Meeting minutes
New PR Triage
spectranaut_: no new PRs this week
WPT Open PRs
TPAC planning - very draft agenda
spectranaut_: to remind everyone, if you go to the ARIA wiki, there's a section called F2F (face to face) meetings. Page has all the information about TPAC/Japan you will find on this Wiki page
spectranaut_: ...we created a Google spreadsheet for arrival times at airports and what hotels we're staying at; if you have the ability to help blind members, it would be helpful
… for every topic, like previous TPAC agendas, we (Valerie/James) would like a lead to introduce the topic, have focused agenda and outcomes for the meeting. For example, the session on accname & name from content, so we need a lead for that
<front-endian-jane> present_
spectranaut_: times in the agenda are times in Japan (we could add a link to a date/time website for figuring out local timezones)
… Japan Standard Time
keithamus: can I move my session to another time (e.g., Monday)?
spectranaut_: Yes, I'll make a note
… you can get back to me on whether there's someone remote that wants to attend (and when is good for them)
sarah: I added another session proposal; Keith may be interested in helping lead that. Related to updates on OpenUI and invokers
keithamus: it might be one and the same; the session around CSS anchor positioning is related. We could turn this session into OpenUI interest for popovers (and that kind of stuff)?
spectranaut_: great
keithamus: we'll go with your (Sarah's) issue as the umbrella issue
spectranaut_: should I slot the OpenUI discussion somewhere?
keithamus: it can stay put (on Tuesday) if everyone's OK with that
cyns: is it possible to have Matt's session as the last session (Monday)?
spectranaut_: yes, that's fine
lola: a few weeks there was an issue for this workshop (#2629), around a11y interop. In the minutes, it says this issue should remain ARIA-AT. My understanding is that this session is not about interoperability testing and a11y interoperability testing, it's simply about ARIA-AT. If yes, I would like to interrogate that a bit
Matt: that is related to ARIA-AT+ which includes ARIA-AT overview as well as other content. Daniel and I haven't finished mapping that out
Daniel: I will follow up with you
lola: is the other content ARIA-AT? Based on the sessions that revolve around ARIA-AT, and ACT (Accessibility Compat Data) are both moving in similar directions. I think this should be a session to discuss interop projects going on at the moment, and collaboration opportunities, duplication of work, etc. Based on the thread happening in a11y interop, it seemed like the discussion was only about ARIA-AT
Matt: won't be just about ARIA-AT
lola: I'll wait to hear from you and Daniel, and overlap between ARIA-AT/ACD
spectranaut_: is this a topic for discussion all a11y testing going on right now?
Matt: we had a to-do item to follow-up; don't want to end up with a session without a clear agenda/outcome
spectranaut_: I think we should figure out points of collaboration with testing, I was looking forward to having this discussion at TPAC. This might two different sessions; we also have the workshop on ARIA-AT later in the day as well
lola: on Cynthia's point, could that session be scheduled for a Europe-friendly time? I'll be there in-person but there are people in Europe who should come to that session. I'm also chairing other sessions that are Monday Europe-friendly times
jcraig: I worry about having the overview sessions later in the week; ideally we have those first and do the detailed sessions that follow that
lola: is it possible to move both of these sessions to Tuesday?
spectranaut_: not at a Europe-friendly time
… if the only Europe-friendly time is the last slot of the day, no way to make this work
lola: Cynthia is one of the key people I want at the session
cyns: I could do the 5AM session
… or could we do the overview at 15:30 on Monday, and workshop on Tuesday?
spectranaut_: no, doesn't work for Lola. We'll move the session to Monday (5AM Europe time)
<chrishtr> w3c/
<chrishtr> w3c/
chrishtr: I requested two slots for discussion items; first one is snap to activate and a11y implications. The second is a menu element, and need help identifying mappings. We have reps from Google to help with those
keithamus: this one is going to need a fair amount of time (in particular the new menu element)
spectranaut_: we'll figure out how much time is needed with the agenda
chrishtr: for snap-to-activate, might not need more than 15 minutes
spectranaut_: Sarah, was this part of your list?
sarah: no, but I can add it; the menu element one is on the OpenUI agenda
spectranaut_: we'll give it its own section
Lucas: for invokers/popover stuff, I'll be participating in person so happy to discuss this there. I hope others from Google are attending, but I'll be there for the a11y stuff
sarah: some of my goals for the invokers stuff is not stuff that will get fed back to HTML; AT behavior hasn't been directly explored
Lucas: agreed
jcraig: for Lola's agenda item, it was Matt's original issue about a11y interoperability. Initially, Matt filed it specifically as an ARIA-AT thing. Daniel/James on a call decided it would be a big overview but Matt's agenda was specific enough that it needed an ARIA-AT meeting
… please edit my comment/strikethrough if needed; Matt's should go back to being an ARIA-AT specific issue
… happy to leave time for other ones (such as ACD). Would love to see a more detailed overview and test run of the Acacia stuff. Rahim/Jamie might want to talk about the WebDriver implementation stuff in a separate session
… I'd like to lead an overview of all the folks working on testing/interoperability, and how they're related. The overview session shouldn't be especially long (such as an hour max). Could also do a 20 minute overview, and then 40 minute discussion
Matt: when I first filed an issue, Daniel and I talked ahead and wanted it to be more than just ARIA-AT. When we had that first discussion in this meeting, we extended it to an 1.5 hours so we could cover the other related testing approaches in the same session and then we would have about 30 minutes to arrive at decisions/next steps. I think we should take this offline, amongst the folks involved, how we want to structure the time
spectranaut_: I'm not sure if we want to have a separate item for this, will follow up via email or Slack
<pkra> I have to drop off early today. see you next week!
jcraig: could they be back-to-back sessions?
Matt: would prefer hashing this out via email, and we might need a meeting. Valerie or Daniel mentioned using the Tuesday (planning meetings)
Daniel: we should be mindful that this session is in collaboration with the Accessibility Guidelines (WCAG) group since it overlaps with their testing efforts. We wanted both groups together so it may take longer to move this meeting
jcraig: if it's a joint meeting, I'd prefer to have the agenda be bookended
Daniel: let's follow-up offline
jcraig: we could potentially do the overview session before lunch, have it be an hour, and then the discussion with AG. Then, the detailed meetings for testing approaches
<cyns> I have to step away for a minute
spectranaut_: we'll take this offline
… please reach out if you have anything else that you'd like to check in about with the agenda. Lots of to-dos from our conversation today so we'll follow up on all of them. Should have the next draft of the agenda by next Monday October 27
Should the default value of aria-live be undefined instead of off?
Rahim: cookiecrook and I spoke - makes sense to me. 2 action items will be updaing the IDL reflection missing value default to undefined and specifying in the aria spec that there will be undefined, off, polite and assertive
Rahim: just curious if folks agree
<Adam_Page> +1
jcraig: that reflected the hot take in last weeks meeting
mattking: +1
<front-endian-jane> +1
mattking: getting the spec to have a clear defintion of off would be good
jcraig: timers etc. are the usage
matt: that expectation is not well conveyed in the spec
jcraig: think it might be there in prose
jcraig: need that interoperable behaviour
Rahim: the end user impact: people are using aria-live=off as per spec. is there a web compat issue in making this type of change
Rahim: webkit - would need no changes
<spectranaut_> ack +
jamesn: my suspicion is that everyone is doing this anyway
jcraig: we should be asking the webcompat quetion every time we make a change - the only place this would be suprising is in cases of things that are defined with live regions that don't have a specific content attribute
jcraig: we should perhaps add a new value for low priority live regions
jcraig: patrick lauke pinged on a thread to ask how to clean up live regions. This is just one of many ways.
<chrishtr> Note: Chromium already defines an undefined aria-live as null (not reported to the browser's AT)
Adam_Page: wanted to bring up Patrick's issue
Adam_Page: would be happy to work on the synonym
<spectranaut_> ack?
<aardrian> check...
spectranaut_: the aria-actions PR doesn't always load (folks chiming in that they're experiencing the same)
feat: aria-actions addition to the ARIA spec
Matt: we've been doing some work in APG related to this; Sarah, I'll send you the PR in APG that Curt is working on. We have questions about expectations about how things are supposed to work, such as the accname changes. We essentially are not sure how we can use that feature of the spec
sarah: I'll find it
Matt: lots of tricky questions related to focus so hoping to get the very specific questions with examples of behaviors added to the aria-actions PR so we have it in context
sarah: should I bring this to TPAC again?
Matt: it might require more discussion if we have time at TPAC. The things that we're running into with the Chrome implementation and the JAWS implementation, as part of APG, there are practical open questions that the new spec fully addresses
sarah: Matt/Adam, will you be at TPAC?
jcraig: we could schedule an adhoc meetings ahead of time? If we could work out separately, that would be great
jamesn: plenty of time on Thursday/Friday
jcraig: is the primary issue focus vs. click? I think I included a reply in that (non-loading) PR. Would love to discuss further with you
Matt: would also like to talk about the touch/cursor experience
spectranaut_: feel free to self-coordinate on discussing these topics
Expose implicit ARIA semantics (browser-defaults and ElementInternals)
benbeaudry: wanted to get a quick opinion on this from the ARIA group. Essentially, we have a bunch of a11y testing frameworks like axe-core, Deque, etc. and MS Insights is based on checking the a11y attributes of DOM nodes, and those tools figure out the errors developers made. Need to know computed values at the end conveyed to AT; could be used as extension, bookmark, script
… need a DOM API to get all the changes in a11y that were made to all DOM elements. When author uses role attribute, for example, we can fetch that value because it's a public API however, starts to break down for attributes set on ElementInternals (used for setting a11y attributes on custom elements). Uses private encapsulation so once set, it can never be read back
… lots of custom elements use ElementInternals so can't test the values (don't even know that attributes are set on them). WHATWG tried opening up ElementInternals (wasn't successful), and they suggested working with the ARIA WG
… good back-and-forth with Jamie Teh around not creating a new API, but rather create something specific ElementInternals without opening up ElementInternals. Do we want to consider this problem of not being able to read all of the attributes set by authors? Or should anything set by authors could also be read by authors as well?
<aardrian> i have to drop, but am excited for the topic we didn't get to (namefrom heading)
spectranaut_: please respond in this issue to discuss for the next meeting
jamesn: let's roll to the first agenda item next week