00:01
<Hixie>
anyway, regarding asking microsoft, i suspect the people who would know and whom i would be able to ask are in the a11y tf, so hopefully they'll reply to rich's e-mail and explain what we need
00:13
<TabAtkins>
an ImageData is static, right? So if I need to do an input->output filter, I can just grab two ImageDatas from the same canvas?
00:14
TabAtkins
will just test it anyway.
00:14
<Hixie>
static?
00:15
<Hixie>
use createImageData() if you need an output image data
00:15
<TabAtkins>
In that if I grab two ImageDatas from the same canvas and alter one, the other won't be affected?
00:15
<TabAtkins>
Ah, that'll work too.
01:10
<Templarian>
I'm going through and helping out extending the autocomplete for an editor and they wanted to put all css3 in... is user-select in the spec? It says it was superseded by a newer document that lacks user-select.
01:11
<Templarian>
So am I to believe that if it doesn't exist in http://www.w3.org/TR/css3-ui/ it shouldn't be added?
01:12
<TabAtkins>
For now, yes.
01:12
<TabAtkins>
Also, css3-ui isn't really supported now, though it's picked up an active editor again.
01:13
<TabAtkins>
(Supposedly.)
01:13
<Templarian>
Well they added all the -prefixes and wanted to just keep them in and just remove them if they don't make it through later on.
01:14
<Templarian>
I'm leaving user-select out and all the ones not in it till later.
01:14
<Templarian>
Thanks. :D
01:27
<TabAtkins>
Hm. Anyone know why I'm seeing an assymmetry here? The code should be perfectly symmetrical. www.xanthir.com/test.html
01:27
TabAtkins
is testing true gaussian blurs.
01:27
<MikeSmith>
cool to see paul_irish getting a hat tip in http://blogs.msdn.com/b/ie/archive/2010/07/15/the-css-corner-better-web-typography-for-better-design.aspx
01:27
<MikeSmith>
bravo Sylvain
01:28
<TabAtkins>
For whatever reason, the white half blurs wider than the black half.
01:32
<Arimil>
What is the difference between the <head> and <header> tags?
01:32
<Arimil>
I know head is required but I always put all my header info in there.
01:32
<Hixie>
they are in no way similar other than by name :-)
01:32
<TabAtkins>
<head> marks the section of the html document that holds metadata and isn't displayed. <header> is just a way of marking up a header section in your content.
01:32
<Hixie>
what tab said
01:33
<Arimil>
Ah so I shouldn't put anything else in head, thanks I'll move it all to a <header> tag.
01:34
<Arimil>
Should I move scripts out of header also?
01:34
<Arimil>
head I mean
01:34
<TabAtkins>
scripts should go in <head>.
01:34
<Arimil>
OK, thanks again.
01:36
<Hixie>
basically anything that you can see in the browser goes in the <body>
01:36
<Hixie>
and <header> goes in the <body>
01:36
<Hixie>
<head> goes before the <body>
01:36
<Hixie>
<header> is a bit like <div>, it's just for heading-related things like <h1>, or the publication date, or whatnot
01:36
<Hixie>
the spec has a number of examples
01:36
<Hixie>
http://whatwg.org/html5
01:36
<Hixie>
search for <header>
01:55
<boblet>
MikeSmith: pity they didn’t link to Craig Mod’s @font-face example page, but I guess it would have made theirs look somewhat tame in comparison
01:55
<MikeSmith>
boblet: add a comment
01:55
<boblet>
MikeSmith: no fallbacks so no x-browser compat either
01:56
<MikeSmith>
big whoop
01:56
<boblet>
MikeSmith: well, they’re demoing WOFF, not @font-face (despite talking about it)
01:57
<MikeSmith>
who cares about x-browser compat?
02:04
<TabAtkins>
Dammit, it looks like my problem is float accuracy. >_<
02:04
<TabAtkins>
I hate it when that happens.
02:05
<TabAtkins>
I guess losing about 8 bits of accuracy when juggling white pixels is significant here.
02:08
<TabAtkins>
Yay, fixed!
02:08
<TabAtkins>
Woo!
02:09
<TabAtkins>
(By cheating - when I'm on the white side, I invert pixel values so the calculations have the same accuracy as the white side.)
02:09
<boblet>
MikeSmith: not browser makers when doing tech demos, that’s fer sure
02:09
<MikeSmith>
heh
02:17
<Arimil>
Is it true that in html5 you don't use self closing tags? For example <br /> is just <br>
02:17
<TabAtkins>
Yes, that was the original syntax for void element (those without end tags). XML invented the <br/> syntax.
02:17
<variable>
Arimil, in the non-XML version: yes
02:18
<TabAtkins>
So, as long as you're using the HTML serialization of HTML, yes. If you're using the XML serialization, no.
02:18
<Arimil>
OK, I'm not using XML serialization. Thanks gotta fix that.
02:19
<variable>
Arimil, IIRC it doesn't actually matter in the non XML version. IE they both work
02:19
<TabAtkins>
In the HTML serialization you're *allowed* to use the <br/> syntax on void elements. (You can't use it on normal elements that just happen to be empty - <div/> in the HTML serialization is just treated as a <div> open tag).
02:20
<variable>
TabAtkins, in the XML serialization <div/> is counted as having an open and a closed tag ?
02:20
<TabAtkins>
variable: Yes.
02:21
<TabAtkins>
In XML, <div/> is the same as <div></div>.
02:21
<Arimil>
Good to know although I most likely wont use it. :p
02:22
<variable>
Someone really needs to write a "html5 for the web author" document.
02:22
<Arimil>
I've never liked xml for some reason.
02:22
<Arimil>
I avoid it whenever possible.
02:22
<TabAtkins>
Good idea in general.
02:23
<variable>
Arimil, XML is perfectly fine for a particular type of content. The problem comes when people try to use XML for a type of content it isn't made for
02:23
<variable>
XML is great when you have an exact format with specific data that must be sent with particular datatypes
02:24
<variable>
its horid for dyamic loosly formed content like HTML
02:24
<Arimil>
variable: I believe your referring to visual studio using it to replace ini?
02:24
<TabAtkins>
XML isn't great even for that - too many meaningless restrictions that bork the document if you mess them up.
02:25
<variable>
Arimil, nope: my sole use for XML ever was for writing an FAQ. I had a series of Sections, questions, and answers
02:25
<variable>
and I used XSLT to convert it to HTML
02:28
<Arimil>
variable: *shrug* It's a good use of XML although like I said I really hate XML and I would have looked for another way
04:09
<micheil>
has anybody tested Safari 5 for websocket draft75 compliance?
04:10
<micheil>
I'm noticing that safari will close connections after N seconds
04:17
<micheil>
where N seconds is 120
04:46
<MikeSmithX>
it occurs to me that a problem with the way many people conceive Web architecture is is that they assume that resources at URIs are not really doing much of anything at all
04:46
<MikeSmith>
but are instead just waiting to be requested
04:47
MikeSmith
is reading through some Web-related papers submitted for the Internet of Things 2010 conference
04:48
<MikeSmith>
or in other terms, most discussions about Web architecture doesn't seem to involve much thought about all the interesting events that could be occurring for a resource at a particular URI
05:46
<micheil>
looks like the timeout issue with safari is due to the server, but it does raise another issue, on close, safari doesn't send a FIN packet back
06:19
<MikeSmith>
I guess I would have expected AtomPub to have some kind of eventing model as its conceptual basis
07:35
<Hixie>
variable: in theory lachy is writing one - http://dev.w3.org/html5/html-author/
07:35
<Hixie>
variable: it's been lacking attention though
07:48
<MikeSmith>
http://diveintohtml5.org/ does a pretty good job too
07:53
MikeSmith
needs to buy copies of http://books.alistapart.com/product/html5-for-web-designers and http://introducinghtml5.com/ as well
08:04
<JonathanNeal>
Bam!
08:04
<JonathanNeal>
Liferay 6 http://sourceforge.net/projects/lportal/files/Liferay%20Portal/6.0.3/liferay-portal-tomcat-6.0.3.zip/download
08:09
<MikeSmith>
JonathanNeal: hey wow
08:09
<MikeSmith>
you work on Liferay, right?
08:10
<JonathanNeal>
I do.
08:10
<JonathanNeal>
And we're an HTML5 toting gang now.
08:10
JonathanNeal
does the HTML5 gang sign.
08:12
<MikeSmith>
JonathanNeal: I think some people outside would say that the HTML5 gang seems to be using the middle finger as its sign
08:13
<JonathanNeal>
How's that?
08:13
<JonathanNeal>
Lemme see if I can get some of my roommates to do the html5 gang sign.
08:15
<MikeSmith>
given the pioneering work that Bruce Lawson and Remy Sharp have done in this area, it seems pretty clear that the HTML5 gang sign needs to involve some partial nudity
08:19
<Workshiva>
Clearly the gang sign is to make a '5' with your fingers
08:21
<MikeSmith>
I think it should just be holding up one five-fingered hand
08:21
<JonathanNeal>
okay lemme try something
08:21
<MikeSmith>
but in the form of an eagle claw
08:21
<MikeSmith>
but if you have only 4 fingers, you can only make the HTML4 gang sign
08:42
<JonathanNeal>
okay i've got it. html5 gang sign.
08:50
<MikeSmith>
reading some of these paper submissions for Internet of Things 2010, I'm getting an education about the difference between using HTTP as a just a transport protocol, and using it as an application protocol
09:05
<JonathanNeal>
http://sandbox.thewikies.com/html5-gang-sign/
09:11
<MikeSmith>
ah JonathanNeal
09:12
<MikeSmith>
you are brilliant, man
09:12
<MikeSmith>
can I tweet this?
09:12
<MikeSmith>
this is solid gold
09:12
<MikeSmith>
complete with the background music and everything
09:14
<MikeSmith>
"2 plus 3 equals 5"
09:15
<ako>
<meta charset="UTF-8" /> <- that's ok now?
09:15
<ako>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
09:15
<ako>
<- no need for that mess anymore?
09:16
ako
fails at pasting
09:16
<annevk>
right
09:16
<ako>
excellent
09:16
<ako>
that's something i can actually remember :>
09:16
<annevk>
you can even do <meta charset=utf-8>
09:17
<ako>
eh... i won't go that far
09:17
<ako>
<:
09:17
<zcorpan_>
you can even do <meta utf-8>
09:17
<annevk>
zcorpan_, uh?
09:17
<zcorpan_>
j/k
09:18
<annevk>
too early :/
09:18
<annevk>
JonathanNeal, lol
09:18
<annevk>
JonathanNeal, we should have you guys at Standards Suck :)
09:21
<Workshiva>
JonathanNeal: Your video is inaccessible to people without speakers :<
09:22
<zcorpan_>
add websrt captions in a <track>
09:24
<ako>
mhmh... does track allow lang attributes? if so, how do you switch between different languages?
09:25
<zcorpan_>
there's srclang
09:28
<JonathanNeal>
How do you expect me to compensate for that, Workshiva?
09:28
<ako>
so, could i use several video files with different audio tracks and also several subtitle tracks which also use different languages?
09:29
<ako>
for example the media splitter i'm using with my regular stand alone player allows me to define priorities for this stuff
09:30
<ako>
well, it's of course an implementation detail, but do the specs point out that one should keep this in mind?
09:30
<MikeSmith>
hendry: you're switching jobs and moving to London?
09:31
<hendry>
MikeSmith: yup
09:31
<MikeSmith>
switching jobs = moving to different company?
09:32
<hendry>
MikeSmith: yes, contracting for Vodafone
09:32
<MikeSmith>
wow
09:32
<MikeSmith>
great news
09:32
<MikeSmith>
working with anybody I know?
09:32
<hendry>
MikeSmith: I think I am working on WAC stuff with Ricardo
09:33
<MikeSmith>
cool
09:33
<MikeSmith>
I see http://twitter.com/kaihendry/status/18518431813 now
09:33
<hendry>
MikeSmith: of course DKA is around. I'm sure to harass him
09:33
<MikeSmith>
yeah
09:33
<MikeSmith>
give that dude a hard time
09:34
<MikeSmith>
he deserves it
09:34
<annevk>
ako, it's not a single video with multiple audio tracks?
09:35
<annevk>
ako, but there's nothing specific for that case at the moment afaik
10:28
<hsivonen>
hendry: what's JIL/WAC?
10:29
<hendry>
hsivonen: it's a JIL/BONDI conversion, now collectively called WAC http://www.wholesaleappcommunity.com
10:31
<hsivonen>
hendry: ok
10:33
<hendry>
It's still not finalised tbh. Though I hope it will all come together.
11:21
<jgraham>
Argh. I really wish javascript had named parameters
11:22
<Workshiva>
Named parameters?
11:22
<jgraham>
like foo(bar=baz)
11:22
<jgraham>
named arguments maybe
11:22
<hsivonen>
jgraham: should be easy enough to pass an object with all the named stuff inside
11:22
<jgraham>
hsivonen: Yeah, that is one possibility
11:23
<jgraham>
In fact I am already using that possibility
11:23
<jgraham>
But...
11:23
<Workshiva>
Or toString the function reference and extract the parameter names, then create an array with the right order and apply
11:23
<Workshiva>
:D
11:23
<jgraham>
Anyway, back in the land of the sane...
11:23
<jgraham>
I have a pair of functions
11:24
<jgraham>
Well a function and a method
11:24
<jgraham>
the function looks like test(func, name, properties)
11:25
<jgraham>
the methos looks like test.step(func)
11:25
<jgraham>
I have now realised I should make it possible to change the this object that is used to call func
11:25
<jgraham>
so I can have the caller pass in a this object
11:26
<jgraham>
for Test.step that is fine: test.step(func, this_obj)
11:26
<jgraham>
but what to do for test()?
11:26
<jgraham>
The logical place to put the parameter would be second test(func, this_obj, name, properties)
11:27
<jgraham>
But the common case is to use the global scope
11:27
<jgraham>
so you would always have to pass undefined
11:28
<jgraham>
Which would be bad
11:28
<jgraham>
Putting it after name would give a weird order and mean you had to pass undefined if you want to set properties
11:29
<annevk>
can't you check whether it is the global scope within test?
11:29
<jgraham>
Whether what is the global scope?
11:30
<annevk>
this?
11:30
<Philip`>
Do test(func, name, properties) and test_with_this(func, this_obj, name, properties)
11:30
<annevk>
is there a scenario where you need test rather than test_async btw with this working?
11:31
<jgraham>
annevk: test() is just for covenience because it causes the test to be run immediately
11:31
<jgraham>
seems useful
11:31
<jgraham>
Philip`: Ugly but workable
11:31
<jgraham>
I could typecheck the second parameter of course
11:31
<jgraham>
That would be the jQuery way
11:31
<jgraham>
But...
11:32
<Philip`>
Or this_obj.apply(func, ...) (or whatever the syntax is) to use the normal language facilities for changing this
11:32
<annevk>
jgraham, it's useful, but do you need "this" working when test is used?
11:32
<Philip`>
s/func/test/
11:32
<jgraham>
annevk: I thought you were the one with the use case :)
11:32
<annevk>
jgraham, my use case is only for test_async
11:33
<jgraham>
annevk: Seems odd to let it work in one case but not the other
11:33
<jgraham>
But maybe that is actually good to enfore
11:33
<jgraham>
*enforce
11:33
<annevk>
testing "this" is always needed in a callback... never really needed it elsewhere in usual tests
11:34
<annevk>
i mean, I can't remember having used it in some kind of sync variant
11:34
<jgraham>
OK I will take the simple solution of not solving the problem and calling it a feature
11:34
<annevk>
also, couldn't I use
11:35
<annevk>
test.step(function(this) { ... }) ?
11:35
<jgraham>
Yeah you could
11:36
<jgraham>
It feels nicer to optionally run the function in the expected scope though
11:36
<annevk>
actually, how would I then access it?
11:36
<annevk>
yeah, if you can do that and tell me how to make it work
11:36
<jgraham>
It's all fine
11:36
jgraham
notices he allowed extra arguments to step() get passed to the function
11:36
<annevk>
so you solve it on your side?
11:37
<jgraham>
That seems unnecessary...
11:37
<annevk>
so what do I do?
11:37
<jgraham>
annevk: Do test.step(function(){}, this)
11:37
<karlcow>
fresh redesign of Zappos http://www.zappos.com/ by HappyCog http://happycog.com/create/zappos/
11:37
<karlcow>
http://validator.w3.org/check?uri=http%3A%2F%2Fwww.zappos.com%2F&charset=%28detect+automatically%29&doctype=Inline&group=0
11:37
<karlcow>
using html5 doctype, but already invalid.
11:38
<annevk>
jgraham, and then the bit between {} can use this without problems?
11:38
<jgraham>
Yeah, I think so
11:40
<annevk>
so I have test.step(function() { assert_equals("data", e.data); this.close() }, this)
11:43
<jgraham>
Yes
11:43
<jgraham>
Although the step is run synchronously
11:43
<annevk>
and if the listener is inlined it is fine right?
11:43
<jgraham>
so you can just close after the test.step rather than in the test function
11:44
<jgraham>
listener?
11:45
<annevk>
test.step(function() { obj.onmessage = function() { test.step(function() {}) } }) or some such
11:46
<annevk>
and a test.done() after the second test.step()
11:46
<jgraham>
Yeah that should be fine I think
11:46
<jgraham>
Hard to read all on one line in irc
11:46
<annevk>
lots of braces
11:46
<annevk>
hard to type too
11:47
<annevk>
all on one line
11:48
<annevk>
jgraham, I need assert_error() or some such for something that should not be reached
11:49
<jgraham>
annevk: Yeah, known bug. Will work on it next
14:06
<zcorpan_>
http://css.dzone.com/articles/python-based-html5lib-firefoxs
14:08
<Philip`>
"html5lib [...] has gone from version 0.1 to 0.9" - wow, a nine times improvement!
14:09
<Philip`>
Should have started at version 0.01 then it'd seem even more impressive
14:10
<zcorpan_>
you should bump the version number with an order of magnitude for each new version
14:11
<Philip`>
Version 1e1, 1e2, 1e3, etc?
14:11
<jgraham>
Um, we are on version 0.90
14:11
<jgraham>
That is 10x larger than 0.9
14:11
<Workshiva>
Yeah, get it right
14:11
<boblet>
MikeSmithX: I got a review copy of HTML5 for Web Designers — will bring it next week
14:12
<boblet>
JonathanNeal MikeSmithX wish I’d been here earlier. respekt, mah homies
14:14
<boblet>
hey anyone know where the ‘HTML5 = 900 pages’ thing came from? trying to fact-check and I only get 519 pages (normal view)
14:14
<Workshiva>
Before all the splits?
14:14
<annevk>
could be the PDF version of the spec
14:14
<Philip`>
PDF?
14:15
<Philip`>
It's an entirely meaningless number anyway
14:15
<boblet>
Workshiva: woah didn’t realise so much had been split off
14:15
<boblet>
annevk: there’s a pdf version!? holy crap the cabal roolz
14:15
<Philip`>
Only slightly less meaningless than counting megabytes
14:15
<Workshiva>
Let's bikeshed over exactly how meaningless it is
14:15
<jgraham>
It's mreo meaningless than counting Mb
14:15
<jgraham>
*more
14:16
<jgraham>
Because Mb directly affect load speed
14:16
<boblet>
Philip`: true, but would like to point out how meaningless seeing as it’s being used in advertising
14:16
<zcorpan_>
but less meaningless than counting cows
14:16
<jgraham>
It is even more meaningless than counting DOM nodes
14:16
<Philip`>
jgraham: Not really, because adding lots of whitespace would make it bigger but it'd probably download in the same time because of gzip
14:16
<jgraham>
Philip`: Pretend I mean gzipped megabytes
14:17
<Philip`>
Counting crows?
14:17
<Philip`>
Oh, cows
14:17
<boblet>
the text-only version should be linked in there too
14:18
<Philip`>
jgraham: That sounds like it's getting closer to measuring the entropy of the document, which is probably a more meaningful measure
14:18
<jgraham>
yes
14:20
<boblet>
provided US letter one is 745 pages. huh — doing a print to pdf in Chrome results in the content only taking 2/3 the width of the page, with smaller text, and clocks in at 519 pages
14:20
<boblet>
but is double the official one’s size
14:21
<boblet>
I guess Prince does a good job
14:39
<boblet>
oof, that’s somewhat embarrassing. “The HTML5 spec is 900 pages and hard to read. HTML5 for Web Designers is 85 pages and fun to read”. Current spec is 745 pages. *Author View* spec is ~ 386 pages.
14:40
<boblet>
I think the polite term for that is “creative copywriting”
14:40
<jgraham>
boblet: Presumably the WHATWG version is longer?
14:40
<nimbupani>
boblet: who is saying that?
14:40
<Workshiva>
We don't do "polite" here :P
14:40
<boblet>
jgraham: those figures are from whatwg version
14:40
<jgraham>
Oh
14:40
<jgraham>
complete.html?
14:41
<jgraham>
:)
14:41
<boblet>
nimbupani: http://books.alistapart.com/
14:41
<nimbupani>
that is sad :(
14:41
<boblet>
well it was one of those gobsmack figures like the 2022 completion date that stuck
14:42
<nimbupani>
yeah. simple view is not necessarily a good idea :*
14:42
<nimbupani>
:(
14:42
<boblet>
just sad that it’s being used (unintentionally or not) in sales
14:42
<nimbupani>
exactly.
14:42
<boblet>
gotta admit, it makes for great copy
14:43
<nimbupani>
i think it trivializes the whole spec.
14:43
<boblet>
although I’d call bs on the hard to read part — whoever wrote that hasn’t read semweb specs (which are obviously written for a higher life form than I)
14:44
<miketaylr>
i think the 900 number came from the super friends
14:44
<boblet>
nimbupani: preeeeecisely. longer is oh so much better, given the spec is uniformly terse
14:44
<nimbupani>
miketaylr: not surprising :P
14:44
<boblet>
miketaylr: that’d figure. the spec prolly was that big back then before stuff got split off, as Workshiva was saying
14:45
<miketaylr>
mmmhmm
14:45
<boblet>
but all the stuff split off is completely outside the topic scope of “…for web designers”
14:46
<boblet>
(PDF still includes Microdata, and that has a 1 sentence mention)
14:46
<zcorpan_>
i think hixie changed the font size and the page margins to reduce the number of pages when people were talking about how many pages it was
14:46
<miketaylr>
heh, http://twitter.com/michai/status/13369668370 points to html5 for designers
14:47
<zcorpan_>
it seems like doing print preview on complete.html in chrome crashes chrome
14:47
<Philip`>
I've noticed far fewer criticisms of the size of the HTML5 spec than of OOXML
14:47
<miketaylr>
aha, found it
14:47
<miketaylr>
http://meyerweb.com/eric/thoughts/2009/09/07/html5-and-you/
14:48
<miketaylr>
boblet: ^^
14:49
<boblet>
miketaylr: thanks — will badger the king ;)
14:49
<miketaylr>
heh
14:49
<miketaylr>
looks like brucel perpepuates the myth here as well: http://www.brucelawson.co.uk/2010/introducing-html5-almost-in-the-can/
14:50
<boblet>
man, gonna have to put my badger suit on at this rate
14:51
<boblet>
woah — 990 pages to 506 pages just based on print style changes? Hixie musta had it set at double leading or something
14:52
<Philip`>
You should count A4 pages, rather than whatever crazy non-metric sizes are used in the US
14:52
<boblet>
Philip`: lol
14:53
<boblet>
was assuming whoever came up with the 900 page figure would have US letter as their default ;-)
14:53
<boblet>
seems i was prolly right too
14:54
<Philip`>
How many pages are taken up by the character references table?
14:55
<Philip`>
Hmm, Firefox tries to put it on 3 pages
14:55
<Philip`>
which is clearly not quite right
14:56
<boblet>
Philip`: only about 8, they’re pretty tiny
14:57
<Philip`>
Oh
15:00
<hsivonen>
Philip`: OOXML should be longer. There are some hand-wavy parts about imitating the behaviors of legacy software
15:00
<hsivonen>
(or at least were in the ECMA version. I haven't reviewed the ISO version.)
15:03
<Philip`>
When I looked at OOXML briefly, it seemed to be a lot of schemas and datatype definitions and examples, and not much that's equivalent to UA requirements
15:04
<hsivonen>
Philip`: well, the length argument was bad anyway, because ODF is even vaguer on processing
15:06
<Philip`>
Seems like most technical complaints against OOXML can apply to ODF too
15:10
<Philip`>
(Clearly we should just use Knuth's new XML-based LaTeX for word processing)
15:17
<boblet>
the print styles for HTML5 are fubar in Chrome. checking in FF now
15:18
<boblet>
(fubar = perfectly readable, but only using half the page width and in a smaller font than official)
15:18
<boblet>
one way to get page count down I guess :)
15:20
<boblet>
or I would if it’d stop beachballing :/
15:24
hsivonen
recommends using Prince for printing the spec
15:25
<hsivonen>
does any browser implement page number counters in generated content?
15:27
<boblet>
hsivonen: Webkit has been working on implementing print-related CSS, but I don’t know how far they are along
15:27
<boblet>
yeah I probably should install Prince huh
15:28
<boblet>
Firefox is currently stuck with a completely white modal pane that came down from under the page title. wonder what it’s meant to contain
15:42
<fagan>
Is what im talking about on the list on topic?
15:42
<fagan>
I hope it is
15:42
<fagan>
its still under the youtube problems with the video tag discussion I hope
16:00
<JonathanNeal>
Goodmorning!
16:16
<OmerSd>
Hi!
16:17
<JonathanNeal>
Hi^
16:17
<boblet>
JonathanNeal: I’m joining your gang yo
16:17
<boblet>
all we need is one more MikeSmithXX and we’ll get the nudie version
16:17
<JonathanNeal>
boblet, oh you heard?
16:18
<JonathanNeal>
We gotta collect all the pictures.
16:18
<JonathanNeal>
Take a picture doing the sign
16:18
<JonathanNeal>
I'll add it to the page.
16:18
<boblet>
JonathanNeal: tweeted yo, we be forming a posse
16:18
<OmerSd>
Have some way change to full screen mode in js?
16:18
<JonathanNeal>
oh it got tweeted, damn, now the url is stuck on my sandbox :P
16:19
<boblet>
lols — redirects are the future
16:19
<boblet>
(or your future)
16:19
<boblet>
OmerSd: standard JS screen resizing code should work fine with HTML5, but a bigger question would be … why would you want to do that?
16:19
<JonathanNeal>
I guess so.
16:20
<boblet>
orz, I thought it was prime-time :D
16:20
<OmerSd>
i wanna make a web appliction
16:20
<JonathanNeal>
HTML5, doctype no, gang sign yes.
16:20
<boblet>
OmerSd: that’s not a good reason. if Gmail resized my browser every time I used it, I’d stop using it
16:21
<OmerSd>
I'll do a option to full screen
16:21
<boblet>
OmerSd: having a button to make the browser window full size may be a nice feature in some situations, but I’d do some user testing before making it the default
16:21
<boblet>
OmerSd: oh ok :)
16:22
<OmerSd>
:)
16:22
<boblet>
OmerSd: well I’ve got no idea but here’s the first thing that came up in Google: http://www.pbdr.com/jscript/windfull.htm
16:23
<OmerSd>
Thanks :)
16:24
<boblet>
for anyone interested in PDF size, the official A4 version was 707 pages, a full version I made with Prince was 709 pages (slightly different copy), and an author vide version clocked in at 378 pages.
16:24
<boblet>
I officially call bullshit on both parts of “The HTML5 spec is 900 pages and hard to read.”
16:26
<OmerSd>
How i can calucate the maximum size that the user can see in the document(without full screen)?
16:26
<Philip`>
boblet: Hmm, you think it's not hard to read?
16:27
<boblet>
Philip`: I think parts of it are still confusing, but the author view isn’t in general no
16:28
<boblet>
Philip`: of course this could just be in comparison to typical W3C specs
16:29
<boblet>
Philip`: it doesn’t have Jeremy’s witty turn of speech certainly, but hard to read is an exaggeration
16:29
Philip`
has been reading it for three years and still thinks it's hard to read :-)
16:29
<Philip`>
though maybe that's because I mostly read the non-author bits
16:30
<boblet>
Philip`: I did have that as a caveat :P the implementor bits seem brutally precise, but I have no idea what they actually mean most of the time
16:30
<boblet>
it’s lucky for the interwebs I am an author not an implementor
16:49
<OmerSd>
Have some api spec for html5 like javadoc?
16:50
<ako>
the biggest problem with that is that the current specs aren't very well suited for automatic generation
16:51
<boblet>
OmerSd: api specs are in the HTML5 spec for basic things http://www.whatwg.org/specs/web-apps/current-work/multipage/
16:52
<TabAtkins>
JonathanNeal: I don't see quite how the H is formed in the gang sign.
16:52
<boblet>
OmerSd: other things that have been split out like Web Sockets are in different specs: http://wiki.whatwg.org/wiki/FAQ#What_are_the_various_versions_of_the_spec.3F
16:52
<JonathanNeal>
TabAtkins, the 3 and the 2 put together make an h
16:52
<JonathanNeal>
they also make a t, the 3 forms the m
16:52
<JonathanNeal>
and any of them could be the l.
16:52
<boblet>
TabAtkins: well you’re obviously not eligible for the posse
16:52
<TabAtkins>
JonathanNeal: Oh, I see it now.
16:53
<miketaylr>
thuglife is hard.
16:53
<TabAtkins>
The problem is too many thugs, not enough signs to throw.
16:54
<boblet>
miketaylr: too funny
16:54
<miketaylr>
:)
16:55
<JonathanNeal>
Updated: http://sandbox.thewikies.com/html5-gang-sign/
16:56
<boblet>
JonathanNeal: btw, your buddy making the peace sign with the palm in? that’s called “the fingers” in Oz & NZ (UK too?) and is equivalent to flippin the bird
16:56
<JonathanNeal>
boblet, ut oh.
16:56
<boblet>
heh, it’s cool :)
16:56
<JonathanNeal>
Maybe I'll update the video tonight.
16:56
<boblet>
it’s a gang sign instructional video after all
16:56
<boblet>
the guy’s expression matches perfectly too
16:57
<TabAtkins>
boblet: Interesting. It's just ordinary sign for when you're ready to chillax while getting krunk in the US.
16:58
<boblet>
yeah, apparently one of the Bushes was waving like that from his limo when he visited Oz. went down a treat ;-)
16:58
<boblet>
prolly Dubya, would be in keeping with general foreign policy
16:58
<TabAtkins>
Clearly Bush was getting krunk in his limo, all knocking back 40s with his bitches.
16:59
<boblet>
hahaha, clearly
17:07
<Philip`>
boblet: (UK too)
17:07
<Philip`>
boblet: (See e.g. http://news.bbc.co.uk/dna/h2g2/A11047132)
17:22
<boblet>
har http://en.wikipedia.org/wiki/V_sign#V_sign_as_an_insult
17:22
<boblet>
a little different to what I remembered, but still very krunk
17:27
<boblet>
MikeSmithXX: btw posted you a bunch of HTML5Doctor flowcharts today. now that I think about it I coulda given them to you next week. oh well. distribute with gay abandon
17:27
<boblet>
nn
18:01
<paul_irish>
TabAtkins: you wanna get a tag team gang sign pic?
18:01
<TabAtkins>
paul_irish: Let's rock this.
18:06
<JonathanNeal>
yeaaaa
18:16
<JonathanNeal>
quite a few submissions.
18:18
<TabAtkins>
Hehe. Me + paul_irish clearly the most attractive of the bunch.
18:20
<paul_irish>
+1
18:21
<miketaylr>
jeeze thanks
18:29
<TabAtkins>
Sorry, miketaylr, but you just can't compete with the raw sexual attraction we pump out.
18:29
<miketaylr>
ahahah
18:29
miketaylr
concedes
19:27
<TabAtkins>
Man, the box model's subtleties are complicated.
19:28
<JonathanNeal_>
fix it, TabAtkins
19:28
<ojan>
TabAtkins: ?
19:28
<TabAtkins>
Nah, the things I'm banging on can't be "fixed" because they're an integral part of the web at this point, and more or less make sense anyway.
19:28
<TabAtkins>
ojan: Working on issue 138, involving what happens when floats and blocks are children of a relpos inline.
19:29
<TabAtkins>
floats and blocks act differently here.
19:29
<ojan>
TabAtkins: fun
19:30
<TabAtkins>
It makes sense, though. Floats position themselves solely based on their containing block (and other floats with the same CB), and the inline doesn't form a containing block for it.
19:30
<TabAtkins>
So the float doesn't move with the inline when it gets relpos'd, but the block does.
19:30
<ojan>
TabAtkins: yuck.
19:36
<TabAtkins>
ojan: Yuck indeed. Floats suck in *so many different ways*. I can't help but think we'd have been better off from the beginning if we'd just defined a way for abspos elements to intrude and push text around them.
19:37
<TabAtkins>
That would have prevented the worst abuses of float-based layouts, too.
19:38
<ojan>
TabAtkins: totally.
19:38
<ojan>
TabAtkins: that could still be worth doing.
19:38
<TabAtkins>
Pretty sure MS has some custom stuff doing that already.
19:38
<TabAtkins>
Don't remember if that was some vendor-specific public stuff, or a private build I saw.
19:39
TabAtkins
notes that it would also probably have increased the drive to make abspos more powerful.
19:39
<TabAtkins>
Now I just need a time machine. First stop, Ben Franklin. Second stop, Hakon and Bert in the 90s.
19:44
<Philip`>
Remember to put a battery in Ben's kite so it'll get recharged and then use it to power the Chron-O-John
19:45
<TabAtkins>
Will do. That's a convenient time to drop in anyway, since I'll be visiting him about his charge convention.
20:28
<TabAtkins>
Question: it appears that Chrome's implementation, at least, of <canvas> only permits integers in the CanvasPixelArray object of ImageData, and floors the value upon assigning a float. I can't find anything in particular to support that in the spec. Is there?
20:29
<Philip`>
setter void (in unsigned long index, in octet value);
20:30
<Philip`>
See WebIDL for conversions to octet
20:30
<TabAtkins>
Ah, octet. Interesting.
20:30
<TabAtkins>
That would explain some of my errors, then.
20:37
<TabAtkins>
Excellent. Switching over to using a number for my intermediate calculations and then assigning to the ImageData at the end fixed everything.
21:43
<JonathanNeal_>
sup thugs
21:44
<JonathanNeal_>
html5 gangstas
21:44
<TabAtkins>
Yo.
23:11
<KevinMarks>
grr at firefox fro not supporting mp3 in <audio> and not failng properly
23:12
<TabAtkins>
mp3 isn't royalty-free. What do you mean "not failing properly"?
23:13
<KevinMarks>
if it refused to play it and displayed the fallback code, that would be lovely
23:13
<KevinMarks>
I coudl put the flash emebd inside the audio tag
23:13
<KevinMarks>
but instead it displays a completely useless grey box with an X in it
23:13
<KevinMarks>
so I am going to have to default to flash
23:14
<KevinMarks>
if I want to use audio declaratively
23:14
<TabAtkins>
That's not how <audio> and <video> are supposed to fail. Their fallback contents are for legacy browsers, not for unsupported codecs.
23:14
<KevinMarks>
yes, adn that supposition si wrong
23:15
<TabAtkins>
It's not "wrong". It can't be. It can be different than what you want, though.
23:15
<KevinMarks>
er
23:15
<TabAtkins>
Anyway, just throw a tiny bit of javascript at it that asks the <audio> if it can play mp3s, and replaces it if it can't.
23:15
<KevinMarks>
I am saying that the spec is wrong
23:16
<KevinMarks>
because it breaks if I write declarative code
23:16
<TabAtkins>
No, the spec can be different thatn what you expect, or suboptimal. It can't be "wrong" in this case, because it's not declaring a statement of fact.
23:17
<KevinMarks>
the spec is misdesigned then
23:17
<KevinMarks>
happy now?
23:17
<TabAtkins>
Yes. ^_^
23:18
<TabAtkins>
Anyway, the discussion about <audio>/<video> fallback has been had several times. It is almost certain not to be changed at this point, because the way you want it to act requires crazy code on the browser side.
23:19
<TabAtkins>
What happens, for example, if your <audio> starts with no playable <source>s, but then you later add one it can play with js?
23:19
<TabAtkins>
Does it stop and tear down the Flash? Does it ignore the new <source>?
23:20
<TabAtkins>
Does this mean that the order in which you insert <source>s and other contents can have an effect on what happens?
23:21
<KevinMarks>
so instead you require crazy code on the client side
23:21
<TabAtkins>
You have a very low definition of "crazy".
23:22
<TabAtkins>
It's like three lines, plus whatever js is necessary to do an embed.
23:25
<Cheery>
hi
23:26
<Cheery>
I've got a question about WebSocket. Are you sure the handshake is good to be GET?
23:26
<hsivonen>
KevinMarks: It's easier to offer Vorbis than to do a Flash fallback :-)
23:26
<KevinMarks>
no it fucking isn't
23:26
<Cheery>
I've understood GET can be cached and such.
23:27
<Cheery>
so it makes me wonder, shouldn't it be POST -method?
23:27
<KevinMarks>
Firefox will happily route an mp3 to a plgin if I use <iframe>
23:27
<Cheery>
for anything else, I happily ignore flash if I can.
23:27
<KevinMarks>
or object
23:27
<KevinMarks>
or embed
23:27
<hsivonen>
KevinMarks: not on Maemo
23:27
<franksalim>
Cheery, a 101 switching protocols response cannot be cached
23:28
<KevinMarks>
course Android is worse
23:28
<TabAtkins>
KevinMarks: Plugins are a completely separate story. They just happened to be on the other side of the "worth compromising principles over".
23:28
<KevinMarks>
it 'supports' <audio>
23:28
<Cheery>
franksalim: which portion of WebSocket -spec makes it a switching protocol?
23:28
<KevinMarks>
except no formats at all
23:29
<hsivonen>
KevinMarks: huh. that's silly
23:29
<TabAtkins>
Agreed, that's pretty silly.
23:29
<KevinMarks>
Look: http://homepage.mac.com/kevinmarks/audiotest1c.html
23:29
<franksalim>
Cheery, the server response HTTP/1.1 101 WebSocket Protocol Handshake
23:29
<hsivonen>
KevinMarks: you are talking about the Android Browser, right? not Firefox on Android?
23:29
<Cheery>
101 -there.. oh IC
23:29
<KevinMarks>
yes, I'm talking about the one with an installed base
23:30
<Cheery>
franksalim: thank you from clarifying. :) I have more questions about websocket.
23:30
<Cheery>
why such complicated handshaking mechanism? why three keys?
23:31
<Cheery>
also, where's UDP?
23:31
<Cheery>
(which would be a welcome introduction)
23:32
<Cheery>
and.. that's it. pretty much other in websocket makes sense to me.
23:33
<AryehGregor>
The handshake is to avoid various security problems. A lot of care is needed, since any web page can trigger a WebSocket request, so the browser needs to be sure it's not being abused somehow.
23:33
<AryehGregor>
(dunno all the specifics)
23:33
<franksalim>
Cheery, UDP would need a different protocol and NAT/firewall/proxy traversal work
23:33
<AryehGregor>
UDP doesn't fit into those requirements at all.
23:34
<AryehGregor>
You'd have to get rid of the handshake, but then you need something else to ensure security (like what?).
23:34
<franksalim>
I am interested in the thinking behind the three key design. Has that been described in public?
23:34
<Cheery>
AryehGregor: not to mention it's connectionless.. yeah.
23:34
<AryehGregor>
I'm sure it's been described in public.
23:34
<AryehGregor>
All the WebSocket development is public as far as I know.
23:35
<franksalim>
AryehGregor, I would be grateful for a link :-)
23:35
AryehGregor
doesn't know where it is
23:35
<AryehGregor>
You'd think websockets.org would mention it.
23:35
<AryehGregor>
At the IETF somewhere, I guess?
23:36
<Cheery>
Maybe I'll have to resort with TCP sockets only.. websockets are anyway already quite all right.
23:36
<AryehGregor>
I think this is it: https://www.ietf.org/mailman/listinfo/hybi
23:36
<Cheery>
planning a multiplayer write into platformer game: http://boxbase.org/fun/knights/
23:36
<AryehGregor>
There's a lot of archived stuff.
23:36
<franksalim>
I found the monster thread about the handshake redesign
23:36
<AryehGregor>
I'm pretty sure it's a recurring topic there.
23:37
<franksalim>
but IIRC, it started with a proposed design and was not designed on list
23:37
franksalim
goes off to read