Meeting minutes
New Issue Triage
jamesn: all PDUB issues.... DPUB group will handle
New PR Triage
w3c/
jamesn: melsumner please merge
melsumner: needs rebase
jamesn: Scott, need more reviewers?
scotto: no, will merge
Deep Dive planning
anyone going to CSUN?
Daniel and James Craig at CSUN.. Matt OOO
keep the meeting... light issue triage
<jamesn> Meeting scheduled 10AM pacific on Monday
<jamesn> admin work - wpt is cross browser JS and python tests so engines can test
<jamesn> not much a11y support previously. one initiative has been the interop testing.
<jamesn> browsers get together and work out what they should be working on.
<jamesn> this year an a11y investigation. someone proposed tree evaluation - but that is lots more. jcraig suggested role and accname mappings. then some investigations on additional properties perhaps.
<jamesn> have posted an extenstion to webdriver - going throught review process right now. kickoff meeting on monday to work out what the success criteria will be. if interested add to web-platform-tests/
<jamesn> cyns: looking at internal browser representation?
<jamesn> jcraig: make sure engine internals get right role and label
<jamesn> jcraig: not platform specific
<jamesn> cyns: may also have found someone for that
F2F agenda planning
Booked rooms in Adobe Seattle office
survey coming in the next few days
need an agenda
a number of issues with the F2F Candidate label on them
talked about it in Vancouver
jamesn: will link the minutes from TPAC
jamesn: I think we decided we're not doing this now. Will drop F2F label
aka aria-actions
will discuss... smhigley in person
Matt_King: yes please discuss
remove F2F
<spectranaut_> jcraig: how individuals contributing aria accname and htmlaam can write automated tests, a plan for organizing tests
<spectranaut_> jcraig, can you make an issue for this and add f2fcandidate?
cyns: if there are googler issues, many are next door to the Adobe Seattle office
everyone: will review issues and tag with F2F Candidate
Matt_King: announcements would be good to discuss
jamesn: something better than live regions
cyns: some Microsoft people were working on it
Aaron: Benjamin B is taking that over
Matt_King: okay to add various "unsolved problem" to the list?
jamesn: yes
Matt_King: will raise F2F Issue and link explainer
(paged content with misnomer "virtual" content)
cyns: good time to have non-ARIA related Accessibility issues
maybe an earlier meeting for those
Matt_King: would be good to meet about popover
Cyns: those people aren't in the Seattle office. was thinking of other issues.
Aaron: I can call in
Explain how values are obtained
jamesn: seems to make sense in accname
Bryan: some cases where there is no better way to define than have a value algorithm
Matt_King: combobox computed value from content of element... referenced accname
can't remove from ARIA until we have a replacement also, in accname or somewhere else
spectranaut_: IMO where this belongs is not as important as the definition: define the value of 3 items: textbox, search box, combobox
Is there a real world example of textbox that doesn't use <input type="text">?
Matt_King: I think some do
spectranaut_: i volunteered to take a start at the definition. any other actionable input?
jamesn: agree the algo def should come first
Matt_King: are there situations where the accname algorithm couldn't be used for value algo?
jamesn: needs to be written down.
spectranaut_: spec says ambiguous things ~"get the value as if it were a button"
is that even correct?
<melsumner> I merged a few of the "editorial" and approved PRs on ACCNAME
Bryan: this is more broad than accname comp. when you return the value, your going to get other references from accname... e.g. aria-labelledby should not be used as the value comp in most cases
Matt_King: next steps?
spectranaut_: refocus on the cases we need to define... anything other than input for textbox, combobox, searchbox? maybe contenteditable="plaintext-only"
Matt_King: if the contenteditable has role="textbox | combobox | searchbox" we need to define
<Zakim> jcraig, you wanted to ask if anyone disagrees with adding a value algo?
Matt_King: not against a value algo
<melsumner> I wonder if this should be a F2F convo?
scotto: agree need the value algo... have the concepts of valuenow, value text, spinbutton.. include those too
Bryan: I was opposed to using accname to expose the value... separate algo seems necessary
jamesn: proposal to add value comp algo to accname, alongside name comp and description comp
consensus
<Matt_King> +1 to covering other roles like spinbutton with a value algorithm so that valuetext can take precedence over valuenow
Does the "Presentational Roles Conflict Resolution" always apply?
aria-hidden error case: disallow aria-hidden on root or document element (e.g. <html> or <body>) agendabot]
arigilmore: summarizes w3c/
difficult to disallow this... blocking authors from doing this
scotto: seems like this is an ARIA in HTML spec
<spectranaut_> belongs in aria-html: https://
jamesn: asking browser to not expose
<spectranaut_> can we say in the aria spec for aria-hidden "not allowed on root node in the host language"
<scotto> maybe a bit more complicated than just the author / conformance rule in aria in html. maybe something that would been needed to be mentioned in AAMs then?
<spectranaut_> jcraig: re what anne said we can leave the relection and not map it in the browser