00:38
<terinjokes>
does XHR2 no longer care about ArrayBuffers as a request body?
00:40
<terinjokes>
annevk: ^^
07:32
<annevk>
terinjokes: sure does
07:32
<annevk>
Hixie: pretty sure, yes
07:33
<Hixie>
why?
07:33
<annevk>
Hixie: there's nothing about event listeners that goes back to the microtask thing, we just iterate through them synchronously and run them
07:34
<Hixie>
how do you run the callbacks?
07:34
<annevk>
Hixie: through IDL's "call ..." thingie for that
07:34
<Hixie>
http://heycam.github.io/webidl/#es-invoking-callback-functions ?
07:34
<annevk>
Hixie: yeah
07:36
<Hixie>
huh
07:37
<Hixie>
when did they stop calling "clean up after running a callback"?
07:37
<annevk>
Hixie: note that in https://www.w3.org/Bugs/Public/show_bug.cgi?id=27046#c5 I asked for clarification about this since I thought I had the wrong mental model
07:38
<Hixie>
maybe this changed at some point
07:38
<annevk>
It does seem quite weird to me to run microtasks between two registered event listeners that run in the same task
07:38
<Hixie>
i could have sworn we wanted mutation observers to run after each event listener
07:38
<annevk>
Though smaug always maintains that is the model, so maybe...
07:38
<Hixie>
at some point i thought heycam|away had webidl calling the "clean up after running a callback" logic
07:38
<annevk>
Hixie: yeah I thought so too, but then it wasn't really supported by any specification
07:38
<Hixie>
maybe bz changed it?
07:39
<Hixie>
oh well
07:39
<Hixie>
bed time for me
07:39
<Hixie>
nn
07:40
<annevk>
have a good night
07:40
<annevk>
I might try to sort it out once more if I have some time left
07:49
<terinjokes>
annevk: i just don't see it in the version up on the site
07:50
<annevk>
terinjokes: "void send(optional (Document or BodyInit)? body = null);"
07:50
<annevk>
terinjokes: "typedef (Blob or BufferSource or FormData or URLSearchParams or USVString) BodyInit;" (if you follow the BodyInit link)
07:51
<terinjokes>
i promise you
07:51
<annevk>
terinjokes: http://heycam.github.io/webidl/#common-BufferSource (if you're curious about BufferSource and look into IDL to find out)
07:51
<terinjokes>
that does not look like a link to me
07:52
<annevk>
terinjokes: you need to hover the IDL block these days I think, we made the styles a bit less distracting for when you're just reading
08:00
<annevk>
Ms2ger: ta, done
08:58
<annevk>
Does anyone else get a lot of these GitHub user surveys?
08:58
<annevk>
Is there a way to opt out?
09:05
<JakeA>
annevk: I got one this morning
09:23
<mathiasbynens_>
annevk: you didn’t respond to this part “https://url.spec.whatwg.org/#origin reads as though the default port is included.”
09:25
<mathiasbynens>
fwiw, i always thought origin didn’t include implicit/default ports, only what’s explicitly part of the URL
09:28
<annevk>
mathiasbynens: I made a mistake, origins are tuples, not strings
09:28
<annevk>
mathiasbynens: and the tuple includes the origin
09:28
<annevk>
mathiasbynens: includes the port, oops
09:29
<mathiasbynens>
ok, so the url spec is correct then
09:29
<mathiasbynens>
so https://example.com/ and https://example.com:443/ are same-origin?
09:35
<annevk>
mathiasbynens: yes
11:37
<MikeSmith>
https://w3c.github.io/webappsec/specs/credentialmanagement/#security-considerations seems to be kind of defining a Somewhat Similar Origin Policy
11:38
<MikeSmith>
I wonder what other specs might have the same need
11:39
<MikeSmith>
i.e., for this Credential Management it's about credential sharing among non-exact origins, but it's imaginable that there might other use cases for sharing across non-exact origins
11:40
<MikeSmith>
well I guess there are lots of other use cases, if you listen to the people who think the same-origin policy is a problem
11:41
<MikeSmith>
"MUST NOT offer credentials to an origin in response to request() without user mediation if the credential’s origin is not an exact match for the calling origin"
11:55
<smaug____>
annevk: what is the current plan, will shadow dom be merged to dom, or will it stay a separate spec?
11:56
<smaug____>
(just thinking about bug filing )
12:06
<zcorpan_>
Ms2ger: in quirks spec source i have this:
12:07
<zcorpan_>
<h2>Quirks that maybe can be removed from implementations</h2>
12:07
<zcorpan_>
<dl>
12:07
<zcorpan_>
<dt>In quirks mode, text-decoration is propagated into floating and absolutely positioned elements</dt>
12:07
<zcorpan_>
<dd><p>Opera, WebKit and IE9 (except in compat view quirks mode) don't support this quirk.</dd>
12:07
<zcorpan_>
(commented out)
12:07
<Ms2ger>
There we are
12:09
<annevk>
smaug____: there's no good plan for shadow DOM yet I think
12:09
<annevk>
smaug____: there's not really consensus on what it should look like
12:11
<smaug____>
ok, so keep filing bugs on shadow dom spec for now, not dom spec
12:12
<annevk>
smaug____: yeah
12:12
<annevk>
smaug____: I'd like to do custom elements first, but we can't seem to get agreement on how to do upgrades :-(
12:13
<smaug____>
right
12:13
<smaug____>
I haven't been following that too closely
12:18
<annevk>
Okay, so I tested and yes, there are bugs around microtasks and callbacks
12:18
<annevk>
Baaah
12:27
<annevk>
Filed https://www.w3.org/Bugs/Public/show_bug.cgi?id=27874
12:28
<annevk>
Note that mutation are from 2011
12:32
<smaug____>
don't understand what that bug is about
12:34
<annevk>
smaug____: you can't explain when the callbacks are supposed to run from reading the specs
12:34
<smaug____>
oh, spec text issue
12:35
<annevk>
smaug____: well it's a bug against a spec
12:35
<smaug____>
hmm, webidl?
12:35
<smaug____>
why webidl
12:35
<annevk>
smaug____: it defines running callbacks
12:36
<smaug____>
ah, right...
12:36
<smaug____>
ok, I would add some hook to webidl and make HTML spec to use that to invoke end-of-microtask stuff
12:36
<smaug____>
HTML or DOM
12:36
<annevk>
yeah there is such a hook
12:37
<annevk>
however, the problem is that we don't want to invoke it all the time
12:37
<smaug____>
yes, only when there isn't any other script on stack
12:37
<smaug____>
and that would be up to the HTML or DOM to spec
12:37
<smaug____>
I guess HTML
12:37
<annevk>
such a weird model
13:18
Ms2ger
pokes dglazkov
13:37
<annevk>
JakeA: did you talk to devdocs.io about AppCache vs SW?
13:37
<annevk>
JakeA: seems like they're lacking TLS
13:39
<JakeA>
annevk: I haven't, but shall. But yeah, I guess TLS is an initial blocker. If they've got it working with appcache, they may not want to move to something with worse support
13:39
<annevk>
JakeA: well, we're going to remove AppCache from Firefox
13:40
<JakeA>
annevk: I'll reach out to them
13:48
<annevk>
JakeA: that site is really fast btw
13:49
<JakeA>
annevk: it's great isn't it? Just in time for my 23hr flight tomorrow too
13:51
<annevk>
JakeA: Oceania?
13:51
<JakeA>
annevk: Yeah, Sydney
14:57
<JakeA>
annevk: re https://github.com/slightlyoff/ServiceWorker/issues/609 - I can't see how client objects can have onmessage, windows have no access to their client
15:14
<iandevlin2>
annevk: appcache is being removed from Firefox?
15:59
<annevk>
iandevlin2: not yet, but with service workers coming the writing is on the wall
15:59
<iandevlin2>
annevk: ok, interesting
15:59
<Ms2ger>
"enemies of the heir beware"?
15:59
<JakeA>
Appcache has some security howlers. Chrome's security team would like to see it gone
16:01
<annevk>
Ms2ger: I'm not sure I comprehend the question
16:02
<jgraham>
annevk: Not a Harry Potter fan?
16:02
<annevk>
iandevlin2: I suspect there will be an announcements of sorts, once we've confirmed that service workers actually works
16:02
<annevk>
jgraham: been a while
16:03
<jgraham>
Think Chamber of Secrets
16:05
<annevk>
heh
16:40
<wanderview>
JakeA: can appcache be polyfilled with SW? (I'm not too familiar with appcache api)
16:41
<JakeA>
wanderview: appcache allows one tab to be using version n, and another to be using version n+1. I think that's the only part SW (deliberately) can't do
16:42
<wanderview>
JakeA: version of the SW script, right? I assume they could be using separate cache objects for versions of the site
16:43
<JakeA>
wanderview: yeah, but you can't tell which tab you're serving, so you can't give them items from different caches
16:43
<JakeA>
& there can only be one active SW in a registration
16:43
<wanderview>
JakeA: oh... right... that makes sense :-)
16:44
<JakeA>
You can get really close though. You can probably make it work under the hood by finding a way to allow that bit
16:45
<JakeA>
Eg, if an appcache SW creates a new registration for each version, but at the same scope, that would work (you can't do that with the currently exposed API)
16:46
<JakeA>
But that means the old tab would be using the old registration & active worker, whereas the new tab uses the new registration & active worker
16:46
<wanderview>
JakeA: or we could just de-orbit appcache... I was just wondering if polyfill was easy or not
16:46
<JakeA>
heh yeah
17:24
<annevk>
JakeA: can't you do that with Client objects?
17:26
<annevk>
Man, the pseudo-security on www-tag just keeps going
17:26
<annevk>
I wish there were at least some memes to alleviate some of the 386 pull
17:50
<TabAtkins>
Does somebody have Chrome Canary and can tell me what this logs?
17:50
<TabAtkins>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3381
17:50
<miketaylr>
TabAtkins: heres what i got https://gist.github.com/miketaylr/ecbf5d97337dc871c90e
17:50
<TabAtkins>
Cool, it's still lolbroke.
17:51
<TabAtkins>
Just checking before I file a bug.
17:51
<TabAtkins>
That is one *broke-ass* background serialization.
17:52
<TabAtkins>
miketaylr: What's your Canary version?
17:52
<miketaylr>
TabAtkins: Version 42.0.2282.0 canary (64-bit)
17:52
<TabAtkins>
OS?
17:52
<TabAtkins>
Not that it matters, but I'm filling out the fields.
17:53
<miketaylr>
OSX Yosemite
17:53
<miketaylr>
10.10.2
18:06
<wanderview>
Domenic: so if we have the request.stream() to reify to a js stream... and request.text() to reify to text... does this mean that request.stream() will set bodyUsed to prevent request.text() from being called?
18:06
<wanderview>
even if the data has not been read?
18:07
<Domenic>
wanderview: bodyUsed is a whole 'nother thread of questions... but I think that is @horo-t's proposal. It seems somewhat self-consistent at least.
18:07
<wanderview>
I think thats how I would expect it to work if I was a developer... "I can only call one of these reify methods and then bodyUsed is set"
18:07
<Domenic>
Yeah.
18:07
<wanderview>
makes sense to me
18:07
<wanderview>
Domenic: sorry I'm always so confused about this stuff on github... I was once familiar with node streams... but its all a bit mixed up for me now
18:08
<Domenic>
wanderview: no problem! it's confusing stuff!!
18:08
<wanderview>
can't keep track of how things have changed recently
18:15
<wanderview>
annevk: why is request.text() a method and not a getter?
18:15
<annevk>
wanderview: because it returns a fresh promise
18:16
<annevk>
wanderview: request.text === request.text wouldn't hold
18:16
<wanderview>
annevk: I guess I would like to see a method like request.stream() so that it can set bodyUsed and return a stream on first call... then reject because of bodyUsed after that
18:16
<wanderview>
which of course can be done with a getter... but would be confusing
18:16
<annevk>
I see, rationale in the ticket?
18:16
<annevk>
I have to go now :-(
18:17
<wanderview>
annevk: k, thanks
18:26
<aklein>
annevk: re: microtasks, did you see my comment?
20:14
<Ms2ger>
So has anything been made public about this Spartan thing?
20:24
<miketaylr>
Ms2ger: seems like the public message is "more details soon"
20:26
<smaug____>
something new about spartan?
20:26
<TabAtkins>
And holograms.
20:27
<wanderview>
they've said "not webkit"
20:35
smaug____
wonders what is happening @nokia. All the activate@w3c employees are gone?
21:16
<jarek>
Hi
21:16
<jarek>
Is <template> element allowed inside <svg> element?
21:17
<jarek>
The spec says it should inherit from HTMLTemplateElement : HTMLElement
21:17
<jarek>
it would make sense to make it more generic (TemplateElement : Element)
21:19
<Ms2ger>
No
21:20
<jarek>
SVG already has <defs> element which is has very similar functionality
21:20
<jarek>
but I guess it's already too late to rename HTML 5 <template> to <defs>
21:28
<jamesr___>
the html/svg ns split is not very useful nowadays
21:38
<tantek>
jamesr___: it was never useful, but rather XML/SGML dogma.
21:39
<jamesr___>
true, but dogma is less useful these days (not that it was terribly useful technically before, but it had more weight)
22:00
<TabAtkins>
jamesr___: I keep trying to get people to agree to merge the namespaces, but they keep whining about "compatibility" and other such nonsense. ^_^
22:13
<annevk>
aklein: yeah, thanks for that
22:13
<annevk>
aklein: makes total sense, so it's just up to bz or heycam to tie things up
22:27
<aklein>
annevk: sounds good. I was just worried that you actually wished the ordering was always 1, 2, 3, and that I definitely didn't want to mess with
22:29
<annevk>
aklein: if I had better understood things from the get go I might have argued for that, but at this point I'd just like to have the implemented rules written down
22:30
annevk
read Spartan as SPARTAN, Opera's test infrastructure
22:30
<wanderview>
annevk: whenever I try to comment on one of those stream github issues... I end up feeling like a complete idiot... not sure my opinion should really matter much :-)
22:32
<annevk>
wanderview: please just view it as having a very different perspective, which is very valuable
22:33
<wanderview>
annevk: I just want someone to tell me what bodyUsed means if someone has read half the body through the stream.... its really weird for me if bodyUsed==false there
22:34
<annevk>
wanderview: I think several views are conflated
22:35
<wanderview>
annevk: I gotta run to dinner, but I'll pick up being confused in the morning! good night!
22:35
<annevk>
wanderview: e.g. I think Domenic wants text() et al to just consume the stream (whatever is in there at the moment) and decode as appropriate (ignoring anything like bodyUsed)
22:35
<annevk>
wanderview: hah okay, nn
22:41
<gsnedders>
annevk: Spartan: better tested than SPARTAN.