W3C

– DRAFT –
(MEETING TITLE)

22 October 2025

Attendees

Present
ChrisCuellar, elizabeth, IsaDC, james, Joe_Humbert, jugglinmike, Matt_King, mfairchild, mmoss
Regrets
-
Chair
-
Scribe
jugglinmike

Meeting minutes

Review agenda and next meeting dates

Matt_King: Next AT Driver Subgroup meeting: Monday November 17, but this is expected to change.

Matt_King: Next CG meeting: Thursday October 30

Matt_King: Requests for changes to agenda?

IsaDC: Perhaps checkbox?

Matt_King: I didn't add checkbox because switch is already present

Matt_King: howard-e and carmen have previously requested topics, but we'll wait on those since they can't attend today's meeting

Matt_King: Hearing no further requests, we'll stick with the agenda as planned

Current status

Matt_King: There are six plans in "draft review" (though due to a clerical error on my part, it appears that there are only five)

Matt_King: The "Switch with HTML checkbox" and "two-state checkbox" plans are ready for people to work on

Matt_King: The spin-button example is coming up

Running test plan for Tabs with Automatic Activation

Matt_King: We'll skip this topic because Hadi isn't present today, so it's down to IsaDC or me to resolve the conflicts. It should probably be me

Running test plan for Tabs with Manual Activation

IsaDC: There is a new version. I fixed some VoiceOver commands

Matt_King: I took care of that

Matt_King: On the VoiceOver test plan, you merged a fix to test #2

Matt_King: So we have the run by Joe_Humbert--it's complete now, except for test #2.

Joe_Humbert: I'll get test #2 by next week

Matt_King: It is currently ready for Dean to tackle, though Dean isn't present today

Matt_King: That's it for VoiceOver

Matt_King: We need to let Dean know that it's ready for him

Matt_King: Let's look at the 5 conflicting NVDA results. These are conflicts between Joe_Humbert and Dean

Matt_King: Starting with test #4: this is one where you both recorded the same output. However, you had different assertion verdicts

Matt_King: So, "shift+tab". The mode switching one, dean was saying that it was working for him

Joe_Humbert: That could just be because I missed it. I'll have to double-check

Joe_Humbert: I'm guessing that I marked it wrong. I know that "shift+tab" generally changes the mode

Joe_Humbert: I'm guessing that I marked it "yes" on the previous test that was similar

Joe_Humbert: I'll change my answer after I double-check it

Joe_Humbert: It's weird that my browser version is listed as 99.0.4.8. I guess I'm just verifying the bot results. That's a very old version!

IsaDC: Is that what the bot is running?

Joe_Humbert: I don't know

Matt_King: Test #13 is another one where you both have the same output. Here, you reported a difference in negative side effects

Matt_King: "The tab panel boundary was ignored." And "the focus moved to the content of the tab panel"

Matt_King: I'm curious about why you recorded that as a side-effects. Don't we have an explicit assertion for that? If the assertion fails, then we shouldn't need to track this with a negative side-effect

Joe_Humbert: It has been a very long time; I can't recall why. Generally, I report failures in this way when I want to make sure we discuss it here on this call, but I can't say for sure

Matt_King: I think dean's should be changed, here.

Joe_Humbert: Based on this, I'm kind of confused why it was not announcing the tab panel (or property page, in this case). I think it did for the tabs key, so I was confused why it doesn't do it for the arrow key

Joe_Humbert: That might be why I put a note in for the previous test--just to make sure I wasn't missing anything

Joe_Humbert: The next two conflicts are the same

Matt_King: There's a part of me that is wondering... "Property page"--is that really...? I guess if you're an NVDA user...

mmoss: This reminds me of VoiceOver reading grids as labels

james: I think there's a fine line. We explicitly are not in the game of dictating what screen readers say. I would say that semantically, there is no such role within NVDA as "tab panel". If you search the source code for "tab panel", you won't find anything. They are treated as "property pages"

james: perhaps NVAccess assumed that if you are switching to NVDA to JAWS, you would be better served by familiar vocabulary (although JAWS has since changed their terminology)

Matt_King: Instead of bringing NVDA up to date, we'll bring dean back in time and get aligned with NVDA

Matt_King: So that's three of these conflicts

Joe_Humbert: The next two (for tab in both browse mode and focus mode), it's the same issue

Matt_King: Okay, so for the very first conflicts, you'll validate.

Joe_Humbert: For the second, I'll remove the side-effect

Matt_King: And for the next two, we'll have to talk with dean

Joe_Humbert: That just leaves test #15

Matt_King: So this is "activate a tab in a tab list" and it's related to the space

Joe_Humbert: The output we reported does not match

Joe_Humbert: If this is not an issue, I'm happy to go back and remove a negative side-effect

Matt_King: For JAWS, we put a moderate side effect

Joe_Humbert: I reported it as severe, but I can change that

Matt_King: Yes, this is definitely moderate. It's just redundant speech

Joe_Humbert: Okay. I'll also re-test just to verify that it's still an issue

[Dean joins the call]

Matt_King: In test #13, for "down arrow" and both of the "tab" commands, the word "property page" is in the output, and that is NVDA-speak for "tab panel". You marked it as failing for "conveying the 'tab panel' role"

dean: We both recorded the same thing. There's no "tab panel" in that output

Matt_King: But the term "property page" is.

dean: Ah, okay. So I need to change all three commands to passing. I can do that

Matt_King: Then, in test #15, we would like you to re-run and check the output for the space bar

Matt_King: Joe_Humbert's output is showing that "selected" is showing multiple times

dean: I've seen that before elsewhere, so I'll check it again

Matt_King: Now I'm wondering if we actually want to record the negative side-effect for this one

Matt_King: This is tabs with manual activation, so pressing the space bar generates a state-change event: the tab goes from "not selected" to "selected". I guess NVDA is reading the result.

Matt_King: James, I'm curious about your thought on this. What do we expect for a speech response in this case?

Matt_King: You're pressing the space bar on a tab that's not selected

James: As an NVDA user, I would expect to hear the word "selected" (The change in state), and that's it

Matt_King: Well, we can call it excessively verbose because after reading the state, it reads the name, the role, and the set again

Matt_King: We only have one assertion here, and that is that the change in state is conveyed

james: There is a persistent issue in NVDA where, when you activate a tab, it reads the tab again

Joe_Humbert: When I re-ran the setup for this test, it puts you into forms mode. You have to run the setup again

Matt_King: That's expected because it's a tab that receives focus

Joe_Humbert: I'm not hearing "select" a second time, so it might have just been the bot results

Joe_Humbert: I'll change my results

Matt_King: Okay, so Dean, you have nothing to do on test #15

Joe_Humbert: The left-arrow test still has all my results, but the shift+f keypress does not.

Matt_King: I don't know how that happened

dean: I changed my results for test #13, so I'm showing no conflicts, now

Matt_King: Okay, great

dean: I was seeing weird results, so I updated, and then, my results seemed accurate. But they still didn't match with the bot

Matt_King: If you unassign yourself, will that delete it?

Dean: That's how I did it before. I had completed only a couple tests, unassigned myself (wiping it), then started a bot run, and finally re-assigned the bot run to myself

Matt_King: You can do that again.

ChrisCuellar: The bot is still behind (we still need to update it), but that seems like it isn't the issue here

Matt_King: If we collect using the bot at version 15.6.1, then if the bot results are accurate then you can just leave it

Dean: sure. It doesn't matter if we have the same version as long as we get the same results

Joe_Humbert: I just need to re-do the one test, correct?

Matt_King: That's right

Joe_Humbert: I'll do that ASAP

IsaDC: That just leave me to finish the JAWS test run. I'll do that later this afternoon

Matt_King: Perfect

Running tests for Switch Example Using HTML Checkbox Input

Matt_King: We have three bot runs available, and we don't have any testers

Matt_King: We're looking for two volunteers for each screen reader

dean: I'll do VoiceOver

Joe_Humbert: You can sign me up for all three. I'll have time for that over the next two weeks

dean: I can do NVDA eventually, but I have three test plan runs assigned to me already, so it may be a bit

mmoss: I can run one of these for switch.

mmoss: The test queue isn't loading for me at all. I'm getting a 502 Bad Gateway

Matt_King: Same for me. It was loading correctly just a moment ago

ChrisCuellar: We saw a 502 last week. It's loading for me right now, for what it's worth, though it is a little slow

ChrisCuellar: I'll file an issue for my team to look into offline

mmoss: I have access to all three screen readers

Matt_King: Maybe we should assign you and Joe_Humbert to JAWS (because dean doesn't have access to JAWS)

Matt_King: Then we can have Joe_Humbert and dean on VoiceOver

Matt_King: Then, for NVDA, we can wait and see who is ready first

Matt_King: But we can't do that right now because the site is down

Matt_King: I think that by next week, we'll have both "tab" plans done and we'll be working on "switch." Hopefully we can start working on "checkbox"

Matt_King: I think that covers it. There is a lot more potential work in the automated-update space. Elizabeth finished one.

Matt_King: Thank you for completing the "action button" test plan with VoiceOver!

Matt_King: Now, we have these other plans in the test queue. Do you have availability to work on one in the next week?

Elizabeth: yes, I saw that the "action menu button" is available in the automated updates tab, so I assigned myself to that.

Matt_King: Right, you did the "active descendants" one. If you would like to take the other, that would help

Matt_King: The "command button" test plan for VoiceOver is showing up as a manual test run, but it was an automated test run...

ChrisCuellar: Please file an issue and we can take a look into it

ChrisCuellar: I'm filing an issue for the test queue performance

dean: There was a big lag when I tried to assign it to myself. It wouldn't work at all

ChrisCuellar: We have an issue for the lack of feedback when you reassign. That issue is still on our queue; we just haven't gotten to it, yet

Matt_King: Thank you everybody. We're making some great progress, and I appreciate all of it!

Minutes manually created (not a transcript), formatted by scribe.perl version 246 (Wed Oct 1 15:02:24 2025 UTC).

Diagnostics

Maybe present: dean

All speakers: ChrisCuellar, dean, Elizabeth, IsaDC, james, Joe_Humbert, Matt_King, mmoss

Active on IRC: jugglinmike