00:48
<dglazkov>
good morning Whatwg!
01:13
<boogyman>
hi
10:10
<annevk>
Given WebRAW http://blog.bitops.com/blog/2013/06/04/webraw-asmjs/ and WebP, maybe allowing JavaScript to be an image decoder is not too bad...
10:11
<annevk>
Would have to sandbox it somehow so it also works for cross-origin images. And needs a story for sniffing.
10:14
<annevk>
Hixie_: Giving https://www.w3.org/Bugs/Public/show_bug.cgi?id=22185 priority would be great
10:17
<jgraham>
annevk: Well you could scope it to same-origin only, perhaps
10:17
<annevk>
jgraham: that's not acceptable
10:17
<jgraham>
Why not?
10:17
<annevk>
jgraham: because that doesn't match what the platform can do
10:17
<jgraham>
It seems strivtly better than it not working at all
10:18
<annevk>
jgraham: I think a worker without networking APIs or postMessage would be workable
10:18
<jgraham>
And matches what people retrospectively wish the platform had done
10:18
<annevk>
jgraham: the only thing it can do is decode a stream
10:19
<jgraham>
Making up a new kind of global scope for each new feature doesn't sound like fun
10:19
<MikeSmith>
fyi for anybody using w3c-test.org: I just spent an hour on there trying to unwedge apache problems due to debian upgrade bullshit and I have it running again but I very likely broke some other behavior of the web server in the process. So if you notice any problems please let me know.
10:20
<annevk>
jgraham: parallelize all the things
10:21
<annevk>
might be a problem even with that, you could encode data in output width/height
10:59
<annevk>
matjas: contacted DH
10:59
<annevk>
matjas: cc'd you
11:14
<annevk>
Ms2ger: yo
11:14
Ms2ger
appears
11:15
<annevk>
Is "The term context object means the object on which the algorithm (typically a method or attribute) being discussed was called. When the context object is unambiguous, the term can be omitted." clear enough you think?
11:15
<annevk>
Ms2ger: http://lists.w3.org/Archives/Public/www-dom/2013AprJun/0154.html
11:16
<Ms2ger>
Hm
11:16
<Ms2ger>
A method/attribute isn't really an algorithm, though
11:18
<annevk>
so just or the things?
11:18
<Ms2ger>
"typically defining the behavior of a method or attribute" maybe?
11:18
Ms2ger
is not good with words
11:18
<annevk>
"The term <dfn>context object</dfn> means the object on which the algorithm, method, or attribute being discussed was called."
11:18
<hallvors>
context object === this object ?
11:18
<Ms2ger>
Yeah
11:19
<hallvors>
text is still a bit clumsy, in that you can not call an attribute
11:19
<jgraham>
"accessed"
11:19
<annevk>
thanks jgraham
11:20
<Ms2ger>
But can you access an algorithm?
11:20
<hallvors>
but if you mean the "this object" perhaps you could just reference a suitable definition in ES?
11:20
<hallvors>
though they might not do one for properties..
11:20
<Ms2ger>
Maybe ... on which an algorithm, method, or attribute getter or setter ...
11:20
<annevk>
Ms2ger: yeah
11:21
<Ms2ger>
hallvors, well, it isn't a ES object :)
11:21
<annevk>
JSIDL will make most of this obsolete
11:21
hallvors
lacks context for the context object discussion
11:21
<annevk>
hallvors: see the link earlier
11:21
hallvors
is recontextualising
11:22
<jgraham>
JSIDL: Pretending it's turtles^Wjavascript all the way down since 2013
11:24
<annevk>
jgraham: not very different from pretending there's no hardware limits
11:25
<jgraham>
Well sure, it might be a good pretense, I don't know
11:26
<jgraham>
Personally I am waiting for the time when we can design APIs where DOM objects hold references to JS objects
11:27
<annevk>
jgraham: like promises?
11:27
<Ms2ger>
Is Safari going to implement?
11:27
<annevk>
Ms2ger: implement what?
11:27
<Ms2ger>
Promises
11:28
<annevk>
Dunno, I'd assume so
11:28
<jgraham>
Doesn't Blink still have the same issue?
11:29
<jgraham>
So "is anyone other than Gecko going to implement DOM Promises", perhaps?
11:32
<annevk>
Blink is implementing for sure
11:34
<jgraham>
Interesting
11:35
<annevk>
Microsoft will too
11:35
<jgraham>
Last time I heard from abarth they had no plans to change thier memory management to deal with DOM->JS->DOM cycles
11:35
<jgraham>
This was pre-blink of course
11:36
<annevk>
abarth basically recommends implementing most directly in JS
11:36
<annevk>
which makes sense, as it'll end up there anyway
11:37
<jgraham>
I guess that might solve the problem for them
11:55
<hallvors>
annevk: "The plan is for debugging tools to solve this." - don't forget that it's common for web apps to log errors happening client-side and send the details home
13:23
<annevk>
hallvors: if at the end of your chain of futures you use .chain(e => log(e)) you'll be fine
13:24
<hallvors>
OK
13:25
<annevk>
sorry, .catch(...)
13:40
<annevk5>
marcosc: http://sysapps.github.io/sysapps/proposals/Contacts/Contacts.html is broken
13:47
<marcosc>
annevk: thanks, will fix
13:48
Ms2ger
liked the "[editing] Editing spec is no longer online" thread
13:48
<annevk>
Timestamps indicate darobin was quickest, but slowest to appear in my inbox :/
13:49
<Ms2ger>
He was first for me
13:49
<Ms2ger>
Then you, then Art
13:49
<darobin>
there may be more replies to come
13:49
Ms2ger
replies
13:49
<annevk>
Given the lag, yes
13:49
<darobin>
Ms2ger hasn't responded yet, for instance
13:50
<annevk>
I should have learned by now not to reply to those emails
13:50
<annevk>
But every now and then...
13:50
<darobin>
yeah so should I
13:50
<darobin>
but then no one replies and it's not nice either
13:55
<hallvors>
guys, just relax and enjoy random syncronicity
13:56
<annevk>
hallvors: I think that's the problem, there'd be less 386 if I was relaxed
14:16
<manu-db>
darobin: HTML5 DOM question for you about the language associated with nodes. Does the document language expressed via the HTTP Header "Content-Language" get used by the DOM in any way? We're trying to figure out if the language specified in an HTTP Header is transferred in any way to a DOM environment.
14:17
<manu-db>
Hixie_: same question ^^
14:17
<manu-db>
darobin, Hixie_: From a read of the spec, it doesn't look like it is... wondering if it's somewhere else in the spec.
14:20
<Ms2ger>
<p>If neither the node nor any of the node's ancestors, including the <span>root element</span>,
14:20
<Ms2ger>
have either attribute set, but there is a <span>pragma-set default language</span> set, then that
14:20
<Ms2ger>
is the language of the node. If there is no <span>pragma-set default language</span> set, then
14:20
<Ms2ger>
language information from a higher-level protocol (such as HTTP), if any, must be used as the
14:20
<Ms2ger>
final fallback language instead. In the absence of any such language information, and in cases
14:20
<Ms2ger>
where the higher-level protocol reports multiple languages, the language of the node is unknown,
14:20
<Ms2ger>
and the corresponding language tag is the empty string.</p>
14:20
<manu-db>
Yes, already read the spec...
14:20
<manu-db>
Let me rephrase the question.
14:20
<GPHemsley>
annevk: See also: https://github.com/GPHemsley/tiff-js , https://github.com/jnordberg/gif.js
14:21
<darobin>
manu-db: my understanding is that it does; at the very least setting Content-Language through <meta> certainly works
14:21
<manu-db>
Is there a way for an HTML+JavaScript app to get the "Content-Language" http header.
14:21
<darobin>
ah, that's a completely different question
14:21
<annevk>
GPHemsley: cool, yeah we should do this, but we need to figure out a way to deal with the width/height thing
14:21
<manu-db>
basically, I don't think that what's in the HTML spec is implementable in a JavaScript environment, is that true?
14:21
<Ms2ger>
No
14:21
<darobin>
presumably you mean the one that was used for the current page?
14:21
<manu-db>
and if so, what should we say in our specs when determining the language.
14:22
<Ms2ger>
But you have to implement HTTP as well
14:22
<manu-db>
darobin: yes
14:22
<manu-db>
Ms2ger: what do you mean, "you have to implement HTTP as well"?
14:23
<annevk>
Man, scumbag language. Almost always wrong, yet keeps wasting a bunch of time with regards to how to set it declaratively and how its inheritance chain works.
14:23
<Ms2ger>
HTML can be implemented in any language you want, but you can't necessarily pick just one section and implement it based on the rest of the platform
14:23
<manu-db>
darobin, Ms2ger: Here's the actual issue - Microdata and RDFa processors written in JavaScript need to be able to get the language of the current node... if the language is set via "Content-Language", there is no way for them to get at that information... so, should the spec just say that all Microdata and RDFa implementations shouldn't pay attention to Content-Language?
14:24
<annevk>
manu-db: I think the theoretical answer is that such implementations would be non-conforming until that primitive is exposed.
14:24
<Ms2ger>
I don't see why
14:25
<darobin>
no no, you don't want to say that implementation should ignore Content-Language just because you don't happen to have that information right now
14:25
<manu-db>
annevk: the only workaround we've found is for the processor to do an XHR request for the exact same document and get the Content-Language like that, but - super kludge (and you don't know if you're making the same request the browser did)
14:25
<darobin>
that's one way
14:25
<manu-db>
annevk: Is there a plan to get at the Content-Language primitive?
14:26
<darobin>
manu-db: well you seem to have a use case :
14:26
<darobin>
:)
14:26
<manu-db>
darobin: This discussion is going in a direction I didn't want it to go in. :)
14:26
<annevk>
manu-db: there's an open bug for it
14:27
<annevk>
manu-db: see DOM Standard's open bugs
14:27
<darobin>
manu-db: can't you use XPath for that?
14:27
<manu-db>
annevk: ok, thanks... reading now.
14:27
<manu-db>
darobin: I'm not hip with the XPath - how would you do that?
14:27
<annevk>
manu-db: it comes up every now and then, my main problem with doing it is that language seems so theoretical
14:28
GPHemsley
is still wondering what to do with http://wiki.whatwg.org/wiki/AAFC_Metadata_Application_Profile
14:28
<manu-db>
annevk: Yeah, I have the same issue with it - don't know of a single case where this would be useful... but we have a LC comment that we need to answer in a fairly logical way.
14:28
<annevk>
(other than as a rendering hint to work around Han unification, but we have heuristics for that)
14:28
<darobin>
ah, no, you can test for a language in XPath but not get one
14:28
<darobin>
that's a shame
14:28
<annevk>
(non-standardized heuristics, someone has to fix that)
14:29
<Ms2ger>
darobin, don't need xpath for that either
14:29
<annevk>
GPHemsley: seems fine
14:29
<annevk>
GPHemsley: it's not spam
14:29
<darobin>
isn't language also used to pick the proper glyph in parts of CJK?
14:30
<annevk>
darobin: see my () comments above
14:30
<darobin>
Ms2ger: yeah, CSS does that too. I thought XPath's lang() did more
14:31
<darobin>
annevk: ah, duh, hadn't seen that line about Han unification
14:31
<GPHemsley>
annevk: But does it belong on the wiki? And is it public domain?
14:32
<darobin>
I guess language information could also be used for automatic translation and for spellchecking of forms/editable
14:37
<manu-db>
annevk, darobin, Ms2ger: Ok, so just to be clear - I'm going to respond with something to this effect: "At present, it is impossible for a JavaScript implementation to deterministically retrieve an HTML5 element's language if no language information is provided in the document, but the language was set via the Content-Language HTTP Header. These Javascript-based implementations will continue...
14:37
<manu-db>
...to be non-conforming in this specific case until the DOM exposes the language of an element [DOM bug #16489]. For this reason, document authors should specify the language of the document in the document itself to ensure that all HTML/RDFa implementations are capable of retrieving the language of the element." Is that response aligned with each of your thoughts on the matter?
14:38
<annevk>
manu-db: yeah
14:38
<manu-db>
annevk: ok, thanks - much appreciated. :)
14:40
<darobin>
manu-db: yeah that makes sense
14:52
<manu-db>
darobin, annevk: Outlined the use case for DOM element language property: https://www.w3.org/Bugs/Public/show_bug.cgi?id=16489#c1
14:53
<darobin>
Using HTTP for metadata sucks. Film at 11.
15:12
<manu-db>
Anyone here working at Google w/ the Gmail+schemas folks? You may want to tell them that their HTML+Microdata markup is so broken that it will be mis-interpreted horribly when processed by a conforming Microdata implementation: https://developers.google.com/gmail/schemas/embedding-schemas-in-emails -- or is putting a <span> element with embedded <time> elements a "thing" in HTML5 now?
15:13
<manu-db>
(that is, putting <time> in <span> in the <head>)
15:13
<manu-db>
(to see the problem, scroll down halfway on the page, click the "Microdata" example tab)
15:18
Ms2ger
wonders about http://twitpic.com/cvmib9
15:18
<manu-db>
darobin: actually, just to sanity check - there are no plans on supporting arbitrary flow/phrasing content in HEAD are there?
15:21
<darobin>
manu-db: you mean like <head><p>blah</head>?
15:21
<manu-db>
darobin: I mean, specifically: <head><span><time .../></span></head>
15:22
<manu-db>
(like what they're doing in Gmail now)
15:22
<darobin>
mmmm, not that I know of, no
15:22
<darobin>
that would be... problematic, unless I'm missing something
15:22
<manu-db>
darobin: I'll tell them to stop doing that, then.
15:22
<manu-db>
the current algorithm will move that <span> block into body, right?
15:23
<darobin>
yeah
15:23
<darobin>
unless there's something I'm not thinking of, <head> is really restricted
15:39
<tantek>
manu-db - you care about emails that are "Best Views In Gmail" ?
15:39
<tantek>
"Best Viewed In Gmail"
15:39
<tantek>
clearly we need buttons for that to embed in email signatures
15:42
<manu-db>
tantek: I don't quite understand your question. I care that Google is asking developers to create invalid markup. There's nothing that specifically limits that usage to Gmail. It could work with any e-mail client that supports HTML content (which is most of them).
15:43
<manu-db>
tantek: They also claimed that their partners didn't like Microdata, which is why they adopted JSON-LD as the primary way of marking up that data.
15:44
<tantek>
Google is suggesting invalid markup in email? Tell me again about how often HTML email validates.
15:46
<manu-db>
tantek: Well in this particular case, the markup could be moved around by the HTML processing algorithm in such a way that the Microdata expresses something completely broken (like if they did <title itemtype="..."> ... <span itemprop="reservation" itemtype="..."></span></title>
15:47
<manu-db>
(meaning, what the developer thinks the markup means is going to be interpreted by a Microdata processor in the wrong way)
15:48
<manu-db>
tantek: Guidance to developers should really be - "just put it in the body"
15:48
<manu-db>
tantek: but they probably didnt' want to do that because they wanted the JSON-LD and Microdata to look the same.
17:10
GPHemsley
wonders why Ms2ger thinks the Mozilla HTML5 bug doesn't serve a purpose
17:13
<Ms2ger>
Do you have any?
17:13
<GPHemsley>
that's a personal question
17:13
<GPHemsley>
;)
17:14
<GPHemsley>
but yeah, isn't it used as a tracking bug for all the other bugs? it's a good way (in theory) to keep track of all the HTML spec-related bugs in one place
17:15
<GPHemsley>
particularly because it's one of the blockers for the living standards bug
17:15
<GPHemsley>
s/particularly because it's/it's also/
17:15
<Ms2ger>
There's way too many such bugs to have them all tracked that way
17:17
<GPHemsley>
well, this tends to be a meta bug to track meta bugs, IIUC
18:05
<GPHemsley>
(annevk:) (In reference to bz's WebAppSec post which I didn't get CC'd on) <object> is a hybrid element that could be any one of image, audio/video, or plugin contexts, and the spec says that.
18:06
<GPHemsley>
I don't think it's a nested browsing context, but I could be wrong
18:06
<GPHemsley>
(I'm not on the WebAppSec mailing list)
18:08
<GPHemsley>
oh, apparently <object> is actually an image, a nested browsing, or a plugin context
18:08
<GPHemsley>
spec doesn't mention audio/video
18:09
<GPHemsley>
but it seems the <object> definition could use some updating to tie in better with mimesniff (though admittedly mimesniff is not in the best state to be tied in with yet)
18:10
GPHemsley
needs to get things out of his head and into the spec
20:04
<Hixie_>
anyone understand https://www.w3.org/Bugs/Public/show_bug.cgi?id=21870 ?
20:04
<Ms2ger>
Sure
20:05
<Ms2ger>
Look at the "Content attributes:" part of the green box
20:05
<Hixie_>
yeah?
20:05
<Hixie_>
what about it?
20:05
<Ms2ger>
Last line only has the name, no description
20:06
<Hixie_>
under <menu> ?
20:07
<Ms2ger>
No, button
20:07
<Hixie_>
oohhhhhhh
20:07
<Hixie_>
thanks
20:07
<Ms2ger>
Np
20:07
<Ms2ger>
You should really start looking at the links ;)
20:08
<Hixie_>
i do...... i swear
20:08
<Hixie_>
sometimes
20:12
<jgraham>
Just upgraded critic on critic.hoppipolla.co.uk
20:12
<jgraham>
It is possible that breakage has occured; please report any issues
20:30
<Hixie_>
mounir: ping https://www.w3.org/Bugs/Public/show_bug.cgi?id=11937#c14
21:31
<GPHemsley>
Hixie_: For future reference, what's the quickest way to get your attention on something? (Choices include IRC, e-mail, bug, skywriting, etc.)
21:32
<Hixie_>
IRC if i'm around
21:32
<Hixie_>
e-mail direct to ian⊙hc (not cc'ed to a mailing list) if i'm not
21:33
<Hixie_>
if you need to catch me when i'm not working, g+ is a good way to get me too
21:33
<GPHemsley>
heh... when are you not working?
21:33
<Hixie_>
like last month :-)
21:33
<GPHemsley>
ah, ok
21:34
<GPHemsley>
and I can judge whether you're around on IRC by whether you're "away"?
21:34
<Hixie_>
no
21:34
<Hixie_>
i'm always /away :-(
21:34
<GPHemsley>
oh
21:34
<Hixie_>
i use it to track what people are saying in other channels
21:34
<GPHemsley>
how does that work?
21:35
<Hixie_>
i have a bunch of things that cause irssi to add stuff to the away log
21:35
<Hixie_>
like if anyone says "hixie"
21:35
<Hixie_>
and then every now and then i look at hte away log
21:35
<Hixie_>
but i can't do that if i'm not /away as far as i can tell
21:36
<GPHemsley>
I see... that seems rather hackish
21:36
GPHemsley
uses X-Chat
21:36
<Hixie_>
very hackish
21:36
<GPHemsley>
but anyway, since you're here now
21:36
<Hixie_>
anyway, you can tell if i'm on irc by pinging me and seeing if i respond
21:36
<GPHemsley>
what were your reservations about defining "resource" again?
21:37
<Hixie_>
dunno how to do it well
21:37
<GPHemsley>
what were the consequences of doing it poorly, again?
21:40
<Hixie_>
depends on how it's used
21:41
<Hixie_>
if you say "a resource is an apple" and somewhere else you say "browsers much support downloading resources" then you've just required that browsers support downloading physical objects
21:46
<GPHemsley>
Hixie_: and if I stick to the digital realm with something like "a resource is a container of data, often represented as a byte sequence, that may have one or more items of metadata associated with it"?
21:47
<Hixie_>
so a bucket with dictionaries in it is a resource? :-)
21:49
<GPHemsley>
"...a digital container of data..."?
21:50
<Hixie_>
so when i hold an apple in my hand? :-P
21:50
GPHemsley
slaps Hixie_ with a circa 2003 trout
21:50
<Hixie_>
:-P
21:51
<GPHemsley>
I will force user agents to download your hand if you don't watch out
21:52
<GPHemsley>
in other news, do you have any opinion on codifying these other contexts? http://wiki.whatwg.org/wiki/Contexts
21:53
<GPHemsley>
I want to make sniffing talk only about contexts
21:53
<GPHemsley>
and have each context have its own sniffing algorithm
21:53
<GPHemsley>
and then not have any "generic" or "general" or "default" sniffing algorithms
21:56
<GPHemsley>
Hixie_: ^^
21:56
<Hixie_>
i'm reading, i'm reading :-P
21:56
<Hixie_>
patience boy :-P
21:56
<GPHemsley>
ah, I always worry about you disappearing :P
21:57
<Hixie_>
a fair concern
21:57
<Hixie_>
i think the Contexts thing is fine assuming it doesn't affect the interface to other specs
21:58
<GPHemsley>
I think it may help them too
21:58
<GPHemsley>
but my first question is really, do you want to put them in HTML, or should I put them in mimesniff?
21:59
<GPHemsley>
annevk et al. might be able to tell you more about how it relates to CSP and Fetch and other things
22:00
<GPHemsley>
see e.g. http://lists.w3.org/Archives/Public/public-webappsec/2013Jun/0027.html
22:02
<GPHemsley>
basically, the idea is that you can't fetch something from an HTML page without having a context to fetch it through
22:02
<GPHemsley>
(I'm picturing "context" as a tunnel or a tube or a lens)
22:03
<GPHemsley>
same with CSS
22:09
<Hixie_>
GPHemsley: mimesniffm i think
22:09
<Hixie_>
or fetch
22:12
<GPHemsley>
yeah, maybe fetch
22:12
<GPHemsley>
but I think HTML should at least mention them
22:13
<GPHemsley>
because they tend to be associated with particular elements
22:13
<GPHemsley>
I plan to make a list of them, if no one beats me to it
22:13
<Hixie_>
i'd rather not do anything that is purely editorial in the near future
22:13
<Hixie_>
trying to deal with feedback of a more serious nature first
22:13
<Hixie_>
since i'm getting behind on it
22:14
<Hixie_>
"the perfect is the enemy of the good", and all
22:26
<GPHemsley>
oh, I wasn't viewing it as purely editorial, but ok
22:28
<Hixie_>
oh
22:28
<Hixie_>
what's the normative effect?
22:36
<GPHemsley>
it's a hook, essentially
22:36
<GPHemsley>
for other specs to latch on to
22:36
<GPHemsley>
and to make clear how to treat fetches
22:36
<GPHemsley>
(you can only fetch within a context)
22:38
<Hixie_>
i don't really see what this gains us over what we have now
22:52
<GPHemsley>
clarity and explicitness?
22:53
<Hixie_>
right, that's editorial :-)
22:53
<Hixie_>
i'm not denying that editorial improvements have value
22:53
<Hixie_>
just that right now, for me, they're not a priority
22:53
<GPHemsley>
alright
22:53
<Hixie_>
because they're always lower priority than actual normative issues, and right now i'm unable to keep up with normative issues
22:53
<GPHemsley>
but I'll proceed as if somebody plans to add them somewhere at some point
22:53
<Hixie_>
with reasonable lag
22:54
<Hixie_>
if you need more important stuff to work on, we have plenty of stuff that will have more impact :-)
22:54
<GPHemsley>
I think this is important within the realm of mimesniff
22:54
<GPHemsley>
though it's debatable where mimesniff itself falls on the importance scale
22:55
<GPHemsley>
I think the organization of mimesniff is dreadful right now
22:56
<GPHemsley>
so adding this context stuff will make it less dreadful
22:56
<GPHemsley>
but if you have other ideas, I'm all ears
22:57
<GPHemsley>
(like I said, I plan to have a separate sniffing algorithm for each context in that list)
22:58
<GPHemsley>
(the style context needs one the most, I think, of those that don't already have one)
22:59
<GPHemsley>
though I do agree that not many people seem to care about mimesniff
23:01
<Hixie_>
GPHemsley: i think mimesniff is closer to done than you do :-)
23:01
<GPHemsley>
perhaps
23:01
<Hixie_>
GPHemsley: put it this way, if you had to edit mimesniff and the entire html spec, how much would you care now? :-)
23:01
<GPHemsley>
oh, yeah, no, I get it :)
23:02
<GPHemsley>
but if you have other things for me to do, let me know
23:02
<GPHemsley>
I'm pondering the GIF thing
23:02
<GPHemsley>
but that's probably mostly done, too
23:03
<GPHemsley>
(I have to admit, though, that I feel like I've only just gotten a handle on mimesniff in the last few weeks)
23:03
<GPHemsley>
(mentally, that is)
23:04
<Hixie_>
there's a big list of things that need doing, see the wiki :-)
23:04
<Hixie_>
i can help out more with that tomorrow morning if you want
23:04
<Hixie_>
right now i gotta go soon
23:04
<GPHemsley>
alright
23:04
<GPHemsley>
though I think the specs todo page could use an audit itself
23:05
<GPHemsley>
(and I'm not sure how many of the things on that list I'm actually capable of tackling)