00:01
<othermaciej>
when does that happen
00:03
<Philip`>
othermaciej: It says things like "If the image argument is an HTMLImageElement object whose complete attribute is false, then the implementation must raise an INVALID_STATE_ERR exception." referring to drawImage one time, and createPattern another time
00:04
<Philip`>
and I associate test cases with bits of the spec by copying the relevant spec text, and it gets all confused when trying to work out where in the spec that text came from if it's there twice :-(
00:05
<roc>
crikey, the IE layout team is huge
00:05
<othermaciej>
how many people are on it?
00:06
<othermaciej>
ad how do you know?
00:06
<roc>
I'm just counting the number of Microsoft people who've popped up in www-style and w3c-css-wg
00:06
<Hixie>
Philip`: heh
00:07
<Hixie>
roc: i'm not sure it's really working in their favour, to be honest
00:07
<jcranmer>
roc: there are some people on /. who think that MS's IE people don't contribute to CSS/HTML AT ALL
00:07
<Hixie>
i still can't get the 7 months it took to get the IE team's feedback on XHR
00:08
<Hixie>
still can't get over, even
00:08
<jcranmer>
Hixie: that's probably their backlog ;-)
00:08
<Hixie>
they were promising it any time now the whole time along
00:08
<Hixie>
and at the end even put it out under a clickthrough license before finally sending it to the list
00:09
<Hixie>
it's not like they said "it'll take 2 years" or whatever
00:09
<othermaciej>
roc: probably a lot of them are "Program Managers", not actual developers
00:09
<hober>
I thought this was nice: http://lists.w3.org/Archives/Public/www-archive/2008Jun/0114.html
00:10
<roc>
maybe so, but still makes for a big team
00:12
<Hixie>
i really wish i could see inside their day to day workings to see what's really going on
00:12
<Hixie>
maybe someone can sue them and we can get some e-mails out in discovery like in comes vs microsoft
00:12
<Hixie>
(it's clear from those e-mails that they really were "evil" back in the late 90s)
00:13
<Philip`>
Hixie: That's why you should quit Google and get a job at Microsoft
00:13
<Hixie>
uh huh
00:16
<Philip`>
It'd be much more productive to work from the inside
00:16
<roc>
I count 7 people who seem to be doing technical work appearing on the CSS lists
00:16
<roc>
of course that's neither an upper nor a lower bound, but still
00:17
<othermaciej>
there also seem to be totally separate AJAX and JScript teams
00:17
<Hixie>
how about the other lists?
00:17
<Hixie>
yeah
00:17
<Hixie>
the "AJAX swat team" has been cc'ed multiple times on the webapps threads
00:17
<Hixie>
sometimes along with half a dozen other @microsoft.com addresses
00:18
<othermaciej>
and I'm like, "they have a whole team for XHR, and their main goal is not to change it in any way?"
00:18
<roc>
of course, if I was writing a new layout engine from scratch in 3 years I'd want a big team too
00:18
<othermaciej>
I wouldn't
00:18
<hober>
well, I bet it's more of an XDR team...
00:18
<othermaciej>
I would want 3 or 4 really good people
00:18
<roc>
I wouldn't at the start but I would at the end
00:18
<othermaciej>
maybe at the end
00:18
<jcranmer>
othermaciej: I don't that'd work for a 3-yr timeline
00:19
<othermaciej>
Safari was created in 1.5 years from project start
00:19
<jcranmer>
s/don't/doubt/
00:19
<othermaciej>
with a surprisingly small team
00:19
<hober>
you had khtml to start with
00:19
<othermaciej>
of course, we were building on an existing engine
00:19
<jcranmer>
othermaciej: but didn't Safari take webkit from elsewhere?
00:19
<othermaciej>
and expectations were lower
00:19
<othermaciej>
but the khtml of those days was far away from what was needed for compatibility, standards compliance, performance, etc
00:20
<Hixie>
woot, finally finished annotating the spec for the url stuff
00:20
<othermaciej>
then again, Microsoft has the old IE engine to look at and borrow code from
00:21
<othermaciej>
I would guess it is lower in code quality than the KHTML of 7 years ago though
00:21
<othermaciej>
if higher in compatibility
00:21
<Hixie>
why are we assuming a rewrite?
00:21
<jcranmer>
othermaciej: that's not a big a boon as you might expect
00:21
<mcarter>
othermaciej, so are you a webkit developer, or are you on the safari team at apple, or are those the same thing?
00:21
<othermaciej>
well, that is how they described it
00:21
<jcranmer>
trust me, someone who's written the same code over three different times
00:22
<othermaciej>
but it sounds like the only major reworking was layout, not other things like scripting or the DOM
00:22
<othermaciej>
and I don't know how truly "from scratch" that was
00:22
<othermaciej>
mcarter: I'm the manager of the WebKit team at Apple
00:23
<othermaciej>
(and I sideline as a developer on the WebKit open source project)
00:23
<Hixie>
i would imagine their "rewrite" is no more expansive in scope than firefox3's gecko "rewrite"
00:23
<othermaciej>
the WebKit team is part of the Safari Team though
00:24
<Hixie>
that is, lots of new code, lots of code replaced, but still an evolutionary path
00:24
<mcarter>
othermaciej, oh hey, thats great! That means there is actually a chance of seeing WebSocket native in a browser in the forseeable future, right?
00:26
Hixie
renames the tcpconnection folder to "websocket"
00:28
<othermaciej>
mcarter: Apple doesn't comment on future product releases
00:28
<othermaciej>
I'm not sure I like the name WebSocket but I am not inclined to paint the bikeshed
00:28
<othermaciej>
I like that Web is in the name though :-)
00:29
<Hixie>
yeah basically i started from SocketConnection, added Web, then dropped Connection
00:29
<Hixie>
i'm open to better ideas
00:30
<mcarter>
I think WebSocket is pretty good. I mean, Web being in the name is nice. Socket implies that its a real duplex stream
00:30
<mcarter>
WebConnection sounds like a conference name or magazine or something
00:31
<Hixie>
hah
00:37
<roc>
othermaciej: there's one strange thing about not commenting on future product releases
00:37
<roc>
how does an open source project not make public comments about future plans?
00:37
<Hixie>
webkit doesn't have releases
00:37
<Hixie>
they do talk about plans, as far as i can tell
00:37
<Hixie>
at least that's my impression
00:38
<Hixie>
i think i'm going to restructure sections 2 and 3 so that instead of being DOM and Structure, they're, like, "infrastructure" stuff, and then "semantics" stuff
00:38
<Philip`>
mcarter: Why should WebSocket have Web in the name, when pretty much every other web technology doesn't?
00:38
<Philip`>
The webbiness is implicit in the context, since you're going to be using it in a web page
00:39
<Hixie>
the whatwg puts "web" in the name of everything it does
00:39
<mcarter>
Philip`, so as not to confuse a WebSocket with a Socket due to the protocol differences
00:39
<gavin>
I agree with Philip`. Also, just "Socket" is too long. It should be called "Sock".
00:39
<mcarter>
really we should name the protocol, and then name the api after the protocol
00:39
<Hixie>
Web Forms 2.0, Web Apps 1.0, Web Controls, Web Binding Language (the "whatwg internal name" for what is published as xbl2), etc
00:40
<Hixie>
the protocol is the Web Socket Something Protocol
00:40
<Hixie>
not sure what "Something" is. Not Transfer, not Transport, not Control
00:40
<mcarter>
given that the protocol name is the Web Socket Something Protocol, I think WebSocket sounds like a good name for the api
00:40
<mcarter>
Web Socket Stream Protocol?
00:40
<Hixie>
that works
00:40
<Hixie>
or Web Socket Streaming Protocol
00:41
<Hixie>
then again, that might make people think of streaming video
00:41
<mcarter>
yeah, thats a good point
00:41
<takkaria>
Web Socket Socket protocol?
00:41
<mcarter>
Web Socket Framing Protocol?
00:41
<Hixie>
Web Socket Duplex Framed Text And Binary Data Protocol
00:41
<Hixie>
WSDFTABDP
00:42
<takkaria>
just put a - after the WS and you can make people even more confused
00:42
<Hixie>
wsdftadbp://damowmow.com:81/news
00:42
<mcarter>
wow
00:42
<Hixie>
(though i think the scheme should just be ws:// actually)
00:43
<Hixie>
oh right, Web Socket Duplex Protocol is what i was thinking earlier
00:43
<Hixie>
WSDP
00:43
<Hixie>
or just Web Socket Protocol :-)
00:43
<Philip`>
Hixie: All those "Web"s are only in the names, not in the APIs
00:44
<hober>
I hope ws:// annoys soap people :)
00:44
<mcarter>
Philip`, really though, the reason that WebSocket should have the word web in it is that historically the Web hasn't had sockets. Calling something a web socket will be really good for establishing a strong recognition as to what the technology is and how it changes web applications
00:44
<Hixie>
what mcarter said
00:44
<Philip`>
Fair enough :-)
00:45
Philip`
wonders if he'll be able to use it for peer-to-peer communication in the future
00:45
<Hixie>
not the way the current protocol is set up
00:45
<mcarter>
Philip`, what are the use cases of peer-to-peer connections?
00:46
<mcarter>
I was trying to justify the need for peer connections last night
00:46
<roc>
games
00:46
<Philip`>
mcarter: Sharing music illegally
00:46
<roc>
well, some games
00:46
<Hixie>
peer to peer across multiple subnets would be interesting, though a whole different problem than what the Web Socket Protocol solves
00:47
<Hixie>
not clear we could even use the same handshaking mechanism
00:47
<Hixie>
you have to deal with NATting and all kinds of issues
00:47
<Hixie>
(as in NATting on both ends)
00:47
<roc>
content distribution
00:48
<roc>
I was talking to some friends who work at a peer to peer content distribution company, mostly video. They're distributing a browser plugin to do it
00:48
<Philip`>
PC games often seem to work around NATs by using a third server to help initiate the connection, and that sometimes seems to work alright
00:48
<roc>
it's always nice if we can avoid the need for plugins
00:48
<Hixie>
well opera has built in bittorrent support :-)
00:49
<roc>
I don't think we want to write bittorrent into HTML5
00:49
<Philip`>
(though that probably only really works for UDP, not TCP, unless your NAT/firewall is dumb)
00:50
<Philip`>
(and I don't think HTML5 wants to reinvent TCP on top of UDP)
00:50
<Hixie>
that's all we need
00:50
<Hixie>
reaching even further down the stack to reinvent the web!
00:50
<Hixie>
maybe i can get vint to contribute
00:51
<Philip`>
Use case: I want to write an internet router in HTML
00:51
<Hixie>
feel free to file that in bugzilla
00:52
<Philip`>
Who needs fancy Cisco boxes when I could use a commodity PC running Firefox?
00:56
<takkaria>
I think a <scene> tag is more important than writing internet routers in HTML
00:56
<Philip`>
I want a <scone> tag
00:56
<hober>
I'm surprised Dmitry didn't beat Philip` to proposing one...
00:57
<jmb>
Philip`: that'd just open a huge bikeshed about how to pronounce it
00:57
<takkaria>
it wouldn't, you clearly pronounce it as "'s gone"
00:59
Philip`
fortunately agrees with takkaria
01:14
<Hixie>
don't look now, but the top of the spec right now is a mess.
01:14
<Hixie>
in case anyone is wondering what's going on
01:15
<Hixie>
i'm in the middle of reorganising everything again
01:37
<othermaciej>
roc: I can comment on project plans, but I can't comment on when or whether Apple will ship anything in a product
01:38
<othermaciej>
roc: although I don't have a formal authority role in the open source project or anything so I can only speak based on my knowledge, not as an official project representative
02:04
<roc>
the distinction seems rather fine
02:07
<mcarter>
Hixie, there may be some value to allowing the client to set headers on the WebSocket handshake
02:18
<othermaciej>
I can say "the WebKit project is interested in FooML, and John Doe might be working on it soon" but I can't say "Safari 5 will ship in 2 months with full FooML support"
02:18
<roc>
yeah
02:22
<othermaciej>
and yes, there could be confusing borderline cases, but that's the rough outline
02:23
<othermaciej>
the WebKit project is a technology provider, not a product organization
02:23
<othermaciej>
which I guess is somewhat unusual in the realm of open source projects
02:23
<roc>
not really
02:23
<roc>
the Linux kernel is the same
02:24
<othermaciej>
well I guess the fact that they ship releases is ultimately not very relevant to either what people get from Linux distros, or what totally bleeding edge fanboys run (which is probably straight out of someone's git tree)
02:25
<othermaciej>
so yeah, I can tell you we're working on the scheduler but not when Red Hat will ship those changes
02:25
<othermaciej>
to draw the analogy
02:25
<roc>
I'm actually surprised you haven't had to do some labelling of builds to indicate which ones are better than others
02:26
<othermaciej>
of nightly builds you mean?
02:26
<roc>
of any builds
02:27
<roc>
someone said you don't do releases
02:28
<othermaciej>
well we have release/update branches which are generally managed by a specific port/vendor, but the project as a whole just develops on trunk, and we try to keep it stable enough that nightlies or even trunk SVN builds are usuable
02:29
<othermaciej>
the only releases that count are done by vendors/packagers though
02:29
<roc>
how do they know when to branch?
02:29
<othermaciej>
(well I guess others might look at it differently and say HEAD is the only release that counts)
02:30
<othermaciej>
they know when to branch by whatever their release cycle is, and hopefully by participating in the project enough to know what revisions are relatively stable and what fixes may need to be merged from trunk
02:30
<roc>
I see
02:31
<roc>
thanks
02:31
<othermaciej>
in practice many (though not all) vendors try to branch from the same point as Apple's releases, but sometimes with different fixes on top of that
02:42
<takkaria>
there is now an up-to-spec C HTML5 tokeniser
05:35
<Hixie>
i'm amused with the people who are complaining about http://lists.w3.org/Archives/Public/public-html/2008Jun/0180.html
05:35
<Hixie>
e.g. in http://www.w3.org/mid/20080619005249.GI5647@sideshowbarker
05:35
<Hixie>
it's not clear to me what exactly the complaint is
07:24
<Dashiva>
"'quality of argument' is a qualitative statement, showing support for an argument reinforces the argument"
07:24
<Dashiva>
In other words, if you repeat it enough times it becomes true?
07:35
<takkaria>
I would have thought showing support for an argument is only a quantitive measure, not a qualitative one
07:36
<takkaria>
sounds like a category mistake to me
07:49
<gsnedders>
Hixie sure is doing well at not doing the URL section :)
07:52
<Dashiva>
It's like seeing myself procrastinate studying for exams, only way beyond my league
08:41
<shepazu>
takkaria: you're assuming that the number is attached merely statistically... in fact, each +1 is the result of a qualitative judgment by a person on the subject matter, not merely an iteration of an instance
08:42
<shepazu>
Dashiva, in other words, if enough people agree with it, it may be worth taking a further look at it
08:43
<shepazu>
this isn't physics, folks... it's not modeling the real world... it's an interactive exercise in what direction people want the web to move
08:47
<Dashiiiva>
And someone has to say stop when people are suggesting to move into brick walls :)
08:50
<takkaria>
shepazu: good counterargument
08:51
<shepazu>
Dashiiiva, agreed, but it's not being done correctly yet
08:51
<shepazu>
takkaria, thanks
08:52
<shepazu>
fwiw, I wasn't sitting here thinking of that the whole time... I just saw your comment :)
08:53
<shepazu>
mikesmith is doing good work in trying to impose a little order, mind you
08:54
<takkaria>
shepazu: I study philosophy, I'm in the business of arguing, not in the buisness of being right or wrong :)
08:55
<shepazu>
takkaria, then your qualitative assessment on my argument is worth even more :)
08:55
<shepazu>
... or less
08:55
<shepazu>
... or both!
08:56
<Dashiiiva>
Superposition until the specific judgement is observed
08:58
<shepazu>
one of my smartest friends studied philosophy as his major... he knows about 5 languages, including japanese and some chinese... he now writes pen-and-paper RPGs, and jokes that he's a "dozenaire" :)
08:58
<takkaria>
I don't know about all that, I just appreciate good arguments :)
08:59
<shepazu>
heh
09:01
<GDashiva>
takkaria: Philosophize on the futility of quantitative judgement in a self-selecting, non-representative group of people :)
09:10
<takkaria>
actually, I've spent a fair bit of time recently thinking about consensus vs. dictatorial decision-making
09:10
<shepazu>
what's your conclusion?
09:11
<takkaria>
well, for the individual, consensus is best, but for the organism as a whole, some dictatorship/domination is required
09:11
<takkaria>
not an earth-shattering conclusion by any means, really
09:12
<takkaria>
but a similar problem exists when thinking about freedom of expression; for the individual, no restrictions on it are best, but for a society as a whole, some may be justified
09:12
<takkaria>
as such, there is an inevitable and eternal conflict where some people will be beaten down at the expense of others
09:12
<takkaria>
er, s/be beaten down/the dominating voices/
09:13
<takkaria>
and that there's not much one can do about it, really. bit pessimistic but there you go
09:13
<shepazu>
heh
09:14
<shepazu>
I prefer to think of it in terms of guides rather than controllers
09:15
<shepazu>
a smart leader knows when and how to follow the will of his constituents
09:18
<takkaria>
that only works in groups small enough, I'd say
09:19
<Hixie>
i wonder how to split the spec up in terms of defining documents and defining the HTMLDocument object and defining the actual members
09:19
<Hixie>
same with elements, content models, HTMLElement, and the actual members
09:20
<MikeSmith>
Hixie: what would be the goal of splitting it in that way?
09:21
<Hixie>
which way?
09:24
<hsivonen>
how come it's OK for aptest.com to host the issue list of the XHTML2 WG but people complain about http://www.whatwg.org/issues/ being outside w3.org space?
09:24
<Hixie>
because i host whatwg.org, i guess
09:25
<Hixie>
same reason it wasn't ok for me to host the css issues list but fantasai can host it
09:25
<Hixie>
aha, this basic structure makes more sense
09:25
<hsivonen>
hmmkay.
09:25
Hixie
has added a top-level section to the spec
09:26
<Hixie>
hsivonen: (real answer is i've no idea, but i've long given up caring. though if you find out, do let me know.)
09:26
<Philip`>
hsivonen: Maybe just because more people are involved with HTML5 than with XHTML2, so if a small fraction objects then that's 0 people in XHTML2 but >= 1 people in HTML5
09:27
<Hixie>
heh
09:31
<othermaciej>
it's more because HTML5 is both successful and controversial
09:31
<othermaciej>
so people opposed to aspects of it are inclined to object
09:31
<othermaciej>
but people opposed to XHTML2 mostly just ignore it
09:40
<MikeSmith>
hsivonen: is there a published doc/spec somewhere that includes a link to that aptest.com list?
09:40
<MikeSmith>
or that otherwise reference it?
09:40
<MikeSmith>
Hixie: same things for the CSS issues list?
09:41
<Hixie>
i linked to it from my blog
09:41
<MikeSmith>
are those just referenced in editor's drafts or other WG-internal docs
09:41
<Hixie>
fantasai's is linked to from the csswg home page
09:41
<Hixie>
iirc
09:41
<MikeSmith>
Hixie: OK
09:41
<MikeSmith>
thanks
09:41
<Hixie>
(i'm absolutely fine with it myself)
09:41
<Hixie>
(and maybe people have complained, i don't know really. haven't heard of any complaints, but i don't follow css closely any more)
09:42
<hsivonen>
MikeSmith: I see references from Virtual F2F minutes
09:42
<MikeSmith>
OK
09:45
<hsivonen>
also in a draft Implemtation Report: http://www.w3.org/2006/07/SWD/RDFa/implementation-report/
09:45
<Hixie>
Philip`: just as a heads-up, when i'm done with this reorg, you'll probably want to update the spec splitter ids
09:46
<MikeSmith>
hsivonen: cool, thanks, that's even better
09:46
<Hixie>
hey did we ever hear back from the xhtml2wg about your request, mike?
09:47
<MikeSmith>
Hixie: I have not heard a peep
09:47
<Hixie>
k
09:47
<hsivonen>
MikeSmith: I don't have a problem with the XHTML2 WG having a tracker outside w3.org
09:47
<Hixie>
and i assume the tag is still looking at my request and the svgwg still looking at the parsing issue
09:47
<MikeSmith>
I guess I could/should bring it up on the Hypertext CG call
09:48
<MikeSmith>
Hixie: your TAG request is the one about the HTML features that affect Web architecture?
09:48
<Hixie>
MikeSmith: only if you think you can't get away with saying "well i asked and got no reply" when we try to publish
09:48
<Hixie>
MikeSmith: yeah
09:48
<Hixie>
the mathml wg was a breeze to work with
09:48
<Hixie>
i wish other wgs were that easy
09:49
<Hixie>
heck the mathml wg actually took over part of the html5 spec in an automated fashion (david carlisle, specifically, iirc)
09:49
<MikeSmith>
no idea of status on TAG thing. as far as SVG WG feedback on parsing issue, that's something else I can bring up on Hypertext CG call
09:49
<Hixie>
eh, don't worry about it
09:49
<Hixie>
there's no rush
09:50
<Hixie>
i'd rather they did a good thorough job, who knows, they might find some idea i missed
09:50
<MikeSmith>
hsivonen: I have no problem with the XHTML2 WG having a tracker outside w3.org either
09:50
<hsivonen>
MikeSmith: fwiw, I just implemented SVG-in-text/html according to Hixie
09:50
<hsivonen>
(parsing that is)
09:50
<MikeSmith>
hsivonen: yeah, saw that
09:50
<MikeSmith>
Hixie: ok
09:51
<Hixie>
i expect one of the main browsers will do it at some point, and once they ship, well, too late
09:51
othermaciej
wonders if Hixie maybe hopes the SVG WG's feedback doesn't arrive until after HTML5 reaches REC with no SVG support
09:51
<Hixie>
if we get to LC and they still haven't replied, i'll just un-comment-out the svg bits
09:51
<Hixie>
and they can send lc comments
09:51
<Hixie>
that gives them more than a year, which should be plenty of time :-)
09:51
<shepazu>
we are working on it, and plan to make a proposal in the next couple weeks... we are posting something public this week to our list
09:52
<MikeSmith>
hsivonen: for the record, I think WGs should have the discretion to be able to link to outside resources in their specs if they care to
09:52
<Hixie>
shepazu: awesome
09:52
<Hixie>
MikeSmith: i'm surprised few people have mentioned the normative links to the whatwg wiki in html5 so far
09:53
<MikeSmith>
well, given that the HTML WG charter specifically mentions WHATWG, I don't think anybody should be very surprised to see those links there
09:54
<Hixie>
logic doesn't always prevail so successfully
09:56
<Philip`>
The "normative links to a wiki" aspect seems like more of a problem than the "normative links to WHATWG-controlled sites" aspect
09:57
<Hixie>
yeah i haven't heard much about that either
09:57
<Hixie>
in fact i've heard more positive noises than negative noises
09:57
<Hixie>
which really surprises me, given how radical an idea this is for the w3c world
09:57
<Hixie>
(e.g. MikeSmith, contrast that idea to what rigo said)
09:57
<Hixie>
(about the copyright thing)
09:59
<Philip`>
It could be considered fine to have those references for now while the spec is under development, because it's just a way of delegating editorship to anyone who has a wiki account, as long as it's folded into the spec in the future so that there's a stable language to implement
10:00
Hixie
gets an internal error on v.nu
10:00
<Philip`>
so there's no need to complain about the wiki references now, but there will be if such a crazy idea makes it significantly further into the process towards REC
10:00
<Hixie>
Philip`: if people are expecting us to expect that to happen, i have news for them :-)
10:00
<Hixie>
but i think the spec is pretty clear about hte intent being permanent
10:01
<hsivonen>
Hixie: with what input?
10:02
<Hixie>
it said you were notified
10:02
<hsivonen>
Hixie: out of heap space
10:02
<Hixie>
input was the spec, but it runs fine when i do it using curl
10:02
<Hixie>
failed when i did it from the web
10:02
<hsivonen>
I guess I need to do some adjustments
10:02
<Hixie>
same source document
10:03
<hsivonen>
Hixie: http://www.whatwg.org/specs/web-apps/current-work/source-whatwg is 404
10:03
<Hixie>
yeah it's transient, only works while the spec is being regenned
10:03
<Hixie>
do you want a copy?
10:04
<hsivonen>
yes, please
10:04
<hsivonen>
http://www.whatwg.org/specs/web-apps/current-work/source works
10:04
<hsivonen>
I wonder if this is a transient issue of someone else validating something big at the same time
10:04
<Hixie>
http://www.whatwg.org/specs/web-apps/current-work/source-whatwg-frozen
10:05
<Hixie>
i was doing it from the command line and the web interface simultaneously
10:05
<Hixie>
web died
10:05
<Hixie>
works now
10:05
<hsivonen>
ok. thanks
10:05
<hsivonen>
I suppose I should expose OOM in the UI
10:23
<Hixie>
ok i've massively (and i do mean massively) reorganised the spec, moving a lot of stuff around
10:23
<Hixie>
(had to do all this to provide a logical place for the URLs section)
10:24
<Hixie>
i'd be interested in feedback on (a) whether i missed anything where paragraphs refer to "above" or "below" or whatever in now meaningless ways, and such problems, and (b) on whether the new structure makes sense to you and is at least as readable as the previous version.
10:24
MikeSmith
gets ready for muffed-up diffs
10:25
<Hixie>
the text didn't change in any meaningful way, it was mostly just moving sections and updating headings, but i did make some very minor editorial changes here and there to keep things sane
10:25
<MikeSmith>
Hixie: request: when you do reorgs, preferably don't actually change any content at the same time
10:25
<Hixie>
yeah i try to keep it to an absolute minimum
10:25
<MikeSmith>
because then the diffs hide/obscure the content changes
10:25
<hsivonen>
hum does this relate to LEIRIs?
10:25
<MikeSmith>
Hixie: cool
10:26
<Hixie>
e.g. just changing "as it says above" to "as it says in the bla section" and things like that
10:26
<Hixie>
certainly nothing substantive
10:26
<Hixie>
hsivonen: who nows?
10:26
<MikeSmith>
holy god... 971K diff
10:26
<Hixie>
yeah like i said
10:26
<MikeSmith>
Hixie: you should apply for a world record
10:26
<Hixie>
major reorg :-)
10:26
<Lachy>
Hixie, why is the section discussing Link elements and relationships within the Web Browser section, instead of within the Element section?
10:26
<MikeSmith>
:)
10:27
<Hixie>
Lachy: it's within the hyperlinks subsection, which is within the section that introduces browsing contexts
10:28
<hsivonen>
Hixie: I thought LEIRIs were like IRIs but allow more control chars and take the LEIRI-to-URI conversion encoding from the context instead of assuming UTF-8
10:28
<Lachy>
and shouldn't the irrelevant attribute be in the Global attributes section?
10:28
<Hixie>
part of my URL plan is to remove all mention of IRIs
10:28
<Hixie>
so...
10:29
<Hixie>
Lachy: seemed more like an interaction thing
10:29
hsivonen
expects reality to suck for my current IRI-wise correct code
10:29
<Lachy>
ok. Other than that, the new order looks reasonable.
10:29
<Hixie>
Lachy: but both those things didn't change with this reorg, so send mail or file a bug if you want those changed :-) (i could be very easily convinced to move them around)
10:29
<Hixie>
Lachy: sweet
10:30
<Hixie>
hsivonen: reality is certainly proving to suck huge genitals when it comes to the writing of the spec, so that wouldn't surprise me indeed.
10:30
<Hixie>
did you know that non-ascii characters are treated differently in the path section than the query section of a URL?
10:30
<Lachy>
ok, I review it a bit more thorougly and see whether or not it works as-is
10:31
<hsivonen>
that I didn't know
10:31
<Hixie>
neither did i until very recently
10:31
<Hixie>
"dismayed" would be an appropriate word to use to describe my reaction to learning this
10:31
<Hixie>
anyway
10:32
<Hixie>
URLs are fun for tomorrow or friday
10:32
<Hixie>
for now i shall sleep
10:32
<Hixie>
if anyoen has input on the new structure, send mail or file a bug, i'll fix the fallout tomorrow
10:32
<annevk>
nn
10:32
<Hixie>
and thanks again everyone for the help :-)
10:32
<Hixie>
nn
11:25
<annevk>
that diff is a mess
16:17
<annevk>
aroben, would delaying script execution be really bad?
16:17
<aroben>
annevk: how do you delay it?
16:17
<annevk>
aroben, similarly to how it's done for layout today?
16:18
<annevk>
several engines at least first compute layout before returning offsetLeft or something for instance
16:18
<aroben>
annevk: in that case you're not waiting for IO, however
16:20
<annevk>
true, don't know whether that matters
16:24
<aroben>
annevk: in general I would think that waiting for a read from disk could take much longer than a layout could
16:25
<annevk>
maybe it should delay but there should also be an event that would be the recommended usage pattern?
16:26
<annevk>
I don't really want sessionStorage / localStorage to become async
16:43
<othermaciej>
annevk, aroben: if you are talking about localStorage, all you have to do is wait for the local storage data (if there is any for that site) to load from disk before you start running the first script
16:43
<othermaciej>
that's the same as waiting for the data for an external script pretty much
16:43
<othermaciej>
not quite the same thing as the offsetLeft case, which is stopping in the middle of script execution
16:43
<othermaciej>
I don't think the API needs to be async, just the implementation
16:43
<aroben>
annevk: if you allow authors to read before the event, then I don't think they'll ever wait for the event
16:44
<aroben>
othermaciej: yeah, waiting to start script execution seems a little better than pausing during script execution
16:45
<othermaciej>
pausing during script execution blocks the UI waiting on I/O, which is bad
16:46
<othermaciej>
and you can also have a master index of which sites have any per-site data, so you'd know up front whether the possible disk read is even needed
16:50
<annevk>
in Opera it wouldn't block the UI
16:50
<annevk>
but I gues that's a problem other browsers have, yes
17:43
<Philip`>
Hmm, does anyone want an IE8 Tech Beta invitation code?
17:48
smedero
waves
17:55
<Dashiva>
Was the xhtml2 wg this possessive about XHTML 1.0 all along?
17:55
<annevk>
prolly
17:56
<annevk>
I guess Steven made sure that the HTML WG charter didn't mention XHTML
17:57
<Dashiva>
Sucks to be xhtml then
17:58
<om_meet>
I guess the XML serialization of HTML we are mandated to do is somehow at the same time not XHTML
17:58
om_meet
shrugs
17:58
<Dashiva>
But good news! They're going to remove appendix C from XHTML 1.0 to distance themselves from HTML as much as possible
17:59
<annevk>
It just happens that our XHTML means HTML as XML
17:59
<annevk>
It's not at all the same! :p
17:59
<Dashiva>
That means everyone agrees no xhtml over text/html
18:00
<jcranmer>
will IE 8 accept application/xhtml+xml, then?
18:01
<annevk>
What does this have to do with IE 8?
18:01
<Dashiva>
I don't think MS cares much about xhtml2
18:04
<hober>
I guess nothing ever came of the XHTML2 rebranding effort
18:06
<Dashiva>
They rebranded?
18:06
<annevk>
they're calling themselves the XHTML WG and are working on XHTML 1.x stuff mostly
18:06
<hober>
"Those design choices have led to XHTML 2.0 having an identity distinct from HTML. With the chartering of the XHTML 2 Working Group,
18:06
<annevk>
the original XForms WG asked to be renamed to Forms WG to make it more clear that they were about all Forms, but then they're calling themselves XForms WG everywhere...
18:06
<hober>
W3C will continue its technical work on the language at the same time it considers rebranding the technology to clarify its independence
18:06
<Dashiva>
That could be misunderstood as even xhtml2 wg having given up on xhtml2 ;)
18:06
<hober>
and value in the marketplace." -- http://www.w3.org/2007/03/html-pressrelease
18:06
<Lachy>
apparently they plan on doing an XHTML 1.2
18:06
<annevk>
I'm not sure if there's any logic
18:07
<Dashiva>
annevk: They want to be Forms WG to own forms, but they want to be XForms WG because that makes it clear they hate all other kinds of forms
18:07
<deane>
annevk, xhtml was in the orignal charter, SP and others made timbl remove it though: http://www.w3.org/2006/11/HTML-WG-charter.html#scope
18:08
<deane>
new charter: http://www.w3.org/2007/03/HTML-WG-charter
18:08
<hober>
ugh
18:09
<annevk>
deane, actually, I think browser vendors might have requested that, because we didn't want to work on modularization
18:10
<annevk>
(and I still don't, seems like a waste of time)
18:11
<Philip`>
Monolithicity wins!
18:11
<Lachy>
annevk, that doesn't make sense. XHTML 1.0 wasn't modularised and if the charter didn't mention 1.1, then there would be no reason to assume modularisation
18:11
<hober>
right.
18:11
<annevk>
Lachy, "the series of specifications previously published as XHTML version 1" includes XHTML Modularization afaik
18:11
<hober>
I mean, it's pretty clear by now that XHTML 1.1 : XHTML 1.0 :: XML 1.1 : XML 1.0
18:12
<Lachy>
yeah, they should have rephrased it to say XHTML, but without referring to the whole series of specs
18:12
<annevk>
I can agree with that
18:12
<Philip`>
hober: Does that mean they're going to backport the XHTML 1.1 changes into XHTML 1.0?
18:12
<Lachy>
but, whatever, it says we need an XML serialisation, and that means XHTML anyway
18:12
<annevk>
I'm not sure I agree that XHTML version 1 excludes XHTML 1.1
18:12
<hober>
Philip`: heh.
18:15
<deane>
annevk, what's the deal with the forms WG? I asked if I could join and it took them 3 weeks to get back to me, then I was asked to give "some backgound info" on myself
18:55
<gsnedders>
jgraham__: Bug in your outliner: the header element has a rank equal to 1.
18:56
<om_meet>
does Opera 9.5 support HTML5 <video> and <audio>?
18:58
<hasather_>
om_meet: no
19:18
<mcarter>
Hixie, how goes URLs?
19:48
<annevk>
deane, I think they (Forms and XHTML2 WG) have a stricter policy on allowing individuals to join their WG
19:50
annevk
likes the WebSocket stuff
19:51
<deane>
annevk, yes, and I think we all know how successful closed groups are
19:52
<mcarter>
annevk, is that in a draft somewhere, or are you referring to the mailing list / irc logs about WebSocket?
19:52
<annevk>
mcarter, just the buzz :)
19:52
<mcarter>
heh
19:53
<annevk>
deane, fortunately for you there are open alternatives :)
19:53
<mcarter>
you can almost feel it just joining the irc channel...
19:54
<Philip`>
We don't keep the riffraff out!
19:55
hober
drafted a personal reply to the xhtml2 wg email
19:55
<hober>
any comments / suggestions / etc. before I post?
19:55
<hober>
http://edward.oconnor.cx/tmp/xhtml2-reply.txt
19:56
<deane>
Some fool thought that the next version of XHTML was going to be XHTML2: http://digg.com/programming/HTML5_differences_from_HTML4?t=7211519#c7212068
19:56
<deane>
so I corrected him :)
19:58
Philip`
tries to work out how to make that link show the actual comment
19:59
<deane>
hober: run it past mike first, I was talking to him today about this
19:59
<Philip`>
Hmm, now it's scrolling me back to the top of the screen every few hundred milliseconds
19:59
<Philip`>
Oh, there it goes
20:02
<hober>
deane: I'll wait a day or so; he's in-channel and can comment on it whenever--I read the logs when I'm not present
21:07
<hober>
Reworked that draft email w/ feedback from h3h, who signed as well.
21:48
<Lachy>
hober, that's quite a well written draft.
21:48
<Lachy>
But I have a small issue with it...
21:48
hober
perks up
21:49
<Lachy>
where you talk about how XHTML2 follows a trajectory from HTML4/XHTML1, it's not really clear to me how they have done that at all.
21:50
<hober>
I struggled with the wording there
21:50
<Lachy>
It seems to me that the XHTML2 group simply discarded the past and started from scratch
21:50
<hober>
Because they clearly believe that they're along such a trajectory
21:50
<hober>
So I tried to describe what such a trajectory might be
21:50
<Lachy>
although, in a sense, it is somewhat inspired by (X)HTML, since it does share a few element names
21:51
<hober>
right
21:52
<hober>
I think rhetorically it's important to acknowledge both groups' stories
21:52
<hober>
whether or not I understand either
21:52
<hober>
Maybe I can dig up a quote from them, and then skip trying to describe it myself.
21:53
<hober>
That sound good?
21:53
<Lachy>
maybe say something about how it doesn't directly build upon XHTML1, it has been inspired by the past
21:54
<Hixie>
hober: the mail from the xhtmlwg is about the "relationship to xhtml1.x" section, not about xhtml2
21:54
<Lachy>
with a goal of building a new foundation
21:54
<Lachy>
ah, yeah. good point.
21:55
Lachy
wonders how to tell the XHTML2 WG that their work on XHTML 1.x is irrelevant, without offending anyone.
21:56
<Hixie>
i wouldn't bother
21:56
<Hixie>
mike will take care of it
21:56
<Lachy>
ok
21:56
<Hixie>
speaking of which, is MikeSmith around?
21:57
<Lachy>
Hixie, he's probably asleep. It's 05:00 in his timezone
21:57
<Hixie>
that doesn't mean much
21:57
<Dashiva>
Not everyone is you, Hixie :)
21:58
<Hixie>
it doesn't mean much for MikeSmith either as i understand it :-)
21:58
<Dashiva>
I must say, MikeSmith is being really chair-y. I hope he sticks to it
21:58
<Lachy>
yeah, he's been good
21:58
<Dashiva>
Things are happening, things are being done
21:58
gsnedders
sits on MikeSmith
21:59
<Dashiva>
(I just noticed the potential cheery pun. Sorry.)
22:00
<gsnedders>
(I just made the obvious, old, joke)
22:14
<Hixie>
annevk: i don't understand http://lists.w3.org/Archives/Public/public-webapps/2008AprJun/0255.html
23:08
<Philip`>
Hixie: I've updated the spec splitter (just removed one id) to work better with the current reorganisation; let me know if I missed any pages and they look stupidly short/long
23:08
<Hixie>
cool, thanks
23:08
<Hixie>
does the new organisation make more sense, do you think?
23:08
<Hixie>
(did you see any obvious mistakes?)
23:09
<mcarter>
Hixie, how would you describe your relationship to HTML5 besides "The editor of the HTML5 specification" ?
23:09
<Hixie>
that's how i would describe my relationship to html5
23:09
<mcarter>
ok, thanks
23:10
<Philip`>
Hixie: No idea - I just looked at the scrollbars, I didn't read the text at all :-)
23:10
<Hixie>
hah
23:10
<Hixie>
fat lot of use you are :-P
23:12
<Philip`>
I could read through the whole spec looking for any parts that lack sense or contain mistakes, and I'll send my feedback some time during 2010
23:14
<Hixie>
that would be awesome
23:15
<Hixie>
though i recommend sending the feedback as you go
23:15
<Hixie>
so that my mail client doesn't crash when you send your 500MB e-mail
23:15
<Philip`>
But then you'll keep changing the spec as I'm reading, and it will take an infinite time before it's stable
23:16
<Hixie>
(2010 is half-way through our last call period)
23:16
<Hixie>
it won't take an infinite time
23:16
<Hixie>
it'll converge on stability
23:17
<Philip`>
1/x converges on zero but takes an infinite amount of time to get there
23:17
<Hixie>
yeah but it doesn't take an infinite amount of time to get to zero plus or minus epsilon
23:17
<Hixie>
and that's my actual target, not zero
23:18
<Hixie>
when the feedback consists only of people suggesting changes to examples, spelling, grammar, etc, we're basically done
23:19
<Hixie>
ok wtf is a URL.
23:19
<Hixie>
it's not a URI, since URIs don't allow non-escaped non-ASCII characters.
23:19
<Philip`>
I suppose epsilon will increase over time, as you get more and more fed up of working on HTML5 and want to just mark it as 'finished' and move onto something else
23:20
<Hixie>
it's definitely not an IRI, since IRIs appear to be UTF-8 only
23:20
<Hixie>
Philip`: probably not, actually.
23:21
<Hixie>
Philip`: but we'll see
23:21
<jcranmer>
Philip`: you can get remarkably resilient
23:27
<Hixie>
you know the web platform has a problem when you can't complete the sentence "The term URL in this specification is used to mean..."
23:27
<jcranmer>
shouldn't it be URI?
23:27
<hober>
heh
23:27
<jcranmer>
I thought URL was deprecated
23:27
<jcranmer>
Hixie: it also means that you have become a language lawyer
23:28
<hober>
"have become"? :)
23:30
<Philip`>
jcranmer: People have thought HTML was deprecated :-)
23:30
<jcranmer>
I meant the term
23:31
<Philip`>
That means fewer people will object if the term gets redefined into meaning something ugly and web-compatible
23:32
<Philip`>
If I'm not wrong, "URI" already has a specific meaning, so it'd be wrong to use that term to refer to something that's not quite the same
23:34
<Lachy>
Philip`, URL already has a specific meaning too. But if I understand correctly, then we're redefining it to be more realistic
23:37
<Hixie>
URI means what RFC3986 says it means, which isn't what I want URL to mean
23:37
<Hixie>
e.g. it doesn't allow non-ascii characters
23:38
<jcranmer>
isn't that idn is supposed to solve?
23:38
<jcranmer>
er, punycode or whatever it is?
23:40
<Hixie>
yes
23:40
<Hixie>
that's RFC3987
23:40
<Hixie>
but it assumes UTF-8
23:41
<Hixie>
which isn't what browsers do either
23:41
<Hixie>
(they use UTF-8 for the path, and the current encoding for the query component)
23:41
<Hixie>
so we can't say URL = URI, and we can't say URL = IRI
23:41
<jcranmer>
isn't this where one should say "screw encodings, everything's UTF8, and if you don't like that, you can have UTF16"
23:47
<Lachy>
Hixie, so what you really need to define is how a URL occuring within a given document maps to a properly encoded URI/IRI, which can then be retrieved.
23:47
<Hixie>
jcranmer: that doesn't really work well with the existing web :-)
23:47
<Lachy>
am I understanding correctly?
23:47
<Hixie>
Lachy: i think you're right
23:47
<Hixie>
i think i need hsivonen
23:48
<Hixie>
specifically, i want to know if he thinks that non-utf-8 escaped octets in a URI, or non-ascii in a query component in a non-utf-8 document, should be non-conforming
23:50
<Lachy>
what do you mean by a non-utf-8 escaped octet?
23:53
<Hixie>
http://example.com/%FF
23:53
<Hixie>
for example
23:55
<Lachy>
re bug 5772, I think rb is trying to make a distinction between the id="" attribute and attributes of type ID. He wants to preserve the meaning of the type ID which requires uniqueness, and redefine id="" so that it isn't of type ID and doesn't require uniqueness
23:55
<Hixie>
wish he'd say so then :-)
23:56
<Lachy>
but it's rather confusing, since HTML5 isn't defined in terms of a schema language
23:56
<Hixie>
aah, you said the word schema! 5 minutes in the corner!
23:56
<Hixie>
oh crap, so did i.
23:56
Hixie
joins Lachy in the corner.
23:56
<Lachy>
:-D
23:59
<Lachy>
but what he's asking for is a bad idea since it creates problems for getElementById(), which depends on uniqueness for being totally predictable, as the DOM spec leaves behaviour explicitly undefined in the case of duplicates
23:59
<Lachy>
that should, of course, be defined in DOM5 though