W3C

– DRAFT –
ARIA WG

15 May 2025

Attendees

Present
Adam_Page, ChrisCuellar, CurtBellew, filippo-zorzi, Francis_Storr, katez, melsumner, Siri
Regrets
-
Chair
-
Scribe
Adam_Page

Meeting minutes

New Issue Triage

spectranaut_: core-aam#249

spectranaut_: this issue is about how Webkit does something differently than other browsers with aria-level="0"
… I’ll assign to myself
… and might agenda it if we need to talk about it

spectranaut_: aria#2529
… scott opened a PR, so we can talk about it then

New PR Triage

spectranaut_: aria#2533
… to update role="scrollbar" AXAPI platform mappings
… we need to confirm whether this has been implemented
… and land this PR
… just representing reality
… I’ll also assign myself to this one
… probably don’t need more than 1 reviewer for this kind of PR

jamesn: agreed

spectranaut_: aria#2532
… editorial

jamesn: I’ll review this one

spectranaut_: aria#2531
… talked about in editors meeting
… pkra has a way forward
… will talk about when it‘s ready for review

spectranaut_: aria#2530
… closing #2529 which scott opened
… clarifies that placeholder takes priority over aria-placeholder

scott: Giacomo noticed the lack of clarity in a WPT test
… it just calls out “the placeholder” but doesn’t disambiguate
… it’s worth noting that in current reality the placeholder wins out over its ARIA counterpart, which is the opposite of many other ARIA attributes
… so the PR makes that abundantly clear

BryanGaraventa: would this cause confusion?

scott: yes, there’s risk of that, and that’s why we explicitly call it out in this change

JaneFulton: perhaps the ARIA spec may need to explain the rationale for when/why ARIA features win out over native features or vice versa

scott: checkbox is another element where the native checked attribute wins out over ARIA; ARIA just lets us “say” that something else is true

<jamesn> Therefore, to prevent providing conflicting states and properties to assistive technologies, host languages MUST explicitly declare where the use of WAI-ARIA attributes on each host language element conflicts with native features for that element. When a host language declares a WAI-ARIA attribute to be in direct semantic conflict with a native feature for a given element, user agents MUST ignore the WAI-ARIA attribute and instead use the

<jamesn> host language feature with the same implicit semantic

jamesn: the ones that can’t be overridden, HTML should specify, otherwise the ARIA feature should win
… do we think that this is the correct behavior?
… are there scenarios where we might want a different placeholder for assistive tech?
… if this isn’t allowed are we shutting out a real use case?
… placeholders aren’t normally super critical to any functionality

JaneFulton: a phone number field is one example where the format is shown in a placeholder; maybe it’d be helpful to give AT an alternative placeholder

jamesn: and the format is probably useful to AT

scott: if we do want to change this behavior, we’ll need to change in all browsers
… because all browsers match what this PR says

spectranaut_: yes, this PR represents reality
… there’s no implementation
… so if we want to propose a change, that would be a much longer process

<melsumner> Why was this implemented if not defined in spec?

spectranaut_: for now, we should just merge this to make the spec accurate and clear
… and if someone wants to advocate for a change, then it can follow on

scott: agreed

<Zakim> jamesn, you wanted to state that I'm happy to document reality in that case

spectranaut_: for this PR, we’ve got 2 reviewers, can we get one more?

JaneFulton: I’ll review

WPT Open PRs

spectranaut_: scott, you opened wpt#52480

scott: this is old
… the original issue is from 2022 or 2021
… because there was talk in HTML about how people should not need to set src attribute if they’re using srcset instead
… in recent testing, most of the WPTs are passing
… I created a ton of variations
… so this would be great to get reviews

jamesn: looking at it now, I see duplicate srcset attributes — lines 48–58

Deep Dive planning

Cancelled June 5th Meeting

spectranaut_: just a quick PSA that the next 2 meetings are happening, then June 5 is canceled

Need a means of suppressing speech output during focus transitions within a SPA

<scott> i think i removed all the dupe srcset attributes. thanks james

spectranaut_: this issue was opened last week
… need a means of suppressing speech output during focus transition within an SPA
… can anyone summarize the issue or open questions?

melsumner: I’d expect a screen reader user to be comfortable aborting the output when they want (e.g., press Ctrl)
… we do solve for this in the JS framework I work in
… but I agree with Dominic that implementation of the navigation API is the current correct answer; not something with ARIA

scott: they brought up a very reasonable use case of Slack and other web apps that do have these transitions
… where it’s not desirable to have constant loading messages being announced
… I get that, but — like melsumner said — people can stop those announcements
… and I suspect many users would rather those announcements be present
… if a transition is too long (e.g., 5 seconds, even less), then it’s definitely sensible to still have a loading announcement
… and there are already solutions for this, like the navigation API, but I don’t think ARIA should be doing anything with this

spectranaut_: if anyone agrees with the OP’s request, please feel free to weigh in — otherwise, we’ll mark this as “won’t fix” for now
… okay, I’ll write a summary in this issue and close it out

Update: fix inconsistency around aria-setsize=-1

spectranaut_: this is Rahim‘s
… this has 3 approving reviews
… so we can talk about whether to merge it
… it’s also waiting for implementation
… do we know whether Webkit is the only one that needs to change?

scott: not sure
… there’s no way to make a WPT to test this

spectranaut_: would anyone be willing to confirm the behavior of browsers and determine which one(s) needs to be updated to match this proposed spec change
… it would need to be someone by who can check a11y API or use a screen reader to confirm

scott: should be able to determine with just screen reader

scott: the ideal behavior, talking to JAWS/NVDA, would be an announcement of “X of nothing” or “X of indeterminate”, when it currently says “X of 12” (when only 12 are shown on the screen, but more are asynchronously loaded)

Prioritizing work of the ARIA WG and how to make the most use of meetings

<melsumner> TY for calling that out, it's a good reminder.

spectranaut_: feedback is welcome
… if you think the ARIA WG could work more efficiently
… if you’re joining meetings but don’t contribute because of a blocker, I’d like to hear about that
… and get people more involved
… very broad topic
… if you have thoughts now, would love to hear them
… but I also hope to continue working on this over the next few weeks
… maybe we could do some kind of prioritization of open PRs

lola: this is more of a process thing, and each WG does things differently
… it’s unclear to me personally how things that *are not PRs* get added to the agenda
… do we submit an issue?
… do we contact spectranaut_ or jamesn privately?
… also, what constitutes a deep dive topic
… triage is important and helpful
… but what about ideas/issues that fall outside triage?

spectranaut_: the process for agenda’ing — any WG can add the “agenda” tag to an issue
… so that’s how things get on the agenda
… for new members, we sometimes hold orientation
… it’s been a while

spectranaut_: there’s a documentation folder in GitHub
… anyone who’s not had an orientation, please speak up

<spectranaut_> Adam_Page: one thing I've noticed for issue and Pr review, and for a little while we were looking at stale issues, sort the issues in reverse

<spectranaut_> Adam_Page: this is a sweet spot in the middle, things that have changed recently, a burst of activity

<spectranaut_> Adam_Page: I try to reach out to people, but maybe there is something structural in the process to surface thoses

spectranaut_: people can always add “agenda” to resurface things

scott: spitball idea
… as one way to stabilize people’s opportunity to talk, and to prevent new things from taking up too much agenda time...
… maybe we don’t necessarily agenda+ every issue that comes through triage
… maybe we leave it up to individuals here to take assignment of the issue, and then they can agenda it if they want

spectranaut_: awesome — noted, thank you

ChrisCuellar: +1 to scott
… in terms of facilitation, people have the most energy at the start of a meeting
… and issue triage can sap that energy
… it can be harder for newcomers to understand what’s happening when the first thing in the meeting is triage
… so I like the idea of starting with PRs
… the co-chairs could be more free to prioritize things

<Zakim> smockle, you wanted to ask about regrets (especially weeks ahead), choosing a scribe (zakim pick a victim?), pace of assigned work, focusing on unblocking big efforts

spectranaut_: thank you

smockle: a few topics going back
… I did do an orientation
… and now I have new questions
… for example, I wanted to send regrets while on vacation, but didn’t know how to do it, or whether it matters
… don’t know the right venue

<Francis_Storr> Jane, in Zoom, asked to be added to the queue.

smockle: we could also zakim pick a victim to encourage more diverse scribing

<melsumner> I feel like it takes about 15 seconds for someone to volunteer for scribe, it's just mostly silence so it feels longer. :shrug:

smockle: I also have a backlog of issue across repos and have a hard time juggling that and being willing to take on new work
… makes it tough for me to feel engaged in new work
… would also love to get a big picture sense of “what’s in flight”
… TPAC did help a lot with that
… but bring that throughout the year

<Zakim> spectranaut_, you wanted to talk about prioritization labels

<Zakim> jamesn, you wanted to suggest not triaging every week and to talk about regrets

jamesn: maybe we could try not triaging every week
… or just triage critical issues as they arise
… for regrets
… we appreciate them, but don’t track them
… so don’t feel bad about not sending them
… if you see the agenda and really want to participate but know you can’t attend, that’s especially when regrets are helpful

spectranaut_: and when you’re on vacation, we can often see that or deduce it

<spectranaut_> Adam_Page: +1 clay saying I don't know the right venue for questions

<spectranaut_> Adam_Page: I miss that we don't have a conical real time chat somewhere.. is it IRC or chat?

<spectranaut_> spectranaut_: formalize the slack?

<cyns> can someone drop a link to the slack?

JaneFulton: reiterate that it would be nice to not have to get to everything in one meeting
… spread it out a bit more
… sometimes discussions will get rushed
… helpful for newcomers to hear some conversations again
… having Slack and realtime chat is great, but need to accommodate people who can’t be available, or are in different time zones, etc.

<melsumner> If anyone can figure out how to do keep meaningful track of conversations about topics, especially those with a long history....that would be nice.

JaneFulton: putting guardrails into that so that essential information isn’t only captured there

scott: +1 to jamesn: maybe just 1 triage meeting a month
… Francis_Storr and I are part of the WCAG WG and we have a weekly check-in to determine issues that need to move to the next stage

Francis_Storr: the backlog that we have every Friday is super useful
… we use a big kanban board, and I think that could work well here

spectranaut_: please do share that with us

cyns: why don’t we use IRC?

spectranaut_: cultural
… some people are used to it

<Francis_Storr> AGWG backlog taskforce board: https://github.com/orgs/w3c/projects/56

Minutes manually created (not a transcript), formatted by scribe.perl version 244 (Thu Feb 27 01:23:09 2025 UTC).

Diagnostics

Succeeded: s/is only/isn’t only/

Maybe present: BryanGaraventa, cyns, jamesn, JaneFulton, lola, scott, smockle, spectranaut_

All speakers: BryanGaraventa, ChrisCuellar, cyns, Francis_Storr, jamesn, JaneFulton, lola, melsumner, scott, smockle, spectranaut_

Active on IRC: Adam_Page, ChrisCuellar, CurtBellew, cyns, filippo-zorzi, Francis_Storr, jamesn, katez, lola, melsumner, scott, Siri, smockle, spectranaut_