05:36
<MikeSmith>
Domenic_: what do gulp and browserfy have to do with streams?
08:36
<zcorpan>
hsivonen: iirc https://code.google.com/p/chromium/issues/detail?id=329531 applies to gecko, too
08:40
<hsivonen>
zcorpan: thanks
08:48
<MikeSmith>
zcorpan: "I propose we don't address this" from sof.. why? because it will break things?
08:49
<zcorpan>
MikeSmith: s/will/is not proven not to/
08:49
<MikeSmith>
ok
08:49
<Ms2ger>
From ap?
08:50
<MikeSmith>
sigbjorn from oepra
09:20
<annevk>
SimonSapin: I thought :scope was still a thing
09:20
<annevk>
SimonSapin: I thought it moved to Selectors or some such, TabAtkins?
09:25
<Ms2ger>
http://dev.w3.org/csswg/selectors/#scope-pseudo
09:34
<annevk>
\o/ Chrome is going to remove obsolete DOM stuff. Sadly they still refer to it as DOM4
09:39
<MikeSmith>
annevk: I think that's just because Dominic didn't get the memo that it should be called something else
09:39
<annevk>
MikeSmith: he does refer to the WHATWG copy so there's that, indeed
09:40
<annevk>
I had to commute today and be in at 9AM. That shit is hard
09:44
<MikeSmith>
annevk: hard to find a parking space big enough for your cadillac
09:44
<annevk>
Can of sardines is nothing compared to the Circle line
09:46
<annevk>
Central line, that is, still getting used to the terminology
09:46
<MikeSmith>
could be worse: You could be living in Mountain View
09:48
<Ms2ger>
Or Tokyo? :)
09:48
<MikeSmith>
heh
10:10
jgraham
wonders where annevk is
10:10
<annevk>
jgraham: http://theodi.org/
10:11
<jgraham>
Why?
10:15
<hsivonen>
annevk: whoa. the Thunderbird compose encoding menu is worse than I expected
10:16
<hsivonen>
https://bugzilla.mozilla.org/show_bug.cgi?id=943252#c9
10:18
<annevk>
jgraham: the W3C TAG is meeting there
10:19
<jgraham>
annevk: Oh, sucks to be you
10:20
<annevk>
hsivonen: ugh
10:27
<annevk>
hsivonen: sorry I haven't been responsive lately, it's not going to get much better until mid-March or so
10:27
<annevk>
hsivonen: I plan to focus on Encoding stuff next week
10:33
<hsivonen>
annevk: OK.
10:33
<hsivonen>
FWIW, I'm inclined to think that Thunderbird shouldn't bother the user with composition encoding UI at all
10:33
<hsivonen>
but my first goal is getting cruft out of m-c. not redesigning TB
10:40
<annevk>
I think email inevitably ends up being a web thing, so I'm not too worried about Thunderbird
10:45
<hsivonen>
annevk: the reason why I end up writing essays for the TB bugs is that the mailnews dependencies make it hard to remove code that's dead code in Firefox
10:46
<annevk>
Understood, the dependencies are sad, especially so long after the Firefox fork
10:46
<annevk>
The RDF-in-Gecko apologists are also making me somewhat sad
10:49
<hsivonen>
annevk: that thread took an unexpected turn
10:49
<krijn>
MikeSmith: thanks voor the retweets! :)
10:57
<annevk>
hsivonen: the last guy also frequently comments on www-tag, not unexpected
10:58
<annevk>
hsivonen: felt similar to how suddenly a bunch of XSLT apologists found out blink-dev exists
11:04
<hsivonen>
and this RDF stuff doesn't even have anything to do with the Semantic Web while XSLT has something to do with using XSLT
11:07
<annevk>
I guess there's a point to be made that the dev.platform discussion is worse :-)
11:07
<annevk>
http://w3ctag.github.io/capability-urls/2014-01-03.html is pretty awesome by the way
11:11
<MikeSmith>
krijn: yup
11:24
<krijn>
MikeSmith: :|
11:25
<MikeSmith>
nice to use twitter for something useful
11:31
<MikeSmith>
"The nightmare scenario would be that the web platform ceases to be a platform at all, and simply becomes an amalgamation of JS bindings to various operating system APIs of varying capability"
11:31
<darobin>
hahaha
11:31
<darobin>
MikeSmith: where that from?
11:32
<MikeSmith>
darobin: https://groups.google.com/a/chromium.org/d/msg/blink-dev/4jBAnIVwrt0/v2qzjSNn_E8J
11:32
<darobin>
oh, I haven't read that thread yet — it seemed to hold promise :)
11:34
<MikeSmith>
darobin: maybe but I much prefer the "should we use auto?" thread on webkit-dev
11:35
<darobin>
haha, I started muting that one
11:37
<darobin>
MikeSmith: Jon Rimmer does have a very good point though
11:38
<darobin>
I mean, we've been talking about providing some sort of low level access to hardware, of the USB API kind, instead of creating dedicated APIs for all sorts of things
11:38
<darobin>
which makes a lot of sense in many ways — but what happens if the generic hardware API starts to become meaningless?
11:40
<annevk>
MikeSmith: has there been any recent interesting webkit-dev thread?
11:40
<annevk>
MikeSmith: given the way Apple works I expect most of it to be on secret-webkit-dev
11:45
<darobin>
annevk: it does feel that way reading the list
11:46
<MikeSmith>
darobin: bingo (about if generic hardware API starts to become meaningless)
11:46
<MikeSmith>
annevk: there have been some interesting ones yeah
11:46
MikeSmith
looks back
11:47
<annevk>
darobin: you need something more generic than USB or Bluetooth
11:47
<annevk>
darobin: it needs to be on the level of "implement a driver for a connected device"
11:48
<darobin>
annevk: yeah, but that still likely relies on OS-level assumptions. Also, I am always suspicious of that level of abstraction
11:48
<annevk>
darobin: which I guess comes down to having something similar to USB or Bluetooth for the web, and if your device wants to be compatible it needs to support that
11:48
<annevk>
darobin: you need to start from the assumption that the web is the OS
11:48
<darobin>
annevk: or there is a browser mapping of that to USB
11:49
<darobin>
annevk: yeah, I started from that assumption a long time ago :)
11:49
<MikeSmith>
annevk: anyway the src-N thread on webkit-dev was gold. looking forward to more like that
11:49
<darobin>
but still, it needs to be resilient to architectural changes
11:49
<annevk>
MikeSmith: fair
11:49
<annevk>
darobin: might just be called HTTP in the end
11:49
<MikeSmith>
QUIC
11:49
<darobin>
annevk: that's all very handwavy :)
11:50
<darobin>
hardware is, well, hard
11:50
<MikeSmith>
oops I mean SPDY
11:50
<annevk>
darobin: once you go down the route of two pieces that need to be connected over a network, you end up there pretty quickly
11:51
<annevk>
darobin: at first you think you need to have something low-level because nothing is capable, like WAP to HTML, and at some point you realize you were wrong
11:51
<darobin>
annevk: I know, and in fact there's very interesting work in that area. Though at this point they've made rather brutal optimisations
11:52
<darobin>
annevk: I will be amused when this all leads to EXI being supported in the browser (it's a common component in ZigBee's HTTP hardware stuff)
11:52
<darobin>
(it's one of the "brutal optimisations")
11:54
<MikeSmith>
darobin: somebody told me HTTP2 people were looking at EXI for imspiration for header compression
11:55
<darobin>
MikeSmith: I'm not sure if I'd want them to go that way, but it could in fact yield interesting results
11:55
<MikeSmith>
and today I see https://github.com/twitter/hpack but haven't looked at it the code
11:55
<darobin>
I suspect the complexity might be excessive though
11:55
<MikeSmith>
darobin: I dunno, Hiro told me that. Nakajima. and he's following it alll pretty closely
11:56
<darobin>
one of the advantages of using EXI is that people can compete on encoder quality while remaining interoperably decodable
11:56
<MikeSmith>
yeah that would be a good advantage
11:56
<darobin>
MikeSmith: I had mentioned it to the SPDY folks a long time ago when they had started working on this but they were dismissive
11:56
<zcorpan>
at least the "punch in the face" thread was a bit funny
11:56
<MikeSmith>
a lot has changed since then I guess
11:57
<darobin>
well, I guess that if someone involved in it has understood that the "X" in "EXI" is accidental, then advocacy can make headway from there
11:58
<darobin>
if EXI ever makes it into HTTP2 I think I'll retire
11:58
<MikeSmith>
zcorpan: "punch in the face" thread?
11:58
<MikeSmith>
https://twitter.com/pmarca/status/419257504913448960 btw
11:58
<darobin>
that's enough evil for a single lifetime
11:59
<zcorpan>
MikeSmith: http://www.w3.org/mid/01EDE259-9B0C-47BE-BA09-536877558EDD⊙gc
12:00
<MikeSmith>
zcorpan: ah yeah that guy was telling it like it is, really
12:02
<darobin>
yeah that was pretty funny :)
12:02
<darobin>
jgraham: I'm done reviewing https://critic.hoppipolla.co.uk/r/526, I'll leave the python bits to someone else
12:03
<jgraham>
darobin: Given the number of comments that's probably just as well :p
12:03
<darobin>
heh
12:03
<jgraham>
I might have to quibble over some style issues, but generally great work, thanks
12:03
<darobin>
well I did read the Python code, but I don't have much to say beyond the fact that I find it weird you find it easier to include a complete HTML serialiser rather than just use a template :)
12:03
<darobin>
jgraham: you're more than welcome :)
12:04
<darobin>
I can see where the style issues are coming from; I mean, it's JS written more or less as if it were Python
12:58
<Ms2ger>
jgraham, I agree with darobin for once (on ||) :)
13:00
<jgraham>
Sigh
13:00
<jgraham>
Well if everyone disagrees with me I will change it, but I don't think it's at all clearer
13:00
<MikeSmith>
uh oh things are heating up on that review. I feel a "punch in the face" message possibilty
13:01
<jgraham>
So I don't understand *why* people like it. Because it's slightly shorter? Because it feels clever?
13:02
<MikeSmith>
idiomatic
13:02
<jgraham>
("slightly shorter" is honestly the best legitimate reason I can come up with, so if there is a better one I would like to know)
13:03
<Ms2ger>
Doesn't evaluate the x in x ? x : y twice
13:03
<jgraham>
MikeSmith: But I don't understand why it's idiomatic
13:03
<Ms2ger>
And the comparison with python doesn't make sense, it isn't "condition and value_if_true or value_if_false" but "value_if_true and value_if_true or value_if_false"
13:04
<jgraham>
Ms2ger: That isn't an actual problem in any case here and could be solved with the use of a variable if it was
13:04
<MikeSmith>
jgraham: I don't either in this case, really.
13:04
<jgraham>
Ms2ger: In python people have written value_if_true or value_if_false
13:04
<jgraham>
It works in exactly the same way
13:04
<Ms2ger>
Well yes
13:04
<Ms2ger>
Let's do that, then :)
13:04
<jgraham>
But these days I would expect anyone to use the ternary operator in Python
13:05
<jgraham>
(in either case)
13:05
<jgraham>
Certianly I would always write value_if_true if value_if_true else value_if_false
13:05
<Ms2ger>
I know you would :)
13:06
<Ms2ger>
I'm not convinced about anyone, though
13:13
<darobin>
whoa, I hadn't realised Ms2ger agreed with me
13:13
<darobin>
I might have to change my mind
13:14
<Ms2ger>
:D
13:14
<jgraham>
Good, then maybe we can end up with code that isn't only accessible to seasoned js veterans :p
13:15
<darobin>
oh please, this is an entry-level idiom
13:15
<jgraham>
But seriously, I am bored of arguing about this
13:15
<jgraham>
If you are prepared to forego this change resolve the issues
13:15
<darobin>
jgraham: hey! you're the one who just reopened the argument :)
13:15
<jgraham>
Otherwise I will change it
13:16
<darobin>
jgraham: IIRC the only problem I care about strongly is the asynchrony one
13:16
<darobin>
because that's actually a bug
13:17
<darobin>
the rest is IIRC just good practice stuff
13:17
<jgraham>
So the only other thing in the review that I recall thinking is better as-is was some of the public-vs-private stuff
13:18
<darobin>
public-vs-private?
13:18
<darobin>
oh you mean putting all your variables as attributes?
13:18
<jgraham>
Weren't there some things that you complained were only used from within the class?
13:18
<darobin>
oh, yes
13:19
<darobin>
it's not so much that I mind, but in reviewing the code if I see a method exposed I expect it to be something that's used from outside somehow
13:19
<darobin>
so for instance I'd wonder why load() doesn't take a path
13:19
<darobin>
it looked like you split out the init code into methods rather systematically, but it's not clear that they could actually meaningfully be used on their own
13:20
<jgraham>
I don't think they can, but the system was nice
13:20
<darobin>
jgraham: I don't reckon this is code that will get used by other code as a library, so it likely doesn't really matter in the long run
13:20
<jgraham>
No, I don't think anyone else will be using this for anything
13:20
<darobin>
but you never know, so I reviewed as I'd review any piece of JS
13:21
<jgraham>
Yep, like I said it's much appreciated
14:28
<zcorpan>
yoav: your way to write attributes is backwards (to what i'm used to, anyway) :-)
14:29
<yoav>
zcorpan: that sounds very likely :)
14:31
<yoav>
I normally don't use the @ thing to write attributes, so when I do, I get it wrong
14:33
<yoav>
zcorpan: Regarding https://github.com/ResponsiveImagesCG/picture-element/issues/89 , it requires direct changes in the HTML spec, right? Or are there existing hooks that can allow that?
14:34
<zcorpan>
yoav: i don't think it requires changes to the HTML spec in particular
14:37
<zcorpan>
yoav: the bug is asking to change "When asked to update the list of source sets for a given picture element picture, user agents must do the following:" to "When asked to update the list of source sets for a given img element img, user agents must do the following:" and then write the algorithm in terms of that
14:37
<zcorpan>
yoav: the difference is subtle, i guess
14:38
<yoav>
zcorpan: Yeah, I missed that
14:40
<yoav>
If no external hooks are required, I agree with you on this issue
15:08
<Ms2ger>
"USPTO.GOV still has URLs that return HTTP/0.9 responses."
15:08
<Ms2ger>
Yay, back-compat
17:00
<gsnedders>
Ms2ger: Compat with HTTP/0.9 is actually wonderful. Try and parse the response as HTTP/1.1; if you fail, treat it as HTTP/0.9.
19:19
<Hixie>
why does http://software.hixie.ch/utilities/js/live-dom-viewer/saved/2725 not work in firefox and safari?
19:19
<Hixie>
am i being dumb?
19:25
<jory>
Hixie: Seems like it's working in FF, AFAICT.
19:25
<Hixie>
what do you get?
19:26
Hixie
updates his nightly
19:26
<Hixie>
i get no message back even on the latest nightly
19:39
<jory>
Hixie: Actually, on reading what the code is actually supposed to do, it's not working in Chrome or FF... looks like setting <body.onmessage> doesn't actually set up your listener properly.
19:39
jory
remembers that consistency does not imply functioning, cause things can be consistently broken too
19:41
<zcorpan>
Hixie: i think offsetWidth is not useful to test since it has an explicit "am i rendered?" path and returns 0
19:42
<Ms2ger>
Hixie, zero with setting the IDL attribute in Fx
19:49
<Hixie>
jory: i get a message in chrome, at least
19:49
<Hixie>
zcorpan: yeah, possible. not sure what is right to test.
19:49
<Hixie>
zcorpan: see thread in whatwg
19:49
<Hixie>
Ms2ger: ?
19:51
<Ms2ger>
document.body.onmessage = ... works
19:51
<jory>
Hixie: Yeah, it works in Canary, just not the... normal... one.
19:51
<jory>
Public one? I don't know what the term for the standard issue is.
19:53
<Hixie>
Ms2ger: weird
19:53
<Hixie>
jory: ah yeah, i'm only testing dev here
19:54
<Hixie>
i believe they call their variants stable, beta, dev, and canary i think
19:57
<zcorpan>
Hixie: maybe window.matchMedia. it returns null in firefox when the frame is invisible. blink returns the right object with .matches == true for '(width:0px)' as the query
19:57
<Hixie>
ah, not familiar with that API
19:58
<Hixie>
like i said in the thread, though, i'm happy to spec stuff if nobody wants to own this on the CSS side.
19:58
<zcorpan>
yeah i'm just popping in here to be clever :-)
20:00
<Hixie>
:-)
20:01
<zcorpan>
svg <switch> might also work
20:03
<zcorpan>
or maybe that didn't do media queries
20:19
<zcorpan>
MikeSmith: http://krijnhoetmer.nl/irc-logs/whatwg/20140107#l-387 - something in particular in the thread being gold? (pointer?)
22:37
<Ms2ger>
Hixie, dom.spec.whatwg.org appears to be down
22:37
<Ms2ger>
annevk-cloud, ^
22:38
<krijn>
Haha, noobs!
22:38
<Ms2ger>
Also, how do you spec "in the DOM"?
22:52
<Hixie>
Ms2ger: wfm
22:53
<Hixie>
Ms2ger: how do you mean, "in the DOM"?
22:53
<Hixie>
Ms2ger: if you mean "in a document", HTML has some macros for that
22:53
<Ms2ger>
Furthest ancestor is the document
22:53
<Ms2ger>
And seems to be up now, yes
22:54
<Ms2ger>
And "in a document" was what I was looking for
23:26
<dbaron>
Hixie, so a number of years ago I remember thinking that the intent of the IDN (and maybe also IRI) specs was that different "slots" (places where a URI went) would have to specify individually whether they accepted the new features. Looking back now, I can't seem to find any backing for this understanding in the specs. Do you remember having a similar understanding, or was I just overinterpreting?
23:28
<Hixie>
i have the same understanding
23:28
<Hixie>
i don't really see how else it could work, i mean, you can't force places that only ever accepted ASCII to suddenly accept Unicode.
23:29
<dbaron>
looking at the specs now... it seems like IDN defines an "IDN-aware slot" concept... but that IRI then says IRIs can have unicode hostnames and essentially says they're IDN-aware
23:29
<Hixie>
IRIs are always IDN-aware. but not everywhere accepts IRIs.
23:29
<dbaron>
hmmm... pretty sure <a href="..."> deals with Unicode these days
23:30
<dbaron>
even though it didn't at one point
23:30
<Hixie>
well <a href=""> just references the URL standard, which bypasses all of the URI and IRI specs.
23:30
<dbaron>
right... I'm trying to finish a blog post about versioning that I started 7 years ago
23:30
<Hixie>
heh
23:30
<dbaron>
and one of the points I wanted to make about how some of those specs were doing it wrong
23:31
<dbaron>
but I'm having trouble finding evidence for that
23:31
<dbaron>
I think the thing we thought they were doing was wrong, but I'm no longer convinced it's what they say.
23:31
<Hixie>
with <a href=""> we added idn support by first updating all the software, basically. Also, we always accepted IRIs there, in a poorly-defined (until recently) way
23:31
<Hixie>
i dunno what doing it right would really look like for IDN. it's a tough space.
23:31
<Hixie>
there's better examples of "doing it wrong" for this kind of thing.
23:32
<Hixie>
XHTML2, e.g., but also XML 1.1
23:32
<dbaron>
I was going to use XML 1.1
23:32
<Hixie>
also people are always wanting to add versions to their protocols even though you can never really do anything with them
23:32
<dbaron>
though if I could use URI/IRI it would lead to much easier-to-read examples
23:32
<Hixie>
e.g. websocket ended up with some version nonsense
23:33
<Hixie>
despite my best efforts
23:33
<Hixie>
(sigh IETF)
23:34
<Hixie>
URI and IRI had lots of problems, but versioning isn't one of them, i think
23:34
<Hixie>
IDNA vs Unicode is a better example of versioning issues
23:35
<Hixie>
but anne is the one to ask about taht
23:37
<dbaron>
k, thanks for the help