feat: add IA2 and ATK mappings for text edit maxlength

[smhigley]

Part of the initial work for w3c/aria#1119

Splitting out each API into a separate PR for ease of review and discussion. I've combined IA2 and ATK here since they generally seem to have the same approach re: Object Attributes, and existing object attributes look like they have the same keys across both APIs

---

Preview | Diff

[joanmarie]

Is there going to be a corresponding ARIA property now or in the future? If so, will it be aria-max-edit-chars?

In general, when ARIA properties are exposed as object attributes, the rule tends to be aria-foo is exposed as foo. So if the answer to my second question is "yes", then this change looks good to me. But if the ARIA property will have a different name, please figure out what that is first and update this mapping accordingly.

Thanks for checking!

[smhigley]

@joanmarie I believe the planned aria attr would be aria-maxlength, should the object attribute just be maxlength then?

max-edit-chars was one of the ones @aleventhal suggested, but that was just a quick chat and not a super in-depth dive. Also if you want this to be a draft until the ARIA attribute is in PR, or go directly to Core AAM, let me know. I just didn't want to hold up browser & screen reader work till the ARIA attribute is added.

[joanmarie]

@joanmarie I believe the planned aria attr would be aria-maxlength, should the object attribute just be maxlength then?

I think so, yes -- unless @aleventhal has a compelling reason why it should be exposed to screen readers as max-edit-chars. As the Orca maintainer, I cannot think of one. But I don't know if there are concerns on the Windows side of things.

I also think this will make it possible for you to get at least one ARIA implementation "for free" (i.e. Gecko). I just loaded the following in Firefox:

data:text/html,<div role="textbox" aria-maxlength="50" aria-doesnotexist="oh really?">foo</div>

Then I used the Accerciser accessibility inspector in Linux to query the properties of the AtspiObject that corresponds to the div:

• acc.getRole() returns <enum ATSPI_ROLE_ENTRY of type Atspi.Role>
• acc.queryText().getText(0,-1) returns 'foo'
• acc.getAttributes() returns ['display:block', 'maxlength:50', 'tag:div', 'doesnotexist:oh really?', 'xml-roles:textbox']

Gecko doesn't sanity-check or rename ARIA attributes to object attributes for ATK or IA2 -- other than stripping off the "aria" prefix -- unless there's a strong reason to do so. 🎉 😁

I just didn't want to hold up browser & screen reader work till the ARIA attribute is added.

Understood. And I don't want to hold up spec work. Perhaps check with @jcsteh at Mozilla regarding his preference and thoughts about max-edit-chars being the attribute to be exposed for HTML?

[jcsteh]

I'd prefer the object attribute to be named according to the ARIA attribute. For example, if it's aria-maxlength, I think the object attribute should be maxlength. Beyond that, I don't have a strong preference what the ARIA attribute is called.

[smhigley]

@joanmarie / @jcsteh updated to maxlength!

[joanmarie]

LGTM. Thanks! I'll add this to my Orca TODO list.

[jcsteh]

I assume that when we get aria-maxlength, this mapping will be updated so that it simply refers to the ARIA mapping and core-aam will absorb the platform specific guidance here?

[scottaohara]

@jcsteh yes, that would be the end goal. if an aria attribute is created, then the mappings go into core aam, and html aam is updated to point to that.

[scottaohara]

closing as this PR has been superseded by w3c/aria#2273