00:07
<webben>
zcorpan: something like http://pastebin.com/mc5b47f9 only longer
00:13
<Hixie>
kingryan: pure code
00:14
<Hixie>
kingryan: e.g. you have one class per element type, and you create a DOM tree using those classes, then you just interogate the root "are you valid", and the root asks its kids the same question, after having checked them for validity according to the rules for that type of node, etc
00:14
<kingryan>
so, instead of having a table that says "for element a, forbid these attributes", you'd have code that gets run when encountering element a that checks for the attribute?
00:14
<kingryan>
ah, ok
00:58
<Hixie>
gah, i replied again
01:05
<kingryan>
anything but a reply!
03:04
annevk
reads www-archive
03:30
<Hixie>
hm, john boyer left the htmlwg
03:32
<othermaciej>
Hixie: your new proposal is sounding better
03:32
<Hixie>
great!
03:32
<othermaciej>
Hixie: but it's hard to think through all of the implications
03:33
<othermaciej>
Hixie: will have to read it and send comments
03:33
<Hixie>
k
03:34
<othermaciej>
Hixie: Gears does handle a few details that aren't in this, like handling <input type="file"> when offline
03:34
<Hixie>
yeah the plan is to do that by just providing an API on HTMLInputFile that exposes the file data
03:35
<othermaciej>
that's definitely good for relatively small files
03:35
<othermaciej>
probably not as good for large uploads
03:35
<othermaciej>
in Safari at least, we stream the file straight from disk in the middle of the form data when uploading, which is pretty much necessary to handle large files sanely
03:36
<Hixie>
true
03:36
<othermaciej>
you'd also need some control to dump the data back into (maybe the file input API can also offer API to set data in the same format that it allows get)
03:36
<Hixie>
but the gears API isn't much better than HTMLInputElement being done, as far as i can tell
03:37
<othermaciej>
I don't remember how it works
03:37
<annevk>
othermaciej, you can just post using XHR
03:37
<othermaciej>
it might not be any better
03:37
<Hixie>
well you can send your data using any backend sync system you do
03:37
<Hixie>
doesn't have to be a real form
03:37
<othermaciej>
that's true
03:39
<othermaciej>
I guess this doesn't need to tie all that closely into the offline cache
03:40
<Hixie>
yeah, i feel better about keeping them orthogonal
03:42
<annevk>
Hixie, should template= and ref be "super" global?
03:43
<Hixie>
yeah they probably will be
03:43
<Hixie>
bbl afk
04:13
<MikeSmith>
Hixie - do you of any existing browser support for inputmode?
08:41
<zcorpan>
webben: that's an interesting table. so you want Country to apply to all cells in the column but EUROPE and ASIA?
10:31
<hsivonen>
if Opera has supported media queries since version 7, why isn't the query in validator.nu's style sheet applying in Opera 8.5 (S60r3.1)? (it is in 9.2 on desktop and in Mini 4 beta)
10:43
<peepo>
oops
12:02
<annevk>
zcorpan, isn't scope=col already implied?
12:03
<annevk>
it already works for my tables for instance... (which don't have scope=col)
12:20
<zcorpan>
annevk: not with the smart colspan algorithm if you have headers lower down with the same colspan as the one in thead
12:23
<annevk>
hmm, I thought leading spaces were not allowed currently for integer values (such as tabindex=)?
12:24
annevk
looks through the validator tests
12:24
<annevk>
is looking through*, even
12:25
<zcorpan>
"A string is a valid integer if it consists of one of more characters in the range U+0030 DIGIT ZERO (0) to U+0039 DIGIT NINE (9), optionally prefixed with a U+002D HYPHEN-MINUS ("-") character."
12:25
<annevk>
indeed
12:30
<hsivonen>
can anyone suggest a collective term for errors that caused validation to end up in an indeterminate state due to problems outside the checked document (validator bugs, IO failures, unusable schemas)?
12:31
<zcorpan>
"internal error"?
12:31
<zcorpan>
or well, doesn't have to be internal
12:31
<hsivonen>
zcorpan: I considered that, but a bogus outside schema isn't exactly internal
12:31
<hsivonen>
zcorpan: neither is outside server failure
12:32
<zcorpan>
indeed
12:32
<annevk>
can't you distinguish between them?
12:33
<hsivonen>
annevk: I'm tring to come up with a forward-compatible format that has a two-level taxonomy of messages
12:33
<hsivonen>
annevk: with the assumption that the higher level doesn't need to change but the detailed level can change without breaking clients
12:34
<hsivonen>
so that I'd have "informative", "document error" and "not-document error" as the frozen high level
12:35
<hsivonen>
and I could introduce a new error type as a subtype with legacy clients still able to figure out the rough category of the new error
12:36
<hsivonen>
the semantics being: no document nor non-document errors: valid
12:36
<hsivonen>
no non-document errors but one or more document errors: invalid
12:37
<hsivonen>
one or more non-document errors: indeterminate
12:38
<hsivonen>
does that make sense or am on the side of astronautics?
12:38
<zcorpan>
makes sense
12:39
<zcorpan>
"non-document" seems like a good term also
12:39
<annevk>
it would be useful to know though if my schema is bogus for instance
12:48
<annevk>
is the person who did <event-source> for WebKit around?
12:49
<hsivonen>
annevk: that would be <non-document type='schema'>
12:49
<annevk>
there's http://tc.labs.opera.com/html/event-source/ and http://tc.labs.opera.com/html/parsing/event-source/
12:51
<hsivonen>
basically, I want to do a native XML format and a native JSON format in a forward-compatible way and say that I reserve the right to break HTML/XHTML/text scraping at whim without notice
13:19
<Philip`>
annevk: That was G0k, I think
13:21
<annevk>
ah ok, I'll guess I'll ping him when he's around
15:09
<hsivonen>
in the XML output format, a message may have these optional subitems: short message (may contain XHTML <code> and <a>), long advice (may contain rich XHTML narrative) and extract from the source
15:09
<hsivonen>
what kind of markup would people on this channel prefer?
15:11
<hsivonen>
explicit: <error><message>Blah <h:code>div</h:code></message><elaboration><h:p>BlahBlah</h:p><h:p>BlahBlah</h:p></elaboration><source>&lt;div&gt;</source></error>
15:11
<hsivonen>
or something less verbose with implied relationships?
15:12
<hsivonen>
e.g. merging message and elaboration and stating that the first para is the main message
15:12
<zcorpan>
using a heading for the message, perhaps?
15:12
<hsivonen>
interesting
15:13
<zcorpan>
<error><h:h1>Blah <h:code>div</h:code></h:h1><h:p>Blah...
15:13
<hsivonen>
my initial though is that they'd be non-heading-like in content
15:13
<hsivonen>
basically, the message part is what you see now
15:13
<hsivonen>
and elaboration is going to be a new feature
15:14
<hsivonen>
I'm thinking of leveraging the liberal WHATWG copyright license and dumping extracts from the spec
15:15
<zcorpan>
<details>? :)
15:15
<hsivonen>
zcorpan: that could be a good idea for the (X)HTML output
15:17
<zcorpan>
having <message><elaboration><source> in the xml format seems good for me, i'm just brainstorming :)
15:18
<hsivonen>
zcorpan: ok. :-)
15:19
<hsivonen>
speaking of <details>
15:19
<hsivonen>
has safe is it in current browsers and are there forward-compatible open-source canned scripted emulation packages?
15:19
<hsivonen>
s/has/how/
15:26
<hsivonen>
I wonder if I should add redundant data as a precomputed success/failure/indeterminate verdict even though the client could compute the verdict from the messages
15:32
<zcorpan>
iirc <legend> is dropped in some browsers
15:33
<zcorpan>
haven't seen any scripts to emulate <details> that uses <details>
15:34
<zcorpan>
thinking about it, reusing <fieldset> might have a better compat story
15:34
<zcorpan>
but might confuse some ATs
16:53
<annevk>
Lachy, you would set up a redirect for http://blog.whatwg.org/faq (try it again here as there's less traffic)
16:54
<Lachy>
annevk, remind me on Sunday
18:59
<kingryan>
has anyone seen markp around?
19:16
<gsnedders>
kingryan: you ever have any issues with standalone in the XML declaration with feeds?
19:17
<kingryan>
hi gsnedders
19:17
<kingryan>
I'm not sure what you're asking
19:17
<gsnedders>
like <?xml version="1.0" standalone='yes'?>
19:17
<gsnedders>
and the fact that standalone=yes
19:17
<kingryan>
no, not really
19:17
<kingryan>
not sure I've ever seen that in a feed
19:18
gsnedders
looks up the bug report he got
19:18
gsnedders
shrugs
19:18
<gsnedders>
seems to be fixed in the feed
19:19
<gsnedders>
probably no need to set standalone=no in everything
19:19
<gsnedders>
(I await for XMLists to go mad at me for even suggesting that)
19:30
kingryan
doesn't even know what standalone is for
19:32
<gsnedders>
kingryan: http://www.w3.org/TR/xml/#sec-rmd
19:41
<kingryan>
thanks gsnedders. I'm still not sure I understand it, but I think that might be ok
19:42
<gsnedders>
kingryan: read the validity constraint, that basically summarises it
21:23
<Hixie>
don't forget to fill out http://www.w3.org/2002/09/wbs/35125/TPAC2007/registrants#html if you're going to the f2f btw
22:02
<a-ja>
anyone familiar with current state of browser support for figure element?
22:03
<zcorpan>
no support anywhere yet afaik (although it degrades pretty well)
22:04
<a-ja>
fairly...with a bit of css lovin
22:04
<a-ja>
actually, looks like opera kestral gets it right,,,,,or mostly
22:06
<annevk>
the draconian answer is that if you use XHTML5 it will work quite good (given that you add stuff like display:block)
22:07
<a-ja>
isntf figure supposed to be block (like a <p>) ? and a legend within a figure should be inline?
22:08
<annevk>
yeah, but it's not natively supported yet and the initial value of the display property is inline
22:09
<annevk>
not sure about <legend> in <img>, I would expect the typical presentation to be block level I think, so that it appears under the image
22:09
<zcorpan>
seems <legend> creates an empty element in kestrel that is display:block... pretty much like <br>
22:09
<zcorpan>
(unless it's in a <fieldset> that is)
22:10
<annevk>
<legend> parsing is kind of buggy everywhere...
22:10
<zcorpan>
indeed
22:11
<a-ja>
certainly is in ff....legend content ends up as figure's content in dom (and no legend in dom)
22:14
<a-ja>
from you're comment re: <img>.....are you implying that <img></img> allowing fallback is in html5, rather than <img/> 's alt ?
22:17
<a-ja>
this is what i've been looking at, fwiw: <figure><img alt="fallback text"/><legend>caption text</legend></figure>
22:18
<annevk>
what comment regarding <img> are you referring to?
22:19
<a-ja>
<annevk> not sure about <legend> in <img> .....
22:19
<annevk>
oh, meant <legend> in <figure>
22:19
<a-ja>
ah
22:19
<annevk>
my bad
22:20
<a-ja>
np...you just saved me some re-reading :)
22:22
<zcorpan>
hmm. opera doesn't support display:table-caption on <legend>s
22:25
<a-ja>
seeing some text-shadow and box-shadow oddities in kestrel, too, btw.
22:26
<a-ja>
haven't come up with a simplified test case yet though
22:27
a-ja
will when he gets some more free time
22:27
<annevk>
box-shadow?
22:27
<a-ja>
um....just text-shadow, actually
22:27
<annevk>
too bad :)
22:28
<zcorpan>
box-shadow would be nice :)
22:29
<a-ja>
kinda mostly works on safari beta (win).....a little shaky with border-radius
22:34
<zcorpan>
yeah, the new border stuff creates a lot of edge cases
22:35
<Philip`>
Was that meant to be a pun? :-p
22:35
<a-ja>
w/ border bg images, too, still? <haven't tried lately)
22:35
<a-ja>
heh
22:36
a-ja
groans
22:38
<a-ja>
well....tks for the info, folks.....later
22:38
<a-ja>
keep up the good work!
22:39
<zcorpan>
Philip`: it wasn't :)