<Chuck> meeting: AGWG-2025-03-18

<ChrisLoiselle> regrets for second hour, customer call.

scribe+

Chuck: welcome to the first AGWG after CSUN, hopefully all who joined enjoyed themselves
... would anyone like to (re)introduce themselves?

announcements

Chuck: I don't have announcements, others?

alastairc: we still have timezoen issues, next few weeks the time of this meeting will be different for those not in the US

Chuck: any new topics?

<Frankie> I request a review of the standards of conduct.

UI-context (ex-views) definition

<Chuck> https://github.com/w3c/wcag3/discussions/286#discussioncomment-12480952

alastairc: we had a longer meeting at CSUN Monday
... we got fairly far in our discussion of what we were calling 'views'
... we got to a place where the people who were there were fairly happy with it
... problem this is trying to solve is… how to solve certain requirements, eg if you have functionality that needs to work with keyboard, doesn't mean each component needs to be accessible, but the place you're accessing the functionaltiy from needs to be
... it could also support task flows, but will discuss that later
... other scoping terms we have: product/website, task/process (could be across things), then UI-context (could replace web page potentially)
... during discussion, we felt we all kind of know what we mean
... one of our conclusions was that it would be helpful to separate interface/layout from the content of the interface
... potentially this separation of surrounding interface from bits of content makes it a lot easier to understand
... programmatic doesn't always match visuals so we kind of need to capture both
... should also work for scenarios when no URI is available
... one scenario we kind of went with is to take layout/components as a 'context'
... so the definition proposed: “The layout and components available (visibly or programmatically) at one time, including dynamic components that appear without making the previous layout & components unavailable.”
... and then we had two notes too
... first “Content / information that is available only visually or only programmatically is likely to fail requirements.” and then “Controls which “flow with the text” do not disqualify a block of content from being content rather than a new UI context.” would need some explanation
... when we looked at examples, these (in the doc) were the most complex we could come up with, and they became easier with this kind of definition
... [shows slide with different examples with tabs and navigation]
... the surrounding layout doesn't change
... we would consider navigating between different tabs within the same page as the same 'context'

<cahall> Well there's vertical vs horizontal tabs right?

alastairc: we also looked at Slack… where there are components on the left for switching between Slack instances, another layer for swithing between different types of content and then another layer for switching between channels
... we'd consider it all the same context, but a change to different context if a lot of the UI is not available
... if an addition of more content means you can't access the previous content, we'd see that as a change of context

<GN015> What if panels are shown one by one (because of resized text, for example). Is then a new panel a new context?

chuck: cahall said there's vertical vs horizontal tabs right?

<Zakim> mbgower, you wanted to say what about a tear sheet?

alastairc: it depends

mbgower: just wonder where tear sheet are in this?

GreggVan: could you describe what you mean?

mbgower: something between modal dialog and separate window, it's almost like a process overlay?

alastairc: it would be the implementation… there's a gray area between modal and non modal dialog?

GreggVan: when you said context you meant ui-context, right?

alastairc: yes

GreggVan: when content pops up… two questions we need to ask are… 1, does the old ui context disappear from the DOM or is it still there? to the screenreader does it look like the old context is still there with new elements or does it all disappear? 2, @@@
... what concerns me, what happens in the DOM and what people using mouse use, would be in sync… if they happen at the same time it's a problem?

<DJ> scribe+

<DJ> hdv: I struggle with the difference between layout vs component in this draft definition

<DJ> ... there are pages which only do layouts or only do components

<DJ> ... i really like the concept though, but if we could clarify this that would be excellent because those terms are overloaded

<DJ> ... also the word "dynamic" might be confusing here

<DJ> ... dynamic vs static has no clear difference here, many grey areas

<DJ> ... also for me, there is no grey area between modal and nonmodal

<DJ> scribe: hdv

<DJ> scribe+

<DJ> alastairc: all of the terms we could possibly use could be misunderstood by somebody

<DJ> hdv: true, but to me the word "component" is especially overloaded by web devs

<DJ> scribe: hdv

<Rachael> From Explainer: "Items are the smallest testable unit. They could be interactive components such as a drop down menu, a link, or a media player. They could also be units of content such as a phrase, a paragraph, a label or error message, an icon, or an image."

Jon: if the navigation and tabs at the top were at different pages, would they be different UI contexts? or would they be assumed to be the same?

<JenStrickland> Regarding the assertion we should not use components, web components are a pre-defined thing: https://www.w3.org/community/webcomponents/

Jon: there could be two pages that are completely different, it seems they would be interpreted as the same ui-context?

alastairc: one of the things we discussed is: do we need a change of content to be defined?

<Zakim> cahall, you wanted to ask are there issues with using both vertical and horizontal tabs together?

cahall: in the examples, there were examples where tabs were horizontal and others where they were vertical, is there a difference between them?
... how does that look programmatically?

alastairc: don't think orientation matters, it's about whether the components stay consistent

<Zakim> alastairc, you wanted to suggest that we have a guideline that says "if you have something visually modal, it has to be programmatically modal as well".

alastairc: I suggest we have a requirement to say that something that is visually modal that it is programmatically modal too

<Zakim> mbgower, you wanted to say I feel like the definition is maybe missing a concept of parallel sections that have their own tab rings but have a method of moving between sections

<mbgower> Like non-modal dialogs, modal dialogs contain their tab sequence. That is, Tab and Shift + Tab do not move focus outside the dialog. However, unlike most non-modal dialogs, modal dialogs do not provide means for moving keyboard focus outside the dialog window without closing the dialog.

mbgower: I think modal vs non modal depends on who you talk to. See the quote I just pasted
... if I interpret this APG info, there are different modes possible on a page
... most people are probably on IRC… it is in one tab ring… although there is an input where I can put information, on the right hand there are users… the tab ring is constrained to what I would call view… if I get to another section, there is a keystroke to go to another section. If you move to a section it is going to move view in some way
... I see more and more interactions on the web that are desktop like
... in that mode of operation i am talking about, with independent sections, they preserve your keyboard in that other section
... Slack behaves like that

GreggVan: Hidde, you talked about component and navigation… maybe define it as 'action component'. First, when you open up a modal dialog, you often can still read the background, sometimes you can move it around… so the page isn't gone, only interacting with it is not possible
... if you can't interact what's in the DOM but screenreader can, that's something we should look at

<mbgower> Folks should try using Slack with a keyboard versus this irc client. They are demonstrably different in how the operate, to the point where I believe they are different UI contexts in Slack, but only 1 in irc.

<alastairc> I'm not sure if/how/why that would affect this definition though?

GreggVan Jon, the tabs problem you talked about, within the current definition of web page, you can have an entire web app there. We have to talk about the definition here

GreggVan: mbgower, you talked about the tab ring, I don't think we should have that define context
... you can define keyboard navigation in something else, if you did that, the evaluation unit would change
... you can also get trapped, as soon as you get trapped that would be a ui-context, but don't think that would be a functional thing?

<Chuck> qq+ to do a time check

<mbgower> I think your use of "trapped" is a little loaded there, Gregg :) It is intentionally contained in a separate modality

GreggVan: in summary, we're not there yet, but up until now, there was a wishful dream that couldn't be better than a web page, I think it has crossed over to being a ui-context as we're defining it, it looks like it might

<GN015> what is a tab ring in this context?

GreggVan: we should watch 3 things… 1, whether 'action components' change, 2, whether the DOM stays the same (expanded vs changed, 3, …
... re the tab questions, if thery are different pages they are different UI contexts

Chuck: time check, we have a second agenda item before subgroups, how much time does it need?

<mbgower> A tab ring is the number of items that are reachable by pressing Tab. At some point you will return to your starting point. That is your tab ring.

Rachael: intent was only to get to that if we had time

<Zakim> Chuck, you wanted to react to GreggVan to do a time check

<Zakim> Chuck, you wanted to do a time check

ChrisLoiselle: two points on the definition: I don't know if 'appear' in the definition implies that appear leads to that as well?
... do we need to define that or not?
... the first note about content… 'only visible' or 'only programmatically' if you were defining that in the definition, is where I was coming from
... on the second conversation… when I hear component and tab ring and other kinds of workflows… in this Google documents, if we talked about panes, there are different types of panes, that are different navigation methodologies
... maybe we can explore that more?

<mbgower> +1 panes/sections/regions/panels all are in the same area

<mbgower> We used to call them frames :)

Jon_avila: thinking from perspective of people interprerting a conformance report… it would be helpful to know which UI contexts have issues than having a UI context that involves conceptually different things
... whatever we can do to help people interpret conformance claims… having units that are broken up into reasonable chunks, if there's a conformance issues there it is helpful for people interpreting

<Zakim> alastairc, you wanted to say a problem with the definition would be if we can't agree what a UI-context is, but whether it has a wide scope or a narrow scope for particular

alastairc: re Jon_avila , I agree that is a problem and something we want people to be able to do… currently if you're in a single page app, it's the same problem but potentially bigger
... we need some kind of methodology to capture different kinds of content
... it would be useful if folks can try applying this definition, if you have something you're working on at the moment. How easy do you find to apply this?
... how easy do you find this as the basis for doing conformance?

<Zakim> hdv, you wanted to mention inert web platform feature

alastairc: whether there's a wide or narrow scope compared to web page is not as much of a problem, a problem would be if we can't agree what is captured by this definition.

<Zakim> GreggVan, you wanted to say "if what we currently have does not aleady require it - then I think we need a new provisions that explicitly says ' what is visiblly present,

<mbgower> I previously quoted the APG for modal dialog. here's where I got that quote https://www.w3.org/WAI/ARIA/apg/patterns/dialog-modal/

GreggVan: ui-context is a term we use, when we have modal dialog it would be a new context

<Rachael> +1 to both directions of matching

<alastairc> scribe+

GreggVan: I think we need a requirement like alastairc's suggested, that we need equality between what's visually modal vs programmatically modal

[ <dialog> with showModal DOM method and the inert attribute from the HTML spec already have that parity' ]

<alastairc> hdv: I've tried applying this, and ended up with a lot of UI-contexts to get a reasonable amount of stuff.

<alastairc> scribe-

<GN015> "equality between what's visually modal vs programmatically modal" - equality or equity?

GreggVan: I am still wrapping my head around content that is visible behind a 'modal' element

alastairc: I hope this conversation helped, Hidde.
... this gives us within guidelines a way to know what the target is
... scoping conformance claims is the second big thing

<alastairc> https://github.com/w3c/wcag3/discussions/286#discussioncomment-12480952

alastairc: and agreed with Gregg, agree there should be a guideline to say that information that is visually modal is also programmatically modal and vice versa

Task Flows / Processes https://github.com/w3c/wcag3/discussions/294

Rachael: this conversation is dependent on views
... we wonder if we do need a new definition, are both process and task needed. Some comments said we don't need both.
... something that's not clear in the 2x definition is that it is end to end and could go out of the existing application
... what I'd like people to do is to go to discussion #294 and think about whether or not the 2x process definition is what we need in WCAG 3… we need it for scoping conformance and for supporting our different requirements
... I'd encourage you, and we can discuss it in the meeting next week, what else do we need in order for process to work for these different requirements?
... the other open question is: if you have a process and that process spans different views, does only the process part of the view need to be tested or only the process part?
... any questions? otherwise we'll discuss next week

Sub-group work

scribe-

This is scribe.perl Revision VERSION of 2020-12-31
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00)

Succeeded: s/joiend/joined/
Succeeded: s/he layout/The layout/
Succeeded: s/megamenus/tear sheet/
Succeeded: s/also/... also/
Succeeded: s/consistent?/consistent/
Succeeded: s/ChrisLoiselle/ChrisLoiselle:/
Succeeded: s/as is @@@/a problem would be if we can't agree what is captured by this definition./
Succeeded: s/information that is visually modal is also programmatically modal/information that is visually modal is also programmatically modal and vice versa/
Default Present: Chuck, jeroen, alastairc, Jennie_Delisi, kenneth, joryc, maryjom, jspellman, kevin, dan_bjorge, hdv, Laura_Carlson, MJ, Detlev, filippo-zorzi, LenB, JenStrickland, Makoto, shadi, Léonie, JeanneEC, kirkwood, Ryladog, GreggVan, Frankie, tiffanyburtin, giacomo-petri, Jon_Avila, Francis_Storr, stevef, tink, .5, Eric, ChrisLoiselle, ShawnT, Kimberly, DJ, mbgower, julierawe, ljoakley, cahall, AlinaV, sarahhorton, jtoles, Rachael, Carrie, GN, Todd, Poornima
Present: Chuck, jeroen, alastairc, Jennie_Delisi, kenneth, joryc, maryjom, jspellman, kevin, dan_bjorge, hdv, Laura_Carlson, MJ, Detlev, filippo-zorzi, LenB, JenStrickland, Makoto, shadi, Léonie, JeanneEC, kirkwood, Ryladog, GreggVan, Frankie, tiffanyburtin, giacomo-petri, Jon_Avila, Francis_Storr, stevef, tink, .5, Eric, ChrisLoiselle, ShawnT, Kimberly, DJ, mbgower, julierawe, ljoakley, cahall, AlinaV, sarahhorton, jtoles, Rachael, Carrie, GN, Todd, Poornima, GN015
Found Scribe: hdv
Found Scribe: hdv
Inferring ScribeNick: hdv

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.)