Open julienw opened 9 months ago
It looks like it's been changed in https://github.com/w3c/html-aria/pull/402 (found that from the change on MDN in https://github.com/mdn/content/commit/e31cb5978e9f3c731c49db9ed0a15795b870e141#diff-25921a273c422a060a3204a4c8e46fb9d42be0ce77679b9d1e5094c9fc627985)
Should the fix be made in https://github.com/A11yance/aria-query/blob/main/src/etc/roles/literal/formRole.js ?
Looking at:
I’m inclined to agree, all the specs point to the form element having the form role regardless of accessible name. As you describe the only condition is regarding whether it’s also considered a landmark.
In addition to the update to aria-query, perhaps worth removing the comment in https://github.com/eps1lon/dom-accessibility-api/blob/main/sources/getRole.ts#L30 as well (although no code change needed as seems already assumed a form
role)
Some more context:
The last one means that WAI-ARIA is somewhat in conflict with the AAM specs currently as it still states that an accessible name is required on a form (something that is a bug in in aria-query
as it has accessibleNameRequired: false,
currently).
If we assume that WAI-ARIA is updated (e.g. in 1.3) then we could ignore that aria-query
has been inconsistent / incorrect till now on this one point.
Prior issue #1021
Testcase:
Result:
@testing-library/dom
version: 9.3.4In the previously mentioned bug, @eps1lon explained that this was following the roles from html-aam. However it looks like they've been updated, see https://www.w3.org/TR/core-aam-1.2/#role-map-form-nameless. It looks like that a
form
element without an accessible name gets a computed roleform
too. However they mention that it shouldn't be exposed as a landmark. Is that where my interpretation is wrong?Thanks for your feedback!