[go: up one dir, main page]
More Web Proxy on the site http://driver.im/

W3C

– DRAFT –
ARIA WG

29 February 2024

Attendees

Present
Adam_Page, Daniel, Francis_Storr, giacomo-petri, jamesn, jcraig, keithamus, Matt_King, pkra, scotto, siri_, StefanS
Regrets
BryanGaraventa, RahimAbdi
Chair
spectranaut_
Scribe
pkra

Meeting minutes

present_

<spectranaut_> scribe?

New Issue Triage

spectranaut_: editorial issue. Daniel responded with respec issue

adam: just wanted to connect the dots

jcraig: I'll add a note regarding contrast modes

jnurthen: let's add it for editors' meetins

spectranaut_: aria 2132

jnurthen: can we ask Laurence?

scotto: I'll leave a comment.

spectranaut_: next aria 2129
… we looked at this last week.
… aria 2127. We have 2 PRs landed without author tests.
… once we have tests, we can make PRs on the validator repos.
… these are good first issues.
… can someone take them on?

Adam_Page: I can take one.

adrian: what are the technical requirements?

spectranaut_: Java.

jcraig: happy to help set things up

adrian: I can take the other.

New PR Triage

spectranaut_: html-aam 533 about hidden elements

scotto: this closes an older issue.
… from last week's discussion
… hidden labels in html cannot work because they're not in the accessibility tree.
… this PR clarifies this.
… not just label but table caption and legend fieldset
… do we want to adjust how the name calculation works?
… it might stop the algorithm too early.
… e.g., Firefox will pick up a title from the input. I feel that's right.

spectranaut_: Adam_Page made some tests?

Adam_Page: no, but I'll take a look again.
… I'll review

jcraig: I'll also review.

scotto: I didn't make WPT tests because I wasn't sure what we want to assert.

WPT Open PRs

spectranaut_: sarah wanted to do a deep dive on aria 1119. but let's wait until we can schedule with her

Deep Dive planning

Upcoming Agenda - CSUN Week

<scotto> no csun for me

<aardrian> I will be at CSUNATC

<Adam_Page> I’m going to CSUN

spectranaut_: please let us know if you're going to CSUN so we know if we should cancel.

<katez> not going

<giacomo-petri> no CSUN this year

<Summer> not attending

jnurthen: let's keep the meeting.

Consider providing a way for authors to customize the announcement of state

<spectranaut_> w3c/aria#2085 (comment)

giacomo: while working on the PR, I realized that there are two attributes doing similar things
… but right now the spec only deals with one of them
… both override the default semantics of the element

Matt_King: I didn't understand the question on rowindex etc
… can you explain what a single value should mean to do both

giacomo-petri: we need aria-valuetext to change the default value of switch or progress bar
… this seems similar to the grid situation

Matt_King: row and col index text need to be specified together
… we chose not to use valuetext because they can be on the same cell.

jnurthen: that's my recollection as well.
… regarding switch. is there a case where we need to know what the other value is?
… if we have one thing, we'd need to know both.

giacomo-petri: right, could be. progress bar is similar - percentage or small/medium/...
… this would be a good feature but perhaps it's not needed.

jnurthen: I think I agree
… I feel that even if we added it, AT wouldn't use it.
… I'd prefer to wait for AT to demand the other value.

giacomo-petri: the other problem with the switch role: if value text is specified, then we also require now value to be specified
… that doesn't make sense for a switch.
… hence my proposal

Matt_King: so aria-valuetext on a switch, you wouldn't add checked/unchecked?

giacomo-petri: that's the question. It seems it's always an active state.

Matt_King: that's the reason why it requires value-now.
… we had a lot of discussions in APG about this
… e.g., day of the week on a slider.
… valueNow could be 1/7, 2/7 etc

but that doesn't make sense and adds no value for the user.
… and it works fine anyway.
… we've discussed raising an issue here. But I don't want to expand the PR.
… it seems like you wouldn't need to require aria-checked.

<Zakim> scotto, you wanted to react to giacomo-petri

scotto: I like the idea of using aria-valuetext because I like the idea. But it's not even a state. It seems more like a new attribute.
… and it's not just for switch.

<Zakim> jamesn, you wanted to agree with Matt that I think we could remove that requriement and to ask if switch in APIs required a checked attribute

jnurthen: one of the complications is HTML's definition of value and how people are using it in the real world.
… checkbox's values is never what anybody talks about.
… I agree with Matt that we should remove the requirement for value-now
… that would also allow us potentially to use it here if we end up wanting to.
… I wonder if accessibility APIs need it.

jcraig: does the ARIA taxonomy require this as well?
… I think it does.

jnurthen: right. But we could also change that.

jcraig: feels like one of these features where we should tag it for privacy
… if it's switched around

jnurthen: feels like any form would fall for the same

spectranaut_: it seems we agree valuetext is not the right thing

<mario> I agree with th use of statetext

Matt_King: random guess: state-text. but still leaves the questions.

scotto: for roledescription we don't remove the role either?

Matt_King: true because it's critical.

scotto: right but I think you still need the initial state. otherwise which maps to which.

Matt_King: is that the same argument as with value-now?
… oh, when you're processing the "actual" value?

scotto: yeah. value-now is what a human understands but you still need to be able to fallback to the underlying actionable state of the component.
… otherwise it'll be just "value" without any actual state.

<siri_> we don't use pressed attribute if the label changes like play/pause

Matt_King: the JS will need to know the value, but will AT need to? It seems like a switch, e.g., swapping cm and inches, won't need to expose that.

mario: for switch checked/not-checked but it's not easy to understand why/how it maps to red/green.

jcraig: I still don't feel why we couldn't re-use value-text here.
… we know what to announce based on value-now but what we announce would be value-text.

<Zakim> jcraig, you wanted to ask why not have aria-valuetext require one of [ aria-valuenow | aria-checked | n… ]

jcraig: seems easier to expect value-text to require things as needed.

Matt_King: so you'd expect an event from value-text change?

jcraig: I could see implementations work either way
… but requiring checked makes sense, e.g., for styling.

giacomo-petri: sometimes you don't have green/red - e.g., cm vs inches is "always on".
… I see that it can be problematic from a development point of view.
… it's not true/false but a choice.

mario: but what is the change? it's always checked?

Matt_King: I think a cleaner implementation of aria-checked isn't possible?
… could you not style based on the value-text text?

spectranaut_: I think so but it wouldn't apply to all switches.

Matt_King: but then you'd have to document checked matching a specific style.

<jcraig> https://w3c.github.io/aria/#switch

jcraig: I had written something that was trying to match HTML phrasing
… for the cm vs inches. what would we do in HTML?
… that should affect what we should do in ARIA

<Zakim> jamesn, you wanted to ask how we envision this tying in with the native switch efforts going on

<jamesn> https://webkit.org/blog/15054/an-html-switch-control/

jamesn: the blog post mentions checked as attribute. but it seems like thumb and track will be standardized for similar purposes.
… I don't know how we will tie that together.

<jcraig> .sliding-track::before {

<jcraig> content: "ON" / "";

<jcraig> left: -40px;

<jcraig> }

<jcraig> .sliding-track::after {

<jcraig> content: "OFF" / "";

<jcraig> right: 0;

<jcraig> }

jcraig: [pasted the example]
… not sure what the empty string should mean.
… but we may not need this new feature if HTML provides it.

jnurthen: feels like accessibility wasn't entirely thought through yet.

scotto: that's why I raised it.
… anyone can change it and it no longer matches checked/no-checked.

<aardrian> Not merged yet: https://github.com/whatwg/html/pull/9546/files

jcraig: but it seems like the feature seems to already support it?

scotto: implementation goes off of role switch with checked/unchecked state.
… not from css content.

<jcraig> content: "label" / "alt";

scotto: that's just how it's visually rendered
… if you drop the alt, you'd sometimes get "on on"

jcraig: but if label/alt is supported, it feels like an implementation bug.

jnurthen: one is the before and the after on the track - how do you connect that in to checked/unchecked?

jcraig: right.
… thanks for clarifying.

spectranaut_: we're at the hour. do we have enough to move a bit forward?

mario: we have to battle through the HTML accessibility

scotto: they're working through the pseudo content

jnurthen: agreed and they should know that they shouldn't.

adrian: my tests were something like "accname on or off switch"

jcraig: accname included both before/after.
… we might have to modify the PRs and WPTs.

scotto: the same with Open UI. People kept doing on/off whichever way they wanted. But everything pulled it in since it's name from content.

spectranaut_: we're at the hour.
… there's more to discuss.

jcraig: thanks again to scott for raising this.

<aardrian> My announcement tests with TP 185: https://adrianroselli.com/2021/10/switch-role-support.html#Update02

scotto: [...]

Minutes manually created (not a transcript), formatted by scribe.perl version 221 (Fri Jul 21 14:01:30 2023 UTC).

Diagnostics

Succeeded: s/adam/Adam_Page/

Maybe present: adam, adrian, giacomo, jnurthen, mario, spectranaut_

All speakers: adam, Adam_Page, adrian, giacomo, giacomo-petri, jamesn, jcraig, jnurthen, mario, Matt_King, scotto, spectranaut_

Active on IRC: aardrian, Adam_Page, dmontalvo, Francis_Storr, giacomo-petri, jamesn, jcraig, katez, keithamus, mario, Matt_King, pkra, scotto, siri_, spectranaut_, StefanS, Summer