00:05
<Hixie>
fantastic
00:05
<Hixie>
for china, gecko uses GB18030, chrome uses GBK, and windows uses windows-936
00:05
<Hixie>
note that windows-936 isn't even in encoding.spec.whatwg.org
00:06
<Hixie>
and wikipedia says of it, "Originally it was identical to GB 2312, and expanded to cover most part of GBK with the release of Windows 95; now superseded by Code page 54936 (GB 18030)."
00:07
<Hixie>
"GB18030 also maintains compatibility with Windows Codepage 936, sometimes known as GBK, which is Microsoft's extended version of GB2312, with the exception of the euro sign which is given a single byte code of 0x80 in Microsoft's later versions of GBK and a two byte code of A2 E3 in GB18030"
00:07
<Hixie>
fricking encodings
00:10
<Hixie>
i love it. gecko, chrome, and Windows all use different names for shift_jis.
00:16
<Hixie>
only gecko has the locale 'ku'
00:16
<zewt>
Hixie: "and windows" do you mean IE?
00:16
<Hixie>
i did a google search for [locale ku] and all that comes up is firefox packages
00:16
<Hixie>
zewt: i mean vista default code pages - http://msdn.microsoft.com/en-us/goglobal/bb896001
00:17
<Hixie>
(ku is kurdish, it seems)
00:43
GPHemsley
wonders what the context is
01:26
<GPHemsley>
serialize( type, subtype, parameters )
01:26
<GPHemsley>
" The MIME type portion of a parsable MIME type is the result of serializing the type and subtype that result from parsing the parsable MIME type and null parameters. "
01:27
<GPHemsley>
Does that make sense? It's supposed to mean: serialize( type, subtype, null )
01:27
<GPHemsley>
For reference: " A parsed MIME type is the result of parsing a parsable MIME type. A parsed MIME type is made up of a type, a subtype, and a dictionary of parameters. "
01:40
<GPHemsley>
"A parsable MIME type is a MIME type for which the parse a MIME type algorithm does not return undefined. Every parsable MIME type has a corresponding parsed MIME type, which is the result of parsing the parsable MIME type. A parsed MIME type is made up of a type, a subtype, and a dictionary of parameters. "
01:40
<GPHemsley>
"The MIME type portion of a parsable MIME type is the result of serializing the type and subtype of its parsed MIME type with null parameters. "
01:44
<GPHemsley>
tell me github is down again
01:45
<GPHemsley>
" 1:45 UTC Major service outage. "
01:45
<GPHemsley>
argh
02:07
<GPHemsley>
Hixie: If you could bump the wiki server to PHP 5.3.2 or higher (preferably 5.4.x), I could update the wiki to the latest MediaWiki software. (With our current version of PHP, I'm restricted to the 1.19.x line we're currently on.)
02:07
<Hixie>
let me look
02:08
<Hixie>
odd, the only options i get are labeled "old, not recommended"
02:09
<GPHemsley>
hmm
02:10
<Hixie>
wtf
02:10
<GPHemsley>
that's very strange
02:11
<GPHemsley>
well, no rush
02:11
<GPHemsley>
I just realized recently I hadn't been keeping the wiki software up to date
02:12
<Hixie>
looks like the machine doesn't have a later php, let me see if i can fix that...
02:13
<Hixie>
no idea. let me send a support request.
02:14
<GPHemsley>
k
07:01
<matjas>
annevk: see logs starting here http://krijnhoetmer.nl/irc-logs/whatwg/20130611#l-936 (if you haven’t already)
07:02
<matjas>
annevk: also, any news from DreamHost yet (without me CC’ed, perhaps)?
07:02
<annevk>
matjas: I thought they reactivated the account. The main problem was you compiling Node.js which they don't allow
07:03
<annevk>
matjas: search for "support⊙dc" in your spam maybe? Email went out June 6
07:03
<matjas>
annevk: ah, that explains it. thanks!
07:17
<annevk>
I should probably read the logs... Been somewhat busy last couple of days.
07:52
<nessy>
annevk: nice effort in the captions discussion, btw!
08:03
<jgraham>
Can I be grumpy for a moment, and mention that the presentation of this whole "Extend the Web Forward" thing really distracts from the useful technical discussion? Presenting a point of view as a "manifesto" prominently featuring a list of endorsments comes across as a little self-aggrandising and amounts to a form of argument by authority. Also, asking people to tweet some hashtag suggests an intention to devolve technical discussions to popularity
08:04
<jgraham>
Note that I'm not complaining about (or commenting on) the technical aspects; only the presentation, which in my eyes leaves good people looking worse than they deserve.
08:06
<annevk>
nessy: thanks
08:06
<annevk>
nessy: mostly tried to relay the concerns from rillian et al
08:07
<annevk>
kinda sounds like it'll be okay either way
08:11
<annevk>
jgraham: Are you talking about the document itself or the blog posts around it?
08:13
<jgraham>
annevk: To a certain extent both. Although I'm sure I haven't seen all the blog posts
08:15
<annevk>
jgraham: having a new set of principles to rally around is kind of a hard concept to communicate
08:15
<annevk>
jgraham: see e.g. the HTML design principles for what happened last time
08:15
<annevk>
jgraham: 1JS didn't go smoothly either I believe, but I'm not really familiar with that discussion
08:16
<jgraham>
The HTML design principles worked extremely well
08:16
<annevk>
jgraham: yes, but getting the people on board who were not already on board were extremely skeptical of them is what I mean
08:17
<foolip>
annevk, nessy, which discussion is that?
08:17
<annevk>
jgraham: there was quite some resistance to "this is how it is"
08:17
<annevk>
foolip: [redacted]
08:18
<jgraham>
annevk: I think this approach is even worse for people who don't buy into the philosophy
08:18
<jgraham>
It's practically self-identifying a clique
08:20
<hsivonen>
Are there some AC meeting minutes that I get to read somewhere?
08:20
<annevk>
hsivonen: https://lists.w3.org/Archives/Member/w3c-ac-forum/2013AprJun/0430.html
08:21
<annevk>
jgraham: you mean with people writing their name at the end of it?
08:21
<hsivonen>
annevk: thanks
08:25
<annevk>
jgraham: might be easier to understand your feedback if you got more concrete, either here or in private
08:32
<jgraham>
annevk: With the list of names, the fact that there wasn't really any public discussion, etc. At least with the HTML design principles we could point to the fact that they were extracted from what we were doing anyway, and were formalised in public.
08:35
<annevk>
jgraham: so this similarly follows from public activity: web components, navigation controller, and promises
08:35
<annevk>
jgraham: it was written down as some higher level set of principles to get more people on board
08:35
<annevk>
jgraham: and as an explanation of what is happening
08:39
<jgraham>
Right, the specs are public (although suggesting NavigationController has been designed in public is somewhat laughable; a high bandwidth F2F makes sense, but locking it up in a private repo for months afterwards is inexcusable), but there has been no public discussion about whether the high level principles make sense. They have just been presented as a fait accompli. And, to circle back to the original point, that presentation has a strong element
08:58
<SteveF>
interesting discussion in the AC minutes, ashame its memeber only
08:58
<hsivonen>
Quite [redacted] that [redacted] is [redacted] in the AC meeting.
08:59
<Ms2ger>
I'm quite [redacted] that you feel that way
08:59
<annevk>
[redacted]
08:59
<Ms2ger>
[DRMd]
09:01
<annevk>
jgraham: I basically agree with that
09:01
<annevk>
jgraham: slightlyoff thinks brainstorming in public before the details are done might get you dragged down somehow, but that's certainly not my experience
09:02
<annevk>
jgraham: I recommend raising this somewhere
09:38
<hsivonen>
annevk++ for the AC meeting.
10:03
<nessy>
annevk: nothing new really, but you are right - I don't expect people that want to continue working on WebVTT to join the TTWG because WebVTT there will only be taken though to rec - the real development in my eyes continues in the TTCG
10:06
<nessy>
foolip: just a discussion of the new proposed charter for the TTWG in the AC meeting - I think it's ok to mention that a discussion took place
10:07
<annevk>
nessy: btw, you're an AC rep these days?
10:11
<matjas>
annevk: fyi, dreamhost seems to have unblocked my account now. thanks for the help!
10:12
<nessy>
annevk: yes, my new employer made me their AC rep - really interesting new role!
10:14
<darobin>
AC rep is the best thing to do in W3C
10:14
Ms2ger
read "<annevk> yes, my new employer made me their AC rep - really interesting new role!"
10:14
<Ms2ger>
That was rather surprising
10:14
<darobin>
whenever you're pissed off, in a bad mood, or whatever, you can just post a big grumpy rant on ac-forum
10:14
<darobin>
it's rather liberating
10:14
<darobin>
even better if drunk
10:15
<hsivonen>
darobin: with the implication that your rant is backed by some $$$.
10:15
<Ms2ger>
That's the main difference with Bj�rn / www-archive, I guess
10:15
<darobin>
hsivonen: better, you're advising, so it has to be constructive
10:16
<darobin>
you have options, too, it doesn't have to be about the team; it can be about AB reps, who are elected by you and therefore take everything politely :)
10:16
<nessy>
darobin: I just stand on the sidelines and shake my head on most of those discussions
10:17
<darobin>
nessy: :)
10:17
<darobin>
you're too kind
12:33
<TabAtkins>
zcorpan: Added text to the parser that allows the algos to be invoked directly on strings.
12:34
<zcorpan>
TabAtkins: thanks
12:34
<TabAtkins>
For "parse a CSS value", it looks like you've already got a token stream, and are just matching it against the grammar, right? If so, Syntax doesn't need to do anything further.
12:34
<TabAtkins>
If you want a token stream to match against the grammar, invoke "parse a list of component values" first.
12:35
<TabAtkins>
Go ahead and put the <an+b> serialization rules in CSSOM, yeah.
12:38
<zcorpan>
TabAtkins: "parse a CSS value" is invoked with a string
12:39
<zcorpan>
by http://dev.w3.org/csswg/cssom/#dom-cssstyledeclaration-setproperty
12:40
<TabAtkins>
Okay, so since grammar productions ultimately simplify to tokens (/component values), you should invoke Syntax there.
12:40
<zcorpan>
"parse a list of component values"?
12:41
<TabAtkins>
Yes.
12:41
<TabAtkins>
Which'll produce a list of tokens + functions and blocks.
13:06
<zcorpan>
should i say to match the list against the grammar for the property? is that always a boolean result or can a property alter the list?
13:08
<hsivonen>
does anyone happen to have examples of pages that require BOM to override HTTP in order to work? The pages were broken when I made the Gecko change have migrated away from UTF-16 since then
13:14
<zcorpan>
TabAtkins: ^
13:21
<SimonSapin>
TabAtkins: is "a string" unambiguously Unicode and not bytes?
13:24
<zcorpan>
from CSSOM's point of view it's DOMString i.e. 16-bit units
13:25
<zcorpan>
not sure what should happen to lone surrogates
13:25
<SimonSapin>
uh
13:25
<zcorpan>
iirc document.write just lets lone surrogates through
13:25
<SimonSapin>
well, anything non-ASCII is a "name character" for the tokenizer, so it should round-trip
14:24
<sangwhan>
media.readyState has a note about being able to jump between readyStates discontinously - does that mean a compliant implementation can jump straight from HAVE_NOTHING to HAVE_CURRENT_DATA?
14:25
<sangwhan>
(While bad, I've seen a corner case where this happens. Just wondering if that can be considered a compliance issue)
14:25
<hsivonen>
Is there a summary of the <hgroup>/<subhead> bikeshed in a few sentences?
14:25
<Ms2ger>
"W3C tries to assert dominance"
14:26
<Ms2ger>
Is one "a few"?
14:26
<hsivonen>
Ms2ger: W3C or SteveF?
14:27
<Ms2ger>
Dunno, I look from afar
14:27
hsivonen
moves to the next thread
14:28
<SteveF>
hsivonen: bikeshead is good, nothing to do with dominence
14:29
<SteveF>
hsivonen: trying to work out if people really want/need some way to identify a subheading (that works)
14:30
<SteveF>
hsivonen: my take is that people should make use of 'custom elements'
14:31
<hsivonen>
SteveF: custom elements seems like a bad story for something as common for static text
14:31
<hsivonen>
depends on whether one believes commonly-undestood copyable and pasteable markup has value
14:32
<SteveF>
hsivonen: i agree, but think they will be used to make all sort crazy semantic stuff
14:32
<hsivonen>
semantics of custom elements will be an illusion. lots of 386 ahead :-(
14:33
<SteveF>
semantics of native elements is often an illusion
14:33
<Ms2ger>
Semantics is an illusion, sheeple
14:33
<SteveF>
unless they actually do something useful
14:33
<hsivonen>
SteveF: yeah
14:35
<SteveF>
hsivonen: if you want to read something about subhead i try to put it into perspective here http://lists.w3.org/Archives/Public/public-html/2013Jun/0025.html
14:37
hsivonen
notes "semi-mythical outline algorithm" :-)
14:38
<hsivonen>
SteveF: thanks
14:38
<SteveF>
np
14:38
<hsivonen>
I'm unhappy to find more MPEG-2 references in my backlog of HTML WG email.
14:44
<jgraham>
I'm pretty sure that along the axis of "less valuable to have have known semantics" to "more valuable to have known semantics" headings are on the "more valuable" end
14:45
<jgraham>
Saying "custom elements" is more or less like saying "just use <font>"
14:47
<Ms2ger>
People here might be interested in https://groups.google.com/a/chromium.org/forum/#!topic/chromium-dev/wDV9JHs0mBA as well
14:47
<darobin>
I agree that custom elements aren't a good way of handling subheaders
14:47
<darobin>
slightly better than <font> though :)
14:48
<jgraham>
<font class="subheading"> if you want it to be readable :p
14:48
<SteveF>
darobin: the custom elements comment was a light hearted one
14:48
<darobin>
SteveF: it seems to be discussed seriously though :)
14:49
<jgraham>
Although really people will be writing templates, so more like {{subheading}}foo{{/subheading}} that expands to <font size=24 face="Comic Sans">foo</font>
14:49
<SteveF>
jgraham: i have been trying to work out how custom elements work, i believe you can extend an element like thus <p is="fancyparagraph></p> which leaves the base semantics (as far as acc API) is concerened, intact
14:50
<darobin>
SteveF: you don't need web components for a custom element that has no behaviour beyond styling
14:51
<SteveF>
darobin: right, but people want a way to define stuff like custom semantic elements to make them more real and legit
14:51
<darobin>
mulling it over a bit, I reckon that <subwhatever> would mostly be useful if it had sane default styling (unless there are e.g. outlining use cases I haven't thought of)
14:52
<SteveF>
darobin: did you read the email i pointed henry to just before?
14:52
<jgraham>
SteveF: I might have lost track of custom elements a bit, but last I heard that syntax was only there to placate Hixie and everyone else planned to use something like <x-foo> (I am probably a lot behind though)
14:53
<jgraham>
They might even have dropped the x- bit
14:53
<darobin>
SteveF: hmmm, so you thinking about something like https://gist.github.com/anonymous/be98f147a2f885257108 ?
14:54
<Ms2ger>
I hear they now use <foo-bar>
14:54
<darobin>
SteveF: yes I read it, I know the UC
14:54
<SteveF>
jgraham: right, and I have been looking at the examples in polymer, all of which appear to be of the interactive UI with no useful information exposed and no keyboard interaction
14:55
<darobin>
polymer is a rather impressive undertaking, but the examples could be better
14:55
<SteveF>
darobin: yes that sort of thing
14:55
<SteveF>
darobin: now they are only examples, but doesn't auger well
14:56
<darobin>
SteveF: the thing is, given that to get that you'd have to either inline the <element>...</element> or use <link rel=import href=path/to/element.html> you're going to need to use many subheadings before it becomes more useful than <span role=subheading>
14:57
<darobin>
I don't think it bodes that bad
14:57
<SteveF>
darobin also <element name=sub-head extends="p"> is supposed to mean the it uses the pelement thingy in the DOM
14:57
<darobin>
those are clearly hacker examples, it doesn't necessarily reflect usage
14:57
<darobin>
right, yes, that would work
14:58
jgraham
has the opinion that <hgroup> might not be perfect, but it is likely good enough, and certainly not so bad as to be worth the months of anguish
14:59
<SteveF>
jgraham: no anguish
14:59
<darobin>
more like some form of vague angst
15:00
<darobin>
the kind you'd find in a 1920s Austrian experimental film
15:00
<SteveF>
jgraham: i don't think a formal feature is needed but if it was hgroup would not be my choice...
15:01
<jgraham>
Well there was a poll of some sort and a WHATWG/W3C fork and multiple(?) extension specs and lots and lots of email that I stopped reading. Which seems quite like anguish to me.
15:11
<GPHemsley>
hsivonen: See also: https://gist.github.com/jonathantneal/5672276
15:12
<GPHemsley>
hsivonen: SteveF's mailing list post doesn't take into account the discussion that JonathanNeal and I had which suggested separate use cases for <hgroup> and <subline>.
15:15
<GPHemsley>
hsivonen: (Note that I don't necessarily agree with all that JonathanNeal wrote in that document; you might want to check the logs from around that time to see the full discussion.)
15:27
<hsivonen>
GPHemsley: thanks
16:08
<SteveF>
jgraham: "Well there was a poll of some sort and a WHATWG/W3C fork and multiple(?) extension specs and lots and lots of email that I stopped reading. Which seems quite like anguish to me." think it was a fair indication that it was a flawed feature
16:09
<Ms2ger>
subhead?
16:10
<SteveF>
Ms2ger: thats not a feature its an idea
16:10
<jgraham>
SteveF: Or an indication that people are apt to get hung up about really unimportant things and overlook significant ones
16:11
<SteveF>
jgraham: depends on what you think is important
16:15
<SteveF>
jgraham: there was an intransigence in the W3C wg at that time that is no longer evident, which makes it easier to get stuff done or undone as the case may be
16:35
<Hixie>
jgraham++
16:35
<Hixie>
bummer, missed anne again
16:35
<Hixie>
i got up at like 7am and i still missed him
16:35
<Hixie>
what crazy schedule is the man on!
16:36
<Hixie>
hsivonen: want me to move to Traditional vs Simplified, or should I leave the region names from Vista's list?
16:36
<Hixie>
hsivonen: (i'm happy to go whichever way you prefer on this)
16:38
<SimonSapin>
Hixie: I think he’s still in Japan
16:40
<Hixie>
any idea if he's getting up late or early?
16:40
<Hixie>
i'll try again this evening i guess
16:41
<SimonSapin>
we also that that new thing called email ;)
16:41
<Hixie>
wat
16:41
<SimonSapin>
s/that that/have that/
16:44
<Hixie>
(actually one of the things i want to ask him is why he didn't reply to my e-mail :-P )
16:44
<SimonSapin>
oh
16:45
<Hixie>
GPHemsley: ok so apparently i'm on debian 4 and to move to newer php we have to upgrade to debian 6
16:45
<Hixie>
this seems like a win in general
16:45
<Hixie>
however
16:45
<Hixie>
however
16:45
<Hixie>
expect EVERYTHING to break this week
16:45
<Hixie>
:-)
16:46
<SimonSapin>
Why stop at Debian 6? 7 is stable
16:47
<Hixie>
maybe dreamhost haven't got 7 ready yet, who knows
16:52
<reyre>
Hixie: for the webvtt API is there any validation things that we need to do? like the end time of a cue can't be greater then the start time of the cue? the spec doesn't say anything about that
16:53
<Hixie>
validation where?
16:54
<reyre>
Hixie: so when a user sets the endTime, or any of the properties on a WebVTTCue, should there be any validation taking place?
16:55
<Ms2ger>
What does the spec say? :)
16:55
<Hixie>
not per the spec, currently
16:56
<reyre>
Hixie: hmm this my bad.. the WebVTTCue spec seems to be okay with setters..
16:56
<Ms2ger>
On another note
16:56
<Ms2ger>
http://dev.w3.org/html5/webvtt/ is ugly :(
16:56
<Hixie>
reyre: iirc we intentionally made it possible to set negative-time cues, because otherwise it's hard to set the times on a cue (you have to figure out which of the two you have to set first based on the old values and new values)
16:56
<Hixie>
reyre: all the other algorithms in the spec should, in theory, handle them gracefully
16:57
<Hixie>
(endTime is on TextTrackCue, not WebVTTCue, btw)
16:57
<reyre>
Hixie: yeah, my bad about that. okay thanks for the info
16:57
<Hixie>
not an problem at all, please do feel free to ask :-)
16:57
<Hixie>
better safe than sorry :-)
16:58
<reyre>
Hixie: :) sounds good
16:58
<Hixie>
and it's not unusual for this kind of question to find errors in the spec :-)
16:59
<reyre>
Hixie: yeah i've encountered a couple of those already heh
17:30
<reyre_>
Hixie: so in the WEBVTT data model we have a writing direction, but in the API we have 'vertical' property
17:30
<reyre_>
how do those two relate?
17:30
<Hixie>
that's more a question for nessy, she's the editor now. but let me see if i can answer your question, one sec...
17:31
<Hixie>
reyre_: is the definition of 'vertical' not clear?
17:31
<Hixie>
reyre_: i'm confused as to the question
17:31
<Hixie>
what's not clear?
17:32
<Hixie>
abarth: a webkit bug in the parser is described here https://www.w3.org/Bugs/Public/show_bug.cgi?id=22183
17:34
<reyre_>
Hixie: i guess what's not clear is if writing direction and the vertical property are the same thing
17:34
<reyre_>
which i'm thinking they are
17:34
<Hixie>
how is that not clear?
17:34
<Hixie>
(are we looking at the same spec?)
17:34
<reyre_>
Hixie: http://dev.w3.org/html5/webvtt ?
17:34
<Hixie>
is http://dev.w3.org/html5/webvtt/#dfn-dom-texttrackcue-vertical what you're reading?
17:34
<reyre_>
yep
17:34
<Hixie>
i don't understand how that leaves any room for interpretation
17:35
<reyre_>
"On setting, the text track cue writing direction" that makes me think there is a property called 'writing direction' instead of the vertical property where it is actually stored
17:35
<reyre_>
"On setting, the text track cue writing direction must be set to the value given"
17:36
<Ms2ger>
Could you file a bug to make that an enum too?
17:36
<Hixie>
reyre_: ??
17:36
<reyre_>
Ms2ger: isn't the other bug to make the kind _not_ an enum? (if that's what your referring to)
17:37
<Hixie>
reyre_: what do you mean by "property"?
17:37
<reyre_>
Hixie: sorry i mean 'attribute in the webidl'
17:37
<reyre_>
so the vertical attribute
17:37
<Hixie>
reyre_: the attributes don't store anything
17:37
<reyre_>
:/
17:38
<Ms2ger>
reyre_, so there's one bug to make HTMLTrackElement.kind not an enum; the other things like that should be, though
17:38
<Hixie>
there's a "text track cue" thing, which is represented in JS by a "TextTrackCue" object
17:38
<Hixie>
the "TextTrackCue" object is just an API that exposes, in various ways, the values of the "text track cue" thing
17:38
<Ms2ger>
reyre_, and you should think about attributes as getter/setter pairs, that helps :)
17:38
<Hixie>
one of those values is "the text track cue writing direction"
17:39
<Hixie>
the "vertical" attribute on the TextTrackCue interface has a getter and setter which poke at values of the "text track cue" thing
17:39
<Hixie>
specifically, the "the text track cue writing direction" value, but that's an implementation detail
17:39
<Hixie>
as in, it could easily have been multiple values
17:40
<Hixie>
reyre_: does that make sense?
17:40
<reyre_>
Hixie: yep, i'm just trying to piece it together with how we do things in Gecko
17:41
<reyre_>
Ms2ger: i'll file a bug :)
17:41
<reyre_>
thanks Hixie
17:42
<Hixie>
np, sorry if i came over as rude :-/
17:42
<reyre_>
Hixie: not at all :)
17:42
<Hixie>
phew
17:53
<Hixie>
anyone know if you can do a bugzilla search that excludes bugs that have open dependencies?
17:54
<Hixie>
heycam|away: ping https://www.w3.org/Bugs/Public/show_bug.cgi?id=22218
17:55
<Hixie>
Ms2ger: https://www.w3.org/Bugs/Public/show_bug.cgi?id=22221 ?
17:55
<Ms2ger>
Yessir?
17:56
<Hixie>
why is it inconsistent?
17:56
<Hixie>
i thought you _liked_ enums
17:57
<Ms2ger>
Yes
17:58
<Ms2ger>
But not for reflected attributes, I realized
17:58
<Hixie>
oh is it that it throws an exception instead of ignoring?
17:59
<Ms2ger>
Ignores instead of transparently setting the content attribute
17:59
<Hixie>
oh ok
17:59
<Hixie>
right
18:04
<Hixie>
i wonder if there's a way to get the behaviour we want while still using enums
18:04
<Hixie>
i guess not
18:04
<Hixie>
oh well
18:04
<Hixie>
no biggie
18:04
<Ms2ger>
The point of using enums is that your prose doesn't see invalid values, no? :)
18:04
<Ms2ger>
The point of using enums is that your prose doesn't see invalid values, no? :)
18:05
<Hixie>
mostly the point is just to be clearer about intent
18:06
<Ms2ger>
I guess you could say that too
18:15
<GPHemsley>
Hixie: Good to know. I'm surprised they haven't done things automatically. Did you put in for the upgrade?
18:15
<Hixie>
yeah
18:16
<Hixie>
i expect they minimise upgrades to avoid breaking stuff
18:16
<Hixie>
it does explain why i've been stuck with such an ancient perl, though
18:16
<GPHemsley>
do we have a lot of scripts on the server?
18:17
<GPHemsley>
(of any kind)
18:19
<jgraham>
Debian 4? Isn't that the version that Jesus used?
18:19
<Hixie>
GPHemsley: yes
18:19
<Hixie>
the server hosts ~60 domains and subdomains
18:19
<GPHemsley>
oh
18:19
<GPHemsley>
this should be fun
18:20
<Hixie>
including e.g. software.hixie.ch
18:20
<Hixie>
which has all my games and tools and so on
18:20
<GPHemsley>
games?
18:20
<Hixie>
stuff i do in my spare time
18:20
<GPHemsley>
you have spare time?
18:20
<Hixie>
some
18:20
<Hixie>
none of the games are finished...
18:21
<GPHemsley>
ah
18:21
<jgraham>
Hixie: I hear there are plenty of specs looking for editors ;)
18:21
<GPHemsley>
heh
18:21
<Hixie>
jgraham: if i spend too much time editing, i burn out :-)
18:21
<Hixie>
jgraham: gotta keep a balance :-)
18:41
<Hixie>
man it's hard to work out how the url parser can ever fail
18:44
<Hixie>
in fact i think the answer might be it can't
18:44
<Hixie>
which is fascinating
18:46
<Hixie>
i wonder if that's intentional
18:46
<Hixie>
more questions for anne!
21:41
<Hixie>
http://test:test/ huh
21:44
<Hixie>
awesome, spec agrees with firefox
21:44
<Hixie>
bummer, sicking was useful and gave me more things to test
21:45
<Hixie>
(https://www.w3.org/Bugs/Public/show_bug.cgi?id=20580)
21:53
Hixie
writes a test case and finds every browser does the same thing
21:53
<Hixie>
clearly i wrote my test wrong
22:25
<Hixie>
so since adding the referer info to the bugs, i'm finding that many of the bogus bugs have no referer field at all. https:->http: maybe? (searches maybe?)
22:25
<Hixie>
e.g. https://www.w3.org/Bugs/Public/show_bug.cgi?id=22285
22:46
<TabAtkins>
If they have no referrer, it's likely that they... weren't referred. They were direct submissions against the submission url by probing spambots.
22:46
<TabAtkins>
If they have no referrer, it's likely that they... weren't referred. They were direct submissions against the submission url by probing spambots.
22:49
<Hixie>
i mean the referrer of the spec, not of the file-bug.cgi script
22:49
<Hixie>
the referrer of the latter seems to always be one of the specs
22:50
<Hixie>
(anecdotally, more often the w3.org/TR specs for the crazy bugs, more often the whatwg.org/html multipage spec for the annoyingly insightful and hard to fix bugs)
23:02
<reyre_>
Hixie: so the WebVTTCue webidl doesn't have [setter throws] on the alignment, vertical, and position attributes, but down lower it says that they are supposed to throw on setting
23:02
<reyre_>
is that correct?
23:03
<Hixie>
what is [setter throws] ?
23:04
<reyre_>
Hixie: i could be wrong about this, but i've seen before in other webidls that when an attribute could throw an error on setting it has a [SetterThrows] above the attribute ?
23:04
<Hixie>
ah
23:04
<Hixie>
that's new to me
23:04
<Hixie>
i don't think i've put that on any of the webidl i've ever written
23:05
<jsbell>
It's old
23:05
<jsbell>
A couple years (more?) ago the need to express throwing behavior in WebIDL was removed.
23:05
<jsbell>
All done in prose now.
23:06
<reyre_>
hmm okay thanks jsbell
23:06
<reyre_>
thanks Hixie
23:22
<rillian>
reyre_: [SetterThrows] survives in the mozilla webidl compiler as an extension to control code generation
23:22
<rillian>
so it's needed in the implementation's webidl, but not the spec
23:22
<rillian>
https://developer.mozilla.org/en-US/docs/Mozilla/WebIDL_bindings#Throws
23:23
<rillian>
i.e. we omit the error object reference in the binding call when that decorator is not present
23:23
<Hixie>
seems like it'd be safer to have [NeverThrows], so the code gets generated in the case where someone didn't think to check
23:24
<rillian>
I think they'd notice when they implemented the exception, because there'd be no way to throw it.
23:25
<jsbell>
Blink and Blink have something similar.
23:25
<rillian>
and concensus was (we wanted that) most things didn't throw
23:25
GPHemsley
wonders if HTML shouldn't have its own custom icon like the other specs, to differentiate from WHATWG itself.
23:25
<jsbell>
Er, Blink and WebKIt
23:26
<reyre_>
rillian: awesome
23:26
<reyre_>
i think marcus is good to go for that then
23:27
<rillian>
great
23:27
<msaad>
yeap
23:30
<GPHemsley>
Hixie: @WHATWG seems not to have tweeted any changes after r7953
23:31
<Hixie>
that's anne's department
23:32
<GPHemsley>
naturally