00:14
<annevk>
does anyone know what i7 stands for in the new Intel product line?
00:14
<annevk>
(press release doesn't say)
00:19
<weinig>
annevk: nice
02:45
<takkaria>
if anyone has an example of a page which ends up needing to be reparsed because of html5's charset algorithm when a meta tag is encountered, I'd very much appreciate a pointer to it
03:03
<Hixie>
takkaria: <script>alert('A');</script><!-- xx megabytes xx --><meta charset="C">
03:03
<Hixie>
takkaria: where A is a byte that is interpreted differently in two encodings
03:03
<Hixie>
and C is the one of those encodings that isn't hte one the browser picks by default
03:03
<takkaria>
ah, I get you. I thought it might be mostly to do with scripting
03:03
<Hixie>
doesn't have to be scripting
03:03
<Hixie>
could be: A<!-- xx megabytes xx --><meta charset="C">
03:04
<Hixie>
it's just more obvious with an alert() because then you get to see that it shows up twice
03:04
<takkaria>
right! got you, thanks
03:04
<Hixie>
np
03:05
<takkaria>
there is a web browser equipped with the HTML5 parsing algorithm now, though it's not got scripting so isn't the most useful browser for testing all the bits that require scripting :)
03:05
<Hixie>
:-)
03:05
<Hixie>
is scripting on the agenda?
03:06
<takkaria>
at some point in the future, I believe, but nothing very soon
03:06
<Hixie>
k
03:06
<takkaria>
otoh, I expect there'll be useful feedback from users of NetSurf when pages break
05:50
<Hixie>
sweet
05:50
<Hixie>
in IE you can insert multiple nodes into the document (adjacent to the root element) using insertAdjacentHTML()
05:53
<Hixie>
and iAH() doesn't work with table bits in IE
06:22
<Hixie>
anyone got an old version of IE? like IE7 or IE6?
06:22
<Hixie>
i'd like http://junkyard.damowmow.com/336 tested with one of those
06:23
<krijnh>
024578A9 (IE6)
06:23
<Hixie>
thanks
06:23
<krijnh>
021453786A9 (IE5.5)
06:23
<Hixie>
wow, cool, thanks
06:24
<krijnh>
Run with IETester..
06:24
<Hixie>
IETester?
06:24
<krijnh>
http://www.my-debugbar.com/wiki/IETester/HomePage
06:24
<Hixie>
neat
06:25
<krijnh>
No idea if it's reliable :)
06:25
<Hixie>
yeah i was about to say, i wonder if that's reliable with javascript: and iframes
06:26
<krijnh>
It's the only way for me to get some different versions of IE running on Vista
07:21
<aboodman2>
Hixie: I keep wishing that support for conversations via MessageChannel was more built-in
07:22
<Hixie>
how do you mean?
07:22
<aboodman2>
like, when you get a message, you always get an object you can reply to the sender with. like it is built into the api.
07:23
<aboodman2>
It's just a little thing, but I was wondering if you could think of some way to make it work.
07:23
<aboodman2>
I keep wanting the sender to be able to do:
07:23
<aboodman2>
var port = worker(or iframe).sendMessage("foo");
07:23
<aboodman2>
and then the port represents the conversation
07:24
<aboodman2>
and on the other side, the worker (or iframe) could do globalObject.addEventListener("message", function(e) { e.port.sendMessage("reply") }, false)
07:24
<Hixie>
a port _is_ a conversation
07:24
<aboodman2>
right, i know
07:24
<Hixie>
if you want to split the conversation off, just send a port with the message
07:25
<Hixie>
and use that port for the subconversation
07:25
<aboodman2>
i know, it seems slightly less elegant for some reason to send the port as an argument
07:25
<Hixie>
i don't understand you example fragments
07:25
<Hixie>
your
07:25
<aboodman2>
one second
07:33
<aboodman2>
k, here is how it works now:
07:33
<aboodman2>
var channel = new MessageChannel();
07:33
<aboodman2>
var worker = createWorker("foo.js");
07:33
<aboodman2>
channel.port1.addEventListener("message", function(e) { ... }, false);
07:33
<aboodman2>
worker.sendMessage("init", channel.port2);
07:33
<aboodman2>
here is how I wish it would work
07:33
<aboodman2>
var worker = createWorker("foo.js");
07:33
<aboodman2>
var port = worker.sendMessage("init");
07:33
<aboodman2>
port.addEventListener("message", function(e) { ... }, false);
07:34
<aboodman2>
the worker code would look the same as it does now. e.port would contain a port that could talk back to the return value from sendMessage()
07:35
<aboodman2>
I do not know how to make this work with your goal of wanting to be able to send arbitrary ports though.
07:35
Hixie
reads (sorry, had a cat petting emergency)
07:37
<Hixie>
note that |port.addEventListener("message", function(e) { ... }, false);| is better written as |port.onmessage = function (e) { ... };|
07:37
<aboodman2>
why?
07:37
<aboodman2>
because otherwise i need to call start()?
07:37
<Hixie>
it's shorter, and doesn't require calling start()
07:37
<Hixie>
but that's orthogonal
07:38
<Hixie>
i agree that it would be neat to have a method that creates a port and includes it
07:38
<Hixie>
and returns a matching port
07:38
<Hixie>
you would likely not want that to happen every time, creating a port is presumably not free
07:39
<Hixie>
btw, it's now worker.port.sendMessage(), since sicking wanted a separate Worker object returned from createWorker()
07:39
<aboodman2>
yeah.
07:40
<Hixie>
we could have port.postMessageWithNewChannel('message') which returns a port
07:40
<Hixie>
though that name sucks
07:41
<aboodman2>
it seems like it would be possible for implementations to know if the port is used and delay the hard work until then
07:42
<Hixie>
true
07:42
<Hixie>
maybe
07:42
<aboodman2>
another thing.
07:42
<aboodman2>
let's say you send port2 to a friend.
07:43
<aboodman2>
and then you start doing things like:
07:43
<aboodman2>
port2.onmessage = function(e) { alert('hi') }
07:43
<aboodman2>
that's totally nonsensical, right?
07:44
<Hixie>
yeah
07:44
<Hixie>
once you send port2, it'll never receive a message again
07:47
<Hixie>
so should i make postMessage() always create a new channel and send a port?
07:47
<Hixie>
seems expensive to me
07:47
<Hixie>
even if theoretically it could be optimised
07:48
<aboodman2>
no
07:48
<aboodman2>
I'm not sure about sicking but a lot of my feedback has been wrong based on not totally understanding what you were trying to do or what the spec proposed.
07:49
<aboodman2>
so i'm just registering the wish that conversations were more elegant because that is a real problem with gears (and postMessage) right now
07:50
<aboodman2>
but i don't have a concrete suggestion
07:52
<Hixie>
maybe port.startConversation('message'); which returns a port and sends a new port
07:52
<aboodman2>
yeah, i think something like that would be cool
08:30
<Hixie>
aboodman2: ok, added startConversation()
08:30
<aboodman2>
don't do that
08:31
<aboodman2>
i was speaking theoretically
08:31
<Hixie>
you said it would be cool
08:31
<Hixie>
i agree
08:32
<Hixie>
man, you people need to stop telling me that things would be good ideas if you don't think they're good ideas
08:32
<Hixie>
anyway, it's in now, but easy to remove if we decide we don't want it
08:32
<aboodman2>
what was the expense you were thinking of to create a port?
08:34
<Hixie>
creating a port and sending it requires creating a MessageChannel and three MessagePorts instead of just creating the two ports already entangled in the right thread spaces
08:35
<aboodman2>
can you get from a port object to a channel?
08:35
<Hixie>
no
08:35
<Hixie>
a channel is basically just an array of two ports
08:36
<aboodman2>
right, so you just need to create the port to return
08:36
<aboodman2>
the port seems like it just wraps a unique integer
08:36
<aboodman2>
and then on the other side you have to create the other port, which can be done lazily
08:36
<Hixie>
right
08:38
<Hixie>
but without startConversation(), the author has to create a new messagechannel, then call postmessage, which creates a third port
08:38
<Hixie>
hence startConversation() is cheaper
08:39
<Hixie>
i wish sicking would send feedback
08:39
<Hixie>
he was all like "we have to have this done in 11 days or it's all over!" last week
08:39
<Hixie>
barely heard three e-mails out him since
08:39
<Hixie>
out of him
08:40
<gavin_>
he was at blackhat last week, fwiw
08:41
<Hixie>
yeah
08:41
<Hixie>
so? :-)
08:41
<Hixie>
i was off sick last week, didn't stop me from responding to all his mails :-)
08:44
<gavin_>
it's harder to be around a computer 24/7 at blackhat than it is when you're off sick :)
08:45
<gavin_>
casino security doesn't appreciate laptops at the blackjack tables
08:45
<Hixie>
i'm just saying'
08:46
<Hixie>
er, sayin'
08:46
<Hixie>
he was volunteering to edit it because it was so urgent
08:46
<Hixie>
presumably because i was a bottleneck
08:46
<gavin_>
yeah- did you emailhima bout it?
08:47
<aboodman2>
Hixie: it's pretty hard to digest these.
08:47
<Hixie>
aboodman2: hm?
08:47
<gavin_>
I think you'd get better results with that than with griping on IRC when he isn't her e:)
08:47
<Hixie>
i'm just ranting, i'm sure he's replying as fast as he can :-)
08:48
<Hixie>
i didn't e-mail him specifically to ask why he hadn't replied, i assumed he had good reasons :-)
08:48
<Hixie>
aboodman2: these what?
08:54
<aboodman2>
Hixie: specs
08:55
<Hixie>
sure
08:55
<Hixie>
i made a bunch of demos which hopefully help explain the worker model a bit
08:55
<aboodman2>
yes, they help a lot.
09:42
<Lachy>
I finally found a solution to allow me to stream olympics from NBCs website, even though I'm not in the USA! :-)
09:43
<hendry>
Lachy: which is? proxy? ?:)
09:44
<Lachy>
I forward all connections using ssh through my web server which is in the USA
09:44
<Lachy>
using: ssh -D 8080 -fN lachy.id.au - and then setting my setting the SOCKS proxy in my browser to localhost 8080
09:45
<Lachy>
http://ubuntu-tutorials.com/2008/06/18/tunnel-web-and-dns-traffic-over-ssh/
09:47
<gavin_>
there are no australian sites that offer internet feeds of the ollympics?
09:47
<gavin_>
perhaps http://www.cbc.ca/olympics/ would work better?
09:48
<gavin_>
oh, I forgot you're in norway
09:50
<Hixie>
well my 1.5TB disk just started makin horrific noises
09:50
<Hixie>
so i guess it's time to get a drobo
09:50
<hendry>
Lachy: i'm IP restricted from BBC news covereage, even though I am in the UK http://www.flickr.com/photos/hendry/2753042810/
09:52
<gavin_>
I guess the broadcasters are forced to implement some sort of geographical restrictions
09:52
<hsivonen>
Hixie: http://www.whatwg.org/specs/web-apps/current-work/source-whatwg is 404
09:52
<hendry>
gavin_: forced by "exclusive deals" and greed
09:53
<Hixie>
hsivonen: yeah, it's only not 404 while i'm running the gen script
09:53
<Hixie>
hsivonen: the markup just had <dfncode> or something along those lines
09:53
<Hixie>
no embedded nulls or anything
09:56
<hsivonen>
Hixie: ok
09:58
<Lachy>
gavin_, yahoo7.com.au/Olympics has streaming, but I need access to server in Australia that I can tunnel through. I would prefer that, cause that only requires flash. I had to install silverlight to watch NBC, and I wish I didn't
09:58
<gavin_>
no norwegian sites?
09:59
hendry
is there is a flickr or youtube of short mp3 clips, so i can embed sound into my blog with (urg) flash (fall back to <audio> would be a plus)
09:59
<Lachy>
not that I know of. But even if they did, they would have norwegian commentary, not english
09:59
<Lachy>
there is also SOPCast, which has live streaming, but that only seems to have a chinese language stream
10:00
<Lachy>
and its windows only, so I can only run it in virtual machine
10:02
<Hixie>
ok, drobo purchased
10:02
<Hixie>
i'm fed up of disks dying all the time, hopefully this will make it a non-issue
10:02
<Lachy>
Hixie, I was thinking about getting myself a drobo too
10:03
<Hixie>
they seem pretty sweet
10:03
<Lachy>
I don't have problem with disks dying, but I am running out of storage space, so I need an economical solution that is cheaper to setup than a RAID
10:07
<Hixie>
can't you just buy a new disk and add it to the usb bus?
10:08
<hsivonen>
my HFS+ filesystem holding my real-world test data and Time Machine backups just died
10:08
hsivonen
wishes HFS+ were more robust
10:08
<hsivonen>
I guess next time I should put the real-world data on a disk image instead of stressing the catalog file of the concrete fs
10:09
<Hixie>
i made a point of not putting anything but my time machine backups on the time machine drive
10:09
<Lachy>
I already have 2 external 1TB drives
10:09
<Hixie>
(that's the one that just started making weird high pitches screeching noises)
10:09
<hsivonen>
Hixie: probably a good idea
10:09
<Hixie>
(i just turned it off without checking if it was ok)
10:10
<Hixie>
(don't want to cause it to fail before i have something to replace it)
10:10
<hsivonen>
my Time Machine disk never fills up, because HFS+ dies first and I have to restart Time Machine history
10:12
<Hixie>
my 1.5TB disk was down to a few dozen GB
10:12
<Hixie>
but unrelated to that i've had hell getting it to work over the network
10:13
<hsivonen>
I have high hopes for zfs
10:13
<Hixie>
my disk failures have always sounded very hardware-related
10:13
<Hixie>
i had a dell laptop once whose motherboard had a bug that was actually destroying hard disks
10:14
<Hixie>
and my mac mini has had two disks die on it (both maxtor)
10:14
<Hixie>
(including this latest one)
10:15
<Hixie>
my plan once i have the drobo is to get spinrite and put it on an old pc i have lying around and see if i can get the dead maxtor 750GB disk and the potentially dead maxtor 1.5TB disk up and working again
10:15
gavin_
can't recall ever suffering from a hard drive failure
10:15
<Hixie>
and if so, use them as an external disk for my girlfriend's timemachine instead of going over the network
10:16
<Hixie>
anyway, bed time first
10:17
<Hixie>
btw anyone know why <script src=""> doesn't support javascript: URLs? (zcorpan sent mail late last year about it)
10:20
<gavin_>
does it not work in gecko?
10:20
gavin_
would expect it o
10:21
<Hixie>
it does nothing
10:22
<Hixie>
http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E...%3Cscript%20src%3D%22javascript%3Aalert(%27test%27)%3B%27test%27%22%3E%3C%2Fscript%3E
10:23
<gavin_>
well, "alert" isn't valid JS
10:24
<gavin_>
that reutns alert.toString()
10:24
<gavin_>
which would justredefine alert
10:24
<gavin_>
...with invalid JS code
10:24
<gavin_>
nevermind
10:30
<webben>
Hixie: Do you have a view on whether, if RDFa in XHTML becomes a Recommendation (currently it's CR), it would be conforming with HTML5 to use its elements and attributes in an XML document also using the new RDFa attributes, which are also in http://www.w3.org/1999/xhtml ( http://www.w3.org/TR/2008/CR-rdfa-syntax-20080620/ ) ?
10:30
<webben>
or are these drafts fundamentally incompatible?
10:31
<Hixie>
sorry i don't quite follow the question
10:32
<webben>
Hixie: I guess it boils down to: if some elements and attributes are placed into the HTML namespace by another W3C recommendation, can they be used in an XML document also using elements and attributes in the HTML namespace which are defined by HTML5?
10:34
<webben>
Would that break document conformance with HTML5, for instance?
10:34
<hsivonen>
Hixie: I'm unable to reproduce the dtcode bug
10:35
<annevk>
webben, it would
10:36
<webben>
annevk: Would it break conformance in a different way than (for example) including some elements in a completely different namespace (I dunno, MusicML or something) would?
10:37
<webben>
annevk: i.e. would it have the same relationship to HTML5 as the XML documents include XHTML and MathML do to XHTML 1.0? http://www.w3.org/TR/xhtml1/#well-formed
10:37
<webben>
*as XML documents that include
10:37
<hsivonen>
webben: as implemented, they both get the same type of error message
10:38
<hsivonen>
interestingly, my "eRDF5" trial balloon got Warnocked, even though a couple of people had tried to stir discussion about RDFao
10:38
<hsivonen>
s/ao/a/
10:39
<hsivonen>
Hixie: oops. I can reproduce the dtcode thing now
10:39
<hsivonen>
will investigate
10:40
<webben>
hsivonen: Thanks.
10:52
<mcarter>
does anyone know what a browser is supposed to do when (in an xhr response) it receives data ending with a partial unicode character -- should it trigger a readystate 3 with only the last complete character?
10:53
<annevk>
it should simply try to decode it
10:53
<annevk>
the partial character will turn into U+FFFD
10:54
<hsivonen>
mcarter: do you mean really ending or the first buffer containing a partial character with potentially more data coming up?
10:54
<mcarter>
hsivonen, actually, I'm curious about both cases
10:55
<annevk>
same for both
10:55
<Hixie>
webben: it has the same conformance status as web forms 2 vs html4
10:55
<Hixie>
webben: using it would be conforming to the spec that makes up those attributes, but not to html5 proper
10:55
<hsivonen>
mcarter: if the stream ends, I'd do what annevk said. If the stream doesn't end, I'd hold onto the partial character expecting more data and behave as if that character hadn't yet been received as far as scripts go
10:56
<webben>
Hixie: That seems a reasonable answer. :) Thanks.
10:56
<Hixie>
webben: it doesn't matter what the namespace is, though
10:56
<Hixie>
webben: from the point of view of conformance
10:56
<mcarter>
annevk, how could the web work at all if it turned it into a U+FFFD character? sometimes unicode characters get split into two seperate packets, and half of the character might come through before the other half
10:56
<webben>
Hixie: okay
10:57
<hsivonen>
mcarter: that is, I'd expect the partial character stay inside the decoder and the decoder not outputting more data until it has at least one complete character to output
10:57
<mcarter>
hsivonen, that makes sense
10:57
<Hixie>
webben: i mean, the spec could say it's the xhtml namespace, or that it's the foobar namespace, html5 doesn't care, html5 just cares about "metadata elements" and "phrasing elements" and so on, the categories
11:00
<hsivonen>
fwiw, I think RDFa could be mostly harmless if it used full URIs instead of CURIEs and, therefore, didn't need to know the namespace mapping context at each node
11:00
<hsivonen>
I'm strongly opposed to importing any CURIE or other qnames-in-content stuff into HTML5
11:01
<Hixie>
RDFa would be mostly harmless in the same way that products in sky mall magazines are mostly harmless
11:02
<Hixie>
they're solving a problem that people think they have when they're stuck 30,000ft up for 12 hours, but when they get down to earth, they don't find it actually fits in with their workflow
11:04
<Hixie>
i should totally make a sky mall catalogue of web standards
11:04
<annevk>
:)
11:06
<hsivonen>
I'd be interested to know how RDFa ended up in the HTML WG charter
11:07
<annevk>
probably similar to how the word XForms ended up there
11:08
<annevk>
or ITS
11:09
<hsivonen>
Hixie: dtcode issue fixed
11:09
<Hixie>
cool, what was it?
11:10
<hsivonen>
Hixie: no longer saying that <dtcode> cannot be mapped to XML
11:10
<Hixie>
right i mean what was the source of the bug? :-)
11:11
<hsivonen>
Hixie: I had forgotten the NCName check and only checked for customness when name policy was allowing XML violations
11:32
<annevk>
Hixie, insertAdjacentHTML is not an attribute
11:32
<annevk>
Hixie, the sentence outerHTML and insertAdjacentHTML attributes should probably say "members"
15:35
<zcorpan>
i find it amusing that people have opinions about the <script src=javascript:'...'> thing
15:44
<annevk>
maybe there was not enough e-mail
17:02
<zcorpan>
firefox adds a playbackrate=1 attribute to <video> elements
17:02
<takkaria>
fun times
17:02
<hsivonen>
zcorpan: do you mean in add the attribute to teh DOM even if absent from source?
17:03
<zcorpan>
hsivonen: yes
17:03
<zcorpan>
if you have <video playbackrate=2> in source firefox still says it's 1
17:03
<hsivonen>
zcorpan: that sucks. is there a bug report already?
17:03
<zcorpan>
if you set video.playbackRate = 3 firefox updates the content attribute
17:04
<zcorpan>
dunno
17:05
<hsivonen>
not with the obvious word in description, at least
17:06
<zcorpan>
same with <audio>
17:06
<hsivonen>
I'll file a bug.
17:07
<zcorpan>
cool
17:11
<hsivonen>
aargh. my tokenizer loop has gone over 8000 byte codes again
17:16
<hsivonen>
https://bugzilla.mozilla.org/show_bug.cgi?id=450115
19:10
<annevk>
did @whatwg.org just get spam?
19:11
<annevk>
http://googlemobile.blogspot.com/2008/08/google-translate-now-for-iphone.html -_- instead of actually making a normal Web page that just works they add special trickery for the iPhone
23:40
<takkaria>
oops, I almost replied to RB then