00:12
<karlcow>
http://twitter.com/ndw/status/25527799850
00:13
<karlcow>
>I have joined the HTML Working Group. May the fates have mercy on my soul. -- Norman Walsh
00:30
<Dashiva>
Souls are deprecated
02:24
<annevk>
karlcow, looking at some of his other messages; more drama?
02:25
<annevk>
joining when everything is pretty much done is a bit late though...
02:26
<annevk>
but on the other hand we can never have enough feedback
02:26
<annevk>
the more feedback we get the more viewpoints we can consider
02:47
<Hixie>
what DOM should <option><button><option> give?
02:48
<othermaciej>
what do browsers do with it?
02:49
<Hixie>
historically they treated <option> as a void element or ignored it altogether
02:49
<Hixie>
so they're not especially helpful
02:50
<Hixie>
<option><div><option> creates the same DOM as <option><div></option><option>, which is the same as <option><div></div></option><option>
02:50
<Hixie>
but <button> has weird scoping rules
02:51
<Hixie>
i guess <option><details><option> and <option><h1><option> has the same problem as <button> actually
02:53
<Hixie>
maybe i just need to make </option> close more forcefully
02:57
<Hixie>
ok two options that i can see:
02:57
<Hixie>
1. make </option> as powerful as </div>
02:58
<Hixie>
2. make <option> only imply </option> if the current node is an <option> element
02:58
<Hixie>
option 2 seems safest
03:38
<Hixie>
wtf is webkit doing with form.enctype
03:38
<Hixie>
<form enctype="multipart/formdata"> returns "multipart/form-data"
03:38
<Hixie>
as does:
03:38
<Hixie>
enctype="multipart/anything"
03:39
<Hixie>
enctype="anything/form-data"
03:39
<Hixie>
enctype="form-data"
03:39
<Hixie>
enctype="multipart"
03:40
<Hixie>
enctype="form-data/multipart"
03:40
<Hixie>
enctype="not-form-data-at-all"
03:40
<Hixie>
basically it seems to match anything that contains either "multipart" or "form-data"
09:33
<annevk>
lots of closed bugs
09:35
<annevk>
ooh, December 8
09:35
<annevk>
for some reason I thought the "all addressed" date was October 1
09:44
<annevk>
Hixie, should HTML5 register application/x-www-form-urlencoded ?
11:28
<annevk>
http://ie.microsoft.com/testdrive/HTML5/DOMParserXMLSerializer/Default.html -- because responseXML is stupid?
11:28
<annevk>
and since when are they in HTML5?
11:29
<annevk>
oh, border-radius is part of HTML5 too, I see
11:30
<annevk>
So HTML5 now means "Web" with emphasis on new features?
12:02
<jgraham>
Hixie: Re: Bug 10427 (object scoping), we have a couple of different reports of broken sites due to that behaviour. I will dig up the URLs at some point (maybe tomorrow)
12:51
<karlcow>
annevk, maybe not more drama. I do not think ndw is a drama person. But maybe frustration at a point for him. We will see.
12:52
<karlcow>
plus I'm not sure there is a list of all discussions which involves XML and what are the specific issues he would like to see addressed.
12:53
<karlcow>
Maybe that would be a start, an open list of issues and I'm pretty sure for many of them, there was already a discussion about it in the mailing lists.
12:57
<AryehGregor>
<annevk> So HTML5 now means "Web" with emphasis on new features? <-- Dude, that's what it's meant for years now.
12:58
<Dashiva>
Except when it means uppity kids destroying the web
12:58
<jgraham>
You're doing well if you get "emphasis on new features"
16:35
<Hixie>
anne, if you read this, HTML doesn't really defien application/x-www-form-urlencoded, it just says how to encode it
16:37
<karlcow>
http://www.opera.com/portal/unw/ I wonder
16:37
<karlcow>
"The red-hot world of technology gets a breath of fresh, cool air in Oslo on October 14, 2010."
16:40
<karlcow>
and the source is not more explicit either <meta name="description" content="Information about Opera’s history, vision, executive team, and more" />
17:47
<Hixie>
the more i hack the foreign lands the more i think i need to just get rid of the secondary mode and just use the "reset insertion mode" algorithm
17:52
<Hixie>
anyone remember ms2ger's e-mail in bugzilla?
17:52
<Hixie>
MikeSmith: why does the autocomplete for e-mails not work in the w3c's bugzilla?
17:53
<Hixie>
ah, @gmail
18:12
<annevk>
so if HTML5 does not define it... who does?
18:22
<Hixie>
define what?
18:22
<Hixie>
oh the form submission type
18:22
<annevk>
yeah
18:22
<annevk>
application/x-www-form-urlencoded
18:22
<Hixie>
well in theory bjorn's draft
18:23
<Hixie>
i'm not sure what really needs defining
18:23
<annevk>
well... he's only doing application/www-form-urlencoded
18:23
<Hixie>
i guess we could define how to interpret it, but people seem to be doing fine without it
18:23
<Hixie>
what's application/www-form-urlencoded
18:23
<annevk>
it's a "cleaned up version"
18:23
<Hixie>
implemented by nobody?
18:23
<annevk>
some idea from 2006 he just resurrected
18:24
<Hixie>
bjorn is weird
18:24
<annevk>
nope, but it has some attractive features, like omitting empty parameters, and using ; rather than &
18:24
<annevk>
but I don't think it's worth it
18:24
<Hixie>
like a cross between julian and ms2ger -- all the pedantic quality of ms2ger, with all of the practicality of julian
18:24
<annevk>
so yeah
18:24
<annevk>
hahaha
18:25
<Hixie>
ok well i guess we can define the x- type
18:25
<annevk>
so yeah, if we define how to write and how to parse the media type is done
18:25
<Hixie>
can you file a bug with bullet points of what i need to spec?
18:26
<Hixie>
or send me a mail or something
18:26
<Hixie>
with the highlights
18:26
<Hixie>
i can turn it into prose
18:26
<annevk>
it ties in with abarth <a>.queryParameters proposal too
18:26
<annevk>
i guess
18:26
<annevk>
ok
18:27
<Hixie>
oooh
18:27
<Hixie>
yeah
18:27
<Hixie>
adam's thing
18:27
<Hixie>
interesting
18:27
<Hixie>
we should pull adam into this
18:27
<Hixie>
that's what we should do
18:27
<Hixie>
have adam's URL spec define this
18:28
<Hixie>
annevk: forget filing a bug or whatnot
18:28
<Hixie>
annevk: let's just give this to adam
18:28
<Hixie>
annevk: as you say he's already speccing the parsing
18:28
<Hixie>
s/say/say,/
18:37
<AryehGregor>
Has anyone given any thought on how browsers could stop the evercookie PNG cache trick?
18:38
<annevk>
Hixie, yeah, makes sense that it would define that type
18:39
<annevk>
AryehGregor, can't that be done with any kind of file and XHR?
18:39
<AryehGregor>
Yes, the PNG part is immaterial.
18:39
<AryehGregor>
It's a cache poisoning attack, sort of.
18:39
<annevk>
(solution: don't cache)
18:39
<AryehGregor>
What, don't cache anything?
18:39
<annevk>
i don't think there's much else you can do
18:39
<AryehGregor>
That's my initial reaction too.
18:45
<karlcow>
(other not practical solution: don't use javascript)
18:47
<jcranmer>
karlcow: howis that not practical? ;-)
18:48
<karlcow>
hehe.
18:49
<annevk>
ooh, Dexter is back
18:49
<karlcow>
the last 6 months or one year I have tried different strategies with regards to cookies to see how the browsers UIs make it difficult to really have a control on them.
18:50
<annevk>
Hixie, hsivonen was suggesting foreign lands could just be a namespace check?
18:50
<karlcow>
There is a lot of things which could be done on the side of PIM with regards to browsers
18:51
<annevk>
Hixie, also, you wrote "<code title="">annotation-xml</code> in the SVG namespace"
18:51
<Hixie>
annevk: yeah, dunno how that would work though
18:51
<Hixie>
oops
18:51
<annevk>
Hixie, I think that should be MathML namespace and needs a <span> too probably
18:51
<Hixie>
yeah
18:52
<annevk>
AryehGregor, regarding Web == HTML5; you know, it never really got to me that much
18:53
<annevk>
AryehGregor, here in my ivory tower; most web developer friends also get the distinction, but when I was at the Adobe thing the other day...
19:23
<Hixie>
man what is it with people describing solutions before problems
19:25
<aho>
too much jeopardy
20:00
<annevk>
the type=date impl in WebKit is weird
20:02
<annevk>
or in Chrome, that is
20:02
<AryehGregor>
It's crazy.
20:03
<AryehGregor>
It implements it like a number-picker.
20:03
<AryehGregor>
The person developing HTML5 forms support for Chrome appears to believe that it's okay to dump semi-functional behavior into releases.
20:04
<AryehGregor>
First no validation UI, now this. Blech.
20:13
<itissid>
hey I wanted to ask about this sandbox feature i read on the blog... What does an iframe gave to do with insecure content? Culd it nt be just anywhere other than the iframe? Unless i am missing something fundamental here...
20:20
<AryehGregor>
itissid, if your insecure content isn't in an iframe, that's what srcdoc is meant to handle.
20:42
<itissid>
Here is my story, its a bit long,pls bear... I am making a chrom extension that saves pages as HTML strings and displays them later... The only way to display after safely is using dataURL scheme and then do a location.replace(). The problem with this is it wipes out all the scripts of the previous page..
20:42
<itissid>
Although this protects me from some XSS rather than I use the much more vulneralble document.write()
20:43
<itissid>
I cant run those scripts if i use the dataURI scheme..
20:45
<itissid>
So Some one suggested i use sandboxing
20:46
<itissid>
Now what i want to understand is that if i have a an HTML string can i limit its cross-origin policy using sandbox...
20:48
<itissid>
Cause i want to use document.write() as it preserves the scripts in the head tag
20:50
<annevk>
AryehGregor, it appears validation is now simply not enforced?
20:50
<AryehGregor>
annevk, yes, thankfully.
20:50
<AryehGregor>
Until they can get some real UI for it.
20:51
<AryehGregor>
Have you seen Firefox's UI? It's in nightlies, and it looks pretty slick.
20:51
<annevk>
AryehGregor, it sucks a bit that it comes in pieces
20:51
<annevk>
the outline? looks nice, yeah
20:51
<annevk>
I wonder what they'll do for <input type=date>
20:51
<annevk>
but I guess that's not going to make Firefox 4
20:52
<AryehGregor>
What comes in pieces?
20:52
<AryehGregor>
I don't think the outline is good for stuff that the user hasn't touched yet.
20:52
<AryehGregor>
But the UI generally looks good. Unlike Opera's, no offense.
20:53
<annevk>
HTML5 forms support comes in pieces
20:53
<AryehGregor>
Well, as long as it falls back gracefully, it seems fine.
20:53
<annevk>
it gets more and more painful to decide whether to use a library for a control or not at this point
20:54
<AryehGregor>
I don't think it's such a big deal. If the built-in stuff isn't good enough, use the library. Better yet, use a library that uses whatever native stuff is available.
20:55
<AryehGregor>
Firefox's UI for manually-set errors is good so far, a library could use that to implement date-pickers and such seamlessly.
21:01
<AryehGregor>
Hixie, there is no "diff given below" at http://www.w3.org/Bugs/Public/show_bug.cgi?id=9828
21:02
<Hixie>
yeah my script fails to submit the link whenever the page has non-UTF-8 text on it
21:02
<Hixie>
not sure how to fix it
21:02
<AryehGregor>
What would cause such a crazy bug?
21:04
<annevk>
AryehGregor, it's not a big deal, it just means that detection is getting more complicated over time
21:04
<AryehGregor>
annevk, seems inevitable.
21:08
<AryehGregor>
Hixie, An error occured while submitting your comment. Please let ian⊙hc know.
21:08
<itissid>
AryehGregor: Hey, I may seem like a noob but does the following seem reasonable. What i want to do is to maintain the javascripts in the original browsing context. Now the problem is that location.replace('NEW HTML PAGE IN STRING') would destroy that. So If I created an Iframe element and put my 'NEW HTML PAGE IN STRING' in that , along with a sandbox attrribute, it would protect me from XSS...
21:08
<itissid>
...attacks?
21:08
<Hixie>
AryehGregor: yeah bugzilla is screwed right now
21:08
<AryehGregor>
itissid, I don't really know anything about those APIs, sorry.
21:08
<itissid>
Anyone else?
21:08
<Hixie>
AryehGregor: i can pass you the script (perl) if you want to figure it out :-)
21:09
<AryehGregor>
Hixie, no thanks, I don't do Perl.
21:09
<Hixie>
:-0
21:09
<Hixie>
:-) even
21:10
<annevk>
itissid, why not iframe.srcdoc = htmlstring ?
21:10
<annevk>
(and also set sandbox, of course)
21:11
<itissid>
annevk: So you suggest I add that small iframe element to the HTML of my original browsing context?
21:12
<itissid>
I am making a chrome extension so i dont need to worry about other browsers..
21:12
<Ms2ger>
Hixie, in http://html5.org/tools/web-apps-tracker?from=5508&to=5509, annotation-xml in the *SVG* namespace?
21:14
<annevk>
itissid, wherever you want to display this HTML I suppose
21:14
<annevk>
itissid, not sure if srcdoc is supported by Chrome
21:15
<annevk>
Ms2ger, already reported
21:15
<Ms2ger>
annevk, thanks
21:16
<AryehGregor>
No one has objected to decentralized-extensibility, or is the survey software broken or something?
21:17
<Hixie>
can someone translate http://www.w3.org/Bugs/Public/show_bug.cgi?id=9817 for me? I have no idea what the problem is.
21:17
<Hixie>
Ms2ger: yeah anne pointed it out
21:17
<Hixie>
thanks
21:17
<annevk>
AryehGregor, has someone objected to centralized?
21:17
<annevk>
I'm planning on objecting, unless I don't have to
21:17
<annevk>
well, objecting
21:18
<annevk>
saying it's highly unlikely we'll implement it
21:18
<annevk>
but maybe that wasn't needed with this proposal; I should read it again
21:20
<AryehGregor>
The only non-zero-edit proposal that got to vote is sane enough that I don't think it's totally out of the question that it would be accepted if there weren't a lot of strong objections.
21:21
Ms2ger
would rather not be crossed with julian
21:21
<Hixie>
hah
21:21
<Hixie>
so nobody can explain http://www.w3.org/Bugs/Public/show_bug.cgi?id=9817 for me?
21:22
<annevk>
Hixie, I looked at it (sparing Leif's comments)
21:22
<annevk>
Hixie, I'm not quite sure either
21:22
<annevk>
Hixie, if this is about tabindex I do not get it and something else does not appear in the specification
21:25
<Hixie>
ok i'll mark it NEEDSINFO again
21:25
<Hixie>
i wish i didn't have as much trouble understanding bugs filed by a11y people
21:26
<Ms2ger>
Don't we all?
21:28
<AryehGregor>
Presumably they also wish they didn't have as much trouble being understood by us.
21:30
<Ms2ger>
I try to avoid presuming things about what the a11y people want
21:31
<AryehGregor>
I don't think it's much of a presumption that they want Hixie to change the spec, and that they've figured out that sometimes he doesn't get what they're asking.
21:31
<Hixie>
from what i hear they think i'm lying when i say i don't understand
21:31
<Hixie>
so...
21:31
<Hixie>
anyway
21:32
<Craig`>
hey guys, haven't really done much socket programming, is it possible to create an irc client using websockets?
21:32
<Ms2ger>
Also, self-closing tags in html!
21:32
<Ms2ger>
Craig`, no
21:32
<Ms2ger>
See the note in the spec that says exactly that
21:32
<Craig`>
actually thinking about it i think i remember reading that note
21:32
<Philip`>
Hixie: Maybe ask for suggested replacement spec text? Perhaps that'd make their meaning more explicit
21:33
<Craig`>
actually, probably, i should read the html5 spec then move on from there, leaving websockets 'til last
21:33
<Hixie>
Philip`: that usually just confuses me more, because i end up with text that is clearly wrong, but still no idea why they want it
21:33
<Philip`>
Oh, it already says "<summary> should not support tabindex" which seems an explicit suggestion
21:39
<annevk>
twitter has so much going for it
21:40
<annevk>
take this: http://twitter.com/johnfoliot/status/25607580283 followed by http://twitter.com/terrahawkes/status/25608513143 both in my retweet stream thanks to @mattur :)
21:40
<itissid>
annevk: Ok srcdoc support does not exist for chrome...
21:41
<itissid>
Even something like this src="data:text/html-sandboxed,%3Chtml%3E%3Ch1%3EHello%20World%3C/h1%3E%3C/html%3E" wont work..
21:42
<annevk>
itissid, at this point it's prolly best to ask the Chrome guys
21:43
<itissid>
Hmm
21:50
<itissid>
annvek: Just so i know i am testing this correctly I made a small iframe with a sandbox attribute and a link to google... I put this code http://jsbin.com/olale3 in a local file and fired it up on my browser... If sandboxing works that should not allow me to navigate to google right?
21:52
<annevk>
why not?
21:52
<Hixie>
annevk: hahahaha that's an awesome reply
21:52
<Hixie>
spot on
21:54
<itissid>
as in there is a link to google in that HTML text in the iframe... The idea is that I cant make cross-origin requests..
21:55
<annevk>
itissid, that's now how it works