00:05
<rniwa>
Hixie: what would be a good step to move undo manager spec on either whatwg or w3c repository?
00:06
<rniwa>
good steps*
00:16
<Hixie>
isherman-book: it's on my list of things to get to in due course
00:16
<Hixie>
isherman-book: so if "wait" isn't a good answer for you, i guess "implement, convince other vendors to implement, and ship" is probably it :-)
00:17
<Hixie>
rniwa: you ok maintaining the spec?
00:17
<rniwa>
Hixie: YES
00:17
<rniwa>
Hixie: in fact, i'm going to start implementing it in webkit and write tests for it
00:17
<Hixie>
rniwa: any idea if AryehGregor would be ok with having it under his editing CG?
00:17
<rniwa>
Hixie: but it's easier for me to convince the webkit community if it were "official"
00:18
<rniwa>
Hixie: not sure.
00:22
<rniwa>
AryehGregor: ping
00:23
<isherman-book>
Hixie: Ok, wasn't sure if there was anything I needed to do other than wait
00:24
<Hixie>
isherman-book: i'm unfortunately under a lot of backlog
00:25
<Hixie>
isherman-book: trying to clear that out asap
00:25
<isherman-book>
Hixie: no problem :)
00:26
<Hixie>
oh it's a problem :-)
04:10
<tantek>
hey what's this? http://www.w3.org/community/html5spec/
04:11
<tantek>
as recently announced: https://twitter.com/#!/w3c/status/169266573192863744
05:51
<MikeSmith>
http://dev.webinos.org/specifications/draft/vehicle.html
05:53
<MikeSmith>
http://webinos.org/blog/2012/02/08/webinos-demo-series-1-vehicle-api/
05:53
<jamesr__>
how old is this?
05:54
<jamesr__>
somebody should tell them about constructors for events
05:57
<MikeSmith>
jamesr: from November it seems
05:58
<MikeSmith>
don't recall having seen that doc discussed anywhere
05:58
<MikeSmith>
though might have been in Device APIs WG
05:59
<MikeSmith>
this is from some kind of EU/EC project I guess
06:14
<MikeSmith>
oh man
06:14
<MikeSmith>
http://dev.webinos.org/specifications/draft/events.html
10:27
<remysharp>
Is it defined anywhere that elements with IDs should become global variables on the window with the same name?
10:27
<remysharp>
I know browsers still do it, but I wasn't sure if it was part of any spec or if it should never be relied on (I generally keep away from relying on it)
10:31
<zcorpan>
http://www.whatwg.org/specs/web-apps/current-work/multipage/browsers.html#named-access-on-the-window-object
10:32
<roc>
Gecko doesn't do it in standards mode
10:35
<zcorpan>
roc: what mode? you mean no-quirks mode? you're not making sense! (https://www.w3.org/Bugs/Public/show_bug.cgi?id=15950 )
10:36
<roc>
it's late, can you give me some kind of sarcasm emoticon?
10:36
<zcorpan>
sorry, </sarcasm>
10:37
<roc>
thanks, just checking
10:37
<remysharp>
zcorpan: okay, so technically it's totally legit - I had assumed for a long time that it was just a dirty thing browser (read: IE) did
10:37
<roc>
it is a dirty thing
10:37
<remysharp>
no doubt about that
10:37
<roc>
zcorpan: in that bug, you should accuse Hixie of ignoring reality
10:38
<remysharp>
confuses the hell out of folk I teach as to why their code works, yet the variable wasn't declared
10:38
<zcorpan>
roc: feel free to comment on the bug
10:38
<remysharp>
then they go asking me why did I declare the variable if it's already there...
10:38
<roc>
I'm fighting enough battles at the moment, thanks
10:38
<zcorpan>
remysharp: it's not totally legit since browsers are considering limiting it to quirks mode
10:38
<roc>
remysharp: "so it'll work in Firefox"
10:39
<remysharp>
zcorpan: good stuff. that makes me feel like it's a little more sane
10:43
<jgraham>
remysharp: Oh, well I'm sure you will be dissuaded of that notion soon
10:44
<remysharp>
jgraham: aye, I never expect to feel sane about how browsers do things for longer than a couple of hours. Otherwise I'd just be mismanaging my expectations :)
12:15
<Taggnostr>
jgraham, thanks for the code! I haven't tried it yet, but I'll try to find some time to take a look
12:16
<jgraham>
Taggnostr: np. We should probably ship something like that with html5lib
12:16
<Taggnostr>
the award for the broken markup of the day goes to: <img width=902 height=250px src="/sites/default/files/images/homepage/ISL_cordless_902X250.jpg" /*qui manca alt--capire*/ />
12:17
<Taggnostr>
that "comment" means "/*here alt is missing--understand*/"
12:18
<Taggnostr>
probably they asked someone to validate the page and he didn't even know how to write an html comment and wasn't even smart enough to figure out what the error that the validator gives means
12:18
<Taggnostr>
so he left a js comment in the middle of the tag as a remainder
12:19
<bga>
where is get lastest chromium? anyone from google plz answer me
12:19
<bga>
http://build.chromium.org/f/chromium/snapshots/Win_Webkit_Latest/
12:19
<bga>
is bit outdated
12:32
<myakura>
bga: http://download-chromium.appspot.com/ will help you.
12:32
<myakura>
(I'm not a Googler though.)
12:34
<bga>
myakura thanks
12:40
<annevk>
http://lists.w3.org/Archives/Public/public-html/2012Feb/0083.html o_O
12:40
<annevk>
if that passes
12:40
<annevk>
lol
13:39
<bga>
gsnedders are you here?
15:28
<annevk>
man why are [ and ] not allowed in the frigging query component in URLs?
15:28
<annevk>
that's nonsense
15:29
<ksweeney>
or the domain? I would love http://[][][][][][][].com
15:29
<ksweeney>
:D
15:29
<annevk>
the difference is that does not even work, whereas putting them elsewhere works perfectly fine
15:44
<annevk>
abarth: you around?
15:44
<annevk>
abarth: should CORS require Origin:null or Origin:<chain of origins>?
15:44
<annevk>
abarth: would have been better if the Origin RFC decided on this :(
15:45
<annevk>
abarth: if you're not around I'll prolly check in "null"
15:45
<annevk>
Velmont: ^^
16:17
<annevk>
haven't heard anything from ALA in years, suddenly two articles
16:23
<annevk>
whoa Zurich is now more expensive than Oslo or Tokyo
16:23
<annevk>
http://www.economist.com/blogs/graphicdetail/2012/02/daily-chart-7
17:59
<[tm]>
annevk: +1 to the Recovering from XML Community Support Group
18:12
<[tm]>
"the main reason is that from the beginning we (the XML crowd) have been arrogant, over confident and have made a significant design error."
18:14
<othermaciej>
where's that quote from?
18:16
<StoneCypher>
othermaciej: http://eric.van-der-vlist.com/blog/
18:26
<[tm]>
"we have been proven wrong and that HTML has been enough to scale up to the amazing applications we see today."
19:41
<annevk>
http://blog.ioactive.com/2012/02/ssl-traffic-analysis-on-google-maps.html is pretty interesting
19:41
<annevk>
would SPDY make that more complicated because it all becomes a single stream?
19:52
<Hixie>
annevk: only if you were actually using more than just google maps at a time, as far as i can tell
19:52
<Hixie>
annevk: if i understand correctly, the only real solution is to pad all requests and responses and make them on a strict schedule
19:53
<Hixie>
annevk: hard core encrypted channels just send continual noise at the max bandwidth and when they want to send data they replace the noise with the data
19:53
<Hixie>
annevk: assuming a good encryption algorithm, my understanding is that you then can't tell what's being sent
19:53
<Hixie>
annevk: of course you can still tell a lot e.g. that there is a connection, so that both sides exist, which is an issue e.g. in war situations
19:54
<Hixie>
abarth would know more
20:06
<annevk>
sounds pretty fascinating
20:13
<annevk>
I wonder when w3.org will go TLS-only
20:23
<abarth>
traffic analysis is really hard to defeat
20:23
<abarth>
at least without cover traffic
20:23
<Hixie>
isn't it essentially impossible without cover traffic?
20:36
<annevk>
Microsoft research paper writes "XmlHttpRequest" o_O
20:37
<abarth>
Hixie: i try not to say "impossible" :)
20:37
<Hixie>
fair enough
20:37
<abarth>
i think Tor tries to defeat traffic analysis without cover traffic
20:37
<abarth>
but it's unclear how successful they are
20:38
<Hixie>
don't they use other people's traffic as cover traffic?
20:38
<abarth>
aren't you the one who says "the impossible just takes longer?"
20:38
<Hixie>
i mean it's not as good as a continuous stream, but...
20:38
<abarth>
Hixie: yes, they mix traffic from multiple sources, but they have problems because that traffic can be removed statistically
20:38
<abarth>
its more like noise
20:39
<Hixie>
wow, interesting
20:39
<abarth>
so, then you get into shaping the traffic by adding queues and delays
20:39
<abarth>
etc
20:39
<abarth>
there's a whole literature about it
20:39
<Hixie>
i'm sure
20:41
<annevk>
is it even possible to design a system that does not have these flaws?
20:41
<annevk>
can you measure anything other than amount of data?
20:43
<Hixie>
timing
20:43
<annevk>
oh, how fast the connection is? how often requests are made, etc?
20:44
<Hixie>
for example, if a site sends a ping every 2s in one mode and a ping every 2.5s in another mode, even if they're the same size, you can tell which mode the app is in
20:44
<Hixie>
just by looking at when the packets are sent
20:44
<annevk>
I guess that can tell you quite a bit
20:44
<Hixie>
the general topic here is called "side channel attacks".
20:44
<Hixie>
q.v. http://en.wikipedia.org/wiki/Side_channel_attack
20:45
<Hixie>
power consumption is another one that that page mentions
20:45
<annevk>
that is, size plus timing seem like vectors that will turn out to be pretty unique
20:45
<Hixie>
e.g. you can tell what's going on based on how much power you are pulling -- if it's just idling, it'll use less power than if it's doing heavy computation
20:46
<Hixie>
also in the case of TLS communication on the web there's things like DNS requests, the IP address of the communication's destination, etc
20:47
<Hixie>
e.g. you can tell if someone is logging in on big sites because they get redirected through a separate login service
20:48
<Hixie>
even if the data is mixed in with noise like i was suggesting earlier, you can also learn things by watching what happens if you flip random bits in the stream
20:48
<Hixie>
since the receiving end might just discard the noise rather than verify that it wasn't tampered with
20:48
<Hixie>
(encryption typically avoids this, q.v. integrity)
20:49
<Hixie>
encryption systems, i should say
20:49
<Hixie>
encryption itself doesn't
22:06
<gsnedders>
bga: I am here now, if at all relevant.
22:09
<_bga>
gsnedders i tried to hack js. Add Proxy object that throw error is property is undefined instead silent null value before Object but failed because engines isnt ready to such tricks
22:10
<_bga>
chrome just stop working, ff fall to recursion
22:11
<gsnedders>
bga: My brain can't parse what you just said.
22:14
<_bga>
i changed fundamental behavour. in js {}.foo == undefined but i do Object.prototype.__proto__ = some proxy object. so now {}.foo throw 'Getting unexisting property'
22:14
<_bga>
understand?
22:14
<_bga>
i try to make js safer
22:17
<gsnedders>
_bga: Ah.
22:17
<gsnedders>
_bga: Object.prototype.__proto__ is not like __proto__ on anything else in SpiderMonkey, IIRC.
22:18
<gsnedders>
(Though quite what its semantics is I am unsure.)
22:20
<gsnedders>
Not really surprising that trying to mutate the Object prototype object's prototype doesn't work.
22:20
<_bga>
Object.prototype.__proto__ = {a: 1, __proto__: null}; ({}).a == 1 work in v8 too
22:22
<_bga>
but main idea that new apis open such abilities so engines will be ready
22:25
<_bga>
iirc opera's String#concat work incorrectly w/ deleted toString/valueOf
22:26
<gsnedders>
Shouldn't as of 11.60
22:30
<isherman>
tantek: You previously expressed some technical concerns about http://wiki.whatwg.org/wiki/Autocomplete_Types on Google+. Wanted to check, have your concerns been addressed by renaming the tokens to match hCard where possible?
22:30
<_bga>
gsnedders btw what your opinion? is js should move to full dynamic (lisp way) or static ?
22:31
<_bga>
i guess - js was born as dynamic and should move to better dynamic
22:31
<gsnedders>
_bga: Dynamic what?
22:32
<gsnedders>
_bga: Dynamic typing?
22:33
<_bga>
runtime access to local hash table of vars (prototyped chained), object has many prototypes, instance eval
22:33
<_bga>
runtime access to stack
22:34
<gsnedders>
_bga: Okay, in the sense of it being a dynamic language.
22:34
<_bga>
make stack[stack.length - 1].locals.a = 1 possible
22:34
<gsnedders>
_bga: On the whole I like the balance it has now.
22:35
<gsnedders>
Proxies I'm still on the fence about whether I think they're a good idea, because they have the same semantic issues as host objects can (e.g., [[HasProperty]] and [[GetProperty]] may not match up).
22:39
<_bga>
i can not balance between dynamic and static world. one my ego like full dynamic for experiments, other my ego want extreme static for real use
22:41
<gsnedders>
_bga: Runtime access to stack used to be possible in SpiderMonkey (through __scope__, I think it was called) and arguments.caller (so arguments.caller.__scope__, for example).
22:45
<_bga>
hm
22:46
<_bga>
bad that is dont work anymore
22:48
<gsnedders>
_bga: Stuff like that is hell for performance given dynamic typing, though
22:48
<gsnedders>
_bga: You get any TC for concat with deleted toString/valueOf?
22:50
<_bga>
now work corectly
22:52
<_bga>
gsnedders so if you want perfomance (or community/market force you) => way of more static stuff
22:53
<gsnedders>
_bga: We can't really make it less dynamic, though, more just optionally static