02:13
<Hixie>
wow, you sure have a lot of tests
03:49
<Hixie>
woot, i finally found a real bug in the parser tests
03:54
<Hixie>
oh wow, no, it's a spec bug
04:12
Hixie
fixes the spec
04:12
<Hixie>
looks like validator.nu already does this per the new spec
04:13
<SamB>
cool
04:13
<SamB>
just pretend it was always this way
04:34
<Hixie>
bummo, finally got to a test that relies on the script data less than sign state
04:36
<caitp>
bummo
05:44
<Hixie>
woops, the last checkin's message is bogus
05:44
<Hixie>
damnit
05:44
<caitp>
quick, force-push
05:45
<caitp>
before it's too late!
05:45
<Hixie>
don't think svn supports that, does it?
05:46
<SamB>
Hixie: no, but I think you *can* edit the commit message
05:46
<Hixie>
yeah, using svnadmin apparently
05:46
<Hixie>
i wonder where the repo actually is...
05:47
<SamB>
could do strange things to git mirrors ...
05:47
<Hixie>
aha, here we go
05:47
<Hixie>
eh, it's the checkin message
05:47
<Hixie>
how bad could it be
05:48
<SamB>
history would diverge depending on whether or not they saw the commit with the old message
05:49
<caitp>
i'm sure svn has full support for being terrible just like git does
05:52
<Hixie>
there, history has been revised.
05:52
<Hixie>
la la la.
05:55
<Hixie>
oh great, i found a bug in my tokeniser that the tokeniser tests didn't catch
05:55
<Hixie>
oops
05:55
<Hixie>
ok well now that i've broken everything it's time for bed. nn.
06:40
<Ms2ger>
"RESOLVED: Specs that define obsolete features don't need to test those features to exit CR."
07:10
<annevk_>
foolip: apparently Hixie changed a commit message in SVN, will that have bad effects on your mirror?
07:11
<annevk>
Ms2ger: what, for real?
07:14
<Ms2ger>
Part III of the Seoul notes
07:14
<annevk>
Oh wow, this was not from the HTML WG?
07:14
<annevk>
o_O
07:15
<Ms2ger>
No, CSS
07:15
annevk
is not impressed
07:16
<Ms2ger>
You should know I don't pay attention to the HTMLWG :)
07:16
<annevk>
So is the table display model now obsolete?
07:48
<foolip>
annevk: how did he do that?
07:48
<foolip>
but more importantly, which commit?
07:48
<annevk>
foolip: the last commit I think
07:49
<foolip>
let me check
07:50
<foolip>
do you know what the old and new message was?
07:51
<annevk>
foolip: old is what http://html5.org/r/8665 has
07:52
<annevk>
foolip: http://krijnhoetmer.nl/irc-logs/whatwg/20140609#l-142
07:52
<annevk>
Ms2ger: with Anolis, why if I have <dfn>origin</dfn> I cannot refer to an origin defined elsehwere?
07:53
<annevk>
Ms2ger: why is data-anolis-spec not stronger?
07:53
<Ms2ger>
Mm
07:53
<Ms2ger>
That's probably because those are implemented in different processes
07:55
<Ms2ger>
Try https://pastebin.mozilla.org/5378908
07:55
<annevk>
I'll hack around it
07:56
<annevk>
I want to switch to Bikeshed at some point anyway
08:05
<foolip>
annevk: html-mirror is now in sync with svn, but you'll probably need to force update your end since it's not a fast-forward commit
08:07
<annevk>
foolip: is there a way I can always do that?
08:07
<annevk>
foolip: so I don't have to do it if it happens again
08:10
<foolip>
yeah, you could fetch and reset --hard in your script
08:24
<annevk>
foolip: I can't get it to change
08:24
<annevk>
foolip: I tried git fetch --force; git reset --hard
08:25
<annevk>
foolip: also tried git pull --force
08:26
<foolip>
annevk: try simply git fetch and git reset --hard origin/master
08:26
<foolip>
(assuming you're on the master branch)
08:29
<annevk>
foolip: that seems to have fixed it
08:30
<annevk>
foolip: should I just change my script to that then?
08:30
<annevk>
I guess if Hixie does it again I'll do that
08:34
<zcorpan>
heycam|away: doesn't webidl allow single quotes? https://dvcs.w3.org/hg/csswg/rev/aec2cc899bab
08:35
<zcorpan>
http://heycam.github.io/webidl/#prod-string
08:35
<Ms2ger>
Apparently not
08:35
<zcorpan>
oh well
08:36
<zcorpan>
i actually run a webidl checker in the build process but it's flooded with missing interface type definitions so i missed that thing
08:42
<foolip>
annevk: there's a small risk that if my mirror breaks completely, reset --hard on your end will blindly follow along, leaving the web interface also broken until someone notices
08:42
<foolip>
I don't think it's going to matter much what you do, it'll be some work either way
09:02
<annevk>
foolip: okay, left it as is for now
09:09
<zcorpan>
annevk: the css "obsolete" was in the context of features that are non-web and must not be implemented by browsers
09:09
<zcorpan>
Ms2ger: ^
09:11
<zcorpan>
i guess it wouldn't hurt with negative tests, but OTOH i think it's fine to exit CR always
09:27
<Ms2ger>
I see
09:32
<annevk>
Maybe I'll merge XMLHttpRequest in sooner. There's a lot of potential code sharing between XMLHttpRequest and fetch()
09:33
Ms2ger
looks what fetch() looks like
09:36
<Ms2ger>
Ah, GlobalFetch, there it is
09:37
<JakeA>
annevk: response.url, what did we decide that would be for constructed responses?
09:37
<annevk>
JakeA: the empty string, underlying object would be null
09:37
<Ms2ger>
annevk, do you want to implement Promises in Servo? :)
09:37
<zcorpan>
annevk: how do you suggest i fix addListener without invoking addEventListener?
09:37
<annevk>
zcorpan: you can use the underlying concepts just as addEventListener does
09:38
<annevk>
Ms2ger: not really
09:38
<zcorpan>
annevk: ok
09:38
<Ms2ger>
Would be nice if Servo did fetch() first
09:38
<annevk>
Servo should do Fetch first
09:39
<Ms2ger>
Not it :)
09:39
<JakeA>
annevk: So, if a page's CSP wanted to prevent constructed responses, we already have ways to detect that? We'd just need a CSP step after step 8 http://fetch.spec.whatwg.org/#http-fetch
09:40
<annevk>
JakeA: yes the idea for SW / CSP is to have a new step around there
09:40
<JakeA>
cool
09:41
<annevk>
JakeA: we still need to check if we do all the correct things with regards to redirects and such
09:41
<annevk>
JakeA: also currently response.url will be overwritten by request.url so we might need some distinct flag
09:41
<annevk>
JakeA: there's a bunch of things there we need to walk through carefully
09:42
<annevk>
I haven't done that, I'm mostly working on defining the API
09:43
<JakeA>
annevk: Is there any reason to overwrite response.url with request.url if the fetch is invoked recursively? That may be a good way to preserve it for the final CSP check
09:44
<annevk>
JakeA: yes, to make sure it's correct with respect to redirects
09:44
<annevk>
JakeA: otherwise e.g. responseURL on XMLHttpRequest would be wrong
09:44
<JakeA>
gotcha
10:17
<annevk>
JakeA: http://fetch.spec.whatwg.org/#fetchbodystream
10:18
<JakeA>
annevk: wfm!
10:31
<annevk>
JakeA: the main outstanding issue now is header representation
10:32
<annevk>
JakeA: I guess I'll try to tidy up everything aside from that
10:32
<annevk>
JakeA: if you could get some people to chime in on https://github.com/slightlyoff/ServiceWorker/issues/300 that'd be great
10:53
<toydestroyer>
Hi! Anyone here from Russian Google office in Moscow?
11:04
<JakeA>
annevk: Is anyone aside from Jonas against your proposal?
11:05
<annevk>
JakeA: not sure, also unsure whether HeaderMap is still the best name if it isn't really a map
11:05
<annevk>
Maybe just Headers
11:08
<JakeA>
annevk: Headers works, or HeaderData like FormData. I prefer just Headers.
11:08
<JakeA>
annevk: I think the proposal is spot on though. I don't think we need to make it a Map
11:08
<annevk>
Should we namespace all these classes? FetchHeaders, FetchRequest?
11:09
<annevk>
At the moment some are namespaced, some are not
11:14
<JakeA>
annevk: not keen on FetchRequest etc. Should it be HTTPRequest, or is it safe to assume http for browser requests?
11:14
<JakeA>
Although HTTPHeaders is clear
11:14
<annevk>
JakeA: HTTP is wrong for fetch()
11:14
<annevk>
JakeA: fetch() takes data URLs and such too
11:15
<JakeA>
annevk: Yeah, but when I do new Response(body) I'm creating an HTTP response
11:15
<JakeA>
anyway, I'm happy with just Response
11:15
<annevk>
JakeA: I'm not sure I see it that way
11:15
<annevk>
JakeA: e.g. Fetch (the spec) creates responses for data URLs too, they look like HTTP responses, but naming them such seems wrong
11:17
<JakeA>
fair
11:22
<jgraham>
annevk: I agree with your proposal fwiw
11:57
<Domenic>
annevk: so to('blob') ignores its argument if someone has previously called to('arraybuffer')? I think throwing an error would be better?
12:01
<Domenic>
errr return a rejected promise, of course
12:15
<Domenic>
annevk: > If the given value is not in the range 200 to 599, throw a TypeError. // should be RangeError
12:17
<annevk>
Domenic: you mean it would only work once?
12:18
<annevk>
JakeA: ^^
12:18
<annevk>
Domenic: happy b-day btw
12:19
<JakeA>
Domenic: Ohh, happy birthday! Are you allowed to drink in pubs yet? :P
12:19
<Domenic>
annevk: thanks :). and, gtg, but yes, once you consume the stream, it's all gone. can discuss more later, maybe will open an issue or something
12:19
<JakeA>
annevk: If streams can only be consumed once, then to() should reject on second calling
12:43
<annevk>
JakeA: we could make to() store the contents somewhere, but it's prolly better to just throw it away I guess
12:44
<JakeA>
annevk: throwing away sounds better. More compatible with streams and better for memory
12:53
<annevk>
fixed
12:56
<zcorpan>
mathiasbynens: you must be new here (re https://github.com/ResponsiveImagesCG/picture-element/issues/198 )
12:56
<mathiasbynens>
zcorpan: i guess I should go see the topic ^ now?
12:56
<zcorpan>
mathiasbynens: yes :-)
12:58
<zcorpan>
mathiasbynens: so in this case i'd expect people to use <picture> and test in a browser without support for <picture> (using picturefill or so) and conclude that everything is fine
12:58
<zcorpan>
but maybe it's not much of an issue
12:59
<zcorpan>
(or they test in a browser with support for <picture> but don't test dragging)
13:59
<annevk>
Domenic: should Response.redirect() throw RangeError too?
14:24
<caitp->
aitp
14:25
<caitp>
irc clients mang, they're not very good
14:50
<MikeSmith>
hey caitp
14:52
<caitp>
good morning MikeSmith senpai
14:52
<MikeSmith>
heh
14:54
<MikeSmith>
caitp: been meaning to ask you about https://github.com/w3c/web-platform-tests/pull/981
14:54
<MikeSmith>
Does that need review from annevk maybe?
14:55
<MikeSmith>
oh I see Simon commented on it
14:55
<caitp>
there was an issue opened about it on fetch or xhr, I forget
14:56
<MikeSmith>
ok
15:06
<Domenic>
annevk: ya RangeError there too I think.
15:06
<Domenic>
annevk: use GitHub issues and link to them from the top :)
15:07
<annevk>
Domenic: don't really want to migrate the existing setup
17:33
<hemanth>
From v0.11.6 I'm waiting for direct proxies in node..now we are at v0.11.13 still they havn arrived..:/ but why?
17:43
<Domenic>
hemanth: nobody is working on proxies in V8.
17:44
<Domenic>
https://code.google.com/p/v8/issues/detail?id=1543
17:46
<hemanth>
Domenic, Holy Goodness, it was reported on Jul 7, 2011! :(
17:47
<Domenic>
hemanth: implementer priorities are what they are... age of a feature doesn't really make much of a difference.
17:48
<annevk>
2011 is nothing
17:49
<hemanth>
Why are they finding it hard to sync up!? AFAIK SpiderMonkey is the only one which is almost up to the mark...
17:50
<zewt>
people have a lot to work on?
17:50
<hemanth>
But why have Proxy.create()? Rather remove it.
17:51
<TabAtkins>
annevk: Context for "obsolete doesn't need to be tested" is some property values that we're only defining in an informative appendix as having been supported by some printers, never browsers.
17:51
<Domenic>
SpiderMonkey also seems more willing to push out mostly-compliant, but unoptimized, versions of ES6 features.
17:51
<Domenic>
hemanth: they don't have Proxy.create() unless you turn on flags
17:51
<annevk>
TabAtkins: yeah, zcorpan mentioned that, I guess that's fine
17:52
<hemanth>
Domenic, Yeah, in any case for most of the ES6 features we need to turn on the flag
17:55
<hemanth>
I can understand it's tough for all the implementer to be in sync, but atleast when they are implementing a new spec, they can try to be, right? Each time a new spec is being implemented again the same mistakes are repeated....
17:55
<TabAtkins>
annevk: Yeah, things which need to be supported by browsers obviously need testing, even if they're obsolete and not to be used by authors.
17:55
<hemanth>
Sorry, but I'm finding it hard to understand the underlying problem.
17:56
<annevk>
hemanth: wrong channel to complain about V8
17:56
hemanth
sits at a corner.
17:57
<hemanth>
No response at #node.js; annevk. I'm not complaining, just trying to understand...
17:58
<hemanth>
<pachet> hemanth: because proxies are evil and will cause all of your friends to stop talking to you
17:58
<hemanth>
^ heh heh from #node.js
17:58
<annevk>
hemanth: there's no infinite resources for a project so there is some balancing between new features, improving existing features, fixing bugs, perf, security, etc.
17:59
<annevk>
hemanth: then there's been ongoing discussion on the proxy design over in TC39 over the past couple of years which will likely lower priority for implementing it, etc.
17:59
<annevk>
hemanth: age of the bug doesn't really matter, there's decade old bugs that haven't been fixed yet
18:01
<hemanth>
Thanks for the insight annevk.
18:02
hemanth
makes a note in his diary : "So it's all about implementer priorities and finite resources."
18:03
<caitp>
that's a recurring theme around here
18:03
<hemanth>
:)
18:04
<Hixie>
well "If the next token is a U+000A LINE FEED (LF) character token, then ignore that token and move on to the next one" is a huge pain in the neck...
18:04
<Ms2ger>
Indeed
18:06
<Hixie>
i guess the tree constructor just has to set a flag on the tokeniser
18:15
<hemanth>
Can anyone state an example for an exotic object?
18:15
<Hixie>
Window
18:15
<Hixie>
(right?)
18:15
<Hixie>
(or no?)
18:15
<Ms2ger>
[]?
18:15
<hemanth>
Window heh heh yup, [] not sure.
19:42
<annevk>
yay, feedback from Microsoft
20:00
<annevk>
JakeA: Domenic: if to() reads all, what happens if you then store it?
20:01
<annevk>
JakeA: Domenic: there won't be anything, that seems kinda sad
20:41
<Domenic>
annevk: what do you mean, "then store it"?
21:55
<estellevw>
on http://www.w3.org/TR/html-markup/input.button.html the constraints and admonitions make it seem like list is a valid attribute. While it does not mention the list attribute as a valid attribute in the first half of the document, adding "The list attribute of the input element must refer to a datalist element." on the button type page makes it seem like it would actually do something. Does anyone know who edits these pages, and if it makes sense to bu
21:55
<estellevw>
them to remove that line?
21:57
<Domenic>
estellevw: did you see the abstract at http://www.w3.org/TR/html-markup/Overview.html ?
21:58
<estellevw>
nope. i did not. thanks
22:01
<estellevw>
wow. those pages aren't as informative as the archive. I need to do some updating. (Anyone have a time traveler I can borrow so I can find the time.)
23:52
<Hixie>
hm, that's odd
23:52
<Hixie>
my mail to es-discuss didn't make the archives?
23:56
Hixie
tries sending it again
23:59
<Hixie>
hey anyone know how we're doing with dropping mutation events?