[aapi] Minutes: UAI TF Meeting Tue 13 January 2015

Link: http://www.w3.org/2015/01/13-aapi-minutes.html

Plain text follows:

   [1]W3C

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

                               - DRAFT -

           Protocols and Formats Working Group Teleconference
                              13 Jan 2015

   See also: [2]IRC log

      [2] http://www.w3.org/2015/01/13-aapi-irc

Attendees

   Present
          Joseph_Scheuhammer, Joanmarie_Diggs, Bryan_Garaventa,
          Rich_Schwerdtfeger

   Regrets
   Chair
          Joseph_Scheuhammer

   Scribe
          joanie, clown

Contents

     * [3]Topics
         1. [4]ACTION-842: (Cynthia) Create a diagram to support
            section 1.2 A11Y vs DOM tree.
         2. [5]ACTION-1544/1545/1546: (Cynthia/Joanie/David) How
            to expose rowgroup role.
         3. [6]ACTION-842: (Cynthia) Create a diagram to support
            section 1.2 A11Y vs DOM tree.
         4. [7]ACTION-1373/ISSUE-441: (Cynthia) Test case for
            aria-posinset with only some explicit aria-setsize.
         5. [8]ACTION-1533: (Joanie) Investigate minimal
            javascript at-spi2 test automation.
         6. [9]ACTION-1515/ISSUE-678: (Rich) Discuss role="group"
            at caucus telcon.
         7. [10]Continue with ACTIONs/ISSUEs for core-aam 1.1:
            https://www.w3.org/WAI/PF/Group/track/products/23
         8. [11]Action-1320
     * [12]Summary of Action Items
     __________________________________________________________

   <trackbot> Date: 13 January 2015

   <clown> agenda: this

   <joanie> scribenick:joanie

ACTION-842: (Cynthia) Create a diagram to support section 1.2 A11Y vs
DOM tree.

ACTION-1544/1545/1546: (Cynthia/Joanie/David) How to expose rowgroup
role.

   <clown> action-1545?

   <trackbot> action-1545 -- Joanmarie Diggs to Make sure the
   rowgroup mapping for atk/atspi is accurate -- due 2014-12-23 --
   OPEN

   <trackbot>
   [13]https://www.w3.org/WAI/PF/Group/track/actions/1545

     [13] https://www.w3.org/WAI/PF/Group/track/actions/1545

   JS: Likely, the only one here who can say anything is Joanie.

   RS: Is this where we decide if it goes to section?

   JS: Might be. This didn't used to be mapped at all.

   <clown> issue-635?

   <trackbot> issue-635 -- Determine if uaig mappings for rowgroup
   are correct -- open

   <trackbot> [14]https://www.w3.org/WAI/PF/Group/track/issues/635

     [14] https://www.w3.org/WAI/PF/Group/track/issues/635

   JD: Why would we want to map this? We don't for HTML.

   JS: I think at some point Mozilla decided to map it to group in
   IA2 at least.

   RS: I don't know why; it just happened.
   ... Actually, it *may* be that Freedom Scientific asked for
   everything in the DOM be in the accessibility tree so they
   don't have to keep going back to the DOM.
   ... They want to try to get rid of that.

   JS: The point of the accessibility tree is to not include
   things that are needed.

   <clown>
   [15]http://rawgit.com/w3c/aria/master/core-aam/core-aam.html#ro
   wGroup

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

   JD: I'd like it to be not mapped.

   JS: It's currently mapped to ROLE_PANEL.

   <clown> scribenick: clown

   JD: Panel is a generic group of stuff.
   ... You need a fairly predicatable accessibiltiy tree.

   RS: What if we take role="section", and make it non-abstract.
   ... And then add sections to the tree.

   JD: Tables do not contain sections, unless they are in a table
   cell.

   RS: There is a <rowgroup> in html.
   ... If there is a need to map that to a role.
   ... Then, map it to a seciton, a generic container that has no
   real meaning.

   JD: But a div is a block of text.
   ... I don't want this in a table heirarchy.
   ... I think the correct mapping of rowgroup is "not mapped" on
   ATK/AT-SPI.

   RS: I don't want <div>s in the hierarchy if they are
   meaningless.

   JS: A <div> is a generic container.

   RS: You should just look at such <div>s and discard them.
   ... What does AX does this?

   JS/JD: it's not mapped.

   RS: We should ask Alex why rowgroup was mapped by FF.

   <joanie>
   [16]http://rawgit.com/w3c/aria/master/html-aam/html-aam.html

     [16] http://rawgit.com/w3c/aria/master/html-aam/html-aam.html

   JD: I have updated my action to say that rowgroup is not mapped
   for ATK/AT-SPI.

   [17]http://rawgit.com/w3c/aria/master/html-aam/html-aam.html

     [17] http://rawgit.com/w3c/aria/master/html-aam/html-aam.html

   [18]http://rawgit.com/w3c/aria/master/html-aam/html-aam.html#el
   -tbody

     [18] http://rawgit.com/w3c/aria/master/html-aam/html-aam.html#el-tbody

   <joanie> scribenick: joanie

   JS: According to the above, (reads mappings for tbody)
   ... So why is rowgroup in ARIA?

   <clown> [19]http://w3c.github.io/aria/aria/aria.html#rowgroup

     [19] http://w3c.github.io/aria/aria/aria.html#rowgroup

   RS: Alex. So people could create custom grids.

   JS: I can see it for treegrids, but grids?
   ... I have my answer to ATK/AT-SPI2. Assign the action to me,
   Joanie, and I'll change the mapping
   ... I'll try to do this in the next week, so January 20th.

ACTION-842: (Cynthia) Create a diagram to support section 1.2 A11Y vs
DOM tree.

ACTION-1373/ISSUE-441: (Cynthia) Test case for aria-posinset with
only some explicit aria-setsize.

   JS: Cynthia is not here.

ACTION-1533: (Joanie) Investigate minimal javascript at-spi2 test
automation.

   JD: I've not had time to get to this yet. Sorry!
   ... I'm pushing this back to 3 March as it's a nice-to-have.

   <clown> action-1533?

   <trackbot> action-1533 -- Joanmarie Diggs to Investigate and
   possibly create a minimal/skeleton javascript at-spi2 listener
   with the aim of automating the 1.1 tests. -- due 2015-03-03 --
   OPEN

   <trackbot>
   [20]https://www.w3.org/WAI/PF/Group/track/actions/1533

     [20] https://www.w3.org/WAI/PF/Group/track/actions/1533

ACTION-1515/ISSUE-678: (Rich) Discuss role="group" at caucus telcon.

   <clown> action-1515?

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

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

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

   RS: I was on the call, I'm trying.

   JS: This was set to Dec 9th, so I thought we should re-raise
   it.

   RS: I think Cynthia said we should get back to her in a month,
   so February.

   JS: February 24th?

   RS: If we can have these things be role of section. Then HTML5
   section maps to section. Unless you put a label on it, you map
   it to role region.

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

     [22]
http://rawgit.com/w3c/aria/master/core-aam/core-aam.html#role-map-group

   RS: Div then would also map to this. It makes it very clean.
   There's some structure preserved.

   JS: We're talking about the group role.

   RS: On the Mac, AXGroup would then not be for every single div,
   etc.

   JS: So you don't need to fix the mapping for group, but
   everything else that maps to group?

   RS: On the Mac, yes.

   <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> [23]https://www.w3.org/WAI/PF/Group/track/issues/678

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

   JD: I'm not sure it will be quite that straightforward on the
   Mac, but I agree that the approach makes sense.

   RS: We need to have a chat with James about this.

   RS and JS: Maybe we can discuss this at an ARIA call with him.

Continue with ACTIONs/ISSUEs for core-aam 1.1:
[24]https://www.w3.org/WAI/PF/Group/track/products/23

     [24] https://www.w3.org/WAI/PF/Group/track/products/23

   <clown> action-1320?

   <trackbot> action-1320 -- Joseph Scheuhammer to Consider
   mapping the "offscreen" api properties in the situation of
   aria-hidden="false" on non-rendered elements. -- due 2014-12-23
   -- OPEN

   <trackbot>
   [25]https://www.w3.org/WAI/PF/Group/track/actions/1320

     [25] https://www.w3.org/WAI/PF/Group/track/actions/1320

Action-1320

   JS: This was my attempt at coming up with a compromise.
   ... Firefox didn't want to take aria-hidden:false that it
   should be exposed in the accessibility tree if it had no
   display properties.
   ... The use case I provided at the time was that it was a
   better way to accomplish the off-screen stuff.

   <clown>
   [26]https://lists.w3.org/Archives/Member/w3c-wai-pf/2014JanMar/
   0001.html

     [26]
https://lists.w3.org/Archives/Member/w3c-wai-pf/2014JanMar/0001.html

   JS: (reads text from above link)

   RS: I've seen where aria-hidden is of value when it's true.

   JS: Do you have any problem with a mismatch between the display
   properties and aria-hidden?

   RS: You do have a mismatch already.

   JS: They use an object attribute, but they don't remove it from
   the accessibility tree.

   RS: ATs ignore it when they see aria-hidden="true". I don't
   know about false.
   ... Want me to ask Freedom Scientific?

   JS: Sure

   BG: Steve and James want a literal opposite of
   aria-hidden="true".
   ... So when you have something with display:none,
   aria-hidden="false" is still exposed to ATs.
   ... JAWS is already doing this. At least for IE.

   JS: To follow-up, Firefox doesn't want to do it. So I asked
   them to treat it like off-screen positioning.

   <clown>
   [27]http://rawgit.com/w3c/aria/issue-688/aria/aria.html#aria-hi
   dden

     [27] http://rawgit.com/w3c/aria/issue-688/aria/aria.html#aria-hidden

   JD: My problem with aria-hidden="false" is that it essentially
   forces screen readers to have some sort of alternative
   presentation.
   ... Because any screen reader which is relying upon presenting
   the content to the user as the user navigates via native
   user-agent support will never encounter aria-hidden="false"
   content.
   ... In the case of off-screen content, it's just a sad hack in
   my opinion.
   ... As soon as we bless aria-hidden="false", it's like we're
   saying it's valid and appropriate to have special,
   separate-but-equal content.

   <clown> issue-688?

   <trackbot> issue-688 -- Aria-hidden=false is ambiguous
   regarding inheritance to descendant elements -- raised

   <trackbot> [28]https://www.w3.org/WAI/PF/Group/track/issues/688

     [28] https://www.w3.org/WAI/PF/Group/track/issues/688

   JD: And if that is done, screen-readers that do not provide a
   separate-but-equal rendering will seem "broken" wrt the spec.
   ... Thus you are forcing non-native interaction to be
   implemented for all screen readers wishing to be conformant.
   ... As a result, I find aria-hidden="false" objectionable. It
   is NOT the same as aria-hidden="true".
   ... For what it's worth.

   RS: Do I have an action item to content Freedom Scientific or
   not?

   JS: Ask if they want it offscreen or not in the tree.
   ... Offscreen is things like list items which are scrolled off
   screen.
   ... Another option is negative coordinates, so the sighted user
   never sees it. But the AT does.

   BG: The use case for off-screen is the skip links.
   ... In that case it needs to be in the tree.

   JS: I'll create an action some time tomorrow for you (Rich)

Summary of Action Items

   [End of minutes]

Received on Tuesday, 13 January 2015 21:16:01 UTC