Meeting minutes
New Issue Triage
spectranaut_: editorial UIA... no need to discuss
dmontalvo: seems leftover
if jcraig does not object, will remove
PDF-AAM?
jamesn: i will flag this to some others
implementations need to update: webkit and gecko have open bugs
scotto: steve merged.. chrome has already implemented
Rahim: cross-reference to spec where defined
jamesn: css display also has a text node definition
New PR Triage
editorial. jcraig to r+
two approved reviews... need a 3rd
WPT Open PRs
Deep Dive planning
<CurtBellew> Need to drop.
Holiday Schedule: No meeting Nov 23, Dec 28, (Dec 21 XOR Jan 4)?
spectranaut_: No meeting Nov 23 is US Thanksgiving
No meeting Dec 28... and possibly Dec 21 or Jan 4
a couple votes to cancel Dec 21. no objs
Ambiguity in AccName LabelledBy section: "[if] current node is not already 'part of' [sic]…traversal" agendabot]
handling multi-selection / selection confirmation of action for menus
scotto: maybe longer than 10m... less than deep dive
postpone until shorter agenda
Ambiguous cascade of role invalidation rules (including dependency label computation) complicates implementations
<Rahim> role="table grid", role="switch checkbox"?
<Rahim> These are the ones I see in the wild most often...
wg agreed to add role comp algo to the aria spec
Matt_King: questions the value of role fallback
Ambiguous but normative requirement about hidden nodes is hidden by default in AccName
<spectranaut_> Matt_King: are there any implementations that do this/
<spectranaut_> jcraig: I think user agents don't implement this
<spectranaut_> jcraig: I might have only written a local test...
<spectranaut_> Matt_King: my concern is that if someone implements it, and people made accessible names that rely on this feature
<spectranaut_> jcraig: seems like there is enough in hidden that is so inconsistent that backwards breakage in one browser is ok
<spectranaut_> Matt_King: if it is only implemented in one, then I wouldn't be concerned
<spectranaut_> jcraig: this is not implemented across the board
<spectranaut_> Matt_King: if its in chromium that might be concerns
<Zakim> jamesn, you wanted to propose I create an editorial PR to expose all that hidden stuff in Accname
Title should not contribute to accName for generic elements
scotto: should title on generics contribute to accname? pros and cons to this, but it's unclear what the author intended.
Matt_King: please do make the accname change that title should not contribute to accname
Matt_King: I see legitimate use of title on generics (e.g. container divs) but not for naming
jamesn: what about fallback error correction, where a generic div takes focus but has no title
Matt_King: if it's explicitly named
Matt_King: sometimes divs become focusable due to css overflow: scroll
or auto
Rahim: curious about what is the stance of aria naming of generics
if there guidance on when authors put it on a generic
Matt_King: author error
scotto: came down to when it was focused, you'd get different results
jcraig: namefrom: prohibited in the chars table for generic role
coryjoseph: when there is a role added to span/div, title is allowed correct?
scotto: yes...
coryjoseph: there are loads of web apps. will this cause breakage?
Matt_King: the experience will be improved
scotto: probably
jcraig: concerned that if we require implementations to NOT expose this, that's hiding some author provided string that may help an AT user
scotto: maybe title should be exposed in another non-name property like description or otherwise
Matt_King: should this get a deep-dive label?
spectranaut_: yes... will schedule it later
no deep dive next week
<spectranaut_> s/core-aam#207 two approved reviews/aria PR #1931 two approved reviews/