00:09
<Philip`>
Wow, I'd forgotten how great Verlet integration is when compared to Euler integration
00:17
<Philip`>
http://canvex.lazyilluminati.com/misc/physics.html - I don't know why it doesn't work nicely in Opera - the circle rendering is broken and it only goes at 10fps :-(
00:18
<Philip`>
Also it's extremely jerky in FF3, but at least FF2 works fine...
00:22
<MasterLexx>
okay, now i am not anymore so sure of the success of html 5 http://www.webdevout.net/tidings/2007/04/23/the-whimzical-world-of-html-5/
00:24
<MasterLexx>
i hope those html5 guys will make something good out of it, but i don't think this will happen in the next time
00:25
<Hixie>
we're the html5 guys :-)
00:31
<MasterLexx>
ohh
00:32
<MasterLexx>
hmm, i hope you guys fix those things
00:35
<Hixie>
what do you want us to fix?
00:42
<MasterLexx>
i meant those things described there, like the bugmode
00:42
<MasterLexx>
and the thing with no version information.... why not? i don't think it can harm.
00:42
<Hixie>
the bugmode was just an idea someone proposed, we're not doing it
00:43
<Hixie>
why would you want version information? it's just more stuff for you to remember when you write HTML pages
00:43
<Hixie>
and most people don't bother with it anyway
00:43
<MasterLexx>
browsers have to bother with it
00:43
<Hixie>
why?
00:43
<Hixie>
they haven't up to now...
00:43
<MasterLexx>
it's not bad, at least let is a option!
00:43
<Hixie>
i don't understand what you would want it for
00:44
<Lachy>
only IE is going to bother implementing a bug mode switch
00:44
<MasterLexx>
i meant, i could still be there as optional emlement
00:44
<Hixie>
right, but what would it be good for?
00:44
<Hixie>
there's no point having an optional thing if it isn't used for anything
00:44
<Hixie>
just makes people who don't understand HTML think it's important and confuses them
00:45
<MasterLexx>
as wrtitten in the article, what if something goes wrong? what if one docuemnt has to be handeled in another manner?
00:45
<MasterLexx>
(sorry for my grammar, i am german)
00:45
<Hixie>
i don't understand why you would want documents to be handled in a different manner
00:45
<Hixie>
wouldn't we want just one set of rules?
00:46
<Hixie>
having multiple different sets of rules just seems like it would be really confusing for authors
00:46
<MasterLexx>
what if in a future html 6 some elemets are deprecated...
00:46
<MasterLexx>
i don't think so, you have one set of elements and attributes that you use, like in xhtml 1.0 strict, i don't use others...
00:46
<Lachy>
elements still have tob e supported, even if they're deprecated
00:46
<Hixie>
deprecating them doesn't mean browsers stop supporting it
00:47
<Hixie>
e.g. browsers still support <p align="">, but align="" isn't in HTML5
00:47
<MasterLexx>
but then this is some sort of quirks mode
00:47
<Hixie>
no, it's just the one mode
00:47
<Hixie>
we're trying to remove quirks mode as much as possible
00:47
<Hixie>
just have one mode
00:48
<MasterLexx>
oh god
00:49
<MasterLexx>
i think, there will be always other versions and things that are left out of new ones...
00:49
<Hixie>
they'll be left out of the language, but that doesn't affect what browsers have to do
00:49
<Hixie>
browsers just have one set of code that handles all versions
00:49
<MasterLexx>
nahh okay, i think i will wait a bit and see what heppens to html5
00:50
<MasterLexx>
i don't think i understand much of this, but it would meen that there will be a mess of elemets and attributes that all always have to be supported by browsers, there will never be a clear html this way
00:51
<Hixie>
yes, browsers will always support old documents
00:51
<Hixie>
having a version doesn't change that
00:51
<madmoose>
Browsers that want to be backwards compatible have to support all the old stuff anyway.
00:52
<Hixie>
note though that even though browsers have to support old stuff, it doesn't mean new versions of html can't be "clean"
00:52
<Hixie>
e.g. HTML5 doesn't have <tt> and <center> and so on
00:52
<Hixie>
even though browsers support <tt> and <center>
00:52
<MasterLexx>
okay okay, i read today the first time about html 5 in the news on selfhtml... i think i need to know more first
00:52
<madmoose>
Microsofts argument for a version number (if I understand it correctly) is that they want to support old bugs in their implementation, but their release cycle isn't going to coincide with html versions anyway.
00:54
<Lachy>
Microsoft's argument is flawed
00:54
<Lachy>
they want to perpetuate every bug in every single browser version, for all time
00:55
<madmoose>
Well that's their prerogative.
00:55
<madmoose>
It doesn't have anything to do with html version numbers though.
00:56
<mpt>
gsnedders, iCab doesn't have alerts asking you if you want pages to render per spec or per Web, but it does have checkboxes for it
00:57
<mpt>
hsivonen, on the Web at least, starship names are possibly more common than ship names
00:59
<MasterLexx>
what?
01:00
<MasterLexx>
bye and good night
08:28
<annevk>
"
08:28
<annevk>
HTML5 is that spec. That was the original goal of the WHATWG effort, and
08:28
<annevk>
continues to be this goal. There are already other groups writing
08:28
<annevk>
specifications that don't take today's content into account, e.g. XHTML2.
08:28
<annevk>
Those specs will be ignored. I have no intention of writing a spec that
08:28
<annevk>
will be ignored, it seems like a spectacular waste of time and of the
08:28
<annevk>
human race's resources."
08:32
<Hixie>
hm?
08:33
<annevk>
I liked the way you phrased it
08:34
<annevk>
btw, it seems like <script for= event=> needs to be implemented :(
08:36
<Hixie>
it's not backwards compatible
08:37
<annevk>
it's implemented in IE and Firefox and is used by plugins
08:37
<annevk>
executeSql() isn't backwards compatible either for that matter :)
08:37
<Hixie>
it's used in firefox?
08:38
<Hixie>
executeSql doesn't cause scripts to run when they shouldn't
08:38
<annevk>
yeah
09:21
<mikeday>
some of the imperative pseudo-code/prose in the HTML5 spec is rather difficult to follow
09:29
<annevk>
you have to be a native en-GB-x-Hixie speaker
09:30
<mikeday>
that could be awkward
09:31
<mikeday>
it's rather frustrating, as you can almost represent all this stuff with regular expressions
09:31
<mikeday>
which would be a lot easier to check that you've got it right
09:32
<mikeday>
I'm stuck on the "get an attribute" step
09:32
<annevk>
if you read carefully it shouldn't be a problem
09:32
<mikeday>
the code gets rather long and fiddly
09:37
<annevk>
you don't have to follow the specification
09:37
<annevk>
you just have to ensure that A > algorithm > B is identicala
09:37
<jgraham>
mikeday: Charset sniffing? Yeah, I didn't like doing that either
09:37
<annevk>
actually, A > B > C
09:37
<annevk>
where B can be English prose or C
09:37
<annevk>
or Python, etc.
09:38
<mikeday>
right
09:45
<mikeday>
the charset sniffer seems to parse attributes on close tags, eg. </foo bar="baz">
09:45
<annevk>
yeah
09:45
<annevk>
tokenizer does that too
09:46
<mikeday>
does the tokenizer parse them then throw them away?
09:46
<annevk>
yes
09:46
<mikeday>
phew :)
09:49
<mikeday>
what I meant about the spec is that the description is actually lower level than the code
09:49
<mikeday>
it's easier to implement a high level spec with low level code than the reverse.
09:49
<annevk>
hmm
09:49
<annevk>
isn't the spec at the same level?
09:50
<annevk>
charset sniffing needs to be done at the byte level
09:51
<mikeday>
I mean it's very imperative
09:51
<mikeday>
the spec is longer than your python implementation, for example
09:51
<mikeday>
and much harder to understand
09:53
<annevk>
so how can it be done in the same way being as exact as it is now?
09:53
<annevk>
if it can be improved I suppose it should be done
09:54
<mikeday>
that's what I'm trying to figure out
09:54
<mikeday>
using regular expressions would be an obvious place to start
09:54
<mikeday>
and it's easy to see how that could be done for most of it
09:55
<mikeday>
for example: <![^>]*>
09:55
<mikeday>
that's the regular expression to match <!DOCTYPE...> and similar junk and skip over it
09:56
<mikeday>
similarly for comments: <!--([^-]|-[^-]|--[^>])*-->
09:56
<mikeday>
defining it using a regular expression means that there is no ambiguity about <!--> vs. <!---> vs. <!---->
09:57
<mikeday>
start tags and get attribute are a little trickier, but I think can still be done
09:58
<mikeday>
to match the beginning of a start tag: </?[A-Za-z][^\s><]*
09:59
<mikeday>
(or end tag, as it has an optional slash)
10:00
<mikeday>
the resulting regular expressions could just about be pasted straight into lex or perl or whatever
10:01
<annevk>
however, it makes it harder to introduce <!--> as comment
10:02
<mikeday>
very easy, just change the regular expression
10:02
<mikeday>
much easier than editing the prose
10:03
<mikeday>
actually, the expression I pasted before is a little bit too strict
10:03
<annevk>
if you're going down that route you have to define what subset of regular expressions and such...
10:03
<mikeday>
<!--([^-]|-+[^>])*--> would be better
10:03
annevk
prefers the prose actually
10:04
<mikeday>
that's not a problem, many many specifications use standard subset of regular expressions or BNF
10:04
<mikeday>
oh well :)
10:05
<annevk>
regular expressions also encourages a particular implementation
10:05
<annevk>
which isn't necessarily good
10:06
<annevk>
meeting
10:06
<mikeday>
hmm, I think the current prose is more biased towards a particular implementation
10:06
<mikeday>
whereas regular expressions can be implemented directly, or used as specification for hand written parser
10:12
<mikeday>
ahah, regular expression for attributes: [\s/]*([^<>][^=\s/<>]*)[\s]*=[\s]*('[^']*'|"[^"]*"|[^\s<>]*)
10:13
<mikeday>
it may look like line noise, but you don't have to be an en-GB-x-Hixie speaker to understand it :)
11:27
<annevk>
mikeday|away, you have to be crazy instead :p
11:32
<mikeday>
it's not really as nasty as it looks :)
12:10
<gsnedders>
mikeday: and how do you define error handling with ABNF/regex?
12:11
<mikeday>
you don't have to
12:11
<mikeday>
charset sniffing doesn't do error handling
12:12
<gsnedders>
mikeday: the majority of the tokeniser does, though
12:12
<mikeday>
the tokeniser is more complex
12:13
<mikeday>
but even that could be "more formalised" than it currently is
12:13
<gsnedders>
mikeday: then where were you going to use ABNF/regex? in the document conformance section?
12:13
<mikeday>
I'm just looking at charset sniffing now
12:15
<mikeday>
currently it is defined imperatively: go forward one byte, if it's this then goto step 1, otherwise step 2...
12:15
<gsnedders>
trying to work out how to implement things without using regex when given like that can get rather complex
12:17
<mikeday>
eh? you mean if I give you a regular expression you don't know what to do with it, or what?
12:17
<gsnedders>
I mean trying to implement a complex regular expression without using regular expressions can be hard to work out
12:18
<mikeday>
well, not really; regular expressions can be transformed into code mechanically
12:18
<mikeday>
unlike prose.
12:19
<Dashiva>
The regular expressions have to be created from prose first, though
12:19
<Philip`>
Would BNF be easier to understand and less line-noise-like than regular expressions, but still work about the same?
12:20
<mikeday>
yes
12:20
<gsnedders>
BNF is slightly easier, yes
12:20
<mikeday>
you would create definitions for commonly used bits
12:20
<mikeday>
avoid regular expression style shorthand
12:20
<mikeday>
more like the EBNF in the XML spec
12:21
<gsnedders>
I have a preference of ABNF, personally
12:21
<mikeday>
Comment ::= '<!--' Char* '-->'
12:21
<mikeday>
(as a rough example)
12:24
<annevk>
tree construction is now 63 lines...
12:24
<annevk>
two simple functions :)
12:25
<mikeday>
sounds good!
12:26
gsnedders
still doesn't really know how to parse URIs without using regular expressions, when you don't know how many parts of the URI you have
12:27
<mikeday>
you mean like optional query parameters or port numbers?
12:27
<gsnedders>
or schemes, or authority, etc.
12:28
<mikeday>
it's pretty horrible
12:29
<mikeday>
:)
12:29
<gsnedders>
I can think of ways of doing it, but all very expensive
12:29
<gsnedders>
and in a non-cached interpreted language…
12:30
<mikeday>
the overlap between URLs and filenames makes things even worse
12:31
<mikeday>
c:foo is a filename, not a URL with scheme "c"
12:31
<mikeday>
of course, http://slashdot.org is a valid filename on Linux...
12:32
<gsnedders>
what FSes is it not valid on?
12:33
<gsnedders>
NTFS?
12:39
<annevk>
so it will become a little longer I'm afraid
12:39
<annevk>
need to deal with after the last closing tag too
12:39
<annevk>
oops
12:42
<gsnedders>
only two exams left this year!
12:58
<mikeday>
Windows won't like http://... as a filename I suspect
13:02
<Dashiva>
\/:*?"<>| are illegal
13:02
<mikeday>
but anyway, if you try to use "http://..." as a filename, many programs will treat it as a URL instead
13:03
<mikeday>
'cos it starts with http:
13:03
<mikeday>
can't really blame them
13:20
<gsnedders>
Mac OS X won't like http:// as a filename, but HFS+ doesn't care. You can actually create a file called that going through the POSIX layers of OS X
13:45
annevk
wonders why mike uses both tabs and spaces
13:45
annevk
doesn't like that
13:47
<MikeSmith>
annevk - mike = me?
13:48
<annevk>
no, mikeday
13:48
<annevk>
in his libhtml project
13:51
<MikeSmith>
maybe because he inherited parts of the code from somebody ...
13:52
<MikeSmith>
happens to me sometimes
13:52
<annevk>
seems like he's starting from zero
13:52
<annevk>
(there's not much there so far)
14:04
<annevk>
XML5 sort of works now :)
14:05
<annevk>
there's no spec yet, but there's an implementation
14:06
<Dashiva>
What are the major changes so far?
14:07
<annevk>
I don't have DOCTYPE nodes (DOCTYPEs do affect processing of entities), I got </>, I got error handling
14:07
<annevk>
s/got/have//
14:08
<annevk>
I need to make testcases at some point and once I tweaked stuff a bit more I want to set up some online demo so people can generate trees for themselves
14:36
<Philip`>
zcorpan: http://canvex.lazyilluminati.com/misc/physics.html (probably works best in Firefox) seems to be the desired kind of concept, and it's not particularly complex, though I still don't think it could get close enough to Elasto Mania without having their code to copy :-(
14:37
<annevk>
Philip`, btw, instead of using toDataURL and drawImage() one could use getImageData and putImageData
14:38
<met_>
Philip`looks nice
14:39
<zcorpan>
Philip`: cool!
14:39
<Philip`>
annevk: You can't really use get/putImageData if you want to save the image between browser sessions, because it's not unlikely that the user will change their screen resolution or switch to a different browser, and then ImageData will no longer correspond to the correct size (and you have no way of knowing what the correct size would be)
14:41
<Philip`>
Also it seems incredibly inefficient saving ImageData to disk, because it'll be dozens of bytes per pixel, whereas toDataURL is nice since it's about the best compression possible
14:41
<Philip`>
(Er, except it's base64-encoded, which kind of damages that a bit)
14:41
<annevk>
the first is an argument for having canvas pixel == image data pixel and the second, yeah fair enough
14:41
<Philip`>
(but not a lot)
14:42
<annevk>
doesn't have to be base64 encoded in theory
14:43
<Philip`>
Hmm, can you get Unicode URIs with no limits on whatever characters are included? Then you could do image/png;base65536,<...> which'd be better
14:43
<annevk>
Philip`, I managed to jump out your physics model
14:43
<annevk>
I'm falling hard and forever :)
14:43
<met_>
me too
14:43
<Philip`>
The jumping-through-walls thing is an intentional feature ;-)
14:43
<met_>
on the left side
14:44
<Philip`>
because apparently that's what Elasto Mania does, so I'm just doing the same, though I'm probably using larger timesteps so it's a more visible problem
14:44
<Dashiva>
Philip`: Might as well stick to base256, no? :)
14:44
<Philip`>
Dashiva: Depends on whether it's being stored on disk as UTF-8 or UTF-16
14:45
<Philip`>
Actually, neither of those would work because they've got funny non-codepoint values :-(
14:45
<Dashiva>
Yep
14:45
<Philip`>
If you're storing ISO-8859-1 or UCS-2, then it'd work as base 256/65536 nicely
14:47
<Philip`>
annevk: If you had canvas pixel == image data pixel, why would one want to use get/putImageData instead of toDataURL/drawImage?
14:48
<annevk>
security
14:48
<Philip`>
Ah, right
14:50
<Philip`>
...though toDataURL/drawImage should have the same security permissions/restrictions as ImageData (I think), except that current implementations don't do that
14:50
<annevk>
It's still unclear to me if you can somehow get an unsafe data: URL
14:53
<Philip`>
If you've got any kind of data: URL at all, how is being able to new Image()/drawImage/toDataURL any worse than simply transmitting the data: URL string directly to somebody?
14:57
<Philip`>
If you've got an image which was loaded from an external data: URL (or in another document or something) and you can't actually get the data: URL string out from it, then I think that's handled under different security rules and it's given a different origin so you can't drawImage/toDataURL it, so maybe that handles the problems (but I should probably leave security issues to people who understand them :-) )
14:58
<annevk>
I suppose data: URIs might be "safe" then
14:58
<annevk>
as long as you don't use data:image/svg+xml, or something
15:41
<zcorpan>
btw, the presentation went well yesterday
16:02
<gsnedders>
exams over! w00t!
16:05
<Philip`>
Just until next year? :-)
16:07
<zcorpan>
html5.org seems to be down :(
16:09
<hasather>
zcorpan: hmm, that happens fairly often
16:10
<MikeSmith>
zcorpan - presentation?
16:10
<zcorpan>
MikeSmith: yeah, http://www.robertnyman.com/2007/04/26/geek-meet-may-2007-html-5-and-xhtml/
16:12
<hasather>
zcorpan: probably because of this: http://www.dreamhoststatus.com/2007/05/23/more-dns-issues/
16:21
<zcorpan>
hasather: ok
16:32
<annevk>
zcorpan, what questions did you get?
16:33
<annevk>
zcorpan, and what about the other presentation?
16:35
<zcorpan>
annevk: mostly about document conformance things (like <font>), versioning (and doctypes), if html5lib could parse tag soup, whether the presence/absence of "/" in void elements need to consistent...
16:35
<zcorpan>
when html5 will be ready/used/implemented
16:36
<annevk>
it's used and implemented... ready? well...
16:36
<annevk>
:)
16:36
<zcorpan>
jarvklo talked about xhtml and the mobile web, mostly
16:37
<zcorpan>
he had some good ideas about what can be done server side, like checking conformance without using a third party
16:37
<annevk>
how XHTML is fucked up there?
16:38
<zcorpan>
not really
16:39
<zcorpan>
more that the w3c and wap forum (or what it is) have finally come to an agreement about what markup to use for mobiles
16:39
<annevk>
oh that
16:40
<annevk>
I think vendors have mostly decided that HTML is to be used for mobiles
16:40
<annevk>
(Vendors being Apple, Nokia and Opera here...)
16:40
<zcorpan>
ok
16:40
<zcorpan>
i'd encourage you to talk to jarklo directly though
16:41
<annevk>
the actual news from "wapforum" is that they're no longer going to subset and then superset W3C specs
16:41
<annevk>
like they did with WML2 and XHTML Mobile Profile
16:41
<zcorpan>
indeed
16:41
<annevk>
not that XHTML is now recommended or something
16:42
<zcorpan>
i thought it was
16:43
<annevk>
yeah, it probably still is
16:43
<zcorpan>
on what grounds i don't know, really
16:44
<annevk>
"latest, greatest"
16:44
<annevk>
Anyone know how to escape !- inside a Linux terminal?
16:45
<othermaciej>
\?
16:45
<annevk>
I'm trying to test comments but doing 'python test.py "<!-- -->"' doesn't work
16:45
<othermaciej>
backslash or single quote
16:46
<annevk>
I tried backslash, didn't work
16:46
<annevk>
Single quote doesn't seem to work either...
16:46
<othermaciej>
what shell are you using?
16:47
<annevk>
Ah wait, backslash does work
16:47
<annevk>
The problem is that the backslash ends up in my Python script...
16:47
<othermaciej>
echo "<"\!"-- -->"
16:47
<annevk>
I'm using "GNOME Terminal 2.x"
16:47
<othermaciej>
does what you'd want in bash
16:48
<othermaciej>
the terminal is not the shell
16:48
<annevk>
ah ok
16:48
<annevk>
that does work
16:48
<annevk>
thanks
16:58
<annevk>
twitter is funny... it escapes your text input and then counts the number of characters and complaints about it being over 140...
16:59
<Dashiva>
Even better with forum software showing title previews, first 10 chars or so, after escaping
17:00
<Dashiva>
Here we &nb... ->
17:00
<annevk>
the thing is that the client side check does count the visual characters and it therefore does get submitted and ends up in their database
17:15
<annevk>
whatwg is down
17:16
<hasather>
annevk: DH is having DNS issues
17:21
<annevk>
k
17:36
<zcorpan>
html5.org is up again
17:38
<zcorpan>
perhaps i should blog about the presentation at blog.w.o
17:39
<annevk>
you have a blog now? :)
17:39
<zcorpan>
blog.whatwg.org
17:41
<annevk>
heh
18:16
gsnedders
has been given the best reason _ever_ for using <em> for italics even though they've been told better — "we prefer it".
18:19
<Hixie>
if anyone can work out the code ni http://img.shopping.com/jfe/JavaFrontEnd-fe55.1.1-874/js/tabMenus.js gets called in a way that actually does something, i'll give them 500 points.
18:19
<Hixie>
a sample page that gets that js file is http://www1.uk.shopping.com/xPP-hobs--neff-price_range_630_800
19:02
<Philip`>
http://66.102.9.104/search?q=cache:4IXIf9DhazkJ:zhulianxing.blogspot.com/feeds/posts/default/7688979022477086776 - back in February it had <a ... onmouseover="startShow(&#39;1&#39;,&#39;6&#39;);setMenusSize(&#39;8&#39;);" onmouseout="sMH();hIfr();" ...>
19:04
<Philip`>
Uh, March
19:04
<Philip`>
(probably)
19:06
<Philip`>
(I really don't know why http://zhulianxing.blogspot.com/ posted shopping.com's front page on their blog...)
19:07
<Hixie>
aah
19:07
<Hixie>
interesting
19:12
<bewest>
what will 500 points get me?
19:12
<nickshanks>
ian, did you see my mailing list suggestion? is it workable?
19:13
<Hixie>
bewest: dunno :-)
19:13
<Hixie>
nickshanks: which one?
19:13
<zcorpan>
http://blog.whatwg.org/html5-geekmeet
19:14
<Hixie>
bewest: thanks, though, that's very helpful
19:14
<nickshanks>
ian: one document or two
19:17
<Philip`>
"Does that mean that you can convert any old HTML document to XML by feeding it through html5lib?" - not really, since it seems a significant number produce DOMs that can't be serialised to well-formed XML
19:17
<Hixie>
nickshanks: when did you send it?
19:17
<Hixie>
oh i see it
19:17
<nickshanks>
about 20 mins ago. i can't receive emails now, so haven't seen if anyone replied
19:17
<Hixie>
haven't gotten to that yet in my e-mail
19:17
<Hixie>
:-)
19:17
<zcorpan>
Philip`: yeah, true
19:17
<Hixie>
there's a reply
19:17
<nickshanks>
but since you popped up in #webkit, i thought i'd ask :)
19:18
<zcorpan>
Philip`: you can probably convert it to XML5 however ;)
19:18
<Philip`>
(Actually, I'm not sure how significant a number - at least there are some with <!--------> which I think can't be done, and some with invalid characters in attribute names, but I never looked at what the actual problems were and if they could be worked around easily)
19:19
<bewest>
I don't think that javascript is used on that page
19:19
<bewest>
Hixie: it looks like legacy stuff that someone new is trying to clean up
19:19
<bewest>
Hixie: looks to me like thereis more than one author involved in the javascript across the site
19:20
<Hixie>
seems likely
19:20
<Hixie>
oh well
19:20
<bewest>
sometimes new people may not like javascript that goes out onthe site... but they can't necessarily just get rid of it because it's not clear what might break
19:22
<Philip`>
shopping.com on archive.org shows that the tabMenus thing wasn't there six months ago, so it's a new addition (and then half-removal)
19:22
<Hixie>
fun
19:22
<Philip`>
http://wordsandpictures.dyndns.org/cgi-bin/parsetree/parsetree.py?source=%3Cspacer+type%22block%22+width%3D%221%22+height%3D%221%22%3E%3C%2Fspacer%3E - aha, that's one I found that couldn't be serialised to XML
19:23
<Philip`>
Can XML5 do that? :-)
19:23
<Hixie>
hm, the tag wants to talk to me
19:23
<Hixie>
that can't bode well
19:25
<Philip`>
Hallucinating about speaking HTML tags?
19:26
<Hixie>
no, the w3c tag
19:26
<Hixie>
apparently there is "significant interest" in me joining them for lunch next week
19:26
<Dashiva>
Bring a food taster
19:27
<Hixie>
it'll be in a google cafeteria, so i think i'm safe
19:28
<Dashiva>
Polonium or whatnot, I tell you
19:29
<othermaciej>
Hixie: perhaps they want you to say hello to their little friend
19:35
<nickshanks>
Dashiva: that's the soviets silly
19:35
<nickshanks>
soogle isn't microsoft
19:35
<nickshanks>
-s +g
19:35
<Hixie>
nickshanks: replied
19:35
<nickshanks>
what did you say? :)
19:36
<nickshanks>
(i can't get email atm.)
19:36
<Hixie>
you said that an option would be to "simply have two views of the spec"
19:36
<Hixie>
i said that it was from "simple"
22:32
<jgraham>
html5lib now has BeautifulSoup support (at last. I really made a meal of implementing it)
22:32
<Hixie>
what's BeautifulSoup?
22:33
<jgraham>
http://www.crummy.com/software/BeautifulSoup/
22:33
<jgraham>
(we just build a tree in that format, the BeautifulSoup parser is irrelevant, obviously)
22:34
<Hixie>
ah ok
22:34
<jgraham>
Simon Willison asked for it at XTech so he could use his getElementsByCSSSelector function with html5lib
22:39
<Hixie>
ah
22:50
<zcorpan>
is there a reason why http://wiki.whatwg.org/wiki/HTML5_Presentations doesn't link to hsivonen's slides?
22:52
<hasather>
annevk: oh, I'll update it now
22:55
zcorpan
adds a link, if it shouldn't be there then revert it... :)
23:01
<nickshanks>
while you're att it, you can translate the page to swedish ;-)
23:06
<zcorpan>
lol
23:16
<zcorpan>
hsivonen: the expansions of XSD and XHTML5 in your thesis say "XML" in <abbr title> in your thesis
23:38
<mpt>
The perils of invisible metadata
23:41
<Dashiva>
I wouldn't say title is invisible
23:41
<jruderman>
just hiding
23:44
<mpt>
well, yeah
23:44
<mpt>
Semi-visible
23:44
<mpt>
like <title> in most browsers
23:44
<mpt>
hence "Welcome to Adobe GoLive 5" etc
23:46
<nickshanks>
if title was in the body, would that be a Good Thing ?
23:46
<nickshanks>
(assuming it had been that way since HTML 1)
23:46
<Hixie>
it's not where in the markup that matters in this case
23:46
<Hixie>
since the element is by definition not rendered in the main page
23:47
<Hixie>
and the main page title (H1) is often not an appropriate <title>
23:49
<mpt>
I think probably the only way to fix that problem is for browsers to not have toolbars at the top of the window
23:50
<Hixie>
we kind of have that now with the tab bars
23:50
<nickshanks>
well, titles go in the tab bar, and that's only 20 chars wide or so
23:50
<mpt>
|That's not reall...|
23:50
<Hixie>
i don't see the toolbar going away
23:50
<nickshanks>
tabs are not real? what! waaaaaa
23:51
nickshanks
holds his head in his hands*
23:51
<mpt>
-ly a solution :-)