- From: Dael Jackson <daelcss@gmail.com>
- Date: Wed, 22 Mar 2017 21:11:19 -0400
- 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.
=========================================
REC spec steps tête-à-tête
--------------------------
- Cascade 3
- fantasai reported that there are edits that need to be made.
- gregwhitwoth with port the Microsoft tests into the test
suite.
- Conditional Rules
- dbaron & zcorpan need to incorporate edits.
- Values & Units
- Approval for the <position> edits will be added to next
week's agenda.
- Background and Borders
- There are working group approved edits that need to be made
to the spec and then have a republish.
- tmichel will put together an implementation report.
- Transforms
- About half the open issues need feedback from people who
have worked on SVG. smfr will reach out members of the old
SVG working group to set up a call, perhaps during the old
SVG telecon time.
- smfr will need assistance in evaluating the test suite.
- Flexbox
- Open items are actively being discussed on telecons
(including today) and after that there will be a
republication ask.
- gregwhitworth and/or gsnedders will evaluate the test suite
coverage.
- CSS UI
- astearns and fantasai will split the work of reviewing the
tests Florian submitted.
Publication of Paint as FPWD
----------------------------
- RESOLVED: Publish FPWD for CSS Paint
- RESOLVED: Short name is fill-stroke with a title of Fill and
Stroke
Flexbox
-------
- Though there was agreement that the use case in issue 401
(https://github.com/w3c/csswg-drafts/issues/401) was broken by
the change, the group felt that the change was throughly
considered and still the right choice.
- RESOLVED: No change on issue 401
- fremy will open bugs on those browsers that do not conform to
spec when sizing images with intrinsic aspect ratio. These
bugs will reference issue 1112
(https://github.com/w3c/csswg-drafts/issues/1112)
in case a behavior change is needed.
- fantasai requested more review of the flex basis and box-sizing
issue (https://github.com/w3c/csswg-drafts/issues/316) to see
if there's actually a compat risk to fixing the behavior to be
as intended.
- The remaining issues that need discussion will be added to next
week's agenda so that Flexbox can be republished.
===== FULL MINUTES BELOW ======
Agenda: https://lists.w3.org/Archives/Public/www-style/2017Mar/0067.html
Present:
Rachel Andrew
Rossen Atanassov
David Baron
Bert Bos
Tantek Çelik
Alex Critchfield
Benjamin De Cock
Emil Eklund
Elika Etemad
Simon Fraser
Tony Graham
Dael Jackson
Brad Kemper
Vladamir Levantovsky
Chris Lilley (IRC Only)
Peter Linss
Myles Maxfield
Thierry Michel
Michael Miller
Theresa O'Connor
Anton Prowse
Matt Rakow
Melanie Richards
Hiroshi Sakakibara
Jen Simmons
Geoffrey Sneddon
Alan Stearns
Greg Whitworth
Regrets:
Dave Cramer
Daniel Glazman
Florian Rivoal
Scribe: dael
astearns: Let's get going
astearns: Is there anything anyone would like to add to the agenda?
<fantasai> astearns: Publication of Paint as FPWD?
astearns: I'll add the publication as the item after rec steps.
Thanks fantasai.
REC spec steps tête-à-tête
==========================
astearns: We had a bunch of chatter on public & private lists for
some including fonts and variables.
astearns: There are a few blanks we need to add in.
Cascade 3
---------
astearns: So does anyone know next steps for cascade 3?
<gregwhitworth> testsuite was what we discussed last right?
<fantasai> astearns, cascade 3 needs an edit to drop scoped styles
and it needs a testsuite
astearns: [reads gregwhitworth & fantasai]
astearns: I'll put that as next steps that we'll have those edits
to reduce scope. Who is going to work on test suite?
<fantasai> astearns, tests exist, but perhaps not in the right
format, e.g. IIRC dbaron said Mozilla's tests are in
mochitest format
<dbaron> reducing the scope, eh?
gregwhitworth: We reviewed last time. We have extensive tests we
can provide, but need to port. There is one thing
in cascade we don't support.
astearns: Is that on you?
gregwhitworth: Yeah. Sure.
<fantasai> what's unsupported?
astearns: [reads fantasai ]
astearns: Can we get a volunteer to convert those?
astearns: Lacking a volunteer we'll find someone. We have three
next steps.
Conditional Rules
-----------------
astearns: Conditional rules. What's the next step?
<fantasai> astearns, Conditional Rules needs republication
<fantasai> astearns, it's taken some edits. Has outstanding
resolution to publish
dbaron: I suspect needing tests is at least a big piece.
dbaron: I haven't looked at the state of the suite.
astearns: [reads fantasai]
astearns: fantasai, what needs to be changed?
<fantasai> astearns, yeah, on the call, can't speak atm, dbaron
should have updates
dbaron: I have another thing. The OM stuff needs work. They have
competing models and we really need some pieces of one and
some of the other.
astearns: So outstanding edits, can I add you, dbaron, as the
person to review and edit?
dbaron: Me and zcorpan probably.
dbaron: It might come down to deleting stuff from conditional
rules because OM is in a better state. We also need to
figure out which spec should have it.
astearns: What about test suite? Do we have tests in various
engines?
dbaron: I'm sure Gecko has some, I don't know what format or
amount of conversion.
astearns: Can you find someone to make that assessment?
<gregwhitworth> I'm sure we all have tests, converting them all to
work in the wg is hard to promise
dbaron: I could eventually. I'm at a state where I can't promise
anything before next F2F
<gregwhitworth> this is me too
astearns: That's fair. It's good to have that there are test to
convert, even without a person.
astearns: [reads gregwhitworth]
Values & Units
---------------
astearns: On to Values & Units. What needs to be done there?
astearns: Who should know?
<fantasai> astearns, needs republication after <position> edits
<fantasai> astearns, also needs a test suite compiled
astearns: [reads fantasai]
<fantasai> <position> edits are done; needs review
Rossen: Which <position> edits?
<fantasai> https://lists.w3.org/Archives/Public/www-style/2017Mar/0054.html
astearns: [reads fantasai]
astearns: Looks like we need the edits on the agenda and then
republish. I'll put that on me to add to the agenda next
week.
<fantasai> astearns, see link above for edits & commentary on said
edits
Backgrounds & Borders
---------------------
astearns: Backgrounds & borders has had test suite work. It would
be great to have someone review the state of the test
suite and see if recent edits have been good.
astearns: Who can take a look at the B&B tests?
<fantasai> spec needs republication as well; a handful of edits
<fantasai> but nothing significant
Rossen: Have there been changes recently? It seems the spec has
been stable for quite some time.
Rossen: It seems to have fallen off the radar. There's quite a bit
of interop and we need to button up the test suite.
<fantasai> https://drafts.csswg.org/css-backgrounds-3/#changes
Rossen: I see fantasai points out there's a handful of edits.
Rossen: I'm assuming they're mostly editorial and I'm curious to
know if we need any kind of resolution or if they're the
result of resolutions.
<fantasai> They're not editorial, but they're WG-resolved.
<fantasai> but again, see message wrt <position>
<dbaron> Mozilla has a lot of B&B tests, some of which may have
been contributed to the WG before the WG accepted
reftests... and thus might need to be re-contributed...
astearns: [reads fantasai]
Rossen: Okay.
Rossen: In that case is those are shovel ready we just need to
make it happen and republish.
astearns: That and get someone to review test suite.
Rossen: Right.
Rossen: I think the test suite is on everyone. I would be
surprised if there's an impl on call that doesn't use
testing as part of their practice. Those are the tests we
want.
astearns: But assigning evaluation to everyone doesn't get us
anywhere. Getting one person to evaluate and put a
report together if we're ready is good.
Rossen: Right. I mis-understood.
astearns: From IRC there's additional tests for Mozilla. We need
more test review in this process because each spec needs
review of if the test suite is in a good state.
Rossen: I see bradk, Bert, and fantasai as editors.
Rossen: bradk or Bert is this something you can take as an action?
tmichel: A few months ago I worked on the test suite and tried to
evaluate coverage.
tmichel: My opinion is that most of the features are covered but
today we have about 80% of the tests passing 2 impl.
tmichel: The impl work is pending. It's improved a lot in a year.
astearns: tmichel can you put together an impl report to show
what's not passing?
tmichel: Yeah, I can do that.
astearns: Thank you.
tmichel: Are you interested in only the ones that are not two, or
are you interested in only one?
astearns: All tests with NOT 2.
tmichel: Thank you
<bradk> thanks tmichel
Transforms
----------
astearns: Transforms. What is the next step? There were a bunch of
issues resolved in seattle.
smfr: There's 27 open issues. A fair portion need SVG work. We'll
have to go through those with SVG representation. I don't
know what that will be.
Rossen: We have a large bit of SVG representation.
fantasai: I think we should make an explicit effort to pull in the
former SVG WG people that aren't here and get them
involved. We need to coordinate a specific SVG meeting
or telecon. Maybe we can slot it into the old SVG time
slot. We won't be able to get them all to the F2F. We
could probably do a telecon.
Rossen: I like that idea.
Rossen: I think sending an e-mail to the SVG ML to get things
going would be great. We'll see if we can re-use the SVG
telecon.
Rossen: smfr out of the 25 how many need SVG? All?
smfr: Maybe 12 or 15-ish?
Rossen: About half.
fantasai: Paint has a bunch of issues that need SVG too.
Rossen: That's fine. Sounds great. Let's reach out and go from
there.
Rossen: We didn't cover...do we have any test suite for transforms
L1? If not can we get that moving?
smfr: I think we have some contributed, but I don't believe
they've been brought together for review. I don't have
bandwidth for that.
Rossen: So you'll need help evaluating what's there and do another
round of calls for tests.
gsnedders: From memory there were problems with references not
matching.
Rossen: Okay.
Rossen: gsnedders is this something you can run through and see if
there are any good tests and next week we can discuss?
astearns: gsnedders needs to finish the web-platform-tests
transaction first.
Rossen: Fair enough.
<gregwhitworth> when does that happen?
gsnedders: Plan is to do it on Tuesday.
gregwhitworth: Will we be working out of wpt at that point?
gsnedders: Yes.
gregwhitworth: Okay, cool.
<gsnedders> gregwhitworth I'll send out a further email on Friday
<gregwhitworth> gsnedders thank you so much!!
<gsnedders> gregwhitworth confirming that yes, it's going ahead
roughly when I said so (okay okay a day late)
Flexbox
-------
astearns: Next is flexbox. It looks like someone recently
reviewed, but no one can remember where that evaluation
went. Did that happen?
astearns: And is that the next step?
fantasai: It needs republication. There's a couple of open issues.
We do need the test suite compiled and evaluated. All
these need to happen.
astearns: A couple of the flex issues are on agenda. Is there
anyone that can volunteer for test suite?
<skk> Regarding test (not specific to any specs), I read minutes
that there lacks QA people in WG. If it's OK, I want to
announce broadly as much as possible that in Japan. If there
is a website regarding this, it's easier for me to tell
people in JP.
skk: Regarding what I put in IRC...There's no website. If there's
a website it's easier for me to announce that we need QA in
Japan.
<dbaron> https://wiki.csswg.org/test ?
<gsnedders> dbaron: and http://web-platform-test.org/
astearns: Let me talk to you off the call for what would help with
recruiting QA people in Japan.
<skk> It's happy to communicate in email. Thanks.
<fantasai> skk, astearns, feel free to loop me in; gsnedders might
be helpful, too, he knows where existing docs are :)
astearns: We still need a volunteer.
gregwhitworth: I can do both the list dive and the review. I can
see what we have and what we have internally.
gsnedders: I've also spoken with people at looking in the next
month.
astearns: Depending on gregwhitworth timing perhaps you two can
collaborate.
CSS UI
------
astearns: Next is UI. I believe next step is review test suite
from florian
<tantek> someone other than the editors preferably :)
astearns: This is smaller, this is just getting florian's
submissions into the suite. And as tantek mentioned it
should be a non-editor.
astearns: fantasai, are you and editor?
fantasai: Nope.
astearns: Can I assign the review to you?
fantasai: Um...I guess.
fantasai: My current priority is to get grid, flexbox, and
display:all published.
astearns: I'm hoping this is a small task. It's reviewing
florian's work.
<tantek> I'm sensing that fantasai is already quite overloaded,
would help to get another volunteer!
<tantek> also this feels like something that a new-ish person
could review
<tantek> so if there's a new WG member that wants to try an "easy"
review task, this would be a great one to do!
<tantek> we have a few new WG members right?
gsnedders: In a general sense I'd like us to land as many open
pull requests as possible in the next week as pulling
them over in manual work.
astearns: That is true. And I doubt fantasai will get to it in
that time. It would be great to have another volunteer.
[silence]
fantasai: I'll give it a try tomorrow.
astearns: Once you do look fantasai see if there's anything you
and I can split up. I'll volunteer.
gsnedders: I was planning on looking at the 2.2 test in the next
few days because they have silly amounts of comments
and I don't want to move them over.
<gsnedders> https://github.com/w3c/csswg-test/pull/1139
astearns: So as you look at 2.1 tests if there's anything you can
throw to me I'll try and help.
<bdc> (I'm also happy to have a look, although I'm not entirely
sure yet how all of this works)
<gsnedders> bdc: http://web-platform-tests.org/reviewing-tests/index.html
astearns: So that's everything. We have next steps for all specs
on the list. I'll send a summary to the private list.
We'll try to make this quicker and easier as we go.
Publication of Paint as FPWD
============================
<fantasai> https://lists.w3.org/Archives/Public/www-style/2017Mar/0059.html
fantasai: TabAtkins and I think we're ready. We merged in heycam's
paint spec. We finished last week. There's a ton of
issues, but I think it's ready for FPWD.
astearns: Anyone have anything to add?
Rossen: Did anyone besides the editors review?
<ChrisL> I can take a look at Paint.
fantasai: Amelia sent a bunch. Someone else sent some changes. We
should look at that set as an issue, but it shouldn't
hold up FPWD. It was mostly syntax adjustment.
<ChrisL> +1 to fpwd
fantasai: We did go over it in detail in, I believe it was the
Sydney F2F so the design of features has been discussed
with WG.
Rossen: Thanks.
astearns: Objections to FPWD for Paint?
<tantek> ship it!
RESOLVED: Publish FPWD for CSS Paint
fantasai: Can we get short name approval?
astearns: Are we going with paint?
fantasai: Yeah. Someone, used to be plh, has to officially notify
the webmaster it's okay.
<tmichel> yes Plh is still the right person
<ChrisL> shortname approval is part of the fpwd transition request
astearns: plh is still right according to tmichel and ChrisL said
we put the short name as part of the request. Objections
to paint as the short name?
RESOLVED: Short name is paint.
<ChrisL> ok I will do the transition request
Rossen: And the name will be ... ?
dbaron: Sorry, I had a hard time getting off mute. Is short name
css-paint or paint?
fantasai: Just paint. That's the naming convention of fxtf.
dbaron: It's a broad short name.
fantasai: It's fill and stroke properties from svg.
dbaron: Okay...I guess.
<tantek> I thought Houdini had a Paint also
<ChrisL> paint api
<tantek> this will be confusing
fantasai: I'll take alternatives. I just don't have one.
Rossen: fxtf-paint?
dbaron: Let's not put fxtf in the name of a spec when it may be
about to disappear.
<gsnedders> +1 on what dbaron said
Rossen: Yes, but I strongly agree paint is broad.
dbaron: Paint is fine. Let's stick with it.
fantasai: If there are better ideas let us know and we'll prepare
the publication next week. Post the better idea to the
ML.
Rossen: Module is fill and stroke?
fantasai: Yes. That's clearer and easier for people. But as a
short name 'fill and stroke' is a bit long.
<tantek> why not "fill-stroke" then :P
tantek: I'm a little concerned about confusion with Houdini's
paint. If this does not have something to do with
Houdini's we shouldn't have short names that make it seem
that way.
<Bert> (good point, tantek)
Rossen: fill-stroke seems pretty good.
astearns: Yep. Any concerns on fill-stroke?
fantasai: My only concern is if that ends up too specific. I'm not
100% sure.
astearns: More specific is probably better than too general.
fantasai: True.
astearns: Objections to short name of fill-stroke?
<bdc> +1
RESOLVED: Short name is fill-stroke with a title of Fill and Stroke
Flexbox
=======
Absolute Positioned Elements in Flex Containers Removes Common Layout
Patterns
---------------------------------------------------------------------
<astearns> https://lists.w3.org/Archives/Public/www-style/2017Mar/0067.html
<fantasai> https://github.com/w3c/csswg-drafts/issues/401
fantasai: We got an issue where somebody is complaining about a
change we made. In an earlier draft we have a behavior
that static position is where it would have been if it
was in flow. The major problem with that is
implementations were making it take space when you were
using space around or other alignment properties.
fantasai: That kind of violates the idea that abspos things
shouldn't effect flow layout. There was lots of
suggestions and discussion and we ended up with the
current spec.
fantasai: This broke some use cases. This was a library with
vertical dividers and they did it with abspos and they
couldn't do that anymore.
<dbaron> what do implementations do today?
<dbaron> I'm skeptical that we can just go change behavior for
this stuff at this point
fantasai: They filed an issue. They're unhappy with the change. I
don't have a strong opinion, but implementor have an
incentive not to change and authors are frustrated they
can't get the static position they want.
fantasai: This is for the WG to discuss.
dbaron: What do implementations do today?
Rossen: We all match the behavior we agreed on. We started by
computing static position as was requested by this issue.
Later we reverted to 0,0 when order came into play. That
added impl complexity. At this point all impl compute
static position as the origin of the flexbox. For what
it's worth there's already some content dependency on this
given the amount of flexbox content.
<bdc> safari does the right thing imo
Rossen: In short, we all do what the spec says. And there's
content that will break if we change impl.
Rossen: Also, the 0,0 static position was not random. There were
long discussions. That was not a quick decision to wave it
off.
fantasai: I think it was largely because there were a lot of
complications on meaning for 'where it would have been'.
There was a justification problem where if you justify
and there's an abspos in the middle where does it go. We
punted all this complexity and defined static position
as if the abspos were the only item in the flex
container.
Rossen: Correct. Also one thing worth pointing out is there's an
easy work around to achieve the same by adding the 0 size
flexbox with relative position and have whatever you want
inside to achieve the same behavior. It sucks to add
another element, but it's a totally achievable work around.
<rachelandrew> It seems to be an issue with this particular
library - I'm not hearing the same issue a lot from
other authors
astearns: I'm not hearing any impl interest in changing behavior
to help this use case. Does anyone want to speak in
favor of this change?
eae: We'd rather not change.
fantasai: If we're starting over we could dig in further, but at
this point it will be quite difficult to make a change
for something that's not super compelling. Another issue
is I don't know how you interpret this in grid and we
try and keep them consistent.
astearns: In effect we're resolving the static position of abspos
elements is different in flex and grid then in previous
layout systems.
Rossen: But that resolution has been recorded and made in the past.
<tantek> resolving to keep prev resolution is usually a good thing
astearns: Right. We're resolving no change and that out new layout
system is different from those in the past.
astearns: So objections to resolving no change?
RESOLVED: No change on issue 401
Sizing images with intrinsic aspect-ratio: ¿harmonize with grids?
-----------------------------------------------------------------
<astearns> https://github.com/w3c/csswg-drafts/issues/1112
fremy: Basically this issue came up when I was pursuing Manuel
Rego's question on the images change in grid. He asked if
this applied to flex. I was wondering what we do and I
found we aren't interop.
fremy: When we have images the decision to flex is different
across browsers. I wanted to point it out. It seems like
Edge is per spec. I wanted to make sure people are okay
with spec and, if so, we can file a bug.
astearns: Opinions on FF and Chrome behavior?
fremy: I think if no one reviewed the test case it's unlike
there's a conclusion. It's a good time to ping people and
make sure there's review.
astearns: I think next step is for you to write bugs.
fremy: That's fine. I can do that.
dbaron: If you write bug, please link to this issue in both
directions.
fremy: Okay
fantasai: I think there's two things where there are bugs, but if
anyone wants to think about what we should be doing
here...there is a distinction in flexbox where if you're
larger then intrinsic you get one behavior and smaller
is a different behavior. Anyone with an interest can
look at the behavior as spec and do we need to tweak
flex, or grid, or leave as-is. That's the question at
hand.
ACTION fremy to write bugs on other browsers to get a response on
if they can change to match spec (for issue #1112)
<trackbot> Created ACTION-837
Flex basis and box-sizing
-------------------------
<fantasai> https://github.com/w3c/csswg-drafts/issues/316
fantasai: There's another flex issue that needs input ^
fantasai: Not to resolve now.
fantasai: We have a problem with how flexbox works. The goal was
if you say flex and then an int the goal is the flexbox
would take that int's space. So if you want it even you
can do all 1, or you could do 2 and 1 to get thirds. But
that doesn't work right now because the effective flex
basis is not 0, but the sum of the margins and borders
and padding.
fantasai: It would be nice to fix the spec to get expected
behavior of exact proportions. I don't know if that's
web compat or how to evaluate this.
astearns: So you would get the precise behavior only if box-sizing
is set to border-box?
fantasai: For sure that would work. Flex basis is set to follow
content box-sizing. If you're using border box sizing it
should account for the entire border box.
dbaron: I guess some of the question is what stage can things go
negative. Normally with box-sizing we have that the
content box can't be negative.
fantasai: I'm not sure what case would be negative.
dbaron: Flex basis of 0 and border box then your content box would
be negative.
astearns: If you have borders.
fantasai: Yeah.
<dbaron> It's also not clear to me if we could take a compat hit
here at this point
<tantek> yeah
fantasai: End result to maintain the variant where content box is
never negative the flex would have to impose the sum of
borders and padding as a minimum. So content box won't
be negative but flex basis used value could be negative.
We use the outer size of the flex item in calculations.
astearns: Both dbaron and TabAtkins mentioned that for compat we
probably can't change. I'm thinking this might be next
level of flexbox to have an additional switch to get the
desired behavior.
fantasai: I'm worried, but I'm not sure we're in that situation. I
think more flexbox use an equal number instead of aiming
for a ratio and that's the only cases where you would
have a change. If the padding is small comparatively you
wouldn't really notice the change. It might be an issue,
it might not. I do think we really screwed up in
relation to the intention for it.
fantasai: I'm interested in what people think about compat and
what's a good way to evaluate.
astearns: Alright. Thanks for bringing this up. It's good to get
people thinking.
Publication
-----------
asteanrs: Next topic is percentage [max-]width|height and
intrinsic sizes.
astearns: We have 3 minutes
fantasai: That's scary. Let's just resolve to republish flexbox as
it's out of date?
<tantek> yes please resolve to republish
astearns: That's possible. It's CR?
fantasai: Yes.
fantasai: Let me pull the issue list.
<fantasai> https://drafts.csswg.org/css-flexbox/issues-cr-20160526
fantasai: I do have a DoC, but we have a couple more issues.
astearns: We're close but not quite for republish.
<tantek> ok
astearns: Let's get those issue on next week's agenda.
astearns: I think we should call it for the week, looking at the
remaining issues.
astearns: Thanks everyone for calling in. I'll follow-up on a few
of the rec steps that need some people assigned.
astearns: I'll also send a summary.
astearns: Thanks everyone.
Received on Thursday, 23 March 2017 01:12:24 UTC