W3C

– DRAFT –
ARIA and Assistive Technologies Community Group

10 August 2023

Attendees

Present
Isa, James Scholes, Matt_King
Regrets
-
Chair
Matt King
Scribe
Sam_Shaw

Meeting minutes

Review agenda and next meeting date

<Matt_King> View agenda at https://github.com/w3c/aria-at/wiki/August-10%2C-2023-Agenda

<Matt_King> Next meeting on Wednesday August 16

Current Test Work

Matt_King: Action Menu Button

Matt_King: Isa and Joe have completed the whole test plan, no conflicts, I think we can publish

Isa: I will publish it

Matt_King: Disclosure Navigation Menu Example

murray_moss: I will perform my testing tomorrow

Matt_King: Hadi you said you can finish soon as well

Matt_King: Currently for Disclosure Navigation Menu Example on JAWS and Chrome there are 21 conflicts

Matt_King: I wonder if we should review those conflicts before Hadi finishes up

Hadi: I think some plans are collecting too much info, For example interaction mode or reading mode. For reading mode, we drive little piece by piece, and forget our target. As we enter a list or our target, I see in our tests in virtual mode to satisfy our pass, we are requiring it read the parent item info and the item info.

Hadi: This would make a very verbose screen reader

Matt_King: I wonder if you are interpreting the test the same way, for this test when you are instructed to press the down arrow three times, when you record your output you need to record the output for each press of the key

Hadi: okay that makes sense, I did not do that, which may explain all the conflicts

Hadi: Another issue, for a test that uses the letter I to go to the next item, I think it reads to much

Matt_King: I think you're right, we had that discussion with Vispero and we decided to change the way we write our tests

Matt_King: We will be able to have one test for each of the commands, which allow the i key in this case to say less

Matt_King: I'm glad you're working on this test, when we apply the next round of changes to the app, we will want you to test it and let us know your feedback

Hadi: I have one more question. Isablel sometimes I retest something to make sure my observations are correct, sometimes I notice a change. When we are testing #25 in the Disclosure test, when I do that it says " the focus will be on the about disclousre, then in it should be in interaction mode and by default it is expanded. Sometimes when I rerun the test I see the about disclosure is expanded and sometimes not, do you see that?

Isa: Yes you need to refresh and press F5

Hadi: Yes I do that, I refresh the page and rerun the test

JS: It could be an more generic issue with the browser, or it could be an issue with a test. For that we need a specific example and an issue to filed, let us know if we can help with that.

Matt_King: I think I have observed something like that before as well Hadi, everytime thats happened, if I close the window and reopen it the issue is resolved.

Hadi: When I press insert tab it tells me all the correct info, when I press insert tab again, it doesn't announce the state, I think it might be a JAWS bug.

Matt_King: Before we move on, Isa I think we need Hadi re-record his test results right?

Isa: Yes! I think so

JS: He will need to rerun the tests

Matt_King: Hadi can you go back and rerun the test?

Matt_King: I think he just needs to rerun the test with conflicts

JS: It may not just be those, any test that includes multiple commands will need to be rerun

JS: I just reproduced the JAWS bug Hadi mentioned.

Isa: We're not catching that in our tests because we don't have you press the command twice

Matt_King: Correct, we don't want our tests to capture that, it would be a much different kind of test then

Hadi: This is a severe bug

Hadi: Though this is outside the scope of our tests, I thought this might be a good place to raise the issue

JS: I think this raises a good point, if someone collects an output that is outside the scope of our tests, Matt what should we do with those issues?

Matt_King: I think we can only report issues that are within the scope of our tests. For other issues I think Hadi you will need to report them outside this group. We are happy to help you and can pass it along, but it can't be reported in our app

Hadi: For the unexpected behavior's, should I remove those?

JS: Yes please, if its not related to the test please remove those

Matt_King: We only have 18 minuets left, we wont be able to discuss everything

Matt_King: Lets skip Update on coming app changes

New Test Format

Matt_King: These changes are moving quickly

Matt_King: First what is happening. Right now we have some csv files that are in the test repo and they define what you see when you run a test. We are going to change those files and how they are built, which will change what you see on the webpage

Matt_King: This will mean we will need to rerun every test

Matt_King: This doesn't mean that the data we have isn't valuable, it has gotten us to where we are with screen readers and getting feedback from them

Matt_King: Why are we doing this? We'll one reason is the one you raised Hadi.

Matt_King: Right now we can't have one test with command within the test have different assertions. This problem shows up in many different ways

Matt_King: Another Challenge we have, we only have required and optional for assertions. Many W3c standards have must, should and may. I think we will be in a better negotiating position if we make a distinction before must and should

Matt_King: We also go some feedback about the reports, that it wasn't clear what a partial support like 80% means. Are there parts of that pattern that shouldn't be used, but are? Our report should detail this, users of our reports would find this clarification helpful when deciding whether to use something or not at all

Matt_King: I think those are the two biggest changes

Matt_King: There are some other smaller changes

Hadi: For the must and should discussion, have people already decided which assertions should be must and which should be should?

Matt_King: Not yet, those will be ready for you to review in a couple weeks

Hadi: I'm concerned if you are going to much into telling the screen reader how to behave. If we tell one SR to do things this way, what's the difference between that and apple telling Jaws what to do?

Hadi: I think it should be the users choice, to display the order of the list for example. I don't have to order the list in the way the screen reader tells me

Matt_King: I think thats getting a little into the wild west of what Screen readers have been forever

Matt_King: There has to be some amount of agreement for the Must items, for Should there will be more room for options and room for the users choice

Matt_King: That is the essential problem we are solving with this project, what does everyone agree on?

michael_fairchild: I'm going to open a can of worms. I know our initial goal was not to follow standards because we thought that terminology may turn people away. Have the tides turned in that regard that we are turning more toward standards and their language like must and should?

Matt_King: I think there are more and more people clamoring for a standard

Matt_King: At the conference I went to in Seattle, there were three Screen reader vendor JAWS there, and they we're being dissenters. For a while there was a movement to remove should or may, but that was recently reversed. So yes the tides are changing

JS: I also want to add, that when we say something is required, we are saying the screen reader Must have something to pass a test. We are actually making more exceptions for Screen readers to create behavior's the users wants that are May have, like Hadi's example of where an Item is in a list, some people will find this helpful, some will find it verbose. We are now able to say, we think this is something you should offer, but you don't need to

JS: The aim isn't to force screen readers to all act the same, but to provide more feedback to vendors, which is based on feedback from them

Matt_King: I'm so excited about these changes, James I made a bunch of changes and comments yesterday.

Matt_King: This is coming, we can talk more about it next week, there are links in the agenda with more info to issues

Matt_King: Okay thats as far as we can get today, good discussion

Matt_King: Thank you everybody

Matt_King: James I'm trying to wrap up this spec by Monday to send it to Bocoup

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: Hadi, JS, michael_fairchild, murray_moss

All speakers: Hadi, Isa, JS, Matt_King, michael_fairchild, murray_moss

Active on IRC: Matt_King, Sam_Shaw