00:01
<Philip`>
"wow. fascinating. depressing." - that's an appropriate response to pretty much any attempt to look at a random collection of real-world markup
00:02
<Dashiva>
Or humanity in general :)
00:02
<Philip`>
I hope nobody is going to notice that there's some weird euc-jp decoding issues on my table-summary-values-dotbot page where it inexplicably misses the '"' at the end of an attribute value
00:03
<Philip`>
(I assume it's an euc-jp decoding issue because all the offending values look kind of Japanese or something foreign like that)
00:03
<Philip`>
(I can't reproduce the problem with parsetree.validator.nu but I don't care enough to debug it further)
00:04
<Dashiva>
Shouldn't the URL show whether it's Japanese or not?
00:04
<Philip`>
I guess so
00:04
<Philip`>
but I didn't look very closely
00:05
<Philip`>
and I'm not going to look now because it's a 5MB page and I have to pay the equivalent of half a chocolate digestive biscuit per megabyte of traffic
00:06
<Dashiva>
Well, I just clicked the expand link in your summary table :)
00:07
<Philip`>
You could look for some of the values lower down that have strings like "<tr>" in them
00:08
<Philip`>
and see if those are .jp pages
00:08
<Dashiva>
Yup
00:08
<Hixie>
Philip` needs a better network service provider
00:08
<Hixie>
one that charges a flat rate
00:08
<Lachy>
I'm surprised by how well Shelly wrote those emails about summary. Is she the same Shelly who's been complaining about RDFa stuff?
00:08
<Hixie>
and not per bit
00:09
<Dashiva>
Maybe they charge per byte, so it still costs the same if he only downloads 7 bits
00:09
<Philip`>
Hixie: I have one of those, but the wireless router appears to be dead, so I'm using a mobile broadband thingamy instead
00:10
<Hixie>
Philip`: fix the router :-P
00:10
<Dashiva>
Philip`: I checked a few different ones, they all contain Japanese sites. I can't say every site listed is Japanese, but it seems likely.
00:10
<Philip`>
Hixie: I can't fix it myself - all I can do is complain to the owners, and they're asleep now :-p
00:11
<Lachy>
although, I really hate it when people claim that "... the folks who really want this attribute gone, really have no alternative solution in mind." because that's just complete nonsense.
00:11
<Hixie>
Philip`: ah
00:12
Philip`
found that most "flat rate" mobile broadband thingamy plans had a limit of 3GB per month, and cost £15/month, which isn't so good
00:13
<Philip`>
(so I got one that's £15/GB and doesn't expire for 6 months, which is better value when I don't use it much)
00:13
<takkaria>
Philip`: hm, is that 3G USB stick-type stuff? who did you get yours from?
00:14
<Philip`>
takkaria: It is
00:14
<Dashiva>
Bah. Why is it so hard to get that stuff like autoplay are author hints, the UA is perfectly capable of overriding.
00:14
<Philip`>
takkaria: It's a Vodafone one
00:15
<takkaria>
Philip`: ah cool, sounds like what I want really
00:16
<Philip`>
takkaria: (Also it works on Linux, with a couple of configuration files found via Google)
00:30
<jjlee>
what's the best way to get character indices of tags out of html5lib?
00:31
<jjlee>
I had a quick look a few weeks back, and suspected that copy and pasting chunks of html5lib into my code was likely my best option
00:31
<jjlee>
any other ideas?
01:20
<ezyang>
So, is the dash-dash-bang-angle comment terminator change solid now?
01:20
<ezyang>
(If it is, we should add tests)
01:22
<Hixie>
i think so
01:23
<ezyang>
Great!
04:33
<Hixie>
uh oh
04:33
<Hixie>
i've reached microdata e-mails
04:34
<Hixie>
time to go home i think!
04:35
<Dashiva>
But it's not even midnight yet!
05:21
<Hixie>
http://digg.com/comedy/W3C_s_Response_to_the_sarcasm_Tag_in_HTML_5_PIC is sad
05:22
<Hixie>
first of all, whoever is behind the w3c twitter account clearly doesn't know html5 well
05:23
<Hixie>
secondly, clearly the people commenting there don't either :-P
05:27
<ezyang>
Yaya
05:27
<ezyang>
But you should take a deep breath
05:28
<jruderman>
you disagree with the reply from the w3c account?
12:44
<zcorpan>
Hixie: maybe <hgroup> without <h1>-<h6> descendant should be ignored altogether instead of being equivalent to <h1>
15:54
gsnedders
wants resizable text fields in all browsers…
15:54
gsnedders
grumbles
17:25
<Philip`>
<li><a href...>Menu item <!--[if IE 7]><!--></a><!--<![endif]--> <!--[if lte IE 6]><table><tr><td><![endif]--> <ul> Sub-menu items... </ul> <!--[if lte IE 6]></td></tr></table></a><![endif]--> </li>
17:25
<Philip`>
Conditional comments let people do scary things
17:52
<krijnh>
Philip`: almost every example on http://www.cssplay.co.uk/ uses that
17:54
<krijnh>
I think it started with http://www.cssplay.co.uk/menus/dd_valid_2.html
18:02
<gsnedders>
Philip`: WTF?
18:26
jgraham
wonders if he should clean up the table summary long terms solution page so that it isn't so biased
18:27
<jgraham>
Or failing that at least so that it makes sense
18:28
<jgraham>
And actually has the proposal I made rather than a worse variation on it
20:47
<gsnedders>
Is there anyway to get a a Document object from an HTML string?
20:47
<gsnedders>
*a string containing HTML
20:48
<Philip`>
document.write into an iframe?
20:49
<gsnedders>
how do I do it in an iframe?
20:50
<gsnedders>
I mean, the iframe will already have html/head/body…
20:52
<Philip`>
Do what the Live DOM Viewer does, I guess
20:52
<gsnedders>
What does that do?
20:52
<gsnedders>
:P
20:55
<Philip`>
document.writes into an iframe, I think
20:56
<Philip`>
(after calling document.open)
20:57
gsnedders
always gets confused when he looks at the Live DOM Viewer's source :P
21:16
<webben>
(OT): http://www.w3.org/TR/CSS2/syndata.html#charset Anybody know what CSS2.1 means by "byte" here?
21:17
<webben>
I'm trying to implement CSS2.1 character encoding sniffing, and I'm just looking at the character encodings that involve transcoding.
21:17
<webben>
Starting with GSM 03.38.
21:17
gsnedders
notes this would be easier if he hadn't just crashed his web browser
21:19
<webben>
What's confusing me is that whereas you can grab octets for (say) UTF-8 bytes (i.e. octets) and match against their regexes, GSM 03.38 takes packs 7-bit codepoints into octets.
21:19
<webben>
so if you just naively grab the octets, they won't match that regex
21:19
<gsnedders>
Well, if it is codepoints then doesn't that give you the Unicode string?
21:19
<webben>
you actually have to decompose them to septets.
21:20
<webben>
gsnedders: I'm using codepoints loosely. These aren't unicode codepoints.
21:20
gsnedders
thinks this would be easier to answer if he actually knew what GSM did :)
21:21
<webben>
There the abstract hex values in the GSM 03.38 character encoding, which can be represented into 7 bits, which are then packed into octets such that one octet contains bits from multiple codepoints.
21:21
<webben>
http://pda.etsi.org/exchangefolder/ts_100900v070200p.pdf is the GSM 03.38 spec.
21:22
<webben>
and http://www.dreamfabric.com/sms/ is a English-language description of what it does.
21:22
<webben>
I've been wrestingly with this in increasing confusion all day; figured there's quite a lot of spec-compliance and character encoding knowledge in this channel ;)
21:23
gsnedders
would ignore that table in CSS21 :)
21:23
<gsnedders>
Other things make more sense for GSM 03.38, IMO.
21:23
<gsnedders>
Or maybe I'm wrong
21:23
gsnedders
shrugs
21:23
<gsnedders>
Don't ask me!
21:23
<webben>
I've also been wondering: who delivers CSS in GSM 03.38? I can't find any reference to this anywhere else.
21:24
webben
is also amused that i18n's test cases for CSS and charset don't actually cover all these random encodings.
21:24
<webben>
(http://www.w3.org/International/tests/list-html-css#charset)
21:30
Philip`
discovers that if you don't have setuptools installed, Trac will silently choose not to attempt to load any plugins
21:48
<virtuelv>
GSM 03.38 brings back bad memories
23:17
<ezyang>
Hixie, you update the tree-construction and tokenization parts of the spec too much :-(
23:22
<ezyang>
In other news, it would be totally awesome if the diffs would actually tell you what section you were, instead of some bogus function.
23:23
<gsnedders>
Uh, yeah. We discussed this at TPAC. If someone writes the code to do this :P
23:23
<ezyang>
We could cheeze it by adding <!-- function section1header4 { -->
23:53
<gsnedders>
ezyang: We're above that :P