This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
Hey Steve I think one absolutely critical area to think about from an interoperability perspective is handling "interesting" markup combinations. For example, if you have an <li> that does not have a <ol>/<ul> parent, what does that get mapped to? Or, if you have an <li> that has a <ol>/<ul> parent, but the parent has a role attribute that maps it to (say) the "button" role, what does the <li> get mapped to?
I agree that "interesting" markup combinations are something that should be tested against the proposed mappings and the accessible name and description calculation, as part of the work in making sure we've got those right. For example, see bug 25371. But I'm not sure we should be looking at cases of invalid markup. Neither the <li> without a <ol>/<ul> (or presumably <menu type="toolbar">) parent, or an <ol>/<ul> with role="button" are permitted.