00:03
<annevk5>
I'm not quite sure I follow the logic in that e-mail. A draft cannot change because of lack of implementations because it did not change as a result of receiving comments?
00:06
<zcorpan>
annevk5: the email says there were implementations
00:11
<othermaciej_>
annevk5: the indirection isn't quite the same
00:11
<othermaciej>
in a typical use of namespaces, you see "foo:bar", but that really means the ordered pair {http://something.foo.com/FooML, bar}
00:11
<othermaciej>
and the foo has no meaning
00:11
<annevk5>
The more I read that e-mail the less I get it. Oh well.
00:12
<othermaciej>
but binding CSS to an element with class="foo", there is no "more real" meaning than foo, and you won't be misled by looking for literal foo
00:12
<othermaciej>
for example, if you have a rule foo { color: green; } it's an unlikely mistake to try to find all green text by looking for everything with class foo in any document
00:14
<annevk5>
I agree that the CSS comparison does not quite work.
00:31
<othermaciej>
I think the difference is in CSS, you have an indirection to binding style, whereas XML namespaces introduce indirection to binding meaning
01:23
<annevk5>
http://gawker.com/5258524/ o_O
01:26
<Hixie>
i wonder why some of the classificatiosn are censored
02:13
<annevk5>
fwiw, application/microdata+json will probably not fly with the IETF
02:14
<annevk5>
though I could be mistaken I think s/+/-/ will be easier
02:14
<Hixie>
i'm just trying to be consistent with +xml
02:14
<Hixie>
i'm happy to change it to whatever makes them happy
02:30
<annevk5>
should subject="" be allowed to allow explicit assocation with the top-level microdata items?
02:30
<annevk5>
(it already works that way in user agents)
02:30
<Hixie>
say what?
02:31
<annevk5>
e.g. <span item> <span subject itemprop=name>X</span> </span>
02:32
<Hixie>
subject="" prevents the element from ever having a corresponding item
02:32
<Hixie>
(subject="" literally empty, i mean)
02:34
<annevk5>
"Otherwise, there is no corresponding item." ... "The list of elements that create items but do not themselves have a corresponding item forms the list of top-level microdata items."
02:35
<annevk5>
(what I was trying to point out was that subject="" is currently disallowed)
07:21
<hsivonen>
roc: the reftest failures don't suggest any lack of clue--just lack of cross-browser testing
07:21
<hsivonen>
roc: that is, test is for behavior x but also depends on behavior y where y varies among Gecko and IE or Gecko and WebKit
07:22
<hsivonen>
where behavior y is obscure and counter-intuitive
07:47
<Hixie>
annevk5: oh, i see what you're saying
07:48
<Hixie>
annevk5: i don't really think we should allow that, in fact it's probably better just to call out empty subject="" as an error
10:56
<hsivonen>
is there a way to test the precedence of the MS-maintained IE8 mode blacklist without being able to serve arbitrary content at any of the blacklisted domain?
10:57
<hsivonen>
that is, is there a way to edit the blacklist locally?
10:57
<hsivonen>
(an easy way that is)
10:58
<hsivonen>
I wonder if evaluating a bookmarklet when a given site has loaded inherits its blacklist properties
10:58
<othermaciej>
I bet it would
10:59
<othermaciej>
particularly if you use innerHTML to inject content into the site
10:59
<hsivonen>
othermaciej: innerHTML is no good, because the meta thingy works only if the parser sees it before a script has executed on the page
11:00
<othermaciej>
document.write perhaps?
11:00
<hsivonen>
document.write might work
11:00
<othermaciej>
or inject an iframe
11:00
<othermaciej>
supposedly subframes inherit the blacklist state of the top level document
11:00
<hsivonen>
kurafire claims that frames inherit the IE8/EmulateIE7 setting from parent
11:00
<jgraham>
hsivonen: Does it need to be a test that wworks online? Can't you just use /etc/hosts to point the domain at a site of your choice?
11:01
<jgraham>
Or does the balcklist not work like that?
11:01
<hsivonen>
jgraham: good point
11:01
<hsivonen>
jgraham: thanks
11:05
<Philip`>
You could modify the blacklist file in a hex editor
11:05
<Philip`>
(Probably won't work, though)
11:11
<hsivonen>
I'm appalled at the shady sites that doing a live search for windows hosts file turns up in IE8
11:11
<hsivonen>
I wouldn't get junk like that searching for Linux or Mac things
11:12
<Philip`>
Hmm, the recent update to the IE8 blacklist seems to have changed very little
11:12
<Philip`>
Removed: ca.gov daft.ie nexopia.com onet.pl picnik.com pricegrabber.com rightmove.co.uk yandex.ru
11:12
<Philip`>
Added: sat.gob.mx
11:13
<hsivonen>
google search is much more useful
11:13
<Philip`>
(Those are the changes from a couple of weeks ago, vs the initial release)
11:21
<hsivonen>
OK. I had the MS blacklist in the wrong place in my flowchart
11:22
<hsivonen>
umm. no, I didn't
11:22
<hsivonen>
hmm. either I have tested wrong, kurafire has tested wrong, or I have misunderstood what he said
11:23
<Philip`>
Or it's non-deterministic
11:23
<jgraham>
Or some combination of the above
11:30
<hsivonen>
whee! IE=9.9 and IE=99 are indeed different
11:33
<hsivonen>
oops. my test script is wrong for 9.9
11:34
<hsivonen>
9.9 and 99 are indeed the same
12:27
<hsivonen>
it turns out that the IE8 mode sniffing is even more complex than I previously thought
12:27
<hsivonen>
it takes into account the compatibility modeness of the frame parent
12:28
<hsivonen>
the parent can be in IE5 mode and the parent still remembers how it got to the IE5 mode
12:28
<Rik|work>
hsivonen: http://farukat.es/journal/2009/05/245-ie8-and-the-x-ua-compatible-situation
12:28
<hsivonen>
Rik|work: yes, that's what inspired me to test
12:29
<hsivonen>
Rik|work: my testing shows that X-UA-Compatible still allows you to override the mode from frame parent
12:29
<hsivonen>
for iframes at least
12:30
<Rik|work>
i work in a company that got in the compatibility list, we never received the mail to inform us
12:56
<hsivonen>
whew. I've done my share of community service today and have revised http://hsivonen.iki.fi/doctype/ and linked flowcharts
12:56
<hsivonen>
(several times until I got them right. I think)
13:02
<Philip`>
That is a little bit crazy
13:04
<hsivonen>
Philip`: what's crazy? revising or IE's behavior?
13:04
<hsivonen>
that's two hours of my life I'll never get back
13:05
<Philip`>
IE's behaviour
13:05
<Philip`>
and hence the flowcharts
13:05
<Philip`>
Two hours isn't much when you consider how much time you waste sleeping :-)
13:15
<annevk5>
sleeping is not actually wasting time though
13:16
<jgraham>
Indeed sleeping can often be more enjoyable than the rest of the day was :)
13:32
hsivonen
found a reftest that actually tests Gecko-specific weirdness: http://mxr.mozilla.org/mozilla-central/source/content/html/document/reftests/bug448564-2_malformed.html
13:49
<annevk5>
seems that http://my.opera.com/ODIN/blog/2009/05/19/opera-mama-a-sneak-peek-at-headings-images-and-summary confirms that summary="" indeed sucks on the Web
13:52
<annevk5>
https://twitter.com/robburns1/status/1809470040 "the problem is that the Microsoft/Opera cartel seeks to use HTML5 & WhatWG to undermine the web regardless of W3C endorsement"
13:53
<annevk5>
I guess that goes nicely with the vast Apple/Google conspiracy (they even have shared board members under investigation!!)
13:55
<Philip`>
I didn't think Microsoft and Opera were getting along well enough to join a cartel with each other, what with the lawsuits and suchlike
13:55
<Dashiva>
It's a typo, he meant Opera/Google
13:56
<Philip`>
Isn't Mozilla in on this?
13:56
<Dashiva>
Google owns Mozilla, duh
13:58
<Philip`>
I think Google should buy Opera
13:58
<Philip`>
just to complete the set
13:58
<Philip`>
Well, not complete, but make more complete than it would otherwise be
13:58
<Philip`>
It'd be great if they were paying for development of three competing browsers
14:07
<annevk5>
whoa, Google and Safari get the encoding right for http://thebjoernhoehrmannproject.org/ -- nice
14:13
<Philip`>
Why wouldn't they get it right?
14:14
<roc>
if there's a conspiracy, I don't want to be left out
14:16
<Philip`>
/invite roc #whatwg-secret-treehouse
14:17
<Philip`>
Whoops
14:17
<annevk5>
Philip`, they used the text/html algorithm
16:33
<Philip`>
http://broadcast.oreilly.com/2009/05/the-assertions-in-html-5.html
16:39
<annevk5>
Seems he somewhat misunderstood how things came to be :)
16:41
<Philip`>
I'm not sure that what he says is wrong
16:42
<Philip`>
because the constraints in H:TML are largely based on RELAX NG and (maybe?) Schematron implementations
16:42
<Philip`>
which is what he seems to say
16:43
<annevk5>
Not really: "(I don't know whether the designers of HTML limited themselves to Schematron assertions or to the subset in drafts of XSD 1.1.)"
16:44
<jgraham>
Philip`: But by basing his analysis on those parts of H:TML that have been extracted from schema in validator.nu, he implicitly misses anything that could not be written as a schema constraint
16:45
<jgraham>
(for example the constraint on table cells)
16:46
<Philip`>
annevk5: Oh, I missed that bit
16:46
<jgraham>
I guess it is hard to say his conclusions are wrong since the article just ends with no conclusions
16:47
<Philip`>
jgraham: Seems like a useul thing to point out in a comment :-)
17:49
<gsnedders>
What does (a,b)|c mean? the (a.b) is the greatest common divisor, but what about the |?
17:51
<gsnedders>
divisibility.
18:02
<annevk5>
Hmm, for cases such as http://www.w3.org/mid/65D2E21C-44E4-4E0B-B7D3-C635A8854A7D⊙ac it seems better to go with HTML5 directly
18:03
<annevk5>
A lot less markup, far easier for authors, and since neither solution has much adoption yet it doesn't really matter it seems.
18:37
gsnedders
should just give up all hope of ever managing to do any sort of maths related subject, seeming he can't do maths
18:37
<gsnedders>
Oh well, I guess Ill do English
18:39
<Philip`>
Maths is just like English, but with numbers
18:42
<inimino>
and with fewer right answers
18:48
<jgraham>
I guess maths actually has more right answers since there must be a countably infinite number of possible english essays but one could consruct a maths question with a uncountably infinite number of right answers
18:49
<inimino>
fewer right answers as a percentage of answers that are likely to be given, to questions that are likely to be asked
18:50
<inimino>
see, this is like English, where you have to interpret ;-)
18:52
<jgraham>
I'm not really sure how one takes percentages of infinite quantities
18:53
<inimino>
both the set of questions likely to be asked and answers likely to be given are finite sets
18:54
<inimino>
(though I guess "Lebesgue measure" might also be a right answer to that question)
18:54
<jgraham>
I have mo idea how you determined that. It seems that there are an infinite number of possible maths questions at least
18:55
<jgraham>
I guess for english questions there is a limit imposed by length
18:55
<inimino>
but surely not all of them are infinetely likely to be asked
18:55
<inimino>
erm
18:55
<inimino>
s/infinetely/equally/
18:56
<inimino>
for any expressible question or answer in maths there is just as hard a limit on length
18:57
<jgraham>
That seems like a more reasonable objection
18:58
<inimino>
an implication is that answers or questions that can't be expressed in the expected lifetime of the universe are somewhat less likely to be given or asked
19:00
<jgraham>
I wonder if one can still represent an infinite number of mathematical problems using some notation that allows the expression of an infinite number of possiblities in finite space
19:00
<jgraham>
I'm not quite sure that would work though
19:01
<inimino>
sounds hard
19:54
<hsivonen>
http://lists.w3.org/Archives/Public/www-tag/2009May/0056.html
22:05
<gsnedders>
hmm… I don't get what PCRE is doig…
22:05
<gsnedders>
*doing
22:06
<gsnedders>
/\x{D800}/u never matches
22:06
<gsnedders>
http://bugs.php.net/bug.php?id=47526
22:17
<Hixie>
these conspiracy theories are very confusing
22:17
<Hixie>
who am i supposed to be getting my bribe from this month again?
22:19
<gsnedders>
Hmm, html5lib has kinda died since the end of March
22:43
<gsnedders>
hmm, PHP 5.3-RC3-dev takes 0.58s less than 5.2.9 to tokenize the spec
22:43
<gsnedders>
(taking it down to 6.37s)
22:53
gsnedders
gawps at PHP's memory usage when storing arrays
22:53
<gsnedders>
We're talking ~70 bytes of overhead per array value
22:54
<gsnedders>
Which with what I'm doing is 25 times what storing the raw value would
22:54
<gsnedders>
*sighs*
22:54
Philip`
wonders how gsnedders is storing raw values that are 2.8 bits each
22:54
<Philip`>
*bytes
22:54
gsnedders
notes he wasn't being exact
23:11
<annevk5>
http://lists.w3.org/Archives/Public/www-html/2009May/0022.html is this bit of HTML5 so intuitive or did he actually look into it?
23:16
<Hixie>
annevk5: hah, awesome
23:17
<Hixie>
annevk5: you should let him know html5 already does that :-)
23:41
<annevk5>
Hixie, I'm not on that list anymore
23:42
<annevk5>
Hixie, out of curiosity, do you keep some notes somewhere other than /issues/ on what you're planning to work / working on?
23:42
<Hixie>
i have my company-internal okr list which says what i plan to have done by the end of the quarter
23:43
<Hixie>
but basically i just do what people want me to do
23:43
<Hixie>
so if you have something you need doing, let mek now
23:43
<Hixie>
and i'll get to it after microdata
23:47
<annevk5>
what's left for microdata? the remaining comments?
23:48
<Philip`>
Hmm... Microdata can be used to represent tree-shaped data structures
23:48
<Philip`>
HTML documents are tree-shaped data structures
23:49
<Philip`>
Therefore, I think we should have a way to encode HTML documents inside microdata attributes
23:50
<Hixie>
annevk5: quite a few use cases not yet addressed
23:50
<Hixie>
still got lots of mail to go through
23:50
<Philip`>
<div item=p><meta itemprop=style content=background:yellow></div> etc
23:50
<Hixie>
well, not mail, but use cases