00:03
<Lachy>
it's nice when people present problems for which we already have solutions in HTML5 :-) http://meyerweb.com/eric/thoughts/2009/02/12/selector-blocks/
00:03
<Lachy>
scoped stylesheets should work nicely for that use case
00:06
<Hixie>
ojan: the mutation events stuff is all up in the ir
00:06
<Hixie>
air
00:06
<Hixie>
ojan: i don't think it matches firefox
00:06
<Hixie>
ojan: i'm waiting for the dom events spec to mature before looking at them again
00:08
<ojan>
Hixie: got it.
00:16
<jruderman>
scoped stylesheets are coming? sweet
00:17
<jruderman>
i've been wanting that for blog posts
00:27
<Philip`>
Hixie: Am I being stupid in thinking that the logic for opaque headers seems to be the wrong way around?
00:27
<Hixie>
?
00:28
<Philip`>
If I have a table like <tr><th>A<tr><td>B and look at cell B and do the internal algorithm for scanning and assigning header cells, then in the first iteration at step 9.4 the opaque headers list will be empty ...
00:28
<Philip`>
which means that "there is no cell in the opaque headers list anchored with the same (stuff)"
00:28
<Philip`>
which means blocked is set to true
00:29
<Philip`>
which means the cell (A) isn't added to the headers list
00:29
<Hixie>
http://www.w3.org/Bugs/Public/show_bug.cgi?id=6438 ?
00:30
<Philip`>
Oh
00:30
<Philip`>
Yes
00:30
<Philip`>
I agree with that bug report :-)
00:48
Hixie
wonders what to do about his not having a working anolis anymore
00:50
<heycam>
svn up -rPREV?
00:54
<Hixie>
i don't have the code myself
00:54
<Hixie>
it's on pimpmyspect.net
00:54
<Hixie>
pimpmyspec.net
00:54
<heycam>
pimp my spectacles, elton john style
00:55
<Dashiva>
Pimp my spec still doesn't add bling
01:01
<Hixie>
jgraham: disregard my earlier whining. the three regressions i see on pimpmyspec.net (after fixing it to use the new urls etc) are:
01:01
<Hixie>
jgraham: 1. it's slower
01:01
<Hixie>
jgraham: 2. it doesn't omit as many start tags (see the diff i just committed)
01:02
<Hixie>
jgraham: 3. it doesn't use entity references for characters outside of ascii (i don't trust all the servers involved to do utf-8 correctly)
01:10
<Hixie>
jeez windows sucks
08:27
<zcorpan>
the xhtml media types note has " <script>//<![CDATA[" in an example which is invalid xhtml 1.x
08:54
<jgraham>
Hixie: So, I am here now
08:54
jgraham
is still waiting for the internet fairies to visit his flat
08:59
<jgraham>
The repo for pms.net is http://hg.hoppipolla.co.uk/hgwebdir.cgi/aquarium/ although that has almost no history for reasons best explained as incompetance
09:17
<zcorpan>
hey it's friday 13th today
09:18
<zcorpan>
http://lists.w3.org/Archives/Public/public-xhtml2/2009Feb/0032.html
09:22
<yecril71>
krijnh! Would it be much trouble to turn the channel in the log path to a hyperlink?
09:23
<yecril71>
I think the METER element would be appropriate for specifying price.
09:23
<yecril71>
And INPUT[type=currency], where appropriate.
09:24
<hsivonen>
yecril71: in practice, people don't use a widget for price. <meter> is the gauge widget.
09:27
<annevk5>
Hixie, any particular reason why C in filename had to be lowercase?
09:30
<fearphage>
annevk5: may i pm you an opera-centric question?
09:30
<annevk5>
i suppose
09:31
<fearphage>
don't be so enthusiastic next time :)
09:48
jgraham
hopes Hixie is not trying to gen the spec right now
10:01
<jgraham>
Hixie: OK the ascii thing is fixed. It is now another option :(
10:01
jgraham
hates options
10:02
<jgraham>
I don't know why it should omit fewer start tags
10:03
<jgraham>
It does seem a bit slow
10:04
<jgraham>
annevk5: Do you have the old anolis.py from quuz.org somewhere? I could compare notes with that
10:04
<annevk5>
http://anolis.quuz.org/source is still there
10:05
<annevk5>
comparing notes with code I write is not recommended though
10:06
<jgraham>
annevk5: More to see which options were being passed and whether it is really slower
10:07
<annevk5>
fearphage, I'm generally wary of people asking to ask questions ;)
10:18
<krijnh>
yecril71: what do you mean?
10:26
<jgraham>
gsnedders: I assume you are not around
10:28
<fearphage>
annevk5: people seemed to get offended by people pm'ing them freely at times
10:30
<jgraham>
fearphage: Usually the best solution is to just ask
10:30
<jgraham>
"beg forgiveness not permission"
10:31
<jgraham>
Also there are a number of people here who might be able to help with something Opera related (or s/Opera/$Browser/) so it seems a bit unfair to pick on annevk5
10:36
<fearphage>
never thought of it that way
10:36
<virtuelv>
instead of asking to ask a question, I'd say "just ask it"
10:37
<virtuelv>
fearphage: there is also an #opera channel on freenode, I just discovered
10:40
<yecril71>
Krijnh! The log page header contains the path to the log.
10:40
<krijnh>
Yes
10:41
<yecril71>
It would be awesome if the channel component were a hyperlink.
10:41
<krijnh>
Where to?
10:41
<yecril71>
Just at the main page.
10:41
<yecril71>
Just as at the main page.
10:41
<yecril71>
To IRC.
10:41
<krijnh>
/irc-logs/ has that link already ?
10:41
<annevk5>
you mean on individual archive pages?
10:42
<annevk5>
you're not being very clear
10:42
<yecril71>
The channel names on the main page are hyperlinked.
10:42
<yecril71>
Whereas in log pages they are not.
10:43
<yecril71>
I hope you would not have to pay Amazon for that :-)
10:44
<yecril71>
My remark about METER and price was triggered by a request for a PRICE element.
10:48
<Lachy>
krijnh, if you're going to update the IRC logs, then please add the links to the next and previous days in the logs
10:50
Philip`
is usually happy with just editing the URL to switch dates
10:54
<Lachy>
Philip`, that takes too long
10:55
<Philip`>
Lachy: It takes much less time than reading the page
10:56
<Philip`>
and if you're skipping past lots of pages without reading them, changing the URL is quicker than clicking next/previous lots of times
10:58
<Lachy>
Philip`, the use case I'm trying to address is where the transition from one day to the next occurs in the middle of a conversation, splitting it in half
10:59
<Philip`>
Lachy: Ah, okay
11:08
<jgraham>
Hixie: I think something like http://tinyurl.com/b8zs92 is roughly what PimpMySpec used to do for html5
11:09
<jgraham>
I guess I should run a diff against something to be sure but I don't quite know what
11:10
<krijnh>
Lachy: yeah, and Ben's styling too
11:10
<krijnh>
yecril71: done
11:11
<yecril71>
Thx :-)
11:58
<hsivonen>
Does rubys have a license for his SVG icons somewhere?
12:14
jgraham
doesn't hold out too much hope for other people writing table tests
12:31
<Philip`>
jgraham: Yesterday I was kind of attempting to write an independent implementation of the table stuff, so its output could be compared to yours
12:32
<Philip`>
and I could probably mostly finish that today
12:34
<jgraham>
Philip`: Cool. You can now have an entirely trivial set of testcases that cover almost nothing useful
12:37
<Philip`>
jgraham: The test case format is not entirely helpful to me, since it uses HTML and I don't have an HTML parser easily available
12:37
<Philip`>
I suppose I don't actually have an XML parser or any kind of parser at all, but I presume I could find an XML library fairly easily...
12:37
<Philip`>
(Also using HTML is bad because you can't test <tr> outside of <tbody>)
12:42
<jgraham>
Philip`: My assumption is that you have already parsed the table in accord with the table model in the spec
12:42
<jgraham>
s/have/have a way to/
12:43
<Philip`>
jgraham: I violate your assumption :-)
12:43
<jgraham>
Philip`: This is too often the case
12:44
<Philip`>
but I could write a script to parse the test format into something I can understand, so that's not really a problem
12:44
<Philip`>
but the inability to test <tr> outside <tbody> seems like a real problem, because it makes some of the spec untestable
12:45
<jgraham>
Philip`: My second assumption was that it would not be so hard to do that
12:53
<Philip`>
jgraham: Would it be good for the tests to all include the expected slot assignments, rather than simply the headers?
12:53
<Philip`>
since all that stuff is kind of mixed up together, so maybe it's sensible to test all of it
12:58
<jgraham>
Philip`: My goal was to make it as trivial as possible to write the tests. If you can think of a way to do it whereby both things get tested for little increase in complexity, I'm very much open to it
12:59
<Philip`>
jgraham: I suppose that's an issue
13:00
<zcorpan>
have two formats, one html and one xml, where the xml one is used for the no-tbody tests
13:01
Philip`
might just try generating a load of random test cases and comparing the output from implementations, rather than doing any actual work and writing tests by hand
13:01
<zcorpan>
(and flag the xml with e.g. "#input xml")
13:09
<Philip`>
(XML is also needed for testing that non-table elements get skipped properly, I guess)
13:09
<Philip`>
(Though all this is presumably only relevant for the slot-assignment tests, not the header-assignment tests)
13:35
<zcorpan>
http://lists.w3.org/Archives/Public/public-xhtml2/2009Jan/0010.html
13:36
<annevk42>
he's looking for "mu" but if you're working on XHTML2 it seems pretty tough to reach markup enlightenment
13:48
<zcorpan>
http://lists.w3.org/Archives/Public/public-xhtml2/2009Feb/0000.html - wonder why lang is not more attractive than xml:lang if the desire is to have just one of them in the end
14:22
<annevk42>
http://coolepochcountdown.com/
14:33
<krijnh>
Zeker een coole poch countdown :)
14:37
<Dashiva>
It's such a ripoff of 12:34:56 on 7/8-90
14:39
<Lachy>
JohnResig, yt?
14:40
<Lachy>
JohnResig, webkit is still failing 1 test: "FAIL Inside Element: :target selector". There must be a rounding error in your script that calculates 2162 out of 2163 as 100%
14:41
<JohnResig>
Lachy: you have to access it directly
14:41
<JohnResig>
http://ejohn.org/apps/selectortest/#
14:41
<JohnResig>
oops
14:41
<JohnResig>
http://ejohn.org/apps/selectortest/
14:41
<JohnResig>
Lachy: just hitting reload on it is not sufficient
14:42
<Lachy>
oh, why?
14:44
<JohnResig>
Lachy: *shrug*
14:59
<hsivonen>
now I've filled 45 min with boring history and design principles
14:59
<hsivonen>
now I need to figure out a nicer way to use the other 45 min to talk about new features
14:59
<hsivonen>
just enumerating them is boring
15:00
<hsivonen>
and demos are always a risk when presenting
15:00
<jgraham>
OTOH demos are kind of fun
15:00
<jgraham>
You could prerecord the demos using some screengrab software
15:00
<Dashiva>
Nothing says bleeding edge like a non-working demo
15:01
<hsivonen>
do we have a wiki repo of links to demos so that I don't need to write them?
15:01
<jgraham>
where by "screengrab" I mean "one of those things for recording your screen to video"
15:01
<hsivonen>
should I just use Hixie's demos?
15:02
<annevk42>
oh fun, more namespace confusion; I should hang out more often in the Oslo office
15:03
<annevk42>
Core developer just came asking about a certain testcase that tested that namespaced attributes were not considered the same as attributes not in a namespace...
15:03
<Lachy>
hsivonen, I have some video demos
15:04
<Lachy>
hsivonen, http://lachy.id.au/dev/markup/examples/video/
15:36
<hsivonen>
Lachy: thanks
15:37
<hsivonen>
Lachy: your links to Ogg claim to be links to MPEG-4
15:39
<Lachy>
really? I thought I had fixed that bug before
15:39
<Lachy>
d'oh. Either I forgot to upload the changes, or I never fixed it
15:40
<Lachy>
I will fix it later when I get home
15:46
<Lachy>
wtf? PTW is complaining about the spec being published and not knowing about it, but I distinctly remember it being discussed weeks ago
15:46
<Lachy>
besides, it's just another WD. We shouldn't need to go through too much bureaucracy each time they're published
15:46
<annevk42>
somebody else's problem :)
15:47
<Lachy>
indeed.
15:47
Lachy
errects a Somebody Else's Problem field around it and makes it disappear.
16:36
<Philip`>
jgraham: Is http://james.html5.org/tables/table_inspector.html meant to die with an AttributeError on input like <table><tr><th scope="row">A</table> ?
16:42
<jgraham>
Philip`: Yes.
16:42
<jgraham>
No wait, I mean no.
16:42
<jgraham>
:-p
16:46
<jgraham>
Philip`: Fixed?
16:48
<Philip`>
jgraham: It appears so
16:52
<Philip`>
The HTML5 spec doesn't seem to define algorithms in a way that perfectly matches functional programming languages :-(
16:58
<Philip`>
Hmm, the algorithm for processing rows says "To establish which header cells apply to the current cell element, use the algorithm for assigning header cells described in the next section" which sounds to me like you should run that algorithm immediately to find the headers
16:58
<Philip`>
but that algorithms needs to check if the current cell is part of a row group
16:59
<Philip`>
and row groups are only computed after the algorithm for processing rows has completed
17:00
Philip`
wonders if jgraham avoids that problem by processing the whole table before looking for any headers
17:06
<Philip`>
(...or if I'm just misreading the spec)
17:17
<virtuelv>
hm
17:18
<virtuelv>
http://googlewebmastercentral.blogspot.com/2009/02/specify-your-canonical.html
17:31
<Philip`>
That blog post breaks in Opera :-(
17:34
<wilhelm>
Philip`: How? WFM in the O10 alpha.
17:35
<Philip`>
wilhelm: Unclosed <code> makes the rest of the page code-coloured, in 9.63
17:35
<wilhelm>
Ah, didn't see that.
17:36
<wilhelm>
Still broken.
17:50
<gsnedders>
jgraham: You assumed correctly. I was in another city.
17:51
<gsnedders>
jgraham: For patch the second, change "lxml" to "lxml.html" or some such string. Need to cope with both HTML and XML parsers in lxml
17:52
<gsnedders>
jgraham: Equally, I assume you are not here now.
17:54
gsnedders
only looked for two seconds so he hasn't looked very closely
17:56
<Hixie>
uh oh
17:57
<Hixie>
pms changed again
17:57
<Hixie>
uh
17:57
<Hixie>
it's returning nothing
17:57
<Hixie>
that seems bad
17:58
<jgraham>
Hixie: Fuck. Sorry.
17:58
<jgraham>
I need to go though
17:58
<Hixie>
heh
17:58
<Hixie>
no worries
17:58
<jgraham>
Try quuz.org or whatever annevk42's site is
17:58
<jgraham>
I will fix it tommorrow
17:58
<Hixie>
k
17:59
<jgraham>
Hixie: It rturns something for me btw
18:00
<Hixie>
i'll poke at it
18:00
<Hixie>
no worries
18:00
<gsnedders>
http://www.whatwg.org/specs/web-apps/current-work/ — is that meant to happen?
18:00
<gsnedders>
:P
18:00
<Hixie>
that's the error it now returns
18:01
<gsnedders>
jgraham is awesome.
18:01
<jgraham>
gsnedders: :-p
18:01
<Hixie>
i don't understand the error log though
18:02
<jgraham>
gsnedders: BTW, if you want to email me about the patch for whatever reason it is easier to use my email address @hoppipolla
18:03
<jgraham>
(over the weekend)
18:03
<gsnedders>
Hixie: Re: r2812, is the violation of ECMAScript a result of following UTS22?
18:04
<Hixie>
no, unrelated changes
18:04
<gsnedders>
Hixie: That was unclear from the commit message
18:05
<gsnedders>
Esp. the use of the word "Trade"
18:05
<Hixie>
gsnedders: yeah
18:05
<Hixie>
i know
18:05
gsnedders
slaps Hixie
18:05
<Hixie>
i was having issues with anolis at the time
18:05
<Hixie>
still am :-)
18:05
<Hixie>
can you explain what that message meas?
18:05
<gsnedders>
Hixie: No, you were having with jgraham's web service, not Anolis :)
18:06
<jgraham>
Hixie: Does http://pimpmyspec.net/aquarium.py/output?uri=http%3A%2F%2Fwww.whatwg.org%2Fspecs%2Fweb-apps%2Fcurrent-work%2Fsource&process_toc=on&process_xref=on&process_sub=on&newline_char=LF&tab_char=SPACE&min_depth=2&max_depth=6&w3c_compat_xref_a_placement=on&parser=lxml&serializer=html5lib&inject_meta_charset=on&omit_optional_tags=on&use_best_quote_char=on&encoding=utf-8 work for you? Then does it still work if you use the whatwg version of the sou
18:07
<gsnedders>
jgraham: Did we really have so many W3C compat options on by default before?
18:07
<jgraham>
gsnedders: I took the defaults from the command line version defaults, I thought
18:07
<gsnedders>
By default they are all off
18:07
<jgraham>
They are not what HTML 5 was using
18:07
<gsnedders>
HTML 5 only uses "W3C compatible cross-reference anchor placement"
18:07
<jgraham>
Oh well maybe I am an idiot then
18:08
<jgraham>
Or the js messes something up
18:08
<jgraham>
Which is still my fault
18:08
<annevk42>
for everyone in Europe, it's sort of time to go out right now :)
18:08
<annevk42>
or at least get food and prepare
18:09
jgraham
wonders what annevk42 is preparing for
18:09
<jgraham>
Vogons?
18:09
<gsnedders>
jgraham: Also, you probably want to hide all the html5lib serialization options when the serializer is not html5lib
18:09
<Hixie>
jgraham: those options seem to work. i'll figure out how to get the usual ones to work.
18:10
<jgraham>
gsnedders: They were all disabled before but my code didn't like the change to a dropdown and I thought I should do some real work rather than rewrite it
18:10
<gsnedders>
jgraham: heh
18:10
<gsnedders>
jgraham: Is the code in any sort of repo?
18:11
<jgraham>
gsnedders: http://hg.hoppipolla.co.uk/hgwebdir.cgi/aquarium/
18:11
<jgraham>
But it looks like I need to hg ci]
18:11
<jgraham>
Or move my changes from the live copy over or something
18:11
<jgraham>
which I will do tommorrow
18:12
<jgraham>
Hixie: If you let me know exactly which options are b0rked I will try and fix in the morning
18:12
jgraham
really is going
18:13
<Hixie>
jgraham: thanks
18:15
gsnedders
would hack away at it this weekend if it weren't for looming coursework deadlines
18:16
gsnedders
really ought to have started on more of the coursework earlier in the (academic) year
18:16
<Hixie>
jgraham: the problem was i wasn't including all the arguments and so you had no serialiser/parser picked out and crashed instead
18:17
<gsnedders>
http://pimpmyspec.net/aquarium.py/output?uri=http%3A%2F%2Fwww.whatwg.org%2Fspecs%2Fweb-apps%2Fcurrent-work%2Fsource&w3c_compat_xref_a_placement=on&parser=lxml&serializer=html5lib&inject_meta_charset=off&omit_optional_tags=off&encoding=us-ascii
18:17
<gsnedders>
That's probably closer to what you had before
18:19
<gsnedders>
http://pimpmyspec.net/aquarium.py/output?uri=http%3A%2F%2Fwww.whatwg.org%2Fspecs%2Fweb-apps%2Fcurrent-work%2Fsource&w3c_compat_xref_a_placement=on&parser=lxml&serializer=html5lib&encoding=us-ascii
18:19
<gsnedders>
That's I think what you want.
18:19
<Hixie>
i set parser to html5lib and it crashed again, weird
18:20
gsnedders
thinks jgraham should stick to the astrophysics
18:20
gsnedders
ducks
18:20
<Hixie>
the error log is uniquely usless
18:24
<Hixie>
oh the error log has the error in a comment
18:24
<Hixie>
much more useful
18:24
<Hixie>
UndefinedError: HTTPError() has no member named "reason"
18:26
<Philip`>
I guess it's an unreasonable error
18:33
<Hixie>
it seems using html5lib as the parser is all that is needed to fail
18:38
<Hixie>
that, and the script doesn't handle missing parameters or bogus parameters in a helpful way at all
18:42
<dglazkov>
Hixie!!!
18:42
<dglazkov>
have a sec?
18:44
<Hixie>
sure
18:45
<dglazkov>
so get spammed by the sysbot to update my status
18:46
<dglazkov>
I give up: I want do update my status
18:46
<dglazkov>
it's a very persistent little sysbot
18:47
<Hixie>
?
18:47
<Hixie>
"sysbot"?
18:47
<Hixie>
status?
18:47
<dglazkov>
status in HTML WG as invited expert.
18:47
<Hixie>
ah that has nothing to do with me
18:47
<dglazkov>
excellent
18:47
<dglazkov>
who do I bug?
18:47
<Hixie>
you want DanC in #htmlwg on irc.w3.org port 6665
18:47
<krijnh>
It has got to do with Hixie though, cause he invited us all to the HTML WG party :)
18:47
<Philip`>
#html-wg
18:48
<dglazkov>
very helpful!
18:48
<dglazkov>
thanks
18:49
<Hixie>
er, what Philip` said
18:49
<Hixie>
my bad
19:30
<Hixie>
well you gotta give the bespin guys credit
19:30
<Hixie>
they certainly got canvas to do more than i expected them to
19:43
<inimino>
slick.
19:46
<Philip`>
Bespin guys like Lando?
19:46
<Hixie>
Lando?
19:46
<inimino>
hehe
19:46
<inimino>
Calrissian
19:47
<inimino>
Philip`: https://bespin.mozilla.com/
19:47
gsnedders
is way too tempted to go to sleep
19:48
<Philip`>
inimino: Ah, right
20:22
<annevk42>
Hixie, there is no "simple color" that represents 'transparent', is there?
21:00
<Hixie>
annevk42: crap
21:00
<Hixie>
hm
21:01
<Hixie>
how did zcorpan determine that 'transparent' was actually supported?
21:01
<Hixie>
i can't think of a test case that would show this
23:19
<heycam`>
Hixie, checking for it in CSSOM?
23:28
<Hixie>
heycam`: CSSOM wouldn't expose this as far as i can tell
23:28
<Hixie>
heycam`: since 'transparent' is the default
23:44
<heycam`>
but transparent wouldn't be the default if it weren't supported, would it?
23:47
<Philip`>
Wouldn't it be treated as #00a000 if the keyword wasn't supported, or something like that?
23:48
<Hixie>
oooh, good point
23:48
<Philip`>
Compare http://software.hixie.ch/utilities/js/live-dom-viewer/?%3Cbody%20bgcolor%3Dtransparent%3E vs http://software.hixie.ch/utilities/js/live-dom-viewer/?%3Cbody%20bgcolor%3Dtransparenx%3E
23:48
<Hixie>
indeed
23:48
<Hixie>
ok
23:48
<Philip`>
In IE6, the former looks white, the latter looks #0000e0
23:49
<Hixie>
so can anyone think of a way in which browsers differe in their handling of <body bgcolor=""> (empty) and <body bgcolor="transparent">?
23:51
<Philip`>
http://software.hixie.ch/utilities/js/live-dom-viewer/?%3Cstyle%3Ehtml%7Bbackground%3Ablue%7D%3C%2Fstyle%3E%0D%0A%3Cbody%20bgcolor%3D%22transparent%22%3E
23:51
<Philip`>
In IE6 it looks white
23:51
<Philip`>
(whereas bgcolor="" looks blue)
23:54
<Hixie>
well making it map to body { background: transparent; } isn't going to solve that problem...
23:57
<Philip`>
I see quite a number of pages with bgcolor=transparent
23:58
<Philip`>
mostly on table elements, not body
23:58
<Philip`>
but http://windows.radified.com/ has it on body
23:58
<gsnedders>
Can someone try emailing me⊙gc?