W3C

– DRAFT –
ARIA Authoring Practices Task Force

25 February 2025

Attendees

Present
Adam_Page, howard-e, jugglinmike, Matt_King
Regrets
-
Chair
-
Scribe
jugglinmike

Meeting minutes

Setup and Review Agenda

https://github.com/w3c/aria-practices/wiki/February-25%2C-2025-Agenda

Matt_King: Next meeting: March 4

Matt_King: Any requests for change to agenda?

Matt_King: Hearing none, we'll stick with the agenda as published

Publication planning

mk: We need more time for the next publication, especially considering CSUN (which I will be attending). So I've moved this milestone to March 25

mk: Any concerns about moving that out?

howard-e: It works for me

mk: We'll leave it at that, then--for now, at least

PR 3194 - Missing symbol in grid and table properties practice

github: w3c/aria-practices#3194

mk: Adam_Page provided feedback, and we decided to wait to see if the pull request author would respond

mk: It seems we haven't received a response, yet

mk: It would be nice to merge this, though

mk: If it's possible to push a commit to this branch with the change that addresses Adam_Page's feedback, that would be ideal

Adam_Page: I can give it a shot

Issue 3232 - rowgroup for table body

github: w3c/aria-practices#3232

mk: We discussed this before. Both Kurt and I did some testing, and most recently, the reporter reported that they can't reproduce the issue

mk: The "role=none" makes these elements extraneous.

mk: Because of that, they kind of work against the APG's goal to keep examples as simple and as pedagogical as possible

mk: I don't think we want to suggest to anyone that there is any value in having those "rowgroup" divs

mk: So, first question is: is everyone good with calling this a bug (and marking it as a "P3" priority)?

Adam_Page: Sounds good to me

mk: I should probably label this as a good first issue

Issue 3237 - Potential bug: incorrect menu structure in navigation menubar example

github: w3c/aria-practices#3237

mk: The reporter is saying that we have menus with descendants that are menus. That would be incorrect because a menu should be either standalone or inside a menu item

mk: I'd like someone to look at whether or not this is only happening in this one example

mk: I guess this menu has a different structure than our other examples

Adam_Page: I'm comparing "tuition" and "admissions" (because the reporter specifically called those out) with comparable items in the other items (eg. "facts" and "about"), and I'm not seeing any difference

mk: I'm wondering if this is also a problem with the editor menu bar

Adam_Page: We discussed this in last week's ARIA Working Group call. We might want to sit tight while Giacomo follows up on the related spec changes

Adam_Page: In the APG issue, he writes pretty emphatically that it violates the spec. But in last week's ARIA WG call, he posed it as a question, saying that the spec seems unclear on this topic

Adam_Page: If a menu cannot nest another menu, how can we do nested menus at all? A menu item also does not permit a nested menu...

mk: It doesn't?

Adam_Page: I'm scrolling through the ARIA spec of the different roles... Ah, I may be misreading

mk: In our example, the elements with role "menu" are actually linked

Adam_Page: The menus are "ul" elements and setting the "role" attribute to "menu"

mk: So when we have a sub-menu, it is a "ul" contained within the "li" that contains the parent menu item

mk: ...but the parent menu item isn't the "li"; it's the anchor.

mk: To change this, you would actually have to make the "li" the parent menu item, right?

Adam_Page: I'm not sure

mk: If something is a parent menu item, and if it had to contain the menu--if that's the way it's supposed to work...

mk: I suppose it could own it. We could use "aria-owns", right?

Adam_Page: I suppose so. The ARIA spec is confusing about this right now because I don't see any affordance for nesting menus

mk: I guess we wouldn't be able to use links for the parent menu item. Instead of having "li" with "role=none", that would probably be the parent item. It would take a menu, which would be on the "ul" element

Adam_Page: I'm not sure how I would do this, anymore. Though I don't have much experience working with the "menu" and "menuitem" roles

Adam_Page: I'm inclined to let Giacomo clarify the spec since that appears to be an outstanding issue

mk: I actually remember Brian talking about this, but you don't really need the menus to be descendants of... I wonder if there really is alignment on this

Adam_Page: In th spec, there's an explicit section for "allowed child roles" and "menu" is definitely absent

Adam_Page: But I don't see anywhere else in the spec that allows for nested menus, so I don't know how it should be achieved

mk: There's a practical challenge here in that there are probably a lot of menus written in the way that this one is written

mk: so clearly, it at least can work

Adam_Page: This relates to w3c/aria#2438

mk: And that doesn't have a pull request, yet

Adam_Page: I think he just volunteered to take it last week

mk: I'm fine with us just watching for now

mk: I guess when he submits that pull request, he could hopefully reference this issue. That would cause it to float back up to the top of our list

mk: I'm hesitant to call this a bug because I'm curious to see what kind of feedback Giacomo's pull request will receive

mk: I thought that might be straightforward, but it wasn't!

Issue 3244 - Change naming guidance for radio group, grid, table, dialog

github: w3c/aria-practices#3244

mk: I guess this issue could end up having multiple sub-issues--one for each pattern that needs to change

mk: I haven't reviewed all of these to know how many changes might be driven by this

Adam_Page: They asked me to raise this in APG during last week's ARIA call

Adam_Page: I haven't looked too deeply, yet--just at three pattern pages. All three imply that a name is needed, but it's kind of an editorial question

Adam_Page: I don't know if it's on us to encourage any of these patterns to have these names because everyone admits that those are kind of narrow use cases

mk: Yes, we do avoid the normative language entirely and intentionally

mk: the way we handle this editorially is there is one thing we've done in the naming practice--the accessible names and descriptions practice page

mk: We have a table of guidance for every role, and we do distinguish "required" versus "recommended"

Adam_Page: That's very severe language that may not be true for much longer

mk: It's okay to get ahead of that pull request and land a change before it is merged

mk: In the patterns, we do use the word "optional" sometimes.

mk: Mike Gower [sp?] has a pull request related to "tab". I was reviewing that and I decided I need to review other examples to find appropriate wording for consistency

mk: I'm trying to figure out which practice we have previously established, here. I think the right thing to do would be to describe that in this issue

mk: Sometimes, we add notes where we try to specify what's important about the naming

mk: I guess, for this, one person could take it on. I'm not going to put my name on this issue, yet, but I will help spec it out

mk: Hopefully we'll be able to find someone else to work on it

mk: I'm labeling as "required for ARIA spec"

mk: We don't know the timeline for merging that, do we?

Adam_Page: Oh! It's already merged!

Adam_Page: It was merged just one hour ago

mk: Ah, I guess we have to prioritize this, then

Adam_Page: Scott also filed a follow-up issue with IBM's equal-access checker to encourage them to remove the associated test

mk: I'm going to label this "P1"

mk: Because this merged, it goes into the editor's draft. Actually--did we change ARIA? Is it evergreen? Will that change go to TR?

Adam_Page: I don't know. I wish Daniel were here to answer that question

mk: We don't want APG to be out of sync. I kind of wish it wasn't merged until we had aligned on a timeline

mk: ARIA changes this, but it would be really nice if someone from the ARIA Working Group would do the merge, here

mk: Well, for now, I've marked this as "P1"

Adam_Page: We could place a disclaimer at the top of affected pages to give warning about a breaking change in the ARIA spec

mk: I think that would be a whole new can of worms

Adam_Page: Yeah...

Issue 3175 - aria-selected requirements in listbox

github: w3c/aria-practices#3175

mk: so, here's the thing: jaws-test provided an excellent answer to this issue

mk: If I'm understanding this right, this was to specify aria-selected=false in all the items that are not selected

mk: In our example, we don't set an item to "selected" until there is a selected value. This whole thing is about if you have a listbox where nothing is marked as selected--neither true nor false

mk: Is it the case in our example that, once you select something and unselect something, then the selected status isn't persistent?

Adam_Page: Our example does that; I just verified

Adam_Page: Selection persists after you blur away

mk: I think the referenced best practice is not applicable because we do mark something as selected once it becomes selected. And if there's nothing selected when the page loads, then that's an okay default state

mk: This particular listbox doesn't give you an option to unselect. Once you've selected a value, there's no way to deselect entirely--you can only change to a different item

mk: It's sort of like how you can't uncheck a radio button

mk: So I don't necessarily see that as a problem

Adam_Page: I agree

Adam_Page: The issue that jaws-test linked to about a best practice is still an open issue

mk: There's a comment about a second thing. I guess that could be raised as a separate issue

Adam_Page: It's that the "class" attribute does show up on elements with no value. This is kind of a code hygiene issue. Certainly not related to the issue they've raised here

Minutes manually created (not a transcript), formatted by scribe.perl version 242 (Fri Dec 20 18:32:17 2024 UTC).

Diagnostics

Succeeded: s/recently/most recently/

Maybe present: mk

All speakers: Adam_Page, howard-e, Matt_King, mk

Active on IRC: jugglinmike