[CSSWG] Minutes APA + CSS Joint Meeting 2020-10-14: ACT, Media Queries 5, CSS AAM, Navigation [css-snapshot] [mediaqueries-5] [css-nav-1]

  These are the official CSSWG minutes.
  Unless you're correcting the minutes,
 Please respond by starting a new thread
   with an appropriate subject line.

Joint Meeting with APA


  - The recommendation is to use the CSS snapshot ( https://www.w3.org/TR/CSS/ )
      as an additional data point for what specs are stable and what
      are still changing since TR can get out of date.

Media Queries 5

  - There has been progress made on the user preferences section of
      Media Queries 5. Though it's not stable, it is stabilizing so
      now is a good time to review it.
  - Some of the requests on the original wishlist
      ( https://www.w3.org/WAI/APA/wiki/Media_features_use_cases_for_personalization
      have a dependency on creating a security/privacy model so the
      recommendation is to open an issue with TAG to get them to look
      into it further.
  - The items on the wishlist looking to expose a value instead of a
      yes/no answer (such as timeout) are probably better to be a part
      of the Environment Variables spec ( https://drafts.csswg.org/css-env-1/ ).


  - RESOLVED: Close CSS AAM / css-a11y Task force
  - Coordination will continue using the github tags and review as is
      currently happening.
  - It was noted that the first Wednesday of each month the CSS group
      meets at 4pmPT so there would not be a meeting conflict and APA
      members could join for discussions.


  - The Spatial Navigation spec ( https://drafts.csswg.org/css-nav-1/ )
      had some progress on creating requirements for navigation using
  - It has been looked at by both chrome and chromium teams, however
      no one on the call knew the current status or if the work was
      different between chrome and chromium.


  Shadi Abou-Zahra, W3C
  Rachel Andrew, Fronteers
  Rossen Atanassov, Microsoft
  Christian Biesinger, Google
  Amy Carney, Invited Expert
  Tantek Çelik, Mozilla
  Elika Etemad, W3C Invited Expert
  Wilco Fiers, Deque Research
  Becky Gibson, Knowbility
  Paul Grenier, IE
  Brian Kardell, Igalia
  James Nurthen, Adobe
  Joshue O Connor, W3C
  Theresa (Tess) O'Connor, Apple
  Kazumasa Okabe, Yahoo! JAPAN
  Justine Pascalides, Educational Testing Service
  Simon Pieters, Bocoup
  Morgan Reschenberg, Mozilla
  Florian Rivoal, Invited Expert
  Janina Sajka, Invited Expert
  Boaz Sender, Bocoup
  Jen Simmons, Apple
  Alan Stearns, Adobe
  Miriam Suzanne, Invited Expert
  Jason White, Educational Testing Service
  Gottfried Zimmermann, Invited Expert

Scribe: fantasai
Scribe's Scribe: boazsender


  janina: ACT, suffice it to say they try to make the WCAG
          machine-tested to the extent possible
  janina: They've been running into issues with older CSS specs
  janina: They asked if we could talk about that
  <shadi> https://github.com/w3c/wcag-act/issues/477
  <shadi> ^^issues with CSS specs
  Wilco: ACT is tasked not just automations, but also to create clear
         definitions for how to test
  Wilco: We're currently focused on HTML, etc. and applying WCAG to
  Wilco: We're referencing a lot of bits of the specs, translating
  Wilco: So we make a lot of use of CSS specs, ARIA specs, etc.
  Wilco: Everything depends on existing definitions
  Wilco: Quite a number of specs which appear not to be actively
         worked on
  Wilco: Here's a list of specs we've referenced at one point or
  Wilco: Reason we're using them are browsers implemented them
  Wilco: Advice on how to deal with that, and see if any of these can
         be completed at some point

  janina: I wonder if an example of a problem might be helpful
  <fantasai> +1 to example
  Wilco: We have some rules around CSS transforms
  Wilco: CSS Transforms spec, there's a recommendation
  Wilco: ...

  fantasai: Not sure if this is a problem with the TR being out of
            date, or the stage not being advanced enough in the w3c
  Wilco: It seems like the specs are out of date
  fantasai: This is definitely a problem we have. astearns asked me to
            draft a report on this. What we need to do in the csswg,
            which the chairs need to lead, is a project in which we
            systematically update the specs. Thanks for coming to
            complain, maybe it will get the rest of the working group
            to care.

  janina: Is the issue that draft word is making you reticent to rely
          on the current spec?
  janina: Because concerned they might change?
  Wilco: Can't build on them because they're not stable
  Wilco: More permanent status, could be more reliable
  Wilco: confident wouldn't get changed up from under us
  <shadi> +1 to Wilco
  fantasai: https://www.w3.org/TR/CSS/
  fantasai: Another thing that might be useful is the css snapshot
  fantasai: which will let you know a spec is stable
  fantasai: A lot of our specs are not in CR since there are a lot of
            issues in them, but they are being shipped
  fantasai: Snapshot will tell you where our specs are
  <TabAtkins> Will note that the idea that you shouldn't rely on
              drafts is a persistent but untrue idea. Drafts don't
              guarantee stability, but they don't guarantee
              instability either, and *because* TR is often
              out-of-date, the drafts usually contain the most
              accurate info about the bits that are implemented.
  fantasai: Anything in the snapshot is fair game, and probably a few
  fantasai: We triage every year and add things
  Wilco: Sounds good, thanks

Media Queries 5

  janina: We came with a wishlist many years ago
  janina: Excited to see some of the user-oriented specificity for
          contrast etc.
  janina: A lot of this came out of work jcraig was doing
  janina: Wanted to ask about progess on that
  janina: Some comments from APA also
  <Gottfried> Wiki page:
  janina: How are implementations coming along? Does it look promising?

  Rossen: Clarification on what we were discussing back then?
  <jcraig> The user preferences section
  <Gottfried> Spec chap. 11:

  florian: This is looking pretty good. Spec isn't final, a number of
           issues still open
  florian: Things are stabilizing, not stabilized
  florian: but things are landing
  florian: A number of things related to color and contrast
  florian: Addressing preferences for contrast, forced colors mode and
           letting authors know about it
  florian: Preferences wrt colors scheme, reduced motion / transparency
  florian: both on expressing user preference, some of which tied to
           a11y, also colors, fairly rich set of things
  florian: All are progressing in terms of implementation. Not
           finalized but moving along
  florian: More things in that spec also, in particular relating to
           video, these are a bit controversial
  florian: but overall spec is good. Good time to review, not
           finalized so still opportunity to change

  jcraig: A lot of these were indeed things we drafted in Indie UI and
          proposed in CSS
  jcraig: Thanks florian for doing the work
  jcraig: Bunch of other topics we were discussing wrt Indie UI, much
          more detailed media features
  jcraig: but Indie UI has a security/privacy model proposal and Media
          Queries doesn't have that
  jcraig: so we're butting up to the edge of what we can request
          without asking for user permission
  jcraig: Things that have overlap have broader interest in
          mainstream, so less concern over privacy
  jcraig: Some topics are off the table until can restrict access to
          MQ info behind security/privacy model

  Gottfried: When would that happen though?
  Gottfried: Wrt our wishlist, is there any way to deal with this, say
             this is L5 and this goes into L6?
  Gottfried: We also have a few items that are very important, which
             are not a fixed set. E.g. session timeout
  Gottfried: Would be great if an author could basically find out, as
             a variable, how long is the preferred session timeout of
             the user
  Gottfried: Is there any model foreseen where a web author could get
             a value transferred from a system setting and use that?
  <Gottfried> Wish list:
  jcraig: There is no Media Feature Privacy Model being worked on AFAIK
  jcraig: but would be good to link to work
  florian: Indie UI did have some things. There's been a lot of
           discussion of security/privacy in general since then
  florian: I'm not an expert, so I understand the need, but I can't do
  florian: If you have an idea, please propose it. Putting something
           wrong is a better way to get what's right rather than just
  Gottfried: Folks who worked on the old security model, any way to
             transfer to CSS?
  jcraig: It's possible, maybe. There were a number of ideas in that
          one. Wasn't fully fleshed out, and would take awhile to work
          out the kinks, and I don't think CSS is the right landing
          spot for that.
  Rossen: Would recommend opening an issue with the TAG
  Rossen: Would be good to look at security model and see where it
          should land
  jcraig: I think I have an issue filed against CSS also

  Gottfried: If we have things like e.g. values for session timeout
  Gottfried: is there anyone that we could do this with media features
             or other tech
  florian: MQ are not great for ask how many of something, they're
           questions that are answered by yes/no
  florian: You can ask "is the timeout more than 60 sec", can do that
  florian: but if what you want to expose is the number, it's the
           wrong tool
  Gottfried: But we'd still need a model for this value
  florian: To expose values to CSS, we have environment variables.
  florian: If you want to expose to CSS specifically, that's the right
  florian: but for JS, I'm not sure. Probably need something different.
  <Wilco> +1
  <florian> environment variables are at https://drafts.csswg.org/css-env-1/

  <TabAtkins> I haven't been entirely clear on what "session" is being
              talked about here.
  <TabAtkins> Like, a bank website imposing a relatively small
              activity-based timeout?

  <jcraig> This is probably the best Privacy & Media Features issue:
  <Joshue108> Just for the record, also as James mentioned a similar
              spec the PING privacy threat model may be of interest


  Rossen: Is the CSS AAM still a thing
  janina: Yet Another Accessibility API Model
  janina: From the APA side, wondering if that's still something we're
          hoping to get to
  janina: Do we have a timeframe?
  jamesn: I'm not aware of what's going on in that space right now
  Rossen: AAM mostly started from need to go and define a lot of the
          things that were longstanding complaints wrt CSS, and visual
          effects that CSS adds to content
  Rossen: Examples vary from flexbox reordering to being able to
          position something on top of something else with z-index,
  Rossen: When we started this many years ago, there was strong
          interest from both sides

  Rossen: We created a repo and ML, but very little work has happened
  <jensimmons> What's the link to the repo?
  Rossen: What can we do to make the work happen?
  Rossen: Haven't seen many people participate, so don't even know who
          to look to for an answer
  Rossen: Would be a good medium to liaison things that are issues
          with CSS, but given fact that it's been 4-5 years at this
          point and not much has happened
  Rossen: Chances of something happening in the next few years are
          probably the same
  Rossen: so should rethink if this is the best path forward

  jcraig: My approach has been if I have an accessibility issue with
          CSS, I bring it directly to CSS.
  jcraig: I find intermediary groups not so effective, ends up
          duplicating conversations and gaps in knowledge
  jcraig: So find most value in bringing the a11y issues to this group
          and having this group deal with it
  Rossen: That's the approach that many have taken, and one that I
          have preference for
  Rossen: Having entirety of CSSWG look at incoming issues is usually
          productive for us
  janina: I'm fine with that actually
  janina: I know we tried to put a TF together between CSS/ARIA/APA,
          but hasn't quite reached the level of functional
  janina: Might have some opportunity there again, but informal if
          need be
  janina: Amy Carney who is participating consistently and reliably
          with APA going through CSS issues
  janina: not getting much response because of lack of expertise, with
          her help been working through them slowly
  janina: without AAM ...
  janina: We're in a better position to clear issues atm
  janina: fantasai has been helping by flagging specs for APA review
  janina: so working through backlog
  janina: and that seems like the answer atm
  janina: Maybe Amy can be formal liaison
  janina: let's work on better work mode.
  Rossen: We've been diligent about tagging issues with various labels
          in order to attract attention for horizontal review or other
          WG attention
  Rossen: When people open issues in our repo, we try to figure out
          what it's related to, is it a11y-related

  Rossen: Having said everything, putting forward motion of sunsetting
          CSS AAM and continuing in liaison mode going forward?
  janina: I think so! We can resurrect it in the future if necessary.

  jamesn: CSS AAM lives in ARIA WG, nothing happening
  jamesn: Thought we were working about css-a11y TF
  jamesn: but doesn't need css-a11y TF
  Rossen: Great clarification, it is the TF that we are sunsetting

  jcraig: Wanted to remind group that there are two CSS meeting times,
          once per month it is 4pm Pacific
  jcraig: which doesn't conflict, so can choose to discuss issues at
          that time and join the CSS call
  <astearns> the first wednesday of each month has that meeting timing
  [[ Note CSSWG repo has 'Agenda+ APAC' tag for slotting an item to
      that particular timing ]]

  RESOLVED: Close CSS AAM / css-a11y Task force


  janina: We've had some discussion of semantic navigation vs layout
          navigation models
  janina: I think issue never got resolved
  janina: idk if any updates
  janina: Maybe this is another one we drop for now, without someone
          wanting to work on a specific model
  florian: It's a little more positive than that but not completely...
  florian: Jihye and I were involved for awhile on writing a Spatial
           Navigation specification
  florian: which would define a model for what happens when you try to
           navigate up/down/left/right rather than next/previous
  florian: and what kind of APIs exist to allow authors to adjust when
           UA gets it wrong
  florian: Initial stage was polyfill-driven
  florian: we got a model going
  florian: Moved into an exploration in the Chrome implementation
  florian: but don't know what happened after
  florian: Whether it's validating the model or they're moving to
           something else, or if it's done or given up
  florian: Anyone from Chrome know where this went?

  Rossen: Can't speak for Chrome, because don't work for Google
  Rossen: But from Chromium point of view, from discussions we've had
          in the past we have considered this, we have discussed quite
          a bit in various forms
  Rossen: The running model that most engineers have gravitated
          towards are based in enabling spatial navigation on top of
          existing accessibility APIs
  Rossen: and existing accessibility tree
  Rossen: and its ability to describe a visual model as it does for
  Rossen: Need both types of UI pieces on the screen
  Rossen: This has additional benefits
  Rossen: First, it builds on existing platform and systems on every
          native platform
  Rossen: Second, there's an additional incentive for people to
          annotate and make sure their content is an accessible format
  Rossen: so that such navigation models can have easier time with
  Rossen: Broadly speaking, always two major categories
  Rossen: when these discussions take place, intermingled
  Rossen: people have hard time separating
  Rossen: One category is what I'd consider "structured application
          model UI"
  Rossen: Usually composed of not too many top-level items that are
          being presented for navigation
  Rossen: and then there's normal Web, where you have e.g. NYTimes and
          it's a bunch of elements all over the place
  Rossen: Making two consistent and perform in the same expected way
  Rossen: is a very tall order
  Rossen: By default browser will have harder time making something
          like this work from the get-go
  Rossen: but application model, which has more applicability in TVs
          etc, that one is pretty straightforward to make it work with
          this technology
  Rossen: So this is where the discussions are in the Chromium
  Rossen: as of ~6 months ago
  florian: Then we have double homework
  florian: because I know that the model Jihye and I wrote in the spec
           continued to be explored by Chrome engineers
  florian: so we should try to see if that discussion merged into the
           one you discussed, or they're separate and we should
           connect them
  florian: And then if they diverged from what the spec says, then we
           need to make sure things in sync
  florian: I'll note also I used to be sponsored to work on this, but
           no longer am, so can't lead this.

  rachelandrew: This is something that comes up all the time when I
                talk to authors
  rachelandrew: So there is interest from Web developers to do the
                right thing
  rachelandrew: The want to use these tools, but concerned about a11y
  rachelandrew: so need to do something about it
  rachelandrew: People want to do the right thing, but lots of use
                cases for reordering stuff, and tripping up over it
  Rossen: Provided no one on the call from Google who could signal any
          additional progress happening in the area, should table this
          question and answer it in an issue?

  bkardell: Just want to point out, last year in Toronto we had
            someone from Google who came and gave an alternative
            vision for this, which they had prototyped
  bkardell: I've meet with some other people who had some other ideas.
  bkardell: If there's exploration in Chromium, it would be downstream
            from the thing Florian and Jihye worked on

  Rossen: So, we're at time.
  Rossen: This particular topic, best path forward is to ask at least
          someone to run down the status of where the current Chromium
          community is with this particular implementation
  Rossen: If I'm missing anything on this topic at Apple / Mozilla,
          please bring it up
  Rossen: Please raise issues against css-nav spec
  <florian> https://drafts.csswg.org/css-nav-1/

  janina: This has been productive yet again, thank you everyone
  Rossen: We are certainly interested in Web remaining accessible,
          please raise issues and/or talk with the liaison! Thanks

Received on Saturday, 17 October 2020 19:54:14 UTC