00:06
<Hixie>
rillian: does http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2013-January/038705.html still represent the current state of mozilla's metadata-in-video API work?
00:08
<rillian>
Hixie: yes
00:09
<Hixie>
thanks
00:09
<rillian>
which is to say, I haven't done further work on it
01:58
<zewt>
idb thread cc fail
02:09
<jgraham>
Ms2ger: yes, it is the reviewer's responsibility to reopen any issues that weren't actually addressed by the commit (although of course it is polite if the code author helps them out)
02:12
<jgraham>
zcorpan: I didn't really understand your use case, since it seems like you will always be inside another (async_)step when you might do the thing that returns the value you want. But it seems strictly better than the current API
04:03
<kochi>
MikeSmith: yt?
09:01
<zcorpan>
if i were to encode U+D800 as utf-8, ignoring that it's invalid, would it be: ED 9F C0 ?
09:05
zcorpan
finds http://mail.python.org/pipermail/python-dev/2009-April/089134.html
10:25
<Steve____>
FYI: there will be a bug opened by chaals against the html spec about changing the conformance requirments for @longdesc, could you deal with it? as I woul;d like to stay at arms length from it given my roles as a11y chair and the controversy surrounding it
10:27
<Steve____>
oops wrong window ;-)
10:28
<Steve____>
darobin_: thats for you
12:10
<MikeSmith>
kochi: here now
12:35
<MikeSmith>
anybody have a link to the draft that defines custom elements?
12:35
<MikeSmith>
for Web Components
12:38
<MikeSmith>
nm
12:38
<MikeSmith>
found it
12:39
<MikeSmith>
next question: Is there yet a spec that actually defines the <decorator> element?
12:43
<JohnAlbin>
MikeSmith: https://dvcs.w3.org/hg/webcomponents/raw-file/tip/explainer/index.html
12:59
<MikeSmith>
JohnAlbin: thanks
12:59
<MikeSmith>
but note, that's non-normative
12:59
<MikeSmith>
it seems Dimitri has yet to write the actual normative draft for it
13:16
<JohnAlbin>
I just happened to have that doc open in my browser. :-)
13:27
<zcorpan>
what is the correct output of new RegExp('').source per spec?
13:30
<zcorpan>
and new RegExp('/').source ?
13:30
<zcorpan>
"The source property of the newly constructed object is set to S.", ES5 says
13:31
<zcorpan>
two paragraphs above, it says S is to be escaped in some way, but isn't particularly exact
13:31
<zcorpan>
"For example, if P is "/", then S could be "\/" or "\u002F", among other possibilities"
13:34
<zcorpan>
http://people.mozilla.org/~jorendorff/es6-draft.html#sec-15.10.5 has the same text
13:36
<hasather>
zcorpan: V8 has a bug it seems, https://code.google.com/p/v8/issues/detail?id=1982
13:45
<zcorpan>
well, sent an email to es-discuss
13:49
<hasather>
zcorpan: what's not clear by the spec?
13:49
<zcorpan>
hasather: what .source should be when Pattern is something that needs to be escaped, e.g. empty string, a slash, a newline
13:50
<hasather>
zcorpan: it says something about both those cases
13:51
<zcorpan>
hasather: the spec gives *two* examples of what to do for a slash, and no specific requirements
13:53
<hasather>
zcorpan: isn't "The characters / occurring in the pattern shall be escaped in S as necessary to ensure that the String value formed by concatenating the Strings "/", S, "/", and F can be parsed (in an appropriate lexical context) as a RegularExpressionLiteral that behaves identically to the constructed regular expression." a requirement?
13:54
<zewt>
that gives me interop twitches (why doesn't it specify exactly how to escape it?)
13:54
<zcorpan>
hasather: yes, it is, but it gives multiple choices
13:54
<zcorpan>
what zewt said
13:55
<zewt>
specs should specify what to do, not what the end result should be
13:55
<hasather>
zcorpan: zewt: right, it would be better if it just specified. Agreed. :)
13:56
<zcorpan>
not surprisingly, browsers suck in this area
13:56
<zcorpan>
when specs give a clear algorithm, browsers tend to implement the algorithm
13:57
<hasather>
zcorpan: apart from the V8 bug, do browsers differ in this specific case?
13:57
<zcorpan>
with text like this saying "yeah you should do the right thing, y'know!", the implementor probably goes "yeah whatever" and moves on to the next thing
13:57
<zcorpan>
for \n, opera/chrome/firefox all give a literal new line as .source even though it violates the requirement
13:58
<zcorpan>
so they don't differ, but they don't conform to the spec
13:58
<zcorpan>
for / and the empty string, they do differ, actually
13:58
<zcorpan>
firefox and opera returns empty string for empty string
13:59
<zcorpan>
opera and firefox return \/ for /, chrome returns /
13:59
<zewt>
my chrome gives \/
13:59
<zewt>
for console.log(/\//.source) (checking that I'm looking at the right thing)
14:00
<zcorpan>
zewt: try new RegExp('/').source
14:01
<zewt>
yeah that does it ... why are they different? maybe /foo/ just copies in the actual literal
14:02
<zewt>
i guess that points to another thing that might be different and/or cause compat issues--is it okay to always re-escape from scratch in order to meet the requirement, or could changing it when you don't have to (resulting in new RegExp(x).source != x) cause compat issues somehow
14:05
<zewt>
also, this seems like some ancient hack to work around injection issues on sites, but that might be irrelevant to the constructor path anyway (especially if Chrome isn't doing it)
14:07
<zewt>
if the escaping could be killed for the ctor, then you might not need anything special for literals; just say "let .source be the exact literal", since the literal itself must have been escaped in the first place
14:07
<zewt>
anyhow, off to work
14:46
<darobin>
Ms2ger: I can close this without taking it, right? https://github.com/w3c/testharness.js/pull/13
14:46
<darobin>
since I fixed it on the server
14:48
<Ms2ger>
darobin, yes
14:48
<darobin>
Ms2ger: cool
14:51
<jgraham_>
darobin: Can you mark your webidl parser issues somehow? I just endured several seconds of confusion :)
14:51
<darobin>
jgraham_: sorry man
14:51
<darobin>
I've just added a label
14:52
<darobin>
they're actually idlharness — the parser is in another repo
14:52
<jgraham_>
That also confused me :)
14:53
<darobin>
haha
14:54
<jgraham>
Ms2ger: You should drop http://critic.hoppipolla.co.uk/r/10
14:55
<Ms2ger>
jgraham, I think I've done that
14:55
<zcorpan>
oh, i have to subscribe to es-discuss.
14:55
<jgraham>
perfect, thanks
15:54
<jwalden>
zcorpan: http://krijnhoetmer.nl/irc-logs/whatwg/20130318#l-708 btw
15:54
<zcorpan>
jwalden: yeah i saw it, thanks
15:54
<jwalden>
cool
15:55
<zcorpan>
i sent an email to es-discuss. annevk said it might have been decided in the last meeting already
15:56
<annevk>
I remember reading something about it in the conclusion bits of the minutes
16:00
<jwalden>
I can't find anything in March 12-14 or January 30
16:00
<jwalden>
did see new Date(dateObj) is supposed to clone
16:00
<jwalden>
but that was the only date-ful stuff I noticed
16:17
<dglazkov>
good morning, Whatwg!
16:36
<jwalden>
morning
16:38
<zcorpan>
"somebody would need to develop a proposal that completely specifies the required escaping." - any takers? :-)
16:38
<zcorpan>
(re RegExp.source)
16:55
<Hixie>
btw, i posted https://plus.google.com/u/0/107429617152575897589/posts/iPmatxBYuj2 (re drm)
16:57
<annevk5>
Hixie: old news :p
16:57
<annevk5>
great post though
17:19
<TabAtkins>
zcorpan: What's the requirements for RegExp.source? The escaping, while not easy, isn't particularly difficult either.
18:02
<karlcow>
http://inessential.com/2013/03/18/brians_stupid_feed_tricks
18:45
<GPHemsley>
the formatting of TAG IRC logs is terrible
18:52
<marcosc>
jamesr: The scheme in the logo uri has been removed from all TR stylesheets.
18:52
<marcosc>
(you mentioned this as causing problems for http/https docs the other day)
18:52
<marcosc>
I asked the W3C to fix it, and they did :)
18:53
<marcosc>
see: https://www.w3.org/StyleSheets/TR/W3C-ED
19:06
<dglazkov>
annevk: you still around?
19:06
<annevk>
I'm alive
19:06
<dglazkov>
annevk: are we doing Event.path?
19:06
<annevk>
just minuted a session on httprange-14
19:07
<annevk>
dglazkov: dunno, someone suggested it might not be needed if we introduce something else?
19:07
<dglazkov>
annevk: ok, let me look on bug
19:07
<annevk>
dglazkov: I'd prefer we get more usage data first with the whole component thing before we add a bunch of features we later regret
19:07
<dglazkov>
alright
19:14
<jamesr>
marcosc, cool! thanks
19:15
<jamesr>
marcosc, for the record, how did you asked the W3C to fix it?
19:16
<marcosc>
jamesr: because I have a bunch of specs that I figured would also be affected
19:16
<marcosc>
I saw this happen recently in Chrome
19:16
<jamesr>
yeah, but i'm wondering what you actually did. send an email to some alias?
19:16
<jamesr>
i dunno how to report a problem like this
19:16
<marcosc>
ah
19:16
<marcosc>
I just emails sysreq⊙wo
19:16
<jamesr>
aha!
19:16
<marcosc>
emailed*
19:17
<marcosc>
they are fairly responsive
19:17
<marcosc>
otherwise, just annoy MikeSmith or darobin :)
19:17
MikeSmith
wakes up
19:17
<MikeSmith>
whadda need fixing?
19:19
<jamesr>
MikeSmith, already been fixed, but the issue was w3c logos loading over http:// for specs hosted on https://
19:19
<MikeSmith>
ah yeah
19:20
<MikeSmith>
so the general solution when you run into problems is to do what marcosc said
19:21
<MikeSmith>
that is, ping me or darobin (Robin Berjon) here
19:21
<MikeSmith>
or e-mail me at mike⊙wo
19:21
<MikeSmith>
mailing sysreq⊙wo works to but may take longer
19:24
<marcosc>
jamesr: syreq takes longer, but MikeSmith charges more
19:24
<marcosc>
:)
19:24
<MikeSmith>
I just thrive on the love
19:25
<marcosc>
:)
19:26
<jamesr>
sounds like a difficult choice
20:42
<MikeSmith>
dglazkov: if you're around, I wanted to ask if you have a normative draft for <decorator> yet
20:42
<MikeSmith>
and if not, when you might have one -- or maybe, who's going to write it?
20:42
<zcorpan>
TabAtkins: what needs doing is replacing http://krijnhoetmer.nl/irc-logs/whatwg/20130319#l-382 with an algorithm that states what to replace exactly
20:43
<MikeSmith>
dglazkov: maybe something that Dominic could work on?
20:44
<zcorpan>
TabAtkins: what i know so far is "" -> "(?:)", "/" -> "\\/", "\n" -> "\\n" (probably similar for all characters that JS considers as new line)
21:31
<rniwa_>
bot-watchers
21:31
<rniwa_>
ugh...
23:43
<TabAtkins>
zcorpan: Don't you want to escape *all* the \ characters, regardless of what they code for?
23:43
<TabAtkins>
/\n/ selects a newline, after all. It's a valid escape sequence.