05:45
<annevk>
TabAtkins: super awesome
05:51
<annevk>
mpt: http://www.mpt.net.nz/ :(
05:51
<annevk>
mpt: I was still subscribed
05:52
<annevk>
mpt: but not for "My Positive Thoughts" by "Mat"
06:19
<annevk>
hmm
06:19
<annevk>
so Gecko simply does not have a iso-2022-kr encoder?!
06:19
<annevk>
<!doctype html><meta charset=iso-2022-kr><a href="http://dump.testsuite.org/url/query-dump.php?q=test&#x3066;">test</a><script>var a = document.getElementsByTagName("a")[0]; alert(a.search) </script>
06:20
<annevk>
yields utf-8
06:20
<annevk>
Gecko: ?q=test%E3%81%A6 Opera/Chrome: ?%1B$)Cq=test%0E*F%0F
06:23
<annevk>
well, I'm going to define one anyway, objections can go into bugzilla
06:49
<annevk>
teehee http://dvcs.w3.org/hg/encoding/raw-file/tip/Overview.html#iso-2022-kr-encoder
06:50
<annevk>
guess I'll write a blog post about big5 now and call it a week :p
06:50
<annevk>
or maybe address some of those bugs / work on sniffing...
07:46
<nesta_>
good morning :)
07:48
<annevk>
morning
08:21
<MikeSmith>
annevk: http://qa-dev.w3.org:8888/?doc=data%3Atext%2Fhtml%3Bcharset%3Dutf-8%2C%3C%21doctype%2520html%3E%250D%250A%3Ctitle%3E%3C%252Ftitle%3E%250D%250A%3Cinput%2520type%253Dimage%2520value%253Dfoo%2520alt%253Dbar%3E&showsource=yes
08:22
<MikeSmith>
round 3
08:24
<annevk>
it would be awesome if after "Attribute value not allowed on element input at this point" you had ", because type is image"
08:24
<annevk>
is that feasible?
08:24
<MikeSmith>
no
08:24
<MikeSmith>
unfortunately
08:25
<annevk>
okay, how about highlighting the value attribute line in the overview below?
08:25
<MikeSmith>
oh
08:25
<MikeSmith>
that's a good idea
08:25
<MikeSmith>
that's doable
08:25
<MikeSmith>
very good idea
08:25
<annevk>
pretty awesome you put some much into this validator man
08:26
<annevk>
authors are gonna like it
08:26
<MikeSmith>
well one part's that not so awesome about it is we've had this bug open for more than 3 years now
08:27
<MikeSmith>
or rather, I think it was at least 3 years ago that people started filing bugs about this
08:27
<annevk>
Encoding standard has been bugging me for about four years now I think :)
08:27
<MikeSmith>
yeah, it's nice to get some of this stuff done
08:27
<MikeSmith>
I should have fixed this a long time ago
08:47
<jgraham>
So I'm behind. What's the latest state of callbacks in WebIDL (e.g. for event handler attributes)
08:48
<jgraham>
HTML defines interface Function with [Callback=FunctionOnly]
08:48
<jgraham>
But FunctionOnly doesn't seem to be defined in WebIDL
08:53
<annevk>
HTML needs to change that to "callback Function = any(any...)" iirc
08:53
<annevk>
plus something else
08:53
<annevk>
the nullable stuff
08:53
<annevk>
there's an open bug on HTML for that filed by me
09:12
<hsivonen>
jgraham: are the actual ringmark tests on github? I only see harness code there? what am I missing?
09:12
<jgraham>
hsivonen: They are in a different repo. somewhere
09:12
<jgraham>
If you have the facebook blogpost open somewhere it links to both
09:13
<jgraham>
https://github.com/coremob/coremob-tests/ perhaps?
09:14
<jgraham>
Oh no that's the harness
09:15
<hsivonen>
right
09:15
<jgraham>
FWIW Tobie seemed to agree that the current way that prefixes are tested is broken, although I think we disagree on the magnitude of the breakage
09:16
<hsivonen>
it would be nice if the CSS WG rescinded the policy that leads to such breakage
09:17
<hsivonen>
so that a CG didn't need to figure out how to deal with the fruits of the policy
09:18
<jgraham>
Indeed. Sadly it looks like even money on that and hell freezing over
09:18
<annevk>
http://annevankesteren.nl/2012/04/big5
09:18
<jgraham>
annevk: Over 9000 what?
09:20
<annevk>
search for "over 9000"
09:20
<annevk>
but also pointer/code point mappings
09:22
<jgraham>
Sigh, young people.
09:22
<annevk>
oh, I never watched that show
09:23
<annevk>
I just know about the meme
10:04
<annevk>
http://wiki.whatwg.org/wiki/Encoding
11:45
<annevk>
smaug____: we can fix intersectsNode, no problem
11:45
<smaug____>
annevk: you mean Opera?
11:45
<smaug____>
that would be nice
11:45
<smaug____>
the current API is strange
12:17
<zcorpan>
TabAtkins: do we need to allow parsers to abort for syntax errors?
12:18
<zcorpan>
TabAtkins: html needs that because it's incompatible with streaming, but css probably doesn't
12:19
<annevk>
validator might want it
12:19
<zcorpan>
"Conformance checkers are not required to recover from parse errors."
12:24
<zcorpan>
TabAtkins: shouldn't there be a step using the charset of the parent stylesheet or document (that linked the stylesheet)?
12:33
<zcorpan>
which svg attributes take a css number with scientific notation?
12:36
<zcorpan>
"For example, scientific notation is allowed in attributes, including presentation attributes, but not in style sheets." is what i find in svg1.1
12:41
<zcorpan>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1475
12:41
<zcorpan>
seems webkit doesn't support hashless colors in svg attributes at all
12:43
<zcorpan>
i guess that's a good approach, but i dunno how it should be specced
12:44
<zcorpan>
oh, actually, since the quirks spec only applies it to a number of properties, i guess it automatically doesn't apply to any of the svg attributes (unless they map to one of those properties or something)
13:17
<annevk>
wait all of css is now in https://dvcs.w3.org/hg/csswg/
13:17
<annevk>
but they keep using dev.w3.org URLs?
13:18
<annevk>
confusing
13:18
<annevk>
and getting the changelog per draft is somewhat hard too
13:22
<davidb>
ygoat
13:25
<annevk>
smaug____: fixed
13:25
<annevk>
smaug____: dom spec that is, not Opera :)
13:27
<smaug____>
thanks
13:27
<smaug____>
Ms2ger can now update his patch
13:27
<smaug____>
or her
13:28
<smaug____>
we never know
13:28
<annevk>
CORE-45741 is the bug on Opera for those with access
13:28
<smaug____>
its
13:37
<zewt>
somebody splitting a thread to title case a subject: one of the more bizarre things i've seen on mailing lists in a while
13:39
<annevk>
where is this?
13:40
<zewt>
webapps
13:40
<zewt>
subject Re: IndexedDB: Retrieving A Slice Of A Record Value.
13:40
<zewt>
somebody at opera, no less :)
13:47
<hsivonen>
do we have a spec for generating a Document when an image, plug-in, video or audio file is loaded into a browsing context?
13:48
<smaug____>
is there any way to get notified when a spec changes
13:48
<smaug____>
(not whatwg html)
13:48
<smaug____>
annevk: like for DOM4
13:48
<smaug____>
hsivonen: I thought html spec defines that
13:48
<smaug____>
at least for images
13:48
<smaug____>
I could be wrong
13:49
<zcorpan>
smaug____: http://dvcs.w3.org/hg/domcore/shortlog has a feed
13:51
<smaug____>
zcorpan: thanks
13:51
<zcorpan>
TabAtkins: "Otherwise, if the next input character is a digit, or the next 2 input characters are U+002D HYPHEN-MINUS (-) followed by a digit, switch to the number state.", is --1 supposed to go to the number state?
13:52
<hsivonen>
smaug____: OK. then I need to figure out what words to search for. (not "image" apparently!)
13:53
<annevk>
hsivonen: "page load processing"
13:53
<annevk>
"Page load processing model for media" in particular, section 666
13:54
<zcorpan>
TabAtkins: hashless looks if there's a comment between the "+" or "-" delim and the number/dimension/ident token, but your tokenizer seems to not emit comments
13:55
<hsivonen>
annevk: apt section number considering how bad readyState is in Gecko in that case
13:55
<smaug____>
hsivonen: I can't find it either
13:55
<smaug____>
section 666 sounds evil :)
13:55
<hsivonen>
ok. so the spec allows adding random stuff to <head>.
13:55
<zcorpan>
TabAtkins: or, hmm, actually it tries to skip comments, so that's good
13:55
<hsivonen>
I expected the code I saw to be non-compliant.
13:56
<hsivonen>
but seems to be compliant!
14:00
<zewt>
annevk: heh, good luck with sniffing, that seems like something implementors would be loathe to touch (since any change will doubtless break some pages while fixing others)
14:02
<annevk>
doesn't seem that different from any other change
14:02
<zewt>
the nasty part is that in some browsers (IE, at least), it's based on the user's system language ... doesn't matter if that's specced, it's inherently zero interop
14:03
<zewt>
well, it's a lot more fuzzy
14:04
<hsivonen>
zewt: since the system language is IE's UI language, that's not much worse that Firefox's default behavior depending on the Firefox UI language
14:05
<zewt>
hsivonen: i'd phrase that the other way around: "firefox isn't much better" :) (don't recall how FF handles things off-hand)
14:06
<zewt>
i think the big problem is going to be that no matter what heuristics are used, they're going to be less "reliable" for same-language users (users reading pages in their own language) than depending on the user's language, which will make it hard to get browsers to stop doing that
14:06
<zewt>
worth a try
14:07
<hsivonen>
IIRC someone told me in Opera the domain name the document is served from factors into the heuristic...
14:07
<hsivonen>
TLD that is
14:08
<zewt>
heh, yuck--taking a document and putting it on a different website shouldn't make it stop working :|
14:08
<zewt>
but, well, heuristics are yuck by nature...
14:09
<jgraham>
Well opening it in teh same browser confiugured to a different locale shouldn't either, but…
14:09
<hsivonen>
I think using the TLD is less evil than relying on the UI language
14:09
<zewt>
lots of unmet "shouldn't"s when it comes to encoding compat :)
14:11
<zewt>
afk
15:23
<dglazkov>
good morning, Whatwg!
16:43
<kennyluck>
"Easter edition quiz! The CSS 2.1 spec has at least one easter egg. Can you find one?" refers to what?
16:44
<kennyluck>
(There are many errors in CSS 2.1 but none of which I would call easter egg.)
16:44
<TabAtkins>
kennyluck: Read his later tweet!
16:44
<TabAtkins>
I and dbaron and others point out some of them.
16:45
<kennyluck>
TabAtkins, ah. I should have done that. Thanks.
16:46
<kennyluck>
Well that one… isn't very entertaining either :p
16:46
<TabAtkins>
That's because you're an hollow, humorless shell of a person.
16:47
<kennyluck>
= =
16:48
<annevk>
that went dark fast
16:49
<TabAtkins>
^_^
16:50
<annevk>
Velmont: filing a bug on your server-sent events issue might be good
16:50
<annevk>
Velmont: dunno how much Hixie monitors public-webapps these days
17:05
<Velmont>
annevk: It was just as much to also get discussion about it. Should I cross-post to WHATWG?
17:05
<Velmont>
s/discussion/some answers/ :P
17:46
<jscheel>
hello everyone, I'm a little confused on the property attribute of meta tags. If I use the name attribute, the name is validated against http://wiki.whatwg.org/wiki/MetaExtensions, but if I use property, it isn't. What's the difference?
17:47
<TabAtkins>
Is "property" from RDFa or something?
17:47
<webben>
jscheel: Well, they're different.
17:47
<webben>
yes
17:47
<TabAtkins>
That would be why, then.
17:48
<webben>
jscheel: the design of meta@name fights naming collisions with a wiki. the design of meta@property fights naming collisions with namespaces.
17:49
<jscheel>
webben: so, I just want to have <meta name="company:property" content="content"> or <meta property="company:property" content="content">
17:49
<webben>
what are those?
17:49
<webben>
jscheel: What do you want clients to do with that markup? What's it for?
17:50
<jscheel>
webben: just custom settings that will be parsed my a decorator
17:50
<webben>
jscheel: Do you mean you're going to do something in your JS with that data?
17:51
<jscheel>
webben: actually, it will be parsed server-side. I am using html as a theming engine, sort of the way tumblr does it
17:51
<jscheel>
webben: for example, tumblr has this in a theme definition: <meta name="text:Disqus Shortname" content="" />
17:51
<webben>
jscheel: Well, either way, that's what @data-* was designed for.
17:52
<jscheel>
webben: hmm, that's true, they could just be data attributes on the html tag
17:52
<webben>
<html data-yourframework-company="whatever">
17:52
<TabAtkins>
Yeah, if you're just putting information into the markup that you'll use in some scripts in the page, use the data-* attributes.
17:52
<jscheel>
webben: I didn't even think about doing it that way
17:52
<jscheel>
that makes WAY more sense
17:52
<webben>
:)
17:52
<jscheel>
thanks!
17:58
<jscheel>
webben: so, do multiple data attributes work the same way as meta tags? iirc, if you have two keywords metatags, the spec is to concat and uniq them. If I have two data-palette tags on the same element, should the be put into an array when accessed via javascript?
17:59
<webben>
er... i imagine repeating any attribute is an error
18:00
<webben>
jscheel: If you want to embed a list, I'd be inclined to use a simple separator (whitespace often works well).
18:00
<webben>
jscheel: @data-yourframework-palette="green red beige"
18:01
<webben>
commas is another good choice
18:01
<webben>
if you need a list of arbitrary strings, you could embed JSON rather than inventing your own format
18:01
<TabAtkins>
Yes to everything @webben just said.
18:01
<webben>
@data-yourframework-keyphrases='["foo","bar","baz"]'
18:02
<jscheel>
webben: TabAtkins: here's what I've been struggling with: http://pastie.org/3805737
18:02
<jscheel>
I'm trying to make it easy enough for a designer / front-end coder to use
18:02
<TabAtkins>
That first block of code seems fine.
18:04
<jscheel>
TabAtkins: so then just match data-theme-palette-* when pulling the content?
18:04
<webben>
pretty much
18:04
<webben>
(I'd add -yourframework- in to avoid collisions with random JS or whatever your frontenders want to include)
18:05
<jscheel>
webben: ah, true
18:05
<jscheel>
webben: TabAtkins: thanks for working through this with me
18:05
<webben>
yw
18:06
<TabAtkins>
zcorpan: If you see this later, that was a c&p typo. It's meant to be U+002E FULL STOP, not U+002D.
18:07
<TabAtkins>
I presume my brain was infected by the U+002D case immediately prior for recognizing a cdc token, and then I copied it up into the + state.
19:23
<rafaelw_>
hsivonen: any chance you'll have a chance to evaluate dimitri's parser changes for <template>?
19:24
<jgraham>
rafaelw_: (hsivonen probably isn't around right now)
19:24
<jgraham>
But I should look at those too
19:25
<rafaelw_>
Please do. I'd appreciate that. Do you have the email already?
19:48
<jgraham>
It was sent to whatwg, I think?
19:48
<jgraham>
Or was there another more secret one? ;)
19:49
<jgraham>
Oh webapps. But thanksa for re-forwarding it
19:49
<jgraham>
I will try to look tomorrow; kind of late here now
19:50
<jgraham>
My first comment is "have you looked for people using <template> elements in the wild and not expecting magic behaviour?"
19:51
<jgraham>
Presumably this is quite easy to do if you are working for Google :)
19:54
<TabAtkins>
It's not easy, but it's not particularly hard.
19:54
<jgraham>
Well for the rest of us it is particularly hard, so if you are working for Google it is quite easy by comparison :p
19:58
<Philip`>
http://philip.html5.org/data/tag-count-pages.txt says 3 pages out of 130K have <template>
20:28
<kennyluck>
I am surprised to see <html> <head> having the same number of occurrence.
20:50
<Philip`>
kennyluck: That's because the URL is lying, it's counting elements instead of tags
20:50
<Philip`>
kennyluck: and the parser always inserts html and head elements
20:52
<kennyluck>
Philip`, ah, OK.
21:05
<Matt______>
Would this be a good channel to ask a question about WebRTC? If so, is there a webrtc api for sending text/data to other browsers, and not just video and audio?
22:24
<zewt>
Hixie: where should I file a bug for the CSS "balanced" word wrapping mode for WebVTT? (http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2011-December/034026.html)
22:24
<zewt>
(unless it's already filed; dunno)
22:25
<zewt>
(if that has to go to the CSSWG or something, then I'll let somebody familiar with them deal with it)
22:30
<TabAtkins>
zewt: That (or things like that) have been proposed before, but it's an O(n^2) algorithm to do.
22:31
<TabAtkins>
And thus kinda hard to put in CSS, in case people apply it to their whole page, filled with large paragraphs.
22:36
<TabAtkins>
If we put it into Text, it'd have to be something that UAs are allowed to ignore.
22:37
<TabAtkins>
Send it to www-style with a title starting with "[css4-text]"
22:37
<TabAtkins>
fantasai says it was previously in Text 3, but it was badly-defined.
22:42
<zewt>
seems like it shouldn't need to be exponential, at least naively
22:42
<TabAtkins>
It's not exponential. It's n^2.
22:42
<TabAtkins>
That's polynomial, dude.
22:42
<zewt>
not particularly interested in dealing with more mailing lists, though
22:42
<zewt>
d00d
22:53
<Matt______>
is there a webrtc api for sending text/data to other browsers, and not just video and audio? kinda like p2p websockets? I remember reading that webrtc had the ability to send data, and not only video/audio, but I couldn't find much more information on the topic
23:24
arunranga
wonders if it's worth defining "microtask" since we've got "Perform a microtask checkpoint" defined.
23:29
<Matt______>
with regards to my question earlier, I found that instead of webrtc, you could just use shared workers
23:29
<WeirdAl>
hi arun :)
23:42
<zewt>
https://zewt.org/~glenn/wrapping.py naive attempt at balanced wrapping; linear time on the length of the string
23:49
<TabAtkins>
Ah, yes, indeed. I forgot that the n^2 part comes in when you try to balance the justification, too.
23:50
<TabAtkins>
The problem is still actually n^2 even with your simple attempt if you want *perfect* balancing.
23:50
<zewt>
"perfect or nothing" is the cause of more useless programmer headaches :)
23:50
<TabAtkins>
But if you're willing to accept somewhat non-optimal solutions, a greedy algorithm does well in non-degenerate cases and is linear.
23:51
<zewt>
i'll always take a suboptimal solution over none at all
23:51
<zewt>
(and I doubt it's suboptimal enough to matter)
23:59
<TabAtkins>
zewt: ...why are you calculating the width of a word by summing the index of each letter in the alphabet?