00:13
<Hixie>
i wonder where kevin quoted that from
00:14
<Hixie>
doesn't seem to be in rfc2426
00:15
<Philip`>
Hixie: There's this clever new web site known as a "search engine", which lets you answer questions like that
00:16
<Philip`>
e.g. http://www.imc.org/pdi/vcard-21.txt
00:17
<Hixie>
i tried that
00:17
<Philip`>
(That looks like a bad conversion of http://www.imagescript.com/atg/resources/vcard-21.doc)
00:17
<Hixie>
before asking :-)
00:17
<Hixie>
apparently the string i used was broken in some way
00:18
<Hixie>
weird
00:19
<Philip`>
A likely excuse
00:22
<Hixie>
i wonder why that text got lost in the conversion to rfc
01:32
<Hixie>
wtf, vcard contains the lat long position 37.386013;-122.082932
01:32
<Hixie>
which when i look it up
01:32
<Hixie>
turns out to be within cycling distance of where i'm sitting
01:32
<Hixie>
freaky
01:49
<franksalim>
that's 0.7 miles from me
02:13
<karlcow>
[18:07] <othermaciej> Google inventing their own scheme is a demonstration of the main problem with RDF
02:13
<karlcow>
not at all. The problem here is not RDF but power. Small communities can still invent their own vocabularies and that is good. Google here is creating a new vocabulary AND they have the power to impose it.
02:17
<Hixie>
ok, author conformance criteria for the vCard vocabulary are now defined: http://www.whatwg.org/specs/web-apps/current-work/#vcard
02:17
<Hixie>
(it's basically the same as hCard, but defined in detail)
02:22
<heycam>
"The item with the type vcard has an fn property value consists of a string with exactly one sequence of space characters, and it neither at the immediate start nor the immediate end of the string."
02:22
<heycam>
missing a word (maybe "occurs")
02:26
<Hixie>
thanks
02:27
<Hixie>
wow that whole sentence was broken in many ways
07:54
<hsivonen>
hmm. so even the super-smart googlers make mistakes with Namespaces in their RDFa documentation
08:22
<hsivonen>
I wonder what kind of havoc will ensue now that Google uses the de facto VML prefix for another purpose
08:52
<hsivonen>
"<p><strong>Blast 'Em Up Review</strong></p>
08:52
<hsivonen>
"
08:53
<hsivonen>
It's a semantic heading, because it uses <strong> and <strong> is semantic
09:30
Philip`
tries posting somewhere that uses reCAPTCHA, and finds that it rejects two attempts even though the image was pretty clear and he's almost entirely certain he typed the words in correctly
09:48
<Philip`>
http://radar.oreilly.com/2009/05/google-announces-support-for-m.html - xmlns:v="http://rdf.data-vocabulary.org "
09:49
<Philip`>
Honestly, how hard can it be to get a namespace string right?
09:50
<jgraham>
Philip`: Too hard, it seems :)
09:50
<hsivonen>
Philip`: very, very hard, evidently
09:50
<hsivonen>
I have a special program that hooks into accessibility APIs for inputting namespace URIs
09:51
<hsivonen>
(I don't use it for anything else. really.)
09:51
<hsivonen>
(Typinator)
09:51
<jgraham>
(partially because curies doing simple concatenation violates all sorts of assumptions that people have e.g. http://foo/ and http://foo being assumed the same but treated differently)
09:52
<Philip`>
Looks like Google's updated their help pages to include the trailing "/" now
09:52
<hsivonen>
jgraham: in fairness, that not a CURIE bug per se. It's an issue that goes back to Namespaces itself
09:53
<jgraham>
hsivonen: Really? Maybe I don't fully understand either then
09:53
<jgraham>
Well I guess you just mean that a namespace has to be a literal string
09:53
<hsivonen>
jgraham: xmlns='http://foo'; and xmlns='http://foo/'; are distinct namespaces ever since Namespaces
09:53
Philip`
notes that people may make exactly the same mistakes in mistyping HTML5 microdata itemprop values
09:54
<hsivonen>
Philip`: not if the vocabulary designer has the good sense not to mint URI-based vocabularies
09:58
<Philip`>
hsivonen: People mistype non-URI strings too
09:58
<Philip`>
So, maybe not exactly the same mistakes
09:58
<Philip`>
but mistakes with the same effects :-)
09:59
<Hixie>
not quite the same effect
09:59
<Hixie>
it'll break the one occurrence
09:59
<Hixie>
not the whole page
09:59
<hsivonen>
Philip`: I see that as an argument to bring microdata into the sphere of validation instead of keeping it as an anything goes zone with possibly 'spurious triples'
09:59
<jgraham>
Philip`: I think it is a usability issue to use strings that people expect to behave in one way (like they do in their browser address bar) in a context where they are treated entirely differently (as a pure literal)
10:08
<Philip`>
As far as I can tell, the biggest problem with Google's microformats/RDFa support is that they don't provide any way to test it, so it's impossible to tell whether you've got it right or not
10:11
<Philip`>
(...where "you" is a person who's tried to mark up data on their page)
10:14
<hsivonen>
I want to test what happens if
10:14
<hsivonen>
1) I don't declare the prefix v
10:14
<hsivonen>
2) I declare it to another URI
10:14
<hsivonen>
3) I declare the right URI to be bound to another prefix
10:16
<hsivonen>
Considering that Y! hard-coded some prefixes, I think it's too early to say that Google hasn't done the same
10:24
<aja>
is there an example html5 vcard anywhere?
10:25
aja
wants to make sure he's reading microdata stuff correctly
10:28
<Philip`>
aja: There isn't one yet, as far as I'm aware
10:29
<jgraham>
aja: Make one :)
10:32
<aja>
not sure whether/where to use item= or itemprop=
10:35
<Philip`>
<div item=vcard> and then everything else is itemprop
10:35
<Philip`>
as far as I can tell
10:36
<Philip`>
(The spec says "vcard" is a (predefined) type, and everything else is a (predefined) property name)
10:37
<jgraham>
FWIW I think itemprop is a terrible name
10:37
<jgraham>
and that seems like one good reason why
10:37
<aja>
Philip`: that's what i thought, too
10:37
<Philip`>
jgraham: Got any better suggestions? :-)
10:38
<jgraham>
Philip`: property ;)
10:38
<jgraham>
Or key
10:39
<jgraham>
propertyname is rather long
10:39
<aja>
was originally property...changed to itemprop shortly thereafter (why? i dunno)
10:40
<jgraham>
aja: Because property has theoretical conflicts with future versions of RDFa
10:40
<aja>
makes sense
10:51
<Hixie>
aja: http://junkyard.damowmow.com/387
10:51
<Hixie>
except s/property/itemprop/
10:52
<aja>
gotcha....tks
10:53
aja
was pondering how to do include-pattern sorta things
10:53
<Hixie>
aja: subject="" is the only way of doing include-like things yet
10:53
<Hixie>
i was thinking of adding <ref/> in a future version
10:53
<Hixie>
but probably not in html5
10:54
<aja>
hmmm....interesting
10:58
<aja>
Hixie: might subject being possibly plural work?
10:59
<Hixie>
how do you mean?
11:00
<aja>
eg...any org name up on a site's header....being able to be used in multiple vcards further down the page
11:02
<aja>
or single vcard being used on multiple calendar enties...that sorta thing
11:04
<Hixie>
for the latter i think <ref> makes more sense
11:04
<Philip`>
<div item=vcard id=me>...</div> <div item=calendar><meta itemprop=owner object=me>...</div>
11:04
<Hixie>
(if we eventually add it)
11:04
<Hixie>
and for the former, well... use <meta>, i guess
11:05
<Hixie>
do we have any other attributes that take space-separated lists of ids?
11:05
<Hixie>
Philip`: that's more or less what <Ref> would be
11:07
<aja>
only html profile for GRDDL, that i'm aware of, off type of my head.....,.and that's meta now anyway
11:08
<aja>
n/m...ids
11:11
Philip`
likes how Vodafone has developed open source Linux software in Python for their devices
11:11
<Philip`>
(It's not officially supported, but it seems much better than what the other mobile providers have done on Linux (i.e. nothing, as far as I can tell))
11:13
jgraham
likes how Vodaphone have developed Linux software for the mobile broadband thing he has
11:13
<aja>
no Android apps site out there yet? ;)
11:13
<Philip`>
jgraham: That's the specific software I'm thinking of :-)
11:13
<Philip`>
Too bad I can't get it to actually work, though
11:13
Hixie
is starting to get a headache from the iCalendar spec
11:15
<Hixie>
i like how vcard and icalendar are almost compatible but not quite
11:15
<Hixie>
e.g. they have subtly different formats for times and dates
11:15
<jgraham>
Philip`: It seems to both work and not work for me. The whole UI seems to be broken. The actual part that allows me to connect to the internet seems to work
11:15
<Hixie>
or e.g. how iCalendar requires the PRODID line but vcard doesn't
11:16
<Philip`>
jgraham: I managed to connect without that software at all, just using normal PPP stuff, so that's alright
11:17
<hsivonen>
It's a small miracle that calendaring apps interoperate at all
11:17
<Philip`>
Do calendaring apps actually interoperate at all?
11:17
<Hixie>
iCalendar isn't that bad
11:18
<Hixie>
it's better than many web specs on which we have workable (if not stellar) interop
11:18
<Hixie>
doesn't define error handling, but nobody did in the 90s
11:18
<jgraham>
Philip`: Oh well maybe that's what's working for me. I didn't try without the software because I assumed it would fail
11:18
<aja>
hsivonen: latest tryserver builds still for html5 parser?
11:19
<hsivonen>
Philip`: iCal interoperates with Nokia's S60 calendar app and partially interoperates with Ericsson's T series calendaring app
11:19
<hsivonen>
aja: nope
11:19
<hsivonen>
aja: the latest are for the HTML5 namespace correctness patch
11:20
<aja>
hsivonen: the one in your blog entry crashed horribly on win/xp.....linux was fine, though
11:20
<hsivonen>
I should probably pull the old links from the blog entries...
11:27
<hsivonen>
Hixie: when you say "RDF quad", do you mean lang/datatype as the fourth item in the tuple?
11:27
<Hixie>
no, i mean provenance
11:28
<Hixie>
the language and datatype annotations are considered part of the object literal
11:28
<hsivonen>
ok
11:35
<hsivonen>
what's happening with Origin in the IETF?
11:39
<hsivonen>
Hixie: have you addressed all the recent hgroup feedback in the spec? that is, is the outline algorithm stable now?
11:41
<Hixie>
i have at least 16 messages on the topic that i haven't dealt with yet
11:41
<hsivonen>
ok
11:42
<Hixie>
btw if the absolute URL algorithm URIfies, that's a bug
11:42
<Hixie>
are you sure it does?
11:43
<hsivonen>
I'm not quite sure, but I read it twice and I think it does
11:43
<Hixie>
sounds like a bug
11:43
hsivonen
goes read it the third time
11:43
<Hixie>
it might un-INDA and leave the path as IRI
11:43
<Hixie>
(which would also be a bug)
11:43
<Hixie>
it should ToUnicode() imho
11:45
<hsivonen>
Hixie: should script-inserted xml:base be considered in HTML documents?
11:45
<hsivonen>
doh.
11:45
<hsivonen>
there's a note that says yes
11:45
<hsivonen>
there's a bug in Gecko, then
11:45
<Hixie>
spec says yes
11:45
<Hixie>
the spec goes to some lengths to make base URI changes not be expensive
11:46
<hsivonen>
Gecko checks document HTMLness and returns early
11:46
<Hixie>
if you look at the source, there are comments in a bunch of places about implications certain rules have wrt base URI changes
11:46
<hsivonen>
Hixie: it seems to me that steps 6. and 7.1. modify the URL in the algorithm for resolving
11:47
<hsivonen>
8., too
11:48
<hsivonen>
argh. vCard-RDF uses upper-case local names for properties
11:48
<hsivonen>
worse, it also used mixed case
12:04
<Hixie>
hsivonen: send mail to the list (and connoly⊙wo), i'm pretty sure we want the absolute url to be an IRI, off-hand. Not sure what the implications are though.
12:07
<jgraham>
Hixie: I might be missing something trivial but the "resolve a url" algorithm takes the url to be resolved + either an absolute url or an element
12:07
<jgraham>
But the definiton of "absolute url" just says "if the url resoves onto itself"
12:07
<jgraham>
which doesn't specify what you are resolving it against
12:08
<jgraham>
(those bits in quotation marks are not actual quotes)
12:08
<Hixie>
it should probably say explicitly that it means "if it resolves onto itself for any input base url" or some such
12:08
<Hixie>
again, danc is the guy to talk to here
12:09
<Hixie>
he's editing that bit
12:09
<jgraham>
OK
12:09
<Hixie>
ok bed time now
12:09
<Hixie>
nn
12:09
<jgraham>
goodnight
12:09
Philip`
wonders if the question about whether 'absolute URL' is a datatype or is a property of a string has been raised or resolved yet
12:10
jgraham
has been wondering something similar
12:10
Philip`
is currently too lazy to raise it
13:00
<Philip`>
Hmm, the images on the Last Week blog don't work too well for me
13:01
<Philip`>
since they say "You are not able to access this service because Content Control is in place. If you're 18 years or over, you can remove Content Control by contacting your mobile service provider's customer support team."
13:06
<virtuelv>
Philip`: are you saying that your mobile ISP forces a porn filter on you?
13:06
<virtuelv>
that sounds australian
13:21
<Philip`>
virtuelv: Seems like it
13:22
<Philip`>
though it's not really "forces" if I could ask them to remove it
13:22
<Philip`>
but it sounds like they make it awkward to remove
13:22
<Philip`>
and I don't care that much about seeing Mr Last Week's images
13:23
<jgraham>
Philip`: You are not missing much. Indeed it is like the opposite of missing much
13:25
<virtuelv>
Philip`: I concur with jgraham's conclusion
13:25
<virtuelv>
let's just say that he goatse'd the validator icon for the purpose of annoying mpilgrim
16:04
<Wolfman2000>
...umm...sort of a silly question here...is there any reason why <table> tags no longer have the summary attribute?
17:04
<Philip`>
Wolfman2000: It's because of the argument that <table summary> is often used badly, and rarely used well, and is replaceable by alternatives (like <legend> or <p>) that are accessible to users of graphical browsers too
17:05
<Wolfman2000>
Philip`: ...and what of those that use the handicapped browsers?
17:05
<Philip`>
Wolfman2000: They can access content in <legend> and <p> too
17:05
<Wolfman2000>
True. Still...don't tables use <caption>?
17:05
<Philip`>
Um
17:05
<Philip`>
I might mean <caption>, not <legend>
17:08
<webben>
Wolfman2000: i.e. the contention is that an explanation of how a complex table is organized is useful to everybody.
17:09
<Philip`>
The argument also involves the idea that the better way to improve accessibility is by making it as simple and as automatic as possible, instead of by adding complexity through accessibility-specific hooks for non-graphical browsers
17:10
<webben>
Wolfman2000: It's notable that using ARIA you could point at the p with aria-describedby.
17:10
<Philip`>
(because, on average, people use those hooks in incorrect ways (e.g. <img longdesc="some text instead of a URL">) or useless ways (e.g. <table summary="Layout table with 3 columns and 2 rows">), and either harming the user experience or just wasting their own time)
17:11
<Wolfman2000>
...that reminds me Philip`. The new way for alt text on images is supposed to be <img>Alt Text</img>, right?
17:11
<Philip`>
Wolfman2000: No
17:11
<webben>
Wolfman2000: Still alt=
17:11
<Philip`>
Wolfman2000: <img> is still an empty element
17:12
<Philip`>
(except in XHTML2, I guess)
17:12
<Wolfman2000>
...must be thinking of XHTML2
17:12
<webben>
yep
17:13
<Philip`>
HTML parsers won't understand <img>...</img>, and can't be changed to understand it without breaking existing pages
17:14
Philip`
notes that some people argue that the benefits of <table summary> are high and the costs are low and therefore it's worth keeping it in HTML5
17:17
<Philip`>
(Also some people seem to focus on the ability to make a page accessible, rather than on the likelihood of a random page written by an average person being made slightly more accessible)
17:17
<Wolfman2000>
I admit I'm more on the "accessible" side
17:18
<Philip`>
Nobody is on the anti-accessible side :-)
17:21
<Wolfman2000>
Philip`: at times I question that.
18:21
gsnedders
is wondering how to sanely mark up drama
18:24
<Philip`>
Is this like Shakespearean drama or emo drama?
18:24
<gsnedders>
Shakespearean
18:25
<gsnedders>
Philo: "And is become the bellows and the fan/To cool a gypsy's lust." (I.1.9–10)
18:25
<gsnedders>
Problem: marking that up sanely, and getting sane presentation of that.
18:25
<gsnedders>
However, my exam is on Friday so I don't really have time to care.
18:31
<Philip`>
Hmm, Firefox 2 moves <link> and <meta> into <head>; Firefox 3 only moves <meta>
18:48
<gsnedders>
For some reason my spell-checker doesn't know these Shakespearean names.
21:12
jgraham
discovers he has accidnetially caused a very odd spec discussion on a mailing list he wasn't aware of
21:19
Philip`
demands details
21:19
<Hixie>
jgraham: uri?
21:27
<jgraham>
Philip`: It turns out that there is a es5-discuss mailing list in addition to the es-discuss mailing list I knew about
21:28
<jgraham>
After reorting a possible compatibility issue on es-discuss a thread started on es5-discuss which ends with Mark Miller and Douglas Crockford suggesting removing string indexing from ecmascript 5
21:29
<jgraham>
As an alternative to introducing a single compatibiliy-preserving type check
21:29
<Philip`>
They should go the whole hog and remove strings entirely
21:29
<inimino>
there is or has been in the past es-discuss, es3x-discuss, es4-discuss, and es5-discuss
21:30
<inimino>
(at least)
21:30
<jgraham>
inimino: And people complain about whatwg vs public-html
21:30
<jgraham>
At least those two have different participati
21:30
<jgraham>
on requirements
21:32
<jgraham>
Hixie: https://mail.mozilla.org/pipermail/es5-discuss/2009-May/002490.html
21:33
<Hixie>
thx
22:03
jgraham
wonders if wikipedia is supposed to be used as software api documentation
23:10
jgraham
has a certianly-broken implementation of N3 output now
23:16
<Hixie>
http://www.w3.org/mid/1igm05tt25ip8kg10hureru72agmo2n2c9⊙hbhd has the most references of any e-mail i've ever seen in w3c space
23:26
<jgraham>
That is rather impressive
23:38
Philip`
gets the feeling that Björn has much more fun with the formatting of his emails than with their content
23:38
<Hixie>
...why does the iCalendar example for RESOURCE imply that the meeting is going to need a racoon?!
23:39
<Hixie>
what kind of weird-ass meeting were the iCalendar guys going to?!
23:42
<Philip`>
Perhaps it was a meeting to discuss plans to overthrow Cyril Sneer
23:45
<Hixie>
can someone tell me if http://www.ietf.org/rfc/rfc2445.txt says whether the STATUS line can be included more than once per vevent?
23:45
<Hixie>
it looks like it allows it
23:45
<Hixie>
but i may be missing some key line somewhere that says that the default is no duplicates?
23:51
<Dashiva>
A French raccoon to boot...
23:51
<Hixie>
i'm thinking the word has a double meaning but if it does i don't know it
23:52
<Dashiva>
Maybe karlcow knows