- From: Dael Jackson <daelcss@gmail.com>
- Date: Thu, 16 Jul 2015 08:18:19 -0400
- To: www-style@w3.org
Grid Issues
-----------
- The question regarding CSSOM will be deferred for a week to let
people read up on the issue.
bikeshedding user-select: element
---------------------------------
- RESOLVED: We change to contain and make the final decision next
week.
bikeshedding default
--------------------
- Everyone pretty much agreed that the cursor property shouldn't
be altered, so the word 'default' needed to be changed in
Cascade Level 4.
- Some of the top names that were discussed were rollback,
default-value, and rewind.
- Originally the group resolved on changing it to rollback and
having a new note saying other suggestions were welcome.
However, there was a late suggestion in IRC and the group
changed their mind.
- RESOLVED: rename default to revert
caret animation
---------------
- Several developers expressed that they had severe concerns about
animating caret.
- Florian will gather examples from the web of where caret is
being animated and create a proposal that covers those use
cases.
border-boundary with border-image
---------------------------------
- There was continued high interest in figuring out how to use the
existing CSS properties for rounded displays.
- Hyojin will send a follow-up e-mail to the mailing list with
visual examples of what he's looking for border-boundary to do.
===== FULL MINUTES BELOW =======
Present:
Rossen Atanassov
Tab Atkins
Bert Bos
Adenilson Cavalcanti
Tantek Çelik
Dave Cramer
Alex Critchfield
Elika Etemad
Simon Fraser
Daniel Glazman
Koji Ishii
Dael Jackson
Myles Maxfield
Anton Prowse
Florian Rivoal
Simon Sapin
Hyojin Song
Alan Stearns
Lea Verou
Greg Whitworth
Steve Zilles
Regrets:
David Baron
Sanja Bonic
Chris Lilley
Peter Linss
Michael Miller
scribe: dael
glazou: It's almost 5 past, let's start.
glazou: One from Hyojin and one from SimonSapin for extras.
glazou: I don't think Hyojin is on and we need heycam for the
second one, I believe SimonSapin said.
SimonSapin: I'll e-mail and we can do it next week.
Florian: Items 3 and 4 are things we discussed last week and
didn't get a conclusion. I have an action to write an
e-mail, but I've been busy. If we have a different
audience we can talk, but we should wait on the email.
Grid Issues
-----------
glazou: fantasai are there still things to discuss?
fantasai: The first is editorial
fantasai: The other two...row-gap/column-gap I thought we
discussed already...
Rossen: I thought we did discuss that.
glazou: Yes, but there remain 2 in the agenda. shorthand computer
value and stretching grid tracks in content
fantasai: I don't...I don't know about the shorthand one.
<glazou> https://lists.w3.org/Archives/Public/www-style/2015May/0128.html
fantasai: I don't know enough abot CSSOM for that.
glazou: Anyone else have anything to contribute on that one?
bikeshedding user-select: element
---------------------------------
Florian: I think we touched on this. user-select is a property
with an interesting history. What it used to do is
broader from what it does now. I'd rather not get too
much into the history of what it used to do, but due to
the history one value is called element. What it does now
is if you start the selection inside and element and drag
the mouse out the selection will stop at the boundary of
the element.
Florian: It constrains the selection to say within the element it
started in. I think 'element' is confusing, I'd rather
something like contain. I think this is doable because
there's only one implementation; Microsoft since IE10 I
think.
Florian: Other browsers have that behavior in textareas, but they
don't give an explicit toggle. So the name is still
changeable, I think. If people think it can't be changed,
too bad, we'll just confuse people a bit.
<tantek> right, it meant "a whole element inside" but it has not
been implemented as such and we should update it (
functionality, name) accordingly
fantasai: What was the behavior if you select outside?
Florian: Let me check the spec. I think it's in there.
Florian: It's a bit of an open issue because not all browsers do
the same.
Florian: What we have in the spec is if you start before and end
after the element is included. And if you start outside
and try to stop inside the element the selection would
stop on the boundary. For outside you get all or none,
you don't get part.
Florian: There is no interop on that. What's not in flux is if you
start inside you end inside regardless of where your
mouse ends up.
fantasai: Okay.
<tantek> not even sure if that's useful in any use-case
<tantek> maybe consider dropping it absent a use-case?
<fantasai> tantek, the behavior is that of textarea
glazou: Opinions?
Rossen: I don't find the name confusing.
Florian: The reason I find it confusing is within this group it
has several times been confused with the value that means
'select the entire element always', which is 'all'.
Florian: To answer tantek, there is a use-case for this. This is
the value that the property computes to on a content:
editable element. You might want to stick this in an
article so that a selection stops at the end of the
article and they can't get the comments unless they mean
to get them.
glazou: I could live with it, but it doesn't exactly represent the
behavior.
glazou: Bound?
Rossen: That's more geometric.
fantasai: I think Florian's point about people confusing all and
element is what is compelling and it means we should
change. I think contain makes sense and is a bit better
then bound. They mean the same thing, but without the
extra connotations bound has. I don't have other ideas.
<astearns> I think contain is slightly better than element
glazou: Do you have a suggestion for another name Florian?
Florian: Mine is contain, I'm happy with other ideas.
glazou: What do people think?
Florian: I had another suggestion which would be inside.
glazou: I thought about that too.
smfr: I agree either are better than element. Element is ambiguous
because it doesn't say what happens if you start from
outside.
glazou: Do we agree element is confusing? Any objections to that?
glazou: Okay. Let's try and find something else. We decide here or
let Florian decide and I'm in favor of letting Florian
decide.
Rossen: Let's discuss on the list. I'll get the people doing this
on the conversation.
Florian: I'm fine with the list, but it's been there for a month,
but there's only 2 replies. If people will reply I'm
happy with it on the list, but I want to resolve.
Rossen: The person I'm referring to was on vacation so couldn't
reply.
Florian: Okay.
glazou: I suggest shoot for contain for the time being and revisit.
Florian: I'd prefer that
glazou: Rossen?
glazou: It's simple enough to revisit at any time.
* fantasai is in favor of resolving
Rossen: I wouldn't push back too much; I'd prefer we discuss a bit
because the person I want to agree isn't here and I don't
want to have to revisit again. Let's give him a week and
we'll go from there.
glazou: Tentative resolution is we change to contain and make the
final decision next week.
Florian: Okay.
RESOLVED: We change to contain and make the final decision next
week.
<tantek> thanks Florian. Yes your suggestion of contain is fine
IMO. In general I think it's a good idea for the editor
to move forward with a default bikeshedding choice in
such cases and just await group objections.
bikeshedding default
--------------------
<glazou> https://lists.w3.org/Archives/Public/www-style/2015Jun/0252.html
Rossen: What was the issue and what are we bikeshedding?
Florian: The problem is that we pretended it's a CSS reserved
keyword for a while, but while doing that we forgot it's
used in the cursor property. So its taken even though we
thought it wasn't.
Florian: There has been a bunch of suggestions and I don't think
it's practical to remove the property from cursor since
that's shipped.
Florian: My favorite proposal is rollback because it rolls back
the cascades. I had previous proposal default-value which
is wordy.
<BradK> I like native
fantasai: I think reset was another.
glazou: Also native, cascade default.
glazou: I agree about your opinion of rollback, but I think for
non-English speakers it's a little bit complicated. It
implies you know the mechanism of the cascade.
Rossen: Florian can you refresh my memory on what default does for
cursor?
Florian: It gives you the platform default which is the arrow.
It's not the initial value which is auto that gives you
the arrow or the text cursor.
<astearns> "The platform-dependent default cursor. Often rendered
as an arrow. "
Rossen: And why wouldn't we make this property be the default
where it only selects one case of the initial value
instead of holding all of CSS hostage to that one property
value?
Florian: We could make an exception saying the CSS-wide default
keyword can't be used on this property, but it's ugly.
glazou: If I recall we discussed that in the past and said making
[sound lost]
glazou: I think we discussed in the past at least once and said
making an exception for one property-
glazou: It's not desirable to make exceptions.
Rossen: I agree with that, but taking the keyword away from CSS
entirely because that property is also kind of...I don't
know which is the lesser of two evils.
Rossen: It seems the exception for this property which is a
partial exception since the default is one case of the
initial value for the property, where not having the
default keyword which is a good keyword, doing it for this
one property seems the bigger evil.
<leaverou> exception is really awkward and makes CSS hard to learn.
Also, does all: default set cursor in that case?
<Florian> +1 to leaverou
<leaverou> if we make any exception, I would vote to renaming the
cursor default, since changing the cursor is extremely
minor breakage. But I’m in the minority on that one…
<fantasai> leaverou, we're too late for that
fantasai: This isn't a keyword for the initial value, this is for
what the user agent and user stylesheet set. So in the
case of the cursor property we resolved to make auto
less magic so we're relying on the UA stylesheet to
provide the types of cursor. This keyword would get you
back to what the UA said and if we don't allow this you
can't go back.
fantasai: This is a case where you really want that. You want to
be able to roll back. You use the UA stylesheet a lot to
mess with it so this keyword would be useful on this
property.
<Florian> Fantasai is saying what I wanted to say.
glazou: We have two requirements. 1: I don't want to make
exceptions. 2: We have a pretty good proposal even though
people might not understand - we'll teach them. cursor:
default is widely used. I don't want to touch it at all.
fantasai: Another value we could use is reset
glazou: Yeah. Why not.
Rossen: So unset and reset will be hard for people to distinguish.
<leaverou> +1 to Rossen
fantasai: That's the main downside. But it resets the values that
you wanted it to have.
gregwhitworth: What about UA-default?
fantasai: If you're an author what it means is give me the user
level and if you're the user it means UA level.
<fantasai> and if you're the UA it means initial-or-inherit (i.e.
unset)
gregwhitworth: Gotcha. Yeah.
Florian: My personal favorite is rollback for #1 and default-value
for #2
Florian: rollback is clear, but you have to understand what the
cascade is.
glazou: Opinions?
<leaverou> is rollback understood by people who have never worked
with databases?
smfr: Rollback could be a left pointing arrow, so I don't think
it's a good word.
<antonp> smfr +1
Rossen: The suggestion isn't to change the cursor, it's to keep
the value game for the cursor and we need to re-name
default
smfr: Okay.
* fantasai thinks reset and rollback are the best so far
Rossen: It's all about figuring out a better name for default. I
don't think there's a better name for default.
Florian: Yes, but it's taken.
Rossen: Anything that suggests unrolling.
glazou: rewind
<SteveZ> I would prefer "default-value" to "rollback"; I agree
with the comments that suggest "rollback" has too many
possible interpretations
<leaverou> +1 for rollback. reset is too similar to unset and too
vague
glazou: On IRC SteveZ says he'd prefer default-value
* gregwhitworth I agree that this is not desirable to not have a
global default
glazou: It's hard to do strawpoll on a conference call.
<tantek> +1 to reset
<tantek> "default-value" is ugly
fantasai: My problem with default-value is the other global values
don't follow that pattern. It's inconsistent.
<leaverou> +1 to fantasai
glazou: We're spending too much time on this.
glazou: Some people say default-value is ugly, rollback is weird
but acceptable, there's the rewind proposal
* fantasai liked rewind
<tantek> rewind how far?
glazou: default-value, rollback, and rewind are the best we have
on the table.
fantasai: And reset.
glazou: leaverou said it will be confusing and I agree.
Rossen: Yes.
glazou: default-value, rollback, and rewind.
<tantek> unset is confusing all by itself
Florian: I'd accept anything, I prefer rollback.
glazou: Me too.
<tantek> I prefer keeping/using reset over unset
Rossen: default-value or rollback is fine by me.
<Bert> +1 to default-value, can live with the others
<tantek> is unset even a non-jargon word?!?
<antonp> What are the consequences of cursor:default in the real
world suddenly meaning cursor: <default-as-in-the-css-
keyword-"default">?
<leaverou> +1 to rollback here too
<SteveZ> I still think "rollback' will confuse people
<astearns> slight preference for rollback, default-value is OK
<gregwhitworth> I don't like any of them honestly, but prefer
default-value
<Bert> ('defer' is possibly another option)
<tantek> I agree rollback will confuse people - so much jargon!
<Florian> antonp: default is used specifically to remove the magic
aspect of auto, which is what you'd get if we used the
css-wide meaning of default.
<antonp> Florian: thanks
fantasai: tantek has a bunch of comments in IRC. [reads them]
fantasai: I think once we have it, most people will just switch to
this one.
Rossen: The problem with reset is it's too unbalanced. Reset to
what?
fantasai: To what it was in the stylesheet. I think of it as a
chessboard where unset is take all the pieces off and
reset is put them back to the starting places.
<tantek> Rossen, problem of "x to what" exists for reset / rewind
/ rollback
<tantek> -1 rollback
glazou: It resets to what is in the spec.
<BradK> Not considering native?
smfr: All the values are in the spec. You mean the default values?
glazou: Yeah.
fantasai: It doesn't reset to what's in the spec because the user
might have changed something.
glazou: But the spec will say exactly what happens.
fantasai: What happens is it resets to the values that it had
because you touched it in your level of the cascade.
Rossen: Right. rollback, it's basically specificity sensitive
depending on where you are.
fantasai: Origin sensitive, but yeah.
Rossen: Yes.
Rossen: What's why reset is confusing for me.
<tantek> Rossen, for the same reason rollback and rewind are
confusing
<tantek> presumably something already has the value "clear"
<astearns> undo?
<SteveZ> how about "undo"?
<fantasai> +1 to astearns
Florian: I think rollback is better than either. You have to
understand the cascade, but if you do it's clear what it
does. Unset and reset you have to guess what it does.
glazou: astearns and stevez propose undo.
smfr: I don't think it's better than what we have.
Rossen: At the end of the day I think rollback will be the winner.
* TabAtkins is still of the opinion that "it's not precisely
correct for UA and user style sheets" is irrelevant,
and we should just optimize the naming for usage by
authors.
* fantasai is also okay with uncascade, better than some others...
glazou: Let's not spend the whole call.
glazou: Rollback is the proposal. Who is in favor, who objects?
<tantek> -1 rollback
fantasai: I don't really like it
glazou: Can you live with it?
fantasai: Yeah, probably.
<BradK> I can live with it, but prefer 'native'
<tantek> would prefer "ua-default"
<fantasai> tantek: no, I object to that because it's wrong
<leaverou> tantek: that's not what it does.
<tantek> ok to put anything into a draft and see authors / webdevs
scream
<tantek> fantasai, leaverou, ok
glazou: Is there anyone objection against rollback
* gregwhitworth Can we think on it for a week?
<Florian> greg: this is the 3rd call we spend on this.
glazou: gregwhitworth I'm not inclined to do more time on one item.
I don't want to reallocate 30 min.
<tantek> not counting the debate we had on this in NYC
fantasai: I'm happy to put what we decide in a draft along with
other suggestions and collect feedback.
<BradK> 'Roll-back' or 'rollback'?
glazou: Okay, let's do rollback for now and collect feedback
RESOLVED: rename to rollback for now and collect feedback
<BradK> I think people will wonder whether or not to put a dash
into rollback
<tantek> revert?
<fantasai> tantek: good one
<tantek> darn too late :/
* glazou tantek nice
* tantek was actually serious about the 'revert' proposal. oh well.
* glazou tantek fantasai revert is a good candidate
<fantasai> glazou, how about a quick "do we want to change the
resolution" on the call?
* tantek should have not joked with previous proposals.
* fantasai 1minute
<glazou> fantasai: yes
caret animation
---------------
<glazou> https://lists.w3.org/Archives/Public/www-style/2015Jun/0377.html
Florian: In NY it was suggested my caret blink wasn't a good
design and using a caret specific property for animation
would be better. It would have a few keywords to invoke a
few options and you can do custom.
Florian: I've mulled over it and I think this is a good idea to
add.
Florian: There are a few subtle points though.
<glazou> Florian: we can’t hear you
glazou: I think we lost Florian
[Talk of switching topics and trying to find Florian]
Florian: Should caret color be animate-able in addition to caret
animation with the normal animation property. I think yes
because I don't like exceptions and it's easy to define
what would happen.
Florian: Also because if you can do caret color in normal
animations it lets you do things that you can't do in
caret animations. If you animate caret color it's your
job to define the blinking etc. If you're happy with the
blinking etc. you can just change the caret color.
Florian: It's not a major use case, but it's a corner case and I
don't think we need to craft an exception to exclude this.
* smfr wonders if any UA wants to expose caret animations
* gregwhitworth agree with smfr
smfr: Can we ask about if authors should be able to animate the
caret?
smfr: We see it as a part of the system behavior, especially on
iOS. I don't think we want to let the authors control the
caret.
glazou: I have a question about that. In editors the caret differs
if you're overwriting, have RTL or LTR system.
smfr: I think RTL is interesting. There can be subtleties with
editing the caret and authors may not consider that. That's
a good argument for not exposing. In terms of override we'd
like to match the system.
Florian: The question was broader than what I thought we'd talk
about. Starting with shapes, the default is meant to be
auto, but there is a whole bunch of websites that are
tweeking the caret. It's easy to find websites that are
changing the caret.
Florian: Doing it currently requires ungodly amounts of JS. It's
reasonable to providing a value.
Florian: I don't think it should be open to anything, that's why
I'm only proposing the things that are commonly done.
glazou: I tend to agree. Inline editors all have JS code for the
animation of the caret and they're recreating stuff that
could be easily spec in CSS
Florian: So I think caret shape, as long as the default is auto,
it is not that bad.
Florian: For caret animation, there are use cases, though weaker.
Controlling speed of blink isn't an important use case.
However, instead of just blinking between invisible and
one color, but fading instead of blinking or changing
colors is something you see. We could define that and
cover most use cases.
glazou: Since we have concerns from two impl, I'd recommend a list
of websites that do animate a caret, figure out how they
do it, and propose a keyword or value or whatever. Let's
cover exactly what we see on the web. The caret is a
peculiar beast.
Florian: I understand your point. One thing that is not hard to do
as an implementor and possibly quite desirable is to say
do not blink, just be static. It's not hard to implement
and useful for a11y.
glazou: That's what I said.
Florian: That's what I started with, but since caret color is
animate-able you can change the color, but it works
poorly. If we can do blink or don't blink that's
sufficient for most use cases. The animations perform
poorly, though.
glazou: I suggest based on the input from 2 impl, you limit
yourself to existing things we need including the examples
because otherwise we're not going to convince them.
Florian: I'll go back to mail and offer alternatives and variants.
glazou: That okay for everyone?
smfr: Sure.
<TabAtkins> Anyway, Blink is also of the opinion that we probably
don't need special caret-animation functionality, and
blinking should just be done by a property with
specialized keywords.caret-color in a @keyframes is
fine.
border-boundary with border-image
---------------------------------
<glazou> https://lists.w3.org/Archives/Public/www-style/2015Jul/0080.html
hyojin: My topic is border-boundary and border-image
hyojin: I've noted my opinion about the topic on the ML. I think
border-boundary with border-image can be in CSS
Backgrounds 3. I know border-image is complicated, but can
be extended to support round-display.
hyojin: It can be altered to fit around images correctly.
hyojin: I'm wondering if my opinion is reasonable and if there's
any suggestions or opinions.
glazou: Opinions?
<TabAtkins> I don't think it belongs in B&B 3; that's a stable
spec now.
Bert: I have a small opinion.
Bert: I think border-image is the right idea, but the syntax might
be difficult. I think it's the right idea and we should look
at it and see how to fit it into the syntax. I might be
simpler for round displays. That's my intuition.
fantasai: I'm a little confused. Why does it need to be different?
Why not just do the round image. Like if you do a border-
image with round corners it's a circle.
Bert: So you say you don't have to do anything special, just make
sure the middle parts are 0. That might be possible. I was
thinking of a keyword to say the middle parts, but making
them 0, why not.
hyojin: I don't know how can the border-image work, but originally
the border-image works best on the rectangular box. I know
there is no method to draw the border image on the round
display. It can be extended, but I need to describe the
detailed method.
Rossen: Can you re-describe the use case? How is the border-image
supposed to behave for round display?
hyojin: I'd like to draw images around the round display. We have
some use cases to make application on our product, but I
don't know how I can draw the image using CSS.
<BradK> Can't it just use the smallest rectangular that contains
the round display? The display would clip any part of the
image that was outside the circle.
Florian: What are you expecting will happen? A straight image
bends around the circle?
hyojin: If I have some small image and I try to draw it, or
stretch it on the round display corner, is it possible or
not and how can I draw the image using a similar pattern.
glazou: The question is, if you have a rectangular image and you
want to draw in a round border, do you round it or place
it on top of the border.
Rossen: If we use a simple use case, if you want to create the
strikes for the second, so you need 60 around the dial,
and your saying the second makes need to be along the
border, but the source image is a rectangular image with
one strike you'll be repeating?
hyojin: That's one of the use cases, yes. It can be rectangular.
glazou: I think we need images. We need figures to understand what
you want. Can you contribute a reply to your org message
with the images you want to use and the result?
hyojin: Okay, I will do that.
glazou: Let's do that please.
<BradK> Sounds more like image repeated along a path.
glazou: Thank you for joining in the middle of the night.
default bikeshed
----------------
glazou: There's a proposal for revert from tantek and I like it.
<tantek> thank you glazou
<astearns> +1 to revert
<smfr> +1 for revert
<fantasai> +1
fantasai: I'm in favor
<gregwhitworth> agreed
<leaverou> +1 for revert
<dauwhe> +1
<Rossen> +2
<BradK> I like it
Florian: I like it. Had we proposal it earlier and someone said it
would mean something different?
<tantek> yay!
<Bert> 0
glazou: I don't care. Let's get consensus. If people are away,
they're away. Objections to revert?
<Florian> +1
RESOLVED: rename default to revert (ignoring previous resolution
about rollback)
Rossen: On the previous topic, I wanted to tell hyojin that he
should feel free to reach out to any of us if he needs
help with ex. There's a lot of interest in your topic,
it's just hard for use to guess what you need. Reach out
opening and frequently so we can help you out.
glazou: Absolutely.
glazou: Thank you everyone, I'll talk to you next week.
Received on Thursday, 16 July 2015 12:19:18 UTC