User Agent Accessibility Guidelines Working Group Teleconference -- 09 Sep 2010

<trackbot> Date: 09 September 2010

<oedipus_away> scribe: Gregory_Rosmaita

<oedipus_away> scribenick: oedipus

<kford> we should take up the browser item last, I think that is item 5.

Announcements

JA: take up agendum 5 first -- HTML WG deadline for bug submission 2010-10-01 http://lists.w3.org/Archives/Public/public-html/2010Sep/0074.html

GJR: full details of HTML5 timeline: http://lists.w3.org/Archives/Public/public-html/2010Sep/0074.html

http://www.w3.org/WAI/PF/HTML/wiki/Access/pf_requirements

http://www.w3.org/WAI/PF/HTML/wiki/Talk:Access/pf_requirements

Accesskey Replacement Requirements

<inserted> scribenick: kford

Gregory: I need to work with folks to get an access key replacement.
... There were 9 requirements from PFWG and I've logged various replies on a wiki page.
... I need to finish the process of going through 4.1 and making sure it reflects all the requirements that PF has.
... Does that seem like a reasonale process?

JA: Yes.

Gregory asks if there other places in UAAG to look.

Group replying we do not think so but can check again.

Group now thinking need to look under some parts of perceivable section.

Gregory: I'll break this down and hopefully we can discuss this next week.

<Greg> Here is the SC about presenting shortcut keys: 4.1.6 Present Direct Commands in Rendered Content: The user can have any recognized direct commands (e.g. accesskey) in rendered content be presented with their associated elements (e.g. "[Ctrl+t]" displayed after a link whose accesskey value is "t", or an audio browser reading the value or label of a form control followed by "accesskey control...

<Greg> ...plus t"). (Level A)

Verbose Descriptor Requirements (longdesc versus enhanced native describedby)

<oedipus> longdesc: http://www.w3.org/WAI/PF/HTML/wiki/Verbose_desc_reqs - HTML A11y TF to determine if shoud be logged as issue (since there is no mechanism currently in HTML5 that provides for a verbose descriptor) or as a bug

Gregory, as everyoe knows, longdesc has been removed from HTML5. Currently, no replacement for a verbose descriptor.

Gregory: Would be good if people could review requirements.

<oedipus> 1. A programmatic mechanism to reference a specific set of structured content, internal (enhanced describedby model) or external (HTML4 longdesc model) to the document containing the described image.

<oedipus> 2. A way to inform users and authors that a description is present/available.

<oedipus> 3. A device independent way to access the descriptive content.

<oedipus> 4. An explicit provision that accessing descriptive content, whether internal or external to the document containing the image, does NOT take the user away from the user's position in the document containing the image where the verbose descriptor was invoked;

<oedipus> 5. A way to provide user control over exposition of the descriptor so that rendering of the image and its description is not an either/or proposition. (A visual indicator of the description should NOT be a forced visual encumbrance on sighted users by default).

<oedipus> 6. A method to reference a longer description of an image, without including the content in the main flow of a page.

Gregory: Two possibilities. Keep longdesc with some adjustment.
... Other possibility is aria describedby
... Each of these has limitations.
... Bugs have been filed on this.

<oedipus> http://www.w3.org/Bugs/Public/show_bug.cgi?id=10455

<oedipus> http://www.w3.org/Bugs/Public/show_bug.cgi?id=10017

<oedipus> jeanne, bugzilla bug http://www.w3.org/Bugs/Public/show_bug.cgi?id=10525

<oedipus> http://www.w3.org/WAI/PF/HTML/wiki/Verbose_desc_reqs

<scribe> ACTION: kford to look at longdesc issue with Gregory. Other also indicating they will look. [recorded in http://www.w3.org/2010/09/09-ua-minutes.html#action01]

<trackbot> Created ACTION-446 - Look at longdesc issue with Gregory. Other also indicating they will look. [on Kelly Ford - due 2010-09-16].

HTML A11Y bugs relating to UA from Michael Cooper - http://lists.w3.org/Archives/Public/w3c-wai-ua/2010JulSep/0061.html

<inserted> scribenick: oedipus

JA: reply not made to list yet
... sent message to list -- 5 bugs reported by HTML A11y TF as UAWG concerns
... first: 8647 - tab order for IFRAME (when IFRAME processed, tab order in doc should be outside iframe, when get into iframe follow its tab order, when move off iframe, resume tab order at next tab following last tab that put user in iframe
... might be similar to flash issues where things get inserted in DOM tree -- tab into and tab out into regular content and document flow
... compound document implications
... question: in IE and FF can use F6 to move between frames on windows platform -- if in IFRAME should be able to use F6 to jump from IFrame to main content

GL: F6 is cycle through frames -- don't know if concept of main frame / parent exists
... main viewport in which other viewports are contained

KF: what if have page with 2 frames but no iframe

KP: important to have a sure thing -- if toggling between things cannot do command to go to place x and do something

JA: can with AT

KP: cannot do it with speech with a toggle
... useful to have command to put to known place
... really useful thing to be able to directly get to a known place

KF: IE design philosophy -- F6 is a pane-cycling key -- ALT+D (address bar) is known place to return
... if use SHIFT+F6 to navigate panes in reverse

<Zakim> oedipus, you wanted to say need aria role=iframe

GJR: aria role=iframe that would make the iframe a landmark that could be easily navigated to

JA: seems like overkill -- iframe just another type of frame -- can use F6 to jump between iframe
... iframe should show up as part of frames list -- 1 less thing for author to have to do and get wrong

KF: address in HTML5 or in UAAG?

JA: this is one that UAAG needs to address unless go with role=iframe

GJR: role=iframe is a non-starter -- already have a last call draft in place
... there has been talk of ARIA 1.1 as opposed to waiting for ARIA 2.0 for which there is an extensive wish list

JA: recommend adding GL to UAAG if don't already have one

proposed resolution: address bug 8647 in UAAG 2.0

resolution: address bug 8647 in UAAG 2.0

JA: next bug title attribute on iframe
... suggest including advice on having title on IFRAME for navigation
... HTML4 had @name ATs used to identify iframe to user
... AT or other mechanism could use @name -- issue: @name is not required -- if were would be beneficial

GJR: didn't @name get deprecated in favor of @id

JA: name attribute required?

KF: this is on iframe specifically?

http://www.w3.org/WAI/PF/wiki/IFRAME

KF: @name becomes title for embedded document

JA: HomePageReader used to go to URI of IFRAME src and pull out its TITLE -- otherwise used @name value
... what to do?
... iframe should be treated as frames and use a11y stuff used for frames -- UAAG should address how to handle FRAME and IFRAME -- would be nice if @name was required -- right now @name, @id, and @title are optional

KF: for a11y want to use @title versus @name -- @name programmatic concept
... don't think is specific thing UAAG needs to tell HTML5 to change

GJR: can file a bug stating that @title should be required on IFRAME and FRAMES

JA: that would solve a whole host of issues

GJR: what i will i will send a proposed bug report to ua list and then we can discuss before i log it in the HTML5 bug tracker

<scribe> ACTION: Gregory - send proposed bug report on IFRAME and FRAME requesting that @title be required (used by AT to identify FRAMEs and IFRAMEs to user) [recorded in http://www.w3.org/2010/09/09-ua-minutes.html#action02]

<trackbot> Created ACTION-447 - - send proposed bug report on IFRAME and FRAME requesting that @title be required (used by AT to identify FRAMEs and IFRAMEs to user) [on Gregory Rosmaita - due 2010-09-16].

GL: could use a DIV for everything an IFRAME couldn't one?
... can set flow attribute to auto -- IFRAME has fixed size

JS: DIV won't let you embed external document

GJR: points out some issues with iframe resizing -- have to use sandbox attribute

<Greg> From the user perspective there is no difference...

JA: date picker -- a11y concerns expressed for date picker applies to almost everything

KF: question to ask ourselves is do we give enough guidance in SC to make controls accessible?

JA: have that in "everything should be keyboard accessible"

proposed resolution: UAAG will not take up bug 8666 -- covered by UAAG already

resolution: UAAG will not take up bug 8666 -- covered by UAAG already

JA: bug 8682 - tab and reading order for context menus
... should be defined in spec to clarify usage; editor says "don't understand is a platform specific issue"
... addition of context menu into native context menu of particular item or provide one for the user when not provided
... for UAAG, use arrow keys in windows to move through context menu after opening
... do we need an indicator for when author provided context menu different from standard context menu
... think this is a UAAG issue and not an HTML5 issue

KF: agree
... should discuss need for indicator

JA: any disagreement that is UAAG issue and not HTML issue?

resolution: bug 8682 is a UAAG issue, not an HTML5 issue; need to discuss need for indicator for when author provided context menu different from standard context menu

<AllanJ> ACTION: Jim to review notification of existence of author context menu see bug 8682 [recorded in http://www.w3.org/2010/09/09-ua-minutes.html#action03]

<trackbot> Created ACTION-448 - Review notification of existence of author context menu see bug 8682 [on Jim Allan - due 2010-09-16].

JA: bug 8743 --- auditory icons clashing with AT
... icons in HEAD of document could be visual/aural/olfactory/etc.
... are auditory icons meant to be played by UA when page loads
... HTML editor said how icons displayed and act is a UA issue
... agree with editor -- how UA should handle auditory icons should be handled by UA under user control

KF: agree

GL: 1 concern: don't disagree, but when HTML spec defines something but doesn't define how to be used, get diff implementations from different browser devs
... similar to issue with @title -- how to expose?
... personally think is bad not to provide some guidance as to how can be used
... UAAG should outline user control over auditory events

JA: sounds like another configuration thing

GJR: user should be able to decide how things are done (per instance or as universal setting)

resolution: bug 8743 (auditory icons clashing with AT) is a UA issue; UAAG should highlight user control over such events

issue: bug 8743 (auditory icons clashing with AT) is a UA issue; UAAG should highlight user control over such events

<trackbot> Created ISSUE-73 - Bug 8743 (auditory icons clashing with AT) is a UA issue; UAAG should highlight user control over such events ; please complete additional details at http://www.w3.org/WAI/UA/tracker/issues/73/edit .

JA: last bug 8751 - user should have ability to override automatic scrolling
... can use HTML5 to scroll element into view -- when detected, supposed to put element into viewport
... should be way for user to override automatic scrolling because auto-scroll may entail much more work on part of user
... editor said scrolling req is a SHOULD not a MUST -- must bring to user's attention but agnostic on how should be done
... this is a specific scripting thing being added to HTML5 affecting UA behavior
... if in HTML5, may know what is being done and how to control

<AllanJ> http://www.w3.org/TR/html5/editing.html#scrolling-elements-into-view

http://www.w3.org/Bugs/Public/show_bug.cgi?id=8751

JA: if it is in HTML5, UA should know what is going on, so UA should have ability to control whether happens automatically or only on specific request
... if just javascript, UA won't have a clue -- should ask author to assign ARIA politeness level to it so user can make decision to auto scroll or not
... HTML5 has hundreds of these javascript "standardization"
... should all of this stuff in HTML5 will inform what UAs will do with DOM manipulation or will it go into javascripted black hole

JR: dont' mention focus -- strange

GL: think that is intentional

JA: scrolls whole object into view but doesn't put programmatic foucus on object

JR: doing something that will cause user's attention to it, but haven't moved focus

GL: use case: have 2 panes -- 1 where user is working, 1 where info about background tasks; want to scroll 1 pane into view without causing focus to move so user can return to pane 1

KP: googledocs - when delete or paste, jumps to top of doc; WordPress - if delete comment from bottom, jumps to the top, so have to delete them one-by-one

JR: this could be a fix for that?

KP: if user could say "don't move my focus"

GJR: KP are you saying "don't move focus, but you can move my viewport if you bring me back to original focus?

GL: document will be taken by devs as instructing them to scroll in all cases -- title is "scroll into view" and states SHOULD scroll (although non-normative) -- key part is MUST focus but doesn't state how
... acknowledge that user settings may change behavior -- most authors will never read UAAG so would be good to have in HTML5 directly

KF: if we want something need to propose a change

JS: most effective way to get done is propose a change

GJR: 2 steps: 1 log bugs - 2. write change proposal for HTML WG

<Greg> It would be NICE but not absolutely required for the HTML document to acknowledge that user preference settings can change how this is handled. We can and should put it in UAAG but many developers won't read it.

JS: need to reopen bugs

GJR: need to take bug 8751 and put them in a change proposal on HTML WG wiki

KP: user settings can affect this

GL: would prefer them to change wording slightly so that more context available before user starts scroll
... if can find implementations should at least make recommendation; wording in a number of places could allow for that rather than seeming to prohibit it

JA: could you write something up on this GL?
... please work with GJR on this

<scribe> ACTION: Greg - work with GJR to write up wording to address bug 8751 so that highlight that it is allowed and not prohibited for HTML WG [recorded in http://www.w3.org/2010/09/09-ua-minutes.html#action04]

<trackbot> Created ACTION-449 - - work with GJR to write up wording to address bug 8751 so that highlight that it is allowed and not prohibited for HTML WG [on Greg Lowney - due 2010-09-16].

<Greg> Also, it would be nice for the HTML5 document to acknowledge that user preference settings could let the user specify the size of a buffer around the edge of the viewport such that when scrolling to found etc. content that content would not appear at the very edge of the viewport, but instead inset somewhat so the user can see more context around it without having to scroll. UAAG20 could...

<Greg> ...include a low-priority recommendation (e.g. Level AAA) that UA provide such a setting.

JR: author wants info of use to user -- want attention to be on object x -- for mouse user bringing something into middle of viewport, makes easy to mouse links -- if focus left behind, then those links are no closer for non-mouse user
... if telling people attention should be here, focus should be here too?

GJR: will author use HTML5 details element or aria-dialog?

JR: like 2 pane work use case -- if second pane isn't allowed to scroll, not getting full functionality

KP: can you fix if author broke it

JA: could be user setting -- if viewport moves, have focus move with it

GL: what authors want to do is move focus, not scroll into view
... when author explicitly does not want to move focus, would be useful to retain focus in one pane while providing viewport for second

<Greg> The HTML5 document should probably give more guidance on when authors would use scrollIntoView vs. moving the focus (which also scrolls things into view in the default configuration). Authors should almost always move the focus rather than using scollIntoView, except in very specific circumstances where they REALLY want to bring something to the user's attention WITHOUT stealing the keyboard...

<Greg> ...focus.

KP: best case - user gets info - if overriden, notify getting something different;

JR: "before you use scroll into view, make sure you do not want to move focus..."
... "scroll element into view without moving focus to element"

JA: are we covered?

GJR: are we also saying "don't move focus, but you can move my viewport if you bring me back to original focus?

JA: GL 3.10 "help user to use viewports and orient within them"
... indicate viewport position -- all in GL 3.10
... have a lot of that covered

investigate scrollIntoView crossover with ARIA politeness levels and live regions

JA: can scrollIntoView be first thing on page so user is placed in middle of document?
... example: google search puts focus in edit box
... propose Greg and Gregory get wording ready to submit to HTML5 -- find out what is response, then follow up if needed
... need to remember to bump up GL 3.10 to make sure that cover this in implementation of viewports

FIVE MINUTE WARNING

Writer's Surveys

JA: where to start?
... don't know if have enough time to get to any of them

KF: take up next week with survey 3

Browser Implementations

KF: would like WG review of all success criteria -- think we are close, but when reviewed ATAG, noticed that drafting very crisp -- where can UAAG 2.0 be tightened up especially in regards SC

JR: ATAG got major feedback -- may have to cycle back to WD -- cautionary tale for UAAG 2.0

KF: GL you have ATAG comments?

GL: yes, need to divide between high priority and regular priority
... submit as go or wait for comprehensive feedback

JR: would like to get as much feedback as soon as possible -- AUWG having F2F next week to process LC issues
... stabilization draft feedback is what we need
... people have pavlovian response to the term "Last Call"

JS: getting other WAI groups feedback before LC is key

JR: UAAG feedback to AUWG good --
... also received substantive issues from other sources, which is why may have to cycle back through another PWD and another LC

[ADJOURN]

<jeanne> gregory, can we talk about Bugzilla?

<jeanne> sure. What phone number can I call you at ?

jeanne, the bug on bugzilla is: http://www.w3.org/Bugs/Public/show_bug.cgi?id=10525

- DRAFT -

User Agent Accessibility Guidelines Working Group Teleconference

09 Sep 2010

Attendees

Contents