Avoiding ambiguity in specification language beyond the web platform #86
Description
While designing "user intents" for this protocol, we've recognized that describing the state of an AT necessarily involves referencing application user interfaces. The W3C and WHATWG offer a robust language for describing web applications, but it isn't clear if such language is appropriate for use in AT Driver. Since ATs are not web browsers, their understanding of the user interface does not necessarily align with the web platform.
For example, the proposed "activate element" user intent needs to modify the focused area of the document. It's not clear whether ATs' concept of the focused area of the document (or the document itself, or even an element) is normatively aligned with HTML's definition.
That's why we're seeking implementer feedback. When it comes to implementing user intents like "activate element" or "move to next heading", what kind of language would be most helpful (or least ambiguous)?
(I'll add that we aspire for this protocol to one day be useful for automating interactions with native software. While this may be a challenge for another day, we'd like to avoid making decisions which preclude expansion in that direction.)