required owned and context roles

[rdeltour]

It seems there's something wonky with the relationships established with required owned elements, required context role, and role subclasses defined in DPUB ARIA.

ARIA 1.1 has a note in the Required own elements section that says:

An element with a subclass role of the 'required owned element' does not fulfill this requirement. For example, the list role requires ownership of an element using either the listitem or group role. Although the group role is the superclass of row, adding a owned element with a role of row will not fulfill the requirement that list must own a listitem or a group.

This means for instance that doc-biblioentry cannot be used as the unique children of a list, which invalidates Example 6:

<section role="doc-bibliography">
   <h1>Cited Works</h1>
   <div role="list">
      <p role="doc-biblioentry" id="b8cab5dd-bc24-459c-9858-7afa9da69b64">
         John Steinbeck, The Grapes of Wrath (New York: The Viking Press, 1939)
      </p>
   </div>
   …
</section>

In other words, doc-bibliography is only usable on li children of ol and ul.

Can we clarify that with the ARIA folks (cc @w3c/aria)? This looks like an oversight in how DPUB extends ARIA, or maybe even in ARIA's own extensibility.

[mattgarrish]

In other words, doc-bibliography is only usable on li children of ol and ul.

But it wouldn't be, if we follow that logic, would it? Both of those have the implicit role of list so they also couldn't allow listitem subclasses. Strictly interpreting that note, you always have to create new list types in order to create subclasses of listitem, at least outside the core.

I agree that the note adds confusion, especially since it's technically only informative and I don't see the rule expressed normatively, but it can't be solved here. I appreciate why it exists, too, but the problem is that extension modules can't modify the core vocabulary, and the core vocabulary doesn't reference extension roles, so until a role is added to the core there will be this ambiguity.

What we probably need is a way to confirm that the roles are, indeed, expected to be included in the expected owned role list when in extensions ("required owner", or something like that).

But that assumes that the way we created the vocabulary is correct, which I'm holding out hope is the case. If not, then the problem will require reworking the vocabulary.

/cc @TzviyaSiegman @joanmarie

[rdeltour]

See also w3c/aria#748 which proposes a stricter definition of "owned", which would break the multi-ancestors required contexts of roles like doc-biblioentry (which requires both a direct list owner and doc-bibliography grand-owner).