Handling Author Errors: form & region roles

[scottaohara]

closes #1083

adds initial draft content to the handling author error section to call out expected behavior for form and region landmark roles which have not been provided accessible names.

---

Diff

[smhigley]

For form, this looks like a substantive change from what happens currently. It looks like <div role="form"> is still exposed as a form landmark even without a name (checked Firefox and Chromium on this quick example: https://jsfiddle.net/py96hL7b/, and Safari appears to map to Group whether named or not).

Is that something we actually want to change? Out of all landmarks with name required, I can most see form being valuable to expose anyway.

[scottaohara]

Fair point. The intent here was to align how HTML AAM treats unnamed <form>.

If we do not want to make this change then this is really just to better specify what happens with region, and arguably that could just be part of the definition of that role, since it'd be unique to that one alone.

[smhigley]

To be fair, aligning to an unnamed <form> also makes sense. I'm not really against either approach, maybe we should just make sure browser folks are on board with updating the mappings?

[scottaohara]

Adding agenda label to hopefully get this quickly resolved.

Solutions - keep PR as is if people agree that an unnamed form role should behave like HTML's <form> and not be exposed if unnamed. Webkit mapping to group seems odd regardless.

Alternative, we decide that form if not named should still show up as a landmark (in which case, is it really required to name it???). But if so, then maybe this PR changes to note region's behavior solely within the region definition. Not sure if we need to call this out in handling authors errors if it's behavior just for this one landmark.

[scottaohara]

TODO: make sure that this PR indicates that the fallback role for the element is what is to be exposed AND if someone were to do <el role='form article'> that the fallback role in the role attribute would be respected.

[scottaohara]

@MelSumner I have finished my TODO, so this is ready to review again.

@cookiecrook I added you as a reviewer as it was your comment (if i recall correctly) that prompted me to add the new bits to this PR. If you do not wish to review, please ignore.

[MelSumner]

As an aside, I wonder if browsers could add some sort of warning to the console log. 🤔

Anyway, thank you for doing this work

[scottaohara]

question, but should I also update the form and region roles following text:

landmarks.

to

landmarks.

bold "named" for emphasis, but not because i think it should be bold

[smhigley]

@scottaohara that sounds like a good idea to me 👍

[scottaohara]

thanks @smhigley. added some words to those lines for each role.

[chlane]

I agree with this change. If a form exists, it probably represents the purpose of the page and AT users should at least understand they are interacting with a form, even if it is unnamed. Anonymous regions aren't perfect but they still can provide a sense of structure for AT users.