00:16
<dglazkov>
yikes! who killed html spec?!
00:16
<dglazkov>
http://www.whatwg.org/specs/web-apps/current-work/multipage/
00:16
<divya>
lolol
00:16
<dglazkov>
Hixie, I have to do my homework!!!
00:17
<divya>
"Never ascribe to political protest what can be adequately explained by incompetence"
00:17
<divya>
jgraham's words ring true
00:17
<dglazkov>
Hixie: I think divya just called you a dumbass
00:17
<divya>
>_>
00:17
<dglazkov>
:D
00:18
<divya>
Hixie: dglazkov is trying to shift blame so he doesnt do HOMEWORK
00:18
<dglazkov>
<_<
00:20
<dglazkov>
I worry that there's a new 404.html file there. Did multipage version finally die? MULTIPASS! MULTIPASS!
00:33
<kennyluck>
Oh man Hixie, you are breaking the Web.
00:45
<Hixie>
um... sopa blackout? yeah, let's say that's what happened.
00:59
<xtgyal>
Hi, the W3C Validator @ validator.w3.org gives me an error stating <meta name="copyright" isn't registered, redirecting me to add it @ http://wiki.whatwg.org/wiki/MetaExtensions
01:00
<Init--WithStyle->
Hello everyone... i am trying to get some guidance on what tools to use for rendering vector graphics to an HTML canvas element... is this the right place to be asking?
01:03
<TabAtkins>
xtgyal: I suggest following the instructions.
01:04
<TabAtkins>
Init--WithStyle-: I just use SVG when I need vector graphics.
01:04
<TabAtkins>
But otherwise I hear paper.js is pretty cool?
01:04
<xtgyal>
the WHATWG wiki redirected me to here :-p
01:04
<TabAtkins>
Oh, so you need an account created.
01:05
<TabAtkins>
Hixie: ^^^
01:05
<Init--WithStyle->
TabAtkins: I have an engine that I would like to use that is based on the html canvas element... i want to include some extreme zooming gameplay and so I need to either use LOD or vector graphics to provide for that level of zoom
01:05
<Hixie>
xtgyal: /msg me your e-mail address and i'll create one
01:05
<Init--WithStyle->
Is there some way to get the performance benefits of a sprite, while still enabling the zoom-ability of vector graphics?
01:05
<xtgyal>
I'm not sure, if the issue is simply that that attribute is not listed, then yes, is that where the W3C validator is reading from?
01:06
<Init--WithStyle->
some way to "draw" from a vector graphic at run time for multiple levels of detail?
01:06
<xtgyal>
http://validator.w3.org/check?uri=http%3A%2F%2Fnikki.dyndns.info%2Ftemp%2Ftemplate.htm&charset=%28detect+automatically%29&doctype=Inline&group=0
01:06
<TabAtkins>
Init--WithStyle-: Drawing from SVG works in some browsers, though it also taints the canvas in some of them.
01:06
<TabAtkins>
Init--WithStyle-: Several canvas libraries support js-backed vector graphic syntaxes.
01:07
<Init--WithStyle->
js-backed vector graphics syntaxes.. what do you mean exactly?
01:08
<TabAtkins>
Init--WithStyle-: Check out, for example, paper.js. Grab one of their examples and hit the "Source" button in the upper-right corner.
01:09
<Init--WithStyle->
i'm looking at it right now... does it draw the whole thing every frame or only upon the shape changing?
01:09
<TabAtkins>
Other libraries often do something similar - you create some objects and maybe compose them in a scene, and the library handles drawing.
01:09
<TabAtkins>
Init--WithStyle-: I dunno.
01:09
<TabAtkins>
Never used one of them. Just pointing out options.
01:09
<Init--WithStyle->
right... what i'm trying to avoid is having the CPU draw the entire shape on every frame
01:09
<Init--WithStyle->
massive resource cost
01:10
<Init--WithStyle->
instead i would like to use vector processing to composite the shape on the first frame, and then load it as if it were a "sprite" or bitmap image from then on
01:10
<Init--WithStyle->
basically like a single render pass into a bitmap image
01:10
<TabAtkins>
If you can draw it into a separate canvas, you can then draw that canvas into your main one.
01:11
<xtgyal>
Is the issue with the validator or simply it not being listed on the wiki?
01:14
<xtgyal>
@Hixie how do I message you?
01:15
<Init--WithStyle->
TabAtkins: do you know a forum where I could post a more elaborate example of what i'm saying and perhaps get some additional feedback?
01:58
<Init--WithStyle->
TabAtkins: I see what you mean about rendering to a separate canvas
02:01
<Init--WithStyle->
could render an image via SVG, save it to the DOM as a png or such and then pass it into the canvas using drawimage() if I am understanding things correctly
02:01
<Init--WithStyle->
then later on when the image isn't needed anymore, the client could just drop the png off the dom
02:16
<MikeSmith>
wow, hybi wg is rechartering
02:16
<jamesr_>
you know what www-style could use? more deep space nine fan fiction analogies
02:20
<MikeSmith>
well, that's true of all mailing lists, isn't it?
03:47
<Init--WithStyle->
Is it possible to convert a vector image into code that will draw that image automatically?
04:34
<MikeSmith>
dglazkov: you were asking about when Hixie introduced transparent content models
04:34
<MikeSmith>
http://html5.org/tools/web-apps-tracker?from=466&to=467
04:41
<MikeSmith>
hmm, actually it had to have been some time before that change
04:41
<MikeSmith>
that's just the point at which Hixie actually defined it
04:41
<MikeSmith>
but it seems he'd already been using that term before adding the actual definition
05:39
<Init--WithStyle->
when creating an SVG image in a browser, after the first frame, do modern browsers cache and then display that image as a bitmap instead of re-drawing it, provided there is no scaling/alpha changes going on?
07:16
<hsivonen>
https://www.w3.org/Bugs/Public/show_bug.cgi?id=15614 sad SEO
07:19
<MikeSmith>
hsivonen: I guess that one is too subtle for me
07:19
<MikeSmith>
I don't know what the inferred complaint is...
07:41
<hsivonen>
MikeSmith: that distribution global is a validation error
07:42
<hsivonen>
do SEO folks really think that the default mode of the Web is something other than global distribution?
07:42
<MikeSmith>
ah
07:43
<MikeSmith>
holy god
07:43
<MikeSmith>
people actually use that?
07:44
<MikeSmith>
nuts
07:45
<hsivonen>
MikeSmith: I believe some people are using that
07:45
<MikeSmith>
who comes up with this stuff?
07:45
<MikeSmith>
hsivonen: yeah, I see it at http://www.metatags.org/meta_name_distribution and elsewhere
07:46
<hsivonen>
"ALMOST NO INFLUENCE"
07:46
<MikeSmith>
heh
07:47
<Hixie>
i'm looking forward to the day where we go through all these meta names that people registered, and kill off the ones that are wasting people's time
07:48
<MikeSmith>
well
07:48
<Hixie>
not as "unregistered non-conforming please see the wiki to add it" but as "non-conforming obsoloete do not use"
07:48
<MikeSmith>
that would be most of the list
08:15
<hsivonen>
uh oh. dglazkov wants to modify the parsing algorithm
08:15
<MikeSmith>
yeah
08:16
<MikeSmith>
not an encouraging sign
08:16
<MikeSmith>
RED FLAG
08:38
<hsivonen>
activities for today: document.close()
08:39
<hsivonen>
nope, document.write and document.close still aren't fully "done" :-(
08:48
<MikeSmith>
hsivonen: that doesn't sound like fun at all
08:49
<hsivonen>
MikeSmith: it seems to me that these things aren't falling out of use. instead, people are coming up with new uses for them and push the boundaries
08:49
<MikeSmith>
yeah
08:50
<MikeSmith>
I remember seeing some specific case of that a few weeks back
08:50
<MikeSmith>
blog posting
08:50
MikeSmith
tries to recall the details
08:50
<MikeSmith>
anyway, not an encouraging trend
09:49
<MikeSmith>
hsivonen: heads-up lest I forget later: I had to also make a change to the validator site/script.js file to match the updates I made to the s.validator.nu URIs
09:49
<MikeSmith>
so if you don't deploy that directly from the repo sources, it will need to be updated on validator.nu however else you do it
09:50
<MikeSmith>
the change affects whether the parser option is auto-set to XML or not
09:50
<MikeSmith>
the isHtmlCompatiblePreset stuff
10:01
<MikeSmith>
wow, people do actually use the "bring your own schema" feature of validator.nu
10:01
<MikeSmith>
http://bugzilla.validator.nu/show_bug.cgi?id=893
10:01
<MikeSmith>
at least one person
10:02
<MikeSmith>
though he misunderstands where the bug is
10:02
<MikeSmith>
oh
10:03
<MikeSmith>
maybe it's because it's not evaluating the embedded schematron rules
10:20
<hsivonen>
MikeSmith: thanks
10:21
<MikeSmith>
hsivonen: does jing handle embedded schematron rules?
12:02
hsivonen
gets nearer implementing innerHTML for Document for HTML documents but not XML documents
12:02
<hsivonen>
isn't it great how XML involves doing stuff like this twice
12:09
<david_carlisle>
hsivonen: the obvious solution is to get the whole world to switch to xml, and just do the xml one, then you only need do it once.
12:09
<david_carlisle>
or at least it seemed that way in 1998
12:11
<jgraham>
Say XML twice and Bang! david_carlisle appears :)
12:12
<david_carlisle>
jgraham: Actually I just (this week) switched our public facing documentation set away from xml to html(5)(+mathml+svg) (in the next release) then xml can be our private secret
12:15
<hsivonen>
david_carlisle: BTW, I'm starting to believe in versionless MathML. The existence of SVG 1.2 Tiny makes me say SVG 1.1 with a version number, still.
12:17
<david_carlisle>
hsivonen: Ah OK, if that belief results in you wanting any actual code or schema changes from me let me know....
12:18
<hsivonen>
david_carlisle: ok. I will need to reconsider my position on the schema.
12:19
<hsivonen>
Ms2ger: the DOMParser spec doesn't say what document URL the DOMParser-returned doc should have
12:19
<Ms2ger>
What should it have?
12:19
<hsivonen>
Ms2ger: I don't know. That's why I took a look at the spec.
12:20
hsivonen
tried naively to use the spec as a guide for implementation rather than the other way round
12:20
Ms2ger
tests
12:21
<Ms2ger>
And every time I have to do that, I remember how horrible this API is
12:21
<hsivonen>
Ms2ger: the chrome-exposed API is even more horrible than the specced Web-exposed part...
12:22
<hsivonen>
the way it is implemented makes me sad
12:23
<Ms2ger>
Hm, is document.URL supposed to be implemented?
12:24
<hsivonen>
Ms2ger: FWIW, nsContentUtils::CreateDocument says I can't pass nsnull as the URL, so I guess the document URL can't be null...
12:24
<Ms2ger>
I was about to suggest null :/
12:24
<hsivonen>
Ms2ger: not sure. documentURI might work better
12:25
<Ms2ger>
That returns null in Chrome and Opera
12:25
<hsivonen>
interesting
12:26
<Ms2ger>
110 // Note: can't require that aDocumentURI/aBaseURI/aPrincipal be non-null,
12:26
<Ms2ger>
111 // since at least one caller (XMLHttpRequest) doesn't have decent args to
12:26
<Ms2ger>
112 // pass in.
12:26
<Ms2ger>
In NS_NewDOMDocument
12:26
<Ms2ger>
So why doesn't nsContentUtils accept null?
12:27
Ms2ger
tries to remember how XHR works
12:27
<hsivonen>
Ms2ger: I have no idea. I'm just reading the API docs in the .h
12:27
<hsivonen>
Ms2ger: XHR does pass non-null args these days
12:28
Ms2ger
looks at bonsai
12:32
<Ms2ger>
hsivonen, I see no reason to assume that the "must not be null" ever had a reason
12:32
<hsivonen>
Ms2ger: yay
12:32
<hsivonen>
next up: figuring out what to pass as the principal
12:34
<hsivonen>
ooh. I can just read existing source a few lines down
12:42
<Ms2ger>
Hmm, it's only a month ago I touched that spec
12:50
<hsivonen>
Ms2ger: so it looks like Gecko inherits the documnet URL from the DOMParser creator
12:51
<Ms2ger>
Yeah
12:52
<hsivonen>
Ms2ger: interesting that the content type isn't compared ASCII case-sensitively
12:54
<hsivonen>
I'm not complaining. just surprised.
12:55
<Ms2ger>
Er
12:55
<Ms2ger>
It is case-sensitive, isn't it?
12:56
<hsivonen>
Ms2ger: yeah. typo above. I meant *in*sensitively
12:57
<Ms2ger>
Good
12:57
<Ms2ger>
Your comments keep making me doubt my sanity :)
12:57
<hsivonen>
I almost wrote a bug out of logic and was about to make it ASCII case-insensitive like this kind of things usually are
12:57
<hsivonen>
Ms2ger: sorry
12:57
<Ms2ger>
Now, so do my own comments
12:58
<Ms2ger>
So maybe it's not your fault :)
12:59
<hsivonen>
Ms2ger: are you OK with expanding the DOMParser spec with HTML support when content type is text/html?
12:59
<Ms2ger>
I guess
13:00
<hsivonen>
ok. only for the Web-exposed part
13:00
<hsivonen>
I have no intention of supporting HTML in the methods we only expose to Firefox extensions
13:01
<Ms2ger>
Hrm, I want a assert_instanceof that gives less obscure error messages
13:01
<Ms2ger>
jgraham, ^
13:03
<Ms2ger>
Interesting how we return an SVGDocument even if we hit a parse error
13:05
<Ms2ger>
Updated the spec
13:08
<Ms2ger>
Oh, and I wouldn't mind at all if you made DOMParser use nsCString for the content type ;)
13:38
<jgraham>
Ms2ger: Less obscure than what?
13:38
<Ms2ger>
jgraham, "got false expected true"
13:39
<jgraham>
Less onscure than assert_true(foo instanceof bar) then?
13:39
<jgraham>
*obscure
13:39
<jgraham>
Well that sounds possible
13:40
<Ms2ger>
Yeah
13:40
<jgraham>
Although istanceof can be evil in multiple window situations I guess it is fair enough if you use it explicitly
13:41
<Ms2ger>
Maybe have it take an explicit window argument to make the point?
13:43
<jgraham>
Not sure how that would work
13:44
<Ms2ger>
assert_instanceof(foo, window, "DOMParser")
13:44
<jgraham>
But the rightmost argument need not be a property of window directly
13:44
<Ms2ger>
Oh?
13:44
<jgraham>
Although it typically is for DOM cases
14:29
<hsivonen>
Ms2ger: I think DOMParser.parseFromString("...", "text/html"); should always set the charset of the returned document to UTF-8 regardless of <meta>s
14:29
<hsivonen>
to avoid complexity
14:29
<hsivonen>
Ms2ger: ok?
14:29
<Ms2ger>
Makes sense to me
14:29
<hsivonen>
good
16:44
<MikeSmith>
god there is such a huge cost to formal process
16:44
<MikeSmith>
you create a formal decision policy with good intentions
16:45
<MikeSmith>
and it becomes a frankenstein monster
16:55
<dglazkov>
good morning, Whatwg!
16:56
<dglazkov>
hsivonen, MikeSmith: tell me more!
16:57
<MikeSmith>
dglazkov: ask Adam Barth how much he loves implementing parser changes
16:58
<dglazkov>
well, this time we'll be doing it for him :)
16:58
<MikeSmith>
also, as far as when transparent content models first appeared, somewhere around http://html5.org/tools/web-apps-tracker?from=466&to=467
16:58
<dglazkov>
MikeSmith: thanks!
16:58
<MikeSmith>
dglazkov: cool, so you will also be taking over ownership of the parser code and fixing all the bugs that come along!
16:59
<dglazkov>
MikeSmith: of course not! we'll just drop the code and run like a wind :P
16:59
<MikeSmith>
heh
17:00
<MikeSmith>
seriously though, you are going to be facing an uphill battle if you need to add more stuff that requires adoption-agency behavior and such
17:00
<MikeSmith>
just you guys may be willing to implement in your parser code does not mean others will
17:00
<MikeSmith>
*just because you guys
17:01
<dglazkov>
but what would be another way of getting the same result?
17:01
<MikeSmith>
a spec that does not require that result
17:01
<dglazkov>
insertion points are critical in shadow DOM, and having quirky rules about where they may and may not go seems ... deficient.
17:01
<MikeSmith>
I hear what you are saying
17:02
<MikeSmith>
just trying to give you a heads-up
17:02
<MikeSmith>
but maybe I'm just being overly pessimistic
17:04
<dglazkov>
hey, I've been rolling this Web components rock uphill for 2 years now, what's one more uphill battle? :)
17:04
<dglazkov>
would love to hear any feedback on what could be done though
17:04
<MikeSmith>
that rock has being rolled uphill for longer than 2 years now
17:05
<MikeSmith>
or a similar rock at least
17:05
<MikeSmith>
ask Jonas Sicking
17:05
<dglazkov>
yup
17:05
<MikeSmith>
it is nice to be getting closer
17:05
<smaug____>
I thought we were quite close at some point (with XBL2)
17:07
<MikeSmith>
I think Jonas several times had set aside time for implementing XBL2 but always ended up getting pulled away to work on other things
17:07
<smaug____>
yup
17:07
<MikeSmith>
and Indexed DB came up somewhere along the line
17:07
<dglazkov>
smaug____: in this particular case, XBL2 needs the same changes to parser. Shadow DOM spec is basically an extraction of XBL2's shadow DOM functionality.
17:07
<smaug____>
(or he is just lazy :p )
17:07
<MikeSmith>
well
17:08
<MikeSmith>
we know Jonas ain't lazy
17:09
<dglazkov>
smaug____: and I honestly don't want to go on the same merry-go-round about XBL2 and why it's not the right spec. I wrote about it, I talked about it, and demonstrated clearly it's not the right thing. Zbarsky agreed with me, Sicking agreed with me, Deakin agreed with me. Why keep bringing it up?
17:09
<MikeSmith>
dglazkov: I don't think anybody was bringing it up to suggest we go back to that
17:09
<MikeSmith>
just in the context of the history
17:09
<MikeSmith>
is all
17:10
<dglazkov>
MikeSmith: ok :)
17:10
<smaug____>
just because I'm not sure (yet) web components / shadow DOM is the right thing either
17:10
smaug____
needs to find some time to file bugs
17:10
<smaug____>
the event handling certainly needs changes
17:11
<MikeSmith>
dglazkov: I definitely don't mean to discourage you
17:11
<MikeSmith>
quite the opposite
17:11
<MikeSmith>
please keep doing what you're doing :)
17:11
<dglazkov>
MikeSmith: thanks! :)
17:12
<dglazkov>
smaug____: please do! otherwise we'll keep having ivory tower discussions about philosophical uncertainty of the future. We gotta start moving.
17:16
<smaug____>
dglazkov: btw, I don't what to discourage you. I just think we need to be very careful when adding new major APIs to the web
17:16
<dglazkov>
smaug____: I agree. And I know you have expertise. And I want some of that! :)
17:16
<smaug____>
new APIs are proposed probably every month, and luckily lots of them aren't "accepted", at least not immediately
17:18
<smaug____>
(scary, someone thinks I have some expertise)
17:19
<dglazkov>
well, you talk like you do! :P
17:19
<smaug____>
I've been hacking browsers only 9 years
17:19
<dglazkov>
actually, I just recently changed event dispatch section and would appreciate a closer look.
17:20
<dglazkov>
that's close to a 100 in internet years.
17:20
<dglazkov>
you're basically a great grandpa of the browsers.
17:20
<Ms2ger>
Yeah, even Glazman has been around for only 12 years
17:21
<smaug____>
dglazkov: anyway, I'll try to give more feedback...hmm...during next week.
17:21
<smaug____>
I'll meet Boris et al
17:21
<dglazkov>
my first (well, technically second) WebKit commit was just over 3 years ago. Though I was a Web developer since 1995 prior to that
17:22
<dglazkov>
smaug____: cool! you coming to boston?
17:22
<Ms2ger>
Paris
17:22
<smaug____>
DOm stuff in Paris
17:22
<smaug____>
DOM
17:22
<dglazkov>
in Paris, it's Dom Pérignon
17:23
<dglazkov>
it's a non-standard extension.
17:24
<dglazkov>
smaug____: let me know if you want me to VC in or something like that.
17:44
<Ms2ger>
Go, irccloud
17:44
<smaug____>
someone is really using irccloud o_O
19:15
<bga>
http://web.mit.edu/newsoffice/2012/faster-fourier-transforms-0118.html
19:31
Philip`
wonders if the people doing distributed computation like Folding@home have considered purchasing ads from an ad network that lets them run arbitrary scripts, so they can use the CPUs of a huge number of web users for an extremely low cost
19:32
<TabAtkins>
That's an extremely good idea.
19:32
<Ms2ger>
People were doing such a thing when workers were first implemented, but on their own sites
19:43
<Philip`>
Ms2ger: Yeah, the problem there is the lack of scalability - you can't easily get more visitors to your own site, whereas using an ad network gives you EC2-style cloud computing (but probably much cheaper) since the computing power is proportional to your spending and is effectively limitless
19:46
<Philip`>
Apparently ads are often on the order of $1 per 1000 views, and if people spend maybe 10 seconds on a page then you get 3 hours of viewing for $1
19:48
<Philip`>
and each visitor probably has something like 2GHz dual-core, which is probably similar CPU to an EC2 high-CPU medium instance
19:48
<Philip`>
which is, um, about $0.17/hour
19:48
<Philip`>
(on current EC2 prices)
19:49
<Philip`>
so maybe it's not very economical :-(
19:49
<Philip`>
although you could exploit the user's GPU too
21:00
<bga>
http://www.youtube.com/watch?v=tErVvempcCU
21:05
<erlehmann>
oh funny
21:05
<erlehmann>
i has a shell script irc client :3
21:07
<erlehmann>
https://github.com/erlehmann/uii/blob/master/uii
21:08
<erlehmann>
but it still needs query support
21:21
<jamesr_>
Philip`, i've seen pages that spin up worker scripts to do bitcoin mining
21:21
<jamesr_>
i don't think many ad networks let you run arbitrary script, though
21:21
<jamesr_>
that'd be a terrible idea
21:39
<erlehmann>
i can attest to the fact that ad networks are a terrible idea
21:39
<erlehmann>
i know a big german site that hangs occasionally because the ad loading was blocking, multiple document.writes argl
21:48
<Philip`>
jamesr_: I was thinking of e.g. http://seclab.stanford.edu/websec/csrf/csrf.pdf which apparently used ad networks to run scripts ("submitting forms, requesting images, and issuing XMLHttpRequests") on random users' machines
21:52
<erlehmann>
the internet, infinite possibilities!
21:57
<bga>
heh botnet
22:07
<erlehmann>
dat botnet
22:32
<erlehmann>
test :3
22:49
<bga>
passed
23:27
<dglazkov>
http://dvcs.w3.org/hg/webcomponents/raw-file/tip/samples/widget-theming.html