16:25
<zewt>
dreaming of a world where dom events never had the bubbles and cancelable flags
19:44
<Hixie>
zewt: the cancelable flag is a particularly useless one
19:51
<Ms2ger>
Pff
19:51
<Ms2ger>
Half the web is useless
19:59
<Hixie>
i dunno about _that_
20:02
<zewt>
cancelable should be relegated to web-compatibility only and otherwise never even mentioned in specs
20:02
<zewt>
all events should be cancelable unless there's some weird web-compat reason otherwise
20:17
<Hixie>
zewt: it's less work for me to keep just speccing it as is (cancelable if canceling has an effect, not otherwise) than it would be to go through and change everything...
20:17
<zewt>
well, changing the ones that are implemented already is probably not worth it, but new events should just always be cancelable
20:18
<Hixie>
i think it'd be worse for half the events to make sense and the other half to be always cancelable than for us to continue to just do it
20:21
<zewt>
but always being cancelable makes more sense, since the whole concept of uncancellable events doesn't make sense :)
21:04
<Hixie>
zewt: sure it does
21:04
<Hixie>
zewt: i mean, it's not useful, but it makes sense
21:04
<Hixie>
zewt: it just means that the code that invoked the event will not check whether the event was canceled or nat
21:21
<gsnedders>
hsivonen, jgraham, abarth: do we really want tests for <main> in html5lib when it's not in the spec
21:31
<gsnedders>
Also, wow, nice. We now have the tokenizer depending on the parser. Yay.
21:32
<Hixie>
for what?
21:33
<gsnedders>
In html5lib. For the CDATA state.
21:33
<Hixie>
is that all?
21:33
<Hixie>
i thought the parser could switch the tokeniser state for other things too
21:33
<gsnedders>
Yeah, otherwise the parser just manipulates the tokenizer state.
21:33
<Hixie>
e.g. <plaintext>
21:33
<Hixie>
ah
21:34
<Hixie>
what's the difference here?
21:34
<gsnedders>
Tokenizer looking at the parser, not parser mutating tokenizer.
21:35
<Hixie>
ah, k. i guess i forget what's special about cdata.
21:35
<Hixie>
it's been a whiel.
21:35
<Hixie>
while.
21:35
<gsnedders>
Foreign content is top of the stack.
21:35
<Hixie>
ah, right
21:35
<Hixie>
could probably be implemented as state the other way too
21:35
<gsnedders>
Or bottom of the stack, depending on who's listening.
21:35
<Hixie>
oh man, that's something i should have listed as one of my biggest mistakes
21:36
<Hixie>
what a mess
21:37
<abarth>
gsnedders: The <main> thing seems like a tempest in a teapot
21:38
<abarth>
gsnedders: I'm just ignoring it figuring that it doesn't matter
21:40
<gsnedders>
abarth: I'm just somewhat irked by the tests failing :P
21:41
<abarth>
can we move them to a "proposed" directory or something?
21:42
<gsnedders>
Not really. html5lib just names tests *.dat, and there's little attention paid to file names in most runners, AFAIK.
21:45
<abarth>
I mean that we could move the "dat" file to a different directory
21:45
<abarth>
so that it's not run until the behavior it is tested is "accepted" in some sense
21:46
<gsnedders>
Yeah, I guess. hsivonen: you care if we move the main tests to a separate file in a proposed folder?
22:10
<Hixie>
do browsers not implement Element.querySelector()?
22:12
<Hixie>
nevermind, i'm a tool