Meeting minutes
[New Issue Triage](https://bit.ly/3qI01wB)
https://
peter: ancient
jamesn: sometimes you do need to do this, right?
jamesn: the things that required counts, like a posinset, if everything isn't at the same level in the tree, or not a direct child in some browsers, so you don't get the count correct... so intermediate spans and divs are a problem
<aaronlev> Yes, it's possible you'd need it (sorry can't call in right now)
jamesn: there is an issue to resolve this, and to ask browsers ot ignore intermediate generics, but not done yet
jamesn: we don't need to do this in 1.2 because it has been here for years
cyns: 1.3 good first bug?
jamesn: I think its more big/complicated
peter: the examples are bad
jamesn: so at least we should fix the examples in 1.3
peter: seems editorial
peter: because it is just an example. I can do it
https://
jamesn: scott can we close?
scott: yeah
https://
jamesn: maybe we should wait for aaronlev to weigh in
https://
peter: I think this is an important convo
jamesn: 1.3?
jamesn: let's schedule a deep dive
jamesn: and I'll agenda it
scott: this should be talked about in regular meetings because it's hard / maybe impossible for me to join deep dive
<jamesn> https://
jamesn: this seems related
jamesn: lets link them
peter: did discuss bring dpub into core spec?
jamesn: I think we talked about bringing some roles in, but we decided not to
[New PR Triage](https://bit.ly/3noOfFh)
https://
jamesn: joanie can you review?
joanie: ok
jamesn: we will discuss later
[Deep Dive planning](https://bit.ly/aria-meaty-topic-candidates)
jamesn: dec 9 for aria hit point
jamesn: issue 788
cyns: I have a conflict on the 9th but maybe I don't need to be here
[CFCs for ARIA 1.2, ARIA-Practices 1.2](https://lists.w3.org/Archives/Public/public-aria/2021Nov/thread.html)
o/
<MarkMcCarthy> 🎉
<Jemma> 🎉🎉🎉🎉
jamesn: positive notes are great! if you have objections, post
jamesn: practices closes today 5pm boston time, aria closes tomorrow 5pm boston time
jamesn: thanks everyone for all of your work!!!
joanie: the idl stuff I don't know how to write tests for that
joanie: there is one on the backburner for posinset and set size, I was working on implement it in chrome, and should browsers calculate it when it is not specifically provided
mcking: right now says how browser should behave, not in a normative statement, there is an expectation that it should be done
jamesn: there is 4 things listed "at risk" that only have two implementations instead of threee
joanie: also there isn't microsoft test results
jamesn: joanie said they aren't really at risk because they are easy to implement
jamesn: how to get windows test results?
joanie: if jon gunderson still has tool to do it, that will be great
joanie: if nothing else I can do test manually
jamesn: scott, do you know how to get test results
joanie: in the past for getting test results for microsoft, microsoft had a hacky tool for running manual tests
joanie: microsoft would deliver tests
scott: i will talk to melanie richardson after this meeting, she is not at microsoft but might know abou tit
cyns: joanie email me what to ask jon
[Clarify how "otherwise interactive" relates to overriding the none/presentation role](https://github.com/w3c/aria/issues/1628)
jamesn: has this been superceded by jcraig pull request?
jamesn: should we drop this? if an element is focusable, drop presentation
mcking: this otherwise interactive, or focusable, this is all about recovering from author unintended negative consequences of making something role presentation when they shouldn't
jamesn: define when they shouldn't
jamesn: what does it mean to make something interactive
mcking: the consequence of removing this is...? our goal of having the phrase there is to have the browser ignore role presentation in certain situations
mcking: when it would be negative to user
jamesn: but we never defined what that meant
mcking: I'm all for the simplification, unless in practice, browser's are ignoring role presentation?
mcking: if so the consequences might be bad for users if we remove this
jamesn: if an image has a title or tooltip, role=presentation is being ignored
mcking: that is not exactly role presentation
jamesn: but it is similar
mcking: I think an image is different than a div
jamesn: a leaf node vs container node
mcking: now we are back to where we left off last week
mcking: jcraig said something about this
(I missed it ^)
mcking: we need jcraig or aaronlev
jamesn: maybe we should move to next meeting
jamesn: I think we shouldn't approve PR until we discuss this
<melsumner> if implementers approved it we should be okay to merge tbh
mcking: we need firefox's ok too
jamesn: we need a comprehensive def of what we mean by otherwise interactive
aaronlev: I think we should probably leave it as just focusable... but I don't know what jcraig is thinking, like take it out and explain it better later
aaronlev: yeah we need james teh
<jamesn> https://
jamesn: there is an associated issue filed by jawstest, aaronlev I would like your comment
jamesn: I personally disagree and think it should be closed
aaronlev: I need to spend time thinking about it
aaronlev: I think we use focusable in chrome
aaronlev: but how hard is that to implement?
jamesn: we do focus for aria-hidden
aaronlev: at the moment, chrome, if it is focusable at all.... I mark it as invisible but keep in tree
aaronlev: only for voiceover support we remove it from the tree because there is no invisble
mcking: I'm not sure why jawstest would want to change it
mcking: seems problematic for me to change the tree
mcking: we should make what is presentational so volatile
aaronlev: seems to me like we don't have enough information
aaronlev: what is this going to improve? and if we change it, what new errors will occur?
jamesn: can you look and comment and maybe close, aaronlev ?
aaronlev: I will say please re-open if you have a specific reason why this is a problem for end users
[clarify img naming steps](https://github.com/w3c/html-aam/pull/322)
jamesn: scott.... where did we leave this....
scott: I think the last thing that you and I had a side convo about, instead of continuing to treat images with alt as generics or something
I think I messed that up ^
scott: I think this change will cause problems for lots of microsoft sites
scott: we need more data, two implementations following the current spec
mcking: as a user, I think this is one of the reasons that lots of people to choose not to use firefox
mcking: when you have an undesirable or extra chattiness
mcking: you will change browsers
mcking: that is right, none is more clearly semantic
mcking: maybe I should add another comment. I agree with his reasoning for wanting this, but, if the none escape hatch is not there for authors, this is a problematic change
mcking: the point that I didn't make is that doing this at all complicates the world for authors
mcking: alt="" gets rid of things, we have been told this forever
jamesn: screen reader users adding comments is more persuasive than the rest of us
<melsumner> As an author and a template lint maintainer, I really like the idea of alt="" AND role="none"
<melsumner> because if the role isn't there, I can throw an error on empty alt
[summary element role mapping](https://github.com/w3c/html-aam/issues/345)
jamesn: scott, where are we on this?
scottO: lets talk about it
scottO: because it is currently spec button, and we have problems with nested interactions in button, we can request htrml do a spec change, or another idea is to come up with a different role of this...?
scottO: it is exposed as summary for some browsers
scottO: webkit exposed as detailed triangle
mcking: I have seen it exposed as a button, collapsed or expanded
mcking: I swear i have done this a lot
jamesn: if you do have interactive descendants of a button, aaronlev what do you do
aaronlev: we only make a button a leaf if only has text children
aaronlev: the purpose to avoid double announcements
aaronlev: but we don't need to do that for any of the ATs anymore
jamesn: do you do the same for a link....?
aaronlev: if it is aria, we don't truncate descendants
aaronlev: if it is a native role with not possibility of anything interesting in subtree we will truncate
jamesn: if browsers and at can deal with this, is there a new role.... a interactive container with children elements.... that summary/details can use?
<scottO> a bit of an extreme test case, but this is valid html: https://
aaronlev: in the button case it is a descendant, in the summary details is it also a descendent?
jamesn: yes
jamesn: summary can have anything be a child
<melsumner> I have had so many devs do re-writes of elements because they used details/summary
scottO: here is an awful test case with valid html
<melsumner> but WHY is it valid HTML? that doesn't seem correct. It's a nested interactive.
scottO: this is quarky with browser/at combos
mcking: this is not a aria thing it is chrome's bugginess
mcking: in webkit it calls it a summary group
mcking: they made up their own role, whatever
mcking: is that a problem in itself?
scottO: we could ask html to clean this up
scottO: we could make the triangle a button, to separate the button from the test
mcking: oh I didn't realize you can put things in the summary itsself, not the details
<melsumner> YES THAT 100% I agree Matt
mcking: why doesn't html even allow that??????????
someone said "the devil is not in the details"
jamesn: html allows it, someone might want a summary with a link...?
scottO: there was an original allowance, but then they opened the floodgates
<melsumner> I would really like it disallowed, folks are using stuff like this for menu dropdowns :(
mcking: if you have a separate for expand/collapse.. when you expand the summary to be everything
jamesn: even aria people try to do these things, child focusables in a focusable thing
james: what if we have some kind of thing to support this?
mcking: what do you expose when the container has focus?
mcking: same problem as grid cells
jamesn: if authors want it, we should make it a feature
<melsumner> Would we be opening the floodgates for more nested interactives if we did?
mcking: the way screen readers and accessibility trees have evolved will make it hard to fix in a way that is implementable and understandable by screen reader users
RRSAgent make minutes
RRSAgent: make minutes