[AAPI] Minutes UAI TF Meeting, Tuesday 05 July 2016

URL: https://www.w3.org/2016/07/05-aapi-minutes.html

Plain text follows:
   [1]W3C

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

   Accessible Rich Internet Applications Working Group Teleconference

05 Jul 2016

   See also: [2]IRC log

      [2] http://www.w3.org/2016/07/05-aapi-irc

Attendees

   Present
          Joanmarie_Diggs, Joseph_Scheuhammer, Bryan_Garaventa,
          Rich_Schwerdtfeger

   Regrets
          Cynthia_Shelly

   Chair
          Joseph_Scheuhammer

   Scribe
          joanie

Contents

     * [3]Topics
         1. [4]ISSUE-1029 (All) Mappings for separator when it is
            a window-splitter.
         2. [5]ISSUE-1016 (All) Mappings for aria-keyshortcuts.
         3. [6]ISSUE-682/ACTION-1745 (Joseph/All) Overriding role
            of elements with required owned children
         4. [7]ISSUE-661 (All) How to map children of table/lists
            with role presentation.
         5. [8]ISSUE-677 (All) ATK/AT-SPI mapping of
            role="presentation" to ROLE_PANEL.
         6. [9]ISSUE-487 (All) List and define mappings for
            managed states.
         7. [10]Triage issues/actions:
            https://www.w3.org/WAI/ARIA/track/products/23
     * [11]Summary of Action Items
     * [12]Summary of Resolutions
     __________________________________________________________

   <clown> agenda: this

   <scribe> scribe: joanie

ISSUE-1029 (All) Mappings for separator when it is a window-splitter.

   JS: You wanted to go through issues this week.
   ... So I pre-loaded some that seem relevant.

   <clown> issue-1029?

   <trackbot> issue-1029 -- Add mappings for interactive separator
   (window splitter). -- open

   <trackbot> [13]http://www.w3.org/WAI/ARIA/track/issues/1029

     [13] http://www.w3.org/WAI/ARIA/track/issues/1029

   JS: This is an issue to have specific mappings for interactive
   separator.
   ... We have a number of actions for the platforms.
   ... All those mappings are pretty much done.

   <clown>
   [14]https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#r
   ole-map-separator

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

   JS: Here (above) are the mappings.
   ... We spoke about this last week.
   ... (Describes what's in the mappings)
   ... I have not changed the AXAPI mappings at all.

   RS: Is there a splitter role in IA2 or ATK/AT-SPI2.

   JS: No.

   RS: And you removed expanded/collapsed?

   JS: Yes.

   RS: Then I think you're done.

   <clown> action-2089?

   <trackbot> action-2089 -- Cynthia Shelly to Double check
   potential value events on windows when window-splitter changes
   position. -- due 2016-07-05 -- OPEN

   <trackbot> [15]http://www.w3.org/WAI/ARIA/track/actions/2089

     [15] http://www.w3.org/WAI/ARIA/track/actions/2089

   JS: Action 2089 remains.
   ... Cynthia is not here.
   ... And she did not make a note.
   ... Joanie pointed out that maybe we don't want to say "window"
   with splitter.

   [16]http://methvin.com/splitter/

     [16] http://methvin.com/splitter/

   JD: The above has demos and they don't look like windows to me.

   RS: It's really a panel.
   ... In the really old days, you were splitting actual windows.

   JS: How should it be reworded?

   RS: Panels, panes, ...?

   JS: So "focusable pane splitter"?

   RS: What's your definition of "window"?

   JD: I would just cut out "window"

   JS: So "focusable splitter"?

   JD: Yes.

   JS: I will do that.
   ... But we cannot close this until we hear back from Cynthia.

ISSUE-1016 (All) Mappings for aria-keyshortcuts.

   <clown> issue-1026

   <trackbot> issue-1026 -- AAPI mappings for aria-errormessage --
   open

   <trackbot> [17]http://www.w3.org/WAI/ARIA/track/issues/1026

     [17] http://www.w3.org/WAI/ARIA/track/issues/1026

   <clown> issue-1016

   <trackbot> issue-1016 -- Means to get all the aria-kbdshortcuts
   elements -- open

   <trackbot> [18]http://www.w3.org/WAI/ARIA/track/issues/1016

     [18] http://www.w3.org/WAI/ARIA/track/issues/1016

   <clown>
   [19]https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#a
   riaKeyshortcuts

     [19]
https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#ariaKeyshortcuts

   JS: The mapping is at the above URL.
   ... I have everything as people have asked me to.
   ... (Reads mappings)

   RS: Accelerators and mnemonics are different.

   <clown> action-1695

   <trackbot> action-1695 -- Joseph Scheuhammer to Propose UIA
   mapping for aria-kbdshortcuts -- due 2016-03-08 -- CLOSED

   <trackbot> [20]http://www.w3.org/WAI/ARIA/track/actions/1695

     [20] http://www.w3.org/WAI/ARIA/track/actions/1695

   JS: This is what Cynthia told me to do.
   ... (Reads from action)

   RS: Then that's fine.
   ... I thought IA2 had a property.

   <clown> action-1697

   <trackbot> action-1697 -- Joseph Scheuhammer to Propose
   MSAA+IA2 mapping of aria-kbdshortcuts -- due 2016-06-07 --
   CLOSED

   <trackbot> [21]http://www.w3.org/WAI/ARIA/track/actions/1697

     [21] http://www.w3.org/WAI/ARIA/track/actions/1697

   JS + JD: We got this from James Teh.

   <clown>
   [22]https://lists.linuxfoundation.org/pipermail/accessibility-i
   a2/2016-February/002008.html

     [22]
https://lists.linuxfoundation.org/pipermail/accessibility-ia2/2016-February/002008.html

   JS: The above is the email from Jamie.

   <clown>
   [23]https://msdn.microsoft.com/en-us/library/windows/desktop/dd
   318482%28v=vs.85%29.aspx?f=255&MSPPError=-2147217396

     [23]
https://msdn.microsoft.com/en-us/library/windows/desktop/dd318482(v=vs.85).aspx?f=255&MSPPError=-2147217396

   JS: James Teh also cites the above documentation.
   ... But if you want to reopen this

   RS: No. I just want to check. I thought there was something.
   ... I'm looking.
   ... I'm not seeing it.
   ... So it's fine.

   JS: The one that's standing out is AXAPI, which doesn't map it.
   ... So it's not interoperable any more.

   <clown> action-1698

   <trackbot> action-1698 -- Joseph Scheuhammer to Propose AXAPI
   mapping of aria-kbdshortcuts -- due 2016-07-05 -- OPEN

   <trackbot> [24]http://www.w3.org/WAI/ARIA/track/actions/1698

     [24] http://www.w3.org/WAI/ARIA/track/actions/1698

   JS: I actually had the mapping from Chris.
   ... But shortly after he said that, James Craig said "please
   don't map this."
   ... In part, they want to get rid of the "ARIA".

   RS: It's their platform.
   ... It's my impression that Apple wants to move more to device
   independence.

   JS: Can we close the AXAPI action?

   RS: I think so.

   JS: I'm surprised they didn't say to put it in AXHelp.
   ... Anyway, closing this action.

ISSUE-682/ACTION-1745 (Joseph/All) Overriding role of elements with
required owned children

   <clown> issue-682?

   <trackbot> issue-682 -- Clarify what happens to required owned
   childrens' roles when native host language parent is overridden
   with a WAI-ARIA role. -- open

   <trackbot> [25]http://www.w3.org/WAI/ARIA/track/issues/682

     [25] http://www.w3.org/WAI/ARIA/track/issues/682

   JS: We discussed this briefly last week.
   ... This is the case where someone does the following:

   <clown> <table role="dialog"> … </table>

   JS: They use a table to make a dialog.
   ... The question is what happens to the required own elements?

   JS + RS: They become presentational.

   JS: I was asked to add some text regarding this.

   <clown>
   [26]https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#m
   apping_conflicts

     [26]
https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#mapping_conflicts

   JS: The end of the second paragraph of the text at the above
   URL

   <clown> "If the host language element has required owned
   elements, and its native role is overridden by a WAI-ARIA role,
   then treat the required owned elements as having role
   presentation or none. "

   JS: I added the above sentence.
   ... And I have links to the mappings for presentation and none.
   ... I open the floor for comments.

   RS: Well, the problem is... What if you override a table with a
   grid?
   ... So no, that won't work.

   JS: Grid inherits from table, right?

   RS: I don't think you can do a blanket statement on this one.
   ... Is this for particular roles, or not?

   JS: The only ones I can think of at the moment are table and
   list.
   ... Others?

   RS: What if we say something like the structure matches the
   overriding role?
   ... A dialog has no structure.
   ... If you put a grid on a list, that makes it even more
   challenging.
   ... I think I would just special-case a couple of these.

   JS: Which?

   RS: I think what I would do is say that when you're overriding
   a table, the structure would have to align with the overriding
   role.
   ... If there is no structure, they all become presentational.

   JS: I can see someone using a list as well, just to get the
   layout.

   RS: Is it just for dialog?

   JS: No, that's just something I came up with.
   ... Actually, it came up in a conversation between Alex and
   James N.
   ... Alex had said the table remains a table.

   RS: You cannot do that.

   JS: So there was a compromise, to put a div around the whole
   table.

   RS: Fake nodes are a mess though.
   ... I think that's Joanie was talking about last week.
   ... Is it just for dialogs, or other things?

   JS: Dialog is the one mentioned.

   RS: You can make it an authoring error.

   JS: Table is used for layout.
   ... I think Dojo might have even used a table element with role
   button.

   RS: Why don't we just way, with the exception of grid or table,
   if you put a role on top of table, it becomes presentational?

   JS: Or their subclasses? Because there's treegrid.

   RS: Does grid subclass table?

   JS: I think so.

   RS: So if the semantics are the same, of you have a subclass,
   then the descendants become presentational.

   JD: I think what Rich is describing makes sense.
   ... Though due to scribing I'm only *mostly* listening and
   thinking.

   <clown> If the host language element has required owned
   elements, and its native role is overridden by a WAI-ARIA role,
   then treat the required owned elements as having role
   presentation or none.

   JD: That said, if we can come up with text that cleanly/clearly
   expresses what Rich is saying, then it probably makes sense to
   do so.

   <clown> "If the host language element has required owned
   elements, and its native semantics is overridden by a WAI-ARIA
   role, then treat the required owned elements as having role
   presentation or none. "

   <clown> If the host language element has required owned
   elements, and its native semantics or structure is overridden
   by a WAI-ARIA role, then treat the required owned elements as
   having role presentation or none.

   JD: It's really not so much subclass; it's more like when the
   role's required owned elements are in conflict with the native
   required owned elements.

   JS: Above I put a new version of the text in which just changed
   one word.

   RS: But what about gridcell versus cell?

   JS: The difference there is interactive.
   ... In my replacement, I changed "native role" to "native
   semantics" and then to "native semantics or structure"

   JD: I generally agree with what Joseph is saying.
   ... The owned children are what are being trumped.

   <clown> If the host language element has required owned
   elements, and it is overridden by a WAI-ARIA role that has a
   different structure, then treat the required owned elements as
   having role presentation or none.

   JD: And we don't want to have to have a paragraph for each
   scenario.

   JS: What about the above text?

   RS: I think you need to go back to native host language
   semantics.

   <Rich> Proposal: If the host language element is ovveridden by
   an ARIA role whose semantics are not equiavalent to the native
   host language semantics or a subclass of those semantics then
   the required owned elements become presentation or none.

   JS: I want to tweak that slightly.

   <clown> Proposal: If the host language element is ovveridden by
   an ARIA role whose semantics or structure is not equiavalent to
   the native host language semantics or a subclass of those
   semantics, then the required owned elements become presentation
   or none.

   JS: So in the above I just added "or structure"

   RS: I'm okay with that.
   ... You may need to check with Alex.

   JS: I think they're already doing this.
   ... I think eventually he was convinced that the semantics
   really do change.
   ... There was even a weird case where Oracle was using the CSS
   table markup.
   ... So it looks like a table on screen.
   ... And Gecko was not exposing it as a table; but as divs and
   spans.
   ... Anyway, if I make the changes discussed here, can I close
   this?

   RS: Yes.

ISSUE-661 (All) How to map children of table/lists with role
presentation.

   <clown> issue-661

   <trackbot> issue-661 -- UAIG1.1: Specify how to map children of
   table/lists with role presentation. -- open

   <trackbot> [27]http://www.w3.org/WAI/ARIA/track/issues/661

     [27] http://www.w3.org/WAI/ARIA/track/issues/661

   JS: Isn't that what we just did?
   ... Actually, this is different.
   ... I think this is finished.
   ... I have a bunch of actions on this.
   ... All the actions are closed.

   <clown>
   [28]https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#r
   ole-map-presentation

     [28]
https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#role-map-presentation

   JS: The text for role="presentation" already has information in
   it (above URL)
   ... So I think this action is done.
   ... (Reads mapping)
   ... Can I close this issue?

   RS: Yes.

ISSUE-677 (All) ATK/AT-SPI mapping of role="presentation" to
ROLE_PANEL.

   JD: I saw this and closed it.

   <clown> issue-677

   <trackbot> issue-677 -- ATK/AT-SPI mapping of
   role="presentation" to ROLE_PANEL is incorrect. -- closed

   <trackbot> [29]http://www.w3.org/WAI/ARIA/track/issues/677

     [29] http://www.w3.org/WAI/ARIA/track/issues/677

   JS: When I looked at this the other day and thought it was
   closed.
   ... Joanie had raised it a while back.
   ... The mapping is now ROLE_SECTION.
   ... Joanie commented on when it was fixed and closed the
   action.
   ... (Reads Joanie's comment on the action)

ISSUE-487 (All) List and define mappings for managed states.

   <clown> issue-487

   <trackbot> issue-487 -- List managed states and define mappings
   for managed states -- open

   <trackbot> [30]http://www.w3.org/WAI/ARIA/track/issues/487

     [30] http://www.w3.org/WAI/ARIA/track/issues/487

   JS: I think this is defunct.
   ... Actually, no. I think it's a duplicate.

   <clown> issue-706?

   <trackbot> issue-706 -- Describe/define AAPI managed states --
   open

   <trackbot> [31]http://www.w3.org/WAI/ARIA/track/issues/706

     [31] http://www.w3.org/WAI/ARIA/track/issues/706

   JS: And issue-706 has Cynthia's action.

   RS: So she still needs to do it?

   JS: Yes.

   <clown> issue-487

   <trackbot> issue-487 -- List managed states and define mappings
   for managed states -- open

   <trackbot> [32]http://www.w3.org/WAI/ARIA/track/issues/487

     [32] http://www.w3.org/WAI/ARIA/track/issues/487

   JS: My question is: Is issue-487 a duplicate of issue-706?

   RS: Yes.

   JS: Ok, I'll close the former as a duplicate of the latter.

Triage issues/actions:
[33]https://www.w3.org/WAI/ARIA/track/products/23

     [33] https://www.w3.org/WAI/ARIA/track/products/23

   issue-441

   <trackbot> issue-441 -- Normative UAIG requirements for what
   UAs do when aria-posinset or aria-setsize is provided
   explicitly on some, but not all, elements within a set. -- open

   <trackbot> [34]http://www.w3.org/WAI/ARIA/track/issues/441

     [34] http://www.w3.org/WAI/ARIA/track/issues/441

   <clown> issue-441

   <trackbot> issue-441 -- Normative UAIG requirements for what
   UAs do when aria-posinset or aria-setsize is provided
   explicitly on some, but not all, elements within a set. -- open

   <trackbot> [35]http://www.w3.org/WAI/ARIA/track/issues/441

     [35] http://www.w3.org/WAI/ARIA/track/issues/441

   JS: I think Cynthia has an action for this.
   ... What user agents are supposed to do when posinset and
   setsize are only provided on some items in the set.

   RS: Didn't we deal with this in ARIA 1.0?

   JS: No.
   ... We might have opened it in ARIA 1.0.

   <clown> 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 2016-08-02 -- OPEN

   <trackbot> [36]http://www.w3.org/WAI/ARIA/track/actions/1373

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

   RS: Do we have an action?

   JS: Yes, action-1373. Due August 2nd.

   RS: Should we reassign this?

   JS: You want it?

   RS: Sure.

   JS: I would rather ask her first.

   RS: Why can't we leave it as undefined?

   JS: If you don't put it on anything, Firefox figures it out.
   ... But it might not be correct.

   RS: My personal belief is that authors should set them.
   ... I don't know if it's something we should try to fix.

   JS: It was created back in 2011.
   ... But it doesn't say who created it.

   RS: This is before ARIA 1.0 shipped.
   ... I think this is not important.
   ... And it's unreliable.
   ... Because no one is going to do it the same way.

   JS: Her action is to create a test case.

   RS: I don't think this is worth doing.
   ... I recommend we close it.

   JD: Not saying anything at all seems bad though.

   <clown>
   [37]https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#m
   apping_additional_position

     [37]
https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#mapping_additional_position

   RS: It's an error condition.

   <clown>
   [38]https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#m
   apping_additional_position

     [38]
https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#mapping_additional_position

   JS: I just found text in the mapping spec which tells you how
   to fix it (above URL).

   RS: But that's group position.

   JS: Keep reading. Third paragraph.

   RS: So there in or they're out.
   ... But what happens when posinset is there and setsize is not.
   ... If they're not provided, user agents must compute them as
   follows.
   ... But if one of them is provided, then it's up to the author
   to do the setting.
   ... And you might have inconsistent results.
   ... But we shouldn't try to repair them.

   JS: So what the user agent does is undefined?

   RS: Yes.
   ... It's up to the author to maintain these values.
   ... Not the user agent.

   JS: Do you want to add a sentence to that effect, or just leave
   it alone?

   <Rich> Proposal: If the author provides one or more of
   aria-setsize and aria-posinset it up to the author to maintain
   the values and user agent correction of these values is not
   defined.

   JD: Looking at the ARIA spec,
   [39]https://rawgit.com/w3c/aria/master/aria/aria.html#aria-posi
   nset, it says authors SHOULD provide aria-setsize.

     [39] https://rawgit.com/w3c/aria/master/aria/aria.html#aria-posinset,

   RS: Above is my proposal.

   JS: I can wordsmith that.
   ... I'm inclined to put it in the author-error section.

   <Rich> Proposal: If the author provides one or more of
   aria-setsize and aria-posinset it is up to the author to
   maintain the values and user agent correction of these values
   is not defined.

   JS: And then close the issue and action.
   ... And Cynthia's test case action evaporates?

   RS: Yes, because it's undefined.

   JS: I'll triple-check with her though.

Summary of Action Items

Summary of Resolutions

   [End of minutes]

Received on Tuesday, 5 July 2016 20:15:39 UTC