Meeting minutes
<WilliamTennisNFCU> It seems that I have lost the privileged ability of joining the video conference
<WilliamTennisNFCU> I can log in, but it says I don't have privs to the zoom
<WilliamTennisNFCU> 403
<pkra_> https://
<WilliamTennisNFCU> I dialed in
[New Issue Triage](https://bit.ly/3iilYxi)
https://
scott: I'm happy to do this for 1.3
scott: I don't want to go through the issue but I see three different outocmes
scott: we can talk about it at the end of the meeting or next meeting
jamesn: lets put it in next week's meeting
jamesn: the other two issues are related to aria solutions for ruby. the use requirements document in 16.20 covers all the use cases, but doesn't go through potentials to solve. we are scheduling something for TPAC week to discuss. Please look through this if you think it might have impact on implementations, or accessibility APIs, so we can get hte right people to attend.
jamesn and jcraig seem to have ideas already. jcraig: you can check out most of the ruby test cases in the Apple Books app and maybe other readers. Ruby layout is commonly used in in Japanese ebooks.
jamesn: I don't think these are milestonable
[Deep Dive planning](https://bit.ly/aria-meaty-topic-candidates)
jamesn: should we skip and move to discuss TPAC?
jamesn: no proposals
jamesn: no deep dive next week
[TPAC 2021](https://github.com/w3c/aria/issues/1482)
jamesn: cynthia proposed a lot of talks
jamesn: any other ideas? otherwise we will schedule these meetings
[Updating ARIA 1.2 due to IDL implementations (exit and re-enter CR?)](https://github.com/w3c/aria/issues/1598)
jamesn: this is blocking us from getting the new version of 1.2 out
jamesn: I don't know what to do
jamesn: it seems we hit a block where html is stating that what we have is not a thing, the thing they want us to do is not implemented
jcraig: the question I think is that right now the implementations do it this way, if you remove the attribute and check the reflected dom property it returns null. dominic said this can result in a null pointer exception in the IDL. the IDL doesn't have a way to represent what is currently implemented in browser? Or, we are specifying it the wrong way. So what is the right way? I could use help from joanie and aaron
jcraig: IDL doesn't allow you to do what you are doing. But we do have it implemented, because all browsers do the same thing here. Is IDL insufficient?
jamesn: IDL allows null domstring
jcraig: HTML's implementation of relfection doesn't allow it to be used with a nullable domstring
jcraig: but the implementations are clearly doing that
jamesn: should we not use the term "reflection"
jamesn: joanie said we should call it mirroring
jcraig: we could file an issue against the HTML spec
jamesn: but we probably won't get quick resolution
jamesn: the only alternative is pulling the IDL section
joanie: the mirroring comment was snarky
joanie: we have three implementations, and we shouldn't pull it
jcraig: do you think they will file a formal rejection?
jamesn: if they use the term reflect they will
jcraig: if we use a different term we would have define the new terms
joanie: aria reflection is based on IDL but here is one difference
jamesn: aaron, james, joanie and cynthia -- can any of you sort this out?
jcraig: I have to chat with other webkit people to see if there is a path to move forward on this
jamesn: we are a year+ late, this is preventing publishing
jamesn: this is the number 1 urgency from aria
aaron: can we meet in person?
jcraig: possible solutions: (1) remove IDL section, (2) ship it as is (and might get a formal rejection), (3) figure out out to change it, but we aren't sure how to change it to keep the section in and make them happy
WilliamTennisNFCU: I think we should email them and explain
jamesn: dominic is on vacation until oct 12
aaron: can we move it to a note or something?
aaron: its not like aria would break
jamesn: it would be in 1.3 editors draft
aaron: we might want to give is more time
jcraig: we can ship this with a note that says "we have implementations, but it is the HTML reflect has a conflict with this"
scott: I like that idea. Can we ship it is a "informative"
jamesn: I think that doesn't sound great
jcraig: we can't wrap a "infromative" about a "normative" section
jamesn: I'm happy to help arrange things but I defer to people who actually know about IDL
jcraig: I'll commit to chatting with people in webkit who are closer to the IDL implementation
jamesn: can aaron do the same with google?
jcraig: within apple, I'll chat with people to see if they agree with Dominic and sasha and the others in the issue
jcraig: or if they feel this is an issue that needs to be resolved in html
aaron: lets put a general message on the thread he responded to
aaron: to dominic, because he is on vacation, so he doesn't have to reply
[summary element role mapping](https://github.com/w3c/html-aam/issues/345)
jamesn: scott filed and issue, the fact that summary element doesn't have a role mapping.
jamesn: the current mappings to html are kind of broken, summary is mapped to button, and summary allows child content within a summary element. button mapping does not allow that
scott: this has been updated in the last year. it used to allow for fewer child elements, but now you can shove anything in a summary now. if you have an input inside of a summary, in chrome, if you try to put a space in the input it will open and close the summary because it is triggering the button response
jamesn: this comes back to an aria issue where people need to put extra elements within various aria elements that don't allow
jamesn: do we want to add another aria mapping in html aam
jamesn: or do we not want to do that for now? or should with get the html spec changed and not allow this?
scott: I'm open to suggestions. this has been brought up in html before, and there has been no change
scott: maybe implementations need to change, there is two bits to a summary, there is a part where content can be added and there is a button
jamesn: that seems like a reasonable implementation
jamesn: chair hat off, I don't want to try to prohibit things that people need to do. The fact that people are doing this means people need a way to do it in an accessible manner
WilliamTennisNFCU: can we compute the role dynamically based on what is inside of the element
scott: you mean compute the role of the summary element based on it's children?
scott: that could potentially solve one problem
scott: but what is it when it has nested elements?
WilliamTennisNFCU: if it is a button inside of a expand/collapsed thing....
WilliamTennisNFCU: if you put a button inside of a summary element, it will remove the summary role from the summary element
scott: because of the inconsistency with what is being done right now, this is mouse accessible, sometimes keyboard accessible, if we take away the button role, then how could someone using an assistive technology open and close the panel?
scott: your suggestion makes it broken in a different way, but still broken, like it is now
aaron: I think you made changes in chrome to make some of these situations not completely break, right?
aaron: for buttons, we only truncate the descendants if it is a single text child
aaron: not it is the case for all aria roles that have aria-presentational
jamesn: how do the users interact with those?
aaron: it is the same as if you put tabindex=0 to the containing div
aaron: then you can tab inside
scott: virtual cursor will move pass these things after one stop
scott: but if you tab explictly you can frequently tab in
scott: but if you press enter key inside of it, only one element will fire. a link inside the summary and send enter, you might just toggle the summary instead of opening the link
jamesn: we need to accept the reality that this exists, and we need to find a way to make it work better than it does today?
scott: this might be a screen reader issue?
jamesn: if browser says buttons don't have children presentational, then should we remove from the spec?
jamesn: if we don't have implementations of children presentational any more, we should remove it
aaron: there is no current widget that does this
aaron: maybe this should be author advisory
aaron: but we need to add this ability
jamesn: can we leave this for now?
scott: its been like this for years, and I have been personally asked about it many times recently
scott: HTML says explicitly buttons can't contain interactive elements?
jamesn: lets have a deep dive on this, with sarah and scott
WilliamTennisNFCU: do we have an aria role "expandable"?
WilliamTennisNFCU: we might need something other button or link
WilliamTennisNFCU: we don't have a semantic for this scenario
scott: tangental to this, I have had convos with people about how accordions are set up as heading with buttons, and maybe headings need to be focusable if they announce they can be expanded or collapse.
scott: aria sort you need to put on column header, but changing value doesn't do anything, but you have to move away and back to hear the new state, could column headers be focusable elements
scott: we need some way to make otherwise static elements interactive when they are interactive
jamesn: we have has this discussion many times
StefanS: I have brought this up a number of times
jamesn: there is always a lot of pushback from various folks, but we should continue them. the more applications you encounter, the more you see the need for these things
StefanS: when are we going to need to start talking about this redesign sooner than later
StefanS: I would need help with other industry contacts, who shares the same pain
Zakim: next item
[When is hidden content taken into calculation of name and description?](https://github.com/w3c/accname/issues/57)
jamesn: looping back on the hidden content conversation?
jamesn: we were going to try to get some examples together?
aaron: I had a call with ??, this doesn't solve firefox problems, firefox has a special implementation
aaron: chrome, I came to the conclusion we don't need this, we can use the existing algorithm and simplify the code a low
<jcraig> Meet of the discussion is in the PR rather than the issue. https://
aaron: I don't know where webkit is
aaron -- i'm having trouble scribing your summary
jcraig: if this is something like innertext exposed to web authors...
aaron: no this is an implementation detail
aaron: very slight modification to innertext code
aaron: chrome doesn't need changes
jamesn: what about if something is visibility hidden, we run through the innertext thing and we will get weird results?
aaron: I'm just concerned that exisiting content is broken
aaron: it's not broken for most cases
aaron: if aria-labelledby, 99.9% you are trying to hide the text from sited users
aaron: and that works
jamesn: what about an intermediate solution, if you are pointing to node that only has text children, use text children, otherwise use innertext
aaron: innertext does not get alt text
RRSAgent: make minutes