07:29
<AxaliaN>
Hi all
07:29
<AxaliaN>
any autoconfirmed member active?
07:31
<AxaliaN>
annevk?
07:39
<AxaliaN>
zcorpan is a confirmed member, right?
07:40
<zcorpan>
member of what?
07:40
<AxaliaN>
thwe wiki :P
07:40
<AxaliaN>
the*
07:40
<zcorpan>
no idea :-)
07:40
<AxaliaN>
:P
07:41
<AxaliaN>
The wiki says it to be so!
07:41
<zcorpan>
then i guess so. how can i hlep?
07:42
<AxaliaN>
Anyhoo; there's a meta extension that needs to be added to the wiki
07:42
<AxaliaN>
hwo does one go about doing that?
07:42
<AxaliaN>
how*
07:43
zcorpan
looks at http://wiki.whatwg.org/wiki/WHATWG_Wiki:How_to_create_a_user_account
07:44
<AxaliaN>
Ah, I just saw you needed to come on IRC and ask :P
07:44
<zcorpan>
AxaliaN: username? email? (pm if you like)
07:44
<AxaliaN>
Problem 1, I am not an autoconfirmed member :P
07:44
<AxaliaN>
ok
07:48
<zcorpan>
ok created
07:48
<AxaliaN>
Awesome, thanks :)
07:48
<zcorpan>
np
09:24
<annevk>
zcorpan: does Chrome workers deal with data URLs?
09:24
<annevk>
s/does/do/
09:25
<zcorpan>
seems not http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3046
09:25
<zcorpan>
i think blob works though
09:26
<zcorpan>
at least for dedicated, dunno about shared
09:38
<annevk>
blob is fine as they have an origin
09:38
<zcorpan>
do you want to drop data url for workers?
09:40
<annevk>
yes
09:41
<annevk>
zcorpan: see WHATWG list
09:41
<annevk>
zcorpan: we want to make data URLs an opt-in
09:43
<zcorpan>
ok cool, would love to get interop on data urls :-)
09:43
<annevk>
I think we have a plan that works now
09:44
<zcorpan>
i think workers support data urls because blob didn't exist back then
09:45
<annevk>
well back then we didn't have a plan for data URLs that everyone agreed upon
09:45
<annevk>
I wish Google pushed their concerns with regards to them a bit harder at the time
09:45
<annevk>
oh well
09:45
<zcorpan>
do everyone agree on a plan now?
09:46
<annevk>
zcorpan: we're aligning with Chrome on <iframe> and hopefully workers
09:47
<annevk>
zcorpan: I guess I'm hopeful that's the case, haven't actually seen it confirmed
09:47
<zcorpan>
annevk: what about <track>?
09:48
<annevk>
zcorpan: seems like that can work as there are no side effects, right?
09:49
<zcorpan>
annevk: webvtt seems safe enough, but i'm not confident we can keep it to webvtt-only :-(
09:49
<annevk>
bah
09:50
<zcorpan>
there are some tests that use data: for <track> but that's easy to fix
09:50
<annevk>
zcorpan: yeah dunno, it's safe for XHR and <img>
09:50
<annevk>
zcorpan: I would say it should be safe for <track>
09:51
<annevk>
zcorpan: otherwise we might have some issues with that going forward
09:51
<annevk>
zcorpan: whether data URLs are allowed or not
09:53
<zcorpan>
i think chrome doesn't allow data: for <track> but i don't know if that's just a result of the default policy or if it's deliberate
09:54
<annevk>
I'm gonna add a flag to Fetch tomorrow or so and also change the text around data URLs
09:57
<tobie>
annevk: was wondering whether you were aware of https://code.google.com/p/chromium/issues/detail?id=362214 and https://sites.google.com/a/chromium.org/dev/Home/chromium-security/security-faq?pli=1#TOC-Which-origins-are-secure
09:58
<tobie>
annevk: and whether that's something you were considering to add to Url.
09:58
<tobie>
annevk: given there seems to be interest to implement something similar at Mozilla.
10:01
<annevk>
tobie: I have some vague plans on taking over the remaining bits of the Web Origin RFC and stick 'm in some existing spec
10:01
<tobie>
I see.
10:01
<annevk>
tobie: Origin header is already obsoleted by Fetch, URL could define some of the origin serialization algorithms
10:05
<tobie>
makes sense.
10:08
<annevk>
tobie: oh btw, https://w3c.github.io/webappsec/specs/mixedcontent/ will define the secure origin thing most likely
10:09
<annevk>
tobie: Mike West and I have discussed on how to add hooks to Fetch for that and such
10:09
<annevk>
tobie: so that e.g. https to http will return a network error, and other such goodies
10:09
<tobie>
I'll have a look at that. Thanks for the pointer.
10:32
<annevk>
zcorpan: where does <object> pay attention to response's status?
10:32
<annevk>
zcorpan: are you confused with Content-Type?
10:33
<annevk>
zcorpan: although <img> pays attention to that, so maybe I'm missing something
10:33
<zcorpan>
annevk: the spec's default is to pay attention to it (since http requires it)
10:33
<zcorpan>
annevk: <img> opts out
10:34
<annevk>
"HTTP requires it"? What does that even mean?
10:34
<zcorpan>
it's possible that all other features but <object> should be like <img>, i don't know
10:36
<annevk>
E.g. currently 401 requires an authentication dialog for features that opt-in (which due to legacy is all of them, except for a couple new ones)
10:36
<annevk>
But I don't think we've established a list of codes that a response has to treat as if it received a network error unless it opts in/out of that
10:36
<annevk>
s/codes/statuses/
10:37
<zcorpan>
i dunno. i mean the spec in general treats 404 like network error rather than like 200 OK
10:37
<zcorpan>
lunch
10:42
<annevk>
zcorpan: but is that actually true in implementations?
10:42
<annevk>
zcorpan: I seriously doubt it
10:48
<caitp>
it doesn't seem to be true in chromium
11:12
<zcorpan>
caitp: what did you test?
11:13
<beverloo_>
annevk, image is public domain/cc0, go for it :-) (can't find my bugzilla details)
11:16
<caitp>
zcorpan: if you try to load a document which results in a network error, you won't actually have the document opene, instead the browser will render some data:text/html,... to tell you about the network error
11:17
<caitp>
which is not true of 404
11:18
<zcorpan>
caitp: so you tested navigating top-level browsing context?
11:18
<caitp>
yes
11:18
<zcorpan>
ok, well that's not so interesting
11:19
<annevk>
beverloo_: cool
11:19
<zcorpan>
more interesting is what does <embed src> do, what does <link rel=stylesheet href> do, what does <video src> do, what does <script src> do, etc
11:19
<caitp>
i'm interested in an example where a 404 would be treated as a network error, with the exception of weird behaviour of webviews on certain mobile platforms
11:20
<caitp>
or file:// scheme requests
11:20
<zcorpan>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3047
11:24
<caitp>
> If the load failed (e.g. there was an HTTP 404 error, there was a DNS error), fire a simple event named error at the element, then jump to the step below labeled fallback.
11:25
<caitp>
so that part of the spec is saying treat 404 as the same as a particular network error
11:25
<caitp>
but that's not "http"
11:26
<caitp>
it's a particular behaviour of a particular html tag
11:28
<annevk>
zcorpan: interesting
11:28
<zcorpan>
caitp: yeah
11:31
<zcorpan>
seems <embed> is like <img> per spec
11:31
<annevk>
zcorpan: it would be interesting to know what status codes that was triggered with
11:32
<caitp>
in the cases of these elements it might make sense, because you probably don't want to render some domain's particular fancy 404 response
11:32
<caitp>
but it's not the same as saying "in http, 404 is a network error"
11:33
<caitp>
that's the point I'm making, that's all :>
11:33
<zcorpan>
i said "in html" not "in http" :-P
11:33
<caitp>
i could have sworn you said in http
11:33
<zcorpan>
i said "the spec" but i was referring to the html spec, not the http spec
11:34
<caitp>
<zcorpan> annevk: the spec's default is to pay attention to it (since http requires it) ... ] <annevk> "HTTP requires it"? What does that even mean?
11:34
<caitp>
it is very early in the morning here, though
11:34
<caitp>
no coffee yet =)
11:37
<caitp>
<zcorpan> i dunno. i mean the spec in general treats 404 like network error rather than like 200 OK
11:37
<caitp>
ok I see what you're saying
11:38
<caitp>
conflated that with the earlier mention of "http requires it"
11:40
<annevk>
Hmm, I find HTML very vague on this subject. Somewhat clear for 404, but there's a ton of other response codes
11:43
<caitp>
it would probably apply to all of the non-200s (with the exception of things like 302)
11:43
<caitp>
and probably not 204 either
11:44
<annevk>
At that point 3xx are already handled
11:44
<annevk>
Maybe not 304, that depends a bit
11:44
<annevk>
But yes, I think that's the approach CORS takes.
11:45
<annevk>
http://fetch.spec.whatwg.org/#cors-preflight-fetch This does not even consider 304
11:45
<annevk>
I guess we should consider that flattened at that point since it's an HTTP cache lookup
14:08
<darobin>
does anyone implement document.origin?
14:17
<annevk>
dunno
14:17
<annevk>
they should
14:20
<darobin>
annevk: apparently, no one at this point
14:20
<darobin>
though I'll admit I haven't checked Servo
14:20
<annevk>
file bugs
14:21
<annevk>
or create a test
14:40
<Ms2ger>
darobin, not Servo either, patch welcome ;)
16:45
<dglazkov>
good morning, Whatwg!
19:51
<zewt>
fascinating bug title: "Specify the Event Loop integration for various non-deprecated events"
21:13
<IZh>
The bug needs to be Hixied. ;-)
21:15
<caitp>
that could be interpreted in so many different ways, hmm
21:15
<IZh>
Fixed :-)
21:17
<caitp>
oscar wilde would have a field day
22:46
<zewt>
does that mean "specify the things that need to be specified and not just the ones I like"
23:54
<slightlyoff>
hey annevk, wanted to chat about fetch API
23:54
<slightlyoff>
specifically about isSynchronous in the API
23:54
<slightlyoff>
and the body property
23:55
<slightlyoff>
we'd talked about a toBlob() on the response to hand back a promise or similar so that we can get out of the contract that requires synchronous reification