W3C home > Mailing lists > Public > www-style@w3.org > February 2017

[CSSWG] Minutes Seattle F2F 2017-01-12 Part V: CSS2.1, What can go to CR this quarter?, WPT, CSS UI 3 [CSS21] [CSS-UI-3]

From: Dael Jackson <daelcss@gmail.com>
Date: Mon, 13 Feb 2017 20:59:15 -0500
Message-ID: <CADhPm3v4uG==1naOd6pwHNkq2YY44rs0rRHOokxvQn2vvNJniQ@mail.gmail.com>
To: www-style@w3.org
  These are the official CSSWG minutes.
  Unless you're correcting the minutes,
 Please respond by starting a new thread
   with an appropriate subject line.


  - astearns summarized the discussion from the 4 Jan call on how to
      continually keep CSS 2.1 up to date (Minutes here:
  - There was a strong leaning to go with the note approach with
      lots of bikeshedding on what to call the note.
  - RESOLVED: Create CSS2-testing as a NOTE
  - RESOLVED: 301 redirect from /TR/CSS21/ to /TR/CSS2/, and from
              /TR/CSS22/ to /TR/CSS2-testing/
  - RESOLVED: Cherry-pick changes from CSS2-testing/ that have tests
              and 2 impls and port to /TR/CSS2/, Process-wrangling
              to be handled by our Team Contacts.
  - fantasai will write a wiki page detailing this new process.

What can go to CR this quarter?

  - The group talked through specs that should be able to make CR
      soon so they can receive priority on the telecon agendas.
  - There was also a group desire to republish every WD this quarter
      now that there are more tools available to make that easy.
  - RESOLVED: add Xidorn as editor of css-ruby
  - RESOLVED: Florian co-edits css-contain


  - RESOLVED: Move CSSWG tests to WPT


  - gsnedders will review the test suite for completeness.
  - RESOLVED: Defer text-overflow double value and string value
              features to L4
  - RESOLVED: MUST->SHOULD for negative outline-offset
  - RESOLVED: nav-* at-risk
  - RESOLVED: caret-color no longer at risk


Agenda: https://wiki.csswg.org/planning/seattle-2017

Scribe: fantasai


  astearns: Idea is that we have one /TR document that is CSS2.
  astearns: That has everything that is really implemented in
  astearns: And is the reference that everything goes to.
  astearns: Then we also have a /TR document that is a NOTE.
  astearns: That is the same thing, but with proposed errata.
  astearns: Stuff that isn't quite vetted.
  astearns: Aren't tests, or tests aren't passing.
  astearns: And we periodically--much more frequently than
            previously--update CSS2 with whatever in the NOTE is
  Florian: One of the things that we said we'd be doing with
  Florian: Was to delete the things better defined elsewhere.
  Florian: If we are replacing CSS2.1, do we replace with things
           that have deleted sections?
  fantasai: I think we should not delete anything
  fantasai: But we can put a note (not an obnoxious notice, but a
            regular note) at the top of each replaced section/
            subsection pointing to the corresponding section/
            subsection in the replacement spec.
  Florian: What status does the replacement have to be?
  fantasai: CR is fine, replacement spec would be at least as good
            as that in CSS2.1 at that point.
  astearns: What are we calling it?
  fantasai: CSS2.1
  fantasai: It's already REC, we're not renaming it
  Florian: An advantage of deleting parts is that you no longer have
           to errata them.
  fantasai: We'll get there when the whole spec is replaced :)

  astearns: Team Contact, what do you think?
  Bert: Seems okay, yeah. Just unsure what you're going to call it?
  Bert: CSS2 Revision 17?
  astearns: What I was hearing is that we want to update things,
            don't really need to update the title, we'd just have
            dated drafts
  Bert: Yeah, I just think it's weird to have revisions of Revision
        1 all called Revision 1
  plinss: If we want a test suite per update, need new shortnames
  Florian: No, I don't think we should do that. We should just keep
           the test suite updated
  Florian: Shortname should stay the same, Title can probably stay
           the same, and if we really want we can put the date in
           the header but that's obvious already.
  tantek: What if you're doing spec archeology.
  fantasai: That's what VCS is for.

  astearns: Proposal is to work on errata in the main document as
            represented in the NOTE.
  astearns: We cherry pick changes from the NOTE to put into the REC.
  astearns: So we do maintenance in the NOTE, and when things in the
            NOTE are ready they move to the REC.
  fantasai: ...
  tantek: So what you need is a name for the thing that's a NOTE.
  [bikeshedding: css21-future, css21-next, css21-stage]
  astearns: Every time we make a change to the ED, we publish an
            update to the NOTE
  astearns: Once we're sure a change is solid, we move it to the REC.
  tantek: So once the NOTE is stable you move it to REC.
  astearns: No, each individual change. We cherry pick the change.
  astearns: We'll probably batch things a little bit, like every
            quarter move things over.
  Florian: But we could cycle faster if needed.
  astearns: The AC might get annoyed at us.
  Florian: That might be a good problem to have.
  [bikeshedding: css21-cooking, css21-oven]

  dbaron: Btw, the new process allows AC ballots to start at CR
  <liam> [yes, AC ballots now routinely start at CR]
  fantasai: Well, this is a PER cycle, not PR

  astearns: Any other issues?
  Florian: Need to resolve on shortname
  Florian: css21-testing?
  tantek: I'd like a flow chart.

  scribe: astearns

  fantasai: Four copies of the spec - two in drafts, two in TR
  fantasai: have some errata - put it in css21-testing, gets
            published to TR as a NOTE.
  fantasai: This gets listed in the changes section with
            implementation status
  fantasai: as each individual change passes tests, it gets (by
            itself) added to the REC draft.
  fantasai: This gets published to TR and does a PER->REC cycle
  Bert: This only works if the changes are editorial.
  SteveZ: No, it's OK.
  fantasai: Pretty sure it's OK.
  Florian: What role does the NOTE serve in the process?
  fantasai: Process doesn't care about NOTE - we're just editing the
            REC and using the NOTE for staging changes
  <dbaron> https://www.w3.org/2017/Process-20170301/#rec-edited
  dbaron: Last two process revs changed some things.
  dbaron: March 1st says PER doesn't exist.
  dbaron: Two options now - editorial changes with consensus can get
          republished REC with dir's approval
  dbaron: Substantive changes or dissent means a CR.
  <dbaron> https://lists.w3.org/Archives/Public/spec-prod/2017JanMar/0002.html
           is the announcement of the new process

  Scribe: fantasai

  [more process discussion]
  [how long we're required to stay in various stages for Process

  Florian: This is just process churn, which may be justified, but
           isn't something we need to deal with in any case
  *fantasai thinks that rule has been proposed before.
  fantasai: We should just go with this, and ignore the extra work
            behind the scenes by W3C Staff. If it becomes annoying
            for them, they can petition the AB to streamline the
            process of REC maintenance.

  Florian: What do we do with CSS22?
  fantasai: Redirect it to css21-testing
  tantek: Why do we have CSS22?
  fantasai: It was easier to publish a WD than update the REC.

  RESOLVED: Create CSS2-testing as a NOTE

  [discussion about what CSS2's shortname is]
  Florian: We could do a 301 redirect from CSS21 to CSS2

  RESOLVED: 301 redirect from /TR/CSS21/ to /TR/CSS2/, and from
            /TR/CSS22/ to /TR/CSS2-testing/
  RESOLVED: Cherry-pick changes from CSS2-testing/ that have tests
            and 2 impls and port to /TR/CSS2/, Process-wrangling to
            be handled by our Team Contacts.

  ACTION: fantasai Write a wiki page describing this process
  <trackbot> Created ACTION-822

What can go to CR this quarter?

  <fantasai> https://www.w3.org/Style/CSS/current-work
  Florian: What's blocking css-contain?
  astearns: FPWD?
  Florian: Does anyone else have comments?
  <dbaron> https://test.csswg.org/shepherd/testcase/::07e433f30a88/contain-paint-clip-002/owner/dbaron/status/issue/

  tantek: Let's figure out which specs should go to CR this quarter,
          so we can prioritize them on the agenda.
  Florian: MQ is blocked on me and Tab dealing with DoC.
  Florian: Just need to find the time to do it.

  fantasai: Box Alignment still has some issues open, but should be
            possible this quarter. I forget which issues they are at
            the moment.

  Florian: Containment has 2 open issues.
  Florian: Should be able to do CR.
  plinss: Still needs FPWD.
  fantasai: Kidnap TabAtkins.

  tantek: Owner for multicol?
  Florian: I'm standing in for that, in theory.
  tantek: ED has a lot of changes.
  <tantek> https://drafts.csswg.org/css-multicol/
  fantasai: Should be substantially similar to current CR.
  Florian: Difference between ED and CR is unclear.
  Florian: So we need a Changes section, and DoC, and if either of
           these surfaces something large maybe need a WD.
  Florian: If relatively small, then CR is ok.
  tantek: So we know what to do, but not the timing.
  Florian: I'm happy to work on it eventually, but not a high
           priority atm

  fantasai: I think for me the top specs that need advancement are:
  fantasai: second CR for Grid
  fantasai: CR for Box Alignment and Text
  fantasai: second CR for Text Decoration 3
  fantasai: For Text, need about 2 days of editing work.
  fantasai: Selectors is blocked.
  Florian: The thing between TabAtkins and dbaron?
  fantasai: Yeah.
  fantasai: If dbaron can just do the rewrite, that should unblock
            the spec.
  fantasai: Sizing and Display are close to CR, but need discussion
  fantasai: And then I'm supposed to do FPWD of Text Decoration 4
            and Value and Units 4.
  fantasai: And FPWD for Paint, I think.
  fantasai: and css-rhythm.
  tantek: Maybe we don't need to add more CRs to that list...
  <tantek> Paint and CSS-Rhythm are FPWDs then right?
  <fantasai> yeah

  Florian: will-change?
  fantasai: Is CR 2015.
  astearns: This is a good list. I can use that for prioritization.
  <dbaron> there's also transforms, transitions, and animations
  tantek: Anything to go to PR?
  astearns: Shapes needs a new CR.
  fantasai: Conditional Rules needs a new CR as well.

  fantasai: Have an outstanding resolution to take Transitions to CR.
  dbaron: Need to recompile issues list... don't like our tools for
  dbaron: Birtles has been working on Transitions.
  dbaron: There's a problematic edit that we were actioned to make
          that no implementer was willing to make
  astearns: That should be brought up with the WG again.
  dbaron: Did that twice and failed. Could try a third time.
  Florian: Everyone is refusing to implement a change and refusing
           to change the spec?
  dbaron: This was about non-interpolable properties and switching
          at halfway.
  dbaron: Group succeeded in changing for Animations, but
          Transitions are a problem.
  <birtles> we are implementing that in Gecko but expecting it to
            have to be backed out due to compatibility issues
  <birtles> (for transitions)
  dbaron: That's one of the blocking issues on Transitions.
  fremy: We made the change for custom properties.
  dbaron: Yeah, I encouraged birtles to make the change to prove it
          wasn't going to work.
  fremy: For custom properties it would be a problem if it works to
         animate in a browser that doesn't support typing, at least
         it'll flip.
  fremy: No opinion for other properties.
  astearns: Sounds like once you get some compat data, needs to be
  tantek: Doesn't sound like a good bet for CR this quarter.
  fantasai: Would be good to get the WD updated.
  dbaron: I think most of the edits were made, unsure if the option
          G-beta was done. Maybe birtles did it.
  <birtles> not yet
  <birtles> will do it soon
  tantek: Would like to get updated stuff that reflects reality.
          Care less about WD vs. CR
  dbaron: Although Gbeta was animations rather than transitions...

  fantasai: We should have a goal of having all WDs updated this
            quarter. No reason not to, now that we have the tools.
  <tantek> That is a good goal
  astearns: I'm happy to call a resolution for publication update.
  fantasai: I'd make an exception for Selectors, since it hasn't
            been reviewed
  fantasai: But otherwise yeah.
  fantasai: Would like to keep the invariant that drafts on /TR
            have, at the point of publication, the consensus of the

  tantek: A couple of CRs that we should consider taking to PR.
  tantek: One is Flexbox.
  fantasai: I think there's actually open issues on Flexbox.
  dbaron: I'd also like to see Ruby in CR.
  fantasai: Ask me again in Tokyo :)
  astearns: How about asking Xidorn to become a co-editor?
  fantasai: That might be a good idea.
  <fantasai> xidorn, what do you think? :)
  <xidorn> fantasai: I think I'm fine, although I've never been an
           editor before
  <fantasai> I'm happy to help you out, of course. ^_^
  <xidorn> fantasai: thanks
  <astearns> if you are willing to take the time to try, I'd love to
             assign this to you :)
  <xidorn> yeah, I'd like to try :)
  <tantek> xidorn++
  <fantasai> xidorn, I'd love to have you!

  RESOLVED: add Xidorn as editor of css-ruby

  <fantasai> \^_^/

  tantek: scroll snap?
  fantasai: No impls
  tantek: Tests?
  gsnedders: There was some plan to dump a load of tests from blink,
             from what I understood.

  dbaron: Syntax?
  gsnedders: Some of it is hard to directly test.
  tantek: So we think there's implementations for Syntax, but not
  tantek: So Syntax PR is gated on tests.

  [discussion of testing strategy]
  <tantek> gsnedders says we have atomic tests for Flexbox but not
           feature interaction tests
  <tantek> possible to get Flexbox to PR this quarter?
  gsnedders: Despite what test harness says, Firefox fails half the
             writing-modes tests because of anti-aliasing.
  gsnedders: That's a bug in the test.
  gsnedders: It affects several hundred tests.
  gsnedders: Yes, we can go to PR that the test tests the feature
             but is failing because of anti-aliasing.
  fantasai: I think you can run those tests manually, which gtalbot
            has done.
  gsnedders: Yeah. Which is why the test harness results don't match
             what I got when I ran them automatedly.
  astearns: Not very useful to have these tests failing.
  gsnedders: Yeah, 'cuz can't run them in continuous integration.
  gsnedders: I think we can fix this.
  gsnedders: Not sure which ones are failures due to image scaling
             due to me being on a high-DPI screen.
  dbaron: There's a pref to flip that off.
  gsnedders: Fixing an individual test isn't much work... but there
             are hundreds.
  fantasai: If you explain the general issue and its solution,
            I'm pretty sure GĂ©rard can fix the tests.
  dbaron: Also some tests that are not anti-aliasing, but reference
          might be wrong.
  dbaron: e.g. English description is true, but test and reference
          don't match.
  dbaron: Why is test.csswg.org slow?
  plinss: Looking into it.

  fantasai: Status of css contain?
  TabAtkins: Should go to CR
  Florian: There are 2 open issues.
  astearns: I appoint Florian to co-edit

  RESOLVED: Florian co-edits css-contain

  fantasai: You know what should be in PR? Backgrounds and Borders.
  astearns: Blocked on tests?
  fantasai: yeah.


  gsnedders: Status of what we agreed in Lisbon is that the majority
             of the things we agreed to do before the merge are now
  gsnedders: There needs to be a small amount of work finalizing
             some introductory git documentation.
  gsnedders: I think we're ready to do the merge of CSSWG tests into
  gsnedders: I'm going to be able to write the documentation needed
             this month.
  gsnedders: Would like to get the merge done by mid-February at the
  gsnedders: We have an intro to git and stuff.
  Florian: I added the part about how to do reviews.
  Florian: Can be improved, but we have an intro to git for the
           entire process.

  gsnedders: Would like resolution to merge.
  gsnedders: Tests will be in css subdir of WPT, and essentially
             everything as-is.
  gsnedders: In subdirectory because some things that only apply to
             those as a result of keeping the build system working
  fantasai: Does this mean we have to pull down the entire WPT repo
            to edit a test?
  gsnedders: Yeah.
  astearns: There is such a think of a sparse checkout.
  gsnedders: It's around 450MB.
  fantasai: That's not too bad then.

  Florian: What happens to status of all the tests that have been
           merged but not reviewed? How do we tracked that?
  gsnedders: We don't.
  Florian: Browsers are expected to file bugs against the test suite
           to fix that?
  gsnedders: Yeah.

  Florian: What about the comments in shepherd?
  gsnedders: The agreement was to make shepherd read-only, and at
             some point import into github in some way.
  plinss: Don't remember if we were going to try to keep shepherd
          updated or just freeze it.
  gsnedders: I think we were going to freeze it.
  gsnedders: Could in theory, but probably not worth the effort.
  plinss: Shepherd has a lot of metadata searching and other useful
  gsnedders: Some of the invariants in shepherd aren't honored in
             WPT, e.g. unique filenames.
  plinss: It's not quite required, shepherd tracks those as variants
          of the same test.
  astearns: Any objections to making the move now?

  RESOLVED: Move CSSWG tests to WPT

  Bert: What do we do about the many tests that are wrong in
        backgrounds and borders test suite?
  <Bert> http://test.csswg.org/harness/test/css-backgrounds-3_dev/origin-border-box/
  Florian: So whoever takes on task of making test suite for
           backgrounds and borders can use these as an inspiration.
  gsnedders: Or just make a ref.
  Florian: Yes, that's one way of doing it.


  Florian: Both Blink and Gecko have a almost completed
           implementation of caret-color.
  Florian: That made me think about what was left before PR.
  Florian: We have a not complete but reasonably substantial test
  Florian: Would like to ask gsnedders to give an opinion on how
           complete the test suite is.
  Florian: Then with one exception, I think if we exclude all things
           at-risk we're in good shape.
  Florian: Need to do a DoC, file a few bugs, and then pretty close
           to PR.
  Florian: Having not competed the TS, not quite ready, but getting

  ACTION: gsnedders Review CSS3 UI test suite to see how complete it
          is, if there are substantial parts missing.
  <trackbot> Created ACTION-823

  Florian: Things that are at-risk are the string value of the
           text-overflow property, the double value of the
           text-overflow property.
  Florian: They are implemented in Gecko but not anywhere else.
  Florian: Would like feedback from other implementers on their
           opinions of this features.
  Florian: Should we keep these in the draft for soon-ish
           implementation, or push them out?
  tantek: Question for Edge, Webkit, Blink.
  tantek: These features allow picking the string used for the
  tantek: and allow having ellipsis for overflow in both directions
          rather than just the end.
  fremy: Not a priority.
  myles: No comment.

  tantek: Part of what drove these features were i18n feedback
  tantek: e.g. mixed rtl flow of text, might need to specify what
          happens on either side
  tantek: and also not all languages use &hellip;
  tantek: or not all UI situations.
  tantek: We left those in there long enough, if there really isn't
          implementer interest.
  Florian: We have a spec and L4, can move them don't have to delete.
  astearns: Given lack of enthusiasm, sounds like that's the way

  Florian: We have an at-risk of MUST to SHOULD about the behavior
           we expect when you put a negative offset on the outline.
  Florian: This has not seen implementer interest as far as I can
  Florian: There is an outline-offset property. It accepts negative
  Florian: Implementations do different weird things. We have a MUST
           requirement that to not do the really bad things.
  Florian: e.g. if the negative offset is too big, don't make it
  Florian: since it's used as a focus indicator?
  <Florian> https://drafts.csswg.org/css-ui-3/#outline-offset
  <astearns> https://lists.w3.org/Archives/Public/www-style/2017Jan/0019.html

  Florian: Other thing is nav-up/left/right/down
  Florian: These were not at-risk because we had two impls.
  Florian: One was Presto. Which is not going to make any interop
  Florian: Others are TVs.
  Florian: Don't know if they have independent impls, or how
           spec-compliant they are.
  tantek: We think there are both WebKit and Blink forks with this
  tantek: Not in Safari or Chrome, but think it exists in forks of
  tantek: Not sure how to deal with that, cuz dunno how to test them.
  Florian: There's a 2-3 year old test build from Samsung that has
  Florian: If the spec was stuck anyway, could stay in. But don't
           want to be stuck on this.
  Florian: Thinking it should be deferred for L4.
  tantek: Worried that dropping nav stuff will raise objections from
          W3C Members like LJ or Samsung.
  tantek: We could [...]
  tantek: Or we could re-issue CR with those properties at-risk, say
          we don't know how to test, and pressure ACs to vote and
          comment on CRs.

  RESOLVED: Defer text-overflow double value and string value
            features to L4
  RESOLVED: MUST->SHOULD for negative outline-offset
  RESOLVED: nav-* at-risk
  RESOLVED: caret-color no longer at risk

  ACTION Florian DoC for CSS3 UI for new CR
  <trackbot> Created ACTION-824

  Meeting adjourned.
 <div id="DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2"><br />
<table style="border-top: 1px solid #D3D4DE;">
        <td style="width: 55px; padding-top: 13px;"><a
width="46" height="29" style="width: 46px; height: 29px;" /></a></td>
		<td style="width: 470px; padding-top: 12px; color: #41424e;
font-size: 13px; font-family: Arial, Helvetica, sans-serif;
line-height: 18px;">Virus-free. <a
target="_blank" style="color: #4453ea;">www.avast.com</a>
</table><a href="#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2" width="1"
Received on Tuesday, 14 February 2017 02:00:20 UTC

This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:09:06 UTC