02:11
<benjoffe>
I noticed a broken link in this spec: http://www.w3.org/TR/css3-3d-transforms/#animation
02:11
<benjoffe>
"the values are first converted to a 4x4 matrix, then decomposed using the method described by unmatrix"
02:12
<benjoffe>
links to a 404: http://tog.acm.org/GraphicsGems/gemsii/unmatrix.c
02:23
<bencc>
if a websocket fails due to a proxy problem, will it fail immediately or will the browser wait for a timeout?
05:42
<Hixie>
anyone know if the websocket spec is written assuming high-bit first or low-bit first?
05:42
<Hixie>
i can't find anything in the spec that defines it
05:49
<Hixie>
oh nm, it does say "the most significant bit is the leftmost in the ABNF"
07:11
<boblet>
re: the i element’s definition, why does “an idiomatic phrase from another language” contain “idiomatic”? shouldn’t that be “transliterated” instead? cc: Hixie
08:06
<Hixie>
boblet: it means things like "de facto" as used by english speakers
08:11
<boblet>
Hixie: while the current defn is nicely language-agnostic, it seems to rule out non-idiomatic foreign language transliteration in English, which I suspect is the major “traditional” use of italics
08:11
<boblet>
woah, let me rephrase that :)
08:11
<boblet>
… use of italics for foreign language words in English
08:13
<boblet>
Hixie: I’m also assuming that non-transliterated words (eg Chinese, Hebrew etc in native script) should use <span lang=""> over <i lang="">
08:17
<Hixie>
yes for the second
08:17
<Hixie>
not sure about the first, but if you have examples of it, file a bug, it's worth looking at for sure
08:17
<Hixie>
afk for a bit now
09:52
<zcorpan>
Ms2ger: isn't "URL" in implementations prefixed?
09:52
<Ms2ger>
Not for create* in Gecko
09:54
<zcorpan>
i thought create* were methods on URL
09:54
<zcorpan>
and URL was mozURL in gecko
09:55
<bencc>
when a websocket fails due to a bad proxy server, will it fail immediately or will it timeout?
09:55
<zcorpan>
note i didn't suggest to rename createObjectURL
09:56
<zcorpan>
bencc: i guess it depends on how the proxy is being bad
09:56
<zcorpan>
bencc: e.g. if the client can't establish a connection to the proxy, it should fail immediately
09:57
<Ms2ger>
Eh?
09:58
<zcorpan>
bencc: but if the client can connect to the proxy and send the request, but then the proxy doesn't send a response back to the client for some reason, then the client can't know that it failed so will timeout
09:58
<zcorpan>
Ms2ger: maybe i'm missing something
09:58
<Ms2ger>
So in Gecko we've basically got partial interface Window { [replaceable] readonly attribute URLProp URL; }; interface URLProp { createObjectURL(); };
09:59
<Ms2ger>
I think the actual URL API hasn't landed yet
09:59
<bencc>
zcorpan: is it possible that the proxy will not respond?
09:59
<bencc>
zcorpan: I'm referring to the proxy issues with non ssl websockets
09:59
<bencc>
zcorpan: if the proxy has a normal failure, long-polling will suffer from it as well so I'm not worry about it
09:59
<Ms2ger>
Or are you suggesting to spec it like that instead of the silly static methods we've got now?
10:00
<bencc>
zcorpan: I'm trying to understand how to implement the fallback to long-polling
10:00
<bencc>
zcorpan: socket.io saves the fact that ws failed in a cookie so they won't try it again
10:00
<bencc>
zcorpan: but if it fails immediately I'm not going to worry about it at all
10:01
<zcorpan>
Ms2ger: so gecko shipped window.URL but made it replaceable. i see. i thought it was called mozURL
10:01
<bencc>
zcorpan: I've tried with nginx which doesn't support ws and it fails immediately but I'm not sure if that's how all proxies will behave
10:02
<Ms2ger>
No, the parts that are in the URL spec are prefixed, but I don't think it landed yet
10:07
<zcorpan>
Ms2ger: http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1311 - window.URL is defined in my gecko nightly
10:08
<Ms2ger>
Yes
10:08
<Ms2ger>
And it's an instance of MozURLProperty or whatever we called it
10:08
<zcorpan>
sure
10:09
<zcorpan>
what i was trying to say was that i thought it was prefixed
10:09
<zcorpan>
and it wasn't
10:09
<Ms2ger>
Indeed it isn't
10:11
<zcorpan>
although webkit doesn't have URL but has webkitURL
10:12
<zcorpan>
and opera/ie don't have either, right?
10:12
<zcorpan>
i dunno, but my gut suggests that it's still possible to change the name
10:14
<Ms2ger>
Maybe... We shipped it like 8 version ago, though ;)
10:14
<zcorpan>
bencc: i remember a case being reported with proxies not responding with an earlier version of the handshake
10:15
<zcorpan>
bencc: i think -00
10:15
<zcorpan>
bencc: because they wanted complete http requests, and the random bytes wasn't a complete http request, so it wasn't forwarded
10:16
<zcorpan>
bencc: so the server couldn't send a response back
10:16
<bencc>
zcorpan: now that I think about it, you can also have a problem with enterprise firewall which might just drop the packate
10:16
<bencc>
zcorpan: am I right?
10:17
<zcorpan>
bencc: dunno
10:17
<zcorpan>
but it seems reasonable to assume that a proxy might not respond for some reason
14:34
<boblet>
Hixie: here’s the overly detailed bug report re: italics and transliterated foreign words https://www.w3.org/Bugs/Public/show_bug.cgi?id=15786
14:56
<crankharder>
why is this manifest file redirecing my 404/500 redirects to the fallback route?
17:07
<dglazkov>
good morning, Whatwg!
17:29
<AryehGregor>
In WebKit, adding an iframe to the document onload of some other iframe doesn't seem to further delay the load event. :(
18:25
<crankharder>
why is this manifest file redirecing my 404/500 redirects to the fallback route?
18:26
<crankharder>
http://pastie.org/3283796
18:38
<Velmont>
So. Looking at IndexedDB. Firefox accepts arrays as keyPaths. Like: keyPath: ['key', 'a'].
18:38
<Velmont>
I can't really see that in the spec, I've been looking for days.
18:38
<Velmont>
Is it a bug in Gecko, or am I overlooking something?
18:45
<Ms2ger>
"If the optionalParameters argument is specified and has a keyPath property which is not undefined or null, then set keyPath to the value of this property. If keyPath is an Array, then each item in the array is converted to a string. If keyPath is not an Array, it is converted to a string. "
18:47
<Velmont>
Ms2ger: Ah, yes, that's why I started the look in the first place. I was starting to wonder why I was looking for arrays as key paths at all.
18:54
<Velmont>
So then I remember my original question, -- it's not really clear how the keypath as array is used later in the operations. http://dvcs.w3.org/hg/IndexedDB/raw-file/tip/Overview.html#dfn-steps-for-storing-a-record-into-an-object-store says If store does use in-line keys and evaluting store's key path on value does yield a value, then set key to that result. --- which is okay, the key returned might be an array.
19:01
<Velmont>
But following the links to the evaluating steps it's not clear to me that it handles an array.
19:01
<Velmont>
for indexes there is an explicit mention: If index's key path is an Array, then set index key to a newly created empty Array. For each item in index's key path evaluate the item on value. If this does not yield a value don't take any further actions for this index. Otherwise add the value to the end of the index key Array.
19:48
<crankharder>
how long does local storage persist as long as the user doesn't remove it?
19:59
<TabAtkins>
crankharder: As long as cookies do - until the browser decides to evict it.
20:27
<AryehGregor>
Ugh, IE9 doesn't support <img src='data:image/svg+xml,...'>?
20:27
<Ms2ger>
Hah
20:28
<Ms2ger>
IE9 supports what's required in acid2, from what I recall
20:28
<AryehGregor>
Or maybe it just refuses to accept it if it's not base64-encoded or something?
20:29
<AryehGregor>
"A text-representation of the data of the resource, where content is represented as URL-allowed characters. Unsafe characters must be percent-encoded. If sBase64Encoding is set, the data must be Base64 encoded; otherwise, the data is assumed to be represented in ASCII encoding." http://msdn.microsoft.com/en-us/library/ie/cc848897(v=vs.85).aspx
20:48
<AryehGregor>
Okay, why does the blue box not paint on top of the orange one here? <!doctype html><div style="height:100px;width:100px;background:blue;z-index:1"></div><div style="height:100px;width:100px;background:orange;position:relative;bottom:50px"></div>
20:49
<AryehGregor>
Shouldn't z-index: 1 put it on top?
20:49
<AryehGregor>
Putting z-index:-1 on the second div seems to do what I want . . .
20:49
<AryehGregor>
Oh, stacking only applies to positioned boxes.
20:49
<AryehGregor>
Got it.
22:22
<Hixie>
TabAtkins: what's the status on @global?
22:23
<Hixie>
TabAtkins: (the <style scoped> @-rule for making selectors also be allowed to consider elements outside the scope)
22:28
<TabAtkins>
Hixie: Nobody's worked on it, or on addressing scoped styles within CSS at all.
22:28
<Hixie>
k
22:28
<Hixie>
guess i'll just stick a temporary spec in html then
22:29
<TabAtkins>
Ping www-style with it.
22:29
<Hixie>
k
23:13
<Hixie>
looks like the multipage copy not updating is a bug on anne's side, fwiw
23:13
<Hixie>
he'll be back soon so i'm gonna not worry about it
23:41
<Hixie>
TabAtkins: done
23:43
<TabAtkins>
Yup, saw it.
23:46
<dglazkov>
is annevk around?