Re: Conflicting inclusion/exclusion criteria for elements in the accessibility tree (Was: Re: [ARIA] Agenda: March 3, 2016 WAI-ARIA Working Group)

To summarize the current state of the debate, post-teleconference…

Inclusion/exclusion priority would be:

   1. If element is in the current active tab order (i.e., it is rendered
   and either it has an author-supplied positive tabindex value or it is
   normally focusable and does not have negative tabindex) ,
   OR is identified as the aria-activedescendent of another element that is
   in the current tab order, INCLUDE

   2. Else, if element has aria-hidden="true" (on self or ancestor),
   OR is hidden according to the host language semantics (this includes not
   rendered at all) and does not have aria-hidden="false" to override, EXCLUDE

   3. Else, if element has other global ARIA attributes,
   OR a computed role (by default or as set by the author) other than
   none/presentation,
   OR is referenced by an ARIA relationship of another (not hidden) element,
   OR is focusable but not in the current tab order (e.g., a
   tabindex="-1"), INCLUDE

   4. Else, if element has an author-supplied role of none/presentation,
   EXCLUDE

   5. Otherwise, INCLUDE or EXCLUDE according to host language semantics

If any of the inclusion rules (e.g., focusable element, global ARIA
attribute) force an author-supplied role of none/presentation to be
ignored, the element should default to the role specified by host language
semantics.  This is specified in the ARIA 1.1 description of presentation
role, but needs to be more clearly emphasized in Core-AAM.

We were still debating how to handle assigning events to accessible
objects, and whether the user agent should ever create additional
accessible objects based on DOM event handlers.  On the one hand, we don't
want irrelevant elements to be added to the accessibility tree just because
they can have click events or text mutations events and so on; the events
should be reported to ATs by the nearest ancestor element that does have
semantic meaning.  On the other hand, authors aren't always very good at
indicating which elements are actually relevant for interaction.

~Amelia