W3C

– DRAFT –
ARIA and Assistive Technologies Community Group Weekly Teleconference

10 September 2025

Attendees

Present
Carmen, dean, howard-e, Joe_Humbert, jugglinmike, kelly, Matt_King, mfairchild, mmoss
Regrets
-
Chair
-
Scribe
jugglinmike

Meeting minutes

Matt_King: We'll meet on Thursday of next week

Matt_King: However, we will NOT meet the following week (when we would normally meet on Wednesday the 24)

Matt_King: I didn't have time to create an agenda for today's meeting; we'll simply review the test plans in the test queue

Disclosure of Answers to Frequently Asked Questions Example

Matt_King: The only screen reader we have left to work on is VoiceOver

Matt_King: It looks like Joe_Humbert is all the way done

mmoss: I'll have my part done by the end of the day, for sure

Matt_King: Awesome!

Joe_Humbert: I haven't raised any issues in my latest round of testing

Rating Radio Group

Matt_King: We're skipping this for now; it's still on hold

Switch Example

Matt_King: We're only working on VoiceOver on this Test Plan, as well

Matt_King: This one doesn't have any conflicts, either

Matt_King: Elizabeth and Joe_Humbert ran this one

Joe_Humbert: I ran this a couple days ago. Again, I reported no issues

Matt_King: Elizabeth hasn't been here for a little while. I think IsaDC reached out to Elizabeth before IsaDC left, but I don't know the outcome of that

Matt_King: We may have to re-assign Elizabeth's work

Joe_Humbert: I will double-check the commands that I ran to be sure that I accurately reported the "untestable" condition

Switch Example Using HTML Button

Matt_King: In this one, we have five conflicts for VoiceOver

Matt_King: Joe_Humbert's results and dean's results are conflicting

Joe_Humbert: I marked the first as "unexpected behavior" and dean marked it as nothing

Matt_King: This is test 2, which is a "navigate backwards" test. Also 7 and 8

Joe_Humbert: in test 7, I listed unexpected behaviors, but dean did not. We reported the same AT responses and assigned the same assertion verdicts

dean: I think Joe_Humbert is correct on test 7

Matt_King: so on test 1, though, it did find the button?

Joe_Humbert: No, on test 1, since its VoiceOver, instead of saying "no buttons," it goes to the "run test setup" button. We both indicated that

dean: I'm looking at all these, and like Joe_Humbert said, the only difference is that I didn't mark anything as "unexpected behavior." I have no problem changing to what Joe_Humbert reported

Joe_Humbert: When I checked "untestable", it automatically changed the radio button to "unexpected behaviors" to "yes"

dean: I don't think it did that for me

mmoss: I observed the same behavior as what Joe_Humbert just described

Matt_King: And that's the way it's supposed to function

Joe_Humbert: Maybe dean was able to get around that by testing with an earlier version of the application

dean: Or I may have manually changed "unexpected behaviors" from "yes" to "no"

jugglinmike: The "no" radio button should be disabled when "untestable" is checked

Matt_King: Well now, I'm questioning whether or not we should have the "b" command in the test plan...

Matt_King: The reason we included "b" in the first place is because we thought two of the three screen readers were treating switches a buttons.

Matt_King: We were going to make the case that, "hey, it's called a switch button, so shouldn't it be included?"

Matt_King: But at least when it's coded this way, it seems to be working

Joe_Humbert: It looks like it goes to the "run test setup" button

Joe_Humbert: So apparently VoiceOver and JAWS don't work with a switch when you use the "b" key

Matt_King: Maybe we should remove that test

Joe_Humbert: I wondered about this, but I wasn't sure because I haven't done enough testing of switching in HTML contexts to know which buttons work and which do not

Matt_King: I wonder what iOS does when you switch to buttons on the rotor...

Joe_Humbert: You can't use button in native; it only works in HTML

Joe_Humbert: Only a subset of the rotor is available in native

Matt_King: If we remove the "b" command, then the conflicts go away

Joe_Humbert: Some of the conflicts go away. Tests 7 and 8 have different conflicts

Matt_King: I talked to James about this, and we decided to collect the data and decide based on that. Maybe we already have enough data to make a decision

Matt_King: I thought I had tested the "b" key with at least two screen readers and found it to work. Now, I'm questioning my memory

Matt_King: What did NVDA do in test 1?

Joe_Humbert: It worked just fine in browse mode

Matt_King: So NVDA is the only one...

Matt_King: Any change to what Apple does in this space would really be swimming upstream

Matt_King: When IsaDC returns, we'll talk though this

Matt_King: I think we know what the path forward for fixing the conflicts is right now

dean: I've already fixed most of them; I'm fixing the final one was we speak

Matt_King: Okay, as for test 7...

Joe_Humbert: dean and I got the same output. It's very strange--the screen reader says "no action available off"

Joe_Humbert: The "off" is confusing. This comes up in another test. It says "no action available off", but the visible switch updates to the correct state

Joe_Humbert: The visual updates but the screen reader is not reporting that

Joe_Humbert: This is operating a switch that is off to turn it on

dean: You can toggle that thing back and forth (off and on), and you can visually perceive the change, but it will not be announced by VoiceOver

Joe_Humbert: Right

Joe_Humbert: This occurs in test 8. There, dean marked it as "pass" because it technically said "off", but I don't think that "off" is related to any state. Because no matter what you do, it reports "off"

Joe_Humbert: It does report the state when you ask for it. It's this command (and I'm 95% sure on this), it updates the state of the button visually changes, but all VoiceOver says is "no action available off" regardless of the state of the switch

Joe_Humbert: For this test, you get no output when you press "space" or "enter"

Joe_Humbert: It seems that, at least with a switch using an HTML button, there are significant problems with VoiceOver

Joe_Humbert: that's why I marked "unexpected behavior" even though dean did not--I wanted to have this conversation

Joe_Humbert: I don't think I've ever heard the screen reader say "no action available"

Matt_King: It normally says when actions ARE available, but it normally wouldn't say "no actions available" unless you are trying to execute an action

Joe_Humbert: Right. But I've never heard that output from VoiceOver, so I was very confused by this

Matt_King: Isn't there a side-effect in the form that is related to "incorrect or misleading output" or something like that?

Matt_King: The option named "other" would be fine, too

Matt_King: I do think we should be reporting a side-effect, especially when the switch appears to be on but it says "off"

Joe_Humbert: What should we say for the details?

Joe_Humbert: "VoiceOver is announcing incorrect information, but the visual state changes"

Matt_King: That's fine

dean: I wanted to make sure that we had a note somewhere that these things visually change but are not announced

Joe_Humbert: There may be other conflicts in JAWS and NVDA. I'll double-check

dean: I'm going to finish my tests today/tomorrow because I will be out for about a week

Joe_Humbert: So there may be other conflicts that are not this (because we know how to resolve these conflicts, now)

Tabs with Automatic Activation

Matt_King: For this one, we have JAWS and VoiceOver in progress

Matt_King: VoiceOver is done. It says "99% done". I'm not sure why

Matt_King: There are four conflicts--they are in test 2 and test 4

Matt_King: Joe_Humbert and mmoss have different output

Matt_King: You're both using the same versions of the software

mmoss: We have the same output; I just didn't include the entire text

Matt_King: In test 2, mmoss says "you are currently on a text element." I don't see that on Joe_Humbert's output

Joe_Humbert: It is the last line

Joe_Humbert: mmoss just didn't include all of the text output for the two commands--he just included the final line

Matt_King: So for both of you, it read the entire tab panel

mmoss: Correct

Joe_Humbert: I'm happy to update

Matt_King: Oh, right. Now, I understand

mmoss: Instead of navigating backwards into the tab list and onto a tab, it navigated to the tab panel

Matt_King: When you navigate backwards--how in the world did it read that whole panel content...?

mmoss: It didn't read the tab

Matt_King: Did the focus ever reach the tab on "Peter Muller"?

Joe_Humbert: I can't remember

Matt_King: Why does it say "you are on a text input"?

Joe_Humbert: It actually says "you are on a text element", which makes sense because it is a paragraph

Matt_King: Ah. That's a weird thing to say, though

Joe_Humbert: I imagine most people turn that off

Joe_Humbert: Do I need to add an unexpected behavior, or does mmoss need to remove the unexpected behavior?

Matt_King: We should be capturing the whole output, though

Matt_King: There is an unexpected behavior, here. I think the way that mmoss described it looks pretty accurate

Joe_Humbert: Should we change it to "the cursor moved unexpectedly"?

Matt_King: I think that is very much the case. The reading cursor moved in an unexpected manner

mmoss: I'll update accordingly

Joe_Humbert: And once you do that, mmoss, I will update mine to match yours

Matt_King: That was for test 2

Joe_Humbert: There was a similar problem with "shift+j"

Joe_Humbert: It said "form element not found"

Matt_King: Is that also what it did for "j"?

mmoss: No, it was only "shift+j"

Joe_Humbert: I'm happy to update my unexpected behaviors if that's what we need to do

Matt_King: It just says, "no found". Does the reading cursor rectangle move at all? Or does it just stay put?

Matt_King: Actually, what is the reading cursor visually placed on?

Joe_Humbert: I would have to revisit this to say

mmoss: I just tested real quick and confirmed that the visual cursor doesn't move at all

Matt_King: This sounds like an untestable situation similar to the other "shift+j" ones

Joe_Humbert: In the other cases, we were saying just that the cursor didn't move

Matt_King: Okay, yes, I think that's the way forward here, too

mmoss: I'll update accordingly

Joe_Humbert: Me, too

Joe_Humbert: Same conflicts for "navigating backwards", so we can resolve them in the same way

Joe_Humbert: JAWS is unfinished because Hadi hasn't finished his testing. Do we need a third Tester?

Matt_King: We can give it some more time

Joe_Humbert: Hadi and Louis are assigned

Matt_King: Rather than having you do that right now, I think I'd rather add the "tabs with manual activation" to the test queue, then run the bots, then have you get started on that

Matt_King: In the case of Hadi and Elizabeth, I'm going to give them an opportunity to finish. Hadi usually pulls through given enough time--it's just a matter of lining up with his schedule

Matt_King: I'll add "tabs with manual activation" to the queue and start the bots and then re-assign to you, Joe_Humbert

Matt_King: I'll work on getting to that today

Matt_King: Okay, that's all for today. Thanks to everyone for the great testing!

mmoss: I will be out all of next week, but I will be back after that, and I will complete all the testing which is currently assigned to me by the end of the day

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

Diagnostics

Succeeded: s/option/radio button/

All speakers: dean, Joe_Humbert, jugglinmike, Matt_King, mmoss

Active on IRC: Joe_Humbert, jugglinmike, Matt_King, mmoss