Potential need for a new CSS pseudo-class for role matching

[jnurthen]

From @therealglazou on December 29, 2016 11:39

Testing in CSS an epub:type fallback for an absent role attribute is a bit cumbersome. Such a selector would be in the form of (for a cell role):

[epub|type="cell"]:not([role]) {
   ...
}

and it would require a namespace at-rule:

@namespace epub "http://www.idpf.org/2007/ops";

Maybe a new :role() pseudo-class is needed here to ease the pain and deal automagically with that fallback; some UAs could look only at role to resolve it while EPUB Agents would also deal with the epub:type fallback. Just thinking out loud here.... That would of course require joint work with the CSS WG.

Copied from original issue: w3c/aria#506

[jnurthen]

From @AmeliaBR on December 29, 2016 15:41

There is definitely a need for an API to access the computed role of an element. The role itself can have fallback values, and values that are dis-allowed because of native semantics, and of course default roles based on semantics. A CSS pseudo-class has the benefit of exposing this information to both scripts and declarative styling.

[jnurthen]

@IanPouncey something for the CSS Accessibility Task Force to consider?

[digilou]

APA has created an issue for the CSS-WG to see where this conversation needs to go.

[astearns]

Please use this older issue for the csswg: w3c/csswg-drafts#3596