01:00
<rniwa>
Hixie: yt?
02:53
<MikeSmith>
Hixie: can you please check again and see if cvs is still slow for you?
05:57
<annevk>
Hixie: your unrestricted double change looks wrong
05:58
<annevk>
Hixie: you removed the generic NaN/Inifity language and then changed double to unrestricted double (unrestricted double is the one that allows NaN and such)
05:58
<Hixie>
i only changed it for ones where the previous language didn't apply
05:59
<Hixie>
at least, that was my intent
05:59
annevk
looks again
06:02
<annevk>
Hixie: e.g. the scale() method on <canvas> has no language for it
06:02
<annevk>
Hixie: setLineDash does have language for it, but it seems better if we make that double
06:03
<annevk>
same for globalAlpha, if that's still feasible
06:03
<annevk>
and similar methods
06:05
<Hixie>
all the canvas ones have language because there's the default language, no?
06:05
<Hixie>
oh, i guess methods don't
06:05
<Hixie>
but for the transforms there's no need for language
06:05
<Hixie>
you just do the maths with infinities
06:05
<Hixie>
there's separate language for problematic transforms iirc
06:06
<Hixie>
i guess we should add text for NaNs for scale()
06:06
<annevk>
I went through the spec for Infinity and missed that then...
06:07
<annevk>
I only see it for linedash, gradients and globalAlpha
06:07
<annevk>
but maybe it does not need to be explicitly mentioned?
06:07
<Hixie>
filed https://www.w3.org/Bugs/Public/show_bug.cgi?id=17468 for transforms
06:08
<Hixie>
please file other bugs for others
06:08
<Hixie>
gotta go
06:09
<annevk>
kk
06:28
<annevk>
so we're the Web Notification WG with http://www.w3.org/2010/web-notifications/ as home page
06:28
<annevk>
:)
06:32
<annevk>
jgraham: where is that email from bz to public-html?
06:32
<annevk>
jgraham: oh last month
06:33
<annevk>
jgraham: but that one cc'd whatwg
08:26
<zcorpan>
Hixie: the sync image load uses the state of the crossorigin attribute, but the potentially CORS-enabled fetch algorithm ignores the state of that attribute when the URL is same-origin
09:19
<annevk>
matjas: I guess theoretically you could add Registry-HTTP-headers to http://wiki.whatwg.org/wiki/Category:Registries
09:19
<annevk>
matjas: we haven't really invested effort into taking over that IANA stuff however
09:20
<annevk>
matjas: you could also try registering the header through IANA, but that process sucks balls
09:20
<matjas>
annevk: yeah, the latter option doesn’t sound like fun
09:20
<annevk>
(which is why everyone keeps minting X- prefixed headers instead; joy)
09:21
<matjas>
facebook spam: http://wiki.whatwg.org/wiki/Registry-HTTP-headers (by User:Kirayamato)
09:21
<annevk>
whoa
09:23
<annevk>
nuked
09:48
<matjas>
annevk: so can I just add an entry for X-UA-Compatible to the Registry-HTTP-headers page, or what?
09:52
<annevk>
well, 1) although IANA sucks our alternative has not really proved successful thus far (mostly due to lack of effort I think) 2) the HTML specification would need to be updated either way for the registration procedure 3) it would be interesting to hear what hsivonen_ has to say about this issue 4) you could start by adding it to http://wiki.whatwg.org/wiki/PragmaExtensions and mention it is widely used but you cannot be bothered to deal with the
09:52
<annevk>
pain that is IETF registration procedures for IANA registration
09:57
<matjas>
annevk: ok, done: http://wiki.whatwg.org/wiki/PragmaExtensions I’ll await hsivonen_’s feedback before I submit a spec “bug”
10:01
<karlcow>
there is an RFC against the proliferation of X-headers
10:01
<karlcow>
it is a similar story than and -vendor- extensions
10:44
<sm0lm>
Anyone know where the source code of the OPAL selectoracle can be obtained?
10:46
<Stevef>
[tm]: where are you at brotherman?
11:04
<jgraham>
Hixie: (mail sent)
11:04
<jgraham>
Also, I added some more tests for navigation in beforeunload and pagehide, which seem to work the same as in unload
11:41
<sm0lm>
Anyone know where the source code of the OPAL selectoracle can be obtained?
11:53
<annevk>
is http://mxr.mozilla.org/mozilla-central/source/netwerk/base/src/nsStandardURL.cpp Gecko's URL implementation?
12:46
<MikeSmith>
annevk: about the URL spec, Chris Weber has aggregated a bunch of tests at https://github.com/cweb/iri-tests
12:47
<MikeSmith>
and he has a report of some results at https://github.com/cweb/iri-tests/blob/master/results/uri-iri-results.xls
12:52
<annevk>
MikeSmith: reading that it sounds like he made the WebKit tests worse...
12:52
<MikeSmith>
oh
12:52
<MikeSmith>
how so?
12:53
<annevk>
removed e.g. <base> information
12:53
<annevk>
do you have pointers to Adam's original tests?
12:54
<annevk>
https://github.com/abarth/url-spec I guess
13:11
<annevk>
MikeSmith: pubrules throws a fit
13:11
<MikeSmith>
oh
13:11
<MikeSmith>
I will get those fixed
13:12
<MikeSmith>
and I will look from Adam's tests right now
13:12
<annevk>
I might have made a typo that threw pubrules into a fit
13:13
<annevk>
yeah was a typo
13:17
<annevk>
MikeSmith: so yeah, my plan is to work on URLs now
13:17
<MikeSmith>
ok
13:18
<MikeSmith>
the Webkit tests are at http://trac.webkit.org/browser/trunk/LayoutTests/fast/url I think
13:18
<MikeSmith>
not sure if those are essentially the same as Adam's github ones or not
13:18
<MikeSmith>
anyway, I will help with URLs as well
13:28
<MikeSmith>
annevk: about Notifications
13:28
<MikeSmith>
http://dvcs.w3.org/hg/notifications/raw-file/tip/TR.html is what we want to publish, right?
13:28
<annevk>
MikeSmith: yeah
13:28
<MikeSmith>
OK
13:29
<MikeSmith>
I will copy that to the dated TR URL
13:29
<annevk>
is there a problem?
13:29
<MikeSmith>
no
13:29
<annevk>
k
13:29
<MikeSmith>
just in your message to Denis, you had mentioned http://dvcs.w3.org/hg/notifications/raw-file/tip/Overview.html
13:31
<annevk>
ah yeah, I did not have the TR copy at that point
13:31
<annevk>
just the editor's draft
13:53
<MikeSmith>
annevk: OK, I have it up at http://www.w3.org/TR/2012/WD-notifications-20120614/
13:53
<annevk>
thanks
13:53
<MikeSmith>
and I'll work with Denis if anything more needs fixing for pubrules or whatever
13:59
<annevk>
Hixie: I want to change the email address for my account on the HTML spec
13:59
<annevk>
Hixie: to annevk⊙an
14:02
<annevk>
MikeSmith: IRI tests look nice too though
14:02
<MikeSmith>
yeah?
14:02
<MikeSmith>
I guess Chris did all those
14:02
<annevk>
MikeSmith: I was thinking of writing a URL parser in JavaScript and then using tests to compare it with browsers
14:03
<annevk>
MikeSmith: and then writing a spec based on the structure of that URL parser
14:04
<annevk>
MikeSmith: the current spec suffers from quite a few problems; e.g. "s<:s" is a relative URL, not a URL whose scheme is "s<"
14:04
<MikeSmith>
oh
14:04
<annevk>
so starting anew is probably best
14:05
<MikeSmith>
annevk: you might want to look at http://medialize.github.com/URI.js/ to see if you can repurpose some of it or get some ideas from it
14:05
<MikeSmith>
from Rodney Rehm
14:05
<MikeSmith>
https://github.com/medialize/URI.js
14:11
<annevk>
volkmar: "We are willing to discuss any detail of the API as long as we do not include any feature we believe are out of scope." seems somewhat poorly worded
14:12
<annevk>
MikeSmith: doesn't look that usable
14:12
<MikeSmith>
OK
14:12
<volkmar>
annevk: it might be indeed
14:12
<annevk>
but I should indeed look around first
14:13
<volkmar>
annevk: but I believe what I meant is undertandable (at least I hope)
14:16
<annevk>
volkmar: it seems kind of poor form to restrict the discussion to what you believe it should be restricted to
14:29
<AryehGregor>
annevk, [120612 17:30:48] <NeilAway> AryehGregor: I think it's Gecko's only URL type, yes, but we have several URI implementations too ;-)
14:30
<annevk>
AryehGregor: :(
14:30
<AryehGregor>
Are you surprised?
14:31
<annevk>
I blame the IETF
14:31
AryehGregor
blames Netscape
14:31
<volkmar>
annevk: it's indeed poorly worded, we doesn't want to block the discussion
14:31
<volkmar>
the idea is much more that unless proven useful, we don't want to handle those use cases
14:32
<annevk>
okay, cool
14:41
<annevk>
MikeSmith: why do backreferences not work in the URL spec?
14:41
<MikeSmith>
annevk: probably because I muffed something up
14:42
<MikeSmith>
I'll take a look right now
14:43
<MikeSmith>
"Uncaught TypeError: Object function Object() { [native code] } has no method 'push' dfn.js:26"
14:56
<gsnedders>
So, #whatwg, what should be done with Error.stack, given sites rely upon different JS engines having different strings?
14:57
<Philip`>
Add Error.stdStack
14:57
<Philip`>
which is standardised between browsers
14:58
<zcorpan>
gsnedders: [citation needed]
14:58
<Philip`>
(and which is much more convenient to use, e.g. not requiring you to do string parsing to extract data from it)
14:58
<Philip`>
(so that most people will start using that instead of Error.stack, and then nobody will care that Error.stack is inconsistently implemented)
14:59
<gsnedders>
zcorpan: Which part?
15:00
<gsnedders>
On the whole it looks like Error.stack won't be spec'd and a new API introduced that is sane
15:02
<annevk>
gsnedders: not sure what you want #whatwg to do here
15:04
<zcorpan>
gsnedders: what sites?
15:09
<gsnedders>
annevk: Is creating a new API the right way to go, or should most/all vendors break sites?
15:10
<zewt>
(the latter sounds unlikely)
15:10
<zewt>
little's more annoying than not being able to fix a bug because somebody broke your bug reporting :)
15:11
<annevk>
gsnedders: both sound annoying :)
15:12
<zewt>
creating a non-crappy API and ignoring the completely uninteroperable ones in there now seems the way to go
15:28
<AryehGregor>
gsnedders, that implies that there will indefinitely be a compat constraint on browsers that they have to implement the old .stack in whatever manner the legacy pages expect, but it won't be documented. That sounds like a very bad idea.
15:30
<jgraham>
AryehGregor: The compat constraint is something like "if you have the string 'webkit' in your UA, you must return a string formated like X. Otherwise, if you have the string 'gecko' in your UA header, you must return a string formatted like Y, otherwise..."
15:31
<jgraham>
(except probably more complex than that)
15:31
<AryehGregor>
jgraham, well, if that's really needed for site compat and people aren't willing to deploy site-specific workarounds, then that needs to be specced. You can't just have all these pages break in some random new browser that's written per spec.
15:31
<AryehGregor>
Better would be if browsers hacked around known bad pages to reduce compat impact enough that they could change to match the standard.
15:32
<gsnedders>
AryehGregor: the sites already do UA sniffing, so they're already broken in random new browsers.
15:32
<zewt>
can't blame pages for this, heh
15:32
<AryehGregor>
gsnedders, right, so new browsers have to clone an existing UA string, which is what they do in real life.
15:32
<jgraham>
gsnedders: Unless a random new browser implemented some other browser's US header exactly
15:32
<AryehGregor>
But they still need to know how to behave.
15:32
<jgraham>
*UA
15:32
<AryehGregor>
Given the UA string they picked.
15:32
<gsnedders>
Also it's unclear what to do given strict-mode and tail call optimization.
15:33
<gsnedders>
And overspecification could make TCO impossible, which would be bad.
15:33
<jgraham>
(it may not even be UA string based sniffing of course)
15:33
<zewt>
gsnedders: specify how to expose stack frames, but not what the stack frames will be (inlining could also cause frames to disappear, though that can probably be hidden by implementations with some effort)
15:44
<annevk>
meh
15:44
<annevk>
test://test.com/
15:44
<annevk>
does this have a host or not?
15:44
<zcorpan>
"Nobody; OK to take it and work on it"
15:45
<annevk>
.host returns "" in Opera/Firefox; test.com in Safari
15:46
<zewt>
annevk: doesn't the former imply that you could never add new schemes without breaking compatibility? (for what that's worth--not like the addition of a new protocol itself wouldn't have bigger side-effects)
15:53
<annevk>
zewt: seems like it has not been a problem in practice when we added ws: and wss:
15:54
<Ms2ger>
Am I alone in reading "twss"?
16:00
<zcorpan>
annevk: ws: and wss: aren't supported in <a href> so the URL decomposition attributes should treat it as an unsupported scheme, i think
16:04
<Ms2ger>
Whoa, fantasai *and* Hixie on a replying spree is no good for my inbox
16:06
<AryehGregor>
Isn't it great how we have a generic URL standard so that you can parse URLs without having to know scheme-specific details?
16:13
<jgraham>
pimpmyspec.net will be down for a while
16:15
<zewt>
for additional pimping?
18:15
<Ms2ger>
ojan, I don't care much, but I don't think moving mailing lists gains us much
18:16
<ojan>
Ms2ger: yeah, maybe not. i find it frustrating that in practice i need to check both mailing lists anyways because enough threads are incorrectly started on www-dom
18:16
<Ms2ger>
correctly, you mean ;)
18:16
<ojan>
lol, yeah
18:17
<Ms2ger>
I bet Bj�rn will have a say about it
18:17
<ojan>
Ms2ger: i'm not gonna make a big deal of this, but if everyone agrees, then there's no harm
18:17
<ojan>
i certainly don't intend to have a long dicussion about this
18:17
<Ms2ger>
If you want to have a long discussion about it, have it over at public-html? :)
18:23
Ms2ger
hadn't seen someone use "L.S." for a while
18:26
<TabAtkins_>
What's that mean?
18:28
<Ms2ger>
It's Latin, you won't like it
18:29
<TabAtkins_>
Hey, I got nothing against Latin itself. Just against the ridiculous attachment that some grammar assholes from a century ago had to it.
18:31
<hober>
TabAtkins_: you are tilting at windmills. English is weird. Get over it.
18:33
<TabAtkins_>
Hey, the faux-latin bullshit started because of people tilting at windmills. ^_^
18:33
<gsnedders>
TabAtkins_: On the other hand, there is a lot of English which has Romance influence, such as bases. ^_^
18:34
<TabAtkins_>
English stole tons of words from romance languages, yeah. But not its grammar.
18:35
<Ms2ger>
Did you mean "not it's grammar"?
18:35
<gsnedders>
Ms2ger: the Genitive form of it is its.
18:35
<Ms2ger>
Troll succeeded
18:36
<gsnedders>
TabAtkins_: Where does grammar end and lexis begin?
18:36
<gsnedders>
Ms2ger: I thought that was probably what you were doing.
18:37
<gsnedders>
You can trivially argue that closed-word classes (pronouns, for example) are grammatical.
18:37
<TabAtkins_>
gsnedders: I'm only a casual linguist on the internet, so I'm not prepared to argue to that level of precision.
18:39
<gsnedders>
TabAtkins_: Certainly English grammar has to some extent been influenced by Latin (which isn't that surprising, given English's status for the most of the first two millenia)
18:40
<Philip`>
Did you mean "millennia"?
18:40
<gsnedders>
Yes.
18:41
<gsnedders>
(Though yes, the influence on the lexis was far greater)
18:42
<gsnedders>
Today's fun fact: "they" and "them" are derived from Old Norse and not Old English.
18:43
<AryehGregor>
gsnedders, "bases" is Greek, not Romance.
18:43
<AryehGregor>
Oh, it's both.
18:43
<AryehGregor>
Interesting.
18:44
<AryehGregor>
Although it apparently didn't enter English through Romance languages -- it came into Middle English through Latin directly, according to my dictionary.
18:44
<Philip`>
("bases" as in plural of "base" or of "basis"?)
18:44
<gsnedders>
AryehGregor: Very little came from Greek directly.
18:44
<gsnedders>
Philip`: basis
18:45
<gsnedders>
AryehGregor: And Latin is a Romance language.
18:46
<AryehGregor>
gsnedders, not according to Wikipedia!
18:46
<Ms2ger>
Fix Wikipedia!
18:47
<AryehGregor>
"The internal classification of the Romance languages is a complex and sometimes controversial topic which may not have a unique answer. Several classifications have been proposed, based on different criteria."
18:54
<gsnedders>
I'd say the prototype of a language group is part of the group, *shrugs*.
19:31
<jgraham>
Italian must be a romance language, otherwise how would you explain Berlusconi
19:33
<sm0lm>
when are media fragment uris set to get widespread support?
19:46
<jgraham>
http://tools.ietf.org/html/draft-tbray-http-legally-restricted-status-00
19:48
<micheil>
jgraham: how do I say that there's an issue in that document?
19:49
<micheil>
quoting: "The restrictions typically most directly affect the".. make's little sense
19:52
<jgraham>
micheil: I'm not really sure. Perhaps the link marked "email" at the top of the document? Otherwise work out which WG is working on it
19:52
<micheil>
:/