03:07
Aryeh-laptop
experiments with Empathy
03:29
<karlcow>
http://developer.yahoo.com/blogs/ydn/posts/2010/10/how-many-users-have-javascript-disabled/
03:33
<wirepair>
woa interesting stats.
08:47
<hsivonen>
http://www.w3.org/TR/REC-xml/#AVNormalize is still the XML spec section I need to quote the most often :-(
08:53
<zcorpan>
i wonder what happens with the polyglot spec when it's going to be published again since it now uses the html5 doctype
08:53
<Hixie>
shouldn't it be a polyglot document?
08:54
<zcorpan>
yes, that's why it uses the html5 doctype
08:54
<Hixie>
ah, interesting
08:54
<Hixie>
well it can always use the XHTML1.0 one when the pubteam complain
08:55
<zcorpan>
then it's not polyglot anymore according to the polyglot spec iirc
08:55
<Hixie>
ah ok
08:55
<Hixie>
well then
08:55
<Hixie>
this ought to be fun
08:56
<Hixie>
as the pubteam's immovable wall of policy meets the polyglot's unstoppable force, as they say
08:58
<zcorpan>
oh, it seems it allows the xhtml 1.0 doctype
08:59
<zcorpan>
actually the normative part allows html4 doctype without url
08:59
<zcorpan>
that seems like a bug
08:59
<zcorpan>
http://dev.w3.org/html5/html-xhtml-author-guide/html-xhtml-authoring-guide.html#doctype
09:01
<zcorpan>
anyway, i don't care about that document
09:12
<hsivonen>
zcorpan: wait! is there now a normative part in the polyglot guide?
09:13
<zcorpan>
hsivonen: well i guess the document as a whole is non-normative, but it has "must"s
09:13
<hsivonen>
zcorpan: hmmkay
09:14
<zcorpan>
and also "this section is non-normative" and normative references and informative references
09:15
<Hixie>
um... it seems to do a poor job at being non-normative then
09:15
<Hixie>
someone should, like, file a bug
09:15
<hsivonen>
yay for leaky abstactions. when I read the email I just set to public-html (as sent back to me by the list server), I see two instances of ">From" at the start of a line in Zimbra
09:15
<Hixie>
that design bug has plagued the mail world for years
09:15
<zcorpan>
> From is so annoying
09:17
<zcorpan>
at least opera doesn't do that
10:12
<zcorpan>
does space need to be escaped in srcdoc in xml?
10:13
<Hixie>
we had a whole poll about that topic recently
10:13
<zcorpan>
yeah i know
10:14
<zcorpan>
"For a white space character (#x20, #xD, #xA, #x9), append a space character (#x20) to the normalized value."
10:15
<zcorpan>
seems to me that space doesn't need to be escaped
10:15
<zcorpan>
the paragraph about non-CDATA attributes doesn't apply since srcdoc is a CDATA attribute
10:16
<Hixie>
well then i guess we have new information: it turns out it IS too complicated for julian to come up with accurate text after all, and the concerns in my CP were justified.
10:19
<Workshiva>
But surely Julian isn't representative for the average XHTML author ;)
10:22
<zcorpan>
ok so what do i do? just file a new bug?
10:22
<annevk>
so srcdoc really is just a nice syntax for data URLs, right?
10:22
<annevk>
with the media type implied
10:22
<Hixie>
zcorpan: sam's e-mail said something about how to escalate it, i didn't read the details
11:54
<Philip`>
zcorpan: It could be a non-CDATA attribute, if the DTD specifies it to be so
11:57
<Philip`>
(That's very unlikely, but if the goal is to describe how to safely escape attributes then it shouldn't ignore that case, and to avoid being far more complex it should just say to always escape whitspace)
12:03
<kuya>
hello, im using html5lib python version and its showing an error i cant work out
12:04
<kuya>
*an error parsing my html
12:04
<kuya>
could anyone help?
12:06
<jgraham>
kuya: Maybe. Just ask the question :)
12:06
<kuya>
im not quite sure what my question is! ;D
12:06
<kuya>
im seeing 194:11 Unexpected end tag (form). Ignored.
12:07
<kuya>
http://dpaste.com/257740/
12:08
<kuya>
is the relevant bit of html
12:08
<kuya>
the w3c validator doesnt seem to complain on the same code
12:09
<kuya>
(im willing to accept the error might be in a ton of html before that line...)
12:10
<kuya>
actually if i remove that form from the html the error goes away
12:10
<kuya>
so wth is wrong with my form?
12:11
<jgraham>
Possibly it is just the error detection in html5lib gone wrong
12:11
<jgraham>
that is not very well tested
12:11
<jgraham>
I didn't really think anyone was using it :)
12:12
<kuya>
should i try html5lib from svn ... or try and fix html5lib... or just give up :)
12:12
<jgraham>
You could try the latest version from mercurial
12:12
<jgraham>
What are you actually trying to achieve?
12:13
<jgraham>
I mean wy do you care about the parse error?
12:13
<kuya>
im trying to use html5lib as a validator
12:13
<kuya>
is that a dumb idea?
12:13
<jgraham>
Oh, that sounds like a bad idea :)
12:13
<kuya>
ergh
12:13
<jgraham>
I mean it should theoretically tell you about misnested tags
12:14
<jgraham>
But it knows nothing of content models or restrictions on attribute values, for example
12:14
<jgraham>
Maybe you could run a local instance of validator.nu to get the effect you want?
12:14
<zcorpan>
Philip`: if the author uses his own DTD then he doesn't need to read the note
12:15
<kuya>
jgraham: i was hoping to not have to run a whole instance of that... but if its my best bet i guess ill have to
12:16
<jgraham>
kuya: Well it depends on your needs. But adding true validation to html5lib would be highly non-trivial
12:17
<zcorpan>
kuya: http://about.validator.nu/#api
12:17
<kuya>
ok thanks for the info jgraham
12:21
<Philip`>
zcorpan: You're assuming the authors understands all the details of XML, and the purpose of the escaping note was to be for people who don't understand XML (but are using it anyway) and need the HTML5 spec to tell them about it
12:23
<Philip`>
I suppose it just seems odd to give specific advice that is only partially correct
12:23
<zcorpan>
Philip`: i'm assuming that no-one is going to change srcdoc from being CDATA
12:24
<zcorpan>
and if they do then they understand the concequences
12:29
<jgraham>
zcorpan: That doesn't really fit with the "monkies with guns" model of web developers
12:29
<jgraham>
monkeys maybe
13:11
<kuya>
does anyone some js that posts data to validator.nu? for some reason i cant do it
13:13
<zcorpan>
doesn't XMLHttpRequest work?
13:13
<zcorpan>
or a form?
13:14
<kuya>
afaict this should work $.ajax({url:'http://localhost:8888/';, type: 'POST', data: 'some_html', success: function(d) { console.log(d); }})
13:15
<kuya>
but i get nothing back and the request looks totally screwy in firebug
13:18
<zcorpan>
oh, i guess it's because it's a cross-origin request
13:18
<kuya>
it comes back looking like an OPTIONS request (wth that is)
13:19
<zcorpan>
have a look at the xmlhttprequest level 2 spec and the CORS spec
13:20
<zcorpan>
this is because it's cross-origin
13:21
<zcorpan>
if your page is on http://localhost/ then it's cross-origin because the port is different
13:22
<zcorpan>
if you serve your page from http://localhost:8888/ also then it should work without the preflight CORS stuff
13:22
<kuya>
hrm
13:22
<kuya>
cant serve two apps from the same port :S
13:22
<kuya>
maybe i can stick in a little proxy
13:23
<zcorpan>
well, i think validator.nu should support CORS so unless i'm mistaken it should just work
13:25
<kuya>
i dont quite understand what your telling me about CORS
13:30
<zcorpan>
ok. traditionally browsers haven't allowed cross-origin requests in XHR at all
13:30
<zcorpan>
a new spec called CORS changed that by letting the server opt-in to cross-origin XHR
13:31
<zcorpan>
but it does that by doing a pre-flight request for POST before doing teh real request
13:32
<zcorpan>
validator.nu should support this so I don't know why it doesn't work for you, but at least it explains what you see in firebug
13:32
<kuya>
ah ok
13:32
<kuya>
i think i follow
13:32
<kuya>
i certainly see some headers regarding this CORS stuff
13:33
<kuya>
but im no closer to making it actually work hehe
13:33
<zcorpan>
the browser should do the magic for you in theory :)
13:36
<kuya>
i get exactly the same problem trying to do the request to validator.nu so its not to do with my local install
13:50
<kuya>
ah i think i get the problem
14:13
kuya
fail
14:13
<KaOSoFt>
http://www.blambot.com/grammar.shtml
14:13
<kuya>
if anyone can do a cors ajax request to validator.nu please tell me
14:15
<hsivonen>
has the CORS spec changed from the headers that validator.nu sends?
14:26
<zcorpan>
the headers are here http://dev.w3.org/2006/waf/access-control/#syntax
14:32
<hsivonen>
kuya: the spec has changed and v.nu hasn't kept up :-(
14:32
<kuya>
ah ok
14:33
<kuya>
im proxy requests to it now to side step the problem
14:33
<kuya>
im not hacking java
14:33
<zcorpan>
kuya: you could file a bug in bugzilla.validator.nu so hsivonen can fix it at some point
14:33
<zcorpan>
(or someone else)
14:33
<hsivonen>
annevk: just curious, when the header names changed anyway, why didn't you get rid of the "Access-Control-" prefix and use a "CORS-" prefix?
14:34
<annevk>
when did the header names change?
14:34
<annevk>
oh, I see
14:34
<hsivonen>
so now I send Access-Control: allow <*>
14:34
<hsivonen>
is Access-Control-Allow-Origin: * what I need to send instead?
14:34
<annevk>
yes
14:35
<annevk>
the header names changed before I renamed the draft to something more appropriate
14:35
<hsivonen>
annevk: ok
14:35
<annevk>
the header names changed around August 2008, the next publication that called it CORS was in March 2009
14:36
<hsivonen>
I see
14:36
<hsivonen>
so v.nu has been broken for 2 years...
14:36
<annevk>
and nobody noticed...
14:37
<annevk>
I have repeatedly suggested changing the header names, but since Microsoft has some deployment sicking is not really open to it
14:37
<annevk>
Microsoft shipped IE8 around that time, I think that made the whole thing stuck
14:38
<hsivonen>
annevk: I think it's better not to bikeshed names after browsers have shipped
14:39
<hsivonen>
grumble, grumble. looks like I need to use Access-Control-Allow-Methods, too
14:39
<hsivonen>
apparently in addition to Allow
14:40
hsivonen
wonders what the rationale is
14:40
<zcorpan>
maybe you just want to allow GET in the normal case?
14:41
zcorpan
hasn't followed CORS stuff
14:41
<hsivonen>
Hmm. I have access-control-policy-path
14:41
<hsivonen>
that's no longer in the spec, either
14:41
<annevk>
hsivonen, the whole specific opt-in was something from the Mozilla security team
14:41
<hsivonen>
annevk: what should I replace the policy path with?
14:42
<annevk>
there's no replacement
14:42
<annevk>
just need to make sure you allow for each URL
14:43
<hsivonen>
annevk: so I'll just zap that header...
14:43
<annevk>
right
14:50
<hsivonen>
kuya: does it work for you now?
14:51
hsivonen
redeployed v.nu with now CORS headers
14:51
<kuya>
ooh
14:52
<kuya>
thanks hsivonen i will check in a moment
14:55
<hsivonen>
s/now/new/
14:59
<kuya>
i cant seem to construct a request via js that validator.nu will accept
14:59
kuya
just tried hand crafting a multipart request and failed
15:03
<kuya>
hsivonen: would supporting application/x-www-form-urlencoded be very hard?
15:05
<kuya>
oh wow i just did it
15:05
<kuya>
i was being dumb
15:09
<kuya>
hsivonen: CORS now works :)
15:09
<kuya>
hsivonen: im sending the content as the POST body - but adding ?out=json returns nothing
15:10
<kuya>
nope i lieing
15:10
<kuya>
works now with CORS
15:10
<kuya>
woot
18:41
<bluejay2>
is this the chan for html5?
18:41
<KaOSoFt>
Web technologies, mostly, including of course HTML5.
18:43
<bluejay2>
Cool. I've worked on a method to allow a user to download a file encoded in base64 via URL string but of course browsers have a URL size limit, so I want to explore the possibilities of using localstorage for file downloading.
18:43
<bluejay2>
I've read about the file "uploading" via drag and drop, very awesome, but haven't found an example of downloading a file other than plain ascii text
18:56
<TabAtkins>
bluejay2: You'll want the FileSystem API and/or FileWriter API. Neither are implemented publicly yet, though.
18:57
<TabAtkins>
Currently file downloading is usually done by passing the information back to the server and then opening an iframe to the server, where the server can then present the information back with appropriate headers to trigger a download.
19:07
<bluejay2>
TabAtkins: What do you think the ETA is for filewriter support in browsers?
19:20
<TabAtkins>
bluejay2: No clue, but since it's Chrome people writing the spec, it'll probably show up in Chrome/Webkit first.
19:32
<Ms2ger>
"I am usually the one to argue for orthogonality"
19:32
<Ms2ger>
Who can guess the author?
19:38
<jgraham>
trick question?
19:38
<Ms2ger>
Nope
19:43
<svl>
Shelby!
19:56
<Ms2ger>
svl, you win an axiomatic proof!
19:59
<svl>
\o/
19:59
svl
fears
20:01
<Ms2ger>
Hixie, ping
20:53
<Hixie>
Ms2ger: here
20:54
<Ms2ger>
Hixie, I filed a bug to remove the File API exceptions, as they now use a separate counter
21:02
<Hixie>
k
21:40
<heycam>
"polyglot heartbeat" sounds like a terrible condition to have
21:43
<karlcow>
heycam: sounds to me inter-ethnic love :)
22:01
<Ms2ger>
"W3C Invites Implementations of CSS Style Attributes"
22:01
<Ms2ger>
Also, http://twitter.com/sgalineau/status/27374482354
22:02
<annevk>
CSS WG comedy week :)
22:14
<karlcow>
discovering http://www.gnu.org/software/pythonwebkit/
22:15
<karlcow>
"Python-Webkit is a python extension to Webkit to add full, complete access to Webkit's DOM - Document Object Model."
22:15
<karlcow>
the repository seems to be active http://git.savannah.gnu.org/cgit/pythonwebkit.git
22:30
<hober>
heycam: it'd be a good band name