Meeting minutes
<joanie> agenda: be done
<joanie> s/agenda:be done//
[New Issue Triage](https://bit.ly/aria-12Aug-issues)
<joanie> https://
<joanie> Reflecting Element References
joanie: we are going to talk about this later today
joanie: we don't know if this is 1.2 yet
<joanie> https://
<joanie> Fix prose description for where doc-footnote is allowed
joanie: this is basically handled, we had a cfc to have a first public working draft (1.1?), a google implementer had issues, matt garish has a fix. We are going to go with the originally approved draft, but immediately publish an updated working draft. We will have to call for consensus for new stages or phases. Updated a working draft is the least amount of hassle
<joanie> https://
<joanie> Other menuitem "widgets"? menuitemslider and menuitemspinbutton
joanie: soooo 1.4?
james craig: I don't know what a menuitem spinbutton would even look like. Can we ask for screenshots or link to an example
james craig: I'll comment and ask
<joanie> https://
<joanie> Role-specific defaults should be addressed as reflectable in the idl or in normative prose
joanie: we discussed this last meeting. there has been discussion on the issue. should we talk about this during the aria reflection issue at the end?
james craig: 1.3
[New PR Triage](https://bit.ly/aria-12Aug-prs)
<joanie> https://
jon gund will edit, only one test needed
<joanie> https://
joanie: we need to read and think about this for a deeper discussion
james craig: the diff is very smaller
james craig: I think its a bad idea to have the id be the numbering structure, given that we reorder them from time to time, so it's not good for a permenant link. I can make a separate issue for this.
<joanie> https://
there is confusing language in the dpub spec, as discussed in the issue above
<joanie> https://
joanie: I'll review this
<joanie> https://
1591, 1592, 1594, 1599: these PR are validator tests and need to be reviewed by people working on tests
joanie: can spectranaut review?
yes!
joanie: I'll add spectranaut as reviewer to jon's other tests
[Meaty Topic Meeting Reminders](https://www.w3.org/groups/wg/aria/calendar)
joanie: this is a reminder for everyone
<joanie> https://
joanie: on the 19th we have this meeting ^
<joanie> ARIA Deep Dive - ACCNAME and hidden content continued
Joanie: we need aaron, james
cyns: I'll tell him
<joanie> https://
joanie: on the 26th we have this meeting ^
<joanie> ARIA Deep Dive - Layout Grids
<joanie> https://
the agenda includes the related aria github issue
[Meaty topic(s) for September](https://bit.ly/aria-meaty-topic-candidates)
joanie: options: updates to content editable and tab index
joanie: or secondary action on composit ewidget roles
joanie: or role=tree on mobile and touch
cyns: can we talk about that one?
<joanie> https://
<joanie> Should ARIA provide better support for routing in single-page applications?
joanie: is melanie on the call? she was changing affiliations and should be back in the group
<joanie> https://
<joanie> Should dialog/alertdialog be modal or non-modal by default?
cyns: we could probably make progress on this, so we should schedule it soon
<jcraig> Closing the loop on my action. I included that comment on #1210 in March 2020. https://
joanie: dialog/alert dialog, should this be a topic after the next two?
seems like jcraig and cyns agree
joanie: september 2nd for 1210 (dialog/alert dialog)
Action: joanie to set up Sept 2 deepdive: https://
<trackbot> Created ACTION-2162 - Set up sept 2 deepdive: https://
<joanie> AccName trims whitespace but doesn't define which code points are whitespace
cyncs: can we combine the trim white space with next weeks deep dive on acc name?
<joanie> https://
cyns: if doesn't get talked about next week, this is a good candidate for september 9th
joanie: there are articles that don't have headings, issue 899
<joanie> https://
jcraig: this is the facebook feed
jcraig: the article didn't have a label but when you pull up VoiceOVer, you just see "article article article"
jcraig: the hope is to show something more relevant in the rotor list than the above ^
jcraig: I wanted to hear from the working group the best way to differentiate these
jcraig: rather than it be left to the implementors to decide
cyns: if eric can come I'm less necessary
jcraig: this doesn't need a deep dive, it's in ARIA, but maybe it should be in acc name
joanie: its not the same as acc name 132
<joanie> https://
joanie: how can we handle author errors?
jcraig: not considered an author error to not include a label on article
jcraig: because it's not provided there is nothing to display
jcraig: we can reach into the article
jcraig: this is more about flexibility in find the label/description where it is not required
joanie: is the conclusion no deepdive?
jcraig: actually it could be
jcraig: we discussed drafting a PR, but we should have a PR before the deep dive
joanie: I don't think acc name needs to be parallel with aria
joanie: it can progress at it's own rate
jcraig: I'll try a PR
joanie: next week can schedule one of these, because we 3 scheduled already
cyncs: next time lets go through these in reverse order
[Prohibit presentation / none on where it causes conflict resolution](https://github.com/w3c/aria/pull/1494)
ARIA Reflection
<cyns> https://
cyns: I put the relevant bits in this issue
cyns: currently, the aria mix in IDL only include simple aria values
cyns: like strings
cyns: aria-activedescentdent and describedby are not include, id refs or arrays of idrefs
cyns: there is a consensus on how to reflect this properties. I want to know if everyone is ok with this
cyns: at a high level, it allows IDLs that can reflect these. it uses a frozen array. it takes a snapshot of the items in the array
cyns: some edge cases that are unlikely related to triple equals are considered to rare to be concerned with
cycns: the recent change "not sprouting these attributes" allowed us to get consensus
cyns: I like the sprouting of attributes, but its more important to have a way to reflect them
jcraig: it's sometimes impossible to reflect them in strings
jcraig: aria-labelledby elements property will never generate an aria-labelledby dom reflected content attribute
jcraig: only strings, it makes it easy to understand from implementors to authors
jcraig: if aria-lablledby content attribute was in there, but then you update the aria-labelledby attribute, does the content attribute go away?
jcraig: because it is stale
jcraig: should I put in aom or aria issue
jcraig: I'll put in aom
cyncs: does anyone hate this?
jcraig: the issue here is that if you need to label something by something in the shadow root, you need the new interface, and if you looked at the dom tree you won't see anything
jcraig: you have to use the dom interface by checking the attribute
joanie: the accessible tree will have the correct attributes by default
cyns: I head no objects, so I assume there is approval
joanie: what about the nullable strings?
<joanie> https://
joanie: my understanding from trial and error and testing, all user agents act like it's a nullable string. they don't set defaults
jcraig: I put in the request to say everything is nullable, anna said no
jcraig: another person I cc'd in the comment has information about this that is necessary
joanie: if we are allowed to make this change back and magically we have three implementations
joanie: it would be worth pulling aria 1.1 out of CR, making change, and popping back in
jcraig: I'm fine with that.
jcraig: the diff is adding the question marks back in to the string reflective ones
jcraig: some of the other idl issues are related to non-string data times. like the token values or token lists (floats or doubles), if we want to claim that we have implemented the right thing and not risk a backwards compatibility change, then we should do these at a same time
jcraig: we should do on any of the ones that will remain strings, it's safe, but if we update the non strings this would be hard to change in the future
cyns: will we get pushback from html5 because numbers are represented as strings?
jcraig: this is already an issue
jcraig: lets do the easy "?" change on strings
cyns: token list behave as nullable strings
cyns: as far as we can tell the current implementation accidentally match the spec
cyns: we should make the specs match the implementation is the implementation are in agreement
jcraig: all numbers are implemented as strings
jcraig: lets do the easy change because it's already implemented, we can change it later and deal with it
jcraig: the only risk is that someone will complain later
cyns: the html5 working group will have to accept because all implementors already do it
cyns: I will take it
joanie: I think we will have to change some links.
joanie: in the description text
joanie: because it currently says things about default vlaue
joanie: I think we need to do this, but wait until michael cooper and james are back
Action: Joanie to file issue about IDL and pulling ARIA 1.2 out of CR
<trackbot> Created ACTION-2163 - File issue about idl and pulling aria 1.2 out of cr [on Joanmarie Diggs - due 2021-08-19].