W3C

– DRAFT –
ARIA and Assistive Technologies Community Group

18 April 2024

Attendees

Present
Castillo, Del, hadi, Isa, Isa Del Castillo, James, James Scholes, Joe_Humbert, Matt_King, Michael_Fairchild, murray_moss, Scholes
Regrets
-
Chair
Matt King
Scribe
Sam_Shaw

Meeting minutes

Current Status

Matt_King Our goal is Goal: 6 recommended plans by June 30

Matt_King We have 4 plans in candidate review: alert, command button, link, toggle button

Matt_King Radio test plan in draft review.

Matt_King Next up will be the Modal dialog

Matt_King Isa can you give an update on the Modal Dialog and where we are at?

Isa No more questions, we are refactoring and joining tests like we did for Radio Group, which will shorten the plan. We are looking at 10 to 11 tests

Matt_King Do you have an idea when you will be done adding the boundary tests?

Isa I'd like to review them with you next Tuesday on our call. I'm aiming to have them ready for the next CG meeting

Matt_King I think I need to move our meeting next week, I'll reach out to reschedule

Matt_King Okay so maybe by next Wednesday Modal Dialog will be ready

Matt_King Okay, awesome. And if anybody wants to look at what's going on. We do have a open poll request for refactoring the radio group and you can Just you can look at the poll request and there's a preview link in there. Its very close

Candidate Review Issue 1032

Matt_King The next topic is. Canada review issue. 1032 which we've previously discussed and it's about alert.

Matt_King I'm going to try to carefully represent what Vispero shared yesterday

Matt_King I just learned of a new ARIA issue however

Matt_King The issue is related to live regions and alert in particular and that it's unclear what browsers are supposed to be doing.

Matt_King When should the Alter event be triggered? On node creation? on Text insertion in the Dom?

James: I do just want to Raise a bit of a concern. I think Vispero has been incredibly patient throughout this and have been very willing to discuss on several occasions but I think we I think we may be getting to the point if we haven't crossed it already where this is dragging to long. I'm hoping we can make a decision about this soon

Matt_King I agree, I don't want to reduce enthusiasm for this project

Matt_King Essentially Vispero is saying that if the priority of creating the roll alert is treated by ARIA AT CG as a priority 2, that they are concerned how their stakeholders and customers will interpret this. They don't agree in this case that a SR should convey roll alert because of the history of this issue

Hadi: I Think announcing Alert is a Must, but I don't have all the history

James: To Hadi's point is that in this group we have opinions, we need to reconcile with data. Vispero in this case has data from customers that is directing them

James: Vispero is not announcing the alert based on user feedback

Matt_King Now we have evidence why people are using role alert when they shouldn't. But everyone across the ARIA world has said you must use roll alert.

Matt_King This is what came up this morning in the ARIA meeting. People want to fix this

Matt_King It was interesting to see how there wasn't complete alignment on how live regions are supposed to work. That is something that is not clear in the ARIA spec

Joe_Humbert Isn't the only difference between a generic live region and roll of alert, is that it does something with that roll?

James: yes. but practically speaking developers are finding that if they use the roll with a reasonable amount of assertiveness its not being treated that way

Joe_Humbert So its being copted because it works better with assistive technology. thank you

Matt_King So people are using role alert with what they should of been using ARIA live = assertive

Michael Can I propose we change this to a may?

Michael: I think that would make sense based on this history and what Vispero is choosing to do

Joe_Humbert If the intent originally was optional, then I agree may would be fine

James: I think that Vispero has treated this exactly as the original ARIA spec defines.

James: They are going on both user feedback and the spec, and putting the usage patterns third, which I completely agree with. I think we should change this to a may

Michael: I think this will become easier if there is more alignment, we should strengthen the assertion rather than weaken it

Matt_King Great I think this is great way forward

Joe_Humbert How then would a developer indicate to a SR user that something has this importance?

Joe_Humbert Because then there is no way to say this message is more important than another

James I think that would come with the accessibility notifications API, but thats 12 years away

Matt_King One way to differentiate importance by using other things, like an alert sound or moving focus.

James: To your point Joe, and this is Vispero's point there is no way to distinguish between importance.

James: To touch on the other solutions to convey importance that we discussed with Vispero was to play a sound or in braille a flashing dot. The issue was there isn't a way to track these things with automation, and how do we describe the sound for the tester? Especially for users without hearing. The other solution was keep it as a should,

reducing Vispero's score, but with a note explaining why Vispepro chose not to convey this roll. Vispero thought that people may not read this note, or have the expertise to understand. This provides alot of extra cognitive overhead

Matt_King We also just discussed making it a should and communicating that full support doesn't means 100 percent on the should. Because as defined in the definition of should vendors may have a reason for not doing something that is deemed a "should"

Matt_King To move forward, resolution proposal:

RESOLUTION: The assertion that AT should convey role alert will have the priority of "MAY".

Conflicts with Radio test plan

Matt_King There are 4 conflicts, I raised two issues

w3c/aria-at#1052

w3c/aria-at#1051

Matt_King First one is Issue 1052

Matt_King So some people are saying this is a moderate impact, one person saying it was a severe impact

Matt_King Murry and Isa said Moderate, Joe said severe

Matt_King So Joe what is your proposal?

Joe_Humbert I thought it was severe because the user is moving in the opposite direction that they expect. I am not tied to severe, if the group wants to change it I'm fine

Matt_King I'm curious about Isa and Murrays thoughts. My gut reaction is to say this is severe, it has a easy recovery

James: Easy on this webpage, could be more difficult other places

murray_moss I'm comfortable with severe. I thought when I pressed shift J it put me back in the group

Matt_King Yeah so this is a weird one because it only happens right after pressing test setup button, but not after that

Joe_Humbert Is it because the test setup is acting up?

Matt_King The test setup isn't doing anything unusual

Matt_King I'm not sure if this is caused by the set up script. I don't think that these set up scripts are doing anything a normal webpage wouldn't do

James Even it they didn't the setup script just puts focus on something, its not an exotic script

Isa: I marked as moderate because when I repeated it it worked

James So there is a bug in Safari or VO where the focus is being moved

Isa I'm comfortable changing it to severe

Matt_King We agree to change this to severe for tests 2 and 4

Matt_King to make this change Murray and Isa can you edit your results?

murray_moss yes

Isa yes

Matt_King Okay next issue

w3c/aria-at#1051

Matt_King In this case one of the test outputs looks like someone put the tab results into shift tab

Joe_Humbert I might have copied one wrong, I can go update that

Joe_Humbert I wrote forwards instead of backwards

Matt_King I think Murray may have made a mistake in the output

Isa I think it should be forwards

Isa How strange Murray and I made the same mistake

James hang on hang on

Isa okay for test 5, shift plus tab

Joe_Humbert While they look into that, the main issue I raised was that they announced the landmark

Joe_Humbert if that is a normal thing I can close my issue

James: I think there is a significant issue though

Matt_King I want to take a close look at this

James I think VO is the only SR I've used that announces the Landmark that you are using because you've reached the first link

Joe_Humbert I dont want to mark this as verbose if this is standard/expected behavior

Isa I want to rerun this test first

Matt_King I want to look more into this

murray_moss I just reran and it was a typo, however Joe's issue exists

Joe_Humbert okay so we agree this is a moderate issue?

Matt_King yes

Summary of resolutions

  1. The assertion that AT should convey role alert will have the priority of "MAY".
Minutes manually created (not a transcript), formatted by scribe.perl version 221 (Fri Jul 21 14:01:30 2023 UTC).

Diagnostics

No scribenick or scribe found. Guessed: Sam_Shaw

Maybe present: Michael

All speakers: Hadi, Isa, James, Michael

Active on IRC: Joe_Humbert, Matt_King, murray_moss, Sam_Shaw