After <http://trac.webkit.org/r127936>, the fix for bug 94870, multiple accessibility tests began to fail. See the URL for example.
I'll look at these and either fix or revert shortly.
I took a look; no new bugs, just need to rebaseline some expectations.
Committed revision 127963.
In addition to changing expectations, you changed a test: > Updating one test (update-children-when-aria-role-changes.html) that > incorrectly assumed a node with an aria-lable would be ignored. Used title > instead of aria-label. This kind of change requires review.
(In reply to comment #4) > In addition to changing expectations, you changed a test: > > > Updating one test (update-children-when-aria-role-changes.html) that > > incorrectly assumed a node with an aria-lable would be ignored. Used title > > instead of aria-label. > > This kind of change requires review. Sorry. Fixing this now - I'll revert the offending change, fix just the expectations, then upload a patch to fix the test for review.
Reverted r127963 for reason: Inappropriately modified test, and not just expectations, without review Committed r127992: <http://trac.webkit.org/changeset/127992>
Fixed the text expectations only: Committed r127993: <http://trac.webkit.org/changeset/127993>
Created attachment 163013 [details] Patch
Comment on attachment 163013 [details] Patch do you know what kind of role a <span> becomes if it's not ignored all of a sudden. i want to make sure we won't start exposing AXUnknown elements if someone puts an aria-label on a <span> without a role mapping
(In reply to comment #9) > (From update of attachment 163013 [details]) > do you know what kind of role a <span> becomes if it's not ignored all of a sudden. i want to make sure we won't start exposing AXUnknown elements if someone puts an aria-label on a <span> without a role mapping It will be AXGroup. I tested it to confirm, but the logic is at the very bottom of determineAccessibilityRole: if (supportsARIAAttributes()) return GroupRole;
Comment on attachment 163013 [details] Patch that seems like reasonable logic.
Comment on attachment 163013 [details] Patch Clearing flags on attachment: 163013 Committed r128072: <http://trac.webkit.org/changeset/128072>
All reviewed patches have been landed. Closing bug.