00:04
<heycam>
annevk, Domenic_, anything I need to change in IDL?
00:04
<Domenic_>
heycam: It sounds like Promise.cast will get renamed to Promise.resolve
00:04
<heycam>
Domenic_, ok
00:04
<heycam>
Domenic_, in the next ES6 draft I guess?
00:05
<Domenic_>
heycam: yeah
00:05
<heycam>
Domenic_, ok cool
01:30
<dominiquenf>
good evening
08:42
<zcorpan>
is http://www.khronos.org/registry/typedarray/specs/latest/ the current spec for typed arrays or has it moved to ES land?
08:43
zcorpan
finds https://people.mozilla.org/~jorendorff/es6-draft.html#sec-typedarray-objects
09:02
<zcorpan>
yoav: yeah so long as the black-box observable behavior is equivalent it can be implemented in any way
09:02
<yoav>
zcorpan: I agree in 100%
09:04
<zcorpan>
yoav: what do you think this should do? <!doctype html><div id=x><script>x.remove();</script><img src=foo onload=alert(1)>
09:04
<zcorpan>
yoav: (the img will be inserted into the removed div by the parser)
09:05
<yoav>
I think that the src should be triggered as soon as the img is inserted (even if the div is removed)
09:05
<yoav>
The only reason we need to wait, is to see if the parent is picture
09:05
<yoav>
But I completely agree that this area should be well specified, and currently it isn't
09:07
<zcorpan>
ok, then it's different to "inserted to the document"
09:07
<zcorpan>
i'm pondering if i should take a stab at specifying this
09:11
<yoav>
zcorpan: If you're willing, it'd be highly appreciated
09:13
<zcorpan>
yoav: do i get any chocolate? :-)
09:14
<yoav>
zcorpan: I'll see what I can do :)
09:15
<zcorpan>
hehe
09:16
<Ms2ger>
Dammit
09:48
<yoav>
zcorpan: The plot thickens...
09:54
<zcorpan>
yoav: btw downloading immediately in the script case is wrong per spec already even without <picture>, since there's crossorigin and srcset. the spec requires to await a stable state
09:56
<yoav>
zcorpan: HTMLImageElement can tell if it *already* contains such attributes when 'src' is encountered
09:56
<yoav>
It cannot tell what it parent is
09:56
<yoav>
I don't know how "stable state" is defined
09:57
<zcorpan>
yoav: it basically means to let the script finish
09:57
<yoav>
OK
09:57
<zcorpan>
yoav: so that it doesn't matter in which order you set src and crossorigin in a script
09:58
<yoav>
I think that in practice it does...
09:58
<zcorpan>
http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-1.html#update-the-image-data step 8
09:58
<zcorpan>
if it does it's a bug
09:58
<yoav>
Interesting
09:58
<yoav>
Can you comment that on the blink-dev discussion?
09:59
<zcorpan>
sure
09:59
<yoav>
Practically, it means that an HTML parser script just has to add the image to its parent before terminating
10:02
<zcorpan>
the spec says to await a stable state for the html parser case too, which means to wait until the parser yeilds to the event loop, however that's not technically necessary i think since the html parser sets all attributes at the same time as creating the element (this might not match impl reality though)
10:03
<yoav>
zcorpan: Yeah, current parser (at least Blink/WebKit) sets all the attributes upon creation
10:34
<hsivonen>
annevk-cloud: FYI: https://groups.google.com/forum/#!topic/mozilla.dev.l10n/PH7tF9m8vUY (encouraging a UTF-8 default for outgoing email)
10:43
<MikeSmith>
hsivonen: about Client Hints, was wondering why you were asking earlier if howcome had an opinion on it. You think he'd not be favorable?
10:43
<Ms2ger>
CH CH
10:44
<Ms2ger>
Client Hints Considered Harmful
10:51
<hsivonen>
MikeSmith: When I read about CSS Regions Considered Harmful, some of the arguments seemed applicable to Client Hints.
10:53
<MikeSmith>
ah ok
10:57
<zcorpan>
hsivonen: which arguments? problem #2 ?
11:07
<hsivonen>
zcorpan: yes
11:09
<zcorpan>
from what i understand, CH doesn't intend to solve art direction but leaves that to <picture>. however if the user zooms it seems reasonable that the browser should reload the image, dunno what the I-D says about that
11:11
<hsivonen>
zcorpan: and if you use CH on HTML resources, does it mean a reasonable browser becomes Netscape 4.x when resizing the window (includes rotation on mobile)?
11:12
<zcorpan>
does CH apply to any resource? i was under the impression it was only for <img> requests (though i haven't read the spec)
11:12
<hsivonen>
when I argued it was about images, karl said it's not just about images
11:12
<hsivonen>
I don't recall anything in the spec restricting it to images
11:12
<zcorpan>
hsivonen: the DPR doesn't change when you resize the window though
11:13
<hsivonen>
also, sicking argued for a non-image case
11:13
<hsivonen>
zcorpan: width does
11:39
<zcorpan>
hsivonen: how?
11:43
<hsivonen>
zcorpan: resizing the window changes the width of the view port
11:43
<zcorpan>
hsivonen: the width of the viewport doesn't affect DPR
11:44
<hsivonen>
zcorpan: DPR isn't the only kind of client hint
11:44
<hsivonen>
zcorpan: there's another one for width
11:44
<zcorpan>
ah, ok
11:44
<hsivonen>
zcorpan: though the spec isn't really clear on the width of what exactly
11:45
<hsivonen>
see https://bugzilla.mozilla.org/show_bug.cgi?id=935216#c20
11:48
<zcorpan>
the width hint seems like a bad idea to me
11:49
<hsivonen>
zcorpan: both client hints seem like bad ideas to me
14:00
<zcorpan>
jgraham: why doesn't test(function(t) { }) have the test object as the argument (t is undefined)?
14:15
<jgraham>
zcorpan: I don't think there's any particular reason.
14:15
<jgraham>
It's not really clear what the use case is, I guess
14:15
<zcorpan>
t.add_cleanup
14:15
<jgraham>
Oh, yeah so I guess with that there is a use case
14:17
<jgraham>
zcorpan: Want to make a patch? You just need to change the call to test_obj.step() in function test(){} to look like the one in function async_test(){}
14:17
<jgraham>
i.e. add tet_obj, test_obj as arrguments
14:17
<zcorpan>
yeah sure, but will have to wait until monday
14:17
<jgraham>
although I am slightly worried that will make the this object different
14:18
<jgraham>
Dunno if undefined, this_obj would be a better bet
14:20
<zcorpan>
is that a problem for async_test? what is this_obj here?
14:21
<zcorpan>
isn't test_obj the this object for both test and async_test now?
14:22
<jgraham>
I think in test() it's the global
14:23
<jgraham>
Oh, no
14:23
<jgraham>
the code is just strange
14:23
<jgraham>
OK, patch should be quite sae then
14:23
<jgraham>
*dafe
14:23
<jgraham>
*safe
14:23
<jgraham>
Christ
14:24
<zcorpan>
why does async_test run step with three arguments? the function only takes two arguments?
14:25
<zcorpan>
oh wait
14:25
<zcorpan>
nm
14:26
<zcorpan>
yeah so it's just adding ", test_obj, test_obj"
14:26
<jgraham>
Yeah, the first argument is the callback, the second is the this to use and any remaining arguments are the parameters to pass to the callback
14:33
zcorpan
*poof*
15:17
<mathiasbynens>
https://twitter.com/RobbertAtWork/status/429271270052872193 — “<svg> elements in <template> won't work: they're parsed in the HTML namespace. We're gonna need XML to do this template thing properly.”
15:25
<MikeSmith>
oh boy
15:27
<MikeSmith>
well that seems intead like a very good argument for us to change to parsing all SVG using the HTML parser instead of the XML parser
15:27
<stockwell>
I'd like to help by writing some test cases. What's the process for doing that?
15:28
<MikeSmith>
stockwell: https://github.com/w3c/web-platform-tests/blob/master/README.md#contributing
15:29
<stockwell>
<MikeSmith> thank you
15:33
<gsnedders>
Is there any ETA for ES6 being basically done?
16:04
<Domenic_>
gsnedders: http://esdiscuss.org/topic/es6-timetable-and-current-status
16:07
<gsnedders>
Domenic_: Braw.
16:08
gsnedders
remembers this is #whatwg and en-gb is probably better than sco.
16:08
<gsnedders>
so, uh, thanks.
16:09
<Domenic_>
np :P
16:18
<MikeSmith>
Domenic_: so what was the recent big change to Promises? the .cast -> .resolve change?
16:19
<Domenic_>
MikeSmith: yeah. "big" was overstating it from an API perspective
16:19
<MikeSmith>
ah OK
16:19
<Domenic_>
it is kind of big behind the scenes because some planned ES7 changes that some people wanted were decided against.
16:19
<MikeSmith>
oh
16:19
<Domenic_>
which allows simplification of the internal model
16:19
<Domenic_>
(fairly drastic simplication)
16:19
<MikeSmith>
aha
16:20
<Domenic_>
but from an actual consumer POV the only change is that .cast becomes .resolve and resolve dies
16:20
<MikeSmith>
well the behind-the-scenes change sounds like a very good thing
16:21
<MikeSmith>
because simpler usually being better
16:21
<MikeSmith>
and drastically simpler especially so
16:28
<Domenic_>
indeed, i am quite happy
16:29
<MikeSmith>
cool
16:30
<MikeSmith>
jgraham: somebody should come up with a way we can track data about test cases that have resulted in spec bugs being found
16:32
<MikeSmith>
(in context of https://critic.hoppipolla.co.uk/showcomment?chain=1646 -- see the last comment from Ms2ger)
16:32
<MikeSmith>
(and bug https://www.w3.org/Bugs/Public/show_bug.cgi?id=24404)
16:33
<jgraham>
MikeSmith: Yeah. We should track many things that we can't / don't
16:34
<MikeSmith>
yeah I know
16:34
<jgraham>
But I agree that makes the list
16:34
<MikeSmith>
that's why I was suggesting assigning it to the "somebody", who seems to have infinite time
16:34
<jgraham>
:)
16:34
<MikeSmith>
yay
17:29
<reyre>
foolip: ping
17:43
<annevk>
Man, for a moment I thought the microtask discussion was over
17:52
<annevk>
I wish I could answer the range question without doing a bunch of research
17:53
<Hixie>
hsivonen: you around?
17:54
<Hixie>
hsivonen: haven't heard your opinion of the UI proposals in my registry e-mail, would love to have your feedback
18:05
<dglazkov>
good morning, Whatwg!
18:18
<Hixie>
why do dictionaries have constructors?
18:33
<MikeSmith>
Hixie: I'd also like to hear what hsivonen is thinking on that
18:34
<MikeSmith>
in the mean time I will try to write up a reply myself this weekend
18:35
<annevk-cloud>
Hixie: they do? Sounds weird, although they can be returned they do not have exposed names
18:35
<Hixie>
if they have a constructor the name is exposed
18:35
<Domenic_>
are there examples of ones with constructors? seems bad
18:37
<Hixie>
well i assume there was some use case...
18:37
<Hixie>
i'm just curious what it was
19:43
<MikeSmith>
somebody recommended a good IRC client for OSX here the other day
19:43
<MikeSmith>
a newer one, not Colloquy
19:43
<MikeSmith>
I didn't pay attention at the time because I didn't care because I just irssi and screen
19:44
<MikeSmith>
but now I'm trying to help somebody get set up with IRC so now I care, so maybe can somebody remind me what the name was?
19:45
<MikeSmith>
Textual?
20:04
<Ms2ger>
Hixie, I thought the constructors for dictionaries was for points or something, but http://dev.w3.org/fxtf/geometry/ has those as interfaces
20:55
<sicking>
Hixie: shouldn't BroadcastChannel inherit EventTarget?
21:24
<Hixie>
sicking: sounds plausible
21:25
<sicking>
Hixie: event dispatch won't work otherwise
21:25
<Hixie>
sicking: fixed, reload in a minute or so
21:25
<sicking>
thanks
21:25
<Hixie>
np, thanks for catching it
21:42
<annevk>
MikeSmith: LimeChat is nice
21:42
<MikeSmith>
annevk: OK thanks
21:43
<MikeSmith>
I think Textual is based on LimeChat originally
21:43
<Domenic_>
still a big fan of irccloud
21:44
jgraham
doesn't understand why you would move off irssi
21:44
<jgraham>
Oh,for someone else
21:45
<jgraham>
Fair enough
21:47
<Hixie>
i'm likely to work on the web focus model refactoring soon so if people have comments on my proposal, now's the time https://www.w3.org/Bugs/Public/show_bug.cgi?id=23475
23:27
<Hixie>
jgraham: the bug thing still seems out of date?
23:29
<Hixie>
jgraham: yeah, i get a 500 when i do an update request
23:46
<annevk>
I might have comments on how focus works a couple of years from now
23:46
<Hixie>
heh
23:46
<annevk>
When I have time to learn about UI events :-P
23:46
<Hixie>
hopefully it'll be a solved issue by then.
23:46
<Hixie>
the focus part, that is
23:46
<Hixie>
UI events won't be