00:40
<Hixie>
jamesr__: did you ever reply to http://lists.w3.org/Archives/Public/public-whatwg-archive/2013Oct/0013.html by any chance?
00:44
<TabAtkins>
Hixie: s/textArea/textContent/
01:12
<jamesr__>
Hixie: i did not. in order to reply i'd have to figure out whether there are any observable differences between my proposed text and the spin the event loop algorithm
01:13
<jamesr__>
mentally it's a lot easier for me to map the steps of "spin the event loop" on to the actual C++ act of putting an object on the stack that spins the event loop, but that's not how the text really works
01:13
<Hixie>
TabAtkins: context?
01:13
<TabAtkins>
The most recent email you sent, about <output>
01:13
<Hixie>
ah, my bad
01:13
<Hixie>
oh well
01:13
<Hixie>
hopefully he figures out what i mean
01:13
<Hixie>
jamesr__: k
01:13
<Hixie>
jamesr__: not really sure how it differs, but ok
01:13
<Hixie>
jamesr__: :-)
01:14
<jamesr__>
Hixie: if it doesn't, then i'd suggest replacing it with my text just for the purpose of removing invociations of that step
01:14
<jamesr__>
since that step makes the spec incredibly hard to reason about
01:14
<Hixie>
jamesr__: i don't understand your text :-)
01:14
<jamesr__>
ok. well i don't think i fully understand the currently spec algorithm
01:14
<Hixie>
jamesr__: as far as i can tell, it's the same as what the spec says, but using non-spec terms :-)
01:15
<jamesr__>
window.showModalDialog() does something that i can map to the C++ callstack
01:15
<jamesr__>
and as an implementor that's relatively easy to reason about
01:15
<jamesr__>
this parsing step does not map to reentrancy in the C++ callstack
01:15
<Hixie>
jamesr__: "spin the event loop" right now means "suspend the current algorithm, end the task, and, later, queue up a new task that resumes the current algorithm"
01:15
<jamesr__>
so it's a lot harder at least for me to reason about what it really does
01:15
<jamesr__>
yeah but there are consequences about what's higher up on the stack
01:16
<jamesr__>
or equivalently what the consequences of "suspend the current algorithm" are
01:16
<Hixie>
what are the consequences?
01:16
<jamesr__>
i don't believe any major browser today supports continuations for javascript, so "suspend the current algorithm" if the algorithm has javascript on the stack means run a bunch of code on top of that callstack
01:16
<jamesr__>
instead of returning to the underlying native event loop
01:16
<Hixie>
i thought you were proposing that instead of spinning the event loop, we suspend the current algorithm, end the task, and later, queue up a new task that resumes the same algorithm. which seems to me identical.
01:18
<Hixie>
jamesr__: your e-mail didn't suggest a change to showModalDialog() as far as i can tell
01:23
<jamesr__>
no
01:23
<jamesr__>
that was an example
01:23
<jamesr__>
the parser case doesn't have the issues that showModalDialog() has of JS higher up the stack
01:23
<jamesr__>
but it's not obvious from reading the parser text that those issues are not a problem here
02:16
<Hixie>
jamesr__: k. i don't understand the exact issue then.
02:16
<Hixie>
jamesr__: please respond in due course so i can study your answer and fix the spec if necessary :-)
02:17
<Hixie>
jamesr__: (or ping me sometime around 11 and we can chat over lunch or whatever)
04:05
<Domenic_>
annevk: Firefox promises in Aurora are not doing so hot, 84/879 tests pass. http://domenic.me/aplus-tests-against-the-browser/
06:12
<rniwa>
annevk: yt?
07:47
<MikeSmith>
Domenic_: seems like those are not doing so great in Chromium either -- 94 passes, 778 failures
07:48
<MikeSmith>
unless there's some flag I should have set but don7t
07:49
<MikeSmith>
hmm I get different results when I re-run it
07:50
<MikeSmith>
in Chromium, 223 passes, 649 failuress
07:50
<MikeSmith>
with duration: 8.27s
07:51
<MikeSmith>
takes way longer for the tests to execute in Firefox Nightly
07:51
<MikeSmith>
duration: 129.66s
07:51
<MikeSmith>
passes: 237
07:52
<MikeSmith>
failures: 620
07:52
<MikeSmith>
but only runs 98% of the tests
09:02
<Ms2ger>
jgraham, can I use commit --amend with critic?
09:10
<MikeSmith>
jgraham: I notice that critic seems to send some messages without an address in the From header (only a GH username). Seems like it would be nicer if it always put some address for From , even if it's just some dummy default
09:10
<reggna>
Ms2ger: You can, if you prepare critic for a rebase, and critic will create a "merge commit".
09:10
<reggna>
Ms2ger: But it's probably easier to just commit a fixup and do the rebase later.
09:12
<MikeSmith>
also wonder why it sends messages base64-encoded (not that it bothers me -- just curious)
09:47
<zcorpan>
hmm. test(func, title, {help:foo}) should be able to take an array of urls
09:48
<zcorpan>
darobin: did you invent {help:foo} in testharness.js? is it being used? does something break if i use an array?
09:48
<darobin>
zcorpan: that doesn't look like something I'd have invented
09:48
<darobin>
is this the metadata you can add?
09:49
<darobin>
that's plinss
09:49
<zcorpan>
oh, ok
09:49
<darobin>
I am not sure what uses the data
09:49
<darobin>
I would say try using an array and if someone complains you'll have your answer :)
09:49
<zcorpan>
i'll just use an array and update the documentation saying that it takes a string or an array of strings
09:50
<darobin>
yeah that works too
09:51
Ms2ger
dislikes that argument
10:03
<hsivonen>
interesting that Greece and Japan appeared to have opted not to get a Punycode TLD
10:11
<odinho>
zcorpan: You even want multiple help urls on a single test? :|
10:11
<odinho>
Crazy :D
10:17
<hsivonen>
annevk: so the gbk decoder checks the gb18030 flag only in one step. What's hard about determining if that step could always run as if the gb18030 flag was set (unifying the two encodings)?
10:54
<jgraham>
Ms2ger: No
10:54
<Ms2ger>
So it seems
10:56
<jgraham>
Ms2ger: It's effectively an in-place rebase
10:56
<jgraham>
Which it *does* support, but you have to tell it that you are going to do one in advance
10:56
<jgraham>
(in the normal setup)
10:57
<annevk>
hsivonen: the open bugs suggest there might be more differences :/
10:57
<annevk>
rniwa: now...
10:57
<jgraham>
In the GH setup it would have to infer this from the GH API provided data
10:57
<annevk>
Domenic_: that's great, thanks!
10:57
<Ms2ger>
jgraham, I think I did tell it
10:58
<jgraham>
Ms2ger: I think you probably didn't, because the prepare rebase button, which is normally what you use here, doesn't work when you are tracking a branch, AIUI
10:58
<jgraham>
(which you always are in the GH setup)
10:59
<jgraham>
That button just shouldn't be visible in this case
10:59
<Ms2ger>
Fun
10:59
<jgraham>
Ms2ger: Yeah. I really have to put some more work in to critic
11:00
<jgraham>
But there is a rumor that I might get internet at home soon
11:00
<Ms2ger>
Well, I learned something today :)
11:00
<jgraham>
So then it is actually plausible that I will
11:01
<Ms2ger>
Does that mean you spend a lot of time in the office instead?
11:02
<jgraham>
It means I spend a lot of time wishing I had a proper internet connection, and some more time wishing that we weren't also in a cell phone dead spot where 2 bars of signal is the maximum we have observed
11:03
<Ms2ger>
I see
11:37
<zcorpan>
odinho: sure. it's not uncommon to test multiple things together
12:17
<odinho>
zcorpan: Yeah, -- I had several help urls back when we used <meta> and one test per file. That's how the real world (tm) is.
13:21
<jgraham>
In the real world no one actually writes all that metadat
13:21
<jgraham>
a
14:16
<annevk>
hsivonen: https://www.w3.org/Bugs/Public/show_bug.cgi?id=23971
14:28
<darobin>
annevk: has there been any further discussion about the idea of having a sort of web platform conference thingie?
14:29
<Ms2ger>
Don't think so
14:29
<darobin>
I ask because I could get some logistical support from W3C to organise something
14:29
<darobin>
with a very open ended mission in terms of how it's run, etc.
14:30
<smaug____>
could we kill xslt, pretty please
14:30
<darobin>
smaug____: try v3, it adds something like JSON support :)
14:31
<smaug____>
seriously?
14:32
<Ms2ger>
smaug____, then you could remove peterv's only remaining contribution ;)
14:32
<odinho>
jgraham: lol, true.
14:33
<smaug____>
Ms2ger: I would count webidl bindings reasonable major contribution
14:33
<darobin>
smaug____: http://www.w3.org/TR/xslt-30/#json with love
14:35
<darobin>
the example almost makes me like Handlebars
14:44
<annevk>
darobin: so yeah, all I got is it has to be about social interaction, barcamp, loosely organized, for framework authors, complex apps, browser developers, and standards peeps
14:45
<darobin>
that sounds pretty much like what I wrote down :)
14:45
<annevk>
darobin: with focus on getting to know each other, talking through problems and annoyances
14:45
<annevk>
cool
14:51
<Ms2ger>
darobin, *I* never wrote such a library :)
14:51
<darobin>
Ms2ger: not even in Python?
14:52
<Ms2ger>
I doubt it
14:53
<darobin>
Ms2ger: so... you use the DOM? :)
14:54
jgraham
wonders what kind of library Ms2ger never wrote
14:54
<jgraham>
I feel like I am missing something obvious here
14:55
<Ms2ger>
darobin, right
14:55
<Ms2ger>
jgraham, "We've all written small libraries of the kind that Ian was describing, probably several times over."
14:55
<jgraham>
Oh, context is email
14:56
<MikeSmith>
q?
14:56
<jgraham>
darobin: Pretty sure that you bake things in, not boil things in
14:56
<darobin>
Ms2ger: well I guess we all have some of those kinky, painful things we enjoy
14:56
<jgraham>
Boiling is what you do to oceans
14:56
<darobin>
jgraham: my point exactly
14:56
<Ms2ger>
q+ MikeSmith
14:57
<Ms2ger>
annevk, http://html5.org/tools/web-apps-tracker?from=8324&to=8325 is empty
14:57
<hsivonen>
darobin: wow. that XSLT JSON thing was one of those things that confuse joke-or-serious detection without the spec link
14:57
<Ms2ger>
hsivonen, even with the spec link...
14:57
<annevk>
Ms2ger: le sigh
14:58
<darobin>
hsivonen: you clearly haven't been to an XML conference in a while — all they talk about is JSON these days :)
14:58
<annevk>
The spec link fact sounds sad
14:58
<darobin>
hsivonen: that said, JSON support makes a lot of sense for XQuery — I guess it just got into XSLT that way
15:00
<annevk>
Ms2ger: "fixed"
15:00
<Ms2ger>
Ta
15:00
<hsivonen>
darobin: how is JSONx (not a joke apparently!) doing at XML conferences?
15:00
<annevk>
Ms2ger: should really say "Ta" since this is going to happen again soonish at the rate things are going now
15:00
<darobin>
hsivonen: I can't say I have my ear that close to the ground, but I haven't heard about that at all in those circles
15:01
<hsivonen>
darobin: I guess that's a good sign
15:01
<darobin>
hsivonen: processing JSON with XML tools started coming up a few years ago; most people recommend the "Snelson Mapping"
15:01
<darobin>
I doubt anyone needs any kind of standard for that, especially not given support in XQuery/XSLT...
15:01
<darobin>
the MarkLogic people have been supporting JSON forever, you can use it as a JSON DB
15:02
<darobin>
I think they use Snelson internally
15:02
<darobin>
the one thing from that world which I would dearly like is http://jsoniq.org/ for JSON DBs
15:02
<darobin>
(and is possible over IndexedDB)
15:29
<jgraham>
https://critic.hoppipolla.co.uk/r/368 has been stuck for weeks :(
15:31
<Ms2ger>
Since Monday last week ;)
15:32
<jgraham>
In unrelated news, it seems polyglot is the HTMLWG equivalent of a comedy callback. It's carefully timed so that just when everyone has forgotten it, it's brought out again for another laugh
15:32
<odinho>
lol
15:47
<Lachy_>
jgraham, what recent discussion has there been on polyglot? I don't see any on public-html
15:51
<jgraham>
"Preference Survey: Polyglot Note vs Recommendation Track"
16:06
<Lachy>
jgraham, I can't find any mail with that subject on any W3C list.
16:06
<Ms2ger`>
It's on the whatwg list
16:06
Ms2ger`
ducks
16:07
<Lachy>
:-)
16:08
<MikeSmith>
hey it's Lachy
16:08
<Lachy>
Hi Mike
16:08
<MikeSmith>
howdy
22:01
<Hixie>
is there work ongoing into server-push notifications for web apps that aren't open?
22:11
<Hixie>
GPHemsley: any chance we can make the big warnings at the top of http://wiki.whatwg.org/wiki/FAQ get hidden for that page?
22:47
<hober>
Hixie: I don't think I understand the question
22:48
<Hixie>
re server-push notifications?
22:48
<hober>
yeah
22:49
<Hixie>
well like on android you can receive a text message notification even if your text message app isn't running, because the cloud can push a notification to your device
22:49
<Hixie>
do we have anything similar but for web apps in the works?
22:50
<hober>
like https://developer.apple.com/notifications/safari-push-notifications/ ?
22:50
<Hixie>
yeah, but not a proprietary thing :-)
22:51
<hober>
istm a "push api" being worked on in the webapps wg
22:51
<hober>
but I don't know if that's related offhand
22:51
<Hixie>
https://dvcs.w3.org/hg/push/raw-file/tip/index.html ?
22:52
<Hixie>
seems like what i'm looking for!
22:52
<Hixie>
thanks!
22:52
<hober>
np
22:58
<Hixie>
oops
22:58
<Hixie>
marked the last two commits as editorial
22:58
<Hixie>
but they're normative
23:00
<othermaciej>
Hixie: a challenge to making a non-proprietary version is that app-not-running push generally needs an intermediary service of some kind, and therefore requires the web server authenticating to that push server
23:00
<Hixie>
yeah
23:00
<Hixie>
for that reason i was mostly hoping someone else was working on it so it could be Not My Problem. :-)
23:00
<othermaciej>
Push API punts on solving that (and also on defining the protocol you use to talk to the push service)
23:01
<Hixie>
pity
23:01
<othermaciej>
Push API isn't even a suitable generic cover over a not-fully-specified set of non-interoperable protocols like EME is
23:02
<Hixie>
it does seem to be common for spec writers to punt on the hard stuff
23:03
<Hixie>
it's like spec writers think there's glory in writing specs
23:03
<Hixie>
poor, poor misguided spec writers :-)
23:03
<othermaciej>
I don't think we could convert Safari's proprietary push stuff to that API for instance, because it has no mechanism for providing an authentication token to the app server that will then talk to the push server
23:03
<Hixie>
it doesn't provide a way to close the loop?
23:04
<Hixie>
that seems like a pretty big hole
23:04
<Hixie>
i guess send feedback?
23:07
<othermaciej>
adding to TODO list
23:07
<Hixie>
:-)