STR: 1. Start Firefox and NVDA. 2. Open this URL: data:text/html,<ul id="list"><li>item</li></ul><div contentEditable="true" role="textbox" aria-owns="list">text</div> 3. Focus the text box. 4. Press end. Expected: NVDA should say "blank". Actual: NVDA says "list, bullet". While the new aria-owns implementation is definitely a great thing and it does need to expose appropriate embedded objects via text interfaces, I don't think it should affect text interfaces for contentEditable/designMode. For contentEditable, because the user uses the caret as a primary means of navigation, it's important that the text exactly reflects what the user can reach with the caret. The real world impact here is for autocompletes, since many autocomplete lists use aria-owns so that they are "owned" by the text box.

Clarification: I'm not suggesting we change the exposure of the hierarchy, just the embedded objects that are exposed via text interfaces. So, in this case, the list should still be a "child" of the text box. However, the list should *not* be exposed as an embedded object via the text interface on the text box.

CC'ing Steve Faulkner since this might affect some of the ARIA to platform API mapping and may need an addition/change for the text interface specific stuff.

Moving to p3 because no activity for at least 1 year(s). See https://github.com/mozilla/bug-handling/blob/master/policy/triage-bugzilla.md#how-do-you-triage for more information