00:05
<Steve^>
Text within a <footer> must also be wrapped in a <p>?
00:20
<aho>
Steve^, nop. but it might be handy to have some extra hook for styling purposes
00:21
<Steve^>
Aha, Text is flow content
00:22
<Steve^>
Just making sure I can reproduce your answer from the docs :)
00:23
<aho>
well, it's just like a div... just with some meaning tacked on :>
00:24
<Steve^>
True, though I wasn't sure in that content either
00:24
<Steve^>
context
00:25
<aho>
"when in doubt, validate" ;)
00:25
<aho>
makes everything easier
01:13
<Steve^>
For example on this section: http://dev.w3.org/html5/spec/Overview.html#e-mail-state it does not appear to list the actual value for the email state, "email"
01:14
<Steve^>
It is listed in the table above, but it'd be nice to not have to cross-reference
01:40
<Hixie>
"actual value"?
02:07
<Steve^>
The value to go into the type attribute
09:38
<micheil>
Hixie: in what ways is the websocket protocol spec moving forward from the (what seemed like) a break down in communication between WHATWG and IETF
09:38
<micheil>
?
09:46
<annevk5>
micheil, it's a tad more complicated than that
09:47
<micheil>
okay, explain then
09:47
<annevk5>
heh, could have expected that
09:47
<micheil>
because I'm totally lost at the moment as to what's going on.
09:47
<annevk5>
we're still communicating
09:48
<micheil>
okay, it sounded like there was a break down of communication.
09:49
<annevk5>
I'm not quite sure where it is heading though, but everyone is still there and nobody has been ousted
09:49
<micheil>
okay
09:50
<annevk5>
it's mostly that people have somewhat varied views on how extensibility is to be addressed
09:50
<micheil>
well, are we still going ahead with the draft76 style packet format, or moving towards the VI one?
09:50
<micheil>
erm, Framing Take IV
09:51
<annevk5>
ranging from WG in control to wide-open (and also way more complicated)
09:51
<micheil>
hmm..
09:51
<annevk5>
Hixie changed the framing
09:51
<micheil>
well, from an implementors perspective, using the Framing Take IV is a hell of a lot simpler.
09:51
<micheil>
did he?
09:51
<annevk5>
it now uses 64-bit
09:52
<micheil>
okay, you've lost me there; let me read up on the spec.
09:53
<annevk5>
http://www.whatwg.org/specs/web-apps/current-work/complete/network.html#protocol-overview
09:53
<annevk5>
frames are now
09:53
<annevk5>
frame-type frame-length data
09:53
<micheil>
one moment, safari's busy crashing trying to load the compelte one :P
09:53
<annevk5>
with frame-length being a length encoded as 8 octets
09:53
<annevk5>
(the above is multipage)
09:54
<micheil>
okay, cool
09:55
<micheil>
so what, is it still using \x00 and \xFF as the start/end delimeters?
09:56
<micheil>
or are we using just a message is frame-length long
09:56
<annevk5>
a message with a type that is frame-length long
09:57
<micheil>
okay, example?
09:57
<annevk5>
they're in the spec?
09:58
<annevk5>
one octet for type (only text and closing supported currently); eight octets for lengths as unsigned 64-bit int; data blob
09:59
<micheil>
so, example packet may be: <0xFF><11><hello world>
09:59
<micheil>
<11> being a big endian 64 bit int
10:00
<micheil>
<hello world> being UTF8 encoded binary data?
10:00
<annevk5>
yes
10:01
<micheil>
okay
10:01
<micheil>
and, if the content was say, binary, it might be something like <0xFE><11><binary data>
10:02
<micheil>
where <binary data> would be 11 bytes of binary data
10:03
<micheil>
(if binary data were to be supported)
10:03
<micheil>
would it make more sense to have one frame-type reserved for metadata, where an example might be:
10:04
<micheil>
<0xF0><13><Encoding=UTF8>
10:04
<micheil>
which would tell the server that the client expects data to be encoded with UTF8 encoding
10:05
<annevk5>
that requires API changes
10:05
<annevk5>
on top of which we do not want to support more than one encoding for text
10:05
<micheil>
(this would be sent onconnect by the client, but could also be sent by calling ws.setEncoding(enc) )
10:05
<micheil>
where enc maybe a DOMString of either: "binary" or "utf8"
10:06
<annevk5>
seems simpler to just have a different frame type
10:06
<micheil>
okay
10:06
<micheil>
that sounds fair enough then
10:06
<micheil>
but then, how do I change between binary and utf8 data?
10:06
<micheil>
(without requiring massive changes)
10:07
<micheil>
I'm thinking that when the server receives that Encoding=UTF8, it will then encode all out-going messages as UTF8 data
10:07
<micheil>
if it receives Encoding=Binary, it will then encode all out-going messages as raw binary data
10:09
<micheil>
because otherwise the server cannot send binary data
10:09
<micheil>
unless you make everything always binary data, and only have the clients decode to utf8 or something
10:09
<micheil>
but that's not a good direction to go
10:12
<annevk5>
once binary support is there the server can do whatever it wants
10:12
<annevk5>
and the javascript library will just handle binary and strings
10:12
<annevk5>
and in the transition period you can tell the server with a simple text message that your client supports binary
10:12
<micheil>
yeah, but if the data coming in on the "message" DOMEvent is UTF8 encoded, then you'd have to do odd stuff
10:12
<annevk5>
from the application layer
10:13
<annevk5>
a different frame type would be exposed differently in the API
10:14
<annevk5>
(maybe not for all future frame types; but binary and strings get their own primitive; but since JS doesn't have one for binary; we're waiting)
10:14
<micheil>
okay, I think it'd make sense to emit just one event "message"
10:16
<micheil>
and then to change the encoding of those message events, you call a client side API of setEncoding("...")
10:16
<annevk5>
no
10:17
<micheil>
okay
10:17
<annevk5>
just like with XMLHttpRequest send() will just be overloaded
10:18
<annevk5>
and the data attribute of the message event will be a ByteArray (or whatever it ends up being) rather than a DOMString
10:18
<annevk5>
(it's of type any after all)
10:20
<micheil>
okay, so inside ws.send's function, it'd be something like: if(typeof data === "string") { send as string } else if(typeof data === "bytearray") { encode to binary }
10:21
<micheil>
would that also mean that the onmessage api would need to do that same, or would it get given the content type?
10:21
<micheil>
eg, server sends a binary message, then a utf8 message, then another binary message
10:22
<micheil>
would the events still be: message, message, message, message
10:22
<micheil>
and would you still get (data) as the arguments to the function call?
10:22
<annevk5>
it would be just message
10:22
<micheil>
or would you get: (data, binary)
10:22
<annevk5>
but data would be of varied type
10:23
<micheil>
okay
10:24
<micheil>
so I'd need to manually check for typeof data
10:24
<annevk5>
you'll always need to do some kind of check; but yes
10:25
<micheil>
okay
10:25
<micheil>
and there's no more usage of that end delimiter format, right?
10:25
<annevk5>
having said all this, nobody has patched their browser implementation just yet
10:25
<micheil>
there's just one packet format, no high or low byte messages or what ever they were?
10:26
<annevk5>
micheil, right (that's not coming back)
10:26
<micheil>
(I'm just wanting to get up to speed)
10:26
<micheil>
I kinda got lost wading through the 200 odd emails in 1 or so days last week
10:26
<annevk5>
well, the binary/text stuff has remained the same; it's just the packaging that's different now
10:27
<annevk5>
from an API point of view nothing really changed
10:27
<annevk5>
yeah, it's kind of crazy
10:27
<micheil>
yeah
10:27
<micheil>
as long as both text and binary share the same actually packaging around them when sent on the wire, I'm cool with that.
13:00
<gsnedders>
Wikipedia is distracting.
13:12
<annevk5>
and writing a complete test suite is hard
13:12
<annevk5>
even "just" for XMLHttpRequest
16:24
AryehGregor
discovers apt-get install pastebinit
16:29
gsnedders
discovers he is incapable of writing for any period of time now
16:33
<Philip`>
gsnedders: I'd blame Twitter
16:35
<Philip`>
(Also, IRC)
16:35
<micheil>
the main part is getting into the right mental frame to be able to write.
16:40
<gsnedders>
micheil: A length delimited or sential delimited frame?
16:41
<micheil>
>_>
16:41
<micheil>
preferably a length delimited frame
16:53
<Workshiva>
I would prefer sentinel delimited, with the sentinel being "I have nothing more that needs writing"
16:53
<annevk5>
would be nice if dev.w3.org had a favicon
16:54
<micheil>
Workshiva: hmm, that could work
18:10
hdhoang
encounters a page for which html5 parser is more useful than chromium's \o/ ( http://www.ishmael.org/Education/Writings/ )
18:49
<gsnedders>
"Note: There will be at least one such element, by definition." — I don't get how there will be.
18:51
<Ms2ger`>
<p class="note">There will be at least one such element, by definition.<!-- (If there wasn't, then this algorithm wouldn't have been invoked by Web IDL.) --></p>
18:51
<gsnedders>
Ah
18:51
gsnedders
tries to work out why
18:53
<gsnedders>
Oh, the "corresponding name property", I think…
19:06
<gsnedders>
Philip`: You had a bug on SpiderMonkey about Math.sin varying between x86 and x86-64, no?
19:16
<Philip`>
gsnedders: https://bugzilla.mozilla.org/show_bug.cgi?id=531915 ?
19:16
<gsnedders>
Philip`: Yeah, thanks
19:24
<gsnedders>
Philip`: Also, if you want more things to test in your canvas test suite, you could take a look at http://codingforums.com/showthread.php?t=198624
20:10
<AryehGregor>
No one much seems to have commented on the srcdoc issue.
20:10
<AryehGregor>
http://www.w3.org/2002/09/wbs/40318/issue-100-objection-poll/
20:37
AryehGregor
has his doubts about whether srcdoc is actually very useful, to be honest
22:09
<Hixie>
seriously tempted to ask if i can use http://alexkessinger.net/story/html5-brief-one-paragraph as the abstract for the spec
22:35
<Steve^>
I notice that the intention is for the UA to replace the content of <time> with a locale-specific form. I think it'd be helpful to have a set format describing how the UA would do this, or else it is difficult to use the value in the right context
22:36
<Steve^>
For example it may be important to the content to stress that a date was on a saturday. So I could do Saturday <time datetime="2010-08-07">8th August</time>. But for all I know the UA could convert this to Saturday Sat 8th Aug 2010
22:41
AryehGregor
is a little skeptical of the idea of locale-specific rendering here.
22:41
<Workshiva>
Steve^: Where does it say that?
22:42
<AryehGregor>
Hmm, it seems not to? Good.
22:42
<AryehGregor>
Sounds like a bad idea.
22:42
<Steve^>
In the time element text
22:42
<Steve^>
Here is another way that could be marked up. In this example, legacy user agents would say "today", while newer user agents would render the time in a locale-specific manner based on the value of the attribute.
22:42
<AryehGregor>
Oh, wait.
22:42
<Steve^>
^ quote
22:42
<AryehGregor>
That's non-normative.
22:43
<AryehGregor>
What it actually says in normative text is "If the datetime attribute is present, the user agent should convey the attribute's value to the user when rendering the element."
22:43
AryehGregor
is unsure whether that's useful.
22:43
<Steve^>
What does non-normative mean?
22:44
<AryehGregor>
Means "not really part of the standard, just meant to help you understand the actual requirements, ignore it if it contradicts normative text".
22:44
<Steve^>
Oh
22:44
<Steve^>
The text I refer to isn't even inside the non-normative "box"
22:44
<Workshiva>
But there seems to be something in the bindings section
22:45
<AryehGregor>
It's in <div class="example">, thus non-normative.
22:45
<Workshiva>
When the time binding applies to a time element, the element is expected to render as if it contained text conveying the date (if known), time (if known), and time-zone offset (if known) represented by the element, in the fashion most convenient for the user.
22:45
<Steve^>
I didn't read the code...
22:46
<Steve^>
I'm a little confused now to your opinion on the matter
22:46
<Steve^>
Is the spec right or wrong?
22:46
<Workshiva>
Steve^: It looks like you're right
22:46
<Steve^>
I'm right that it's silly?
22:47
<Workshiva>
Maybe
22:47
<Steve^>
As an aside, this non-normative stuff sounds misleading and shouldn't be in the spec if it's not valid
22:48
<Workshiva>
Well, "a locale-specific way" gives the UA a license to do pretty much anything it wants
22:49
<Steve^>
which is impossible for content authors to predict
22:49
<AryehGregor>
File a bug.
22:49
<Workshiva>
I think it's one of those problems that won't actually occur
22:49
<AryehGregor>
Why not?
22:49
<Steve^>
It seems that there are two seperate use cases here: One for text that should be easily cross-referenced to a calendar and the second for showing dates in a locale-specific manner
22:49
<Workshiva>
That UAs will magically all decide on formats that don't include the day of week
22:50
<Steve^>
Sure
22:50
<AryehGregor>
But why do we actually want it to be locale-specific?
22:50
<AryehGregor>
That would just be weird.
22:50
<AryehGregor>
Like a Chinese person seeing dates in Chinese on an otherwise English page?
22:50
<Steve^>
I think these two use cases conflict... with the former I probably want the value of "today" or "Saturday" to persist, with the true date hidden until asked for
22:51
<Workshiva>
AryehGregor: More like "Let the user configure the browser to show a sensible date format instead of that American rubbish"
22:51
<Steve^>
AryehGregor, 10/02/2010 vs 02/10/2010 for 10th Feb
22:51
<AryehGregor>
And if you're trying to guess whether 6/7/2010 refers to June 7 or July 6, you now have to not only take into account whether the site you're on is mainly American or British, but also try to guess whether it used <time>, and if so, what your browser thinks the right order is for you?
22:51
<AryehGregor>
Seems like a recipe for confusion.
22:52
<AryehGregor>
In the markup, but it could still be ambiguous in the display.
22:52
<Workshiva>
If the user wants it to be
22:52
<Steve^>
hmm
22:52
<Steve^>
I think the likely case is that the display will always be informal
22:53
<Steve^>
But I somewhat agree
22:53
<Steve^>
The first use case is the important one
22:53
<AryehGregor>
Which?
22:53
<Steve^>
Calendar integration
22:53
<AryehGregor>
Yes.
22:53
<AryehGregor>
Also <time pubdate> is nice for search engines.
22:53
<Steve^>
yes
22:54
<AryehGregor>
I don't think <time> should be displaying stuff differently from what its contents indicate.
22:54
<AryehGregor>
Authors can script that if they like, it doesn't need to be on the browser level.
22:54
<AryehGregor>
Alternatively, users could configure their browser to do it via an extension or something.
22:54
<AryehGregor>
It shouldn't be mandated by the spec.
22:54
<Steve^>
It would result as <time pubdate datetime="2010-05-25" /> 25th May
22:54
<Steve^>
Seperate, to avoid the browser messing with your styles
22:55
<Steve^>
Thus, you can dynamically show "Today", "Yesterday", "25th May"
22:55
<Steve^>
The browser can't decide that for you, but twitter styles feeds would insist on it
22:57
<Steve^>
AryehGregor, I should file a bug in bugzilla?
22:57
<AryehGregor>
Yes, you can use the form at the bottom of the WHATWG version of the spec.
22:57
<AryehGregor>
Post the URL of the bug so I can voice support.
22:57
<Workshiva>
If you want to track the bug, it's better to file manually
22:57
<Workshiva>
That way you get properly registered as reporter
22:57
<Steve^>
What is the different between the two versions?
22:58
<Workshiva>
The spec form is anonymous
22:58
<Steve^>
Ok, http://www.w3.org/Bugs/Public/enter_bug.cgi ?
22:58
<Steve^>
(Is that the right one for the right spec?)
22:59
<AryehGregor>
Yeah.
23:14
<Steve^>
http://www.w3.org/Bugs/Public/show_bug.cgi?id=10318
23:54
<cardona507>
hola boyz