- From: fantasai <fantasai.lists@inkedblade.net>
- Date: Tue, 13 Nov 2012 22:49:22 -0800
- To: "www-style@w3.org" <www-style@w3.org>
HTMLWG/CSSWG Coordination
-------------------------
Identified several places where coordination is needed:
- Selectors. This is largely handled due to last year's Selectors 4 FPWD,
and subsequent edits that aligned the HTML5 spec with that.
- Scoped style sheets. Various aspects of this are either underdefined
or incorrectly specced. Several CSSWG members recommended to drop it
from the HTML5 timeline. Aside from the HTML syntax for invoking
scoped styles, many things need to be defined on the CSS side,
specifically:
- Handling of scoped selectors (mostly handled in latest Selectors 4)
- Potentially, CSS syntax for changing how selectors are scoped.
- Scoping the cascade (will be drafted into css3-cascade soon, but
no draft yet exists as of TPAC 2012)
- Handling of various globally-scoped at-rules such as @font-face.
The third issue was about communication. Historically, the HTMLWG has
put things in their spec that affect the interpretation of CSS (or, in
some cases, extend CSS) without notifying the CSSWG of such changes or
additions, giving the CSSWG no invitation to review and send feedback
or to incorporate such changes into its own specs. The HTMLWG countered
that the CSSWG should file bugs about things that are wrong (once they
find out about them) and escalate things using the HTMLWG process.
Discussion participants were the CSSWG, two of the HTMLWG co-chairs,
and plh.
====== Full minutes below ======
--- Day changed Mon Oct 29 2012
<RRSAgent> See http://www.w3.org/2012/10/29-css-irc#T08-02-05
<JohnJansen> I'm observing in the Browser Testing and Tools WG this morning.
Agenda
------
Scribe: TabAtkins
glazou: We still have no definition of what a scoped stylesheet is in CSS.
TabAtkins: fantasai and I will take care of that next month.
* plh will join at 9:30am with Paul Cotton to talk about html
glazou: Also HTML added new selectors, etc.
<plh> (ie in 10 minutes)
glazou: PLH is also concerned about the TTA specs.
glazou: The length of time they're taking.
glazou: I tried to say that they were complex specs.
glazou: But they're already well interoperable.
HTMLWG/CSSWG coordination
-------------------------
Present: (at table) Dean Jackson, Glenn Adams, Philippe Le Hegaret,
Anton Prowse, Francois Remy, Rossen Atanassov, Simon Sapin,
Lea Verou, Divya Manian, Tab Atkins, Luke MacPherson,
Alan Stearns, Steve Zilles, Dirk Schultze, Bert Bos,
Leif Arne Storset, Vincent Hardy, Paul Cotton, Daniel Glazman,
Ted O'Connor, Håkon Lie, David Baron, fantasai Etemad,
Aaron Eicholz, Taichi Kawabata, Kazutaka Yamamoto, Koji Ishii,
Peter Linss, Maciej Stachowiak
Present: (in back row) Rik Cabanier, ???, ???, Rebecca Hauck, ???,
Israel Noto Garcia, Jet Villegas, ???, Laurence Mclister
plh: HTML plans to move to CR by the end of the year.
plh: So if there are any concerns from the CSSWG, I'd like to get it now.
plh: Are there any issues from the CSSWG?
glazou: I have three issues.
glazou: The first is organization.
glazou: The liaison between html and css - there has been none.
glazou: The HTML spec concerns bits of CSS related to scoped stylesheets
and selectors which were never discussed with us.
glazou: I call that a big issue, since the htmlwg charter contains a
mandatory liaison with us.
glazou: Don't care what side the problem lies on, but it needs to be solved.
glazou: Second problem is technical.
glazou: The html spec contains scoped stylesheets.
glazou: There's a grammar for scoped stylesheets in html, but we don't
have a formal definition of how they'll work in CSS.
glazou: There's no material the html spec can reference normatively.
glazou: Third point is about CSS pseudo-classes.
glazou: There are pseudo-classes inthe HTML spec - some are related to
pseudos in css3-ui, but we never had a joint group or submission
saying "we plan to include things that are in your charter in
our spec, how can we do that, is that correct, etc."
glazou: So it seems to me the process between the two WGs are completely
broken.
glazou: We don't have these issues with the SVGWG - we work together great.
glazou: So we have two things in the HTML spec that should be defined
on our side.
TabAtkins: On selectors, we've either defined everything or HTML is just
defining the host language part of it
TabAtkins: ... in cases where things are to be defined by the host language
plh: I like to understand if other people in the CSSWG share your opinions.
plh: And some of the CSSWG is in the HTMLWG, so if there's no liaison,
I wonder what's going on.
dbaron: We've been through the form control stuff before, but I agree
with Tab that there's no problem there.
dbaron: I don't think the <style scoped> stuff is defined well anywhere
yet, or if it reflects quite what our idea of how it will work is.
TabAtkins: I will define it
<franremy> (just a note: maybe we also need to speak about shadow dom
and its interaction with CSS, those specs also contain quite
a bite of CSS)
plh: Is <style scoped> handled properly in HTML?
fantasai: First thing is how selectors are scoped.
fantasai: Second is a mechanism to change how they're scoped.
fantasai: Third is how the cascade is scoped.
fantasai: The fourth is how globally scoped things like @font-face are
handled.
fantasai: So there's four aspects of this. We have a definition for the
first, and Tab and I are planning to work on the third.
fantasai: Dunno about the second and fourth.
TabAtkins: The fourth is defined in HTML now - Boris got it fixed.
fantasai: It doesn't matter if the HTMLWG editor is in the room or not,
if they aren't bringing things up to us for review, etc. it's
a problem.
dbaron: I think we know about this - we don't need a formal email asking
us to review it.
dbaron: I don't think we need a formal email to inform us of something
that we've discussed multiple times that we haven't received a
formal email about.
dbaron: The interesting question is, is there anything else we're not
aware of.
glazou: I think we do formal link between groups
mjs: this is first time I hear about these issues
glazou: no
glazou: I sent them as LC comments during the vote
glazou: Never got a reply
Scribe: fantasai
plinss: We sent comments last year as a WG, and the bug was closed
WONTFIX by hixie
mjs: We have a clear process for escalating issues
mjs: We love technical input, but don't go out of our way to solicit it
mjs invites CSSWG members to participate in HTMLWG meetings at TPAC
mjs: Recently had change of editors, trying to be more open to input
PaulCotton: Want to be more factual here. What are the bug numbers
raised on these issues?
PaulCotton: What are the comments that were made for LC?
PaulCotton: If there weren't issues raised as tracker issues, we don't
pay attention to them.
glazou: We discussed on HCG these issues
plinss: https://www.w3.org/Bugs/Public/show_bug.cgi?id=13693
TabAtkins: Raised as email comments; you're not allowed to ignore
those just because bug wasn't filed
glazou: All the comments, through various channels were dismissed.
mjs: for record, bug doesn't mention scoped style sheets at all
Scribe: dbaron
fantasai: I think there are three issues, 2 technical.
fantasai: I think selectors issue is already addressed by selectors4.
glazou: Then HTML is referencing a non-normative document
fantasai: No, it's already FPWD.
plh: Where do you think selectors4 will be in 2014?
fantasai: Probably last call or CR.
glazou: maybe, maybe not
fantasai: So that problem is solved as long as HTML correctly referencing
selectors4.
fantasai: Second issue is about scoped style, 4 sub-issues.
ted: Already at-risk for HTML5.
plh: I would advocate the HTML WG to drop the feature and put in HTML.next.
plh: Sounds too risky to keep in HTML 5.0.
mjs: I'd like a record of the four issues.
TabAtkins, fantasai: in minutes
TabAtkins, fantasai: We should have a draft of cascading part in the
next month.
fantasai: This is a complicated feature that ought to have a stable
definition, not one writen 2 months before CR.
dirk: Since we don't have one implementation, it maybe should not be in
HTML5.0
fantasai: Third issue is the process issue.
fantasai: Basically what's happening is that the CSS WG would like the
HTML WG to take some initiative to contact us when defining
things in the scope of our charter.
fantasai: Whether HTML WG does this by assigning that communication as
a task to our common members or some other process doesn't
matter. Fact is it hasn't happened.
glazou: or send an email
PaulCotton: So in the last 2 years I've tried several times to get HTML
WG to pay attention to progression of docs in other WGs.
PaulCotton: As a co-chair it's pretty frustrating. I try to solicit
input from HTML WG and get nothing.
PaulCotton: Not even from members of other WG.
PaulCotton: So the ??? hasn't worked in the past.
PaulCotton: I think we probably need somebody on point here.
PaulCotton: I think we need somebody to draw attention to these kinds
of things.
PaulCotton: We have new editors now.
PaulCotton: I think with new editors: bugs, interop problems, ... bring
that to your attention.
PaulCotton: I think the other route, bringing your (CSS) documents to
attention of HTML WG isn't going to work.
Scribe: fantasai
dbaron: You probably heard me say this before, but I'm not a big fan of
trying to make thing WG-to-WG communication.
dbaron: I think there is a lot of value in giving notice to WGs about
things that are related.
dbaron: I don't see that the response needs to be from the WG as a whole.
dbaron: If there's consensus about a response, fine, it can be from the
WG as a whole, but I don't worry about the whole-WG response.
glazou: Paul, you say you'd like someone in your WG to coordinate and
give us feedback.
glazou: I'd like the Chairs to do it. That's what we do - when SVG has
something, Doug pings us. The chairs discuss things. It's
informal, but it works well.
hober: In CSS/SVG, we have the FXTF to help discuss issues.
glazou: Even before that.
glazou: It's not hard. It's just a matter of person-to-person email.
It's doable.
fantasai: With regards to feedback on our specs, our responsibility
is to inform you of spec's we're writing that might affect
HTML and it's interaction.
fantasai: I think we've done that, but please point out if we need to
improve.
fantasai: As for formal responses from the HTMLWG, doesn't matter.
Individual WG members were informed, had the ability to respond.
fantasai: but it hasn't happened in the other direction
fantasai: The members of the CSSWG that aren't in the HTMLWG haven't
been informed about things in HTML that affect CSS.
fantasai: In particular, HTML defines things that are fundamentally not
in their charter, that are fundamentally in ours.
fantasai: Which may be a bit over the line.
glenn: In that regard, Hixie has been communicating with me about things
that affect the CSSOM.
mjs: Can someone give me an example of a time that the CSSWG has informed
the HTMLWG about a relevant spec change.
Scribe: fantasai
TabAtkins: element function in images spec, wanted to integrate with
HTML in a particular way
TabAtkins: I sent an email into HTMLWG and hixie fixed it for me
TabAtkins: Done similar things with WebApps
PaulCotton: Hixie is no longer an editor in the HTMLWG
<hober> I believe TabAtkins was referring to
https://www.w3.org/Bugs/Public/show_bug.cgi?id=14028
PaulCotton: If you want to communicate wrt HTML5 spec, it's no longer hixie.
PaulCotton: When wanting changes with CR, need to communicate with
current editors, not hixie
PaulCotton: Question was what happens to changes hixie makes
PaulCotton: Our editors are triaging all of those changes
PaulCotton: For changes that go into 5.1, we're previewing in WG
PaulCotton: But editors in WG are about to produce CR drafts
PaulCotton: Those CR drafts will be only changes that are
interoperability-based
PaulCotton: It's possible that change from WHATWG would make it into 5.1
PaulCotton: Also possible that if it's an interop problem, it would
get into 5.0
PaulCotton: But not guaranteed
glazou: Would like us to use HTCG a bit more for communication. You are
three co-chairs, but rarely attend the calls
glazou: I've been sending status reports for CSSWG even when I did not
attend, and these include changes to our documents. Never
triggered a reaction from you.
glazou: When I said in HTCG that there were problems wrt scoped styles,
there was no reaction from you.
glazou: We have a tool in our hands.
mjs: I find the coordination calls useless
glazou: Then let's do that by email
[HTCG doesn't have regular calls any more, just as needed]
glazou: We've had this communication channel for years; it hasn't been used.
Scribe: TabAtkins
https://www.w3.org/Bugs/Public/show_bug.cgi?id=16169
fantasai: Maciej wanted examples of us contacting the HTMLWG about
changes affecting them.
fantasai: First, we don't include things within the scope of the
HTML charter within our specs.
fantasai: But for things where we think the HTMLWG could do a good
review (such as Image Values or Selectors), we have explicitly
pinged the HTMLWG for review.
mjs: How would you like us to inform you of changes we need in CSS?
Filing a bug?
TabAtkins: Send an email to www-style.
mjs: Can you point to an email you've sent that's an example of what
you'd consider sufficient notice to you?
TabAtkins: The bugs I filed against HTML are examples
fantasai: Hixie has sent us emails asking for specific changes in CSS
specs, and that's exactly how it should be handled.
fantasai: What wasn't handled was saying that we drafted a new CSS
feature and put it in HTML5 spec.
mjs: So what do you need to continue technically to resolve any issues
in HTML?
mjs: As far as I know, all we've gotten so far in the HTMLWG about style
scoped is that it's undefined, from Glazou.
mjs: But today we have four things being about it.
TabAtkins: They're four ways of saying a specific thing is undefined.
glazou: Scoped stylesheets will be a major way to copy-paste stylesheets.
There are multiple deep technical issues to solve about these.
It won't be solved today.
glazou: My take is that it's so undefined right now, and there's so much
on our radar, it's at risk, but you should drop it for now.
glenn: Are there two interop impls?
TabAtkins: No - we've specifically held back our impl because it's not
interop with how we know we want it to act.
glazou: It's so undefined, and will take long enough to do so, that it
should be dropped.
<hober> So, has someone filed a "Drop <style scoped>" bug?
mjs: If best solution turns out that it should be dropped, that's fine.
mjs: But we need a bug and to be informed about this.
mjs: I've heard several rapid-fire problems listed today, but we need
those details to be brought to us.
TabAtkins: The original problem we're bringing up is that nobody told
us about it, or asked us to define it. Just now picking it
up because we have to
TabAtkins: But even within HTMLWG, ppl know that it's not well-enough
defined to be implemented right now. bzbarsky has given
feedback; Chrome is holding back implementation because it
doesn't match what we want to do
TabAtkins: The problem is well-known within HTMLWG
glazou: One concrete example I said is specificity of selectors, don't
know technical solution we choose, need discussion to happen,
but will drastically impact solution. Not a 10-minutes discussion
glazou: Need to find compromise that satisfies everyone.
glazou: And it's on our side, not on yours. It's about how the cascade
works. Not something that belongs in the HTML spec.
glazou: Am I completely mistaken or what?
plh: It seems to me that we have information on the style scoped about
whether to keep it or not.
mjs: I don't think we have. You guys have told us multiple rapid-fire
problems, but they haven't been listed.
TabAtkins: they're all variants of "it's undefined". The specifics don't
matter. You've known it was undefined since we told you a
year ago.
mjs: Okay, so what should we do specifically?
glazou: Drop it.
mjs: Give us that feedback specifically. Send it in.
mjs: Please submit a comment to the HTMLWG saying that and giving rationale.
glazou: I've done that multiple times.
mjs: I looked up your comments, and you didn't actually say that.
glazou: PLH, do we really discuss inter-WG things via bugs only?
glazou: I don't understand why you need more input from us.
PaulCotton: I asked for exact comments. You pointed to a bug.
It was closed, you didn't reopen.
mjs: The HTMLWG has a process. If you engage in it, we'll respond.
But berating us won't solve a problem. If the problem is that it's
underdefined and won't be solved in time, file a bug.
mjs: If you're unwilling to engage in our process, then I won't help you.
* fantasai doesn't understand why hober doesn't just file a bug so we
can get on with it
<plh> --> http://lists.w3.org/Archives/Public/public-html/2012May/0080.html
email from Tab regarding scoped
PaulCotton: I don't quite agree with maciej about ???...
PaulCotton: This spec will be in CR for 18 months at least.
PaulCotton: I think that marking it as at-risk with this evidence,
is probably the right status.
dbaron: I've been on the queue for 5 minutes to say that I disagree
that dropping is the right thing. I'd rather see it stay for
now.
dbaron: If you want a WG opinion, we need to take the time for that
discussion. But I don't think right now is the right time
for that discussion.
glenn: Agree. At-risk is easier than dropping and restoring.
mjs: It's at-risk right now. We can enhance our at-risk list with
annotations for reasoning.
mjs: Getting dependencies right, getting implementors to implement,
define it.
* fantasai doesn't think the "at risk" list is the right place for
"things we think are cool but don't have time to define
correctly"
* fantasai also doesn't understand why it's a problem to drop it here,
given it'll still be in the HTML.next drafts that hixie edits
plh: Other issue is Selectors 4 having some of the HTML selectors.
<plh> --> http://dev.w3.org/html5/spec/single-page.html#pseudo-classes
TabAtkins: The ones in HTML right now are all just the host-language
stuff. The real new things are in WebVTT.
* fantasai is willing to bet that Selectors 4 will get to CR before
HTML5 has enough testcases to actually test the spec
mjs: WebVTT is no longer a HTMLWG deliverable. It's in a CG.
glazou: This is a quite heated discussion, I agree. But we'd love to
help the WG with the HTML spec. We'd love to be pinged when
it's on our scope.
hober: Right now we're trying to remove/drop things, not add.
plh: But in HTML.next, if ever there is something new added, tell the CSSWG.
PaulCotton: The way we're pulling from WHATWG to HTML5.1, the triage
team should probably check for CSS-specific things and send
an email to the CSSWG.
mjs: In the past when stuff got put into the spec, I'll be honest and
say it was largely editor recalcitrance.
Bert: About a year ago I sent a personal note that I think there's a
better way to define the <details> element that makes it easier
to style in pure CSS.
plh: Feel free to bring that up in our meeting Thu/Fri
Received on Wednesday, 14 November 2012 06:53:32 UTC