Meeting: May 5, 2026, @ 9 AM PST

[cookiecrook]

Previous meeting: #226

This meeting will be May 5, 2026 @ 9AM Pacific. The meeting is usually recurring the first week of the month.

Please let @zcorpan know (or comment below) if you'd like to participate in the meeting and he can add you to the calendar invitation (or anyone who is invited can edit). Use the Agenda+ label on issues in this repo to add agenda topics, file PRs in the relevant directories, or comment below.

Next meeting is: #234

[cookiecrook]

Agenda+ to lock down the new time of the meeting.

Update: New Time is the same time (9AM Pacific) but now on the first Wednesday of the month.

[spectranaut]

Agenda+ aamtest PR ready for merge? web-platform-tests/wpt#57696

Agenda+ Starting work on Chrome's implementation of get_accessibility_properties_for_*

[rahimabdi]

Apologies, Stefan and I won't be able to make it today.

CC @cookiecrook

[ccanash]

1. Meeting times 

   • https://doodle.com/group-poll/participate/eV2KvoXe 

   • Val

     • Wednesday at 8 am has 8 yeses and Tuesday has 6 yeses. 

   • James

     • Did they include Jamie ?, because 9 am pacific any day will be out of his range. 

   • Val

     • The others were hard for the Eastern and European time. 

     • Well looks like Wednesday 9 am pacific  is the conclusion. Is that right?

   • Simon

     • There wasn’t an option to keep the current time and day. 

   • James

     • I think there was, if not it was an oversight. 

     • It was in the past options. If you don’t mind double checking Simon. 

   • Val

     • Oh, it’s because there are times in the past. 

   • James

     • That might be why the others were not displaying. 

   • Simon

     • So I guess we are switching to Wednesday. 

   • James

     • Wednesday 9am pacific is the winner. 

2. Discuss whether we can merge the AAM PR. 

   • Val. 

     • I can’t merge them because there are errors in the CI. 

   • James

     • I think we need Simon’s authorization to merge without CI. Was there an RFC for this?

   • Val

     • Yes, and it was merged. 

     • Explanation on why they failed - more information in the issue. 

       • Last one is a recent issue on Firefox. If we start the browser with correct command line and 1 out of 10 times it can’t find the browser. I haven’t been able to replicate it. I wrote Jamie in an email with a lot of details and Jamie thought it might be easier to solve once it has landed. 

   • Simon

     • Sounds good. I haven’t reviewed it but it shouldn’t .. It has been approved by 3 people. 

   • Val

     • Yes, James Graham and others (I missed whom) 

   • Simon

     • I’ll merge, I trust the other reviewers. 

   • Val

     • So exiting! 

   • James

     • Congratulations! 

     • Issue Interop 2026 Accessibility Investigation #202 is the issue we are using for Criteria for this year's goals. 

     • We are at 15%,  under Acacia AAM. It’s the second item in the list. 

     • Tyler is working on the third item for the next 5% 

   • Tyler

     • We actually ended up submitting it in another PR and merging: WebKit/WebKit@3e97afb 

   • James

     • Great, then we got 20% today 

     • Rahim is working on the next implementation which needs to land on gecko first. 

   • Val

     • I want to talk about that next. 

3. Webkit 

   • Val

     • I am working on that implementation on Chrome. There are 2 ways to work on this, through Chrome’s API ? and the Hooks. 

   • James

     • Should I replace Valerie? 

   • Val

     • Yest

     • The way to write this test to surface them on Test Driver internal it is not how WPT.fyi tests are run using the normal infrastructure of WPT in Chrome CI. If we want to see results in WPT.fyi we need to define the shape of the API for Web Driver. And I wonder if that is something we should start working on? 

   • James

     • You are right, I think we had a tracker for it last year. Let me track it. 

     • The plan was to not normalize while we were working on the prototype. I think we have achieved an agreement but maybe Jamie started a PR. 

     • But yeah we should track that and include it in a percentage somewhere. 

   • Val

     • I don’t see a PR for that.

     • How can we move that forward? Should I message JamieT? 

     • [recap of discussion on why we need an API for WebDriver]. 

   • Simon

     • Isn’t test driver internal just glue for the web driver implementation?

   • Val

     • Not necessarily, not on chrome’s infrastructure. In Chrome infra internal it is a way to run Chrome called ?? and you have more access to internal functionalities. It’s a total test build that allows you to have access to some internal.. 

   • James

     • I added a place holder on Interop 2026 Accessibility Investigation #202 and use it as a reminder to look for the PR. 

   • Val

     • I am happy to move this forward too. Will you look for the issue and let me know? 

   • James

     • Yes, if you find it before I do feel free. I don’t want to stand on your way. 

     • Going back to the summary. Thank Val for reminding me to update the issue. 

     • TreeWalker investigation. We added it as a placeholder. It will be tentative, it would be valuable to get the test and see where the difference lay. 

     • Formalizing the property bag. It is related. 

4. WebKit/WebKit@3e97afb 

   • Tyler

     • It might not work, I can help to make it if it doesn’t.

     • On MacOS you can issue that command, … it will force accessibility on. I think you can use it Val. 

   • Valerie

     • And I can also turn it off probably 

   • Tyler

     • If you turn it off it won’t do it unless you restart the browser. 

   • Valerie

     • That’s ok, it is part of the process.

   • Tyler

     • Safari might also read this. That is a problem we can figure out a different way,. 

   • Val

     • That might be a problem for the tests. Did someone tell me that Safari tests don’t run in parallel? 

     • I thought someone said that in the Safari CI infra they don’t run tests in parallel for wpt.fyi tests. 

   • James

     • I think the instances are re-started for different tests. But it's default.. 

   • Val

     • Problem is if you are running multiple Safari tests running on the same machine,. They might all have them turned on, 

   • Tyler

     • Yeah, we might need to fix it. How do other browsers do it? 

   • Val

     • They have a command line to do it with CLI flags. 

   • Tyler

     • I can look into that and see what we can accomplish. I can follow up with James and talk more. 

5. Miscellaneous within Interop 2026 Accessibility Investigation #202 

   • James

     • We'll use this list to adjust throughout the year. There are some easy low hanging fruits like editorial changes. These don’t block or expand tests. 

     • Feel free to take one of these for an assignment. If you want to jump on editorial change as a new person on the project is a good idea. If you want you can ping me on slack and I can suggest some good ones. 

     • The accessibility bounds might be a meatier one that we won’t get through this year. 

     • We update the % on Interop 2026 Accessibility Investigation #202 at the end of each call. 

6. Add WebDriver support for get_accessibility_properties_for_element and get_accessibility_properties_for_accessibility_node to WPT TestDriver #225

*   This is a place holder, we need to change it. 
    
*   Val
    
    *   Does anyone else want to talk about the shape of the API?
        
*   James
    
    *   I would, maybe we could schedule something for later in the afternoon to get Jamie on the call and Rahim. 

• [missed this]

  • James

    • Valerie I can create an issue 

  • Valerie

    • If we do that we would want to write a test for each attribute. 

  • James

    • Formalize the property bag. We put it in here as a stand alone because we won’t have implementation for all these properties right outside of the door. Label and role and check and press are already there. There are these 4 in Jamie’s PR. We could do a PR that includes those PRs and TBDs for the rest of the PRs. We wouldn’t have to test those. 

  • Mike

    • Jamie’s PR also has either one or both parent and children. I am curious among the folks here to what extent we feel comfortable about it> 

    • That seems more open to debate about how those are represented? Are we ready to write tests on that or is it better to lead with a discussion on what the shape should be. 

  • James

    • That is the previous item in the list. So I would say we can write tests as long as they are .tentative or .optional. 

    • We don’t have an ARIA property for something like is ignored. But we have for bounds or various things. There is not an obvious spec. Maybe an additional spec would be a good idea or add a new section to Core AAM inside of the attributes / characteristics table. 

  • Mike

    • When we talk about how the trees differs it is not about the semantics but about the structure. 

    • My initial reading was that this was more about semantic difference, like traditional WPT tests.

  • James

    • I have 2 examples. One of them is that for example you have a div and say overall scroll that runs into a scroll view at least — we get a different .. CSS auto will insert (DIDN”T CATCH THIS IN THE NOTES). 

    • It’s not like that in the other implementations.

    • The other example is CSS generated content. So I expect at least in those two scenarios there will be differences in the tree. 

    • We won’t necessarily expect that for any of the applications that are parent child complex for there to be the same semantic. 

  • Tyler

    • The actual platform tree can change a bit. The implementation of ATSPI in Webkit doesn’t display text in the same way. Even uncontroversial things can be controversial. There can be uncontroversial tests that can be written even in a non-tentative way. Like does this have a parent? That can break accessibility. 

  • James

    • It should allow us to add additional tests later. 

  • Simon

    • CSS AAM could be a good place because it is affected by display rules and generated content, etc. 

  • James

    • What do you think Valerie? IT is just starting. 

  • Val

    • Def, it seems fine. 

  • James

    • You are right, I can’t immediately think of one of these special cases that is not tight to a CSS feature. 

    • Seems like a good place to end it. 

    • Thank you all! 

    • Tyler and I will share with Rahim the updates 

    • Valerie, you will lead the WebDriver API.