Bug 224617 – AX: inverse role synonyms "presentation" and "none" to make "none" the primary

• Status: RESOLVED
• Resolution: FIXED
• Component: Accessibility

[James Craig]

AX: inverse role synonyms "presentation" and "none" to make "none" the primary

ARIA "none" has had very good browser support for a number of years, so the ARIA spec is making it the primary role (over the original name "presentation"). Currently "presentation" is the returned role in WebKit's reverse enum lookup (computedRoleString IIRC?). Since "presentation" will be deprecated, the computed role string should return the ARIA 1.1 "none" role, over the 1.0 "presentation" role. The deprecated "presentation" role should remain in WebKit for backwards compatibility.

[Radar WebKit Bug Importer]

<rdar://problem/76713760>

[James Craig]

related bug 224615

[James Craig]

To clarify where this will result in end-user impact... The WebKit Web Inspector Accessibility Node Inspector will return role "none" or "none (computed)" See attached image of that UI.

[James Craig]

Created attachment 426120 [details]
screen shot

[chris fleizach]

Created attachment 426831 [details]
patch

[Devin Rousso]

Comment on attachment 426831 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=426831&action=review

> LayoutTests/inspector/audit/run-accessibility-expected.txt:11
> -Result: []
> -Audit teardown...
> +Result: Audit teardown...

This seems scary.  I wouldn't expect this looking at your other changes 🤔

[chris fleizach]

(In reply to Devin Rousso from comment #6)
> Comment on attachment 426831 [details]
> patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=426831&action=review
> 
> > LayoutTests/inspector/audit/run-accessibility-expected.txt:11
> > -Result: []
> > -Audit teardown...
> > +Result: Audit teardown...
> 
> This seems scary.  I wouldn't expect this looking at your other changes 🤔
I agree. I was hoping someone would say, "that looks expected"

I don't understand why it appeared, but will dig into it further

[chris fleizach]

Created attachment 427078 [details]
patch

[Devin Rousso]

Comment on attachment 427078 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=427078&action=review

rs=me

> Source/WebCore/accessibility/AccessibilityObject.cpp:2182
> +    // http://webkit.org/b/224617 - prefer "none" as the canonical role to be used for presentationals (instead of "presentation"

Rather than have a link to the bugzilla, can you explain inline why this is the way it is (ideally with a link to a spec if possible)?

[James Craig]

Tagging WPTImpact because this may fail with some of the impending role tests coming with the Interop 2023 Accessibility Investigation.

https://github.com/w3c/core-aam/issues/166

[James Craig]

FYI the link to the spec change Devin asked for is now in this ARIA Core-AAM patch: 
https://github.com/w3c/core-aam/pull/167/files#diff-0eb547304658805aad788d320f10bf1f292797b5e6d745a3bf617584da017051R1097

[James Craig]

Sorry. Diff link is actually:
https://github.com/w3c/core-aam/pull/167/files#diff-0eb547304658805aad788d320f10bf1f292797b5e6d745a3bf617584da017051R1640

So final link if you reference it in the patch or commit msg is:
https://w3c.github.io/core-aam/#role-map-img

[chris fleizach]

Created attachment 465579 [details]
Patch

[chris fleizach]

Created attachment 465580 [details]
Patch

[chris fleizach]

Created attachment 465584 [details]
Patch

[EWS]

Committed 262119@main (d5fbadeac7c4): <https://commits.webkit.org/262119@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 465584 [details].