Meeting minutes
Integrating with BTT
Matt_King: We have a plan to incorporate the ARIA-AT proposal into the BTT's new charter
Matt_King: BTT's charter will be updated in August. Typically, these groups work on their charters for several weeks in advance. That's why we're starting talks now
Matt_King: We're getting our ducks in a row
michael_fairchild: Is there an appetite in that group for this?
Matt_King: We've received some signals
<jugglinmike> : The charter is set to expire on 2023-08-10
<jugglinmike> : Is the charter renewal the only time the group could adopt a new specification?
Matt_King: That's not clear. I suppose they could renew/refresh the charter later on, to include a new spec
An update on automating VoiceOver in macOS
mzgoddard: I haven't been able to get Apple's new API to function as documented
mzgoddard: I've theorized that this may be because we have not yet attempted to build with a property certificate
mzgoddard: I worked with jugglinmike to create such a certificate, but so far, applying that certificate to the build process has not been fruitful
mzgoddard: Separately, I've been looking into the e-speak application as it is a real-world example of software which integrates with the new API
mzgoddard: I've installed that software from the Apple App store and witnessed it working as expected
mzgoddard: However, I have not been able to build the software from source in my development environment
mzgoddard: So I'm still trying to learn which steps we need to adopt in our own work
<jugglinmike> : The support request I filed with Apple has not progressed at all.
Matt_King: How much energy should we put into getting the demo application to work? At what point does this become a blocker on our critical paths?
mzgoddard: I think we can contact Apple, now
mzgoddard: There are things I can keep trying, but I don't think they will deepen the conversation we can have with Apple
Matt_King: My high-level feeling is that you're really reaching for solutions
mzgoddard: That's a fair assessment. I think looking into e-speak still has potential (for instance, trying to work with it in a cleaner environment than my development environment), but it's a lot of murky/gray areas
mzgoddard: Maybe less like "grasping at straws" and more like "reading tea leaves"
Matt_King: I think we should ask Apple so that you can spend your time more effectively
Matt_King: Since we're shooting for automatic results generation in the second quarter
Matt_King: Of course, we have Windows support, so we can move forward with just Windows
Matt_King: I think moving forward like that would be an acceptable risk, given that there are other aspects that we need to advance
Scope of automation
<jugglinmike> : Automation causes us to sub-divide the concept of "test results" into two parts: "AT output" and "interpretation of correctness"
Matt_King: Yes, that tracks with my thinking. We may need to adjust our language a bit--specifically when it comes to "running a test". Does that involve both "collecting output" and "interpreting output"?
Matt_King: We also have to remember that output could be speech, but it could also be braille or any number of other things. It's generally, the "AT response"
Matt_King: The analysis is "Does the AT response conform to the expected behavior?"
Matt_King: Maybe, "AT responses" and "assertion analysis results", and that while these could be generated by different "actors", they are part of the same "execution"
Matt_King: A human may second-guess an AT response and decide to collect the response manually
Matt_King: I think we'll always have a human running every part of every test at least once
<jugglinmike> : We're out of time. I will write up an issue that outlines the topics and issues we still need to discuss, and I'll create a "one-off" meeting to discuss further at this time next week