Meeting minutes
Zakim: next item
New Issue Triage
aaron: On Windows, JAWS and NVDA connect to DOM with IA2, but they can also use ISimpleDOMNode, an older API.
aaron: data-mathml is used by Google Sheets and MathJax. Its used to provide MathML representations of images.
peter: MathJax doesn’t use data-mathml today.
aaron: Okay, then lets close this issue.
aaron: Google Sheets can surface the MathML in a different way.
jcraig: MathJax ≥3 keeps the MathML
aaron: Peter, could you send a snippet of what the code should look like?
peter: Sure
aaron: This is for student tests. Pearson, ETS, etc.
aaron: It’s for testing whether a student can read text aloud, so that text is put in a non-standard brailleonlyregion, so that it isn’t otherwise read aloud.
spectranaut_: let’s agenda that
aaron: data-at-shortcutkeys is put on a non-focusable element. When JAWS sees it, it parses the JSON, then tells the user what keys are available—and tries to avoid key conflicts.
aaron: Do screen readers otherwise have trouble finding these key shortcuts?
matt_king: Many apps have other ways to get to key shortcut information.
matt_king: JAWS has a feature for disabling key shortcuts with conflicts.
matt_king: Let’s agenda this
spectranaut_: Is aria-required and aria-readonly allowed on columnheader only in a grid (not treegrid)?
matt_king: It is an inconsistency.
spectranaut_: We’ll agenda this
scott: This is something that needs to be talked about. In HTML, you can use `placeholder` on number inputs, but in ARIA, we don’t list `aria-placeholder` as an allowed property on comboboxes or spin buttons.
scott: It’s not as simple as “allow these”, because it doesn’t always make sense to do so.
spectranaut_: Can someone add a clarifying comment to the issue?
scott: I can, eventually, though others are welcome to follow-up too.
aaron: I worked this out with Scott; he has a PR.
aaron: You put `aria-setsize="-1"` when you don’t know the number of objects and `auto` can’t compute it.
New PR Triage
<scott> w3c/
https://
scott: There’s an issue open for this, from 3 weeks ago.
scott: It’s a question re: how `autocomplete` and `aria-autocomplete` relate; this PR attempts to clarify that.
scott: This is not editorial.
scott: `<selectedoption>` is a new HTML element for customizable selects.
scott: It represents the value of the chosen option.
scott: We should think about e.g. what happens if someone tries to overwrite the visible text with an `aria-label`.
scott: This should match the select’s value.
scott: This PR needs discussion; it’s a partial PR right now.
Rahim: There is a great explainer, with a link to the implementation in Chrome. Is that linked from this issue?
scott: No. Also that wouldn’t say anything about the mappings.
melsumner: This PR is in draft because I thought there might be additional Prettier options we may want to add.
rahim: This is a migrated PR
spectranaut_: That’s all the PRs since TPAC.
Remaining new issue triage
jcraig: This is follow-up to Rahim’s issue (linked)
aardrian: The documentation for the `<abbr>` (element) and `abbr` (attribute) is confusing.
scott: I tried to resolve this in 2019. The mappings in the spec have always been wrong. Help is welcomed.
aardrian: I was going to file a separate issue about mapping and naming. Thanks for validating this is an issue.
spectranaut_: I’ll assign Adrian to the previous issue and close this one as a dupe.
spectranaut_: We already discussed the PR related to this issue.
aardrian: This will rollup into the larger table work I’m doing.
spectranaut_: That is it for ARIA issues; moving on to other repos.
<pkra> a few more:
<pkra> w3c/
<pkra> w3c/
<pkra> w3c/
<pkra> w3c/
aaron: We discussed this at TPAC; people were generally in favor.
aaron: Rather than requiring `aria-actions`, we can detect these nested-controls scenarios.
matt_king: Does this depend on `aria-actions`?
aaron: It depends on the actions interface of the platform (e.g. IAccessibleAction), not on `aria-actions` itself.
aaron: ATs will need changes to make use of this.
matt_king: Those changes would be in line with whatever we do to support `aria-actions`?
aaron: Yes; it’s the same mechanism.
spectranaut_: Is there anyone we should try to get feedback from, at this stage? Did we talk about focusgroup at TPAC?
jcraig: We talked about it in the context of carousels, and the inertness side-effect.
keithamus: This is being discussed in the OpenUI community groups, but may have stalled.
aaron: This issue can be closed; there is no mapping required. It’s a keyboard navigation behavior.
spectranaut_: And that keyboard behavior would be specified in HTML.
scott: I don’t agree, based on the testing I’ve done, e.g. automatic forms mode. Hyperlinks in a focusgroup trip up NVDA/JAWS. We may need an object attribute.
spectranaut_: I’ll agenda this for further discussion.
spectranaut_: Agenda’d.
aaron: Mozilla is doing this already; Chrome isn’t yet. When you navigate to a page, you can jump directly to specific text, similar to a named anchor, but better.
aaron: We have some mappings, though using an undocumented property (AXHighlight) for Mac, and nothing for UIA.
jcraig: I’ll look at the Mac mapping.