04:57
<JonathanNeal>
before i use the wrong word, is xml a syntax or a language or both?
04:58
<JonathanNeal>
the acronym is language, but i dunno, me bad at english
05:53
<Hixie>
JonathanNeal: it's a metalanguage that defines a syntax for defining language grammars for new languages based on annotated trees and a syntax for expressing those languages
06:02
<bret>
JonathanNeal: a markup language no?
06:03
<Hixie>
you could describe it as a markup language, depending on how you defined "markup language"
10:50
<zcorpan>
intredasting http://www.ietf.org/id/draft-kerwin-file-scheme-13.txt
10:55
<ondras>
On an OpenVMS Files-11 system, append a slash "/" to the URI, and encode the device name as the first segment as per step 5
10:55
<ondras>
heh
12:10
<annevk>
better link: https://tools.ietf.org/html/draft-kerwin-file-scheme-13
12:11
<annevk>
I'm not sure why the IETF keeps maintaining so many different canonical links, especially when everything not tools.ietf.org is so poor
12:30
<foolip>
annevk: sorry about isSameNode(), now the ball is with you :/
12:30
<annevk>
foolip: I hope you're okay with me just leaving the bug open
12:31
<annevk>
foolip: there doesn't seem to be a compelling need to make a decision at this point
12:31
<foolip>
annevk: indefinitely, or what input do you need to resolve it either way?
12:32
<annevk>
foolip: I guess compat issues for Gecko
12:32
<annevk>
foolip: or maybe compat issues for a new browser
12:32
<Ms2ger>
We killed it, didn't we?
12:32
<annevk>
Ms2ger: yup
12:32
<Ms2ger>
Then I'd prefer leaving it out of the spec :)
12:32
<foolip>
Ms2ger: yes, but my attempt to remove it in Blink was not met with the usual enthusiasm
12:33
<Ms2ger>
There's usual enthusiasm?
12:33
<foolip>
I'd say so, yes, when some crufty unused stuff is removed
12:34
<annevk>
Ms2ger: if there's a r+ patch on a bug, it needs a checkin-needed keyword right?
12:34
<Ms2ger>
Typically yes
12:35
<annevk>
Ms2ger: https://bugzilla.mozilla.org/show_bug.cgi?id=340494
12:35
annevk
just added it
12:36
<Ms2ger>
annevk, ... why is that bug closed?
12:36
<foolip>
annevk: you can leave it open, but I'd suggest finding out if IE or Safari are going to remove it
12:36
<annevk>
Ms2ger: hmm not sure, maybe that patch should have gone in a new bug
12:36
<annevk>
foolip: is Travis on the cc?
12:36
<foolip>
annevk: nope
12:40
<annevk>
foolip: okay, added and added hober and left a comment with what I'm inclined to do about this
12:40
<annevk>
Ms2ger: ta
12:41
<foolip>
annevk: thanks
12:57
<Ms2ger>
foolip, do you have tests for https://bugzilla.mozilla.org/show_bug.cgi?id=1109486 ?
12:58
<foolip>
Ms2ger: just the test in https://codereview.chromium.org/791783002/
12:59
<foolip>
we can't run serve.py tests (yet) so it's some silly cgi script
12:59
<Ms2ger>
Want to submit one?
13:00
<Ms2ger>
Not having to write tests is usually a nice carrot ;)
13:01
<foolip>
how trivial is it to echo the Content-Type header using serve.py?
13:02
<foolip>
Ms2ger: I'm afraid a per-spec test wouldn't pass in either Gecko or Blink because we both use application/xml unconditionally
13:03
<foolip>
I guess I could write a test using an XMLDocument to avoid that though
13:03
<Ms2ger>
Trivial
13:03
<Ms2ger>
I suspect .headers would work too
13:04
<Ms2ger>
Having HTML and XML tested both would be even better ;)
13:04
<foolip>
Ms2ger: then will you put back isSameNode() :P ?
13:05
<foolip>
so anyway, I'm trying to write a test for you
13:05
<Ms2ger>
No :)
13:05
<Ms2ger>
And thanks a lot
13:06
Ms2ger
should go out and buy chocolate for the holidays
13:07
<foolip>
Ms2ger: looks like web-platform-tests/XMLHttpRequest/send-entity-body-document.htm is already testing this
13:07
<foolip>
that was easy :)
13:08
<Ms2ger>
Ha
13:10
<jgraham>
foolip: So what do we have to do to get Chromium running tests with the proper server?
13:11
<foolip>
jgraham: I don't know, maybe setting up the hosts is non-trivial, or maybe nobody has tried it yet
13:12
<jgraham>
fwiw we use autoproxy instead of the hosts on infra
13:12
<foolip>
it sure would be nice, right now tests are being rewritten by a script to make paths non-absolute...
13:28
<zcorpan>
MikeSmith: in http://html5doctor.com/html5-check-it-before-you-wreck-it-with-miketm-smith/ "Should pre-HTML5 doctypes be flagged with a warning, in the W3C Validator, now HTML5 is a REC?" - the html4 doctypes are conforming, at least the ones that trigger standards mode
13:28
<Ms2ger>
Lol rec
13:29
<zcorpan>
MikeSmith: next question: "Nothing changed in browsers." isn't quite true, the gecko html parser had problems with it that went away when switching to html5 parser
13:38
MikeSmith
wonders what mechanism the html5 doctors have for printing retractions
13:39
MikeSmith
reminds himself to qualify more statements with "as far as I know" or "don't trust what I'm telling you"
13:47
<MikeSmith>
zcorpan: the transitional one is not conforming, right?
13:47
<MikeSmith>
it see like that's the one most often used
13:48
<MikeSmith>
the validator errors for that one still
13:50
<MikeSmith>
I thought we had been emitting a warning about the conformant ones but I guess not
13:51
<MikeSmith>
last I saw the spec said not to use those in documents
13:51
<MikeSmith>
should not not use
13:51
<MikeSmith>
*not
13:53
<Ms2ger>
MikeSmith, I think that changed to reduce warning fatigue
13:53
<MikeSmith>
ah ok
13:53
<Ms2ger>
And no real negative effect
13:54
<MikeSmith>
I guess I should actually check actual facts sometimes before I say things
14:10
<zcorpan>
MikeSmith: right, transitional doesn't trigger standards mode
14:10
<zcorpan>
MikeSmith: facts are overrated :-)
14:11
<MikeSmith>
zcorpan: I will quote you on that
14:11
<MikeSmith>
but I will misstate it when I quote you
14:12
<annevk>
Domenic: www-tag continues to amaze
14:13
<zcorpan>
looking forward to it
14:14
<MikeSmith>
about URL still?
14:14
<annevk>
Domenic: first saying how you could rely on free public caching and then demanding 24/7 phone support
14:14
<annevk>
MikeSmith: TLS
14:14
<MikeSmith>
ah even better
14:17
<annevk>
MikeSmith: someone is claiming that the move to HTTPS is killing independent hosting providers
14:18
<annevk>
MikeSmith: he's sort of a known troll though
14:20
<MikeSmith>
coming out from the woodwork and all that I guess
14:20
<annevk>
http://www.certificate-transparency.org/ not going over TLS o_O
14:21
<annevk>
MikeSmith: he pops up every now and then and considers himself a spokesperson for the independent developer
14:25
MikeSmith
briefly peruses the www-tag thread
14:26
<MikeSmith>
ah yeah that guy
14:28
<zcorpan>
Ms2ger: the "caching is allowed" check can never fail, right?
14:29
<Ms2ger>
zcorpan, it's not intended to be able to
14:29
<Ms2ger>
Covering all conformance reqs and all; I probably wouldn't write it today
14:31
<zcorpan>
should we replace it with a comment saying it can't be tested?
14:33
<Ms2ger>
*shrug*
14:33
<zcorpan>
yeah
14:34
<zcorpan>
ok let's leave it :-)
15:39
<krijnhoetmerbot>
zcorpan: done (I think)
15:42
<krijnhoetmerbot>
(For new logs that is, not for cached older ones)
15:55
<annevk>
krijnhoetmerbot: when will TLS arrive?
15:55
<krijnhoetmerbot>
Not on this machine I'm affraid
15:55
<annevk>
krijnhoetmerbot: I can help you out with a certificate if that's the problem
15:55
<krijnhoetmerbot>
*afraid
15:55
<annevk>
I see
16:01
<Domenic>
topic should probably be updated to https://whatwg.org/ instead of http://www.whatwg.org/
16:16
<annevk>
hmm, MikeSmith has to make that change I guess, seems locked
16:20
<krijnhoetmerbot>
Also added previous/next day links to the logs, yay
16:22
<krijnhoetmerbot>
Will regenerate last month's cache
16:23
<Domenic>
MikeSmith: no www anymore :)
16:23
<MikeSmith>
ah
16:32
<annevk>
jgraham: what's the status of the service worker test infrastructure?
16:34
<jgraham>
annevk: I think some infrastructure that Google used to write some tests landed
16:34
<annevk>
wanderview: ^^
16:36
<wanderview>
thanks... it would be great to the blink tests uplifted then... easiest way to get it into our tree
16:39
<jgraham>
https://github.com/w3c/testharness.js/pull/82 and https://github.com/w3c/testharness.js/pull/93
16:40
<MikeSmith>
annevk: I guess you've been talking with whoever wrote http://www.chromium.org/Home/chromium-security/prefer-secure-origins-for-powerful-new-features (Mike West) about revising it to use some concept/term other than "secure origin"?
16:41
<MikeSmith>
though the definition of what a "secure origin" is there doesn't seem obviously way off to me from what should be needed
16:41
<MikeSmith>
*Mike West?
16:41
<annevk>
MikeSmith: I work with him on https://w3c.github.io/webappsec/specs/powerfulfeatures/
16:42
<MikeSmith>
(not sure who wrote the document but I was guessing it was him(
16:42
MikeSmith
looks
16:42
<annevk>
MikeSmith: I don't care much for the original Chromium documents
16:42
<MikeSmith>
annevk: maybe the TAG draft should reference https://w3c.github.io/webappsec/specs/powerfulfeatures/ instead
16:42
<MikeSmith>
then
16:43
<MikeSmith>
should I raise a github issue for that?
16:43
<annevk>
MikeSmith: I guess you should
16:43
<annevk>
MikeSmith: or just PR it
16:43
<MikeSmith>
hai
16:43
<MikeSmith>
hmm yeah that's better I suppose
16:45
<MikeSmith>
annevk: btw "Requirements for Powerful Features" by itself is kind of an ambiguous title
16:46
<MikeSmith>
it could be about any kind of powerful feature at all
16:46
<annevk>
MikeSmith: I have tried to bikeshed the terminology around this for a while and stopped caring
16:46
<annevk>
MikeSmith: I only care about getting it grounded in first principles at this point, hopefully someone else can fix the other issues
16:47
<MikeSmith>
makes sense
17:51
<zcorpan_>
krijnhoetmerbot: very nice. thanks
17:55
<krijnhoetmerbot>
zcorpan_: np!
18:21
<wanderview>
JakeA: you mean if script does two XHRs or two Fetch() calls to the same URL, is the network layer smart enough to realize and only do one actual request? is that ok to do? for example, REST API endpoints could have side effects, etc
18:22
<wanderview>
seems its up to content to set the cache headers properly... so second fetch() goes to http cache
18:22
<wanderview>
and we live the race of two simultaneous network events
18:22
<JakeA>
wanderview: hmm, but it doesn't know the headers yet. Does Gecko wait on the pending response
18:22
<JakeA>
ahh ok
18:23
<JakeA>
Yeah, I don't think we can solve this with magic
18:23
<wanderview>
JakeA: I mean... I don't think the network layer can do that without eating potentially significant network calls to REST endpoints, etc
18:23
<JakeA>
wanderview: yeah, it seems like something you'd *sometimes* want to do rather than all the time
18:24
<wanderview>
yea, and I think the cache headers are the way to control that
18:29
<JakeA>
wanderview: if tab 1 requests "/hi", then tab 2 requests "/h1", will the request in tab 2 wait for headers to arrive for the tab 1 request to see if they're a match?
18:29
<JakeA>
I mean cachable
18:30
<wanderview>
JakeA: I don't know... that seems like something without a clear tradeoff... good in some cases and bad in others... my guess is they opt for the simple case and just do the second network request
18:31
<JakeA>
Makes sense. Not sure what Chrome does
18:34
<wanderview>
it would also probably have to be "wait X amount of time for request 1 headers, and do request 2 if that times out"
18:38
<wanderview>
JakeA: ok... wow... we actually do have some waiting/request coalescing for <img> elements... everything else races
18:39
<JakeA>
wanderview: I think the img stuff is actually in the spec
18:39
<wanderview>
ah... ok
18:39
<JakeA>
Good to know though
19:05
<TabAtkins>
annevk: Re: layout boundaries, that's what http://dev.w3.org/csswg/css-containment/ is all about (but even stronger)
20:55
<annevk>
TabAtkins: can we combine that with local resize events to get something like element queries?
21:23
<annevk>
Domenic: jreschke appears to have been coopted: https://tools.ietf.org/html/draft-reschke-objsec-01
21:25
<Domenic>
annevk: I am so confused.
21:28
<annevk>
Domenic: ah, where it says "Intermediary-aware" you want to read "MITM"
21:29
<Domenic>
oh, i seeee
21:29
<annevk>
Domenic: basically at the moment the IETF is not requiring authenticated encrypted connections for HTTP/2.0
21:29
<annevk>
Domenic: as I understand it, Google will require TLS, Mozilla will require OE, and Microsoft might not require either (support the same mess as HTTP/1.1)
21:30
<annevk>
Domenic: the telcons are in favor of a HTTP/1.1-like setup so they can continue tracking you and injecting ads
21:32
<Domenic>
but of course
21:32
<Domenic>
it's unfortunate the incentives here are in favor of microsoft's position
21:33
<jamesr__>
what's OE?
21:33
<annevk>
Ok, so https://twitter.com/hillbrad/status/542740425933852672 is paraphrasing
21:33
<annevk>
jamesr__: MITM encryption
21:34
<annevk>
jamesr__: or less offensive, it'd be Opportunistic Encryption
21:34
<jamesr__>
annevk: you said two things, but they seem to be in conflict
21:35
<jamesr__>
i see. encrypt if everybody in the middle says it's OK to do so
21:35
<annevk>
jamesr__: it's encryption without authentication
21:36
<jamesr__>
pay the CPU cost and complexity of encryption without actually protecting you from attackers. coolbeans
21:36
<annevk>
jamesr__: I've been trying to understand why this is a thing we want the web to invest effort in for a while now
21:37
<annevk>
jamesr__: I believe the argument is that more encryption is better than none and that for Mozilla this was not a lot of effort to support
21:37
<annevk>
jamesr__: and that authenticated encryption comes with a lot of baggage around Referer and Mixed Content...
21:38
<annevk>
jamesr__: which is all true, but that still doesn't mean it's a good idea, but obviously not everyone at Mozilla is aligned on that
21:53
<annevk>
Whoa, Flash download is without TLS
21:53
<Hixie>
is it a signed binary?
21:54
<annevk>
I'm not sure how I can tell
21:55
<TabAtkins>
annevk: Actually, scratch that, I forgot that containment is about containing the painting effects of an element to within its boundaries, not stopping layout from leaking out.
21:55
<TabAtkins>
Stopping layout from leaking out is, more or less, what element queries will end up doing.
21:55
<annevk>
TabAtkins: are we still calling it element queries?
21:55
<annevk>
I guess it kind of makes sense...
21:56
<TabAtkins>
As EQ is just a combination of 1) some way of telling an element to stop paying attention to its contents for layout, and 2) some way for the contents of such an element to query the size of said element, likely via a selector.
22:01
<smaug____>
I wonder where Web Performance WG spec bugs should go
22:02
<annevk>
the /dev/null mailing list they run?
22:02
<smaug____>
ah, github :(