[aapi] Minutes: UAI TF Meeting Tue 6 January 2015

Link: https://www.w3.org/2015/01/06-aapi-minutes.html

Plain text follows:
   [1]W3C

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

                               - DRAFT -

           Protocols and Formats Working Group Teleconference
                              06 Jan 2015

   See also: [2]IRC log

      [2] http://www.w3.org/2015/01/06-aapi-irc

Attendees

   Present
          Joseph_Scheuhammer, Joanmarie_Diggs, Bryan_Garaventa,
          Rich_Schwerdtfeger, Cynthia_Shelly

   Regrets
   Chair
          Joseph_Scheuhammer

   Scribe
          Rich (and Joanie)

Contents

     * [3]Topics
         1. [4]ACTION-1542: (Joseph/Joanie) How mozilla exposes
            presentational tables and lists.
     * [5]Summary of Action Items
     __________________________________________________________

   <trackbot> Date: 06 January 2015

   <trackbot> Meeting: Protocols and Formats Working Group
   Teleconference

   <trackbot> Date: 06 January 2015

   <clown> agenda: this

   <scribe> scribenick: joanie

ACTION-1542: (Joseph/Joanie) How mozilla exposes presentational
tables and lists.

   <clown> action-1542?

   <trackbot> action-1542 -- Joanmarie Diggs to Verify the commit
   for mozilla bug 1013584 and report the atk role to joseph. --
   due 2014-12-02 -- OPEN

   <trackbot>
   [6]https://www.w3.org/WAI/PF/Group/track/actions/1542

      [6] https://www.w3.org/WAI/PF/Group/track/actions/1542

   JS: The background is: If you have a list or table with
   role="presentation", what should the required children be
   mapped to?

   RS: Isn't that mapped to text frames?

   JS: Joanie said they should be mapped as if they were divs, so
   ROLE_SECTION.

   <clown>
   [7]https://bugzilla.mozilla.org/show_bug.cgi?id=1113153#c1%29.

      [7] https://bugzilla.mozilla.org/show_bug.cgi?id=1113153#c1%29.

   JS: I tested right before the holiday.

   RS: They don't have to map to the same role, but they should
   not disappear.

   JS: List items map to ROLE_PARAGRAPH, which is understandable.
   ... But a list item with a child paragraph results in nested
   paragraphs.
   ... Grid cells map to ROLE_TEXT which Joanie says is wrong.

   <clown>
   [8]https://bugzilla.mozilla.org/show_bug.cgi?id=1113153#c1

      [8] https://bugzilla.mozilla.org/show_bug.cgi?id=1113153#c1

   JS: Joanie commented on the bug.

   RS: Do we know how role="section" is mapped in Windows?

   JS: This is about ATK.

   RS: I know, but we'd like consistency across platforms.

   <clown> [9]https://www.w3.org/WAI/PF/Group/track/actions/1542

      [9] https://www.w3.org/WAI/PF/Group/track/actions/1542

   JS: (Reads from note on action)

   (Group discusses text frames and cross platform compatibility)

   RS: I look at it as a container, so why do they implement a
   text interface in ATK, but not IA2. I don't understand.
   ... The text interface is fairly heavy-weight.
   ... What is IE going to do with this?

   <clown> [10]https://www.w3.org/WAI/PF/Group/track/actions/1542

     [10] https://www.w3.org/WAI/PF/Group/track/actions/1542

   RS: For UIA, when you run into a table which is presentational,
   what happens to the table cells?

   CS: I believe they get turned into text, but we could change
   that if there's something better.

   JS: I think that might be the next/related action.

   CS: The current behavior is that they also become
   presentational.
   ... My action is to decide if that's the right behavior.

   <clown> [11]https://www.w3.org/WAI/PF/Group/track/actions/1445

     [11] https://www.w3.org/WAI/PF/Group/track/actions/1445

   RS: As long as we don't merge text items, I think we're ok.

   JS: The above is Cynthia's action item and we decided to expose
   it using the Text pattern.

   CS: This gives you selectability, etc.

   RS: They probably have that done in Firefox as well.
   ... If you do that, you have a consistent interface at least
   here.
   ... Cynthia, what do you do with list items when the list is
   presentational?

   CS: They also become presentational, so they become spans.

   RS: In your accessibility tree, what happens to the list items?

   CS: They become text objects again.
   ... Let me double check.

   RS: I don't know how it hurts you, putting paragraphs in there.

   <clown>
   [12]https://bug1113153.bugzilla.mozilla.org/attachment.cgi?id=8
   538543

     [12] https://bug1113153.bugzilla.mozilla.org/attachment.cgi?id=8538543

   CS: I don't think it makes any difference.

   JS: The above is the test case I used. Open it up and look at
   it.
   ... (Nested paragraphs) don't make any sense to me.
   ... It's not valid in HTML. Now that doesn't stop authors from
   doing it, but technically, you cannot put a paragraph inside a
   paragraph.
   ... That might be why they created div.

   RS: If you make a list presentational, why do we do something
   like that? You could do it on an accordion, or something
   similar.

   CS: I just tested this. The list items become text.
   ... The tables in IE 11 are not honoring presentational. I'm
   not sure why. I'll file a bug for that.

   JS: So the first list item becomes text, does it contain the
   two paragraphs?

   CS: No, paragraphs are peers and also text.
   ... Yes, so it looks like they got flattened.

   RS: I think what they want to do is retain the DOM hierarchy,
   what ever it is.

   JS: In which case the first list item should become a div.

   RS: So we could make them sections (ROLE_SECTION) and they
   should be fine.

   (Group discusses section role proliferation)

   JS and RS: (Points out the need to also look at what happens on
   the Mac)

   CS: Nested divs are also flattened in IE.

   JS: (On the Mac, item 1 is static text with no children.
   Paragraph 1 is static text. Paragraph 2 is static text. They're
   siblings.

   RS: I also see a container group.

   JS: The group is a child of the webarea.
   ... There are 11 items, most of them groups.

   RS: On SVG on the Mac, AXGroup seems to be kind of like a
   section.
   ... If you have a title attribute and no role on it, it becomes
   an AXGroup.
   ... To my knowledge, UIA also has the notion of a section,
   correct?

   CS: We have group.

   RS: Group means a lot more than section.
   ... It's like a group of checkboxes. I think we need something
   separate like a section.

   CS: MSAA is all buttons and toolbars and menus. Not very much
   document-related stuff.
   ... I'm working on improving this for UIA.

   JD: IA2 has an accessible role section.

   CS: IA2 added roles on top of MSAA.

   RS: If you had your choice, what would you have? A document
   section element?

   CS: Probably.
   ... For UI kind of things we have PANE.

   RS: We have a role group that has more meaning on some
   platforms, e.g. like a group of checkboxes.

   <clown> [13]http://w3c.github.io/aria/aria/aria.html#group

     [13] http://w3c.github.io/aria/aria/aria.html#group

   RS: We could potentially have a section element in HTML5 and it
   didn't have a role, you could do it under the covers.
   ... And section defaults to role region.
   ... What does OS X map role of group to? Is it AXGroup?

   JS: Yes.

   RS: I think that's a problem.

   <clown>
   [14]http://rawgit.com/w3c/aria/master/core-aam/core-aam.html#ro
   le-map-group

     [14]
http://rawgit.com/w3c/aria/master/core-aam/core-aam.html#role-map-group

   JD: I think there's an action item for the Mac OS X group role
   issue.

   RS: (Reviews history of IA2 and ATK)
   ... If you had those as sections and paragraphs, you'd still
   maintain you're hierarchy, correct?

   JS: Yes. And I think it's what Joanie is advocating.

   RS: And I think that would work for UIA.

   CS: We're still deciding what we're going to do, but we're
   working on this.

   RS: What if we get IA2 and ATK to go to ROLE_SECTION at least
   for the list.
   ... And then work with Microsoft on this.

   <clown> [15]https://www.w3.org/2005/06/tracker/users/my

     [15] https://www.w3.org/2005/06/tracker/users/my

   <richardschwerdtfeger> RS: What we could do is have the HTML5
   <section> element map to a role of section in the accessibility
   API and when it has a label map it to role=“region”. This would
   align platform accessibility APIs with <section> as well

   <richardschwerdtfeger> RS: We would need to work with Steve
   Faulkner

   <clown> issue-678?

   <trackbot> issue-678 -- What is the ideal mapping for the aria
   group role, such that it does not semantically conflict with
   the generic group role on AAPIs, e.g., AXAPI and UIA. -- open

   <trackbot> [16]https://www.w3.org/WAI/PF/Group/track/issues/678

     [16] https://www.w3.org/WAI/PF/Group/track/issues/678

   <clown> action-1515?

   <trackbot> action-1515 -- Richard Schwerdtfeger to Discuss
   issue-678 at the mon aria teleconference. -- due 2014-12-09 --
   OPEN

   <trackbot>
   [17]https://www.w3.org/WAI/PF/Group/track/actions/1515

     [17] https://www.w3.org/WAI/PF/Group/track/actions/1515

   <richardschwerdtfeger> scribe: Rich

   <richardschwerdtfeger> RS: What we can do is have role=“group”
   map to a group role on all the paltforms and have <div>
   <section> and anything else that is inocuous but where we want
   to preserve the DOM hieararchy we map it to a role of “section”
   in the platform accessibility APIs but we need to coordinate a
   fix with IE and Safari as Safari uses axGroup for both a group
   aria role and for he inocuous hierarchy dom elements. IE does
   something similar.

   <scribe> scribenick: joanie

   RS: In the case of ATK, a group maps to a PANEL.

   JS: I'm going to move this out a month
   ... We also need to see how Mozilla reacts to the bug Joanie
   filed.

   <clown> action-1542?

   <trackbot> action-1542 -- Joanmarie Diggs to Verify the commit
   for mozilla bug 1013584 and report the atk role to joseph. --
   due 2015-02-03 -- OPEN

   <trackbot>
   [18]https://www.w3.org/WAI/PF/Group/track/actions/1542

     [18] https://www.w3.org/WAI/PF/Group/track/actions/1542

   [19]https://bugzilla.mozilla.org/show_bug.cgi?id=1113153#c1

     [19] https://bugzilla.mozilla.org/show_bug.cgi?id=1113153#c1

   JS: Action 1542 is what started this whole discussion.

Summary of Action Items

   [End of minutes]

Received on Tuesday, 6 January 2015 21:03:11 UTC