W3C home > Mailing lists > Public > public-aria@w3.org > February 2016

[aapi] Minutes: UAI TF Meeting Agenda Tue Feb 16, 2016

From: Joanmarie Diggs <jdiggs@igalia.com>
Date: Tue, 16 Feb 2016 16:08:01 -0500
To: ARIA Working Group <public-aria@w3.org>
Cc: "wai-xtech@w3.org" <wai-xtech@w3.org>
Message-ID: <56C38FB1.6000701@igalia.com>
Link: https://www.w3.org/2016/02/16-aapi-minutes.html

   [1]W3C

      [1] http://www.w3.org/

   Accessible Rich Internet Applications Working Group Teleconference

16 Feb 2016

   See also: [2]IRC log

      [2] http://www.w3.org/2016/02/16-aapi-irc

Attendees

   Present
          Joanmarie_Diggs, Joseph_Scheuhammer, Bryan_Garaventa,
          Rich_Schwerdtfeger, Cynthia_Shelly

   Regrets
   Chair
          Joseph_Scheuhammer

   Scribe
          joanie

Contents

     * [3]Topics
         1. [4]ACTION-1667 (All) Progress?
            https://www.w3.org/2016/01/26-aapi-minutes.html#item01
         2. [5]ACTION-1531 (All) IA2 mapping of role="text" --
            latest from Cynthia/Alex/James Teh.
         3. [6]ISSUE-1011 (Rich) Bring Rich up to speed.
         4. [7]ACTION-1569/1585 (Joseph) Add to section on AAPI
            differences highlighting the features of UIA.
         5. [8]ACTION-1593 (Cynthia) Compare implementations of
            AAPI managed states and look for gaps.
         6. [9]ACTION-1691/ISSUE-1010 (Joseph) UIA mapping for
            aria-errormessage and role="error".
         7. [10]Blank Titles in ARIA labels and name computation.
     * [11]Summary of Action Items
     * [12]Summary of Resolutions
     __________________________________________________________

   <clown> agenda: this

   <scribe> scribe: joanie

ACTION-1667 (All) Progress?
[13]https://www.w3.org/2016/01/26-aapi-minutes.html#item01

     [13] https://www.w3.org/2016/01/26-aapi-minutes.html#item01

   <clown> action-1667?

   <trackbot> action-1667 -- Cynthia Shelly to Update the mapping
   spec for aria role region to reflect its landmark inheritance
   as defined in the aria 1.1 spec. -- due 2016-02-26 -- OPEN

   <trackbot> [14]http://www.w3.org/WAI/ARIA/track/actions/1667

     [14] http://www.w3.org/WAI/ARIA/track/actions/1667

   JS: This is not due for a while, but we need updates for role
   region for UIA.

   CS: I don't know the answer this moment, but it shouldn't take
   me long.

   <clown>
   [15]http://lists.linuxfoundation.org/pipermail/accessibility-ia
   2/2016-January/001993.html

     [15]
http://lists.linuxfoundation.org/pipermail/accessibility-ia2/2016-January/001993.html

   JS: There's also some discussion on the IA2 list (see above
   URI).
   ... Rich had said IA2 needed a new LANDMARK role.
   ... Alex said he couldn't add roles himself.
   ... Alex also asked about the ATK discussion (if any).

   RS: ATK already has that role.

   <clown> ACTION: Rich to followup with Alex about IA2 discussion
   re: adding a landmark role, similar to ATK [recorded in
   [16]http://www.w3.org/2016/02/16-aapi-minutes.html#action01]

     [16] http://www.w3.org/2016/02/16-aapi-minutes.html#action01]

   <trackbot> Created ACTION-2022 - Followup with alex about ia2
   discussion re: adding a landmark role, similar to atk [on
   Richard Schwerdtfeger - due 2016-02-23].

   RS: I'll take an action to follow up with Alex.

   <clown> action-2022

   <trackbot> action-2022 -- Richard Schwerdtfeger to Followup
   with alex about ia2 discussion re: adding a landmark role,
   similar to atk -- due 2016-02-23 -- OPEN

   <trackbot> [17]http://www.w3.org/WAI/ARIA/track/actions/2022

     [17] http://www.w3.org/WAI/ARIA/track/actions/2022

   JS: Alex sent that message on the 27th of January.

   RS: I remember seeing this. Has he done anything?

   <clown>
   [18]http://lists.linuxfoundation.org/pipermail/accessibility-ia
   2/2016-January/001993.html

     [18]
http://lists.linuxfoundation.org/pipermail/accessibility-ia2/2016-January/001993.html

   JS: What Alex did is email the IA2 list. (See above URI.)

   RS: Ok, that's on my todo list.

ACTION-1531 (All) IA2 mapping of role="text" -- latest from
Cynthia/Alex/James Teh.

   <clown> action-1531?

   <trackbot> action-1531 -- David Bolter to Bolter to investigate
   ia2 mapping of role="text" -- due 2014-11-18 -- OPEN

   <trackbot> [19]http://www.w3.org/WAI/ARIA/track/actions/1531

     [19] http://www.w3.org/WAI/ARIA/track/actions/1531

   JS: This started out as a David Bolter action.
   ... We started out in IA2 with text frames, then moved to
   static.
   ... But then Cynthia brought up ways to do it in UIA.

   <clown>
   [20]https://lists.w3.org/Archives/Public/public-aria/2016Feb/00
   32.html

     [20] https://lists.w3.org/Archives/Public/public-aria/2016Feb/0032.html

   JS: One email is at above URI.

   <clown> <img alt="foo" role="tesxt">

   JS: It's going to be a text pattern.

   CS: Right, but not a separate text pattern.

   JS: In UIA, there's not going to be any accessible object for
   the image element typed above.

   CS: That's also what happens with span elements.
   ... That's the best way I can think of.
   ... Unless there was a property like a tabindex, we wouldn't
   create a separate accessible.

   RS: So what we want is for the text role to be read inline.

   CS: The text pattern does have characters and character
   navigation.
   ... Narrator supports that.
   ... There are also breaks and the ability to get range markers.
   ... So basically you're turning the img into a span
   (functionally).

   RS: What about caret tracking?

   CS: I'm not sure. This is the only thing I can think of to do.
   I have some doubts about this feature.

   RS: What if we have Larry on the call?

   CS: I'll talk to him first.

   RS: I don't know that the Mac has caret tracking.

   JS: I tested that.
   ... If you're going character by character, it treats the img
   text "foo" as if it were a character.

   RS: That seems like a reasonable thing.

   CS: And also is aligned with how we do Chinese and Japanese
   characters.

   JS: In the accessibility inspector, there are three accessible
   objects: the text before the image, the text of the image, and
   the text after the image.

   RS: So if you were reading text in the middle of a paragraph
   and ran into an image, you wouldn't be saying image, right?

   JS: VoiceOver says image when the role is not present.

   RS: I guess what I'm saying is you don't want to pause in the
   middle; you just want to read the whole thing without pausing.

   (Group ponders what the user experience would be like if there
   are periods in the alt text.)

   JS: What about word-by-word examination of alt text?

   CS: Narrator doesn't support that. I don't know about other
   screen readers.
   ... I have a way to map this, and the user experience seems
   bad.
   ... I'm very nervous about the ramifications of this feature,
   and do not think it is worth the risk.

   JS: I also have misgivings.

   RS: We could consider limiting it to just images.

   CS: That would make it better....

   <clown> <span role="text" aria-label="3 of 5
   stars">★★★☆☆︎</span></p>

   JS: One of the examples in the spec involves 5 star glyphs.

   <clown> <span role="text" aria-label="3 of 5
   stars">★★★☆☆︎</span></p>

   CS: And these glyphs exist in the wild.

   JS: And unicode has them.
   ... And in Firefox you can arrow over each character
   separately.
   ... In Safari, the stars have a single accessible of
   StaticTextRole with a value of "3 of 5 stars".

   <richardschwerdtfeger> <span role="text" aria-label="3 of 5
   stars"><span role=“img”>★★★☆☆︎</span></span></p>

   RS: You could also do the above.

   CS: We create currently a group with "3 of 5 stars" as it's
   name and no children.

   RS: What you could do is force it to be an image....

   CS: If I were authoring this, I'd use role="image" on this.

   JS: That's what Stefan said as well.

   RS: I say we go back to Steve and say you cannot put
   role="text" on something unless it's an image.

   CS: I think we should say it's only for images and then ask for
   a WCAG technique.

   RS: I agree with that.

   CS: I'd do a span with role="image"

   JS: Are we changing the ARIA spec?

   <clown> [21]http://w3c.github.io/aria/aria/aria.html#text

     [21] http://w3c.github.io/aria/aria/aria.html#text

   RS: We could say "in HTML..." in the spec. Because SVG's needs
   are different, and the spec applies to both.
   ... We could say that you can't use this on the text and tspan
   elements.

   <clown> <div role="text">

   <clown> <p>I</p>

   <clown> <p>like</p>

   <clown> <p>turtles</p>

   <clown> </div>

   JD: I think we could do a change in the spec. If nothing else,
   we need to get rid of the paragraph example.

   BG: The one use case for the above is that some authoring
   causes ATs to treat a single button as if it's broken up.

   RS: What if you put an anchor in the middle of that?

   JS: This is approaching the next agenda item.

   <clown> "Use caution when using the text role on structural
   elements. In particular, avoid using the text role on elements
   with interactive descendants"

   JS: (Reads from the spec, text quoted above.)

   CS: Authors are going to mess this up.

   RS: Can we give you, Joanie, and action item to limit this to
   graphics and go back to Steve?

   JS: I am.

   CS: I am.

   RS: I am.

   JD: I am.

   RS: Give me an action to go back to Steve.

   <clown> ISSUE-1011

   <trackbot> ISSUE-1011 -- For role="text", there needs to be
   requirements on user agents for overrides, such as controls,
   events, and other global aria properties, like there is for
   role="presentation". -- open

   <trackbot> [22]http://www.w3.org/WAI/ARIA/track/issues/1011

     [22] http://www.w3.org/WAI/ARIA/track/issues/1011

ISSUE-1011 (Rich) Bring Rich up to speed.

   JS: A couple of meetings ago we concluded there was not enough
   guidance in the spec about the dangers and need to
   limit/override it.

   <clown> ACTION: Joanie to write a proposal about how to modify
   the definition of role text to limit its use. [recorded in
   [23]http://www.w3.org/2016/02/16-aapi-minutes.html#action02]

     [23] http://www.w3.org/2016/02/16-aapi-minutes.html#action02]

   <trackbot> Created ACTION-2023 - Write a proposal about how to
   modify the definition of role text to limit its use. [on
   Joanmarie Diggs - due 2016-02-23].

   <clown> ACTION: Rich to discuss with Steve F about limitations
   wrt to role text [recorded in
   [24]http://www.w3.org/2016/02/16-aapi-minutes.html#action03]

     [24] http://www.w3.org/2016/02/16-aapi-minutes.html#action03]

   <trackbot> Created ACTION-2024 - Discuss with steve f about
   limitations wrt to role text [on Richard Schwerdtfeger - due
   2016-02-23].

   <clown> action-2023

   <trackbot> action-2023 -- Joanmarie Diggs to Write a proposal
   about how to modify the definition of role text to limit its
   use. -- due 2016-02-23 -- OPEN

   <trackbot> [25]http://www.w3.org/WAI/ARIA/track/actions/2023

     [25] http://www.w3.org/WAI/ARIA/track/actions/2023

   JD: I'm happy to write a draft, but as you may recall, when
   this new role was first raised, I brought up my concerns and
   got the impression from Steve and James that I was "wrong."
   ... So I can write text, but I anticipate getting some pushback
   from Steve and/or James.

   RS: I'll talk to Steve first.

   JS: Also, James Teh get into this and says he wants
   IAccessibleText implemented for it, even though it has gotchas.

   RS: Why would he want that if we limit it to images?

   Group: True.

   JS: In the meantime, the mapping is to ROLE_STATICTEXT.

ACTION-1569/1585 (Joseph) Add to section on AAPI differences
highlighting the features of UIA.

   <clown> action-1569

   <trackbot> action-1569 -- Jason Kiss to Create a section that
   describes AAPI differences -- due 2015-03-03 -- OPEN

   <trackbot> [26]http://www.w3.org/WAI/ARIA/track/actions/1569

     [26] http://www.w3.org/WAI/ARIA/track/actions/1569

   <clown> action-1585

   <trackbot> action-1585 -- Joseph Scheuhammer to Add to the
   section on differences among aapis highlighting the features of
   uia. -- due 2016-01-15 -- OPEN

   <trackbot> [27]http://www.w3.org/WAI/ARIA/track/actions/1585

     [27] http://www.w3.org/WAI/ARIA/track/actions/1585

   JS: I was hoping to have this (my actions to add Jason's and
   Cynthia's text about AAPI differences) done.
   ... But I was spending time on combo boxes, so I need to put
   this action off for another week.

ACTION-1593 (Cynthia) Compare implementations of AAPI managed states
and look for gaps.

   <clown> action-1593

   <trackbot> action-1593 -- Cynthia Shelly to Compare
   implementations of AAPI managed states and look for gaps. --
   due 2016-02-08 -- OPEN

   <trackbot> [28]http://www.w3.org/WAI/ARIA/track/actions/1593

     [28] http://www.w3.org/WAI/ARIA/track/actions/1593

   JS: Cynthia, this action is yours.

   CS: I have not done this yet.

   JS: New due date?

   CS: Make it next week.

ACTION-1691/ISSUE-1010 (Joseph) UIA mapping for aria-errormessage and
role="error".

   <clown> action-1691

   <trackbot> action-1691 -- Joseph Scheuhammer to Propose UIA
   mapping for aria-errormessage. -- due 2015-12-31 -- CLOSED

   <trackbot> [29]http://www.w3.org/WAI/ARIA/track/actions/1691

     [29] http://www.w3.org/WAI/ARIA/track/actions/1691

   <clown> issue-1010

   <trackbot> issue-1010 -- Add an error role -- open

   <trackbot> [30]http://www.w3.org/WAI/ARIA/track/issues/1010

     [30] http://www.w3.org/WAI/ARIA/track/issues/1010

   JS: I've completed this. I got the information from Cynthia and
   made the changes.

   <clown>
   [31]http://rawgit.com/w3c/aria/master/core-aam/core-aam.html#ar
   iaErrorMessage

     [31]
http://rawgit.com/w3c/aria/master/core-aam/core-aam.html#ariaErrorMessage

   JS: It's at the above URI for UIA.
   ... (Reads mapping)
   ... And now that I'm looking at the table, there are a lot of
   "TBD"s.

   CS: Looks fine to me.

   JS: Objections from anybody?

   (None)

   JS: This action can be closed then.
   ... I already closed it.
   ... I'm going to close the issue to add an error role. (Issue
   1010)
   ... That makes me worry about the other ones.

   <clown> action-1692

   <trackbot> action-1692 -- Joanmarie Diggs to Propose ATK/AT-SPI
   mapping for aria-errormessage -- due 2015-08-04 -- OPEN

   <trackbot> [32]http://www.w3.org/WAI/ARIA/track/actions/1692

     [32] http://www.w3.org/WAI/ARIA/track/actions/1692

   JD: I know what I want to do for ATK, but we're expected to try
   to keep IA2 and ATK in alignment where possible.
   ... And I've not had time to talk to Alex and James Teh about
   this.

   action-1373

   <trackbot> action-1373 -- Cynthia Shelly to Create a test case
   for when aria-posinset and aria-setsize are provided explicitly
   on some but not all elements. -- due 2016-02-08 -- OPEN

   <trackbot> [33]http://www.w3.org/WAI/ARIA/track/actions/1373

     [33] http://www.w3.org/WAI/ARIA/track/actions/1373

   CS: One of these days.... :)
   ... Give me two weeks. I really want to do it.

   <clown> action-1541?

   <trackbot> action-1541 -- Joseph Scheuhammer to Investigate and
   propose mappings for aria-modal property for axapi. -- due
   2016-02-08 -- OPEN

   <trackbot> [34]http://www.w3.org/WAI/ARIA/track/actions/1541

     [34] http://www.w3.org/WAI/ARIA/track/actions/1541

   JS: This is the mapping for AX API. I've been following the
   issue in WebKit's bugilla.
   ... And they have now fixed the issue. So I need to look at the
   mapping and update the Core AAM.

Blank Titles in ARIA labels and name computation.

   RS: We don't have any special cases for "".

   JS: Example?

   RS: In SVG, if you have a circle and it doesn't have any
   attributes, you don't map it.
   ... But if you have circle with aria-label="", it winds up in
   the accessibility tree.
   ... The SVG group has been discussing this.
   ... I would think we just follow the spec and expose it, not
   taking the empty string into account.

   CS: My one concern is that a lot of authoring tools put in
   empty strings for accessibility attributes.

   RS: But that should be checked by a conformance checker.

   CS: The other thing we could do is say empty strings don't
   count.

   JS: We could do that.
   ... I think we had a test for this in 1.0.

   CS: But this is slightly different.

   JS: I see Alex saying that we're not going to take the time to
   check the string.

   CS: It would be nice if empty strings didn't wind up being
   exposed.

   JS: So what do you want Rich?

   RS: If the author does an empty string, I think the associated
   (shape) element should be in the tree.

   Group: And the author is not especially bright.

   <clown> +1

   RESOLUTION: User Agents will not correct alternative text
   solutions for elements when the text is blank or empty. It is
   an author error.

   <clown>
   [35]https://dvcs.w3.org/hg/pfwg/raw-file/default/ARIA/1.0/tests
   /test-files/name-computation-input/536.html

     [35]
https://dvcs.w3.org/hg/pfwg/raw-file/default/ARIA/1.0/tests/test-files/name-computation-input/536.html

Summary of Action Items

   [NEW] ACTION: Joanie to write a proposal about how to modify
   the definition of role text to limit its use. [recorded in
   [36]http://www.w3.org/2016/02/16-aapi-minutes.html#action02]
   [NEW] ACTION: Rich to discuss with Steve F about limitations
   wrt to role text [recorded in
   [37]http://www.w3.org/2016/02/16-aapi-minutes.html#action03]
   [NEW] ACTION: Rich to followup with Alex about IA2 discussion
   re: adding a landmark role, similar to ATK [recorded in
   [38]http://www.w3.org/2016/02/16-aapi-minutes.html#action01]

     [36] http://www.w3.org/2016/02/16-aapi-minutes.html#action02
     [37] http://www.w3.org/2016/02/16-aapi-minutes.html#action03
     [38] http://www.w3.org/2016/02/16-aapi-minutes.html#action01

Summary of Resolutions

    1. [39]User Agents will not correct alternative text solutions
       for elements when the text is blank or empty. It is an
       author error.

   [End of minutes]
Received on Tuesday, 16 February 2016 21:08:50 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 18:58:20 UTC