Meeting minutes
New Issue Triage
<github-bot> I can't comment on that because it doesn't look like a github issue to me.
jamesn: no new PRs so on to new issue triage.
… first one is editorial. dpub-aria.
… next one aria 2875. different outline algorithm.
… we should probably agenda this
… next aria 2872.
pkra: I filed this because I had piggybacked on the older issue which was discussed but not this one.
… I find it unfortunate that we have specified meter for the HTML parity but there hasn't been uptake in AT. narrator even dropping support in my testing compared to tetralogical.
mattking: that's what aria-at is all about. It allows us to identify and help with this kind of thing.
jamesn: this seems like it could fit in a larger conversation about missing features in AT from the specs.
mattking: aria-at would be a natural path for tha.
jamesn: let's try that.
… next up: align area with anchor from giacomo.
… there's a WPT PR
giacomo: the WPT PR is for anchor but not area.
… I opened it because links are more common but they are so close.
… I think we should align implementations.
… the WPT test demonstrates that there's no consistency, even in chrome, anchor and area differ.
jamesn: just for completeness: do we have a use case?
cyns: decades ago?
jamesn: personally seems like a problem that might not need a solution. If there's time in the group and implementations, then ok.
jcraig: it does seem minor.
giacomo: we found it in our tooling so it seemed worth it.
jcraig: a real use case is a good reason.
giacomo: I'll double check and follow up.
WPT Open PRs
<github-bot> I can't comment on that because it doesn't look like a github issue to me.
jcraig: only 2 new ones. First one is not important. Second we just discussed.
rahim: a few are waiting for review.
… I'll ping separately
jcraig: sounds good.
TPAC Plannning
<github-bot> I can't comment on that because it doesn't look like a github issue to me.
jamesn: if you have a deep dive, please make an issue.
… some are leftover.
… we could repeat them. Three are newer.
… the dev tools one could use more ideas. We might even run some coding sessions.
jcraig: sounds great.
… is there interest in sharing agent specific skills or tooling?
… since agentic coding is so new, people are at varying levels of competency. We've been encouraging people to use these tools. Maybe we can get more people involved.
… share how people use these tools.
jamesn: I'll create a placeholder issue.
jcraig: maybe separate sessions to come back to.
jamesn: relies on people having access to the tools.
… free version probably not enough.
… and maybe people won't be able to get their access up&running.
cyns: breakouts might work?
jcraig: I thought that's for other uses. But same time perhaps.
… projecting might be useful.
jamens: if you have other ideas, please add issues with the label.
… update on openui probably makes a good repeat.
… the other two might not be but take a look.
cyns: lola and I are working on a breakout.
<cyns> w3c/
jamesn: great. send us the time?
Security Horizontal Review
jamesn: jcraig could you review this?
jcraig: PR #2869 ?
jamesn: yes.
Daniel: security contact had suggestions. I can update the PR or open a new one.
jamesn: let's add it.
jcraig: what's the issue?
jamesn: using notify to get people to behave differently to identify AT users.
jcraig: it's "just" cloaking/masking. I can draft a note.
… I wouldn't call this a security issue. There are much bigger problems in the JS event model.
… remember when we pushed back on other review, saying this should then also be added to DOM, HTML etc.
… we filed this as a TAG issue that's just been sitting around.
jamesn: aren't we covering this then by the text for this?
jcraig: maybe not passive but active fingerprinting?
daniel: re jamesn I think so.
jamesn: can we push it back?
jcraig: maybe push back with PRs to HTML and DOM?
daniel: that would block us though.
… if firing up arianotify, would it map it to AAPIs without AT active?
jcraig: it would require root access to know this so then there are bigger problems. It's not about ARIA.
… same for event model.
daniel: maybe invite Simone to a session.
jcraig: or breakout.
daniel: sure.
jamesn: let's set something up.
Allow aria-valuetext for use with combobox role
jamesn: this PR has been reviewed by three people.
… last time, mattking wanted to create a PR.
mattking: right. I think the changes here are too light. There's info missing how this would work.
… I think I can get to those answers without taking up group time.
… I suggest not to merge, depending how urgent this is. I can add my concerns to the PR if that helps?
jamesn: yes please.
Expectations for aria-hidden and focused elements
jamesn: looking at the minutes from the last discussion, there were comments from jamie that we needed to address.
rahim: Aaron mentioned that there is a chromium implementation. Has any other browser done this? I had created a draft for webkit.
pkra: jamie's last comment is from september. Was this discussed more at TPAC?
jamesn: not sure.
… it's interesting to look through the linked issues. Seems like AI review tools are creating these.
https://
pkra: I found a bugzilla bug from last year.
jamesn: I'll add it.
giacomo: I think we should ask for real user feedback. People use it intentionally but forget tabindex -1. Exposing them might be confusing. Unless tabbing, these might be skipped by screenreaders which could make it even more confusing to expose them.
jamesn: I thought this was only for exposing content when focused. Not just focusable content with aria-hidden.
<Zakim> jamesn, you wanted to react to Rahim
rahim: I think that's correct.
… you have to do a tree walk. If it's inside a container, the entire branch should be un-hidden.
… one of the WPTs that need review is about aria-hidden. We might be able to piggy back on that.
<HaTheo> I would be happy to write a WPT.
mattking: this is still an author error recovery mechanism though. I don't think author intent can always be clear when you're recovering.
… so we're looking for a least bad option.
jamesn: right. From Valerie's comment, there's still work to do in Chrome, too.
<Rahim> WPT for testing aria-hidden using label that could be used to test this potentially: web-platform-tests/
mattking: consistency is super important here.
jamesn: ok it seems the spec is ok but this PR is almost too long and old. Does anyone want to try to move it forward?
rahim: I can follow up with scott.
jamesn: thanks.