00:33
<zewt>
apis without specs can be comedic
00:35
<zewt>
x = otherwindow.console.log; x("log"); worked in firefox and ie9 but not webkit, and now we have webkit people wanting to change to making it work while ie10 apparently (untested) doesn't allow it anymore
00:35
<zewt>
bleh, guess i should set up a vm for ie10
00:40
<zewt>
why does google let people run ads for "ie10" that have titles "FREE INTERNET EXPLORER 10" that go to random other things, is there no screening at all
01:04
<heycam>
Hixie, so my opinion on the DOMStringMap thing is that it would be nicer if it were [OverrideBuiltins], so you don't get the other properties from Object.prototype shining through
10:28
<Ms2ger>
http://lists.w3.org/Archives/Public/www-archive/2013Apr/ appears to suggest a ttwf is in progress
10:37
<MikeSmith>
Ms2ger: yeah in Seattle
11:06
<smaug____>
" WebIDL systematically leads to APIs hostile to the working JavaScript programmer." is not very helpful
11:07
<annevk>
where is that from?
11:07
<annevk>
oh you're reading that thread too
11:08
<annevk>
I try to read past the blame game
11:08
<annevk>
it's not very interesting
11:11
<smaug____>
I'm trying to understand the complaints
11:11
<smaug____>
but since the complaints are just "this is not good"
11:11
<smaug____>
it is a bit hard to argue anything
11:11
Ms2ger
has long given up on those trolls
11:14
<annevk>
smaug____: they have been concrete though and given feedback
11:14
<annevk>
smaug____: https://www.w3.org/2006/02/lc-comments-tracker/43696/WD-proximity-20121206/2740
11:14
<annevk>
smaug____: part of the problem is us
11:15
<annevk>
smaug____: we haven't done early outreach and just shipped a bunch of broken APIs
11:15
<annevk>
smaug____: I can totally understand that it makes it feel kinda futile to then make concrete proposals for an alternative
11:18
<smaug____>
well, as an example, it isn't clear what the event handling API should look like
11:19
<smaug____>
it doesn't have to be what jQuery does
11:19
<annevk>
the high-level comment there is that what we have now for these sensor APIs is very bad
11:20
<annevk>
registering a handler activates the sensor
11:20
<annevk>
and removing the handler deactivates it
11:20
<annevk>
event listeners should not have such side effects at all
11:20
<smaug____>
why not
11:20
<smaug____>
if it is not visible to the API user
11:21
<smaug____>
that *if* is rather important here :)
11:21
smaug____
goes to check what sensors API does
11:22
<Ms2ger>
Yay, DAP
11:22
<annevk>
it's totally visible
11:22
<annevk>
Ms2ger: more like yay dougt
11:22
<Ms2ger>
Ugh, dougt
11:22
<smaug____>
editor is Dzung
11:23
<annevk>
smaug____: most of those APIs were copied straight from B2G
11:23
<smaug____>
ah
11:23
<Ms2ger>
You can hardly blame the B2G APIs on WebIDL
11:23
<Ms2ger>
They all use XPIDL
11:23
<annevk>
http://dev.w3.org/geo/api/spec-source-orientation.html is the first one that happened I think
11:23
<smaug____>
uh
11:23
<smaug____>
https://dvcs.w3.org/hg/dap/raw-file/tip/sensor-api/Overview.html is totally non-sense
11:24
<smaug____>
The event handler passed to addEventListener must provide the signature: interface SensorCallback {
11:24
<smaug____>
void ondata (double value, double min, double max);
11:24
<smaug____>
};
11:24
<annevk>
smaug____: haha
11:24
<Ms2ger>
Note "The Device APIs Working Group decided to put this general Sensor API specification on hold and work on each sensor separately (see e.g. the Device Proximity)."
11:24
<annevk>
smaug____: fwiw, I might sensor APIs in a more general sense
11:25
<annevk>
smaug____: things like proximity, orientation, etc.
11:25
<annevk>
s/might/meant/
11:25
<smaug____>
Ms2ger: oh, I should read that boilerplate
11:25
<smaug____>
but good
11:26
<Ms2ger>
It might perhaps be helpful if these APIs weren't all developed in small groups without Web API design experience
11:29
<jgraham>
There is a second problem which is that engine constraints mean that we sometimes can't design the APIs that JS programmers would design
11:29
<smaug____>
we should have kidnapped Andrei in London (saw him when we were going to the Thai-restaurant) and make him to fix geo spec
11:30
<jgraham>
e.g. the weblinkit adversion to having C++ own js objects
11:39
<Ms2ger>
Is weblinkit what we're calling it now?
11:53
<annevk>
smaug____: oh he was there? missed that
14:03
<gsnedders>
jgraham: I don't care what you believe. :)
14:09
<jgraham>
I thinkn forcing imports to be alphabetial is just silly :)
14:10
<gsnedders>
jgraham: I think using an abstract concept of generality is silly.
14:12
<jgraham>
I think being anal about import orders is silly
14:13
<jgraham>
Just move the from foo import bar to before the next empty line and stop discussin git :)
14:13
<jgraham>
*discussing it
14:13
<gsnedders>
I already did.
14:13
<jgraham>
You seem to still be discussing it :)
14:13
<gsnedders>
And I broke the build. :)
14:14
<gsnedders>
I should really run tests locally first. :)
14:14
<jgraham>
Oh, missing a comma
14:15
<gsnedders>
Yeah, pushed fix.
14:16
<jgraham>
So are you planning to rebase this? I have no idea if/how github/critic would cope with that
14:17
<gsnedders>
Will rebase then close pull request manually.
14:17
<jgraham>
OK
14:17
<gsnedders>
At least I think I have to do it manually.
14:17
<gsnedders>
But waiting for Travis CI first. I haven't tested the fixes on Python 3.
14:17
<jgraham>
Yeah, i have the feeling that history rewrites on PR branches wouldn't work so well
14:17
<gsnedders>
Could try. :)
14:18
<gsnedders>
But non-fast-forward stuff is always evil to push.
14:19
<jgraham>
Well it's a rather common critic workflow, as well you know (and typically doesn't cause any problems)
14:19
<gsnedders>
(`pip install lxml` is a real bottleneck on Travis CI, sadly.)
14:19
<gsnedders>
jgraham: But that's the critic repo, which people don't often fetch
14:22
<gsnedders>
Getting us actually passing tests with all the optional extras is making me feel that html5lib might, you know, work.
14:24
<Ms2ger>
It does?
14:27
<jgraham>
gsnedders: I know *why* it works
14:27
<jgraham>
:)
14:29
<jgraham>
So where are all the TestTWF people?
14:30
<jgraham>
Oh, I guess it is only 7:30 am
14:30
<jgraham>
So possibly asleep
14:33
<jgraham>
Someone should tell them to go on IRC in #testing or something. Although I don't know how much I'll be around to help
14:34
<jgraham>
Hopefully a bit at least
14:35
<Ms2ger>
I don't know if I should be around, in that case ;)
14:36
<gsnedders>
jgraham: Also now pushed huge commit removing all trailing whitespace :)
14:37
<jgraham>
Yay! No more red!
14:38
<Ms2ger>
So is github the canonical repo now?
14:39
<jgraham>
Yes
14:40
<gsnedders>
Ms2ger: Well, four repos. :)
14:41
<Ms2ger>
Yay.
14:42
<gsnedders>
If you find anything bogus in the history, file a bug. And then we'll have to decide whether it's worthwhile rewriting history to remove.
14:52
<Ms2ger>
annevk, I get a "term not defined: dom manipulation task source" when generating dom
14:52
<annevk>
Ms2ger: hmm sorry about that
14:52
<annevk>
Ms2ger: I have not updated xref yet because the migration of CORS is not entirely complete
14:52
<Ms2ger>
Ah, I see
14:52
<Ms2ger>
No worries
14:53
<Ms2ger>
I was just looking if html5lib still worked
14:53
<annevk>
k
15:11
<Hixie>
03:21 < heycam> Hixie, so my opinion on the DOMStringMap thing is that it would be nicer if it were [OverrideBuiltins], so you don't get the other properties from Object.prototype shining through
15:11
<Hixie>
er
15:11
<Hixie>
mispaste. that was for the bug.
15:11
<Hixie>
what i meant to paste here was: http://www.reddit.com/r/Python/comments/1c8wzq/htmltagpy_a_new_module_to_wrap_content_in_html/c9eclnh?context=3
15:16
<Ms2ger>
Oh, so *that*'s what's been messing with your productivity ;)
15:19
<Hixie>
messing with my productivity? have you _seen_ the chart for the last week? :-P
15:19
<Hixie>
i've been kicking ass! :-P
15:20
<Ms2ger>
Last week, yes ;)
15:20
Ms2ger
has noticed that in his inbox too
15:21
<Hixie>
hehe
15:21
<Ms2ger>
I wish all weeks were that way :)
15:24
<annevk>
Hixie: fwiw http://angularjs.org/ is what Google promotes, prolly under the "HTML5" banner and allows custom elements, much like web components, which come to think of it, are probably also promoted as "HTML5"
15:31
<Hixie>
annevk: yeah, i know
15:48
<jgraham>
gsnedders: Speaking of people pushing non-ff updates to remotes, deleting remote branches it's tracking upsets critic a little
15:48
<jgraham>
By which I mean "I get email about it"
15:48
<jgraham>
Please at least close the review so that the tracking stops :)
15:51
<gsnedders>
jgraham: It already is closed.
15:58
<jgraham>
Huh, so why is it still tracking?
16:00
<jgraham>
Well when I close it it stops tracking... I guess I should either read the code or ask jl
16:14
<GPHemsley>
annevk: Do you know off-hand what encodings have รง encoded at 0x87?
16:15
<GPHemsley>
annevk: (Search for "07390000" in http://memory.loc.gov:8081/ll/llhb/030/llhb030.txt for a live web example)
16:21
<GPHemsley>
annevk: Oh, hint: IBM code page 850 has it. Not in the Encoding spec, but available in Firefox.
16:21
<Hixie>
wtf https://www.w3.org/Bugs/Public/buglist.cgi?query_format=advanced&list_id=7641&longdesc=213.37.128.77&longdesc_type=allwordssubstr
16:21
<Hixie>
this one IP has filed four of these bogus bugs with just selected text
16:21
<Ms2ger>
Does that URL say "longdesc"?
16:23
<Hixie>
longdesc is what bugzilla calls a comment internally
16:39
<Hixie>
how dod i compare to arbitrary unicode strings in js, with case folding?
16:40
<Ms2ger>
foo.toLowerCase() === bar.toLowerCase()?
16:42
<Hixie>
is thought that wasn't good in all languages or something
17:41
<gsnedders>
jgraham: I am very alarmed. html5lib.treewalkers.etree throws ImportError on loading, yet all tests pass.
17:42
<odinho>
Not really 100% coverage there? :P
17:46
<Philip`>
If it couldn't load any of the tests, then none of them failed, therefore they all passed - seems perfectly logical
17:48
<gsnedders>
tl;dr: keep your try blocks small to not catch exceptions you don't mean to
17:50
<gsnedders>
Is there any way to get text outside of the html element? It's always reparented, right?
17:53
<bholley>
Hixie: yt?
17:54
<Hixie>
bholley: yeah, sup
17:54
<bholley>
Hixie: was just curious about dialogArguments
17:54
<bholley>
The dialogArguments IDL attribute, on getting, must check whether its browsing context's active document's origin is the same as the dialog arguments' origin. If it is, then the browsing context's dialog arguments must be returned unchanged. Otherwise, if the dialog arguments are an object, then the empty string must be returned, and if the dialog arguments are not an object, then the stringification of the dialog arguments must be returned.
17:54
<bholley>
(from the spec)
17:54
<bholley>
Hixie: why do we have to stringify in the non-same-origin case?
17:54
<Hixie>
one sec, switching computers and have to reboot the one i'm switching to. brb.
17:55
<bholley>
Hixie: np
17:57
<Hixie>
k, back
17:57
Hixie
reads
17:59
<Hixie>
oh it's just because otherwise you can leak cross-origin access
17:59
<Hixie>
e.g. if it contains a DOM Node
17:59
<Hixie>
you've just leaked that entire origin
18:00
<bholley>
Hixie: but we're talking about the primitive case
18:00
<bholley>
Hixie: objects are the empty string, right?
18:01
<Hixie>
oh, hm
18:01
bholley
reads the dichotomy as object vs primitive, and assumes that nodes would fall into the object case, but might be missing a subtlety
18:01
<Hixie>
no, i don't think you are
18:02
<Hixie>
maybe it's what browsers do?
18:02
<Hixie>
or maybe i was just absent-minded when speccing that...
18:02
Hixie
looks at blame
18:02
<bholley>
not what Gecko does, haven't checked anything else
18:04
<bholley>
Hixie: I'm asking because I'm trying to simplify our whole setup for this stuff so that we can just do the origin check directly on the jsval when it comes time to define it on the global. But only objects (non-primitives) have an origin associated with them, so it's difficult to know whether to stringify or not in the primitive case
18:05
<Hixie>
looks like this text dates back to the first introduction of the text in the spec
18:05
<Hixie>
so it's probably just a mistake
18:06
<bholley>
Hixie: \o/
18:07
<bholley>
Hixie: want me to file?
18:08
<Hixie>
i'd like us to test other browsers first
18:08
<Hixie>
well, before we change the spec, anyway
18:08
<Hixie>
no harm having a bug on file to track it
18:09
<bholley>
Hixie: ok, I'll write a quick testcase
18:09
<bholley>
Hixie: also, can you clarify what you mean by "named properties other than child browsing context names" in https://www.w3.org/Bugs/Public/show_bug.cgi?id=21674 ?
18:10
<Hixie>
(hahah, the guy who said i should read up on html5 edited his comment to remove that bit of it)
18:10
<gsnedders>
(Where? :))
18:11
<Ms2ger>
http://www.reddit.com/r/Python/comments/1c8wzq/htmltagpy_a_new_module_to_wrap_content_in_html/c9eclnh?context=3
18:12
<Ms2ger>
bholley, global scope polluter
18:12
<Hixie>
bholley: like, window.foo for <a name=foo>
18:12
<bholley>
Hixie: oh god, I sure hope not
18:13
<bholley>
Hixie: certainly not in Gecko
18:13
<Ms2ger>
bholley, fwiw, I think we have a special case for nsIDOMWindows
18:13
<bholley>
Ms2ger: special-case in what sense?
18:13
<bholley>
Ms2ger: as in the way we handle it in AccessCheck.cpp?
18:13
<bholley>
IsFrameId?
18:14
<Ms2ger>
Oh, no
18:14
<bholley>
Hixie: but yeah, in Gecko this just goes straight to the browsing context tree, and I'd be very suspicious of doing anything more general
18:14
<Ms2ger>
I was thinking of something in nsWindowSH::GetProperty, but that's dead
18:14
<bholley>
Hixie: I'll test enumerability now
18:15
bholley
grumbles about SSH lag while writing testcases on his server, but then remembers that he's sitting on an airplane
18:16
<odinho>
bholley: mosh
18:16
<bholley>
odinho: mosh?
18:16
<Ms2ger>
mosh
18:16
<odinho>
my bestest friend. mobile-shell/mosh yes.
18:17
<bholley>
odinho: what is that?
18:17
<odinho>
bholley: No, who knows, try a apt-get install mosh or your equivalent, and check it out ;-)
18:18
<odinho>
bholley: It helps you with laggy connections. Works better in typical mobile situations. E.g. when changing connections, or having very high latency. It has local echo e.g.
18:18
<bholley>
hm, there's no homebrew for it
18:19
<bholley>
odinho: or wait, is this server-side or client-side?
18:19
<odinho>
Yes there is, -- I have used an OS X machine once.
18:19
<odinho>
bholley: You need it on both.
18:19
<bholley>
odinho: brew install mosh gives no results
18:19
<odinho>
http://mosh.mit.edu/
18:19
<odinho>
bholley: brew install mobile-shell maybe
18:20
<bholley>
odinho: bingo
18:20
<odinho>
^_^ The website has an OK first page that will list the main points.
18:21
<Hixie>
bholley: i agree that we shouldn't expose it, but interestingly this means the supported properties differ based on who is accessing the Window...
18:21
<bholley>
Hixie: I don't follow
18:27
<Hixie>
bholley: as in, ('foo' in w) is true if you're same-origin with w and false if you're not, if there's an <a name=foo> in w's document
18:28
<bholley>
Hixie: well, the same holds for expandos, right?
18:28
<Hixie>
yeah, i guess it's the same