Re: @inert and @aria-inert for disambiguating modal states

Dominic,

The problem is that when authors create dialogs with a custom dialog
implementation they steel all the keys and the descendant DOM elements
outside the dialog box become inert (the author decides who receives events
and not the browser). In this case the browser does not know enough about
the application to indicate to an assistive technology that these elements
are inert.

We agree that the handling of @inert is best handled by the browser but in
this case it is not able to. By providing aria-inert the author is
conveying this information to ATs.

I personally would prefer to not include aria-inert as I think this is too
much work for the author but I feel I need to play devil's advocate and
defend the use case for the person who entered this request. I have yet to
see a real issue with @inert and ATs.

We both agree that aria-inert should not change browser functionality - at
all ... any more than I think they author should be fully responsible for
the focused state. These things should be managed by the user agent.

Cheers,
Rich


Rich Schwerdtfeger



From:	Dominic Mazzoni <dmazzoni@google.com>
To:	Richard Schwerdtfeger/Austin/IBM@IBMUS
Cc:	Jason Kiss <jason@accessibleculture.org>, HTML Accessibility
            Task Force <public-html-a11y@w3.org>, "W3C WAI Protocols &
            Formats" <public-pfwg@w3.org>, Alexander Surkov
            <surkov.alexander@gmail.com>, PFWG <w3c-wai-pf@w3.org>
Date:	11/05/2014 02:24 PM
Subject:	Re: @inert and @aria-inert for disambiguating modal states



On Wed, Nov 5, 2014 at 11:17 AM, Richard Schwerdtfeger <schwer@us.ibm.com>
wrote:
  Dominic, @inert is an indicator of state - like aria states and
  properties. It does not have to do the actual function. I would prefer it
  be managed by the browser but there was a request for this feature.


Yes, but it's not just a state indicator - it changes the behavior of the
browser by changing which elements can receive events.

I have no objections to inert, I think it's great and I'd like to see it
supported. My objection is only to making it part of ARIA and calling it
aria-inert - I don't think it fits because no other ARIA attribute changes
the behavior like that.



  That said, we have nothing by way of @inert in SVG.

  Personally, I am not in favor of aria supporting an inert attribute but
  what it would do is say that these parts of the document do not process
  input. To my knowledge, if someone were to create a modal dialog that
  person would need to steel the keys and mouse input so that they are not
  directed to areas outside the dialog box. In that scenario the browser
  does NOT manage the inert state and the information is not conveyed to
  assistive technologies.

  Incidentally, I would prefer that authors use an HTML5 dialog element.

  Rich


  Rich Schwerdtfeger

  Inactive hide details for Dominic Mazzoni ---10/17/2014 10:42:40 AM---I
  think that misses the broader point, which I think JameDominic Mazzoni
  ---10/17/2014 10:42:40 AM---I think that misses the broader point, which
  I think James expressed clearly: changing what's focusa

  From: Dominic Mazzoni <dmazzoni@google.com>
  To: Richard Schwerdtfeger/Austin/IBM@IBMUS
  Cc: Alexander Surkov <surkov.alexander@gmail.com>, Jason Kiss <
  jason@accessibleculture.org>, HTML Accessibility Task Force <
  public-html-a11y@w3.org>, PFWG <w3c-wai-pf@w3.org>, "W3C WAI Protocols &
  Formats" <public-pfwg@w3.org>
  Date: 10/17/2014 10:42 AM



  Subject: Re: @inert and @aria-inert for disambiguating modal states



  I think that misses the broader point, which I think James expressed
  clearly: changing what's focusable and what events fire is beyond the
  scope of ARIA.

  On Fri, Oct 17, 2014 at 6:24 AM, Richard Schwerdtfeger <schwer@us.ibm.com
  > wrote:
        ARIA does not just apply to HTML. It also applies to SVG and MANY
        elements formerly limited to HTML are now showing up in SVG. While
        I agree with you to a point Alex we need to think beyond just the
        HTML host language.

        Are you proposing that every host language duplicate all HTML
        features? ... just asking. I am copying the public pfwg list.

        Rich


        Rich Schwerdtfeger

        Inactive hide details for Alexander Surkov ---10/16/2014 08:04:51
        AM---Hi, Jason. In general ARIA shouldn't duplicate every sin
        Alexander Surkov ---10/16/2014 08:04:51 AM---Hi, Jason. In general
        ARIA shouldn't duplicate every single HTML piece otherwise we

        From: Alexander Surkov <surkov.alexander@gmail.com>
        To: Jason Kiss <jason@accessibleculture.org>
        Cc: HTML Accessibility Task Force <public-html-a11y@w3.org>, PFWG <
        w3c-wai-pf@w3.org>
        Date: 10/16/2014 08:04 AM
        Subject: Re: @inert and @aria-inert for disambiguating modal states




        Hi, Jason.

        In general ARIA shouldn't duplicate every single HTML piece
        otherwise we get into the word where HTML serves presentational
        needs while ARIA servers for semantics. In this particular case it
        means I prefer to have HTML5 @inert attribute over @aria-inert. You
        referred to [1] which claims that @inert was removed because it's
        not supposed to be used in no context of HTML5 dialog element. I
        don't have any contrary instance so that rationale looks reasonable
        with me. Getting back to ARIA, it provides role="dialog" however it
        doesn't have a way to specify the dialog modality. @aria-inert
        could be used for that but that doesn't look like a nice approach,
        it also makes ARIA markup farther from HTML. I think I would prefer
        if role="dialog" carried some extra attribute for modality stuff.

        In short I'm up to @inert attribute if it has use case. I don't
        think I see a good reason why ARIA would need @aria-inert.

        Thanks.
        Alexander.

        [1] https://html5.org/r/8536

        On Tue, Sep 30, 2014 at 9:08 PM, Jason Kiss <
        jason@accessibleculture.org> wrote:
              The PFWG is seeking feedback from the HTML Accessibility Task
              Force on
              the appropriate accessibility API mapping for an @aria-inert
              attribute
              for inert subtrees, and to reconsider the potential of the
              HTML
              "inert" attribute.

              The PFWG recently discussed a possible ARIA property,
              "aria-inert", to
              help disambiguate modal states, for example to
              programmatically
              establish that the underlying content "behind" a modal dialog
              is inert
              and cannot be interacted with, even if it is visible [1].

              The group was considering how such an attribute would map to
              accessibility APIs, and it was noted that HTML used to
              include an
              "inert" attribute, but it was removed [2]. It's noted that
              there
              remains a section on "inert subtrees" in both the HTML5 PR
              [3] and the
              HTML5.1 nightly [4].

              The general consensus in the group is that HTML's @inert was
              somewhere
              in between @aria-disabled and @aria-hidden. Mapping to
              @aria-disabled
              is not appropriate because @aria-disabled applies to the
              current
              element and focusable descendant elements only, not all
              descendant
              elements. Mapping to @aria-hidden, which, depending on the
              browser
              implementation, effectively removes the content from the
              accessibility
              tree, supports a semi-modal behavior of dialogs and menus,
              but is not
              quite correct because the underlying content "behind" the
              dialog or
              menu is actually rendered and visible, and this may have
              implications
              for some assistive technology like screen magnifiers.

              Other use cases for @inert or @aria-inert include carousels
              where you
              interact with one pane at a time but you can still see these
              other
              panes even though they're not active. Even if those inactive
              panes
              were marked up using @aria-hidden, there would still need to
              be a way
              to handle the focusability. Pop-up menus present another use
              case
              where sometimes the background might be inert, depending on
              the
              platform.

              Without @inert, and were @aria-inert to exist, it could have
              backwards-compatibility so that elements with both
              aria-hidden="true"
              and aria-inert="true" would be exposed as "inert, but not
              actually
              hidden." An author could use both to ensure the modal state
              worked in
              browsers that supported @aria-inert, and older browsers that
              only
              supported @aria-hidden.

              Members of the PFWG will chime in if I've misrepresented any
              aspect of
              their discussion on this question.

              Jason

              [1] http://www.w3.org/2014/06/23-aria-minutes.html#item01
              [2] https://html5.org/r/8536
              [3] http://www.w3.org/TR/html5/editing.html#inert-subtrees
              [4]
              http://www.w3.org/html/wg/drafts/html/master/editing.html#inert-subtrees

Received on Thursday, 6 November 2014 00:03:46 UTC