Add note addressing validation errors resulting from aria-expanded on menuitem elements to Menu and Menubar patterns and examples

[mcking65]

Feedback in issue #410
from @sideshowbarker describes confusion caused by validation errors that result from applying aria-expanded to menuitem elements.

Per discussion with task force, add notes to the patterns and examples where use of aria-expanded is recommended to explain that the ARIA specification mistakenly omitted aria-expanded for menuitem and this will be resolved in ARIA 1.2. In the meantime, even though it triggers errors from validators, using aria-expanded is still recommended.

[mcking65]

@sideshowbarker, @Decrepidos, @stevefaulkner, @tmorehouse,

Commits dc4e6ef, 310801c, and 1c2446b add the following note to the design pattern, navigation menubar example, and editor menubar example pages. The task force reviewed this language in today's meeting.

Currently, using aria-expanded on elements with role menuitem triggers HTML validation errors because the ARIA specification does not yet support doing so.
The ARIA working group plans to resolve this issue in the next version of the specification.
Until a version of ARIA that resolves the issue becomes a W3C recommendation,
it is safe to ignore these validation errors.
Alternatively, since only a few browser and assistive technology combinations exploit this feature of the pattern,
it can be omitted from implementations.

Closing this issue as resolved. Please feel free to re-open if you believe we have not adequately addressed the issue.

[sideshowbarker]

I am not satisfied with this resolution.

It’s not OK to tell authors, “Ignore what the ARIA spec says and ignore what the validator says."

And as I’ve noted over in 310801c#commitcomment-24845322, we don’t need for the ARIA spec to be at Recommendation as far as the validator is concerned. We just need for the ARIA WG to provide a spec with an actual normative requirement for this that the validator can then conform to. It doesn’t need to be Rec, and it doesn’t even need to be Working Draft. It can just be an Editor’s Draft.

[mcking65]

@sideshowbarker commented:

I am not satisfied with this resolution.

It’s not OK to tell authors, “Ignore what the ARIA spec says and ignore what the validator says."

Isn't this different from asking people to ignore the spec when there is a clear plan for the spec to resolve the issue?

we don’t need for the ARIA spec to be at Recommendation as far as the validator is concerned. We just need for the ARIA WG to provide a spec with an actual normative requirement for this that the validator can then conform to. It doesn’t need to be Rec, and it doesn’t even need to be Working Draft. It can just be an Editor’s Draft.

While liberating, I am surprised and confused. Since ARIA is a rec track doc, and since features do get pulled from the editor's draft, does that mean that a page that validates today may not validate tomorrow because it uses a feature that validated today but will not tomorrow because it was pulled from the editor's draft and then from the validator?

In the case of aria-expanded on parent menuitems, I believe we have consensus. And, we have enough implementation in browsers and ssistive tech to prove value. So, the risk of a plan change here is extremely low.

@michael-n-cooper, @joanmarie, when will we be able to merge ARIA 1.2 changes into the editor's draft? Possibly before ARIA 1.1 is a recommendation and before we publish the APG?

[michael-n-cooper]

In theory we could set up editors' draft for ARIA 1.2 before ARIA 1.1 is a Rec. However, one major gating factor is the repository split, which we want to do after 1.1 has stabilized and before we start work on 1.2. Also I believe Joanie has in mind a versioning via branches approach that I'm not fully clear on, so we need to get that set up. These things need "sitting down to" but I know Joanie is flat-out trying to get 1.1 past CR right now so wouldn't look at this right away.

I also worry about having a 1.2 editors' draft at URI being used for the 1.1 editors' draft while 1.1 is still under development. We could of course set up a different URI but I don't think we were expecting to.

All this to say, I'm not sure we can promise what you want before Rec, but we can explore doing it in a few weeks if it seems not too crazy for our schedules.

[mcking65]

@sideshowbarker,

We now have an editor's draft of ARIA 1.2 where the menuitem role now states that it supports aria-expanded.

We are planning to publish the ARIA Authoring Practices 1.1 as a Note this week. We have 24 hours left before cutting the publication branch. We could remove these notes about ignoring the validator error. Or, we could revise them to state that it is safe to ignore them until the validator is updated, which will be done in the near future ... or something like that.

[sideshowbarker]

We now have an editor's draft of ARIA 1.2 where the menuitem role now states that it supports aria-expanded.

Excellent — I can get the validator updated today. Will post another comment here as soon as I’ve committed and pushed the change

[mcking65]

Fixed with commit 28bf229 so we can cut our publication branch this afternoon for publication on Thursday.

[sideshowbarker]

I’ve updated the HTML checker sources and pushed the update to https://validator.w3.org/nu/