Meeting minutes
<Greta> +present
scribe pkra
[New Issue Triage](https://bit.ly/34gp5hL)
jnurthen: 1489 from wilco. Seems to have a point. Is it 1.3?
… can we punt to 1.4?
siri: is it false by default?
brian: originally, was intended to expose it to AT but not visually.
… and false would expose it.
jnurthen: but this is about role=presentation.
… button with role=presentation would become a button again with aria-hidden false.
… doesn't seem intentional from user.
… is it 1.4?
mattking: 1.4.
brian: validators should flag this
jnurthen: 1.4 it is.
… issue 1488
… I tend disagree with the proposal.
… any takers?
melanie: I'll take a look.
jnurthen: issue 1486.
… will be in 1.3
… others are on the agenda
[New PR Triage](https://bit.ly/3cboD95)
jnurthen: issue 1491 as discussed previously. Needs more reviewers.
mattking: did we document the decision that aria-labelledby provides equivalent to HTML labels, so we do not need additional role=label?
jnurthen: yes. unless implementors ask us.
mattking: right. this is a coding convenience
jnurthen: right. In a space with plenty of complexity.
… but noted only in minutes. This was never in the stable branch.
mattking: it would be good to make this easier to find for future reference..
… worth it as we (in particular Jon) put a lot of effort into it.
jnurthen: right. the main point is that this doesn't seem anything right now.
… accname PRs.
melanie: I added implementors and Brian as reviewers.
brian: looked through them and will approve them.
jnurthen: PR 1484. I had some comments but forgot to post them.
… as the group, we need to work out how we want to deprecate things.
… in the past, I think we've just removed things. Is that about right?
mattking: going straight for remove seems bad. Deprecate first.
… deprecating globalness was different (and difficult).
jnurthen: this case seems different.
joanie: is it prohibited?
jnurthen: no. just proposing to remove support.
… prohibiting was just for global
joanie: do we expect user agent's to stop supporting it?
jnurthen: yes.
joanie: then I'd remove it.
jnurthen: @william this should make the PR actually simpler. Ping me and we can coordinate.
william: will do.
[Meaty topic for next week](https://bit.ly/3oOnOYI)
jnurthen: would like to propose accname.
… melanie, brian, would next week work?
brian: I'm out.
jnurthen: ok. any other topics?
<jamesn> Secondary actions on items in composite widget roles #1440
sarah: Secondary actions on items in composite widget roles #1440 ?
jnurthen: great. Will do.
sarah: I had a partial PR for this.
… I'll remake it and add it to the issue
[Exit Criteria Testing: Need test cases and results for Authors MUST/MUST NOT statements in ARIA 1.2](https://github.com/w3c/aria/issues/1492)
joanie: aria 1.2 is in CR and we want it to move forward.
… for normative changes, we need verification and documentation of two implementations.
… historically, user agent implementations were the focus.
… we probably should also focus on validators (and should have in the past)
… we also have author-only changes.
<joanie> https://
joanie: we had a comment "the list of test results and implementations is small. Is combobox 1.2 really working out?"
<joanie> https://
joanie: I've created issue 1492.
… pointing to the wiki page.
jnurthen: "both" means 1.1 and 1.2
… some are very similar.
joanie: e.g., aria-owns / aria-controls change in combobox change.
… figure out what do we need to do.
mattking: they might be different from 1.1 MUST
jnurthen: looks like my list is incomplete.
joanie: so we need people to write tests, put them through validators, record results.
… the critical stuff (e.g., combobox) needs to be in the inventory.
… makes sense?
jnurthen: the aria-controls was in 1.1 so maybe my list is ok.
mattking: but is the wording the same?
jnurthen: right, it has.
mattking: this sounds like a real project.
joanie: if you're a new member and have lots of experience with validators, this is for you. Write some HTML, push through validators, record on wiki. Done.
mattking: the tricky part is looking at the author must and creating a failing test case.
joanie: correct. I can help with generating something from the UA tests.
jnurthen: sounds like validator people could help...
harris: I'm happy to help.
jory: I'm happy to help.
joanie: I'll make volunteers assignees, organize yourselves, ping me with questions
greta: I'm in.
[Problematic User Agents MUST NOT statement needs to be qualified](https://github.com/w3c/aria/issues/1476)
<joanie> https://
jnurthen: related to issue 1487
… e.g., aria-label on div. If browsers are forbidden from doing error correction, that seems like a bad ida.
… we need to work out what to do in 1.3
… maybe "may expose them"
joanie: I've been working on this for 6 weeks, huge commits, changing hundreds of (mostly test) files.
… finally got it far enough but now here
mattking: what are the "counters"?
joanie: google looking for content and potential AT that takes advantage of this.
… in retrospect, this was probably a mistake.
… I'm working on a proposal for Google where we log something in the console.
… if authors don't use validators, they'll get this at least.
… if google likes it, this might be a good idea.
mattking: to clarify: if author puts label on generic, there would be output on the console?
joanie: only if something is actively asking for name calculation.
… if AT asks for name, that triggers a calculation in Chromium.
… if authors don't actively test for this, they won't see the warning.
mattking: do validators test with AT?
joanie: yes.
cynthia: but won't help developers who don't care about accessiblity?
joanie: yes. but would google want to spam all authors?
cynthia: probably no.
… but is the group large enough to make this useful?
joanie: right.
mattking: if we all the author must not parts are good for the ARIA-AT project, then we can help.
joanie: we're not looking for aria-labels as ATs. User hits tab, screenreader says "foo". It doesn't know where foo comes from, how came to be.
mattking: but if it didn't get a role or a generic role, we could spit it out?
joanie: it might be a hint.
mattking: currently, we're looking at VO with Chrome, if you put an aria-label on it, it creates an empty group.
… if it's a title, it's treated as an aria-label. So you end up with a lot of empty groups that shouldn't be there at all.
jnurthen: sounds like a Chromium issue.
mattking: if naming is not prohibited on generic. should we change the definition of prohibited to be only probibited for authors.
jnurthen: that's what we had changed.
<msumner> I would prefer "should not" to removing it completely
jnurthen: would encourage them not to expose it.
mattking: I'm in favour of changing it.
joanie: I'm ok with any consensus.
cynthia: I suspect the concerns will persist.
mattking: I spoke with James about this. Seemed like it could be solved with VoiceOver 2.
joanie: orca has logic that says what should have a name and what shouldn't. Text objects shouldn't have them so they're there but ignored.
jnurthen: do we have a path forward?
… "should not" seems one path.
jnurthen: let's jump to accname 57
[When is hidden content taken into calculation of name and description?](https://github.com/w3c/accname/issues/57)
<jamesn> https://
jnurthen: this was a bit shocking but I think joanie is right.
joanie: I'd suggest to read it from last comment up.