Previously, <a> elements were frequently invalidated because
`set_the_url()` was called by `reinitialize_url()`, which is a
preparation step in every HTMLHyperlinkElementUtils function. As a
result, styles were unnecessarily invalidated each time any of these
functions were invoked without changing the URL.
This change causes explicit role=none and role=presentation attribute
values to be ignored in cases where the elements for which those values
are specified are either focusable, or have global ARIA attributes —
per https://w3c.github.io/aria/#conflict_resolution_presentation_none.
This change implements the role-checking requirement from the ARIA spec
at https://w3c.github.io/aria/#document-handling_author-errors_roles
that the “form” and “region” roles are required to have accessible
names — and that if they don’t have accessible names as required, UAs
must treat them as if they’d not been specified at all.
This change causes explicitly-specified role attributes to be ignored in
the case where the specified role is “orphaned” — that is, when its
element lacks a required ancestor with an appropriate role.
Per https://w3c.github.io/aria/#document-handling_author-errors_roles,
determining whether to ignore certain specified landmark roles requires
first determining whether the element for which the role is specified
has an accessible name.
But if we then try to retrieve a role for such elements, we end up
calling right back into the accessible-name computation code — which
would cause the calls to loop infinitely.
So to avoid that — and to have handling for any other future cases the
spec may introduce of such recursive calls that will loop indefinitely —
this change introduces a parameter that callers can pass to cause
role-attribute lookup to be skipped during accessible-name computation.
This change adds a virtual to_element function to ARIAMixin, and
overrides it in DOM::Element so it can then be used back inside
ARIAMixin to get an element when needed (for example, when computing a
role requires checking the roles of ancestors of an element).
This accurately reflects the spec it's implementing. This algorithm is
used in 5 spots in the spec but the old buggy behavior was never
triggered:
* In both ::extract() and ::clone_the_contents(), invocations to this
method are guarded by a check to see if the start node is the
inclusive ancestor of the end node, or vice versa - effectively
resulting in the inequality checks to be accidentally correct.
* In ::surround_contents(), we forego the usage of this algorithm as
stated in the spec, and instead use a correct and more optimized
version that simply compares the start and end nodes.
A lot of words to say: no functional changes :^)
