W3C

– DRAFT –
ARIA WG

18 July 2024

Attendees

Present
Adam_Page, CurtBellew, Daniel, Francis_Storr, giacomo-petri, HaTheo, jamesn, jcraig, katez, keithamus, Matt_King, Rahim, ray-schwartz, sarah, Siri, smockle, sohara
Regrets
pkra
Chair
-
Scribe
Rahim

Meeting minutes

New Issue Triage

Rahim: aria #2286 is for the next editor's meeting

<keithamus> `// prettier-ignore`

daniel-montalvo: I'm assigned to work on #2285

jcraig: OK to forward dupe for aria #2281

jamesn: what are the implications for this?

jcraig: aria-invalid has different missing/invalid value defaults; putting this into the attribute's characteristics table could clarify this

jamesn: we might have different missing value defaults depending on the role; some reticence in engines to do this because if we put this in, then IDL needs to return these (which requires validation by web engines)

jcraig: not a problem for enumerated attributes; the way that IDL works is that they reflect as nullableDOMStrings
… one thing that could cause a problem is the boolean-like value types (e.g., true/false/undefined ones). Another one is aria-invalid which doesn't have a well-defined set of values

jamesn: main concern is around implementation concerns but otherwise, no problem with moving forward with it

keithamus: there's a related issue with ARIAMixin; I'm happy to pick this issue up (aria #2281). I think it'll be a lot of work. Will work with Rahim on this

Rahim: for aria #2279, Rahim/James to work with Anne

jamesn: for aria #2277, looks like an editorial one related to inheritance (e.g., something being inherited and required)

New PR Triage

jcraig: aria PR #2284 is the followup from the orphaned role deepdive and from Scott and I's discussion
… not worried about the technical side, mainly prose. Not introducing a new requirement, just removing a UA requirement that implementors thought was too restrictive

Rahim: aria PR #2172, need other reviewers

keithamus: you can add me as reviewer as well

WPT Open PRs

jamesn: nothing new it appears, anything you need James C.?

jcraig: nothing is new

Deep Dive planning

jamesn: any proposed deepdives for next few weeks? unless it's time-critical for a deepdive, should probably postpone deepdives for TPAC

TPAC Agenda Discussion

<spectranaut_> discussion board for topics: https://github.com/w3c/aria/discussions/2283

<spectranaut_> wiki for draft schedules: https://github.com/w3c/aria/wiki/TPAC-2024-ARIA-Meetings

jamesn: we're thinking about how to start planning the agenda; the agenda is pushed by the group (not the chairs). It's what the group wants to talk about; please come up with ideas for things and ideally, leading certain conversations with facilitation by chairs
… we created a discussion board (aria #2283); if you have an idea that isn't fully fleshed out, you can note that in the discussion board and see if others want to collaborate and come up with larger topic. The other way to do that is to add a tag to an issue, or have an issue that would form a 20-30 min agenda item, add the F2FCandidate label which flags to us (chair) as something people want to talk about at TPAC
… please include comment around why you want to talk about it, and what the target of the discussion is
… need potential/targeted outcomes rather than freeform discussion without direction

jamesn: there's an agenda with tentative schedule for TPAC in the wiki

Clay: should I remove old F2FCandidate label on issues (e.g., aria #1957)?

jamesn: it doesn't have that label but will remove the "F2F" label now
… you can re-add the new label with a comment clarifying why you want to talk about it
… 11 people registered so far from our group (great)

spectranaut_: will be a well-attended TPAC with reps from browsers (and hopefully have AT representation there as well per James N.)

keithamus: is it worth mentioning that there are concessions (e.g., lower fees) for reps?

jamesn: yes, trying all possible ways to ensure good attendance

Update HTML-AAM to use title for description instead of name, when the name is prohibited

aaronlev: basically, I was looking at what we do when an accname is calculated for something that is name prohibited; I tried to not allow authors (within our own internal systems) to have name on name prohibited element
… can put title (attribute) on anything; e.g., if you have a title on an element with name prohibited, we should move it over to accDescription

jcraig: great idea as we shouldn't drop author-provided string

aaronlev: if a white-spaced trimmed version of title vs innerText, then we should drop it. Lots of time people use CSS property (e.g., text-overflow: ellipsis), using it to see "..." additional content

jamesn: what needs changing?

aaronlev: html-aam doesn't say that you can't do this; spec as written says title can be used for name or description (and didn't see anything in accname spec that says you can't do it)

Scott: not explicitly written in html-aam but good idea. There's a couple ways I'm thinking of presenting this information and I've been thinking about ways of doing this, e.g., name prohibited being a single algorithm, rather than element-specific. Also want to make a companion PR for HTML elements that can't be named (e.g., <head>)

Clay: spoke about #506 and a related deepdive, did that happen

Scott: no; although listed as potential deepdive topics

aaronlev: want to have screen reader users test this as well (where name gets moved to description generally if name is prohibited)

aaronlev: curious about the case where the title is the same as the child content; or if the description is the same as the child content (important in the text-overflow: ellipsis case where you're visually showing a tooltip)

Scott: I do think that ignoring the title (as child content) could be something that goes in accname; these are 2 separate PRs

jamesn: lots of definitions of whitespace, whitespace is defined differently in HTML vs. CSS (James C. mentions that ARIA links to HTML definition for whitespace

zakim ,next item

User agents must not support aria-owns across shadow root boundaries

aaronlev: there was some skepticism about this; aria-owns is very challenging generally. It's a useful attribute but lots of things can happen from implementation perspective (e.g., inserting referenced node earlier in the document, aria-hidden nodes)
… if there's no actual use case for using aria-owns across shadow roots (e.g., pointing into a shadow root using aria-owns), then would say not to do for now, and implement it for other use cases (e.g., activedescendant)
… not saying we would never do it; if someone comes up with a use case (but by then we'll have debugged it with other attributes with stronger foundation for how to do it)

jamesn: how do we make it clear to developers that this shouldn't be expected to work?

aaronlev: we could put a console error; we're starting to this more for illegal markup (e.g., aria-hidden on <html> or <body>)

jamesn: also not sure how to spec this. Does anyone disagree think we should 100% do this if we're doing the other attributes?

keithamus: I don't disagree; given that it's only going to be for setting, should it be a console only report error (exception)?
… I think it's possible to raise an exception from a setter method (pretty sure you can do this in IDL). Could we throw an error instead of a console warn

Should aria-valuenow be required if aria-valuetext is used?

jamesn: don't think we came to a conclusion on this

Scott: I want to draw everyone's attention to the comment I added in the issue; per the previous conversation on this, want to reiterate that the proposal is not that there isn't value for aria-valuenow(or used together), there are use cases where they don't want to use valuenow. They may not want to have valuenow value announced as well
… thinking of instances where you have YouTube video player (e.g., hearing "15%" along with valuenow value of timestamp)
… also worked on a range slider component and it didn't make sense to announce the numerical value. E.g., use cases where requiring the attributes doesn't make sense if you're trying to override it

<giacomo-petri> +1

jamesn: I was prepared to advocate for both; the range slider component you mentioned is a compelling example of making it not required (valuenow) if you have aria-valuetext

<spectranaut_> +1

sarah: is that for slider and for progress bar?

aaronlev: for progress bar, there is a percentage (such as NVDA's sonification of loading actions)

jamesn: looks like there's consensus to change it for slider but not progress bar

Scott: I will make a PR to clarify that it's not required for that specific role (slider), but people can optionally do it if they want

jamesn: we should add a note when someone may want to do it as it's not obvious

Clay: to clarify, it is going to be provided (valuetext) unless aria-valuetext is provided?

jamesn: essentially one or the other is required

Adam_Page: I can take a look at that PR

Scott: yes please; add me as reviewer and I'll add what I need to

<Rahim> s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers/

<Rahim> s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers//

<Rahim> s/<Rahim> s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers//

<Rahim> s/<Rahim> s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers//

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

Diagnostics

Succeeded: s/IDL works is that reflect as nullable /IDL works is that they reflect as nullable/

Succeeded: s/aria #2286 is for the next editor's meeting/Rahim: aria #2286 is for the next editor's meeting/

Succeeded: s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers

Succeeded: s/unless it's time critical for a deepdive/unless it's time-critical for a deepdive/

Failed: s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers/

Succeeded: s/can put title (attribute) anything/can put title (attribute) on anything/

Succeeded: s/Lots of time people use CSS property (e.g., text-overflow: ellipses)/Lots of time people use CSS property (e.g., text-overflow: ellipsis)/

Succeeded: s/html-aam says you can't do this/html-aam doesn't say that you can't do this/

Succeeded: s/important in the ellipses case/important in the text-overflow: ellipsis case/

Succeeded: s/lots of definitions of whitespace/lots of definitions of whitespace, whitespace is defined differently in HTML vs. CSS (James C. mentions that ARIA links to HTML definition for whitespace/

Succeeded: s/a useful attribute but lots of things that can happen/a useful attribute but lots of things can happen/

Succeeded: s/also in the spec, not sure how we spec this/also not sure how to spec this/

Succeeded: s/want to reiterate that the proposal is that there isn't value for aria-valuenow /want to reiterate that the proposal is not that there isn't value for aria-valuenow/

Succeeded: s/I will make a PR to clarify that it's not required for that specific role/I will make a PR to clarify that it's not required for that specific role (slider)/

Succeeded: s/to clarify, it is going to be provided unless aria-valuetext is provided/to clarify, it is going to be provided (valuetext) unless aria-valuetext is provided/

Warning: ‘s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers//’ interpreted as replacing ‘aria PR #2172, need another reviewers’ by ‘aria PR #2172, need other reviewers/’

Failed: s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers//

Warning: ‘s/<Rahim> s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers//’ interpreted as replacing ‘<Rahim> s’ by ‘aria PR #2172, need another reviewers/aria PR #2172, need other reviewers/’

Failed: s/<Rahim> s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers//

Warning: ‘s/<Rahim> s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers//’ interpreted as replacing ‘<Rahim> s’ by ‘aria PR #2172, need another reviewers/aria PR #2172, need other reviewers/’

Failed: s/<Rahim> s/aria PR #2172, need another reviewers/aria PR #2172, need other reviewers//

Maybe present: aaronlev, Clay, daniel-montalvo, Scott, spectranaut_

All speakers: aaronlev, Adam_Page, Clay, daniel-montalvo, jamesn, jcraig, keithamus, Rahim, sarah, Scott, spectranaut_

Active on IRC: Adam_Page, CurtBellew, dmontalvo, Francis_Storr, giacomo-petri, HaTheo, jamesn, jcraig, katez, keithamus, Matt_King, Rahim, ray-schwartz, sarah, Siri, smockle, sohara, spectranaut_