Meeting minutes
Review agenda and next meeting dates
<Matt_King> No modifications to agenda.
<Matt_King> Next meeting is is Oct 7
<Matt_King> Daniel: would like to talk about security issues at end.
Current interop reporting status
<Matt_King> No major changes to status test plan phases.
Matt_King: We need to work on the disabled button test plan, the tab indexing isn't working as expected.
Matt_King: Do browsers completely disable the ability to make an attribute with a disabled element? Maybe we wrap the element and set focus on wrapper
Chris: I don't know but it makes sense to me
Matt_King: Ok. maybe I can put something inside button that's focusable
Murray: Yes, a button that's disabled can't have focus. At least in Chrome and Safari
Matt_King: You probably can't focus something inside the element then
Chris: I think the wrapper is a better idea
Matt_King: Ok. Disabled button is up next in terms of what will be ready for testing.
Review agenda and next meeting dates
Issue 1402: TPAC Breakout
lola: I opened the issue on the TPAC GitHub because the deadline is coming up.
… No need for anything concrete at the moment, just the notion that we plan to do something
<Matt_King> github: w3c-cg/
lola: I'll work on some slides and I've put a draft of whaat we'll be covering
… We can assume people are aware of ARIA-AT already but we need to introduce them to the test plan composition
… What that is, how it runs, results, and how results are dealt with
… Then a couple of .calls to action -- we need test writer, platform engineers who can contribute time, and other dependencies
<Matt_King> github-bot, status
<github-bot> Matt_King, This is wgmeeting_github_ircbot version 0.4.8, compiled from df82e3369ccba25dbd82902a16200b6b4276de11, which is probably in the repository at dbaron/
<github-bot> I currently have data for the following channels:
<github-bot> #ab (no topic data buffered)
<github-bot> #aria (no topic data buffered)
<github-bot> #aria-apg (no topic data buffered)
<github-bot> #aria-at (11 lines buffered on "Issue 1402: TPAC Breakout")
<github-bot> no GitHub URL to comment on
<github-bot> #css (59 lines buffered on "[css-link-params] Passing `currentColor` (and other state) without requiring the receiving SVG to have a param")
<github-bot> will comment on w3c/
<github-bot> #cssbotprod (no topic data buffered)
<github-bot> #matf (no topic data buffered)
<github-bot> #svg (no topic data buffered)
<github-bot> #w3process (34 lines buffered on "w3c/
<github-bot> no GitHub URL to comment on
Matt_King: Looks great. I wonder if there are particular sections you want help with. Also, how content is going to be structured
lola: People could have a look at the part about the infrastructure. And there is also stuff about AT Driver
… It may be pertinent to somehow connect the two sessions
Matt_King: I could write a few bullets and then we may end up with some pictures from past presentations
lola: The front-end call out is going to be for the group in attendance, pretty much people not on this call. If you want to contribute to front-end tasks, Chris may be the person to reach out to
… There is quite a lot in terms of maintenance that needs to be done
Matt_King: There are a lot of issues, we haven't done priotization based on the situation we have now
Chris: Happy to help out depending on time
Matt_King: I could put together a project with a list of top-ten issues and then size them (small, medium and large)
… Anyspecific person may be able to choose their issues based on their bandwidth
… Sizing is different than it was a year before
lola: Sure we should do that but I don't want for this to be a blocker for the next two meetings. I'll miss next meeting because it clashes with the interop accessibility
Matt_King: Is putting content in the issue OK?
lola: Sure, I can put them in the slides afterwards
Tri-State Checkbox Test Plan
Matt_King: The issues we discussed last time (1410, 1411) are fixed.
… the Voice Over test plan needs to be rerun
Joe: It's already done
Matt_King: Looking in the test queue. Right now Joe is the only person assigned to Voice Over. NVDA is completely done so we just need to do JAWS and VO and Joe it looks like you're completely done with VO. Amruth needs access to do JAWS test.
Matt_King: Chris did you see the request to help Amruth get set up? I've been unable to add new testers.
Chris: I saw your message, I'll look into it.
Matt_King: We have 2 runs left out of 6. Other than unblocking Amruth, there's no other issue.
HTML Button test plan
Matt_King: Joe has completely finished JAWS, NVDA, and VO.
Joe: NVDA doesn't have the latest version number.
Matt_King: I just installed the new version but I guess I didn't update.
Joe: For JAWS we're using 2026 from here on out because 2025 wasn't an option. I am concerned about further testing with JAWS because it looks like Vispero is going to require an account. It prompted an account when I downloaded 2026
Matt_King: Do you need a liscence for an account?
Joe: No but I'm uncomfortable with the potential tracking that will come from signing up and agree to a licence agreement.
Joe: They may force me to buy a liscence if they see that the work I'm doing is considered profesional.
Matt_King: I wonder if we can have a liscence for ARIA-AT
Joe: They can because we do that at the university
Matt_King: Maybe we discuss with Vispero as a way to support the project. We can also recruit more heavily amongst JAWS user base.
… If you know someone who'd be in a good position to help us in that way, please let me know! We need the same kind of outreach in the VO community.
Joe: I've noticed another issue to do with the Chrome update and only noticed because I'm using JAWS demo lisence. The test web page doesn't work properly in Chrome, the button stays deactivated.
… I make my key combinations and submit but it errors then I have to refresh the page.
Matt_King: one way to do it is to leave the test page window open and just refresh. You don't have to close the window between commands.
Joe: But do I need to reset? I thought closing it was resetting it between commands.
Matt_King: If you refresh it will rerun the setup.
… So you haven't found any issues in the actual button plan which is good news!
HTML Button test plan
Issue 1671: VoiceOver Bot is not running
Matt_King: We're moving back to GitHub Actions, Chris can you give more detail?
<Daniel> github: w3c/
Chris: Yes, I have a draft PR on the GH Actions repo which I should have linked. We've been wanting to move back to GH Actions because they're ephemeral test runs which are created and destroyed as needed. Mac Stadium forced us to update recently and was going to reinstall everything but I've also been tracking GH developments and GuidePup
screenreader software, there's been a lot of work to get screenreader testing working on MacOS again. It's working again for GuidePup.
… I thought it'd be faster to go back to GH and I was right, we just need to update configs. Automated VO is working again on GH hosted runners. 15 works fine, 26 will go through a test plan and triggers the correct controls but we're not capturing the speech output via the harness. I'm unsure where the issue is, so we likely need to update to
acommodate macOS 26 updates.
… Looks like the harness may not run on macOS 26 so I will debug, that's the next big blocker. But all in all good news.
Matt_King: Previously the problem was that the macOS script wasn't getting created properly.
Chris: Yes, but this seems to be sorted on GH's side. So next step is to figure out what the bug is. It's promising that it's running through the whole test plan
Matt_King: Related to the conversation we had where Joe asked to run the test plan locally and I had a question about how the local output could be published via the app.
Chris: I wanted to write some documentation for that but I haven't gotten to it.
Matt_King: Can you help us understand how that would work?
Chris: The machine on mac stadium is just a bare metal mac machine that's running a test plan and in the background it has a GH runner script which sends results back to our GH Actions repo. You need to be able to run the test harness locally, you need the AT-Driver repo installed locally, we have internal stuff too that you'd need to have running.
… The other piece is installing the GH runner but we'd need to sort GH team access.
… We should do the move to the GH CG space then we can try adding memebers to that GH team.
Matt_King: does the test run on the local machine or GH?
Chris: it would be your machine, you'd have to register your machine. We could create a way to run any of self-hosted or GH runners.
… the ARIA-AT app is configured to run the self-hosted workflow. There might be a way to run whatever's available.
… If we wanted to go back to GH hosted runners, we'd need to go back to using the regular VO that was configured for GH.
Matt_King: so we'd need to redeploy the app.
Chris: yes
Matt_King: So Joe running it locally would actually be Joe's machine being able to run things in that his machine is part of an available pool of machines
Chris: What do you mean?
Matt_King: If was running it, it could theoretically run on Joe's machine
Chris: yes, exactly
Security Patches
Daniel: as a recap. Thanks for your patch, it hasn't been merged but there may be other actions that need to be run.
… I saw your email about user capability. I wonder if we can make it more difficult to create a fake user and not created automatically.
Chris: Just an update on the first patch, it got merged into dev branch which usually gets merged into staging but that failed because of outdated scripts on our side.
… I have a patch out that fixes it. Howard has approved it. I was prepared to test on staging and see if we could deploy. I think the next step is to get that working. This is my priority this week so that we can get security patch deployed on production.
… Do we have an issue for fake user?
Daniel: no, just email but can resend
Chris: yes please. That was for local developer convinence but shouldn't be live.
Matt_King: it's still using for development so maybe we can include settings.
Chris: I can look into that next
Matt_King: do we want to have an issue?
Chris: yes, that would be good.
Matt_King: Daniel can you recreate the issue?
Daniel: yes