<zcorpan> https://bocoup.com/blog/how-to-scribe-at-tpac
<zcorpan> might help :)
<scribe> scribe: sarah_higley
matt: reminder, no meeting next week, we'll meet on December 3rd
matt: this is one of the bigger
changes, we're in a good position to land once 1.2 is
done
... we want to revisit here and make sure we have the reviews
we need
... I'm assigned to the editorial review, Carolyn is assigned
to functional review
... it would be really nice if we were able to merge this
before the xmas holidays
... we don't have anyone assigned to visual design review, Zoe
is assigned to the accessibility review
... Sarah on code, and Valerie on test
zoe: I can do both visual and
accessibility review
... I have an alternate visual example, I can have it ready to
look at by December 7th
matt: what Zoe's doing wouldn't change the functionality, javascript, html, or testing
zoe: it will probably change the HTML, but not javascript
(correction: December 2nd, not 7th)
matt: functional review and test review can go forward
matt: for 1.2, there's two
primary goals: the first is to close all open combobox
bugs
... and then it would be good to add an example of a combobox
that opens a dialog, and Jon has exactly that with the
datepicker combo
... he has some work to do before it's ready for review, so the
PR is not ready yet
... let's look at the combobox project and open bugs, and see
how to address those
link: https://github.com/w3c/aria-practices/projects/7#card-28909395
matt: first: extending guidance
for combobox to use searchbox role.
... this might no longer apply
<jongund> https://github.com/w3c/aria-practices/issues/1164
returning to tabbed carousel: Jon is asking issue #1164, handling focus moving to tab panel
jon: our current tab panel and regression tests include that requirement, but there is some debate as to whether that's a sometimes thing or always thing
zoe: I think the confusion with
the tabbed carousel comes from the fact that you can tab to the
panel, but also the anchor that's inside of it
... and the anchor is the same width and height as the tabpanel
itself, which is confusing. So we should either remove the link
and add that later as an enhancement, or remove the tabindex of
the tabpanel itself
... because there's already a focusable element inside of it
that encompasses all the content, so one of them has to go
jon: from my perspective, if I tab into the tab panel and it doesn't tell me I'm in a tabpanel, is that confusing? Will the screen reader announce existence of tabpanel if it doesn't receive focus
matt: this is a screen reader
dependent thing, and depends on how you're navigating. If
you're using the tab key, no screen readers announce leaving
the tab panel
... the purpose of putting focus on the tabpanel itself is so a
screen reader user could start reading through the whole panel
from the top, so it's a good way to place the screen reader
cursor at the top
... in this particular case, I don't think we revised our
guidance
carolyn: so are we never going to advocate that tab panel is a single tab stop widget?
matt: the tab panel is a container that could have multiple tab stops within it
carolyn: we would never say that you should down arrow to get to the panel?
matt: right
carolyn: just checking
matt: you could put anything in a
tab panel, it's a generic container, just like a
group/region/article
... there was someone who suggested adding key strokes for
putting focus on the tab panel. Those are interesting ideas,
but don't have any native equivalent when it comes to tabbed
interfaces, and there's no way to convey the availability of
those key strokes
zoe: if all screen readers
supported aria controls, that would do it
... that is part of the tab pattern
matt: the problem is the screen reader would need to provide a key that works in forms/focus mode
carolyn: or a new aria attribute, that's a new aria-key (laughs)
siri: if I am in a tab widget,
when I listen to the tab, if it is placed in a section with
aria-label then I know I am inside that
... then if I navigate there and select anything with enter, I
know it is selected and focus is sent to the tab panel. And if
the tab panel has a heading that matches the corresponding tab,
then I know I am in the tab panel
... then I can move focus or use arrow keys, and read the tab
panel. Then the next tab is send to the tablist or next
focusable item
... as far as I know, only JAWS will say if I am in a tabpanel,
and announce the end of the tabpanel. Haven't heard NVDA
announce any information about that
matt: that's correct, NVDA does
not tell the boundaries of a tabpanel
... I don't think that should influence our guidance too
much
zoe: I think we can solve this by making JAWS free
matt: back to the carousel
itself, the carousel has two elements that provide focus to
roughly the same content. So the question is, whether or not it
should be option
... there are two ways to go with that: remove the link focus,
which doesn't seem good for regular keyboard users. Maybe it
doesn't make a difference if you fake it
... have enter on the tabpanel activate the link
... no, that doesn't help a screen reader user, because they'll
never press enter on a tabpanel to activate a link. I take it
back
... so, it feels like this is a situation where focusing on the
tabpanel isn't something we should do, and the guidance for
focusing on the tabpanel should accomodate situations like
this. That's my gut reaction
jon: yeah, I think that guidance
has to be in there so we can explain in the example
... maybe if the first element in the tabpanel is focusable,
you don't need to put tabindex on the tabpanel
matt: that's kind of the direction I'm thinking. I think we already have language sort of like that, I'm looking for that in 3.23
siri: if we keep focus on the tabpanel and we label it, a screen reader user will know if they're in the tabpanel or not. If there's a heading, that's fine
matt: I actually don't see any
language in the keyboard section about this
... yeah, we have nothing in the pattern right now at all. Is
that what issue 1164 is saying, that we have nothing in the
pattern but we have examples that do this?
... the thing is to add language to the pattern to explain when
you should do this. It seems that the tabbed carousel is one
where you should not do it
jon: and the thing is when the first element is focusable
matt: if the first element is
focusable and interactive, then you would not do that (focus
the panel), but if the first element was non-interactive, like
a heading
... if you want to draft something Jon, that would be great.
This is for section 3.23
(tablist)
matt: this would be language that
would go in the keyboard interaction section, and corresponding
bullet in the states and properties to say it has
tabindex="-1". Or maybe not?
... actually, I don't think we usually tell people that
carolyn: we tell tabindex. In each example, the section is states and properties and tabindex
matt: I'm talking about in the pattern, not the example page
jon: we also have a note in there that talks about tab issues
matt: yeah, this would just go in the keyboard interaction section of the tabs pattern, section 3.23
jon: so I'll take the tabindex off the panel in the carousel, then
matt: and also update the carousel test
(back to combobox)
matt: we were looking at the
combobox project, going to reorder issues
... closing issue about searchbox with a comment
... the next one: must have a label problem
carolyn: it's "escape key must not clear current input"
matt: this issue is high
priority, right? I think we agree we want to change this in all
the comboboxes so it doesn't clear the input when you press
escape to close the popup
... there's also the optional behavior to clear the input if
the popup is already closed
carolyn: we could put it on only one example, to show both
matt: so outside of APG in the real world, is this a thing that people see?
carolyn: I've also seen where
escape key will restore the previous value, even after the
dropdown is closed
... if I overwrite the url bar in Chrome, and press escape, it
will restore the previous value
sarah: there are some places in Windows desktop apps that clear the value, but it's inconsistent
matt: so should we include that (restoring the previous value) as an option in the text?
sarah: I'd rather not -- if people want to, they can, but we shouldn't float it as a possibility
matt: if we have one combo with the escape behavior, how about grid?
jon: I can take that one
matt: yeah, just make it so it
doesn't clear when closing the popup
... when you hit escape, it should just close that popup if
it's open. That's it
... and one line of change in the documentation below
... next, issue #982, issue with onBlur. Does that still exist
after the fixes last week?
sarah: I can take that, check if it repros, and fix if so
carolyn: the person who opened the issue, offered a PR, we should encourage that behavior
matt: in this particular case, I
want to get a fix sooner
... yeah, if this is reproducable, this is a high
priority
... next issue: authoring practices should specify blur
behavior
sarah: so I think I understand this one: it's whether the currently highlighted option is selected on blur, or if selection returns to the previously selected option
<jongund> I have to leave a little early
matt: we already do this, with
the different types of autocomplete. This says the
documentation doesn't specify it, I guess then it doesn't
specify it clearly enough
... this seems like an editorial issue
sarah: I was just writing about this, so it's top of mind. I think it's not super clear from the docs how autocomplete affects blur
matt: we don't have this in the
keyboard interaction section but we do have it in the
introduction, so I guess this is something we could
improve
... this would be part of the section that details interaction
when the dropdown is open
... so if you tab into the states combobox and you type "AL"
and alabama is highlighted and selected, then if you tab out,
alabama becomes the value
... so it does affect the keyboard interaction
subsections
... there's five subsections, doesn't apply to dialog
... will figure out specifics later
... next issue: visual design review. Simon, how much of this
is left open?
issue 1263
simon: I think I fixed the first three examples, and didn't address the grid example. The grid has color contrast issues at least
https://github.com/w3c/aria-practices/issues/1263
matt: this is higher priority
than the previous issue
... clicking outside the popup doesn't close the popup
sarah: I can take that one
matt: next one, grid combobox: https://github.com/w3c/aria-practices/issues/1266. It seems like the scrolling issues in the grid combobox are pretty bad, I think this should be higher priority
carolyn: I did find it odd that you could type down arrow. I suppose in some search boxes, you can't type down arrow
sarah: oh I remember this, it's actually a bit different, not a blocking problem, just unexpected. This isn't high priority
matt: I'll edit this issue
... I want to skip to the incorrect selection, utah vs.
wyoming: https://github.com/w3c/aria-practices/issues/1268
carolyn: I got this twice, the
first time I ignored it, but then I got it twice
... hard to reproduce, one time it happened after going back to
another window and returning. I can't repro consistently
though
sarah: I kind of want to just refactor combobox
matt: maybe if that happens, this would just go away? :D
carolyn: yeah, perhaps this is related to something else too, so maybe we'll wait on this
matt: Simon is working on this next week, and we need guidance added to this PR
<zcorpan> https://github.com/w3c/aria-practices/pulls?q=is%3Apr+is%3Aopen+label%3A%22Needs+Review%22
simon: there's a label for open
PRs that need review
... these PRs I want to wrap up the week after thanksgiving
This is scribe.perl Revision: 1.154 of Date: 2018/09/25 16:35:56 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00) Present: Matt_King zcorpan carmacleod sarah_higley CurtBellew ZoeBijl siri jongund Found Scribe: sarah_higley Inferring ScribeNick: sarah_higley WARNING: No date found! Assuming today. (Hint: Specify the W3C IRC log URL, and the date will be determined from that.) Or specify the date like this: <dbooth> Date: 12 Sep 2002 People with action items: WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option. WARNING: IRC log location not specified! (You can ignore this warning if you do not want the generated minutes to contain a link to the original IRC log.)[End of scribe.perl diagnostic output]