W3C

– DRAFT –
ARIA WG

27 February 2025

Attendees

Present
Adam_Page, CurtBellew, Daniel, filippo-zorzi, giacomo-petri, katez, keithamus, lola, Rahim, sarah, siri, smockle, StefanS, StefanS1, Summer
Regrets
-
Chair
ValerieYoung
Scribe
giacomo-petri

Meeting minutes

New Issue Triage

spectranaut_ w3c/aria#2458 from Aaron and w3c/aria#2456 from Scott added to agenda

w3c/dpub-aria#70 from Giacomo, raised a spec issue where acc name is required but no aria state/proeprty is allowed

jcraig think it should be addressed

spectranaut_ w3c/dpub-aria#71 from Giacomo

giacomo-petri this is just an editorial, that removes duplication

jcraig this can slide after publication, the 1st not

w3c/aria#2447 from ianthedev

keith: I'll respond to this

spectranaut_: w3c/aria#2445 from sideshowbarker

spectranaut_ already in editors agenda

New PR Triage

WPT Open PRs

spectranaut_ I'll take a look at some of them

jcraig merging web-platform-tests/wpt#50637 as long as all browsers are consistent

spectranaut_ anyone wants to review https://github.com/web-platform-tests/wpt/pull/50507?

spectranaut_ I'll take a look again

Deep Dive planning

spectranaut_: w3c/aria#2427 what about 20th 1hour before our regular meeting?

jcraig: it works for me

aaron: ok for me

(re)Consider name from encapsulation for specific roles

scott: related to the issue about required acc name for specific roles that do not need a required acc name

scott: there are roles that should be named by the context/encapsulation of the element

scott: such as name from heading, caption, etc.

spectranaut_ no objection on that in the group from a theoretical point of view

Should the ARIA WG standardize heuristic-based ignoring of ARIA and related features? And if so, how? agendabot]

jcraig: there are lot of scenarios where the engines align on specific heuristic cases, starting before ARIA existed, with tables, etc. to determine for example if they are real tables or not; webkit similar thing for list, but other implementors do not

jcraig: we had an example on a big site with an aria-modal="true" preventing users to access the rest of the page

jcraig: should we put somewhere (in aria specs or somewhere else) a documentation of how browsers do heuristics?

spectranaut_: to start, we should collect first all the heuristics

spectranaut_: so it's easier to discuss it concretely

<lola> Thank you!

<spectranaut_> ack +

scott: we already have plenty of issues of things that are implemented differently by browsers and are not normatively defined

scott: people would want to know this stuff

Matt_King: agreeing with Scott

jcraig: the introduction of these items in normative should be on case by case basis

Matt_King: one of the most important things is the information about what's happening, is often difficult to find. If there is a list of things documented it will help

jcraig: I would take spectranaut_ suggestion putting it somwhere in non-normative for now

<smockle> w3c/aria#2093

<smockle> https://github.com/w3c/aria/pull/2371/files

smockle: reminded me the aria-labeledby vs aria-labelledby topic

jcraig: we will have it documented somewhere in a non-normative and case by case we will decide to include it in normative

**sorry I was disconnected and I missed what Aaron said**

<spectranaut_> aaron: I'm concerned this is a very big project, maybe we should consider it case by case

jcraig: table seems a good place to start

<spectranaut_> aaron: also maybe these heuristic will be done by AI in the future

<np-at> Or maybe defining the fallback mechanisms for errors?

<np-at> E.G., If error condition A, attempt repair using one of the following in descending priority....

jcraig: someone can create some wpt test for understand how tables are handled by browsers

keith: it's in my list

jcraig: resolution: we decided it's a nice thing to do, assign to jcraig

Matt_King: suggestion: primary audience is authors, we can add a page to the APG to have more visibility. I will be happy to support.

<lola> +1 to that suggestion

smockle: suggestion: in addition to documenting how browsers repairs heuristically, we should also document this by logging messages to the authors (console) with some error code that can be referenced in docs

jcraig: lot of ideas on how browser can improve the authors experience but unforunately it seems a slow process

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

Diagnostics

Maybe present: aaron, jcraig, keith, Matt_King, scott, spectranaut_

All speakers: aaron, jcraig, keith, Matt_King, scott, smockle, spectranaut_

Active on IRC: aaronlev, Adam_Page, CurtBellew, Daniel, filippo-zorzi, giacomo-petri, katez, keithamus, lola, np-at, Rahim, sarah, siri, smockle, spectranaut_, StefanS1, Summer