06:34
zcorpan
marks Future discussions as read
07:52
<jgraham>
Hmm, so roc's delaying the content load event proposals are scaring me, but I'm not entirely sure I can entirely defend my git-reaction of "run away"
07:52
<jgraham>
*gut
07:53
<Ms2ger>
http://www.youtube.com/watch?v=KAp9sFVdERQ
07:57
<jgraham>
I think what frightens me is tying in a script controlled flag to something that is already intimately connected with the lifecycle of the document
07:59
<Ms2ger>
Clearly we need a reallyload event
08:38
<smaug____>
well, you can implement the thing using script libraries
08:38
<smaug____>
some hacky iframe doing document.write or such
08:39
<jgraham>
Sure
08:39
<jgraham>
But there's a difference between "X is possible" and "X is encouraged"
09:04
<SimonSapin>
Is http://www.whatwg.org/specs/web-apps/current-work/multipage/rendering.html missing something like th { text-align: center }? Or is it encoded as prose somewhere in there?
09:07
<Ms2ger>
I think that's in prose somewhere...
09:07
Ms2ger
looks
09:08
<SimonSapin>
searching for "th" doesn’t really help ;)
09:08
<Ms2ger>
User agents are expected to have a rule in their user agent stylesheet that matches th elements that have a parent node whose computed value for the 'text-align' property is its initial value, whose declaration block consists of just a single declaration that sets the 'text-align' property to the value 'center'.
09:10
<SimonSapin>
oh …
09:10
<SimonSapin>
that sucks
09:10
<Ms2ger>
Yeah, it's weird
09:13
<SimonSapin>
I guess WeasyPrint is gonna ignore the part about the parent with an initial value
09:15
<SimonSapin>
I wish the whole chapter could be encoded as CSS to be used in a UA stylesheet, even if that means having a few magic CSS keywords
10:06
<aleray>
what is the license of html5lib-python?
10:10
<SimonSapin>
aleray: the source says MIT
10:21
<jgraham>
Yes, MIT
10:22
<jgraham>
Of course we prefer patches than forks :)
10:32
<zcorpan>
does anyone want to help me importing http://dev.w3.org/cvsweb/html5/html4-differences/ to https://github.com/whatwg/html-differences (preserving history)?
10:34
<jgraham>
zcorpan: git cvsimport seems to be what you want
10:36
<jgraham>
http://stackoverflow.com/questions/11362676/what-is-the-best-way-to-import-a-cvs-repository-in-git-and-use-it-locally
10:36
<jgraham>
I think you also have to have CVS set up right in the first place
10:37
<jgraham>
http://maymay.net/blog/2008/04/15/how-to-import-cvs-code-repositories-into-git-using-git-cvsimport/
10:39
<aleray>
SimonSapin, jgornick
10:39
<aleray>
oups jgraham
10:39
<aleray>
thanks
10:39
<aleray>
I'm writing a writer
10:39
<aleray>
for ConText
10:39
<aleray>
(http://wiki.contextgarden.net/Main_Page)
10:39
<aleray>
super crude
10:40
<aleray>
so maybe I'll wait a little bit more, but since I copied the htmlserializer to start with I will keep the mit license
10:41
<aleray>
no fork on the roadmap :)
10:53
<zcorpan>
jgraham: i tried following the maymay.net steps but get:
10:53
<zcorpan>
Running cvsps...
10:53
<zcorpan>
connect error: Permission denied
10:54
<jgraham>
I am not the best person to help here
10:55
<jgraham>
Since I haven't used CVS for 7 years or something
10:55
<jgraham>
and don't have W3C CVS access
10:55
<jgraham>
But, are you sure that your CVS access is correctly set up?
10:56
<zcorpan>
I logged in with the anonymous credentials at http://dev.w3.org/cvsweb/
10:56
<zcorpan>
~/.cvspass exists
10:57
<zcorpan>
i could do a checkout
10:57
<jgraham>
Hmm
10:57
<jgraham>
I note that no other instructions I can see require cvsps
10:59
<zcorpan>
or maybe they do but don't say so :-)
11:00
<jgraham>
Could be :)
11:01
<jgraham>
(you could also try cvs2git I guess: http://cvs2svn.tigris.org/cvs2git.html)
11:23
<zcorpan>
screw it
11:33
<zcorpan>
MikeSmith: can http://dev.w3.org/cvsweb/html5/html4-differences/ be made read-only?
12:21
<hallvors>
hm.. the problem with learning Git is that all one's mistakes seem to cause somebody else trouble.. Working on https://github.com/whatwg/xhr/pull/2 I'm wondering if I can cut down the long list of commits associated with the pull request
12:22
<hallvors>
the code changes under "Files changed" are near perfect, only those I want included..
12:22
<hallvors>
so why does Git (or GitHub) reference all those commits, and isn't there a way I can delete some of them from the list?
12:22
<Ms2ger>
I think the term is "squash"?
12:23
Ms2ger
pokes jgraham
12:24
<jgraham>
Yes, you can
12:24
<jgraham>
I suggest closing the pull request though; I'm not sure what happens if you rewrite history and then push to a existing PR
12:25
<hallvors>
I've heard rumours of squash.. ;-)
12:25
<jgraham>
The way that you rewrite the history to reduce the number of commits is to use git rebase -i (i.e. --interactive)
12:26
<hallvors>
however, what I really don't understand is that the PR also lists changes like "add more IDs.." that are really quite old..
12:26
<hallvors>
and pushed before I made this branch that I wanted to make a PR from
12:26
<jgraham>
Hmm, well let me start by talking about how to use git rebase if you have a clean branch
12:26
<jgraham>
Then we can figure out if your branch is clean :)
12:27
<hallvors>
hm.. what's a "clean" branch in the first place?
12:27
<hallvors>
8-)
12:27
<jgraham>
So the easiest thing to do is something like git rebase -i master
12:27
hallvors
appreciates having Git gurus around
12:27
<jgraham>
This should bring up a text editor with a list of commits on your branch
12:28
<hallvors>
OK - at what point? after committing stuff to master? to other branch?
12:28
Ms2ger
likes how hg doesn't require gurus to be around :)
12:28
<jgraham>
(protip: make sure that your $EDITOR is not set to vi before doing this)
12:28
<jgraham>
(protip2: :q)
12:29
<jgraham>
hallvors: I am assuming that you made the commits on some non-master branch
12:29
<hallvors>
yes
12:29
<hallvors>
so now I say rebase -i master. OK
12:30
<jgraham>
If you made the commits directly on master a) don't do that and b) look back through history to find some SHA1 before you started adding stuff
12:30
<jgraham>
Well, I guess it might even be easier to take it in two steps
12:30
<jgraham>
"git rebase master"
12:30
<jgraham>
This will move your commmits so they are on top of whatever master is
12:31
<hallvors>
OK - but they should already be.. anyway :)
12:31
<jgraham>
Hmm, let's take a step back
12:31
<jgraham>
Yeah, I think this doesn't all apply to your situation
12:31
<jgraham>
So the full process I would do is:
12:31
<jgraham>
git fetch origin
12:31
<jgraham>
git rebase origin/master
12:31
<hallvors>
I've done that
12:31
<jgraham>
git rebase -i origin/master
12:32
<hallvors>
(with some "merge conflict" resolution where I just told it to use Anne's version and discard any local edits)
12:32
<hallvors>
ah
12:32
<jgraham>
The two rebases are just to make things easier; the first will move your commits on top of the latest upstream without otherwise changing history
12:32
<hallvors>
I haven't done the rebase origin/mastr thing
12:33
<jgraham>
and the second one is to do the history rewrite
12:33
<hallvors>
OK, so that's why changes that predate me creating the branch still are in the PR
12:33
<jgraham>
Let's say you get as far as the interactive rebase
12:34
<hallvors>
thanks, now I understand a bit more :)
12:34
<jgraham>
Then you have an editor listing all the commits on your branch
12:34
<jgraham>
And some instructions down the bottom
12:34
<jgraham>
To the left of each SHA1 is the word "pick"
12:34
<jgraham>
You can change this, depending on what you want to do with the commit
12:34
<hallvors>
yes
12:35
<jgraham>
Notably "squash" will merge it into the previous commit and try to keep the commit message
12:35
<jgraham>
"fixup" will merge it into the previous commit and discard the commit message
12:35
<jgraham>
Deleting the line will remove that commit
12:36
<jgraham>
Changing the order allows you to reorder commits (but note that this can lead to merge conflicts)
12:36
<jgraham>
So you edit this until the branch looks how you want
12:36
<jgraham>
and save it
12:36
<jgraham>
and then you might get another editor allowing you to change some commit messages
12:36
matjas
refreshes http://html-differences.spec.whatwg.org/ like crazy
12:36
<jgraham>
and then history will be rewritten
12:37
<hallvors>
I'll try with a fresh branch and see if I can make a reaaaaly clean PR..
12:37
<jgraham>
Ms2ger: That explains why there are never blog posts from Mozilla people describing crazy gymnastics with mq
12:39
<jgraham>
(or to put it differently: describe how to go from commits A-B-C-D to commits AB'-D'-C' in hg that describe an identical change to the original branch)
12:40
<jgraham>
(and convince me this is something that would be obvious to a newcomer without consulting external help)
12:43
<SimonSapin>
jgraham: not Mozilla people, but still :) http://www.logilab.org/blogentry/88203
12:45
<jgraham>
SimonSapin: Maybe I needed a </sarcasm>? I have seen *many* blog posts about how to achieve simple-sounding things in hg
12:46
<jgraham>
(that phases thing sounds like it wouldn't work well with code review on a branch)
12:46
<SimonSapin>
well, I’ve seen this guy give a talk on hg phases, and it didn’t quite sound simple
12:49
<jgraham>
Right, I think we are agreeing that things aren't always as simple as Ms2ger claimed
12:50
<Ms2ger>
Hg phases make a lot of sense unless you push unfinished commits to a public repo, like we do with our tryserver
12:51
<jgraham>
Or we do with critic
12:52
<hallvors>
Ms2ger, jgraham: I had no intention of starting a religious war between the two of you. :-o Peace :)
12:52
<jgraham>
So basically it's a great idea as long as you don't worry about the real world
12:52
<Ms2ger>
Hah
12:54
<hallvors>
Meanwhile, I've successfully achieved some of what I wanted in https://github.com/whatwg/xhr/pull/3 but not all - right now I'll just wait for Anne getting to it and move on to the next opportunity for trying to make Git understand me and vice versa
12:55
<Ms2ger>
hallvors, you know about vim vs emacs, right?
12:55
<hallvors>
ugh
12:56
<hallvors>
I don't want to "know" more than I do because everybody who really "knows" is fighting, right? :-p
12:56
hallvors
likes peace
12:57
<Ms2ger>
hg vs. git is the same thing, with the difference that your choice actually affects the people you work with :)
13:01
<zcorpan>
matjas: meanwhile i was talking to a colleague and then getting a cup of tea :-P
13:04
<zcorpan>
matjas: (i've asked anne & hixie to create that url)
13:08
<MikeSmith>
zcorpan: about the dev.w3.org diffs doc, I can't make it read-only but I can add a redirect
13:09
<zcorpan>
MikeSmith: ok
13:09
<zcorpan>
MikeSmith: i guess the new url for the w3c editor's draft is https://rawgithub.com/whatwg/html-differences/master/Overview.html
13:10
<zcorpan>
MikeSmith: or we could keep the old url but sync it from github maybe?
13:10
<zcorpan>
either way
13:21
<Ms2ger>
annevk; http://critic.hoppipolla.co.uk/r/86
13:31
<hallvors>
Ms2ger: should I create a new branch for the rest of the XHR test changes (to let you/anne/whoever review and merge the current stuff)?
13:32
<hallvors>
I can either make a new branch and pull request or keep checking in to the current, depending on what you think is better
13:32
<Ms2ger>
I'm not going to review them, I don't think :)
13:33
<hallvors>
so "no opinion", basically? :)
13:34
<Ms2ger>
Yep
13:39
<zcorpan>
hallvors: i'll probably review your changes but i don't mind either way
13:44
<MikeSmith>
zcorpan: better to just redirect it, I think
13:44
<zcorpan>
MikeSmith: OK
14:46
<Ms2ger>
<bz> if I'm going to review the futures spec
14:46
<Ms2ger>
<bz> should I read the promises/A+ spec first?
14:50
<jgraham>
No
16:17
<lecuyer>
where are css decorators defined?
16:58
<MikeSmith>
oh man this is rich
16:59
<MikeSmith>
Opera suing Trond Hansen
16:59
<MikeSmith>
for giving away secrets to Mozilla
16:59
<nimbu>
MikeSmith: OMG
17:00
<MikeSmith>
maybe they should instead be suing him for not doing the actual job he was paid for even while he was working there
17:00
<MikeSmith>
or even bothering to, say, show up for work
17:00
<nimbu>
this is just
17:00
<nimbu>
fucked up
17:00
<nimbu>
so fucked up
17:00
<nimbu>
i dont know what to say
17:00
<MikeSmith>
nimbu: strange times
17:00
<nimbu>
sighhhhhh
17:01
<nimbu>
MikeSmith: i suppose you and I should be wary too :)
17:01
<MikeSmith>
heh
17:01
<odinho>
quite the shock yeah.
17:01
<MikeSmith>
yeah, look out!
17:01
<MikeSmith>
"describes himself as an artist, songwriter, and designer"
17:01
<nimbu>
ahahaha
17:01
<MikeSmith>
in that order, I bet
17:02
<MikeSmith>
to be fair I guess he did some years trying to get some UI innovations into Opera only to have Jon reject all of them
17:03
<MikeSmith>
so maybe he just gave up
17:03
<nimbu>
:(
17:03
<MikeSmith>
would be really ironic for Opera to sue him for "stealing" his own ideas that Opera management rejected
17:03
<nimbu>
gawddd
17:04
<nimbu>
everybody working for Opera please dont open your mouths
17:05
<MikeSmith>
new meme
17:05
<MikeSmith>
"Opera, Please Don't Sue Me"
17:06
<MikeSmith>
wait maybe this is the new business strategy
17:07
<nimbu>
:(
17:07
<MikeSmith>
wait for all Opera employees to implement new ideas at other companies
17:07
<MikeSmith>
then sue them
17:07
<nimbu>
MikeSmith: be careful they may sue you also
17:08
<MikeSmith>
s/Opera employees/former Opera employees/
17:08
<MikeSmith>
nimbu: blood from a stone
17:08
<MikeSmith>
they are welcome to share my debts
17:09
<MikeSmith>
I already drank up all the profits I made from business ideas I stole from Opera
17:09
<nimbu>
ahahahhaha
17:22
<TabAtkins>
gsnedders: I'm using html5lib in Python2.7, but that's largely because the libraries I wanted seemed easiest to install there. On the other hand, I'm probably fine with just not receiving updates, so long as the Python2 version is archived somewhere I can point to. ^_^
17:23
<gsnedders>
TabAtkins: It's not happening.
17:24
<gsnedders>
TabAtkins: It'd be nice for when I get around to rewriting the InputStream (as it avoids the one case we reparse currently), but we'll see how the ecosystem is then.
17:24
<gsnedders>
TabAtkins: Quite possibly will do that and just reparse under Py2.
17:25
<gsnedders>
TabAtkins: Or try and work out some hack to avoid it under Py2, dunno.
17:26
<SimonSapin>
gsnedders: is html5lib dropping python 2 support?
17:26
<gsnedders>
SimonSapin: "It's not happening".
17:26
<gsnedders>
SimonSapin: I was cursing yesterday I wish we could.
17:26
<SimonSapin>
ok, good :)
17:26
<gsnedders>
SimonSapin: Only thing that's at risk is 3.0/3.1 support
17:27
<gsnedders>
And that's mostly because it's untested, rather than anything fundemental.
17:27
<SimonSapin>
nobody uses 3.0
17:27
<gsnedders>
Indeed.
17:27
<SimonSapin>
and I’m thinking of dropping 3.1 support in weasyprint and various libs
17:27
<SimonSapin>
so I wouldn’t miss it
17:27
<gsnedders>
Likely we'll drop "official" support but practically it's unlikely to break.
17:28
<TabAtkins>
I like Python 3. :/
17:28
<SimonSapin>
TabAtkins: you still have 3.2 and 3.3
17:28
<gsnedders>
TabAtkins: 3.0.3/1 aren't everything. :P
17:28
<TabAtkins>
Oh! Okay, never mind then.
17:29
<gsnedders>
TabAtkins: The main reason html5lib might drop old Py3 is the lack of testing on them.
17:29
<gsnedders>
We're well tested on 2.6/2.7/3.2/3.3
17:45
<MikeSmith>
I wonder what "packability of the DOM code" means
18:04
<Garbee>
Can dialog elements be used outside of <th> and <dt> elements? The spec says they are where it *can* be used, but it doesn't say only may be used within. http://developers.whatwg.org/commands.html#the-dialog-element
18:05
<TabAtkins>
"Where flow content is expected".
18:05
<TabAtkins>
The weird special cases are because <th> and <dt> don't accept arbitrary flow content.
18:06
<TabAtkins>
So, any element with "Content Model: flow content;" can contain a dialog.
18:09
<Garbee>
Yea. I'd assume W3Schools simply used a table in their demo since that is pointed out as a use-case for a dialog. But their tip on it is wrong.
18:14
<TabAtkins>
Garbee: W3Schools is often wrong, and often terrible. Don't learn from them. ^_^
18:15
<TabAtkins>
Use a better learning resource, like HTMLDog.com, or pretty much anything, anywhere.
18:21
<Garbee>
TabAtkins, nah, I was trying to figure out an issue someone reported on W3Fools.
18:21
<TabAtkins>
Ah, carry on then.
18:21
<Garbee>
I tend to read the spec itself and then learn implementation from seeing people do things.
18:30
<Garbee>
TabAtkins, Also, I feel obligated to say also that http://www.webplatform.org is a great doc resource. ;) That should be first to mind when recommending somewhere for people to learn from.
18:30
<Garbee>
(Yes, it still has issues but we are working on them.)
18:33
<Ms2ger>
Can you get rid of the references to initProgressEvent, then?
18:33
<Ms2ger>
And the other init*Events, for that matter
18:42
<Garbee>
Ms2ger, Were you talking to me about removing those references from WPD?
18:44
<Ms2ger>
Sounds like work on it?
18:46
<Garbee>
Well, we have references to it. But I don't see any reason to remove them.
18:47
<Ms2ger>
It's being removed from implementations
18:49
<Garbee>
It is a CR spec. So where is confirmation of it being removed? If it is being dropped then it is something I should bring up with the mailing list on how we should deal with it.
18:50
<Ms2ger>
It's not in the spec either: http://xhr.spec.whatwg.org/#interface-progressevent
18:52
<Garbee>
What isn't in the spec?
18:52
<jgraham>
Anyone know anything about pointer events? https://critic.hoppipolla.co.uk/r/90 isn't super-convincing at first glance but I don't know the spec
18:53
<Ms2ger>
assert_equals(typeof ev[name], "DOMString"
18:53
<Ms2ger>
No, not terribly convincing
18:59
<jgraham>
Hmm, I wonder who Cathy Chan is
20:42
<TabAtkins>
Hixie: How do you come up with such believable attack scenarios?
20:43
<Hixie>
is that sarcasm or sincere? :-)
20:44
<Hixie>
i didn't think it was that believable...
20:44
<TabAtkins>
Sincere!
20:44
<gsnedders>
"Being evil" is the answer Tarquin would give.
20:44
<Hixie>
TabAtkins: i dunno man
20:45
<Hixie>
anyone here have opinions on margin collapsing quirks?
20:45
<TabAtkins>
I might?
20:46
<Hixie>
i'm looking at https://www.w3.org/Bugs/Public/show_bug.cgi?id=15993
20:46
<Hixie>
trying to work out whether to do it in some sort of CSS syntax, or in prose
20:46
<Hixie>
i'm thinking prose might be the only sane way to do it
20:46
<jgraham>
I assume zcorpan would have opinions
20:46
<jgraham>
If he was here
20:47
<Hixie>
he's made them known on the bug, luckily :-)
20:47
<jgraham>
Then I was right!
20:47
<jgraham>
Hurrah!
20:47
<jgraham>
;)
20:47
<TabAtkins>
Okay, I don't have an opinion on this, except that it should be defined in Quirks, not in HTML. Put all that shit together.
20:48
<Hixie>
surely "quirks" shouldn't exist
20:48
<Hixie>
should just be in the various specs
20:48
<TabAtkins>
Sure it should - you have a nice, compact list of crazy exceptions.
20:48
<TabAtkins>
Rather than spreading them around.
20:48
<Hixie>
there's "quirks" in the html parser, and i'm pretty sure we don't want the html parser spec split into multiple files
20:48
<TabAtkins>
I mean, really individual quirks, sure, put them next to the stuff they're quirking.
20:48
<TabAtkins>
But margin collapsing in table cells?
20:49
<TabAtkins>
Sure, parsers are different. ^_^
20:49
<Hixie>
almost the entire rendering section in the html spec is just a long list of "quirks"
20:49
<Hixie>
this is just one more in that section :-)
20:51
<TabAtkins>
Well, if it's expressible in CSS like Gecko does, then whatever.
20:54
<Hixie>
CSS uses proprietary selectors to express it
20:54
<TabAtkins>
:moz-only-whitespace isn't proprietary any longer - it's now :blank in the Selectors 4 draft.
20:55
<Hixie>
ooo
20:58
<jgraham>
blank?
20:58
<jgraham>
that's a... surprising name
20:58
<TabAtkins>
Needed a name that was the same as :empty, but not :empty.
20:58
<TabAtkins>
:blank is short and a close synonym, so whatever.
20:59
<jgraham>
Why "the same as empty"?
20:59
<jgraham>
It seems like expressing something different to :empty with a synonym for "empty" is an expressway to confusion
21:00
<TabAtkins>
Shrug, we got :empty wrong.
21:00
<TabAtkins>
But probably can't kill it.
21:01
<TabAtkins>
I think I'm supposed to track usage of it in Blink and see what pops out.
21:01
<jgraham>
So to make up for it you will get :blank wrong too?
21:01
<jgraham>
Interesting strategy
21:01
<TabAtkins>
Hm?
21:01
<jgraham>
s/get it worg/give it a bad name/ is you like
21:01
<jgraham>
*wrong
21:02
<jgraham>
Anyway, I think I should try sleeping or something
21:02
<jgraham>
*if
21:02
<jgraham>
(see)
21:02
<jgraham>
(sleeping++)
21:04
<Hixie>
man, i hope microsoft appreciates the pain i'm going through trying to test their browser
21:07
<Hixie>
there's like, no interop in edge cases around margin collapsing
21:08
<Hixie>
<style>html { background: purple; } body { background: lime; } div { margin: 1px; }</style><div><h1>Test</h1></div>
21:08
<Hixie>
different in every rendering engine
21:08
<lecuyer>
am I the only one who thinks :blank and :empty aren't that bad of names?
21:09
<Hixie>
firefox doesn't collapse even if the div has no margin, chrome/safari collapse if the div doesn't have a margin but do otherwise, IE doesn't give the <h1> margins regardless of the <div>'s
21:10
<Hixie>
man, IE tries hard to avoid giving margins to stuff at the top
21:11
<Hixie>
even with <body><div><span></span></div><div><h1> the h1 still has no margins unless i give the span a border
21:12
<Hixie>
hm, chrome/safari are pretty enthusiastic about this too
21:13
<Hixie>
hmmmm
21:16
<Hixie>
hypothesis: any UA-set margin that would collapse with the <html>'s top margin collapses to zero.
21:16
<Hixie>
(in IE)
21:16
Hixie
ponders how to disprove that hypothesis
21:32
<Hixie>
i cannot find any way to disprove this hypothesis
21:32
<Hixie>
could it really be this simple?
22:03
<Hixie>
TabAtkins: is there a CSS equivalent to :-moz-first-node/:-moz-last-node? (matches same as :first-child/:last-child but only if there's no non-whitespace text nodes before/after the node)
22:12
<Hixie>
man, describing elements that match :-moz-first-node is a maze of twisty negations
22:13
<lecuyer>
Hixie: is that meaning that with <div>here's text<span>hi</span></div> that `div span:-moz-first-node` wouldn't match?
22:19
<Hixie>
lecuyer: yeah
22:20
<Hixie>
anyone got a good term to mean "an element or a text node that is not merely inter-element whitespace"?
22:21
<lecuyer>
Hixie: "unfortunate" ;)
22:24
<Hixie>
I'm going with "purple" until someone has a better term. :-)
22:26
<lecuyer>
Hixie: are you trying to name a pseudo selector or describe TEXT_NODEs in a spec?
22:27
<Hixie>
just trying to describe "an element or a text node that is not merely inter-element whitespace" for spec purposes
22:28
<astearns>
Hixie: didn't you use 'palpable' for something similar?
22:28
<Hixie>
that's something else
22:28
<Hixie>
but yes
22:29
<Hixie>
unfortunately i went to the thesaurus and none of "palpable"'s synonyms are words i can use
22:29
<Hixie>
(https://www.w3.org/Bugs/Public/show_bug.cgi?id=15993 for context)
22:33
<astearns>
Hixie: how about 'substantial'
22:34
<Hixie>
that could work
22:34
<Hixie>
noted on the bug
22:35
<TabAtkins>
Hixie: There is no equivalent to :moz-first-node/last-node.
22:44
<lecuyer>
astearns: how do you feel about arbitrarily nested pseudo-elements?
22:45
<TabAtkins>
Hixie: The phrasing we use in CSS for "purple" is "not an anonymous inline containing only collapible whitespace".
22:51
<Hixie>
TabAtkins: hm, good point, white-space:pre should stop this too, huh
22:51
<Hixie>
though i guess gecko doesn't
22:53
<Hixie>
hm, no, gecko _is_ affected
22:53
<Hixie>
wtf
22:55
<MikeSmith>
.win 14
22:56
<Hixie>
TabAtkins: so:
22:56
<Hixie>
A node is significant if it generates a box that is not an inline box
22:56
<Hixie>
containing only collapsible whitespace.
22:56
<Hixie>
...?
22:56
<zewt>
let's let win 8 run its course
22:58
<TabAtkins>
anonymous inline box
22:58
<TabAtkins>
("inline box" also describes inline elements)
22:59
<Hixie>
is it anonymous if it has an explicit node?
23:00
<Hixie>
so:
23:00
<Hixie>
A node is significant if it is an element or if it generates a box that is not
23:00
<Hixie>
an anonymous inline box containing only collapsible whitespace.
23:00
<Hixie>
...?
23:00
<TabAtkins>
anonymous boxes come from bare text only.
23:00
<Hixie>
bare text has a node :-)
23:00
<Hixie>
aren't table rows and such anonymous too?
23:00
<Hixie>
the ones that are implied?
23:01
<TabAtkins>
Those aren't anonymous inlines. ^_^
23:01
<Hixie>
you said "anonymous boxes"! :-P
23:01
<TabAtkins>
Bah.
23:01
<Hixie>
anyway the dfn above is right, right?
23:02
<TabAtkins>
If you can use HTML node concepts, just say "element node, or text node that doesn't just contain collapsible whitespace".
23:04
<Hixie>
that's what i had, but as you made me realise, it's wrong
23:04
<Hixie>
because what matters is whether the white-space actually gets collapsed or not
23:04
<Hixie>
that is, white-space:pre matters
23:06
<TabAtkins>
Yeah, wrong word. I have the right wording in antoher spec. One sec.
23:07
<TabAtkins>
http://dev.w3.org/csswg/selectors/#the-blank-pseudo "characters affected by whitespace processing".
23:07
<TabAtkins>
Oh wait, you *want* white-space pre to matter?
23:07
<Hixie>
apparently
23:08
<TabAtkins>
In that case, "collapsed whitespace".
23:08
<Hixie>
(i mean, what i actually want is neither here nor there on this issue)
23:08
<TabAtkins>
"doesn't contain only collapsed whitespace"
23:09
<Hixie>
as in:
23:09
<Hixie>
A node is significant if it is an element or if it is a Text node that does not
23:09
<Hixie>
contain only collapsed whitespace.
23:09
<Hixie>
..?
23:10
<TabAtkins>
Now it's easier to phrase it in positive terms: "or if it is a Text node that contains anything other than collapsed whitespace".
23:11
<Hixie>
is "collapsed whitespace" actually defined anywhere? http://www.w3.org/TR/css3-text/#white-space-rules vaguely defines "collapsible", but it seems what i actually want is to invoke hte word "removed" from that algorithm
23:11
<Hixie>
"A node is significant if it is an element or if it is a Text node that contains preserved white space or text that is not collapsible" ?
23:12
<Hixie>
"A node is significant if it is an element or if it is a Text node that contains preserved white space or characters that are not collapsible white space" ?
23:12
<Hixie>
with [CSS] at the end :-)
23:13
<Hixie>
(Note that this means :-moz-empty and :blank aren't the same after all)
23:13
<Hixie>
(unless i'm mistaken)
23:14
<TabAtkins>
i like the last one, though it's verbose. It feels pretty precise.
23:14
<TabAtkins>
You mean :-moz-only-whitespace?
23:14
<Hixie>
yeah
23:14
<Hixie>
i'm wrong though
23:15
<TabAtkins>
Yes, that is a different condition than Moz's - the selector *cannot* depend on whether white-space:pre is set.
23:15
<TabAtkins>
It can only rely on whether characters in the content are "normally" collapsible.
23:15
<Hixie>
the selector doesn't seem to depend on white-space, now i'm just confused
23:16
<Hixie>
oh duh
23:17
<Hixie>
i am dumbo
23:17
<TabAtkins>
Heh.
23:17
<Hixie>
had a newline, made it look like it was margin plus the line with whitespace
23:18
<zewt>
wow, new gmail is even worse than i thought
23:18
<zewt>
i try to scroll down past a compose box, and it scrolls the whole compose instead of letting me leave the compose box scrolled where it is and scroll the whole box
23:18
<zewt>
gripegripe
23:19
<Hixie>
WebKit/Blink are affected by white-space, but Gecko isn't
23:19
<Hixie>
interestingly IE isn't either; I really think my hypothesis above is correct
23:20
<astearns>
lecuyer: I have started to become convinced that pseudo-element use cases are better served by HTML templates
23:20
<Hixie>
TabAtkins: amusingly i tried to give you credit for the big breakthrough i had an now it just looks in the bug like you were wrong, sorry about that :-)
23:20
<astearns>
lecuyer: CSS doesn't do structure, HTML does. Why invent CSS structure when you can re-use HTML?
23:21
<TabAtkins>
zewt: I find the new compose *really* useful. It's an interesting behavioral mix of "as tall as all the contents" and "no taller than the screen", which works really well for me most of the time.
23:21
<TabAtkins>
zewt: In the case you're complaining about, I pop the message out instead, because that's typically what I really want to do anyway.
23:22
<zewt>
i edit replies nonlinearly, and i often want to skim down to a reply (past the point where the compose box landed) then go back to what I was writing, but now I can't leave the compose box where it is
23:22
<zewt>
i guess, or i can leave the cursor in place and hit an arrow key to scroll to it
23:22
<TabAtkins>
That works too (and I'm already used to doing that for text editors, when I need to quickly check another section of the spec for the text I'm currently writing).