W3C

– DRAFT –
ARIA and Assistive Technologies Community Group

02 August 2023

Attendees

Present
Alyssa, JoeHumbert, Sam_shaw
Regrets
-
Chair
Matt King
Scribe
Sam_shaw

Meeting minutes

<Sam_shaw> Agenda: 1. Discussion of new test writing language to eliminate use of terms "interaction mode" and "reading mode" 2. Data integrity assurance planning

Current Testing Status

MK: Right now, we dont have any NVDA testing in the queue, only JAWS and Voiceover. Two plans for VO

MK: For JAWS, Hadi has been assigned some test that aren't complete

MK: For VO, only Isabel has been assigned a test

MK: We don't have an additional person assigned for VO Test Plan Actions Action Menu Button Example Using aria-activedescendant

MK: Joe are you able to test that?

JoeHumbert: Yes I can

MK: Alyssa, have you completed everything you were assigned?

Alyssa: Yes, I don't think I've been assigned anything new

MK: Testing updates. All the test plans will be changing in the next couple of months

MK: Right now there are big changes related to the app, mostly test plan admin updates, that will be deployed in the coming weeks. These will help James and I view more clearly what is happening with the working mode process in the app

MK: As we've been working with the Screen reader devs, we've encountered two new issues. 1 as assertion is associated with a command, that the developer doesn't think should be associated with that command. Example navigating with quick keys with JAWS, they believe the output should be different than navigating with arrow keys

MK: In our internal meetings, we've decided we agree

MK: We also think we need a middle set of requirements, to notate assertions the SR should do, but not absolutely necessary Beauvoir

MK: To fix these things requires some big changes, and there are a some knock on problems that are a result

MK: Some of these plans, like the insert up arrow/tab problem, several of the test plans we have already completed, some we anticipate running into soon with the next plans we want to test.

MK: Also, for testing related to mode switching, we now know we will need the mode switch assertions only on some commands

MK: So those are the issues we are currently blocked on

JoeHumbert: Do you have a timeframe as to when the decision will be made?

MK: Yes, I will have a complete project plan proposal for James and I and anyone else to review, done by Monday. I'm spending all my time on this

MK: Then Bocoup will perform some assessments to prioritize the different changes a couple weeks after that. Then we will focus on the P1 issues, I aiming to have some real progress by september, wrapped up by the end of september

JoeHumbert: Okay, when should I aim to have this testing done for Action Menu with VO?

MK: in the next couple weeks would be great

Discussion of new test writing language to eliminate use of terms "interaction mode" and "reading mode"

MK: There are two problems here, both related to VO. Apple insists VO is modeless. Therefor we have all these tests we navigate in reading mode or interaction mode, then we have separate tests for VO that don't mention reading or interaction mode.

MK: This is a problem because essentially each SR has a different set of tests

MK: The other problem, with the current approach, we aren't really testing VO the way VO users use it, with is we don't turn on quick nav

MK: We also use a bunch of commands that aren't commonly used

MK: The bigger problem is having a different set of tests, which causes some confusion with reporting

MK: I've been thinking about how to solve this problem and I have an idea

MK: I'm thinking of ways we can get rid of that wording

MK: I don't have this written up in the issue yet.

MK: One this VO clearly has that seems analogous, is the voiceover curser and keyboard focus. JAWS talks about the virtual cursor and pc cursor, NVDA talks about application focus and browse mode but I think the use the word pointer

MK: NVDA doesn't refer to a cursor

Alyssa: NVDa has a review cursor and object navigation

MK: Thats right!

MK: The goal is to not use SR specific language

MK: So they all talk about cursors.

MK: Here is a proposal. I went through the discloser plan.

<Matt_King> Navigate forwards to a collapsed disclosure button in reading mode Move reading cursor forwards to a collapsed disclosure button

MK: That is my proposed wording change to the disclosure plan

JoeHumbert: I'm assuming there will then be less tests?

MK: No same amount of tests, just changing the wording. For VO actually we will end up with twice as many tests but they will be shorter with fewer commands

MK: Right now VO doesn't have a test for moving the curser

MK: If we have the navigate forward to x in reading mode test, that would become move reading cursor forward

MK: This would allow us to use the same test for JAWs and VO.

MK: If we refer to the javascript in in the browser as the application that works, and would work for native applications

MK: So that works well for navigation type commands

MK: I think that also works with any SR

MK: So we have four general categories of commands: Navigation, reading, operation, Those are 3 main ones, then some miscellaneous ones like dismiss

MK: Now for the reading ones

MK: These are trickier

MK: We currently have commands in two tests, read info about button in reading and read info about button in interaction mode

MK: Borrowing from the VO test, it has language that says

describe the element in the reading cursor, and describe element with keyboard focus

MK: We say read info about button in reading mode, or read info in interaction mode

MK: For reading mode

MK: Let me know what sounds good

MK: Read info about a button in reading cursor

MK: request info about a button in reading cursor

MK: Describe button in reading cursor

MK: Interrogate a button in a reading cursor

Alyssa: I feel like request would be fine, thats what you are doing

MK: Thats correct

MK: Yeah thats what I was wondering if someone would interpret that that way

MK: What do you think Joe?

JoeHumbert: I agree with Alyssa

Alyssa: Interrogate could be fun!

MK: You say insert tab is requesting info about it. If in a grid, JAWS has a command to give you row and column info, so it would have a different assertion but still asks the same request to info about where you are

MK: Then you don't need a separate test for row and column

MK: You could have one test with many assertions

MK: So when it came to the application cursor, I said request info about a button with application focus

MK: We are kinda saying request info about a button that has application focus, but instead to shorten it I has with application focus

MK: In VO the cursor is a box visually

MK: Alot of times in our documentation we talk about focus, but what we really mean is the cursor

MK: We could say with reading cursor focus instead of with cursor focus

<Matt_King> Request information about a collapsed disclosure button with reading cursor focus

MK: I think I will go with that for the proposal

MK: I think thats a good discussion about reading and requesting info in ways of working

MK: The next set of commands relates to operating

MK: We current have "operate a collapsed disclosure button in reading mode"

MK: My proposal is

MK: " Use reading cursor function to operate a collapse disclosure function"

Alyssa: That is very clear to me

MK: Then the interaction mode version " operate a collapse disclosure function in function mode" would become "use application functions to operate..."

MK: It seems when I apply that wording in operating, replacing operating mode with application function and reading mode with cursor functions it works well

MK: Here's one I haven't figured out yet "Dismiss dropdown in reading mode"

Alyssa "dismiss a dropdown with reading cursor active"

MK: If we do that, then what would be the application version? With VO the reading cursor is always active

MK: "Use reading cursor functions to dismiss a dropdown"

MK: I need to review the purpose of this test though I'm not sure this is the focus

MK: This might also be a case of examining "Do we want this test?"

MK: I will need to review this test

MK: "Use application function to activate a link in the drop down"

MK: The only that seems a little suspect to me is the dismiss one

MK: So this is some good feedback

MK: I'm actually hoping the tests become easier to read. We also need to increase the readability of the tests, so having clearly worded tests will effect how clear the reports are

MK: Okay this will be in an issue

MK: Anyone else have thoughts to what to watch out for?

JoeHumbert: Just be consistent and not choosing new terminology all over the place. So that its easy to jump between test for one AT and tests for another to clearly understand the direction

MK: I hoping that all desktop SR would have the exact same tests with the same words, just the list of commands and assertions may vary.

MK: We currently provide translation " How do you turn off the reading cursor"

MK: This will create some tests we need to rerun but I think we can handle it

MK: Thanks everyone, good discussion. Next weeks meeting is on Thursday

MK: We might be able to skip a meeting next week, I will discuss with James. I think with the state of the test queue and all the work on the application we may not need a meeting

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

Found 'Agenda:' not followed by a URL: '1. Discussion of new test writing language to eliminate use of terms "interaction mode" and "reading mode" 2. Data integrity assurance planning'.

Maybe present: MK

All speakers: Alyssa, JoeHumbert, MK

Active on IRC: JoeHumbert, Matt_King, Sam_shaw