Meeting minutes
New Issue Triage
role for expressingtext messages in book text w3c/dpub-aria#54
pkra: not sure what this would add?
jamesn: feel free to ask the question
will leave for dpub to screen
AccName: reconsider appending space between text nodes and comment nodes w3c/accname#193
jamesn: agree that chromes behavior has some benefit
jcraig: not pressing. milestone as you like
Provide a role for file input type w3c/aria#1926
jamesn: agenda for another meeting
Need whitespace consistency for `computedlabel` in WPT tests w3c/accname#192
jamesn: milestoning
Why are <b> and <s> mapped to semantic roles, but <i> is not? w3c/html-aam#475
scotto: I will change <i> to map to emphasis role
New PR Triage
further clarifications for li role mapping w3c/html-aam#476
scotto: li, if you negate the container list role, orphans the role, so should be generic
jcraig: add me as reviewer. I will link the WPT tests
jamesn: remaining are editorial. no need to discuss.
F2F planning
jamesn: F2F meeting next week agenda and details linked above. if you are not in person please join remotely when you are able according to the agenda
pkra: is that the old agenda?
spectranaut_: it's final unless remote attendee needs a timezone related change
spectranaut_:
<jcraig> s/spectranaut_: //
dmontalvo-mac: anything I need to support Teams meeting?
jamesn: Adobe host team will handle
spectranaut_: don't be late
jamesn: I need to escort people in, so being late will disrupt the meeting
jamesn: looking forward to seeing you all in person
What to do about the rowgroup discrepancy between the spec and implementations?
jcraig: this came out wpt tests
jcraig: very recent changes in chrome... my local version of chrome is better than the wpt server
jcraig: it seems like chrome is "faking" the computed role
jcraig: tbody and thead mostly for printing, not used by accessibility APIS
note: I think this what jcraig said ^
jcraig: we could file bugs for this, but maybe this is unnecessary. maybe the spec should reflect reality and ignore these
jcraig:
cyns: I follow up with aaron on this
jcraig: is this something where it matters? should we have computed role tests that validate this? we could have a special test for ignored roles that would allow differences between the return values. verifyelementignore -- null or empty string or none
jcraig: don't do that work around on other roles
jamesn: I have some concerns about not exposing this in certain cases. rowgroup, when the table body becomes a scrollable region, and accessible node is exposed, if you don't have row group it follows table structure and things break
jamesn: browser should not break when generics are in table, they should ignore it
jamesn: I'm concerned if something should be tested... not exposing rowgroup is going to break things
<Zakim> jamesn, you wanted to state that rowgroup has a useful current behaviour of being a generic that doesn't bork the accessibility tree
jcraig: I saw the follow discussion of what happens if you make it focusable....
jamesn: firefox makes something scrollable focusable by default
jamesn: if you have a table with more rows than are exposed by tbody, you get scroll
jcraig: we should include focusable thead and focusable tbody
jamesn: rowgroup seems like maybe the wrong mapping. maybe we need a table generic
cyns: if this is recent change, I want to find out why and who did it
cyns: I'll talk to aaron
jcraig: that means wpt is running something older than stable... or newer???
BenBeaudry: I found a change
jamesn: I'll contact james teh about that
Consider adding a 'header' and 'footer' role
scotto: read over the minutes and prev discussion
scotto: we have header and footer map to banner and contenting landmarks, but html allows these "semantic" elements in other smaller contexts
scotto: came up in the context of minimum role issue...
scotto: so we don't necessarily want this to drop to a minimum role of group.
scotto: Joanie mentioned it could be a generic group at the platform level with a more specific role description
AXRole: AXGroup
AXRoleDescriptoin: "header"
would be fairly easy in WebKit and Mac implementations
<jcraig> s/AXRole: AXGroup /jcraig: AXRole: AXGroup /
jamesn: does this correct author errors?
scotto: not exactly. <header> outside the body context is still valid HTML
so this would map to something more meaningful than generic or group
jamesn: will this make the SR interface more verbose? excessively chatty?
scotto: hence the suggestion to only expose as a new role description, but still a generic container AT could ignore when irrelevant
<jamesn> "The header element represents a group of introductory or navigational aids."
mk: when are these used most commonly
scotto: in the context of other containers.... e.g. could be a header for a block quote, but would not be appropriate to use a banner landmark mapping in that case
<pkra> complex figures could also benefit from such elements
mk: concerned about the chattiness issue jamesn raised
scotto: our choices are: expose as group, or expose as something more meaningful
shouldn't change any behavior... just whether you hear group or header role desc
mk: if we're going to do something "better" there should be a way for an author consciously choosing this to make sure it's exposed.. compared to random web authoring toolkit that just uses the header element as a generic
for example, sections become regions if they are labeled
scotto: we could make complicated mapping roles like header inside article
scotto: provides a way for AT and authors to do this where they can't today
mk: could require an explicit role=header even on <header role=header>
jamesn: would want the default to be generic
scotto: would not be spoken
jcraig: I would want aria rules and behaviors to be consistent but not overly restrictive to act work around misuse.
mk: concerned with title attire on generic containers cause them to be exposed as meaningful semantic containers
<jamesn> zakim ,close the queue
<jcraig> s/@@@/concerned with title attire on generic containers cause them to be exposed as meaningful semantic containers/
scotto: could do this in AAM similar to region role mapping
scotto: I will write some PRs
cyns: will @ others to the issues
Not sure how to sub in a URL due to the slashes, but above where BenBeaudry said "I found a change" he linked https://
He then followed up in the issue w3c/