the current step

jc: there could be a user command, or AT could just announce, or some new interaction pattern

right now, only get if crammed into label

cs: or both

passive and active retrieval

rs: more than step in process

e.g., nav

on a page, there could be multiple current items, within their respective contexts

<jamesn> +1 to allowing multiple

if an element has aria-context, how is it defined what is the scope of context within which it is current?

<Zakim> joanie, you wanted to ask if this should be used in the context of a landmark

jd: for AT, we need it easy for AT to find

if it´s a state, state change events help

if it´s limited to landmarks, limit to a kind of landmark so AT know where to look

so I lean towards a state with some extras to help out ATs

rs: so for nav, would want to know current item from nav?

jd: from anywhere

there´s a ¨where am I?¨ command - AT needs to know how to answer that question

<jcraig> joanie: wants notifications when this state changes

rs: so a relationship on container to element?

<andrewlarkin> +q

jd: need a way to connect, and way to announce

al: relationship between what is currently active, and @@

<Zakim> jcraig, you wanted to say this is an API implementation detail, not to be defined in ARIA until the APIs support it

jc: maybe every element that is current should have a controls relationship to indicate scope

re easier for AT to detect, that´s for the AAPI implementation to sort

e.g., could support a notification event

but shouldn´t be encumbent on authors

<joanie> +1 to jcraig's statement. I just want to know. I don't think it has to come from authors. :)

cs: needs to be in the core mapping guide

jc: yes, after AAPIs have a hook for it

we have chicken / egg here, want to define in ARIA to push AAPIs, but not so useful until they support

cs: we can put this egg out there

can probably get in, and will get feedback

<Zakim> MichaelC, you wanted to suggest a property on a container pointing to ID of current descendant could define both scope and current item

jg: will anything current have to have accessible name?

<joanie> I agree with jgunder in that we need to define what it applies to

<confusion over tabpanel>

cs: case in point, don´t want to confuse authors

rs: e.g., nav list with toggle buttons

if it´s a state, browsers have to do notification

mc: maybe a property on a container pointing to ID of current descendant could define both scope and current item

jc: if we did that, would have to be sure to get the relationship direction right

container references descendant, or current element references container

need to think through the use cases to help decide which of those to do

e.g., for a link, body could point to current link

but if you´re on a train, wanting to know what current step is, container reference is preferred

rs: having it on the current element itself allows attribute selector to style

<jcraig> rs: attribute selector [aria-current] { } would be nice to use.

<jcraig> mc: aria-currentfor="IDREF"

rs: would using aria-controls overload it?

mc: think so

using on the current object referencing container for which it´s in scope allows styling, and provides boolean-like ¨I´m current¨

jc: AAPI could calculate inverse relationship

cs: would like to be able to use the DOM hierarchy

let UA calculate for and by

jc: would aria-currentfor=¨¨ work?

cs: I guess

think it´s better for UA to calculate because authors can mess that up

<Zakim> jcraig, you wanted to say that'd be vague

jc: could be ambiguous

aria-current=true would tell browser is it´s current for

but ambiguous what it´s current for, could be whole page, or some lower container, list, etc.

could support currentfor (IDREF) plus currrent (boolean)

don´t think a non-referencing attribute would be clear enough

cs: would like to try

we would define a set of roles that are presumed container of a current descendant

jc: so need a ¨current computation algorithm¨

rs: makes me nervous

cs: asking authors to do too much makes me nervous

rs: miscalulation could lead to big problems

jc: not sure authors would understand an implicit relation to ancestors

cs: maybe; we could draft it both was and test via authors

jc: let´s think through these ideas

mc: think current and currentfor together confusing

prefer currentfor=¨¨ if we want calculated relationship

rs: do we agree to do something?

cs: not sure the syntax is right, but agree need something

rs: think about it this week, we´ll pick up next week

cs: also annotations

<richardschwerdtfeger> RRSAgen, make minutes

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.138 (CVS log)
$Date: 2014-05-12 18:33:22 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.138  of Date: 2013-04-25 13:59:11  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/are you for in it?/are you looking for in it?/
Succeeded: s/June 29/May 29/
Succeeded: s/backport right/backport, right?/
Succeeded: s/this is important for apps/this is important for single-page web apps, or anything using frameworks like backbone where all links link to the same page but load specific content that represent a single "page"/
Found Scribe: MichaelC
Inferring ScribeNick: MichaelC
Default Present: Rich_Schwerdtfeger, janina, +1.603.882.aaaa, joanie, Michael_Cooper, Jon_Gunderson, +49.322.110.8.aabb, Cynthia_Shelly
Present: Rich_Schwerdtfeger janina +1.603.882.aaaa joanie Michael_Cooper Jon_Gunderson +49.322.110.8.aabb Cynthia_Shelly
Agenda: http://lists.w3.org/Archives/Public/public-pfwg/2014May/0045.html
Found Date: 12 May 2014
Guessing minutes URL: http://www.w3.org/2014/05/12-aria-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.


[End of scribe.perl diagnostic output]