WAI UA -- 5 Nov 2007

<JR> JA: Welcome to our observer this morning: David Singer from Apple

<JR> JA: And HTML5

<dsinger> hi, pleased to meet you all

<JR> http://www.w3.org/WAI/UA/2007/WD-UAAG20-20071101/WD-UAAG20-20071101.html

<jallan> DS: explaining HTML5 audio and video tags

<jallan> ... 2 layer proposal. user may have preferences...captions, etc.

<JR> DS: User has set their preferences

<jallan> .... author also has ability to provide alternate sources for content

<JR> DS: A set of sources is part of VIDEO

<JR> DS: Some sources have captions etc. some don't

<JR> DS: Some sources would have caprions burned in

<jallan> JR: set of sources. what are the attributes of the source for captions, audio description?

<JR> DS: Tag has accessibility axes...

<JR> DS: Deliberately silent on accessibility axes

<JR> DS: These are attribs to proposed VIDEO elements in HTML5

<JR> CL: Rich talks about Preference metadata

<jallan> JR: IMS global metadata, ACLIP preferences. allows user to say I want captions

<jallan> ... on author side can specify sources available.

<jallan> ... then negotiation between user preferences and author specified sources.

<JR> DS: Are prefs browser level? System level?

<JR> JA: Now done in player

<jallan> JR: what about gracefully degrading. What about 'alt' fall back content if requested content does not meet user minimum

<JR> DS: if you fall off bottom...

<jallan> DS: then render some default information (video, transcript, etc.)

<JR> DS: VIDEO, AUDIO has alt

<jallan> JA: how does the user get to the alternative content.

<jallan> all: discussion of sign language ISO code (ASL, signed exact english, etc.) dialects

<dsinger> Example. If the user has asked for

<dsinger> captions: ?want

<dsinger> high contrast video: ?don't want

<dsinger> and the video element has

<dsinger> The second source will be selected for presentation; the second would also be selected if the media attribute were completely omitted.

<dsinger> Once a candidate source has been selected, the UA must attempt to apply the user's accessibility preferences to its presentation, so that adaptable content is presented appropriately.

<dsinger> Example. If the user has asked for

<dsinger> captions: ?want

<dsinger> high contrast video: ?don't want

<dsinger> and the video element has

<dsinger> The second source will be selected for presentation; the second would also be selected if the media attribute were completely omitted.

<dsinger> Once a candidate source has been selected, the UA must attempt to apply the user's accessibility preferences to its presentation, so that adaptable content is presented appropriately.

<dsinger> Example. If the user has asked for

<dsinger> captions: ?want

<dsinger> high contrast video: ?don't want

<dsinger> and the video element has

<dsinger> The second source will be selected for presentation; the second would also be selected if the media attribute were completely omitted.

<dsinger> Once a candidate source has been selected, the UA must attempt to apply the user's accessibility preferences to its presentation, so that adaptable content is presented appropriately.

<dsinger> http://www.iana.org/assignments/language-tags

<jallan> all: discussion of 'high contrast video'

<jallan> DS: what are the a11y axis for audio and video

<jallan> JR: axis are binary preferences

<jallan> JR: author creates many versions. UA depending on axis, choose what to show to the user

<jallan> JR: 4.5 if user jumps to middle, must respect synchronization, what happens to events that were to have fired previous to the jumpin point

<jallan> 4.5 needs further discussion - see wiki

<jallan> all: discussion of embedded player, or native player.

<jallan> DS: should be finding a way to make the player full functioning inside the UA.

<OedipusWrecked> need access to backplane renderer in order to fire audio events (ACSS) in platform-agnostic manner

<JR> JA: In English?

<OedipusWrecked> me?

<JR> yes

<dsinger> the new html audio and video tags have cue-ranges which cause events to fire on entering and exiting that time range

<dsinger> is that what you want?

<dsinger> which indeed could be used to support accessibility (e.g. flipping the transcript in a parallel div)

<OedipusWrecked> one of the problems charles chen (FireVox) has encountered is that he wants to be able to fire purely aural events without use of a proprietary renderer -- would like to accomplish through OS' default audio renderer (plays event sounds, etc.)

<OedipusWrecked> i will phone in

<dsinger> yes, you will be able to instantiate an audio element in the dom from script and cause it to play

<dsinger> which is ugly in some senses (in that if it's not in the page, the user sees no controls)

<OedipusWrecked> ah, but the ultimate goal is to get CSS control of audio rendering

<dsinger> um, yes, the link between CSS and audio/video and HTML and UAAG is one to explore

<OedipusWrecked> if ACSS includes purely aural events as well as speech, would be bigger buy-in from devs and avoid nasty javascripted hacks

<dsinger> ok

<JR> akim Bridge +1.617.761.6200, conference 82941 ("UAWG1")

<JR> GR: Will be requesting Aural CSS go from Normative from Informative

<JR> GR: Ask to change from "Speech" to "Aural"

<JR> GR: Charles Chen is developing Speech CSS module

<JR> GR: Hoping to get more vendor buyin

<JR> GR: And getting more user control

<JR> GR: Will be posting to WAI x-tech

<JR> GR: Something author needs to do?

<JR> GR: Depends on browser situation...

<JR> GR: CHarles chen says optimal to use ACSS to get access to backplane renderer\

<JR> GR: Systems reluctant....but safer for CSS to do it that Javascript

<JR> GR: ACSS can be trusted

<JR> JA: UAAG fit?

<JR> GR: Mentions it because David was talking about audio elements in DOMs

<JR> GR: Wanted to point out that we are looking at an alternative

<JR> GR: Gives more control to user to suppress background sounds etc

<JR> DS: Background sounds are a big User Interface prob

<OedipusWrecked> captioning should include things marked up as such by ARIA

<OedipusWrecked> e.g. describedby and labelledby

<OedipusWrecked> there is also the HTML5 concept of "alternate" to designate alternate versions of a resource - 4.12.3.1. Link type "alternate" : http://www.w3.org/html/wg/html5/#link-type

<jallan> CL: would be nice if there was a semantic element that associates a transcript with an audio file.

<jallan> GR: explains 'alternate', for transcript of separate language of an audio file

<OedipusWrecked> HTML5 section: 4.4. User prompts

<OedipusWrecked> "alert" "confirm" "prompt" "print" included

<OedipusWrecked> HTML5 Section 4: Web Browsers: http://www.w3.org/html/wg/html5/#web-browsers -- "This section describes features that apply most directly to Web browsers. Having said that, unless specified elsewhere, the requirements defined in this section do apply to all user agents, whether they are Web browsers or not. "

<dsinger> thanks, everyone

<OedipusWrecked> key concept: "default view" - The main view through which a user primarily interacts with a user agent is the default view. The default view of a Document is given by the defaultView attribute on the Document object's DocumentView interface. [DOM3VIEWS] When a browsing context is first created, it must be created with a single Document in its session history, whose address is about:blank, which is marked as being an HTML documents. The Document must have

<OedipusWrecked> 4.1.2. Auxiliary browsing contexts

<OedipusWrecked> It is possible to create new browsing contexts that are related to a top level browsing context without being nested through an element. Such browsing contexts are called auxiliary browsing contexts. Auxiliary browsing contexts are always top-level browsing contexts.

<OedipusWrecked> 4.1.3. Secondary browsing contexts: "User agents may support secondary browsing contexts, which are browsing contexts that form part of the user agent's interface, apart from the main content area. :

<jallan> CL: Instant message: get an audio indicator but focus doesn't change

<jallan> current guideline 5 is very granular. seems outdated. perhaps should have 1 checkpoint with success criteria.

<jallan> compliance discussion

<JR> All: Dicuss GL+success criteria structure like WCAG2 and ATAG2...

<jallan> CL: 3 levels - turn off for explicit request, or off for on load, or off all the time

<OedipusWrecked> HTML5, 4.4. User prompts: http://www.w3.org/html/wg/html5/#user-prompts

<OedipusWrecked> The alert(message) method, when invoked, must show the given message to the user. The user agent may make the method wait for the user to acknowledge the message before returning; if so, the user agent must pause while the method is waiting.

<OedipusWrecked> The confirm(message) method, when invoked, must show the given message to the user, and ask the user to respond with a positive or negative response. The user agent must then pause as the the method waits for the user's response. If the user response positively, the method must return true, and if the user response negatively, the method must return false.

<OedipusWrecked> The prompt(message, default) method, when invoked, must show the given message to the user, and ask the user to either respond with a string value or abort. The user agent must then pause as the the method waits for the user's response. The second argument is optional. If the second argument (default) is present, then the response must be defaulted to the value given by default. If the user aborts, then the method must return null; otherwise, the method mu

<OedipusWrecked> The print() method, when invoked, should offer the user the opportunity to obtain a physical form of the document. The user agent may make the method wait for the user to either accept or decline before returning; if so, the user agent must pause while the method is waiting. (This does not, of course, preclude the user agent from always offering the user with the opportunity to convert the document to whatever media the user might want.)

<KFord> Hello, phong in now.

<JR> Gr: Spec also discusses auxiliary browsing contexts

<JR> GR: Pron sidebars, new panes, etc

<JR> Pron=Probablyu

<OedipusWrecked> HTML5 4.2.3. APIs for creating and navigating browsing contexts by name: http://www.w3.org/html/wg/html5/#apis-for

<JR> Guideline: 5.X Manage viewport focus

<JR> A

<JR> No A

<JR> AA

<JR> Allow configuration so that if a viewport opens without explicit user request, neither its content focus nor its user interface focus automatically becomes the current focus.

<JR> For graphical user interfaces, allow configuration so that the viewport with the current focus remains "on top" of all other viewports with which it overlaps.

<JR> Allow configuration so that viewports only open on explicit user request.

<JR> When configured per provision one of this checkpoint, instead of opening a viewport automatically, alert the user and allow the user to open it with an explicit request (e.g., by confirming a prompt or following a link generated by the user agent).

<JR> Allow the user to close viewports.

<JR> Ensure that when a viewport's selection or content focus changes, it is at least partially in the viewport after the change.

<JR> Configure opening of viewports with a full user interface chrome.

<JR> AAA

<JR> No AAA

<OedipusWrecked> HTML5 4.2.3. APIs for creating and navigating browsing contexts by name: http://www.w3.org/html/wg/html5/#apis-for

<OedipusWrecked> "In ECMAScript implementations, objects that implement the Window interface must have a [[Get]] method that, when invoked with a property name that is a number i, returns the ith child browsing context of the active Document, sorted in document order of the elements nesting those browsing contexts."

<JR> KF: What prob are we trying to solve?

<JR> JA: Starting talking about GL5 ... popup windows

<JR> JA: If popups come up with no user request...make sure confirmed

<JR> JA: BRowser gives opening warning

<JR> JA: Doesn't address lack of chrome sometimes

<OedipusWrecked> respect user preferences for chrome when opening a new instance - this includes scrollbars, address bars, tool bars, and menu (user chooses which ones in preferences)

<JR> KF: In IE7, address bar always showed

<JR> CL: In WCAG, close button required?

<JR> CL: If no chrome close

<OedipusWrecked> respect ALL user preferences for chrome interface when opening a new instance/viewport - this includes scrollbars, address bars, tool bars, and menu (user chooses which ones in preferences)

<JR> JA: Its an overlpa but I thiunk its useful for user agent to add little X to closer

<JR> CL: User agent could override author prefs

<JR> GR: Should say respect ALL user interface prefs when opening new window

<JR> JA: Scrollbars not a req for user agent

<JR> KF: Could mandate some user agewnt interface configuration

<JR> KF: Mandate lots of things in physical world

<JR> KF: In software world do same

<OedipusWrecked> could also say that an option is available via chrome, should be available in new instances

<JR> CL: You can open a modal second browser window?

<OedipusWrecked> often time limitations on popup/notification issued by UA

<OedipusWrecked> need to address disappearing pop-up mechanisms that are informational (if that is what user wants)

<JR> KF: Allow user to config all these things....min bar is apply prefs

<JR> GR: Also need to address dissappearing pop-up mechanisms

<JR> CL, JA: 2.4 Time Independent Interaction prob covers this

<JR> GR: See also 2.4

<JR> JA: KF and I saying same thing...JA: use blunt instrument and say all windows open with same chrome...

<JR> KF: Saying all windows have to have X number of attributes is harder to get people to do

<JR> JA: Saying configurable is harder

<JR> KF: IN which sense

<JR> KF: Not unsolvable prob...

<JR> KF: Harder to say all windows shall have X

<OedipusWrecked> HTML5 4.1. Browsing contexts: "Events that use the UIEvent interface are related to a specific view (the view in which the event happened); the AbstractView of that view is given in the event object's view attribute. [DOM3EVENTS]"

<JR> GR: Browsing contect part of HTML5....

<jallan> KF: shouldn't burden other users who don't need additional chrome items in new windows

<JR> KF: Ability to config 2nd window with sufficient chrome to get same user interface

<JR> KF: Since may be different on different platforms

<JR> CL: IBM has added some things due to wcag2....don't know about scrollbars...we do close

<jallan> KF: in techniques flesh out specifics for chrome - scroll, resize, close as a minimum

<JR> KF: What about some UA comes along with no scrollbars

<JR> KF: Voice browser....alwasys expected certain command...

<JR> KF: But if command wasn't available in the secondary state

<JR> KF: So in the secondary state the command would be available

<jallan> KF: in techniques flesh out specifics for chrome - scroll, resize, close as a minimum

<JR> JA: Summary....2 things....user agent allows config of chrome...and that...

<OedipusWrecked> 4.1.2. Auxiliary browsing contexts: http://www.w3.org/html/wg/html5/#opener

<JR> CL: There already is a toolbars item in UAAG

<JR> CL: Nothing about close, scrollbars or resizing

<JR> CL: Close, resizing in wcag2

<JR> JA: Reading wcag2...."resize and wrap"...

<OedipusWrecked> GJR: enable content wrapping in restricted viewports

<OedipusWrecked> GJR: if a scrollbar is unavailable or not applicable, enable content wrapping to fit the restricted viewport

<JR> JA: Scaling font size still works in some of these secondary windows

<JR> ACTION: JA to Will propose a new checkpoint covering all viewports - what configurations are needed. [recorded in http://www.w3.org/2007/11/05-ua-minutes.html#action01]

<OedipusWrecked> make sure to cross-check with HTML5 4.1.5 http://www.w3.org/html/wg/html5/#browsing especially point 5

<JR> JR: Noticed that scrollbars implicit in normative glossary

<JR> JA: Like bring all these things together into one checkpoint

<OedipusWrecked> HTML5 on Rendering: (section 10) http://www.w3.org/html/wg/html5/#rendering - "Terms that need to be defined include: sizing of embedded content"

<JR> GR: THere is an opportunity on HTML5 section 10...about rendering

<JR> GR: Talking about that this morning...video rendering stuff

<JR> JA: Sometimes better to play media in sep. window

<JR> JA: But David Singer was saying context lost

<JR> KF: Devil's Advocate: People will ask what's so bad about opening it in full client.

<JR> GR: Problem especially in Captcha

<OedipusWrecked> GJR: example of context being lost -- audio renderer opens, grabs focus, have to get back to UA in timely manner to enter CAPTCHA code; needed so that user can rewind, slowdown, etc. to compensate for background noise embedded to foil voice-recog software

<JR> JA: 2 minute warning

<JR> JA: Starting!

<JR> JR: Why does AT want to write to DOM

<JR> CL: One example is longdesc....

<JR> CL: AT could put in a link ie. do a repair

<JR> JA: Taking with raman and chen...

<JR> JA: Could use XUL to fix this

<JR> CL: Security concerns

<JR> CL: In HPR...could write back to DOM...user entered text

<oedipus> plus CSS generated pseudo-labels such as "Start Quote" q:before ( content="Start Quote: "

<JR> CL: No API at the moment for doing this - HPR did it with Javascript statements (in proces)

<JR> JA: Thinks GR's example does not write to DOM?

<JR> CL: CSS does not write to DOM

<JR> JR: Why?

<JR> CL: Because CSS supposed to separate presentation from content

<JR> GR: New CSS3 providing expansion for acronyms

<JR> JA: So even more content being created and not going into DOMs or access APIs

<JR> GR: Right needs to go in

<JR> GR: Can already do OL, UL switch in CSS

<JR> GR: An UA can't tell which

<JR> CL: OK so now we say use XML infoset....

<JR> CL: Now do we expand?

<JR> CL: Can user agent do anything?

<oedipus> DSSSL is a document tree transformation and style language in with many adherents in the SGML community. DSSSL resources on the Web

<oedipus> http://metalab.unc.edu/pub/sun-info/standards/dsssl/draft/

<JR> GR: One way of addressing this...

<JR> GR: Haven't read the draft yet

<JR> JA: Do this by saying it's UL then putting styles on LI

<JR> CL: Could put more responsibility back on user agent for things that it generates

<JR> CL: Need to broaden beyond XML infoset?

<JR> CL: Follows link "infoset"...not very helpful

<JR> JR: Wonder if "semantic information"

<JR> KF: Shouldn't it be if User Agent knows about semantic info...it should convey to DOM and any other APIs

<JR> JA: Is it semantic?

<JR> CL: Yes...

<JR> All: Example of UL, OL discussed

<oedipus> CSS3 Generated and Replaced Content Module: http://www.w3.org/TR/2003/WD-css3-content-20030514/

<JR> JA: Browser wouldn't turn UL to OL might instead go int oLI's ("1."_

<JR> CL: At the moment too HTML centric

<oedipus> 4. Pseudo-elements: "Pseudo-elements also provide style sheet designers a way to assign style to content that does not exist in the source document."

<JR> GR: In generate and replace content CSS module....

<JR> GR: We just say anything that generates content that does not exist in the source document should be included in the DOM

<JR> CL: Or content that is define by something outside of DOM?

<JR> CL: User agent writes DOM and access API

<oedipus> For example, the following rule replaces the content of <abbr> elements with the contents of the element's title attribute:

<oedipus> abbr { content: attr(title); }

<oedipus> The following rule inserts the string "Note: " before the content of every P element whose "class" attribute has the value "note":

<oedipus> P.note:before { content: "Note: " }

<jallan> JR: what about short lived information.

<jallan> ... from multimedia

<JR> CL: Generally no multimedia stuff gets into DOM

<JR> CL: Had CSS problem before with colouring characters etc?

<JR> GR: Yes asking for that, in W3C style sheed for DIFF colours are not named...they are Hex

<JR> GR: Currently not able to search for hex - need named colours

<JR> JA: Range of hex?

<JR> GR: Burden is on ATs to recognize by hex and RGB

<JR> KF: user can understand RGB?

<JR> GR: WCAG rule is do not rely on color alone

<JR> GR: Will report back from Willy Walker (orca)

<JR> ACTION: CL to To broaden the defintion in 6.1 beyond infoset to cover thing like CSS [recorded in http://www.w3.org/2007/11/05-ua-minutes.html#action02]

<JR> CL: Another Problem is that no is complying with Write Access...

<JR> CL: Do we need to keep this?

<JR> KF: Not sure I would want that?

<JR> JA: And if no problem

<JR> CL: In UAAG 1.0 if you can change as user, shouldn't AT be able to do the same thing

<JR> KF: Can do SetValue etc. to MSAA

<JR> CL: So MSAA has some write access by ATs

<JR> KF: At the rendered content level

<JR> KF: not at DOM level

<JR> KF: AT can't check box at DOM

<oedipus> IA2 perspective: http://www.linux-foundation.org/en/Accessibility/IAccessible2/Minutes/20071002

<JR> KF: IE gets info back from MSAA to check a checkbox

<oedipus> MSAA's IAccIdentity::getIdentityString can be used to get a unique ID for an object.

<JR> KF: FF's MSAA implementation prob the same

<JR> GR: I want to let people know about mapping between Icc2 to ATSPI

<JR> KF: ATSPI has a document interface

<JR> KF=GR

<cklaws> http://live.gnome.org/GAP/AtkGuide

<oedipus> http://www.gnome.org/~billh/at-spi-new-idl/html/html/classAccessibility_1_1Document.html

<JR> JA: 10mins till ARIA discussion

<JR> CL: In 6.4, note several GLs should be combined....

<JR> CL: But in general some defs needed....

<JR> CL: Programmatically accessible...

<JR> CL: Add a definition of "programmatic access" and "programmatic read access" based on the WCAG definition of "programmatically determined", such as "exposed to an assistive technology through an accessibility API that is supported by commonly available assistive technology."

<JR> JR: Where we stand on access API's vs. DOM?

<JR> CL: We should consider combining 6.1 - 6.4. The techniques sections describe why there are separate guidelines - separating what info from how to provide, and HTML vs DOM vs non-HTML vs graphical content. But there is a lot of duplication. Need to decide if UAAG should require both DOM and accessibility APIs, especially in light of the fact that ATs don't want to get ARIA info out of the DOM.

<JR> CL: Are we really trying to say EVERYTHING needs to go to both?....

<JR> CL: Or just out of accessibility API?

<JR> CL: Believe Scr Readers use DOM only as backup today.

<JR> KF: Hold on...but in practice those AT prods built on DOM are happy...

<JR> KF: Divergent paths

<JR> CL: In Process goes to DOM, out of process goes to access API

<JR> JR: So new product...would we say both?

<oedipus> GJR: HTML5 VERY DOM-heavy; scripting crucial part of Web Forms 2.0 and XForms

<JR> CL: Yeah - in process, out process is an effeciency thing

<JR> KF: Have to do both!

<JR> CL: Agree

<JR> CL: But we've never asked for full buyin on accessAPI

<JR> CL: Do we want ability of ATs to use either?

<oedipus> +1 to CL's in-process / out-process efficiency

<JR> JA: Lean towards both due to efficiency

<JR> JA: In section 508....lots of accessAPI stuff

<JR> CL,JA,KF,GR: Should be to rely on only one.

<JR> CL: Won't be one to one....

<jallan> JR: should be able to get all a11y information needed from either the DOM or API depending on efficiency

<JR> CL: AccessAPI...one call might make 100 DOM calls

<JR> CL: But can get that info

<JR> GR: Also will be essential for compiound docs

<JR> KF: AccessAPIs job will be info amalgamation

<JR> CL: THird point on 6.1...

<JR> CL: Don't know how to define Adaptive Interface

<JR> JA: We resolved to move forward on DOM and accessibility API

<JR> JA: Maybe tomorrow morning we take up 6...

<JR> JA: Let's stop here

<JR> GR: Regrets for 1.5 to 2 hrs tomorrow

<JR> KF: Will also need to be later tomorrow.

ARIA questions....

<JR> 1. Is there a way that UA can find out about active area refresh rate?

<JR> 2. Is it reasonable for the user to override the authors timing intervals.

<JR> 3. UA group concerned about conflicts between content-based and UA-baes keyboard navigation and control

<JR> 4. How do you envision the ARIA experience for non-AT users? Navigating live regions, etc.?

<JR> 5. Define "adaptive interface"?

<JR> CL: PLUS Do they see a problem requiring full implementation both DOM and accessAPI for all access info.

<jallan> CL: there is no DOM on Linux. Yet, UAAG requires.

<jallan> then perhaps we need either DOM or a11y api

<JR> JA: Back to ARIA questions...

<JR> JA: Sort of thinking of live regions as frames....

<JR> CL: Imporant thing is getting info on changes

<oedipus> mozilla live regions example (alerts from buttons): http://www.mozilla.org/access/dhtml/alert

<oedipus> need FF2/FF3 to test

<JR> KF: Up to AT product...

<JR> KF: Ex. of page listing who is talking

<JR> KF: UA not responsible, AT is

<JR> CL,GR: Bug in dhtml example

<JR> JA: Title on live regions?

<JR> CL: No but needs role?

<JR> GR: PF asked for title to be predefined attribute

<JR> GR: aria title could be added to page, other stuff

<JR> GR: To make more universal wanted to add to XHTML role module

<JR> JA: BAck to my issue...

<JR> JA: THinking of page with stock ticker...

<JR> JA: No title on my stock ticker???

<JR> KF: Agreement with Jim?

<JR> CL: Are we equating with a live portlet?

<oedipus> http://esw.w3.org/topic/PF/ARIA/BestPractices/LiveRegion

<JR> CL: In ticker maybe it's just the price.

<JR> GR: Can used owned-by, described-by, etc

<JR> JA: So title could be used to title live area?

<JR> JA: Brings up issue of when titles collide....link with title plus acronym expansion

<JR> JA: SHould live regions be explicilty and programmatically identified?

<oedipus> http://esw.w3.org/topic/PF/ARIA/BestPractices/LabelDescribe

<JR> CL: Back to Q3...

<oedipus> http://esw.w3.org/topic/PF/ARIA/BestPractices/Navigation

<cklaws> *ckl: 2. Establish and document how the user agent resolves key binding conflicts between the user agent user interface, user agent extensions (e.g plug-ins), HTML elements (i.e. accesskeys), and ?JavaScript functions (i.e. keypress events). If a keystroke is not defined by the user agent user interface,the user agent should pass it on to the user agent extensions, HTML elements, then ?JavaScript functions, in that order. Not sure we should be this prescriptive - j

<JR> CL: Missed anything - not sure about extensions?

<JR> JA: Asked about that...something in Firefox accessibility but no checks

<oedipus> http://www.mozilla.org/access/dhtml/spreadsheet

<JR> GR: Othe day - looking at ctrl-alt-m...kept getting OS's minefield

<JR> KF: Struggled with this on IE

<JR> CL: But should we be prescriptive?

<JR> GR: One problem is keys are often platform specific

<JR> ok...so we are in PF group now...

<oedipus> same dialin info?

<JR> no unfortunately....

<oedipus> ok -- i'll listen to IRC

<KFord> is there phone or should I just do IRC?

<oedipus> i don't think a phone is available -- i think they are in the PF meeting room

<JR> we will be meeting in this IRC channel

<chaals> oedipus, correct

<scribe> scribe:tlogan

<chaals> oedipus, should we try to skype you in?

<JR> Gregory I need your skype name.

UAG Meeting

<KFord> If you can do a skyple conference I'm kellford1967. If not IRC is fine for me.

Jim: First question: Do they see a problem requiring full implementation both DOM and accessAPI for all access info
... Originally UAG written heavily into DOM, in last five years moved to API world
... Out of process technology like screen reader, you want to use accessibility api because more efficient
... In process you want to use DOM
... Thinking of user agent having to do DOM and API implementation so that no matter where you are you can get the information

RS: Depends on OS whether you are in/out of proc. There are platforms that won't have accessibility APIs. For example PocketPC you can get to the DOM but can't get to MSAA because not available on platform
... Some ATs are extensions to browsers like FireVox which accesses DOM directly.
... You must support DOM because its one area that W3C controls. At a minimum support DOM

CL: To support DOM does that mean you have to support both a way to access it in proc and out of proc means to access it?

RS: I don't think it matters, an app can get in process. Took ATVs awhile to figure out how to get in-proc with com app but they do it now.

Jim: Follow-up, I thought Firefox on Linux doesn't have a DOM

CL: They do but they don't have public API to access it

CC: They have DOM Inspector

RS: DOM API can get access to on Linux

CL: AT like Orca can't access DOM, its only available from javascript

RS: Reason JAWS goes into IE for information is because MSAA isn't rich enough for what they want to do. What we did in Firefox we added IAccessible2 on top of MSAA to get everything you need.

CL: We don't need to know state of the world. What is recommended techniques/guidelines for future

CC: AT-SPI has a tree, and it is in parallel with DOM tree on Linux. You can get to it. RS:But you cannot make DOM API calls through AT-SPI on Linux.

CL: Today UAG says you must require DOM support but it does not require support of platform API

RS: If it's available then API must be supported. Did demo of JAWS with Firefox. Went in ran it and everything spoke as if desktop application.

resolution: Requirement is DOM first, then must do platform API support

JA: We will require both

RS: 508 is going way of requiring API support
... Caret support you will need API support for rich editing scenarios on web

JA: UAG we want to give user as much control as possible over what is going on. Time-independent interaction, you know timing with html refresh. With ARIA your refresh rate for whatever region/thing is hidden in javascript somewhere and user agent can't go in and find out refresh rate

CC: Active area something controlled by user interaction. Say we get error refresh. Monitoring of refresh is possible, but exact time is not able to be specified

RS: Scripted applications you have to separate what is a user generated change to content vs what is run in the background
... Things run in the background. Looking at for ODF, if we know something generated in background then we should be able to control the timing of this. If there is refresh need to be able to slow that down.
... Question I have, how much do you want to push off to a WCAG discussion?

JA: This is one area where we have overlap with WCAG. Some timing author can control, me as user I should be able to tell author of content this is too fast for me slow it down

*going to try to reconference in gregory*

AG: Distinction is made between genuine realtime data and criticality information (serious exception, needs to get user's attention regardless*
... Charles has criticality covered through channels. Main channel and Notify channel to allow agent to do filtering. In context of chat application it is easy to get drowned with more input than you can keep up with. Chat mode that is working with ARIA can be browsed/reviewed independent of what is happening in realtime.
... ARIA doesn't have generic buffering policy. But chat scenario available demonstrates different modes available
... Advertisements want more attentions. They are paying for marketing to site. Going to be hard to get this deployed in marketplace given that ads are paying to be brought to user's attention on site

CC: Programatically you know where all the live regions are. Get a list of where live regions are and change them on the fly

AG: Politeness attribute can be set. Politeness should be static property available.

CC: It should be dynamic
... Right now coarse level of control where you can turn all notifications off
... Something being worked on for firevox interface
... Going to be for power users to configure

AG: How do you brief user of what this region is

CC: If there is title for region you give them that, if not then give them text

AG: Reasoning from web security context, if they recognize that ad comes from Nike. They could use that information to set politness level

JA: Started talking about timing and then went into politeness. Are timing/politeness related?

RS: In ARIA we can mark an area as a live area. We can tell you the politeness level of how you process that section. If its "rude" you have to interupt user right now with whatever it is. if "polite" then in that scenario the app can still interupt quite frequently. It would be helpful to be able to set how often more polite interruptions should occur

<KFord> The skype call dropped out. But the audio isn't so clear so I'll just stick with IRC.

AG: JA thinking about timing. Refreshes that happen too quickly to be spoken.

CC: Have two separate queues, queue it up in main or queue it up in notify. main queue items with high priority will step on notify items with low priority

<clc4tts> Correction: You have two separate queues. Things in the main queue will only step on things with lower priority in the main queue. Things in the notify queue will only step on things with lower priority in the notify queue. However, if things of equal politeness are in both channels, then the one in the notify channel will go first.

RS: Live property tells you how significant a change is. *group reviews definition of live property*

CL: Does Live indicate the whole area even if a small section may change?

AG: You should put the property on the small section where changes occur

CC: Polite - if user not doing anything. They are just sitting there. Then AT will present those
... Assertive - If user is doing something, as soon as they are done, speak the notificaton. In FireVox you can tell it to read whole page. Polite wait until whole read page is done. Assertive wait until sentence is done. Rude: Stop user action immediately and add notification
... Assertive typing in form field and they make a mistake. As soon as user pauses it knows to fire notification

TL: How do you define pause?

CC: Up to AT, in FireVox uses 3 seconds

AG: Exact handling not defined in spec. ATs can define whats good for user experience. Timing of breaking into queue is known to AT rather than up to author of content
... No, there is not a way that UA can find out about active area refresh rate

JA: Question: Is it reasonable for the user to override the authors timing intervals

RS: Do you want to correct as UAG if author didn't follow WCAG?

JA: There are so many checkpoints in UAG that say allow configuration.

AG: If DOM exposes read/write interface. Then AT can come in and implement user's policies.

CC: AT out of process can still know where event is coming from. Can always map events from certain section to off

AG: User should be able to set disposition of regions
... Screen refreshes at dangerous rates. Should look at option of having a check on authors so that they can't specify these dangerous levels

JA: User of non-screenreader should be able to specify work related to stopping animations etc.

AG: RS this relates to work you are doing in Fluid
... Control of embedded objects. Does necessary control flow down to embedded objects

JA: Question 3: Concerned about conflicts between keyboard nav and control. Establish and document how UA resolves key binding conflicts between OS, AT's, user agent user interface, user agent extensions, HTML elements(ie accesskeys) and JavaScript functions(ie keypress events)
... If a keystroke is not defined by user agent user interface, the user agent should pass it on to the user agents extensions, HTML elements, then Javascript functions in that order. Not sure we should be this prescriptive?

<oedipus> i updated minefield between the 2 times -- OS stole keystroke in last night's build, not with today's midday build

CM: OS gets keystroke first. If OS doesn't keep it, passes it on to UA. When UA gets it, should they hand it to javascript? Or should they respect UI conventions first? I would argue insane to hand it to javascript first. Every web developer would argue they should get it first (which is what happens today) which means any web app can screw up any browser ui.

CL: We should recommend that UA handles it first for their own UI before giving to javascript?

RS: We started doing device independent events

AG: I think this is pretty fundamental and parallel to web security context problem. At present we should go for what we think is right

<oedipus> bang important needs to belong to user

AG: Think we need three layers.

CL: What about extensions

CM: Browsers handle extensions. Extensions are part of browser level functionality
... Where scripts come in is tricky. I think generally bad design for script to set up its own keyboard reading stuff. It assumes a keyboard and that is just wrong because not all scenarios have one
... AccessKey mechanism in HTML lets you trigger things on old cell phones, easy to make use on any platform. Make that work
... This will be huge shift for web developers. As a browser-maker we are loathe to break web applications because that is fastest reason for people shifting away from user agent
... Browser provides AccessKey and browser tells you what key activates. If i have HTC phone with four way rocker and 2 buttons i can call up accesskey mode and get a list of the accesskeys (tab down to one i want and click it) but on phone i can't generate key event to run a function
... If author users javascript to make up keyboard navigation, browser is at edge of capability of being able to configure itself to make it work

CL: Thats the basis of all these new widgets right?

RS: Why don't we move to more declaritive model of events for authors

JA: Do we get prescriptive about this? What can we say in guideline/checkpoint about this

CM: Say "browser should ensure that browser ui functions are not interuppted by scripts"
... We aren't going to be able to follow this though. We won't be able to make this change quickly because it will require shift in web developers
... if you don't prescribe it, then there will be no goal. this is an aspirational goal.
... still should prescribe it, it wont be quick, but if we get top tier providers doing it the right way. opera can't break facebook, airline site, bank site, most games etc. i can afford to break widgets and other classes of authors i can afford to break. once the big guys doing it the right way, i can tell the people i break to follow their lead

CL: We had guidelines for how to expose what accesskeys exist. we can't do that for javascript

AG: we were discussing a way this morning that you could. if chris made table of keyboard shortcuts in DOM with known role. then it could be read. point is there are ways that this info could be exposed

CM: Easy way is make things access keys. Second way have a role of "access key" in ARIA and have a prescribed way to expose it

RS: nice thing of ARIA it doesn't break existing browsers

<oedipus> http://developer.mozilla.org/en/docs/ARIA_to_API_mapping

CB: Want one access key to open a currently focused mail item. How would i declare that?

AG: may require attribute template. i think you could define generic calling profile for that method

<oedipus> "When a direct API mapping is not possible, a role string is used. In ATK/AT-SPI the role string is exposed via the "xml-roles" object attribute (in fact the original ARIA role string is always exposed there)."

JA: Question: ARIA says Define mapping role/state/properties to an adaptive interface. what is adaptive interface?

RS: Might want to go in and change spreadsheet to headers.

CM: Wii is great example. We rerender interface to match the wii controller

AG: An adapative interface is any interface that has been modified to improve usability for the user

MM: Adaptive interface vs User Agent, do not see a big distinction

RS: With ARIA you get more semantics, so User Agents will be able to adapt more
... If everything was divs and spans what would you do with it?

CL: Adaptive is more about rendering than the API? It was confusing in documentation, the distinction

AG: We want both to be supported

JB: ARIA does lots of great stuff for AT users. What is value-add for non AT user

CM: Take ARIA information to adapt page to a phone. You can change the controls. Able to swap in different controls for phone because it has that awareness

CC: Live regions could be used to provide more granularity over visual notifications to user

RS: With main content defined through ARIA you can reposition main content on the fly rather than having to use skip navs
... XHTML Access Module displayed on screen.
... HTML we could have Role=Access and then have propeties defined for it that XHTML Access Module has
... You can suggest key mapping with this element. Opera could tell you that you can't have that key binding so you can specify a backup

AG: key property is author's nomination for that key

JA: if you have 4 key bindings. then you put 4 access modules on one div or span

AG: correct

CL: HTML5 hasn't looked at this?

RS: They kind of. They have something called a Command element. They don't have support for roles

JA: We can reference Access Element in techniques?

RS: Its a proposal, not a current solution

AG: we have allies for getting intent based events implemented.

RS: maybe charles would want to do an add-on for firevox related to this

CC: need actual test cases to implement this feature

RS: we could create some

AG: CB has the application

CC: what tells you what function actually is

RS: you can use title attribute

CC: you have more propeties than the 4 in the spec

AG: since id, title are globals they are already available on the Access Module element

CC: what about modifier keys

RS: we could do that, need to discuss

CC: there are times when you don't want modifiers, and times that you do

CM: as browser want to offer least pain to activate a control

CC: as page developer, i might want to pick single key

CM: which one are you going to pick given most are already taken

CC: should let user author specify they want to use modifiers. then let browser do remapping if they don't want to honor initial request

CM: currently browser lets author do whatever they want.

CC: let browser specify preferences

CM: should let users define how important specific functions are

CC: author provides mnemonic. go to google reader. to do feed reads press "SHIFT+N" you will navigate down feed tree. it doesn't matter if its CTRL+N or something

CM: want way to have navigation control be a navigation control

CC: let author say "SHIFT+N" "SHIFT+P" for feed traversal "N" for article traversal. take all of that. on mobile phone don't have these keys. should do logical mapping

CM: semantics in roles, hint that both actions are traversals then they should use similar interaction model

<oedipus> mnemonics vary with natural language

<oedipus> what works in lang="en" as a mnemonic won't help if lang="pt"

<oedipus> need a way to identify abstractly before mapping can be accomplished

<oedipus> accesskey tokenization?

<oedipus> what do stored prefs do when move from 101-key UA to 10 or 12 key UA?

<JR> We're back in our room now.

<JR> We're wrapping up

- DRAFT -

WAI UA

5 Nov 2007

Attendees

Contents

ARIA questions....

UAG Meeting

Summary of Action Items

Scribe.perl diagnostic output