Meeting minutes
New Issue Triage
jamesn: timescale for milestones
aam #149
focus in editors meeting
valerie, put a year on it
scotto: would be happy to agenda it
people want to fix this it is unclear what to do
no guidance
specs are contradicting
amm 146
jamesn: remove these from core aam?
definitions that don't exist in spec
valerie will merge
pr 147
aam #145 editorial
jamesn: review
#173 slot element
aria label issue
jamesn: agrees safaria and ff are correct
this is accname issue
we don't specify what happens?
how should an aria label on a slot be treated
?
scotto: happy to take on been thinking about prohibited names
slot should not be interacted with or introduce content to a11y API
new category of name NOT supported
jamesn: <body> would be another
meta tags
scotto: add as formal section to accname in aam
#1829 aria
scotto: from IBM, topic has come up before with them, general confusion about how 'generic' is supposed to be exposed
divs, sometimes exposed as generic other times ignored
conformance checker folks
need guidance
jamesn: assinging for opinion to schne324
scotto: happy to contribute not the person to own
#439 aam
should an img with no src map to img
jamesn: on agenda
New PR Triage
pr #1830, reviewers added
#1827 pr similar
Deep Dive planning
Should an img with no src be mapped to role=img?
aam #439
img with no src
and no alt
browsers do different things
safari is correct, chrome and ff should change
they should not render anything for this case
scotto: there are instances without a src that do have alt text and people add css background images
jamesn: if alt text is set to non empty it should be exposed as an image per the spec
scotto: if alt is meaningful it should be exposed
jamesn: it shouldnt' break
jamesn: Id be happy with changes to aam
jcraig: author workaround is to include a url-encoded spacer gif
spacer gif
scotto: put src=#
jamesn: src='' would be removed
jamesn: browser inconsistency, we are attemptimg to make it work the same which is good
<jcraig> s/include spacer giv/author workaround is to include a url-encoded spacer gif/
pkra: not having src? is that a technique for lazy loading the image (post DOM load)?
scotto: probably author error
jamesn: templates can provide empty src=''
scotto: question for group,,, do we have a preference to indicate generic or presentation
?
img set to presentation with empty string, is that a good idea?
should it be generic instead?
img broken auther added alt text but role presentation caused it to be pruned from the a11y API
scotto: is this a way to set role to img, or to generic, or to none
Matt: how can you have 2 roles?
3 situiations
alt provided but empty -> one
what role should be for img that doesn't have a src?
jamesn: just do not map this at all?
scotto: usaully applied to things we don't have roles for
img could map to none
jamesn: concerned about none
that is the same as alt=''
no src, HTML says it's nothing
jamesn: the case is not a 'decorative image' it 'does not exist to anybody', img with no 'src'
jamesn: generic is better than none
matt: why
it putis it in the tree
matt; presentation image is NOT in the tree, that is what we want
jamesn: I don't want it in the tree
people asking for more info on presentation elements
jamesn: concerned about adding extra things
matt; browsers cant correct, no src
scotto: tabindex-0 makes it generic
Matt: feels like author error
putting it in the tree with generic makes no sense
there is no src, no image
jamesn: if theres a broken src, browsers will try to render text
jamesn: no text in this case
no alt text
scotto: chromium firefox, src is absent or '' nothing renders
any character will render a broken img tag
happens on firefox, not chromium, FF looks like a peice of paper torn in half
Matt if I forgot to put src and using FF
i would't get role none
jamesn: no src, it does not render
jamesn: webkit is rendering img
jcraig: if different than windows we may not be decided
jamesn: good way of moving forward
AccName Role Traversal Proposal
jamesn: don't understand why the focusable concept
focusable elements, logic to traverse the roles
some roles would be exposed regardless
constructs that shouldnt be included
jamesn: can we move to something that is always traversable or not?
matt: is there something in accname about menu button inside a heading
bryan: accname specificed 2h to 2x.i it would include everthing
i don' think that is useful
matt: accname stops at groups
content of group in treeitem won't be in the name
bryan, in accname it does not stop
jamesn: it does not stop
bryan
changed 2f.i to 2f
was told to put it back
algo skips initial statement
jamesn: name from content too restrictive
mirror what browsers do
can we come up with a list
of traversals
bryan, how do you deal with that in the ARIA spec to sync the accname spec
jamesn: why not work on that latert
jamesn: doesn't have to be the end state
make it more correct today
bryan: spec bypasses conditional statement
the accname spec
some devs are interpeting that the beginning of F is ignored
there is nothing in aria that supports ...
jamesn: clarify that conditional is always checked at FI
jamesn: we need something new to determine if a role is traversable
jamesn: divs for example, you want to traverse a generic, but currently it is not allowed
which is wrong
jamesn: if it isn't name from author only
don't traverse
scotto: if no name, you maybe still have to traverse
link that contains an <article> with no name returns a link with no a11y name
jamesn: if parent focusable and no accessible name should drill in
scotto: both are necessary
jamesn: is this just error correction
jamesn: maybe article should allow name from content
bryan it could be anything like a table
table does not support name from content
jamesn: like to make some progress
bryan need help with proposal
addresses, never always
and focusable
jcraig: will discuss with bryan
Should aria-hidden be effected by aria-owns?
jamesn: hidden is looking at 'ownership' rather than dom ownership
scotto: modal dialog, pops up and live regions didn't work because they were inert
aria-owns could pull the live region in the dialog
scotto: could solve this now
matt, more than one way to go
make owns take precedence over hidden
owns references a child of something hidden
scotto: like aria-labelledby describeby referencing hidden content
matt if hidden left it out of tree
chromium marks it hidden and webkit leaves it out of tree?
jcraig: it would not work as intended
jcraig: to follow owns tree
a cross reference not a move in the tree
plan to adopt new meaning of aria-owns
not implemented in safari until other ownership change happens