W3C

– DRAFT –
(MEETING TITLE)

09 October 2024

Attendees

Present
dhamack, elguerrero, IsaDC, jscholes, jugglinmike, Matt_King, mfairchild
Regrets
-
Chair
-
Scribe
jugglinmike

Meeting minutes

<elguerrero> Nice to meet you all! My pronouns are she/they and I forgot to mention I work as an Accessibility Specialist Intern, pivoting from frontend development :)

<elguerrero> Thank you jugglinmike !

Review agenda and next meeting dates

https://github.com/w3c/aria-at/wiki/October-9%2C-2024-Agenda

Matt_King: Requests for changes to agenda?

IsaDC: I have an update about the bots

Matt_King: Okay, let's put that near the top of the agenda

Matt_King: Anything else?

Matt_King: Alright, then

Matt_King: Next Community Group Meeting: Thursday October 17

Matt_King: Next AT Driver Subgroup meeting: Monday November 4

App update

Matt_King: Did the app get re-deployed this morning?

jugglinmike: It did

IsaDC: Now, I can assign test runners to the bot, but the results don't get recorded

IsaDC: The inputs are blank

jugglinmike: Noted. I'll check in with howard-e about this

Matt_King: Were you able to test out to make sure that the "submit" functionality was fixed, IsaDC?

IsaDC: Not yet, but I can run a quick test

Testing action menu button with activeDescendant

Matt_King: We're done with everything except JAWS

IsaDC: That's assigned to Murray, and I didn't want to send them an e-mail until I verified that the bug was fixed

IsaDC: The "submit results" button now works as expected

IsaDC: There's just the bots issue

Matt_King: Great. If you can let Murray know that testing can continue, we can hopefully get that wrapped up and move on to "Candidate"

Testing navigation menu button

dhamack: I got a notice this morning saying that the blocking issue had been fixed. I signed in and verified. Now that that's cleared up, I can crank through the rest of the tests assigned to me

Matt_King: Hadi is not hear today. They are working on JAWS

Matt_King: We're still looking for help with NVDA

IsaDC: Are we going to run the "navigation menu button" with NVDA manually?

Matt_King: Lets see what jugglinmike finds out about automation today

Matt_King: I see that the bot has made progress...

IsaDC: But all the responses are blank

Matt_King: Do you have any guesses about when that might be resolved, jugglinmike?

jugglinmike: I know what you folks now right now, so I'm really reluctant to make any estimates about timelines

Matt_King: Okay, well, I think it's not so bad if they're down for a few days

Disclosure Navigation Menu Plan Update

Matt_King: Here's the preview link: https://deploy-preview-1134--aria-at.netlify.app/review/disclosure-navigation

Matt_King: I first want to talk about the first four tests

Matt_King: In these tests, our primary changes. Previously, we removed a bunch of assertions about the list role and the list items

Matt_King: We also had a bunch of assertions about the list boundaries when you enter a region. Those are still in here, but they're optional

Matt_King: The assertions were totally removed from commands like "B" in NVDA and JAWS. NVDA does speak, though.

Matt_King: If you go to the first test, if you navigate with NVDA from the link, and you navigate to the button, NVDA will tell you about the fact that you're entering the navigation region and that you're entering a list of three items

Matt_King: When you navigate with "b" or "f".

Matt_King: JAWS does not

IsaDC: That's why I removed them

Matt_King: Right, but my position here is that we should just be completely optional. That was it isn't called out for excessive verbosity if those three things are announced (the list boundary, the navigation region name, and the navigation region role)

Matt_King: JAWS does do this for "tab", but these assertions will always be optional for all three commands ("b", "f", and "tab")

Matt_King: That was the first change I made this morning. I think making those optional is fairly non-controversial. I think we've discussed it before

Matt_King: Except for when you're navigating with arrow keys--then they are all required

Matt_King: The only way we'd want to remove them is if we all agreed that NVDA's speech should be considered a bug

Matt_King: I'm not hearing any objections, so we'll move on

Matt_King: Concerning test 9

Matt_King: Navigating into the dropdown. I made the exact same change, but just for the list boundary (because we're not crossing the region boundary; we're only crossing the list boundary)

IsaDC: Okay

Matt_King: There are two "MUST" assertions

Matt_King: If you navigate with "a", it has to say that you're inside of the list, but if you navigate through "u" or "k" with JAWS (for example) then it is optional

Matt_King: That's also a change to 10 (the only difference is the element attribute)

Matt_King: Come to think of it, we could delete one of those two

IsaDC: I vote for removing test 9 so that the current test 10 becomes the new test 9

Matt_King: Okay, sounds good. Shorted test plans are better

Matt_King: Now, test 11. In this one, the change that I made was on "up arrow". We were missing the requirement to speak the list boundary on "up arrow". That was a simple omission, so I added it back in

Matt_King: I did not make any changes to test 12

Matt_King: Regarding test 13: I also added the list boundary assertion, but only for the "down arrow, down arrow" command (or equivalent)

Matt_King: No changes to test 14 (dismissing)

Matt_King: For test 15, you activate a link. That moves focus to a heading on the same page

Matt_King: NVDA says a lot of stuff. I don't remember what VoiceOver does. JAWS communicates the fact that the focus moves inside of a region. It tells you the name of the region, but it doesn't tell you want it landed on.

Matt_King: When you activate a same-page link, it seems to me like there should be some base-level expectation that the screen reader should say something. But the question is, what MUST it say about where it landed?

Matt_King: I'm proposing that it only MUST say the content of the thing that it lands on

Matt_King: the link target might be a heading, but I think we should be careful because the link target can be an anchor on the page which is empty. In fact, it very often is such a "nothing" element

Matt_King: I've asserted here that it MUST say the content of the heading, it SHOULD say the role of the heading, and it MAY say the fact that the heading is in a different region of the page (and what the name of that region is)

Matt_King: That's how I've changed the test. Do people agree?

IsaDC: I agree. We didn't have those assertions before

Matt_King: Yes, I made them up

dhamack: Sounds good to me!

mfairchild: I am also okay with it

Matt_King: So I think only one change remains, and that is deleting test 9

IsaDC: Then we'll have to re-run the whole thing

Matt_King: Right. We can get this merged today and worry about assignments during next week's meeting

dhamack: are we just tabling disclosure for now?

dhamack: I had a bunch of conflicts

Matt_King: I think we're going to have you start from scratch--sorry

Matt_King: We cannot re-assign test plan runs to the bot

IsaDC: That would be great, though

Matt_King: Now that we've used the bots enough and we're so in love with the bots, could we open it up to give anyone the option to run the bot?

IsaDC: I'd have to re-assign them

Matt_King: Right now, that's true. I'm wondering if we can add the functionality to allow folks to run the bot and overwrite what's there

Matt_King: Or even to run the bot for all the inputs on a single test

jugglinmike: I can't speak to the implementation challenge behind that on-the-fly. Beyond that, the only potential issue that comes to mind is an increased risk of running up against usage limits in GitHub Actions (a lot of computational resources are expended simply in provisioning environments)

Matt_King: Ah, yes. Presumably, we could alleviate that particular concern given sufficient credits for Azure from Microsoft (potentially at the behest of mfairchild)

jugglinmike: That could be true, but we are currently blocked on using Azure due to technical hurdles with which we've been trying to receive support from Microsoft

mfairchild: If you need any help resolving those, please let me know

jugglinmike: Thank you! Our plan is to do just that. We'll be in touch

Minutes manually created (not a transcript), formatted by scribe.perl version 237 (Fri Oct 4 02:31:45 2024 UTC).

Diagnostics

Succeeded: s/Disclosure Navigation Menu Plan Update/Testing navigation menu button/

No scribenick or scribe found. Guessed: jugglinmike

All speakers: dhamack, IsaDC, jugglinmike, Matt_King, mfairchild

Active on IRC: elguerrero, jugglinmike