W3C

– DRAFT –
ARIA and Assistive Technologies Community Group Weekly Teleconference

21 June 2023

Attendees

Present
IsaDC, James_Scholes, jugglinmike, Sam_Shaw
Regrets
-
Chair
Matt King
Scribe
jugglinmike, Sam_Shaw

Meeting minutes

Status of current test run work

Action Menu Button Example Using aria-activedescendant

James_Scholes: For JAWS and Chrome, we only have one Tester, that's Isa

James_Scholes: It would be nice if the queue displayed the "date added" because it's difficult to determine what's new

James_Scholes: we should clean up the test queue

Color Viewer Slider

James_Scholes: For JAWS and Firefox, we have two Testers and a number of conflicts

IsaDC: we discussed this last week, the conflicts are related to Hint text

Matt_King: Can we clean this up by running as Hadi?

James_Scholes: Possibly; we just need to determine whether there's an underlying issue to be reported

IsaDC: testing is complete; there is just the concerns with the "hint" text

James_Scholes: We discussed this as editorializing last week. We can resolve the conflicts

Modal Dialog

James_Scholes: Only VoiceOver for macOS and Chrome remains

James_Scholes: all testing is complete, there is just one conflict

James_Scholes: It's regarding whether the role "dialog" is conveyed. IsaDC marked this as "no output", but Dylan marked this as passing, even though the output doesn't mention the dialog.

James_Scholes: The word "dialog" appears in the page title, though, so this is likely why Dylan classified this as "passing"

James_Scholes: I will update Dylan's input

Navigation menu button

James_Scholes: For NVDA and Firefox, all testing is complete, and there are no conflicts. We can move this plan forward!

Test Queue Cleanup

Matt_King: I think it would be really nice if we kept the test queue focused on the tests we actually plan on working on currently.

Matt_King: That is to say: we shouldn't keep a backlog of work in the test queue because that makes it difficult to maintain

Jame_Scholes: I think that makes sense

IsaDC: Agreed

James_Scholes: I think some of these things were added to the queue somewhat aspirationally

James_Scholes: Let's start by removing the final instance of Disclosure Navigation Menu Example for JAWS and Chrome. That was published in 2021

Matt_King: Done

James_Scholes: I can go through and clean things up--the outdated things that aren't published yet

Integrating automation

https://docs.google.com/document/d/10681gMmTM2KmVEw_Je-A4nmRvcqIkB6jgtJ8dt_Pp0M/edit

Matt_King: This design document is fairly technical. What I'd like to discuss here is a little higher-level

jugglinmike: Apologies these documents are tough to navigate with all of the comments and edits

jugglinmike: I think that the issue Matt is talking about, that is most important to consider, is which role is responsible for initiating automated testing

jugglinmike: Right now, looking at the section titled Workflow proposal, the way its written is "When drafting a Test Plan, Test Admins may assign the new Test Plan to a user named "Response Collector Bot." This user represents the automation system."

jugglinmike: The reason behind using a fake user to do this, instead of a term like dedicated UI, is it limits the UI we have to designs

jugglinmike: This would achieve two requirements i have, 1. We only selectively collect results 2. This allows us to prevent a potential confusing state, in which a person is assigned to run a test that a automated machine is running also, how do we do that without the risk on conflicts?

jugglinmike: To bring it back to what Matts wants us to discuss, is should we give individual testers the ability to initiate automated testing?

Matt_King: I imagine a initial use case for automation is for a human tester to use the bot to compare against and confirm their results

Matt_King: I'm not sure how fast this thing will run, but if I imagine opening multiple tabs for each test, and the tester could use the bot as much or as little as they want to help them

Matt_King: We can train everyone on the nuances, but I thought this might be a case that makes it simpler to have users be able to initiate automation

jugglinmike: I'm not ready to comment on that, but one risk is that to the extent the system is flaky, is that automation will be in development, it wont be perfect, it will go through various iterations. A tester would need to be aware of this potential for bugs

JamesScholes present+

Matt_King: We can continue this discussion on Monday in our Automation meeting

Minutes manually created (not a transcript), formatted by scribe.perl version 210 (Wed Jan 11 19:21:32 2023 UTC).

Diagnostics

Succeeded: s/"help" text/Hint text/

Maybe present: Jame_Scholes, Matt_King

All speakers: IsaDC, Jame_Scholes, James_Scholes, jugglinmike, Matt_King

Active on IRC: jugglinmike, Matt_King, Sam_Shaw