02:19
<caitp>
TabAtkins: has there been any mail regarding the :target pseudo-class lately? I don't remember seeing any, but you'd probably be more on top of that?
02:19
<caitp>
as far as I know that should still work, right?
05:11
<zcorpan>
Hixie: so is the current content model ok or should it be "zero or more source elements optionally intermixed with script-supporting elements followed by one img"?
06:30
<zcorpan>
UA string is still race to the bottom. when will it stop? should we standardize on one fixed string?
06:33
<zcorpan>
Hixie: how can i run your pipeline?
06:48
<sangwhan>
If everyone got rid of Mozilla/5.0 in their UA string that would probably save enough bytes transmitted to give one small country free internet
06:54
<roc>
it sure would, because the entire Web would shut down
07:10
zcorpan
notices annevk changed "authors" to "developers" in the fetch spec
07:15
<MikeSmith>
author-developers
07:36
<foolip>
Ms2ger: you there?
07:36
<Ms2ger>
foolip, sir!
07:36
<foolip>
Ms2ger: do you have any juicy info on the compat situation with HTMLImageElement.x/y in Gecko?
07:37
<Ms2ger>
We tried removing it, it broke something
07:37
<foolip>
see this thread: https://groups.google.com/a/chromium.org/d/msg/blink-dev/zpLuYu8tmdA/uA_7TQSQDzQJ
07:37
Ms2ger
looks if he can find the site
07:37
<foolip>
I'm trying to figure out if there's any change to Blink that could bring us closer to agreement between spec and implementations
07:37
<foolip>
these are the relevant Gecko bugs:
07:38
<foolip>
https://bugzilla.mozilla.org/show_bug.cgi?id=98292
07:38
<foolip>
https://bugzilla.mozilla.org/show_bug.cgi?id=116843
07:38
<foolip>
https://bugzilla.mozilla.org/show_bug.cgi?id=587021
07:39
<foolip>
https://bugzilla.mozilla.org/show_bug.cgi?id=731832 too
07:39
<Ms2ger>
Yep, 731832 was the one that made us back it out
07:40
<foolip>
http://web.archive.org/web/20120107051245/http://www.crownhill.com/ looks modern :)
07:40
<Ms2ger>
Yep, that's the one
07:41
<Ms2ger>
Code that used it in http://web.archive.org/web/20110807212111/http://www.crownhill.com/buttons_1/xaramenu.js
07:41
<Ms2ger>
"if(NS4)return;"
07:42
<foolip>
if(NS6){if(mal==0){x=event.target.x-bd;y=event.target.y;dx=event.target.offsetWidth;dy=event.target.offsetHeight;} is the relevant bit, no?
07:42
<Ms2ger>
Yep
07:43
Ms2ger
got distracted
07:43
<foolip>
and ar NS6=(!document.all&&document.getElementById)
07:43
<foolip>
var
07:43
<foolip>
that's some pretty weird logic :)
07:44
<Ms2ger>
If that's the weirdest you've seen... ;)
07:45
<foolip>
what I really wanted to ask is if you can figure out what GetXY() in https://github.com/mozilla/gecko-dev/blob/master/content/html/content/src/HTMLImageElement.cpp actually means
07:45
<foolip>
in my testing it's usually like offsetLeft/offsetTop, but not always
07:46
<foolip>
is this layer thing something that maps to a CSS concept?
07:46
<Ms2ger>
I don't know
07:47
<Ms2ger>
I think roc explained or was going to explain it in the W3C bug to add it to the spec
07:47
<foolip>
ok, I'll see if I can find that bug
07:47
<foolip>
https://www.w3.org/Bugs/Public/show_bug.cgi?id=17844
07:49
<Ms2ger>
Comment 16 there was what I remembered
07:49
<foolip>
I guess that means I should ask roc about the details
07:50
<Ms2ger>
Also https://bugzilla.mozilla.org/show_bug.cgi?id=887660
07:58
<MikeSmith>
hsivonen: Any chance you can re-deploy v.nu and h5.v.nu this week? I fixed a bug that had broken <picture> validation.
08:03
<annevk>
zcorpan: seems to be preferred by some set of people
08:31
<annevk>
JakeA: Googleman, what does "impacts MVP" mean?
08:32
<annevk>
JakeA: oh wait, I know, minimal viable product
08:32
<JakeA>
yeeep
08:37
<tobie>
lean spec writing
08:37
<tobie>
you should do workshops. there's money to be made.
09:28
<hsivonen>
MikeSmith: chances for redeploying this week are good
09:32
<jgraham>
hsivonen: You sound like a magic 8 ball ;)
09:43
<sangwhan>
anyone here with knowledge about gecko's KeyboardEvent implementation?
09:43
sangwhan
nudges smaug____
09:51
<Ms2ger>
mayasuki
09:59
<smaug____>
sangwhan: pong
09:59
<smaug____>
what about it
10:00
<smaug____>
but yes, masayuki knows it best
10:00
<smaug____>
sangwhan: aren't you even in the same timezone as masayuki
10:05
<sangwhan>
smaug____: except i don't know masayuki. is he here or do i have to find him on irc.moz?
10:05
<zcorpan>
Hixie: fixed https://www.w3.org/Bugs/Public/show_bug.cgi?id=25585 but possibly that section is a bit confused now, and it doesn't cover img srcset, and there's overlap with the Introduction examples
10:07
<sangwhan>
Basically i have this odd test that only passes in gecko: http://tests.sangwhan.com/tests/generated_keyevent.html
10:07
<sangwhan>
Can't decrypt d3e fast enough to figure out if that's correct or not
10:07
<sangwhan>
smaug____: ^
10:10
<MikeSmith>
hsivonen: cool
10:12
<MikeSmith>
sangwhan: masayuki not on #whatwg afaik. not sure if he's on mozilla irc much either. seems like he communicates most through bug comments..
10:12
<smaug____>
sangwhan: in moznet
10:12
<smaug____>
and yes, MikeSmith is right
10:13
<smaug____>
sangwhan: note, as far as I know, only Gecko implements large parts of the spec
10:14
<smaug____>
sangwhan: oh, that test is odd
10:14
<smaug____>
it tests legacy keyCode
10:14
<smaug____>
http://mxr.mozilla.org/mozilla-central/source/dom/webidl/KeyboardEvent.webidl#50
10:14
<smaug____>
which aren't, AFAIK, defined to be in the dictionary by any spec
10:15
<smaug____>
but we had to support them, at least for now
10:16
<sangwhan>
smaug____: yes, it is odd. came from a external source, and i have no idea if it should work or not
10:16
<smaug____>
http://mxr.mozilla.org/mozilla-central/source/dom/webidl/KeyEvent.webidl and http://mxr.mozilla.org/mozilla-central/source/dom/webidl/KeyboardEvent.webidl explain that kind of part of Gecko's behavior
10:18
<MikeSmith>
sangwhan: btw I think they have a telcon every wednesday morning at 9am JST, on #webapps on irs.w3.org, and masayuki usually calls into that
10:18
<smaug____>
Well, spec says only "Legacy keyboard event implementations include three additional attributes, keyCode, charCode, and which."
10:18
<smaug____>
sangwhan: you could file a spec bug to clarify this
10:31
<jgraham>
glwt
10:34
<MikeSmith>
https://twitter.com/ronkorving/status/496582799181094914
10:34
<MikeSmith>
"Why the hell doesn't documentFragment support innerHTML? This is nuts."
10:36
<zcorpan>
foolip = rambo
10:36
<roc>
foolip: what exactly do you need to konw?
11:05
<annevk>
MikeSmith: we tried to define it at one point and got stuck iirc
11:19
<MikeSmith>
annevk: stuck on what, I wonder. maybe there's an open bug
11:19
MikeSmith
checks
11:19
<annevk>
MikeSmith: there's something about ShadowTree or whatever it is called these days and innerHTML too
11:20
<Ms2ger>
Yeah, there was something
11:20
<Ms2ger>
Broke jQuery, perhaps?
11:24
<MikeSmith>
founs https://www.w3.org/Bugs/Public/show_bug.cgi?id=16904
12:15
<Ms2ger>
zcorpan, r? https://critic.hoppipolla.co.uk/r/2180
12:18
<zcorpan>
Ms2ger: done
12:18
<Ms2ger>
Thanks
12:19
Ms2ger
merges
12:35
<Domenic>
Why don't we just standardize UA string?
12:40
<jgraham>
The entire world would collapse
12:40
<Ms2ger>
Domenic, there's a reasonable argument that they're useful for statistics on the server side, at least
12:41
<Domenic>
I mean, standardize it to one of the existing ones, not to something sane
12:41
<jgraham>
There's also a reasonable argument that not everything can be feature detected
12:41
<Domenic>
I guess
12:41
<Ms2ger>
Also, people are unlikely to stop having browser-specific code altogether
12:41
<jgraham>
But, more importantly, people *do* use it
12:41
<Domenic>
just seems inevitable
12:41
<jgraham>
And it turns out that changing it breaks sites, really badly
12:41
<Ms2ger>
More likely, they'd use things like var NS6=(!document.all&&document.getElementById)
13:02
<foolip>
roc: I emailed, as you noticed :)
13:02
<foolip>
now I've installed Netscape 4 to see how it originally behaved, since that's what both the initial Gecko and WebKit implementation pointed to
13:02
<foolip>
In case anyone has forgotten, NS4 was awesome
13:02
<Ms2ger>
I wonder if its source code ever escaped
13:28
<annevk>
JakeA: 9AM PDT is what?
13:28
<annevk>
found it
13:28
<Ms2ger>
6pm .nl
13:28
<JakeA>
annevk: 17:00 for me, 18:00 for you I guess
13:28
<annevk>
Ms2ger: .ch
13:28
Ms2ger
has no idea what .ch is
13:28
<Ms2ger>
Timezone-wise
13:28
<annevk>
Ms2ger: same
13:29
<Ms2ger>
Okay, wasn't sure
13:36
<annevk>
jgraham: you like discussing this stuff: https://www.w3.org/Bugs/Public/show_bug.cgi?id=26405
13:36
annevk
can't come up with suitable terms that'll stick
13:38
<tobie>
Votes for a URL primitive.
13:38
<tobie>
^ that would make parsing JS SOOOO much fun.
13:44
<annevk>
I'm not sure what you mean tobie, but that doesn't seem to relate to that bug
13:47
<tobie>
annevk: never mind me. I very often forget to add quotes around urls when writing JavaScript.
13:48
<tobie>
annevk: …which makes me think it would be kind of neat if the language of the Web supported URL primitives.
13:48
<annevk>
url`http://....` could be something I suppose
13:49
<annevk>
Not sure what the latest is on those kind of strings
13:49
<tobie>
Yeah. The syntax is so enticing. I can't wait. </sarcasm>
13:50
<tobie>
Anyway, why don't you call them URL strings and URL objects?
13:52
<annevk>
Yeah, URL input and URL objects was what I had so far
13:52
<tobie>
input's kind of weird imho.
13:53
<gsnedders>
tobie: hey, ES4X is what we really want for parsing
13:53
<annevk>
Well, it's input to the parser
13:54
<tobie>
yeah, but that's very specific to the URL spec.
13:55
<tobie>
If you want to talk about setting window.location to a URL xxx you probably want to use URL string (instead of input).
13:55
<jgraham>
FWIW I also said — string and — object
13:59
<zcorpan>
why not "URL" and {URL}
14:13
<jgraham>
IETF approve of your scare quotes
14:13
<Ms2ger>
[[URL]]
14:13
<zcorpan>
remember to register for tpac everyone
14:13
<zcorpan>
or are you guys not going?
14:13
<tobie>
zcorpan: would you call those handlebar URLs or 'stache URLs?
14:13
<zcorpan>
tobie: fiskmås
14:13
<annevk>
Probably not going to TPAC
14:40
<tobie>
JakeA: trying to wrap my head around the SW/cache streaming. Why are we teeing/cloning the streams instead of piping them through the client and to the cache? e.g. response.body.pipe(e.response).pipe(cache) or something
14:43
<annevk>
tobie: see Fetch for why you need to tee
14:44
tobie
looks.
14:44
<JakeA>
tobie: isn't that treating the browser renderer as a transform stream that returns just the same value?
14:47
<caitp>
some time ago, some guy started applying the word streams to things and telling people that it simplified things --- and then everyone jumped on the streams bandwagon waving their pipes everywhere
14:47
<caitp>
and that's why we're doing it
14:47
<tobie>
JakeA: well, doesn't the renderer immediately buffer the response stream and deserializes into something else? A JS string, a rasterized image, etc? (No idea what I'm talking about.)
14:48
<tobie>
caitp: the emperor's fully dressed, thank you.
14:54
<JakeA>
tobie: it'll transform it into dom nodes, video frame(s) etc etc. It won't buffer the whole response before transforming begins unless the format requires it
14:56
<tobie>
JakeA: makes sense.
15:16
<tobie>
annevk: where is teeing defined?
15:16
<annevk>
nowhere
15:18
<tobie>
ah
15:19
<tobie>
annevk: are there any plans for it to be defined somewhere?
15:19
<JakeA>
tobie: https://github.com/whatwg/streams
15:19
<annevk>
yeah, the plan is to have stream primitives for the platform and a stream API on top
15:20
<annevk>
hopefully the stream terminology can be reused by things that don't necessarily expose it in an API
15:24
<tobie>
oh, so teeing throttles the source's writes to the slowest of its destinations? Isn't that an issue when writing to cache and the client at the same time?
15:25
<tobie>
annevk: not sure what you mean by stream primitives in that case.
15:27
<annevk>
tobie: the underlying model I guess
15:28
<tobie>
annevk: thanks for clarifying.
15:29
<JakeA>
tobie: teeing throttles to the fastest of the destinations and holds the content in memory for the slowest, I believe
15:29
<tobie>
JakeA: upon reading the spec again, it seems that's voluntarily unspecified.
15:30
JakeA
shines the Domenic symbol into the cloudy sky
15:31
<tobie>
But the example implementation throttles all connections, though.
15:33
<tobie>
JakeA: I feel like the goal for SW/cache was to abstract away the underlying streams completely, and that they
15:33
<tobie>
arg
15:34
<tobie>
JakeA: …'re now totally leaking through.
15:34
<JakeA>
tobie: the goal was more to prepare for streams, but be able to ship sooner than that
15:57
<JakeA>
http://www.buzzfeed.com/jimwaterson/pictures-of-politicians-when-they-were-younger
15:57
<JakeA>
Well, that was the wrong channel
15:57
<JakeA>
but enjoy that anyway
15:59
<caitp>
would love to, but it turns out that 802.11n makes it very difficult to enjoy much of anything
16:08
<JakeA>
Hixie: I keep typing ArrayBugger, just in case you're looking for more specs to Britainify
16:10
<jgraham>
I think that how we all feel about js arrays
16:11
<gsnedders>
jgraham: that they're the best thing ever, and you really want JS engines to optimize them into typed arrays?
19:25
<Hixie>
if anyone has ideas of examples we could add to the HTML spec of how to use HTML elements or APIs in Web apps (as opposed to documents), please add them to https://www.w3.org/Bugs/Public/show_bug.cgi?id=25747
20:23
<TabAtkins>
caitp: No, nothing in particular about :target. Why?
20:23
<caitp>
i didn't think so, it was just a bug someone had filed, it's been addressed
20:23
<caitp>
just wanted to make sure :>
20:58
<Domenic>
So apparently DOCTYPE serialization is a mess
20:59
<Hixie>
DOCTYPE serialisation is nice and easy
21:00
<Hixie>
you just output the constant string '<!DOCTYPE HTML>'
21:00
<Hixie>
:-)
21:00
<Domenic>
it is only observable via XHR, and browsers all do the more complicated thing
21:08
<Hixie>
anyone got IE around who can tell me the output of http://software.hixie.ch/utilities/js/live-dom-viewer/?saved=3099 ?
21:10
<Hixie>
Domenic: i've no idea what that issue is about, can you elaborarte?
21:10
<Hixie>
elaborate
21:11
<Domenic>
Hixie: so I maintain jsdom, which is a server-side implementation of the DOM. We are trying to figure out when the user asks us "save this document to a file" how we should serialize it
21:11
<Hixie>
replace the DOCTYPE with '<!DOCTYPE HTML>'
21:12
<Domenic>
So, two minor problems: 1) that is not what browsers do when you serialize via XHR; 2) that is not what DOM S&P says
21:12
<Domenic>
(it is what HTML says)
21:12
<Hixie>
why are either of those problems?
21:12
<Domenic>
they are minor problems. e.g. we should probably fix DOM S&P
21:12
<Domenic>
and we should either fix browsers or fix the XHR spec to reflect browsers
21:13
<Hixie>
i don't understand the relevance of either (1) or (2) to what a server does
21:13
<Domenic>
they aren't really problems for us
21:13
<Domenic>
for me
21:13
<Domenic>
they are problems for the web that I have uncovered in the midst of this investigation
21:15
<jgraham>
Isn't DOM S&P the one that W3C forked and added lots of complexity to for no readon
21:15
<jgraham>
*reason
21:15
<Domenic>
to be fair, they seem to have fixed bugs or holes in some csaes
21:15
<Domenic>
Ms2ger hasn't had time to reincorporate those back into the original
21:17
<Hixie>
Domenic: oh the server side of this is irrelevant?
21:17
<Hixie>
now i'm even more confused
21:17
<Hixie>
what is the question i'm supposed to be answering?
21:18
<Domenic>
Hixie: sorry to confuse. I was just paging you guys in to let you know about a problem that we discovered, not to ask you to help us fix our code.
21:18
<Hixie>
what's the problem though?
21:19
<Domenic>
the two abovementioned problems. DOM P&S contradicting HTML; XHR contradicting reality
21:23
<Hixie>
ah
21:23
<Hixie>
i wonder if anything uses the HTML spec's definition at this point
21:25
<Domenic>
I believe XHR says to use HTML's, but browsers do not do that