01:01
<Hixie>
any opera people here?
01:03
<karlUshi>
which nationality?
01:03
<karlUshi>
http://en.wikipedia.org/wiki/Category:Opera_singers_by_nationality
01:04
<Hixie>
...
01:04
<Hixie>
any opera software employees here?
01:12
<Dashiva>
Hixie: Temporarily, at least
01:13
<Hixie>
hey
01:14
<Hixie>
do you know what the offical Opera opinion is of moving postMessage from Document to Window?
01:14
<Hixie>
is it something you guys can do without breaking too much content? would you have to support both for a while? is that ok?
01:17
<Dashiva>
Let's see...
01:18
<Dashiva>
Last I heard it was "window might be nicer, but not breaking the API is probably better"
01:20
<Dashiva>
You'll have to poke anne (I think) for anything official
01:20
<Hixie>
ok
01:20
<Hixie>
thanks
01:20
<Hixie>
i think i'm gonna break the API, in case you can poke people internally about this to make them aware of it
01:22
<Dashiva>
I'll see what I can do. Not sure what the official Opera policy on shooting messengers is ;)
01:22
<Hixie>
:-)
02:01
<Hixie>
<!DOCTYPE html> a<ruby> b <rt>rt1</rt> <rt>rt2</rt> </ruby>c
02:01
<Hixie>
and
02:01
<Hixie>
<!DOCTYPE html> a<ruby> b <rt>rt1</rt><rt>rt2</rt> </ruby>c
02:01
<Hixie>
render very differently in IE
02:01
<Hixie>
grr
02:11
<karlUshi>
judt because of the space?
02:12
<karlUshi>
I think it would be a good opportunity to include Richard Ishida in the discussion. He's on the list.
02:14
<Hixie>
it's not really a ruby problem, ironically
02:14
<Hixie>
it's an IE parsing problem
02:15
<Hixie>
knowledge of ironically ruby doesn't really help with speccing what to do with parsing ruby :-(
02:15
<Hixie>
er
02:15
<Hixie>
knowledge of ruby ironically doesn't really help, even
02:18
<Hixie>
i wonder if we should make <ruby> a formatting element
02:19
<Hixie>
it's the only way i can see to handle <ruby>a<rt>b<p>c</ruby>d
02:19
<Hixie>
in a way that resembles IE
02:19
<Hixie>
though maybe the solution is to not resemble IE
02:44
<Lachy>
http://www.ibm.com/developerworks/library/x-html5/?ca=dgr-lnxw01NewHTML
02:44
<Lachy>
http://developers.slashdot.org/article.pl?sid=07/08/08/1558225
03:14
<karlUshi>
first comment "And here I was thinking that solved all of my web design problems. Now I might have to learn a second type of tag!"
03:14
<karlUshi>
:)
03:29
<deltab>
that comment as intended: "And here I was thinking that <pre> solved all of my web design problems. Now I might have to learn a second type of tag!"
06:29
<Hixie>
http://junkyard.damowmow.com/287 comes from the new study
06:29
<Hixie>
(about usemap)
06:31
<Hixie>
i think we can throw it out safely
06:32
<Hixie>
but if someone wants to look at how those pages use usemap="" on <input> to see if any of them use it in a way that matters, that'd be great
08:11
<Lachy>
Hixie, the first 6 results from that study don't use the map for anything useful at all. I don't expect to find any that do if I keep looking
08:20
<Hixie>
all the pages have an <input> with type=image with a usemap="" that points to a <map> with a name or ID that matches
08:20
<Hixie>
i didn't check if the <map> had <area> elements
08:21
<Hixie>
lachy: in what way were they useless?
08:22
<Hixie>
hm, gotta go (i'm amused that people randomly twitter single words on whatwg.org btw)
08:22
<Lachy>
some had no <area> elements, some had them without href attributes and others had <area ... href=""> or <area ... href="#">
08:39
<annevk>
Hixie, we don't want to break the API I think
08:40
<annevk>
Hixie, there was this bug report you commented on relevant to that
11:08
<zcorpan>
http://simon.html5.org/test/html/semantics/image-maps/ 003..004 confirms that id/name with usemap is unicode case insensitive
11:09
<zcorpan>
in all 4 browsers
11:09
<zcorpan>
(except in xhtml in fx and saf, where it is case sensitive)
11:11
<zcorpan>
Hixie: ^
17:24
<zcorpan>
Lachy: is http://lachy.id.au/dev/presentation/developing-with-html5/ announced anywhere? can it be added to the wiki?
17:59
<annevk>
http://www.prochoiceforum.org.uk/comm47.asp is another page which is worse in browsers that support <input usemap>
18:00
<Hixie>
hmm
18:01
<Hixie>
guess we'll remove it
18:01
<annevk>
http://www.soe.uwm.edu/pages/welcome/Certification_and_Degrees too
18:01
<annevk>
(if you click advanced search you go to the normal search page except that it doesn't submit any data you typed in)
18:02
<annevk>
although I think that site actually had the intention of making it useful
18:03
<annevk>
http://www.universaldizayn.com/irtibat.php login breaks in Opera and Firefox
18:04
<Hixie>
i'm gonna do the study again but only get pages that have <map>s with <area>s
18:04
<Hixie>
which themselves have both coords="" and href="" and their href="" isn't "" or "#"
18:04
<annevk>
these all had <area>
18:04
<annevk>
oh
18:04
<Hixie>
anything else i should filter out?
18:05
<annevk>
I just looked at those three so far
18:05
<annevk>
I already filed a bug for removal...
18:05
<Hixie>
yeah i'm pretty sure we want to remove it
18:05
<Hixie>
i just want to make sure
18:16
<annevk>
http://www.ljagri.gov.cn/nykj_syjs.asp also has the # problem
18:17
<Hixie>
yeah that one wouldn't be caught in the new survey i just launched
18:20
<annevk>
http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C%21DOCTYPE%20html%3E%0A%3Ciframe%20src%3Dhttp%3A//www.opera.com/%3E%3C/iframe%3E%0Axx%3Cscript%3E%20document.getElementsByTagName%28%22iframe%22%29%5B0%5D.contentDocument.location%20%3D%20%22http%3A//google.com%22%20%3C/script%3E
18:20
<annevk>
is one reason why postMessage might just as well stay on Document
18:20
<annevk>
oh, interesting, it only seems to work in Firefox
18:20
annevk
ponders
18:21
<Hixie>
yeah document shouldn't be accessible cross-origin
18:21
<annevk>
I thought they were the ones having issues with this document being accessible?
18:21
<annevk>
We'd like to keep postMessage() as is, if possible...
18:21
<Hixie>
no they were the ones who said that they had too many security problems
18:21
<Hixie>
oh?
18:21
<Hixie>
why?
18:22
<annevk>
see pm with link to internal bug report
18:22
<annevk>
basically, jl doesn't really see a good reason to break the API
18:25
<annevk>
btw, would be nice if Live DOM Viewer had something like $(id) and $$(tagname) in the global scope or something similarly named
18:41
<annevk>
oops, made a mistake
18:41
<annevk>
http://software.hixie.ch/utilities/js/live-dom-viewer/?%3Ciframe%20src%3Dhttp%3A//www.opera.com/%3E%3C/iframe%3E%3Cscript%3Eonload%3Dfunction%28%29%7Bdocument.getElementsByTagName%28%22iframe%22%29%5B0%5D.contentDocument.location%20%3D%20%22http%3A//google.com/%22%7D%3C/script%3E%20
18:41
<annevk>
does make the redirect in Opera
18:43
<annevk>
lol, that variant throws an exception in Firefox
18:44
<zcorpan>
ie7 as well (when you change contentDocument to contentWindow.document)
18:46
<Lachy>
zcorpan: sure, you can add it to the wiki. I will be blogging about it shortly too
18:56
Hixie
slowly builds the most elaborate cross-domain attempt he has ever made
18:57
<zcorpan>
Lachy: done
19:03
<Hixie>
http://software.hixie.ch/utilities/js/live-dom-viewer/?<!DOCTYPE html><body onload%3D"setTimeout(test1%2C 100)%3B setTimeout(test2%2C 200)%3B">...<iframe src%3D"http%3A//labs.google.com"></iframe><script>function test1() {w('contentDocument%3A')%3Bvar y %3D document.getElementsByTagName('iframe')[0]%3Bw(y)%3Bw(y.contentDocument)%3Bw(y.contentDocument.write)%3By.contentDocument.write('test')%3B}</script><script>function test2() {w('contentWindow%3A')%3Bvar x %3D do
19:03
<Hixie>
(did that truncate?)
19:03
<Hixie>
anyway my conclusion is that only Opera lets you iterate over the Document, and only Firefox lets you access document.write()
19:06
<Hixie>
actually i can't even call document.write() in firefox
19:08
<Hixie>
wait that wasn't the right test
19:10
<Hixie>
annevk: i commented on the bug
19:10
<Hixie>
http://junkyard.damowmow.com/288 is my "testcase"
19:25
<Hixie>
i can't work out how to make <ruby> work
19:26
<Hixie>
specifically, the parsing of <ruby> in IE is not compatible with a non-tree DOM
19:26
<Hixie>
i wonder how much ruby is out there
19:27
<Hixie>
hm, not much
19:27
<Hixie>
it's not in the top 200
20:09
<Hixie>
woot, my first official google blog post - http://google-code-updates.blogspot.com/2007/08/optimisation-data-for-html5-parser.html
20:09
<Hixie>
(also on code.google.com)
21:16
Hixie
launches a study of <ruby> parsing
21:17
<Hixie>
i don't understand how people wrote specs before
21:18
<hsivonen>
Hixie: have you taken a look at the ODF spec?
21:18
<Hixie>
no
21:18
<Hixie>
should i?
21:18
<hsivonen>
well, it certainly is very different from HTML 5.
21:18
<Hixie>
heh
21:19
Hixie
starts looking at adding URIs and titles to pushState()
21:19
<hsivonen>
it is basically a commentary on the schema
21:19
<Hixie>
eh, that's a pretty common way of writing specs
21:19
<Hixie>
HTML4 is pretty much that
21:20
<hsivonen>
not on that level
21:20
<Hixie>
scary
21:22
<hsivonen>
anyway, as far as spec writing goes, this processing model thing as well as error handling seem to be novel :-/
21:23
<hsivonen>
OOXML has detail, but more like a DoS of detail. I haven't figured out yet, if it is useful and sufficient detail
21:23
<Hixie>
but how do you write a "version 2" without being able to study the existing content and examine the cowpaths?
21:25
<hsivonen>
you take the next version of MS Office and figure out what new data structures there are? :-)
21:25
<Hixie>
on another note, i have this small slice of the web that i use for testing my scripts before i run them on the real data, and every time i do a test run over that data my parser skips over these same pages, which it outputs the url of to the console
21:25
<Hixie>
one of them is http://www.online-dating-facts.net/
21:26
<Hixie>
which has so many copies of "<b>Warning</b>: feof(): supplied argument is not a valid stream resource in <b>/home/online/public_html/inc/articles.inc</b> on line <b>4</b><br />" that my script runs out of bits to store elements in
21:27
<Hixie>
it may in fact be an infinite page, i can't quite tell
21:27
<Hixie>
i don't think i've ever reached the end of it
21:27
<Hixie>
(my test data file has a truncated copy of it)
21:29
<Hixie>
anyone have an opinion on whether pushState() should always require a URI or whether we should continue to allow state to be included without a URI?
21:45
<KevinMarks>
Afternoon
21:46
<KevinMarks>
having a bit fo trouble following the rel="bookmark" stuff
21:46
<KevinMarks>
http://www.whatwg.org/specs/web-apps/current-work/#link-type2
21:47
<Hixie>
what don't you follow?
21:50
<KevinMarks>
the implications of sections - if there are no <article> elements, the rel="bookmark" only applies to a subset fo the document?
21:51
<KevinMarks>
or can the bookmark link apply to the whole page?
21:52
<KevinMarks>
concrete example, go to google maps, and do a search. The 'link to this page' url has rel="bookmark" on it, which is intended to mean 'use this for the page permalink, not the top url as the content changed"
21:52
<Hixie>
it applies to the nearest <article>, or if there are no such ancestors, to the nearest <body>, <section>, <nav>, <blockquote>, or <aside> element
21:53
<Hixie>
whichever comes first
21:55
<KevinMarks>
that 'section' discovery is tricky
21:55
<Hixie>
you mean working out which element is a section?
21:57
<KevinMarks>
what seems 'natural' to me is to flood-fill outwards from each rel="bookmark" link until you hit another's boundary
21:58
<KevinMarks>
is the intent of that code to not 'leak' out from the subsections to the filler?
21:59
<Hixie>
i don't recall what the thinking was
21:59
<Hixie>
i think it was to make it possible to mark blog posts primarily
21:59
<KevinMarks>
right
21:59
<Hixie>
and then when you're not doing a post, to mark sections and stuff
22:00
<Hixie>
but maybe we should drop that altogether
22:00
<Hixie>
and make it the nearest <article> or <body>
22:00
<KevinMarks>
no, broadly it makes sense
22:00
<Hixie>
i'm not sure i really see the point of permalinking to a section really
22:00
<Hixie>
i recommend sending feedback to the list (whatwg⊙wo) or me (ian⊙hc) so i look at it more closely, if you want it changed at all
22:00
<KevinMarks>
it is a huge usecase in blogs
22:01
<Hixie>
certainly <article> should definitely scope permalinks
22:01
<Hixie>
i'm talking mainly about <section> et al
22:01
<KevinMarks>
it's the independence of the sction parsing and the bookmark linking that confuses me a bit
22:01
<Hixie>
i don't see a real use case to scope to them
22:01
<Hixie>
i don't understand what you mean by "section parsing"
22:02
<KevinMarks>
http://www.whatwg.org/specs/web-apps/current-work/#outlines
22:02
<Hixie>
oh outlining
22:03
<KevinMarks>
wondering if anyone has implemented that yet - is that part of Sam's html5lib?
22:03
<Hixie>
i don't know of any implementation work short of the small script in the spec itself
22:04
<KevinMarks>
OK, thanks
23:20
<Hixie>
preliminary results suggest we can get away with making <ruby> parse like phrasing elements! sweet
23:20
<Hixie>
let's hope it's representative
23:29
<Hixie>
Ok I can only find about 500 sites (about 12000 pages) that use usemap="" on <input> in a way involving <area> elements that have actual href=""s with possibly useful values
23:43
<Hixie>
continuing the usemap=" saga -- http://junkyard.damowmow.com/289 is the new survey data
23:43
<Hixie>
in case people want to look at them