07:53
<annevk>
in WebKit, document.body.setAttribute() creates undefined="undefined"
07:53
<annevk>
that's still not fixed?
07:56
<annevk>
https://bugzilla.mozilla.org/show_bug.cgi?id=315805
07:57
<AryehGregor>
annevk, that's a problem with their WebIDL bindings, right? Don't all methods do that, pretty much?
07:57
<AryehGregor>
They don't throw on too few arguments being passed.
07:59
<annevk>
AryehGregor: f8dy fixed their IDL bindings, but I guess he never fixed the broken methods
08:01
<AryehGregor>
If he fixed the IDL bindings, how could that not throw? It would have to have the params declared as optional or something.
08:01
<annevk>
right
08:01
<annevk>
make for DOM takes almost 10 seconds?
08:01
<annevk>
what is so difficult?
08:38
<annevk>
the problem with the namespace algorithms in DOM seems to be that they do not account for multiple attributes
08:38
<annevk>
I guess nobody looks at them and that's why we haven't noticed until now?
08:38
<annevk>
http://dvcs.w3.org/hg/domcore/raw-file/tip/Overview.html#locate-a-namespace-prefix
08:38
<annevk>
you could have multiple attributes xmlns:test="x" xmlns:test2="x" etc.
08:39
<annevk>
which local name you return is not defined
08:39
<annevk>
then in http://dvcs.w3.org/hg/domcore/raw-file/tip/Overview.html#locate-a-namespace you have the same problem
08:39
<annevk>
actually, in that case you can have only one such attribute I think
08:40
<annevk>
other question
08:41
<annevk>
should http://dvcs.w3.org/hg/domcore/raw-file/tip/Overview.html#dom-element-setattributens have all those restrictions given that setAttribute() can be used to circumvent them?
08:55
<Von_Davidicus>
Hello...
08:55
<Ms2ger>
Or maybe add the restrictions to setAttribute? Or would that break facebook?
08:55
<Ms2ger>
Also, <!doctype html> on http://www.w3.org/TR/2012/WD-dom-20120405/ , I guess I'll have to believe that they mean it now
08:55
<annevk>
Ms2ger: I just removed those
08:55
<Ms2ger>
Oh
08:55
<annevk>
Ms2ger: nobody implements them
08:55
<Ms2ger>
Mm
08:58
<annevk>
Von_Davidicus: use <section> combined with <h1>
08:59
<annevk>
https://bugzilla.mozilla.org/show_bug.cgi?id=742549 o_O
09:01
<Von_Davidicus>
I thought <h1> was the top-level header.
09:02
<annevk>
that changed with the introduction of <section>
09:03
<Von_Davidicus>
That seems a bit screwy with semantics, but then, I'm easily confuzzled. :)
09:04
<annevk>
HTML is not entirely logical either, so that's fine :)
09:06
<Von_Davidicus>
Oh, don't I know it. HTML seems really simple--right up until you try to -explain- it.
09:09
<Von_Davidicus>
So does <section> essentially reset the header numbering? For example, I could have the headers in the main page, and then have, say, <section class="code_example"> and start the example heading over at <h1>?
09:12
<annevk>
yeah more or less
09:12
<annevk>
Von_Davidicus: reading http://www.whatwg.org/C#outline might help
09:16
<Von_Davidicus>
<!DOCTYPE HTML PUBLIC "ISO/IEC 15445:2000//DTD HyperText Markup Language//EN">
09:16
<Von_Davidicus>
09:16
<Von_Davidicus>
", will it still validate?
09:16
<Von_Davidicus>
And sorry for the linebrakes.
09:16
<annevk>
don't think so
09:17
<annevk>
try http://validator.nu/ to be sure
09:17
Von_Davidicus
was kidding. :)
09:17
<annevk>
I think we do allow some old doctypes
09:18
<annevk>
but better to stick to <!doctype html>
09:19
<annevk>
Ms2ger: about features strings, can you ping eseidel somehow about which he wants to support?
09:20
<annevk>
Ms2ger: he was the one you were talking with this about that right?
09:20
<annevk>
Ms2ger: and maybe ask Gecko too?
09:20
<annevk>
Ms2ger: source code or otherwise :p
09:22
<annevk>
ooh
09:22
<annevk>
Mozilla just moved fullscreen to Element
09:22
<annevk>
hmm
09:23
<Von_Davidicus>
Another question, more serious this time: how will XHTML5 handle character entity references? I tried it and got Firefox's Yellow Screen of Death.
09:25
<annevk>
Von_Davidicus: http://www.whatwg.org/specs/web-apps/current-work/multipage/the-xhtml-syntax.html#the-xhtml-syntax has some details
09:25
<annevk>
Von_Davidicus: best to just use HTML
09:26
<annevk>
Von_Davidicus: or stay away from named entities
09:28
<Von_Davidicus>
Okay.
09:34
<annevk>
http://trac.webkit.org/changeset/111028/trunk/Source/WebCore/dom/Document.idl
09:34
<annevk>
hmm
09:34
<annevk>
both Mozilla and W3C version
09:34
<annevk>
never mind that Fullscreen is not done at the W3C
09:34
<annevk>
at least not that I know of
09:59
<annevk>
haha
09:59
<annevk>
http://www.alistapart.com/authors/�/tantekcelik
10:03
Von_Davidicus
ponders whether or not to turn one of his pages from HTML 4.01 to HTML5
10:09
<Philip`>
Von_Davidicus: You shouldn't bother, unless it'll give any practical benefits to your page's users
10:09
<Philip`>
Better to spend the effort on more meaningful changes :-)
10:11
<Von_Davidicus>
Well, the page hasn't needed a change in a while.
10:16
<Ms2ger>
annevk, I dunno if I talked to anybody at webkit
10:18
<annevk>
Ms2ger: but I do
10:18
<annevk>
Ms2ger: https://bugs.webkit.org/show_bug.cgi?id=76214#c5
10:19
<Ms2ger>
Ah, indeed
10:20
<Von_Davidicus>
Hmmm... if the "checked" attribute isn't allowed, how DO I preselect a checkbox?
10:22
<Von_Davidicus>
or, for that matter, "readonly"?
10:22
<annevk>
it is allowed
10:22
<Philip`>
Von_Davidicus: Who says it's not allowed?
10:23
<Von_Davidicus>
Well, the W3C validator I just checked my page on.
10:24
<Philip`>
It reports no error for me for e.g. "<!DOCTYPE html><title></title><input type=checkbox checked>"
10:25
<Philip`>
(though the spec doesn't allow readonly on checkboxes)
10:26
<Von_Davidicus>
The tags I'm having trouble with are: <input name="Rods_Count" class="rod_count readonly" type="hidden" readonly="readonly" value="0"> and <input name="Rods_Choose" class="rod_check" type="hidden" checked="checked">.
10:26
<volkmar>
annevk: where is window.fullscreen spec'd?
10:26
<Philip`>
Von_Davidicus: checked is meaningless (hence not allowed) for type=hidden
10:27
<Philip`>
Same for readonly for type=hidden
10:27
Von_Davidicus
will thus have to redo some of his PHP.
10:27
<Philip`>
(http://www.whatwg.org/specs/web-apps/current-work/multipage/states-of-the-type-attribute.html#hidden-state-(type=hidden) gives the details)
10:28
<Von_Davidicus_>
And, possible, some of my JavaScript.
10:28
<Ms2ger>
annevk, http://wiki.whatwg.org/wiki/DOM_features should have all those that Gecko supports now
10:29
<Philip`>
Von_Davidicus: Why are you specifying those attributes here, when they don't make sense for hidden inputs?
10:30
<volkmar>
annevk: forget the question... I realize it's not part of a spec...
10:31
<annevk>
volkmar: that attribute does not exist; see http://dvcs.w3.org/hg/fullscreen/raw-file/tip/Overview.html
10:31
<volkmar>
annevk: https://developer.mozilla.org/en/DOM/window.fullScreen
10:31
<Von_Davidicus_>
Because I need them there for my JavaScript to work.
10:32
<annevk>
volkmar: do we need that attribute for something?
10:32
<annevk>
volkmar: it seems you can already accomplish that using matchMedia and for element fullscreen you can check document.fullscreenElement
10:32
<annevk>
it's also incorrectly cased
10:33
<volkmar>
annevk: sadly, seems like document.fullscreenEnabled is cased mozFullScreenEnabled in Gecko
10:33
<annevk>
hopefully that's fixed when they unprefix
10:33
<Ms2ger>
Do we have a bug to unprefix?
10:34
<annevk>
Firefox is standardizing on "fullscreen" after all and so is the rest of the world
10:34
<annevk>
(in UI language)
10:37
<Von_Davidicus_>
Basically, my code assumed that every table row would have those elements in it--and the inputs were hidden if they weren't actually needed.
10:41
<annevk>
Ms2ger: haven't seen anything
10:42
<annevk>
I'm having a hard time to not prematurely merge big5 and big5-hkscs
10:42
<Ms2ger>
I'll file
10:43
<annevk>
cool
10:43
<annevk>
hopefully Gecko can create some sanity in the all the Fullscreen code out there
10:44
<annevk>
I haven't seen much community feedback btw; I guess it works good enough...
10:44
<Ms2ger>
Facebook uses it for photos, I hear
10:44
<annevk>
youtube, vimeo, etc. use it
10:45
<annevk>
but all with lots of code paths because of the prefixes, prefixed events, etc.
10:45
<annevk>
it's quite the mess
10:50
<volkmar>
annevk: fullscreen specs say "HTML defines under what conditions the fullscreen enabled flag is set"
10:50
<volkmar>
but I don't see anything in the html specs
10:51
<annevk>
yeah HTML needs to be updated
10:52
<annevk>
not sure if I filed a bug on that already
10:52
<annevk>
but basically, I think the latest was that <iframe allowfullscreen> sets that flag
10:52
<annevk>
for the document it contains
10:52
<annevk>
and it's set automatically for the top-level browsing context's document
10:52
<volkmar>
annevk: so, to know if a document is fullscreen, you have to do fullscreenElement != null, right?
10:53
<annevk>
to know if an element is fullscreen you do that
10:53
<annevk>
if you want to know if the user used some kind of UI you would use media queries
10:54
<volkmar>
so, a dev can't know if the page is currently fullscreen?
10:55
<annevk>
I just said the opposite
10:55
<volkmar>
sorry
10:55
<volkmar>
i meant, with a js api
10:55
Von_Davidicus
heads offline.
10:56
<annevk>
there's window.matchMedia
10:56
<Von_Davidicus_>
Thanks for the header element info. :)
10:56
<annevk>
volkmar: http://dev.w3.org/csswg/cssom-view/#dom-window-matchmedia
10:58
<annevk>
yay foolip
10:59
<annevk>
one interwebs for foolip please
11:00
<Ms2ger>
Granted
11:01
<annevk>
seems I'm gonna take a break for a couple of hours
11:02
<annevk>
big5 can wait a little
11:03
<annevk>
volkmar: if you think fullscreen status of a document (other than fullscreen element) needs some kind of convenience attribute please email WHATWG
11:03
<annevk>
volkmar: I personally think media queries combined wit the API for them are fine, especially since the majority case will be an element fullscreen
11:03
<annevk>
with*
11:04
<Ms2ger>
What makes people even publish mixed-encoding pages?
11:44
<kennyluck>
Can someone give me some examples of invalid URL?
11:44
<kennyluck>
I know "%" is invalid in IE for one.
13:14
<annevk>
kennyluck: http://a b/
13:14
<annevk>
kennyluck: note the space
13:14
<annevk>
kennyluck: however, we might make make it so that all URLs can be resolved
13:14
<annevk>
kennyluck: and that failures happen at the network layer instead
13:15
<annevk>
kennyluck: kind of depends on the URL work, which MikeSmith is working on and abarth is sometimes
13:15
<annevk>
I might work on that after Encodings is done, dunno
13:46
<annevk>
foolip_: hey
13:46
<foolip_>
annevk, hi
13:46
<annevk>
foolip_: so those 22 pages where all the pages with bytes in the indicated ranges
13:46
<annevk>
s/where/were/
13:47
<annevk>
foolip_: all the pages with lead bytes under 0xA1 or in the 0xC6-0xC8 region
13:47
<foolip_>
annevk, do you think that skews the results?
13:48
<annevk>
not really, those are the ranges for which browsers have incompatible mappings
13:48
<annevk>
and for which big5 and big5-hkscs have incompatible mappings
13:48
<annevk>
everything else they agree on so that does not really matter
13:49
<annevk>
so I think that what you generated is fine
13:49
<annevk>
and just default to FFFD for the rest
13:49
<foolip_>
right, I was quite happy to see that the compatible intersection of big5 was a subset of the compatible intersection of big5-hkscs, save one mapping
13:49
<annevk>
the multiple code point stuff seems hairy though
13:49
<foolip_>
annevk, spec-wise or implementation-wise?
13:50
<annevk>
I guess we can special case it if there's only four
13:50
<foolip_>
The alternative is
13:50
<foolip_>
PUA
13:50
<Ms2ger>
Go wash your mouth
13:51
<annevk>
so
13:51
<annevk>
<U+00CA,U+0304> Ê̄
13:51
<foolip_>
annevk, yes, the official hkscs-2008 mapping listed only those 4
13:51
<annevk>
is there no NFC character for that?
13:51
<foolip_>
annevk, nope, at least not according to python unicodedata
13:51
<annevk>
we could also FFFD it given that we didn't find it in real usage
13:52
<foolip_>
It seems better to follow the spec when there is one, don't you think?
13:52
<foolip_>
or does it create other problems?
13:52
<foolip_>
I'll see what the surrounding characters were, these 4 seem a bit random...
13:53
<annevk>
dunno, I don't really have respect for legacy encoding standards and it makes both the encoder and decoder more complicated
13:53
<foolip_>
annevk, the surrounding characters are also various latin letters with diacritics, so it seems about right
13:53
<annevk>
so far all encoders work by just passing in one code point at a time and getting back some bytes
13:54
<annevk>
for this it would need state
13:54
<foolip_>
can't it just return more bytes?
13:54
<foolip_>
oh, encoder?
13:55
<foolip_>
why should we spec an encoder for big5?
13:55
<annevk>
URLs and <form>
13:55
<foolip_>
meh, so that's already web-exposed?
13:55
<annevk>
yes
13:56
<foolip_>
I'll see if the individual code points can be represented in Big5...
13:57
<annevk>
we could also not support it in the encoder
13:57
<annevk>
and if they're not in big5 they would turn into entities in the common case
13:57
<annevk>
and support them as some kind of special path in the decoder
13:57
<foolip_>
nope, big5-hkscs has no combining characters it seems, but python does actually support encoding these properly
13:58
<foolip_>
I would suggest special-casing them and seeing if implementors complain
13:59
<annevk>
does any browser implementation support them now though?
14:00
<annevk>
because if they don't, I'm not sure it's worth adding
14:01
<foolip_>
they all use PUA
14:01
<annevk>
oh right, except for Opera?
14:01
<foolip_>
right
14:01
<foolip_>
specing PUA seems worthwhile avoiding, IMHO
14:02
<annevk>
fair enough
14:02
<foolip_>
decoding them correctly seems sensible, for the encoder I'm quite sure just dropping them would do
14:03
<foolip_>
or what do encoders do with code points that can't be represented?
14:03
<annevk>
k
14:03
<annevk>
depends I think
14:03
<annevk>
form submission makes them entities
14:03
<annevk>
URLs might use "?"
14:03
annevk
forgot
14:03
<annevk>
I haven't really looked at encoders in detail yet
14:03
<annevk>
there's lots of XXX for encoders
14:03
<foolip_>
ok, so encoding the first half and escaping the second as an entity might do the job
14:05
<annevk>
where did you find the HKSCS standard btw?
14:05
<annevk>
(agreed btw with the encoder suggestion)
14:06
<foolip_>
annevk, http://www.ogcio.gov.hk/en/business/tech_promotion/ccli/download_area/mapping_table_2008.htm
14:06
<foolip_>
I also checked it into the git repo
14:08
<annevk>
k sweet
14:08
<annevk>
I'll take a stab at updating the big5 algorithm
14:09
<foolip_>
annevk, great, I'm quite sure this would be an improvement even with the missing mappings
14:09
<annevk>
foolip_: 8862 => <U+00CA,U+0304> is that a big5.json like index? the 8862 number?
14:09
<foolip_>
annevk, the bytes \x88\x62, I used the notation in the HKSCS-2008 mapping table there
14:09
<annevk>
ah
14:10
<foolip_>
(and in list of missing mappings)
14:10
<annevk>
right
14:11
<foolip_>
annevk, I'll have a look at trying to categorize the missing mappings and perhaps asking the Chinese HTML IG if they know more.
14:11
<foolip_>
happy easter!
14:12
<annevk>
ah yeah, that might be a good idea
14:12
<annevk>
you too!
14:12
<foolip_>
some of the missing mappings smell a lot like they could be de-facto big5 extensions or something
14:12
<annevk>
oh and btw, we might standardize this through the W3C i18n WG at some point
14:12
<foolip_>
mkay
14:12
<annevk>
there's a lot of big5 extensions
14:13
<annevk>
http://en.wikipedia.org/wiki/Big5 has a bunch on that front
14:13
<annevk>
http://coq.no/character-tables/chinese-traditional/en too
14:15
<foolip_>
annevk, thanks, I'll have a look
14:46
<annevk>
http://dvcs.w3.org/hg/encoding/raw-file/tip/Overview.html#big5
14:50
<davidb>
at
14:50
<Ms2ger>
mozilla.com?
14:51
<annevk>
I guess I should fetch that script from foolip
14:52
<annevk>
generate myself a big5 index
15:17
<Velmont>
Ms2ger: Ohmy. :-) I should also test it again, -- but I've been on vacation for a while now (still am) -- so don't have my nice setup that I've got at the office.
15:17
<Ms2ger>
Pff, vacations
15:18
<Ms2ger>
Those are good to get work done :)
15:18
<Velmont>
Hehe. -- Well, I managed to fsck up the only two lines I changed, missing a var and actually not doing what I said I should do in an older comment. :S
15:19
<Ms2ger>
That's better than messing up lines you didn't change
15:19
<Velmont>
Guess that's true ;}
15:37
<annevk>
hmm
15:37
<annevk>
the first 900 or so positions have no mapping
15:38
<annevk>
weird
15:38
<annevk>
I now understand why foolip was wondering about that a little more
16:12
<fkm>
Hi there. I have a problem with the validator not accepting code that is valid according to the specs. The validator tells me that "The text content of element time was not in the required format: The literal did not satisfy the date or time format. <span><time>1991</time> ...de la planète Mars</span>". But according to the specs, the time element accepts "A valid non-negative integer representing a year <time>2011</time>".
16:12
<fkm>
The page in question is http://dev.snemelk.ch/pages/bonfire/iam/artist.html
16:12
<fkm>
The specs I have consulted are http://www.w3.org/TR/html5/the-time-element.html and http://www.whatwg.org/specs/web-apps/current-work/multipage/text-level-semantics.html#the-time-element
16:12
<fkm>
Am I missing something or is this a bug in the validator? I tried validator.w3.org and also the underlying validator.nu to see if the problem is with the implementation by the W3C but both tell me the same thing.
16:15
<Philip`>
fkm: http://bugzilla.validator.nu/show_bug.cgi?id=913 sounds like it might be relevant
16:16
<Philip`>
MikeSmith may know
16:17
<[tm]>
fkm: i think the validator code is not yup to date with the spec
16:18
<fkm>
Ok. In that case, there is still hope for me yet :-)
16:18
<fkm>
Thank you both for the feedback!
16:18
<[tm]>
either file a bug or comment on whatever existing one there might be
16:19
<[tm]>
time element should really just be dropped
16:19
<fkm>
I have prepared an email to the W3C validator mailing-list. But I guess I could post it on the validator.nu bug-tracker.
16:19
<[tm]>
it was a mistake to add it to begin with
16:20
<[tm]>
fkm: yeah
16:21
<fkm>
I kind of like the time element. At least more than the new b, i and, u elements ;-)
16:21
<[tm]>
yeah well
16:21
<[tm]>
those aren't new
16:21
<Ms2ger>
annevk, so what are these code points supposed to map to now?
16:22
<annevk>
Ms2ger: not in a map, they'll become U+FFFD
16:22
<annevk>
Ms2ger: already implemented in Gecko
16:22
<Ms2ger>
Oh
16:22
<Ms2ger>
If I don't need to do anything, go ahead :)
16:22
<annevk>
well, not sure about trunk
16:22
<annevk>
did you change encodings already?
16:22
<Ms2ger>
I haven't changed much lately
16:24
<fkm>
[tm]: I know. But with HTML5 they have a new, "semantic" meaning. But it is not my place to question the HTML specs. I am too much of a rookie to go there.
16:24
<Ms2ger>
Hah
16:24
<Ms2ger>
It's the rookies that need to question the spec
16:24
<Ms2ger>
Because we all got used to it
16:24
<fkm>
I will keep that in mind :-)
16:25
<Ms2ger>
(Not on the meaning of b and i, though... Those have been discussed to death :))
16:26
<annevk>
fkm: it's always best to question things, but beware that you might not find logic, especially with web standards
16:26
<annevk>
I guess I could update my test
16:32
<[tm]>
if one of you in Belgium or Netherlands is interested in doing a three hour tutorial in Belgium this summer our fall, lemme know
16:32
<[tm]>
i doubt there is much money involved
16:33
<[tm]>
if any
16:34
<[tm]>
Benelux
16:34
<[tm]>
your Belgian villages have some crazy names
16:35
<annevk>
Benelux is actually 3 countries
16:35
<annevk>
not a place
16:35
<[tm]>
oh
16:36
<[tm]>
shows WTF i know
16:36
<annevk>
BElgium/NEtherlands/LUXembourg
16:36
<[tm]>
ah
16:37
<[tm]>
Hmm so i guess this is in the Netherlands
16:38
<[tm]>
+31
16:39
<[tm]>
man i can't super the crusade against alcohol
16:41
<annevk>
yeah that's NL
16:41
<[tm]>
alcohol is the only thing that makes attending most of these meetings and conferences tolerable
16:42
<[tm]>
telcons too
16:42
<TabAtkins>
It does kinda suck when half the conf occurs in bars that are too loud to have good discussion in, but feeling excluded by drinking at all is venturing too far into the "accomodate my personal preferences!" territory.
16:44
<annevk>
solve all the social problems!
16:45
<fkm>
Concerning the problem with the validator. Should I add my example as a testcase or an additional comment?
16:45
<[tm]>
the suggested alternative of after meetings at a coffee shop doesn't really cut it
16:45
<TabAtkins>
...huh. I have absolutely no idea what this "Hypertext Layout, Reading Speed and Comprehension" thread is about.
16:46
<TabAtkins>
(Also he doesnt' use serial comma ARGH)
16:46
<TabAtkins>
As far as I can tell, someone decided to just collect a bunch of articles and say "Hey, these articles describe things vaguely related to what CSS3 Text cares about. Enjoy!".
16:46
<[tm]>
that's like going to an AA meeting
16:47
<[tm]>
TabAtkins: is that from the www-style list?
16:47
<TabAtkins>
Yes.
16:48
<[tm]>
i think you're the only one still subscribed to that list
16:48
<TabAtkins>
hahaha
16:48
<hober>
not the only one
16:49
<[tm]>
ok two of you
16:51
<[tm]>
hober: btw you rock mightily for fighting the good fight on html WG CPs
16:51
<TabAtkins>
Indeed.
16:51
<TabAtkins>
I gave up.
16:51
<TabAtkins>
Not worth the effort when it's a coin-toss anyway.
16:52
<[tm]>
no comment
16:52
<[tm]>
expletive deleted
16:54
<hober>
[tm]: thanks! sometimes i think i'm the only one who takes seriously the charter requirement to seek convergence with the whatwg
16:55
<annevk>
Ms2ger: http://dump.testsuite.org/encoding/single-byte-test.html seems not much has landed
16:56
<Ms2ger>
Sounds right
16:57
<eseidel>
Ms2ger: what about feature strings?
16:58
<Ms2ger>
eseidel, we'd like to put a static list in DOM4
16:58
<Ms2ger>
And then lock the door and throw away the key
16:58
<TabAtkins>
Yes omg
16:59
<annevk>
http://wiki.whatwg.org/wiki/DOM_features has a start
17:00
<eseidel>
Ms2ger: I support
17:01
<hober>
yes, feature strings need to DIAF
17:01
<eseidel>
TabAtkins: ideally would like to only return YES for things which are requried for web compat
17:01
<Ms2ger>
Can either of you fill in the blanks for WebKit? :)
17:01
<eseidel>
Ms2ger: and NO for everythgin else
17:01
<Ms2ger>
Agreed
17:01
<TabAtkins>
Yeah.
17:02
<Ms2ger>
But first, data! :)
17:02
<eseidel>
Ms2ger: http://trac.webkit.org/browser/trunk/Source/WebCore/dom/DOMImplementation.cpp#L220
17:03
<eseidel>
Ms2ger: but it should be very very easy to write some simple js to test
17:03
<Ms2ger>
On the wiki! ;)
17:03
<eseidel>
Ms2ger: sorry, wiki's and I don't get along :)
17:03
<eseidel>
I'm a reader, not a writer :p
17:04
<annevk>
that is prolly sufficient
17:05
<annevk>
i'll just dump the link in the wiki
17:06
<Ms2ger>
eseidel, I love how it builds a hashset in isEvents3Feature, for one string :)
17:07
<annevk>
whoa, SVG went nuclear with features
17:08
<annevk>
what kind of code did they expect authors to write?
17:08
<Ms2ger>
Well
17:08
<shepazu>
annevk: huh?
17:08
<Ms2ger>
You can use them from markup too, iirc
17:08
<TabAtkins>
Yeah, the feature strings are used in <switch>
17:12
<eseidel>
Ms2ger: this is performance critcal code you're readnig here
17:12
<eseidel>
Ms2ger: we have our best men on it
17:12
<eseidel>
Ms2ger: namely, me. :) (I think I wrote that section)
17:13
<Ms2ger>
Ah, so that's why we're winning the browser wars ;)
17:14
<eseidel>
oh, I just thought we were all singing kareoke at the local browser pub
17:14
<eseidel>
(and SVG was drunk in the corner, spouting about fonts and feature strings)
17:14
<Ms2ger>
Then I'd feel excluded :(
17:14
<Ms2ger>
They don't like letting AIs into bars...
17:14
<TabAtkins>
You and bz can go have AI fun somewhere.
17:14
<TabAtkins>
(Pretty sure the thing I met at SXSW was just a skinvelope controlled by bz.)
17:15
<Ms2ger>
We've done that enough when he was building me
17:16
<Ms2ger>
Also, TabAtkins, when are you going to write the box tree spec? :)
17:17
<TabAtkins>
After I write the parsing spec, which I'm in the middle of right now.
17:17
<TabAtkins>
Or actually, roughly in the beginning, but still.
17:20
<[tm]>
He refused to accept vice in exchange for his lulz
17:25
<annevk>
shepazu: just that has so many
17:25
<annevk>
TabAtkins: kind of jealous now
17:27
<annevk>
"WAI-Engage is an open forum for responsive development of material
17:27
<annevk>
supporting web accessibility, including support for Web Accessibility
17:27
<annevk>
Initiative (WAI) resources."
17:27
<annevk>
I don't even know what that means...
17:41
<Ms2ger>
TabAtkins, including zcorpan's quirks stuff?
17:42
<TabAtkins>
Ms2ger: Um, sure!
17:42
TabAtkins
has to actually look at that spec.
17:45
<moo-_->
I have this really ambitious idea: I'd like that sites could suggest color and background colour for tabs besides the site icon. I know it's not going to happen very easily, but if it were to happen how I should take this idea forward. creating chrome + ff add-ons supporting this, promoting them like mad and then hope it is adopted by the browsers themselves some point?
17:46
<moo-_->
this would be mostly for usability. add-ons like colorfultabs do this already, but you need to manually configure in the sites.
17:46
<TabAtkins>
That sounds like a bad idea that I would hate.
17:47
<jamesr>
TabAtkins: "That's a bad idea. You should feel bad."
17:47
<Philip`>
Sounds like it should be as popular as IE's scrollbar colouring CSS
17:47
<TabAtkins>
Well, moo-_- shouldn't feel bad. I'm just saying, I would hate it.
17:47
<TabAtkins>
And since I'm clearly representative of all users...
17:47
<jamesr>
Philip`, you mean WebKit's? IE has that too?
17:47
<TabAtkins>
Pretty sure we copied IE there.
17:48
<moo-_->
TabAtkins: ok, let me ask then is there anything we could do to make tabs / windows / bookmarks / etc. more distinctive besides their favicon?
17:48
<Philip`>
jamesr: http://www.javascriptkit.com/howto/scrollbarcolor.shtml
17:48
<moo-_->
like all slashdot.org tabs would have (non-offending) green background
17:49
Philip`
didn't realise WebKit had done it
17:49
<jamesr>
Philip`, WebKit has pseudoes
17:49
<annevk>
moo-_-: there are some extensions for that
17:49
<jamesr>
http://www.webkit.org/blog/363/styling-scrollbars/:
17:49
<moo-_->
annevk: I am aware of that
17:49
<annevk>
given that Apple tries to get rid of the scrollbar creating pseudos for that was a silly idea
17:50
<moo-_->
annevk: but my point was more like that websites themselves could suggest the colours
17:50
<moo-_->
hmm
17:50
<annevk>
moo-_-: http://wiki.whatwg.org/wiki/FAQ#Is_there_a_process_for_adding_new_features_to_a_specification.3F
17:50
<annevk>
moo-_-: might help
17:50
<moo-_->
annevk: thanks
17:50
<moo-_->
this is a starting point, at least
17:50
<moo-_->
:)
17:50
<annevk>
moo-_-: if there's interest from authors, browser UI people, no phishing problems, no security problems, it might have a chance...
17:51
<annevk>
moo-_-: but I doubt it somewhat
17:51
<annevk>
moo-_-: the extension could use some information from the page as a hint for the color
17:52
<dglazkov>
good ${time_of_day}, Whatwg!
17:53
<moo-_->
annevk: yep. my first idea was to have (all) extension authors to agree on some way to extract this information from HTML <head> <meta>
17:57
<annevk>
moo-_-: fwiw, Opera has this experimental thing with just having large tabs that show thumbnails of the pages
17:57
<annevk>
I don't use it because I have a smallish screen, but it has some potential I think
17:58
<moo-_->
annevk: I tried that feature... nice but takes little screen estate
17:58
<moo-_->
annevk: IE9 + Win8 also do this tab bar thumbnail thing
17:58
<moo-_->
but it is not good for visually grouping which pages came from which sites
17:58
<moo-_->
you still need to "read" thumbnails
18:01
<Ms2ger>
Good, dglazkov
18:01
<dglazkov>
${time_of_day | "morning"}
18:16
<annevk>
hsivonen: I like how it says "Note: I’ve also added the prefixes versions –moz, –webkit & –o for Mozilla, Webkit & Opera to be compatible with the other browsers. Never forget to do the same!" and then continues to just do that
18:16
<annevk>
hsivonen: but it's better than most browser advocacy
18:22
<TabAtkins>
Hrm. I always forget what the \237 indicates in the CSS syntax definition of "non-ascii".
18:22
<TabAtkins>
(Defined as [^\0-\237].)
18:22
<annevk>
octal
18:23
<annevk>
which is 159 or 0x9F
18:23
<annevk>
iirc
18:24
<TabAtkins>
Yeah it is. I just don't understand what the significance of 0x9f is.
18:24
<TabAtkins>
Oh, I was looking at an extended ascii chart, not at unicode.
18:25
<TabAtkins>
A0 is where the real characters start again.
18:25
<annevk>
it should be 0x7F
18:25
<annevk>
well dunno
18:25
<annevk>
CSS is weird
18:25
<TabAtkins>
0x80 to 0x9F are all non-printable control characters.
18:25
<TabAtkins>
So I guess it's sensical to cut those out too.
18:25
<TabAtkins>
But it's not really "non-ascii" then. ^_^
18:26
<annevk>
well yeah, and there's a ton of other code points that are not actually "usable"
18:26
<TabAtkins>
Yeah.
18:26
<TabAtkins>
I'm guessing it was just a useful block to cut out, since it's adjacent to the ascii block.
18:26
<annevk>
I can already see them discussing this in a committee meeting
18:27
<annevk>
that's actually a rather funny thought
18:28
<TabAtkins>
hehe
18:29
<annevk>
only one decoder to go
18:29
<annevk>
never imagined it would take this long, but it's starting to look pretty neat
18:33
<TabAtkins>
Huh. Why is parsing so complicated?
18:33
<Hixie>
dialog.show()/showModal() or dialog.open()/openModal()? I was going to use show() but the convention is to have a close() method and onclose handler
18:33
<TabAtkins>
It looks a lot easier when you use a grammar.
18:33
<TabAtkins>
Using "open" has an interesting parallel with <details open>
18:34
<Hixie>
we are going to have <dialog open>
18:34
<Hixie>
maybe that's a good reason not to use open()
18:34
<Hixie>
it would clash with the IDL attribute reflecting open=""
18:34
<Hixie>
looks like most APIs use showModal()
18:34
<Hixie>
as opposed to openModal()
18:35
Hixie
decides to stick with show
18:35
<annevk>
TabAtkins: yeah, CSS parsing is vastly more complicated than it needs to be
18:35
<annevk>
TabAtkins: mostly because of the grammar and nobody looking at handwriting a tokenizer
18:36
<Hixie>
css parsing is also complicated because the grammar didn't define any of the error handling
18:37
<Hixie>
so the error handling became semi-random, like html's
18:37
<Hixie>
(well, it defined some error handling, but only within the "forward-compatible grammar" set)
18:38
<TabAtkins>
annevk: Really, the complication is just that it's an ad-hoc grammar with lots of options at a lot of points, rather than the relatively simply HTML grammar (ignoring quirks and scripts).
18:38
<Hixie>
good thing i wasn't drinking anything when i read _that_
18:38
<TabAtkins>
Hixie: Dude, go look at your description of the types of tokens. There's, like, 5 of them.
18:39
<Hixie>
ignoring quirks and scripts it's pretty simple, i agree :-)
18:41
<annevk>
I especially hate that escapes can occur everywhere
18:42
<annevk>
I would have removed that in 2000, if I was smarter back then
18:44
<Ms2ger>
Oh man
18:44
<Ms2ger>
If I had a time machine...
18:45
<Ms2ger>
Killing Hitler? Nah, I'd lock Bert and howcome in a room for a few hours
18:45
<TabAtkins>
...I'd create a Turing oracle and break open computation levels, brute-forcing my way to immediate godlike-AI.
18:45
<Hixie>
hah
18:46
<TabAtkins>
(And then be killed by it, because its value system doesn't recognize life as a useful thing.)
18:46
<TabAtkins>
(Or as a "thing" at all.)
18:53
<Hixie>
options for aligning these dialogs... use aps pos and define the static position as one that happens to result in the alignment i want; define that the element is in a positioning regime that ignores 'position' and 'float'; invent a new 'position' keyword that works like 'relative' but starts with the appropriate alignment rather than the static position (that one is harder since from CSS the new keyword would be useless without a way to define the link from one e
18:57
<Hixie>
hmm
18:58
<TabAtkins>
That cut off at "from one el" for me.
19:00
<Hixie>
from one element to another
19:01
<Hixie>
should it be possible to turn a <dialog> that's already open into a modal dialog? or should .showModal() do nothing or throw an exception if it is already open?
19:01
<TabAtkins>
So, the element() function provides a way to link one element to another. But right now it's defined to only take an id selector.
19:02
<Hixie>
(and should the answer be different if it's open and modal or open and not modal?)
19:02
<Hixie>
TabAtkins: the link between the elements is given by the argument to the show() method, it's not in CSS
19:02
<TabAtkins>
Hixie: Okay, so it's magic from CSS's perspective. That's fine.
19:02
<Hixie>
right
19:03
<Hixie>
i'm just saying a new keyword would be kinda silly since there'd be no good way to use it in other contexts
19:03
<Hixie>
new position keyword
19:03
<Hixie>
i'm thinking either these elements should be forced into a new regime, or i should just use abs pos with a redefined static position
19:03
<TabAtkins>
Actually, I think you should just define a new position value.
19:04
<Hixie>
what would it mean if used on a random div?
19:04
<TabAtkins>
It's just as magic, but it can be picked up by CSS later.
19:04
<TabAtkins>
I'm thinking it woudl be equivalent to 'relative'.
19:05
<Hixie>
could work
19:06
<Hixie>
hmm
19:06
<Hixie>
how do i have a different rule for dialogs that were given an alignment system and dialogs that should just center themselves?
19:06
<Hixie>
could be magic, i suppose
19:06
<TabAtkins>
Magic for now.
19:06
<TabAtkins>
I'll review after you lay down some text.
19:07
<TabAtkins>
To make sure it's not hostile to what I want to do with 'position'.
19:07
<TabAtkins>
Lunch now to grab an interview candidate.
19:07
<Hixie>
hehe
19:07
<Hixie>
later
19:15
<annevk>
http://www.w3.org/2012/02/timed-text-wg-charter
19:15
<annevk>
o_O
19:15
<Ms2ger>
Wow
19:15
<annevk>
"A Section Edition of the Timed Text Markup Language (TTML) 1.0 Recommendation"
19:15
<Ms2ger>
LC: N/A
19:16
<Ms2ger>
CR: N/A
19:16
<Ms2ger>
PR: N/A
19:16
<Ms2ger>
Rec: November 2012
19:16
<annevk>
boom
19:16
<annevk>
headshot
19:16
<Hixie>
why "o_O" ?
19:17
<Hixie>
it's pretty normal for wgs to have their charters extended like this...
19:17
<Hixie>
(by "normal" i mean "common place" not "sensible")
19:17
<Ms2ger>
Well, charters
19:17
<Ms2ger>
The former was already ruled out
19:18
<annevk>
given https://www.w3.org/2000/09/dbwg/details?group=34314 it's highly surprising this is being extended
19:18
<annevk>
the o_O was for W3C continuing to put effort into something the community is not really behind
19:18
<Hixie>
um...
19:18
<Hixie>
the w3c does that all the time
19:18
<Hixie>
for some definition of "community"
19:18
<Hixie>
e.g. rdf...
19:19
<Ms2ger>
Heh, I don't think I've seen "not in good standing" before
19:20
<annevk>
Hixie: true, this feels a bit different, more like another XHTML 2.0 group
19:21
<annevk>
maybe I should reply with some comments
19:22
<Hixie>
xhtml2 lasted years past the point of the community not being really behind it
19:22
<Hixie>
xforms' charter only ran out last month
19:23
<Hixie>
(and their timeline has dates until september 2012, so i doubt they'll be closed)
19:23
<Hixie>
just let the group do it's stuff, the alternative is they come and try to force their ideas into our stuff :-P
19:23
<Hixie>
its
19:25
<Ms2ger>
(Also, for the interested people without member access: https://www.w3.org/2000/09/dbwg/details?group=34314&public=1)
19:25
<annevk>
ah thanks Ms2ger, didn't know that was available
19:26
<Ms2ger>
"The participants list is available publicly"
19:26
<Ms2ger>
4th bullet
19:26
<annevk>
Hixie: fair enough
19:27
<annevk>
Ms2ger: fascinating
19:27
<Ms2ger>
No kidding
19:44
<fkm>
I have another question about the validator. I am not sure if I understood the concept of flow and phrasing content correctly. The way I understand it, flow content is the same as block elements and phrasing content is the same as inline elements or innerHTML. Following this reasoning having inline elements as direct descendants of the body should not be valid. But the validator does not complain.
19:44
<fkm>
Here's an example: http://dev.snemelk.ch/tests/html5-content_model.html
19:44
<fkm>
Is there a flaw with my reasoning or is this another bug (out of sync problem) in the validator?
19:46
<annevk>
you should read the definitions more carefully probably
19:46
<fkm>
Ok. I will do that :-) Thank you!
19:47
<annevk>
and try to forget about block/inline-level
19:48
<fkm>
I will try. But old habits die hard ;-)
19:51
<Hixie>
fkm: to find out what is allowed in <body>, go to the body element's definition in the spec, and check it's "content model" description
19:51
<Hixie>
fkm: follow the links from that to see what the terms mean
19:52
<fkm>
Ok
19:52
fkm
has much to learn :-)
19:52
<fkm>
Thank you both for the pointers!
21:58
<zewt>
heh, gmail's gui editor has become next to unusable
21:58
<zewt>
delete one line of quotes, the entire block disappears
22:00
<TabAtkins>
Solution: stop using a gui editor.
22:01
<TabAtkins>
Or use the gui editor in the way most people do, bottomquoting the entire preceding message.
22:01
<zewt>
that's too much of a pain in the ass with gmail, because if you don't send the mail in gui mode, it sends a horribly pre-wrapped 80-column mail
22:02
<zewt>
so i'd have to switch back and forth each time (or manually formatting-kill every time)
22:04
<Hixie>
pine.
22:05
<zewt>
gross
22:05
<zewt>
i think i used that in highschool, heh
22:05
<Hixie>
that's what i use for all my e-mail. :-)
22:06
<zewt>
that must be why it takes you six months to reply to mail :P
22:06
<Hixie>
nah, it's the other way around
22:06
<Hixie>
it'd be longer with a gui editor
22:14
TabAtkins
just realized he doesn't have to match the grammar exactly with his parser; he just has to capture the same behavior.
22:14
<TabAtkins>
Duh.
22:16
<TabAtkins>
In particular, the grammar goes through some contortions to express things like "selectors can be basically anything!" and to produce a good structure for error-recovery.
22:36
<Hixie>
heycam|away: you should provide a spec subscription model for webidl like i have for html :-)