W3C

– DRAFT –
(MEETING TITLE)

01 May 2025

Attendees

Present
carmen, dean, hadi, IsaDC, james, Joe, jugglinmike, Matt_King, michael_fairchild, mmoss
Regrets
-
Chair
-
Scribe
jugglinmike, Carmen

Meeting minutes

title: ARIA and Assistive Technologies Community Group Weekly Teleconference

Review agenda and next meeting dates

https://github.com/w3c/aria-at/wiki/May-1%2C-2025-Agenda

Matt_King: Requests for changes to agenda?

Matt_King: We'll switch the order of the topics to accommodate Joe, who will be arriving for the second half of this meeting

Matt_King: Any other topics to add?

Matt_King: Hearing none, we'll stick with the ones on the agenda

Matt_King: Next Community Group Meeting: Wednesday May 7

Matt_King: Next AT Driver Subgroup meeting: Monday May 12

Current status

Matt_King: We made some progress over the week

Matt_King: We got the conflicts resolved for radio group with roving tab index--thanks to IsaDC

Matt_King: That's now the 15th plan in candidate review

Matt_King: We have some updates on some of the other things that are on hold

Matt_King: The APG Task Force has a change for the vertical temperature slider to change

Matt_King: We know what the changes will need to be--we took the value out--so we could probably change the test plan partially (we only have to wait to update the references)

Matt_King: There's a change in the example, and that will change the number of arrow key presses required for all screen readers (taking it from three to two, consistently for all of them)

Matt_King: Actually, the only thing we could really update ahead of time is the CSVs, so it's probably not worth it. Let's wait for it to be published.

Matt_King: I'm hoping it will be a pretty quick change

Matt_King: The disclosure test plan, at least when it comes to JAWS, the changes won't be published until the July release of JAWS. There are still more fixes to do. We may be moving anything with a same-page link further out into the schedule. I haven't adjusted the schedule accordingly, yet

Matt_King: There's a pull request for the next "disclosure" test plan. It's waiting on my review, so it may be ready for the test queue next week

Issue 1214 - AT version recorded for reports where bots collected some of the responses

github: w3c/aria-at#1214

james: We started talking about this at the very end of our last meeting

james: We briefly discussed the idea that we care what the bot was running with and we also care about what version the human was using

james: I think we landed that ideally both pieces of information would be stored, but that maybe the bot's version wouldn't be included in reports

Matt_King: I think there are two decisions to make here. The first: do we want to record both versions? The second: what should we show in the run history and the reports (based on decision 1)?

Matt_King: It seems like it might be simpler to say that one takes precedence over the other and that the human has the final say

Matt_King: Are there any downsides to having the human's version overwrite the bot's version

Matt_King: It feels problematic to allow the bot's version to overwrite the human's version

james: I agree

Matt_King: So it's just a matter of: do we record both, or do we give precedence in some way?

<Carmen> jugglingmike: overwritting is a more practical idea. Having a log of the events it's something we have been talking for a the last few months.

Matt_King: If you visit the "reports" page and review the history, you can find the name of the tester and the date and time they completed their run

Matt_King: That wouldn't need to change if the human tester's version takes precedence

Matt_King: the only UI that would need to change is when the human tester opens a test plan run where the bot has done some work, we would want to make sure we're recording the version that the human is using. I suppose we can use the same warning prompt that we've already built--asking if the user wants to change it

Matt_King: It would be like the prompt that we get right now when an admin edits a report

Matt_King: It would also be displayed whenever a human opens a test plan run that was touched most recently by a bot

Carmen: makes sense to me!

Matt_King: This issue is in the ARIA-AT repository. You could move it to the ARIA-AT App repository so that you don't have to make a brand new issue

Issue 1240 - Reporting feedback on bot output and performance

github: w3c/aria-at#1240

Matt_King: IsaDC raised this today

Matt_King: Joe was raising issues for VoiceOver bot problems

Matt_King: I changed the title on those three issues, from "Feedback:" to "VoiceOver Bot Feedback:"

Matt_King: I've done that a few times in the past as well; I don't know if we want to continue doing that moving forward

IsaDC: Those issues have an impact within our data

IsaDC: We haven't documented how to deal with them

Matt_King: When we get feedback on a test plan, they get linked to the test plan. And we have to close all the issues on a test plan in order to advance it in some circumstances

Matt_King: So this workflow creates issues that are tied to a test plan when they actually should be associated with a bot

Matt_King: I assume it's really helpful to have information about the specific test plan version and test when people are giving feedback on the bot

Carmen: It is. We use that information to replicate the issues

Carmen: Perhaps I can do something to move them?

Matt_King: You could edit the description of the issue and remove the HTML comments (which are hidden when the issue is rendered on the page)

Matt_King: If those get deleted, I don't know if it would delete the linking in real time, or if the link information is static

Carmen: I don't know either, but I can ask howard-e and get back to you

Matt_King: Sounds good. howard-e will know

Matt_King: This solution is a little hacky, but it's probably better than writing additional code

carmen: Hopefully we will soon reach a world where consistency issues are no more!

IsaDC: The other part of the concern was that, for example when we have an issue raised against the test plan itself (e.g. to fix the command), we respond to say that we're working on it. How does this work with the bots?

Carmen: I add it to the project, but I don't know if that would be visible to you. Would you like me to add a comment?

IsaDC: That would be lovely

Testing of Rating Radio Group

Matt_King: We have 8 NVDA conflicts and 4 VoiceOver conflicts

mmoss: I just finished the VoiceOver test a few minutes ago

mmoss: I ran the test a long time ago, and so I had some old results. I updated those, and there are no longer any conflicts

Matt_King: I'm going to mark it as "Final", then!

Matt_King: Now, we're just down to NVDA

Matt_King: Maybe "insert+tab" should be the only command

Matt_King: I'm kind of wondering how dean got the result that he reported

Joe: I had to change the results from the bot

Matt_King: I believe that what you did manually was correct, and it looks like the bots command didn't match

Matt_King: The bot was behaving as though it were in focus mode

Joe_Humbert: No, that's the label for the group. If it was in the wrong mode, it would say something like "one star"

Matt_King: Oh, you're right! The bot was just in the wrong place...

Matt_King: I think there are actually two problems, here

Matt_King: There is a bot problem, and there is a problem with the test plan (I think we should remove the "insert+up arrow" from this test)

Matt_King: If we remove that command, there will be no more conflicts

Matt_King: "Insert+up arrow" isn't really a "read all information" in all contexts

Joe: That makes sense to me

Matt_King: I know we left it out in other test plans for this particular test--the "request information" test

Matt_King: So, test 14--you're right, it is here. "Insert + up arrow". My guess is that in that test plan, the "pizza crust options" are not all together and NVDA is separating

Matt_King: Was it "disclosure navigation menu"? There was a whole bunch of stuff all on one line in that

Matt_King: I have definitely done this somewhere

Matt_King: It's strange that NVDA is not reading all of the disclosure buttons on the same line in the "disclosure navigation menu" (at least, when all of the buttons are collapsed)

Matt_King: There is some other plan where we removed "insert + up arrow". I remember performing the delete myself and discussing it with james and IsaDC

IsaDC: Just to confirm, I got the same results as Joe for "insert + up arrow"

IsaDC: I agree with removing this command

Matt_King: NVDA also offers "Insert + tab" to do this

Matt_King: I don't think that this is a failure of NVDA to do what it says it does for "insert+up arrow"

Matt_King: Okay, we're aligned. We'll fix this by removing "insert + up arrow". That will remove these conflicts, and that will complete this test plan

Re-run of JAWS for color viewer slider

Matt_King: We were doing only JAWS for this one. Two testers, Joe and Hadi, are 100% complete. There are four conflicts

Matt_King: Joe is getting the min and max output, and Hadi is not

Matt_King: Perhaps the JAWS version is different. We don't show version information in the "conflicts" page (we may want to change that--it has sometimes been an issue)

Matt_King: When we show these conflicts, it might be good for us to--where we show the output, we could also show the browser version and AT verison

Matt_King: We could add at version and browser version after the output column

IsaDC: Yes!

IsaDC: And on that note, there is no way for us to return to the test queue (other than pressing "back"). There's no button to go back to the queue

IsaDC: There are no breadcrumbs here (unlike in the reports)

Matt_King: It would be good to add some breadcrumbs there

Matt_King: I've also wanted the particular AT to be part of the title of the page. In this case, that would be "Conflicts in JAWS results for {name of the test plan}"

Carmen: I can write up an issue

Matt_King: Let's confirm the testers' versions of JAWS

Hadi: I was using version 2025.2504.89.40 etc (the latest version published in April)

Joe: I am now running the latest, today. I don't know what I was running when I was running this test plan. I would have to double-check

Joe: I may have been running a slightly older version

Matt_King: There's a possibility that they may have regressed support for min/max in the April release

Joe: I can re-run to double check. It shouldn't take very long--it's just a couple of test, and it's just that keystroke

Matt_King: Great

Matt_King: It looks like the test queue is soon going to be empty. However, if there is a plan in the wings here, it could get merged and updated today if I get on top of my game sufficiently

Matt_King: That would be a disclosure test plan

Matt_King: there might be some value in a feature to be able to mark something as "on hold" in the test queue in order to prevent people from working on them

Matt_King: Like an admin function where we mark it as "on hold", and it disables... something. Perhaps the "continue testing" button, though I don't know if we want to completely block the ability to access the test. Perhaps just a warning that lets viewers know that the test plan is on hold...

Carmen: I can write an issue for that and present it to the team

Matt_King: Anyway, I think we're still good to go forward with this report

james: I'm concerned that people won't know what it's conveying

james: e.g. "one" versus "won"

Matt_King: Okay, well, we will have this new "disclosure" test plan ready very soon. Is anyone available to take up more testing?

dean: I will do NVDA or VoiceOver; whatever you need

Joe: You can sign whatever to me

mmoss: I also have availability in the coming week

Hadi: I'm available to do JAWS testing on the disclosure plan when you have it ready

Hadi: I may not be able to join on the Wednesday meeting, but if you notify me via e-mail, that should be fine

Carmen: We have an issue with the harness right now, and the bot is not working. I will send a message to the team when we know that it's fixed

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

Diagnostics

Succeeded: s/half/half of this meeting/

Succeeded: s/start/star/

Maybe present: Joe_Humbert, title

All speakers: Carmen, dean, Hadi, IsaDC, james, Joe, Joe_Humbert, Matt_King, mmoss, title

Active on IRC: Carmen, jugglinmike, Matt_King