W3C

– DRAFT –
ARIA WG

28 May 2026

Attendees

Present
aardrian, Adam, Daniel, filippo-zorzi, Francis_Storr, giacomo-petri, HaTheo, Jacques, jcraig, katez, Rahim, sarah, scott, smockle, Stefan
Regrets
-
Chair
-
Scribe
aardrian, jamesn

Meeting minutes

zakim next item

New PR Triage

<github-bot> I can't comment on that because it doesn't look like a github issue to me.

jamesn: editorial for the first one

daniel: i'll fix it

jamesn: fix role attr definition. Mostly editorial, but would appreciate reviews.

jamesn: long-standing issue about who defines role attribute (maybe a bit of concept)

jamesn: If you have both attribute and concept in the same spec, the references go to the attribute, which isn't what we always want.

jamesn: reviewers?

spectranaut: I'll review

adam: I'll review

rahim: I thought role was a separate spec at some point. Any history?

jamesn: It was an ARIA precursor, but ancient history.

jcraig: Was mostly tied to XML before ARIA.

jcraig: something about hair

jamesn: Two chores, so we can skip the next two.

jamesn: Without meeting last week, may have... nothing else.

New Issue Triage

<github-bot> I can't comment on that because it doesn't look like a github issue to me.

jamesn: A long issue triage since it's been a month.

jamesn: 2797: "headings should not be labeled by author" has some feedback

jamesn: should we close it?

scott: I appreciate it brings up issues from the past that have gone nowhere.

jamesn: We have duplicate issues.

scott: Maybe not the landmark part.

scott: Though he may have meant exposing the heading when navigating to a landmark.

jamesn: Do we need to talk about this? It seems like a duplicate.

<Rahim> Curious how this interacts with nameFromHeading as well

spectranaut_: We should agenda this.

jamesn: 2796: Imapact of AI tech... should we agenda this?

jamesn: You can also add stuff

spectranaut_: The intent was to collect thoughts

Daniel: We can set aside time for discussion and bring it back to the issue.

jamesn: CSSAAM 20: reorg repo labels

Daniel: I'll look

jamesn: CSS-AAM #19, we can skip that for now but leave comments.

jamesn: ARIA 1.3 WD is for later

jamesn: 2791: posinset, asking for a deep dive. Matt, do you think it's needed?

Matt_King: Or you can handle it in a regular meeting and then see if it warrants a deep dive.

jamesn: I'll agenda for next week.

Matt_King: I just remembered -- we need specific people for the discussion. SR reps.

jamesn: 607 simply aside mappings; it's on today's agenda

jamesn: 2785: overscroll, deep dive last week

jamesn: html-aam 606, for HTML spec with readonly attr on fieldset.

scott: I provided questions, but it's not a perfect analogue. Nothing to do right now.

scott: I can ask the mappings for the proposal.

jamesn: html-aam 605: whitespace-only alt attribute

jamesn: Just adding tests.

giacomo-petri: Yes, and maybe a question for ARIA in HTML

giacomo-petri: For an empty image alt or one with white space

jamesn: Can you file against ARIA in HTML?

giacomo-petri: Yes.

giacomo-petri: I suggested a change to a PR to mention the examples to discuss how browsers expose them.

jamesn: Yeah, that's inconsistent and we should have test, suggest how they should be implemented.

jamesn: Are we agreed on implementation?

giacomo-petri: The PR says empty alt should have the same as alt with white space.

<Rahim> https://infra.spec.whatwg.org/#ascii-whitespace

Rahim: HTML spec has a definition of white space, is that what you want to use Giacomo?

jamesn: This would also need... in HTML-AAM?

scott: We already have this in HTML AAM, but this addresses a gap.

jamesn: So it says white space?

<smockle> https://www.w3.org/TR/html-aam-1.0/#el-img-empty-alt

scott: We aligned on this a while ago to address how aria-label behaves.

scott: We didn't have tests to cover it. Giacomo found that.

Rahim: Did we decide where the tests live?

scott: Randomly into WPT folder [something about shooting from a cannon]

Rahim: Giacomo you proposed roles.html?

giacomo-petri: Yes. And to match previous test cases. I'm happy to add more.

jamesn: Why is the ED old?

jamesn: Is this not the correct ED? Why does the URL point to it?

jamesn: I'm looking at the aria-slash URL.

scott: I've never once put "ARIA" into that URL. I thought it just redirected.

jcraig: CLicking the latest ED shows correct URL

jamesn: 2782: draft menu element proposal, for deep dive.

WPT Open PRs

<github-bot> I can't comment on that because it doesn't look like a github issue to me.

jcraig: The newish ones have active comments and review threads.

spectranaut_: Did you see my more recent comment about adding tests to AAM for role?

jcraig: I see them. From two weeks ago.

spectranaut_: Oops, might be a draft comment, unpublished.

Deep Dive planning

<github-bot> I can't comment on that because it doesn't look like a github issue to me.

jamesn: Do we want to schedule posinset / setsize?

jamesn: Would need afternoon meeting?

Matt_King: Yes, if we want to include Jamie or someone from NVDA.

Matt_King: SHould reach out to as many as possible.

jamesn: Put in the thread who you want to get in the meeting, then we can request times.

Matt_King: Ok

Publishing 1.3

jamesn: Daniel?

Daniel: We published FPWD in 2024, with a bunch of PRs merged, some with normative changes.

Daniel: In order to finish charter year, we should publish WD as soon as possible.

Daniel: I know ARIA Actions is not ready, but we can publish what we have now.

Daniel: If we reach consensus with ARIA Actions before CR, we can still point reviewers there.

Daniel: To move ARIA to REC by end of year, need to do soon.

jcraig: ARIA Actions is implemented behind runtime flag in Safari Tech PReview.

jcraig: I think we need second implementation

Jacques: I'mw working on Chromium implementations.

jcraig: We shouldn't move unless we know there's an active implementation during CR

Daniel: Yes, it should be on the calendar to avoid it being at risk.

Daniel: If second implementation doesn't happen, we can withdraw feature.

Daniel: But it's still not in the ED.

jcraig: I thoujght you said you _Were_ going to merge

Daniel: No, I want a publish WD with what we have, without ARIA Actions.

jcraig: It would be nice to have ARIA actions.

jcraig: CR isn't the final month-long review?

Daniel: No. CR gets you 10 months, and then we ask for move to REC (for another month.)

jcraig: Are you far enough along, JAcques?

Jacques: My goal is to get this done within 2 months and then origin trial.

spectranaut_: Runtime flag is fine

Jacques: I think implementation works in most platforms. But two months.

Matt_King: We should address Daniel's question as posed given outstanding questions that can impact timelines.

jcraig: Fair. I withdraw my rush.

spectranaut_: We cannot delay this WD publication. We go with what we have.

<Jacques> that sounds good to me.

spectranaut_: Then we can ask others to review WD

jcraig: We shouldn't rush it. Go to CR without this hanging over us.

jamesn: THe question is publishing WD for wide review

jamesn: Go to CR as soon as wide review is done.

jamesn: Do we need a formal resolution?

Daniel: It would be nice.

RESOLUTION: group agrees to publish a new Working Draft of ARIA for wide review

Reviews for Android Mappings

spectranaut_: I snuck this in

spectranaut_: I want to get a resolution on Android mappings.

spectranaut_: Do I need a consensus to merge this?

jamesn: ANy objections?

RESOLUTION: to merge 2682

RESOLUTION: to merge 2682

Simplify aside mappings

Matt_King: This came up in APG TF a month or so agao.

Matt_King: Looked at updating guidance on complementary landmarks

Matt_King: It seemed more complex than needed.

Matt_King: The proposal is that an aside element only becomes a landmark when it's named or is context of body

Matt_King: So it would be closed to other HTML elements that map to landmark regions

Matt_King: Otherwise it's a far more complex algorithm.

scott: Part of the history is that I asked for help with browser bugs. We've already made changes in ARIA spec to align.

scott: I don't agree it's too complex. Only change seems to not restrict to main element.

scott: Feels like undoing work that's been decided but only partly implemented.

Matt_King: We some texts related to header and footer.

scott: I don't know what that's in reference too.

scott: I don't know what that's in reference to.

scott: But those are also partially resolved with section header/footer.

spectranaut_: We talked about this in the past and try not to go back too often on our previous decisions.

spectranaut_: THis is the last 10% taking 90% of the time because browsers are slow to implement.

spectranaut_: I caution against changing past decisions.

aardrian: Was the proposal to expose regardless of where in DOM and unnamed?

<Jacques> I'm always happy to help folks contribute to chromium! Just reach out.

Matt_King: No, only if in context of body, otherwise it would require a name.

Matt_King: ARIA says you can have a complementary inside main. But this is closer to what we require for section to be named.

Matt_King: Every pull quote becomes a complementary region, because they use it wrong.

Matt_King: So we end up with tons of complementary regions.

aardrian: WOuldn't that be an implementation issue?

Matt_King: That would be an HTML AAM decision.

scott: The previous decision was that we would keep it mapped to complementary both scoped to body or main

jamesn: If we want to do this, then we need sufficient evidence.

<aardrian> +1

jamesn: Web sites where this would cause a problem and web sites where it would be better.

Matt_King: I don't know how you prove the case of degradation.

jamesn: Search for a bunch of sites with this pattern and test them.

jamesn: If you'd expect them to be complementary, then you should investigate how it was done.

Rahim: I'm curious why this wouldn't be an author requirement or best practice instead.

Matt_King: Authors don't know it will become complementary by default.

jamesn: If you want to move this forward, I think you need examples where it breaks things.

Matt_King: Ok. I think we can find some way to do that.

jamesn: Let's leave it until there's more evidence. Nobody seems to object if there's evidence.

feat: aria-actions addition to the ARIA spec

sarah: I'm curious what Matt meant that there's a bunch of open issues.

sarah: I found one, but things seem stable.

Matt_King: I submitted a review just before this meeting, but there may be 3 I raised.

sarah: Are you sure they aren't addressed? I fixed one and two before the meeting.

sarah: They were editorial and I added wording that was covered in the spec.

Matt_King: I'll re-review.

sarah: I left one because it's the focus question in 2691. It's already in Core AAM

Matt_King: Thank you, I will re-review.

Matt_King: Then our primary thing is to address 2691.

Matt_King: SHould there be language related to focusability of referencing element?

sarah: The referencing element needs to be focusable, not referenced.

Matt_King: We need an author MUST for the element.

sarah: It's heavily implied?

jamesn: Sounds like a review comment.

Matt_King: Yes, I'll leave that when I re-review.

sarah: For 2691, we don't define behaviors like focus movement in ARIA or Core AAM up until now.

sarah: We should nail this down, but I'm curious how it fits into our specs.

Matt_King: It's an extension of author's ability to manage focus.

Matt_King: When we first wrote managing focus, it was related.

sarah: We have wording about author focus management, but we lack browser-focused focus management behavior.

sar

jamesn: We're at time...

<Jacques> where can we continue this conversation?

sarah: If I were to draft wording, how would it look?

jamesn: Time.

jamesn: We've had ARIA Actions on agenda a lot, but we haven't had the people.

jamesn: Agenda it again.

jamesn: REviews are helpful outside of meeting.

Summary of resolutions

  1. group agrees to publish a new Working Draft of ARIA for wide review
  2. to merge 2682
  3. to merge 2682
Minutes manually created (not a transcript), formatted by scribe.perl version 248 (Mon Oct 27 20:04:16 2025 UTC).

Diagnostics

Succeeded: s/We cannot delay CR/We cannot delay this WD publication/

Maybe present: jamesn, Matt_King, spectranaut, spectranaut_

All speakers: aardrian, adam, daniel, giacomo-petri, Jacques, jamesn, jcraig, Matt_King, rahim, sarah, scott, spectranaut, spectranaut_

Active on IRC: aardrian, Adam, Daniel, filippo-zorzi, Francis_Storr, giacomo-petri, HaTheo, Jacques, jamesn, jcraig, katez, Rahim, sarah, scott, smockle, spectranaut_, Stefan