Consider role tests to ensure child roles via aria-owns are not inadvertently orphaned.
In w3c/aria#2166, the ARIA working group agreed (among other things) to leave orphaned roles an an implementation detail, neither requiring nor preventing implementations from trying to recover author intent in the case of these author errors.
Some context here:
However, @rahimabdi pointed out yesterday, we're missing WPT tests for the following:
- an otherwise orphaned element (like a detached
<span role="option">) isn't actually orphaned if it's included in a listbox elsewhere via aria-owns.
- other similar aria-owns related parent/child relationships
<div role="listbox" aria-owns="id1"></div>
<div role="option" id="id1">option text</div>
In this case, we should probably ensure in some WPT tests that the browsers are exposing the listitem role and other child roles in similar contexts due to aria-owns. We have some aria-owns tests in /accname but not /wai-aria/role`
Please feel free to modify this title if you have a more clear way to state that scenario! 😅
Consider role tests to ensure child roles via aria-owns are not inadvertently orphaned.
In w3c/aria#2166, the ARIA working group agreed (among other things) to leave orphaned roles an an implementation detail, neither requiring nor preventing implementations from trying to recover author intent in the case of these author errors.
Some context here:
However, @rahimabdi pointed out yesterday, we're missing WPT tests for the following:
<span role="option">) isn't actually orphaned if it's included in alistboxelsewhere viaaria-owns.In this case, we should probably ensure in some WPT tests that the browsers are exposing the
listitemrole and other child roles in similar contexts due toaria-owns. We have somearia-ownstests in/accnamebut not /wai-aria/role`Please feel free to modify this title if you have a more clear way to state that scenario! 😅