00:00
<TabAtkins>
What mail is coming from w3c.org?
00:01
<annevk>
public-webapps email
00:02
<annevk>
nn
00:18
<tiglionabbit>
how can you tell the difference between a user dragging a file or a folder into the browser?
00:18
<tiglionabbit>
in either case, event.dataTransfer.items[0].kind == "file"
01:11
<jamesr>
oh wise wizards, consider this test: http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1805
01:12
<jamesr>
i can write a selector to match the span with id ":foo" by \ escaping, but querySelector("#\:foo") barfs in firefox/chrome/opera. TabAtkins thinks this is just a bug in querySelector which i tend to agree with but does anyone know of a good reason for the current behavior before i file a batch o' bugs?
01:13
<jamesr>
i get really suspicious when i see a seemingly bizarre behavior that has good interop
01:13
<TabAtkins>
I'll support myself and say that, no, there's no reason, and it's definitely a bug.
01:14
<TabAtkins>
Note that this interop is probably on something very bad for authors, since it's an exception rather than being silently ignored, and there's no way around it. (Omitting the escaping slash has the same effect.)
01:14
<jamesr>
even ie9 throws here
01:14
<TabAtkins>
Jeez, it even fails when you do "#\3a foo".
01:15
<jamesr>
yeah it's damn picky
01:15
<jamesr>
is http://www.w3.org/TR/css3-selectors/#grammar supposed to be up to date?
01:15
<TabAtkins>
I'm guessing there's an eager unescaping phase that runs before the syntax checker.
01:15
<jamesr>
it's on /TR/ so i suspect no
01:15
<TabAtkins>
Probably, yes. Worst case, check Selectors 4.
01:27
<Hixie>
jamesr: querySelector("#\:foo") is exactly equivalent to querySelector("#:foo")
01:27
<TabAtkins>
jamesr: There is no grammar-based reason. The escaping thing is part of CSS/Selectors, and I recall there being an example of using a weird character in a selector somewhere, either in 2.1 or in Selectors.
01:27
<Hixie>
jamesr: try querySelector("#\\:foo")
01:27
<TabAtkins>
Sigh, I forgot about that. Goddam double escaping.
01:27
<jamesr>
aha!
01:28
TabAtkins
has memories of \\ in a PHP string to escape \ in a PHP regex to select a .
01:28
<zewt>
has any (real) language besides python attempted to solve that? heh
01:28
<TabAtkins>
You mean with Python's raw strings?
01:28
<jamesr>
writing code to deal with windows paths in any c-style language is fun
01:28
<zewt>
jamesr: it's easy. "c:/foo/bar"
01:29
<Hixie>
matching urls with regexps is fun for a similar reason -- /.../
01:29
<TabAtkins>
Most path apis accept either direction of slashes, no?
01:29
<jamesr>
yes, most. :)
01:29
<TabAtkins>
Hixie: Most decent languages (not JS ;_;) let you use different delimiters for the regex.
01:29
<zewt>
Hixie: that's why most things ... ^ that
01:29
<jamesr>
sed -e 's?check/this/out?arbitrary/delimiters/baby?'
01:33
<Hixie>
yeah, i usually am using perl so it's no biggie once i realise the pattern is gonna contain a path/url
01:34
<Hixie>
but i often don't notice until after the first set of syntax errors :-)
06:20
<zcorpan>
LOOOOOOOOOOOOOL https://www.w3.org/Bugs/Public/show_bug.cgi?id=19087
06:23
<Hixie>
aw man
06:23
<Hixie>
i'd better check with ben goodger to see if he agrees that it's offensive
06:23
<Hixie>
the word appears 12 times in the spec!
06:24
Hixie
ponders just reassigning it to the htmlwg :-P
06:26
<jgraham>
Oh look, W3C are hiring someone to work on tests
06:26
<jgraham>
Or testing, I guess
06:27
<jgraham>
Seems like a nice idea, but the requirment for fluent japanese cuts down the potential candidates by a lot :)
06:28
<jgraham>
(I wonder if it is one of those things where they already know who they want for the job and have written the advert to match)
06:33
<heycam>
I think it's more likely that they need to hire someone through Keio for whatever reason
06:35
<heycam>
I've never heard of a NZer being offended by the term kiwi before
06:45
<doublec>
I guess it depends if the term kiwi refers to be peeled and eaten
06:57
<jgraham>
doublec: You are saying that "I'm going to eat you like a kiwi, kiwi" is offensive, but "I'm going to eat you like a kiwi-fruit, kiwi" isn't? :p
06:58
<jgraham>
(or am I misunderstanding the bug? AFAICT it doesn't claim that New Zealanders are offended by "kiwi", except when it is used to denote a fruit)
06:59
<jgraham>
(which I agree doesn't make any sense)
07:02
<doublec>
I'm not offended but I assume the bug reporter didn't like the "peel the kiwi's skin, then eat the kiwi's flesh"
07:30
<othermaciej>
my favorite part is how an Australian is reporting that New Zealanders may be offended by the term "Kiwi"
07:36
<mpt>
As a kiwi, I'm not offended when people talk about eating kiwis, I just hope this age of cannibalism will one day come to an end.
07:39
<hsivonen>
at least the tag cloud example has context that suggests fruit rather than bird or people
07:40
<mpt>
Aww, tag clouds, how very oughties
07:40
<zcorpan>
"we will encourage modularity as a preferred way to approach introducing new features into the 5.1 release."
07:41
<Ms2ger>
Was the TAG involved?
07:48
<annevk>
no man, modularity is Team-speak for small specs
07:48
<annevk>
small specs with holes inbetween
07:49
<annevk>
I have oft wondered where that modular desire comes from. No young people talk about that stuff
07:50
<Ms2ger>
Should get to rec faster
07:51
<annevk>
Microsoft Program Managers?
07:51
<zcorpan>
gotta let each feature mature at its own pace and have its own community around it (i.e. 4 new mailing lists)
07:51
<annevk>
maybe the funding is per mailing list
07:51
<zcorpan>
that might explain things
08:42
<annevk>
hehe, I think I'm about to own wtw.tf
08:51
jgraham
wonders what a wt is
09:31
<annevk>
jgraham: I'm helping out dino
09:31
<annevk>
https://twitter.com/grorg/status/156849367448428545
09:37
<hsivonen>
since when has Opera not escaped < and > in attribute values in the innerHTML getter?
09:45
<zcorpan>
hsivonen: about since http://html5.org/tools/web-apps-tracker?from=2362&to=2363
09:46
<annevk>
nice URL there
09:48
<zcorpan>
hsivonen: iirc that page does something unusually stupid like eval()ing the innerHTML of an <a href=javascript:...> or some such, which didn't work when we escaped <> (and nobody else escaped them *for href=javascript:*)
09:49
<hsivonen>
zcorpan: we have a bug about the UI for a dark-matter Cisco appliance breaking because we stopped escaping <>
09:49
<hsivonen>
I gather it tries to parse innerHTML output as XML
09:50
<zcorpan>
hsivonen: ok. i guess we're screwed either way :-(
09:50
<hsivonen>
Don’t worry, your network is safe with Cisco.
09:57
<hsivonen>
by quick black-box testing, it seems the WebKit special case applies if the attribute reflects as URL and the URL has the scheme javascript
10:05
<zcorpan>
yeah i recall something along those lines as well. old gecko something similar.
10:06
<annevk>
I wonder who's going to edit XHR for the W3C
10:08
<zcorpan>
hsivonen: which attribute does cisco use here?
10:18
<hsivonen>
zcorpan: DOMParser barfs on <operator name="<" value="LESS_THAN"></operator>
10:18
<hsivonen>
this would be a non-issue if DOMParser did XML5
10:21
<hsivonen>
they have a massive “XML” island inside <xml>…</xml>
10:21
<zcorpan>
ok and the other page expected not-escaping for <a href=javascript:> (iirc)
10:24
<jgraham>
hsivonen: OOI, what are your travel plans for TPAC?
10:29
<hsivonen>
jgraham: plane from Helsinki to CDG, TGV from CDG to Lyon Part-Dieu
10:30
<hsivonen>
arrival on Sunday, leaving on Saturday
10:30
<jgraham>
OK
10:30
jgraham
is doing Test The Web Forward in Paris on the Friday/Saturday and so will also travel by train on the Sunday
10:31
<hsivonen>
this Cisco thing seems to expect there are three browsers: IE, Firefox and Safari
10:31
smaug____
will be in Lyon also Sun-Sat
10:31
<smaug____>
hsivonen: fun
10:32
<smaug____>
hsivonen: thanks for looking the issue
10:32
<hsivonen>
smaug____: now we need to decide what to do about it
10:32
<hsivonen>
which is harder
10:32
<jgraham>
Someone should do a list like http://www.kalzumeus.com/2010/06/17/falsehoods-programmers-believe-about-names/ but for web browsers
10:32
<annevk>
jgraham: how much is the train?
10:32
<annevk>
I guess I could fly into Paris and then fly back from Lyon
10:32
<jgraham>
annevk: That probably depends, it's the last bit that I haven't booked
10:33
<hsivonen>
annevk: depends on time of day
10:33
<jgraham>
Seems to be somewhere in the range of 35-60 eur
10:33
<darobin>
the train can be relatively cheap if you book it ahead
10:33
<annevk>
ah okay
10:33
<darobin>
it's 2h — prolly shorter than flying to Lyon
10:33
<darobin>
(all told)
10:34
<hsivonen>
also, Lyon Part-Dieu has the airport code XYD
10:34
<smaug____>
hsivonen: I hate special cases
10:34
<hsivonen>
for plane-train bundle tickets
10:34
<smaug____>
hsivonen: so I like the current behavior
10:34
<hsivonen>
but I didn’t know that when I booked
10:34
<darobin>
hsivonen: you mean Lyon Saint-Exupéry — Part-Dieu is the train station
10:34
darobin
bbiab
10:34
<hsivonen>
smaug____: I like the current behavior, too
10:34
<hsivonen>
darobin: the airport is LYS
10:35
<hsivonen>
darobin: that magic code is for use for CDG–XYD train when bought through Air France ticketing
10:35
<SimonSapin>
Paris-Lyon by plane, you would spend as much time in the airports than in the air. The TGV train is 2h but you just have to get there a few minutes before.
10:36
<hsivonen>
also, TGV from CDG is cheaper than taxi from Saint-Exupéry to city center
10:36
<annevk>
"People’s names are all mapped in Unicode code points." is interesting
10:37
<annevk>
(that statement is false per the link from jgraham, fwiw)
10:37
<hsivonen>
annevk: cruel parents in this day and age, IMO, in that case
10:37
<smaug____>
silly me, I have flight from Paris to Lyon. Didn't know the train would be that much better.
10:38
<SimonSapin>
annevk: does that mean that Unicode needs more characters?
10:38
<hsivonen>
SimonSapin: no. Then parents who want to be special would invent something new that’s unique on not in Unicode
10:38
<annevk>
I don't know what it means
10:39
<hsivonen>
I’m pretty sure it means Han/Kanji names that are deliberately an unconventional combination of quill strokes to be unique
10:39
<annevk>
probably that the world is odd and full of unknown things, lots of which is created by us
10:40
<smaug____>
hsivonen: I wonder which all attributes webkit escapes
10:40
<jgraham>
Also consider the artist sometimes known as prince
10:54
<darobin>
hsivonen: ah, of course — I obviously never go through AF to get a train ride here :)
10:55
<darobin>
smaug____: if you're already on a flight and just connecting through Paris, then doing the last leg on a flight is okay
10:56
<darobin>
it's mostly shorter if you're in Paris, since the train is centre to centre
10:56
<hsivonen>
darobin: properly ticketed train is CDG to Lyon center
10:57
<darobin>
hsivonen: I know, but in that case I doubt it comes out shorter than flying to SaintEx (or by much of a diff)
10:57
<hsivonen>
shorter might not be the only thing to optimize
10:59
hsivonen
traveled from Finland to Switzerland and back without flying this summer
10:59
<darobin>
having gone from Grenoble to Å on a train, I can't easily disagree
11:00
<hsivonen>
the only part that was worse than a plane was Baltic buses from the worse company
11:00
<hsivonen>
(Baltic buses from the better company were better than plane)
11:00
<darobin>
heh
11:01
<darobin>
in my case we had a problem with the two boats that did the last leg having rammed one another in the port, and so being routed through another boat that went to the north of the Lofoten
11:01
<darobin>
which in turn went through a massive storm during fish migration period
11:01
<hsivonen>
Better company: http://www.luxexpress.eu/en Worse company: http://ecolines.net/en/
11:02
<darobin>
fish migration means that you can use sonar, which can be problematic when there are reefs
11:02
<darobin>
so the boat had to make way using spotlights — it took a while
11:02
<darobin>
I thoroughly enjoyed it though — but I think other passengers would've preferred to fly ;)
11:06
<annevk>
hmm, maybe I should finally add chunked to XHR
11:07
<annevk>
either that or work on that query API
11:11
<zcorpan>
hsivonen: http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1807
11:13
<zcorpan>
smaug____: ^
11:17
<zcorpan>
maybe only a few of those need to be supported, since javascript: doesn't work in most places anyway (at least in opera)
11:17
<hsivonen>
zcorpan: supports my hypothesis of what the rule is
11:17
<hsivonen>
zcorpan: thanks
11:17
<zcorpan>
hsivonen: except <html manifest> doesn't have a reflecting IDL attribute
11:18
<hsivonen>
oh
11:19
<zcorpan>
well... per spec... webkit has one
11:20
<zcorpan>
that reflects as a URL
11:22
<annevk>
why did they add one? :/
11:22
<annevk>
hmm
11:24
<zcorpan>
i think innerHTML should use a fixed list instead of hooking in to "reflecting"... like: a href, iframe src, form action, frame src
11:24
<zcorpan>
or just a href if that's enough for compat
11:26
<zcorpan>
unfortunately this issue is hard to get compat data on
11:30
<smaug____>
I think innerHTML shouldn't escape
11:30
<smaug____>
if IE doesn't escape
11:31
<zcorpan>
if you evang Cisco, WFM :-)
11:52
<AryehGregor>
And everyone with outdated Cisco firmware, too?
13:19
<annevk>
seems nobody ever replied to other browsers being interested in chunked
13:29
<annevk>
btw
13:29
<annevk>
I don't like "Creating" for URLs at all
13:30
<annevk>
I want "Parsing"; what do I call what is named "parse" now?
13:30
<annevk>
the sweet generic "process"?
14:03
<zewt>
"Sorry but it is not late. The spec is still a working draft" ... heh
14:13
<zewt>
annevk: should the "remove any leading and trailing..." step be in create or parse? currently it seems like url = new URL(" http://foo.com "); may have a different result than url.href = " http://foo.com ";
14:13
<annevk>
I think it should be added to setting .href too
14:13
<zewt>
annevk: also the URL ctor doesn't define the case where base is a DOMString rather than a URL
14:14
<zewt>
in what case wouldn't you want it in parse?
14:14
<annevk>
setting pathname
14:14
<annevk>
zewt: I thought it did define that
14:15
<zewt>
it's probably somewhere else
14:15
<zewt>
oh yeah creating only deals with URL base so I imagine that's wherever the actual ctor is defined
14:18
<annevk>
I'll fix setting .href and rename Create to Parse and Parse to Process
14:20
<zewt>
annevk: btw. in encoding's "Unless stated otherwise emitting an encoder error", that might want to say something like "unless stated otherwise by the caller"; i interpreted that as "unless stated otherwise in the place that says emit an encoder error"
14:20
<zcorpan>
won't that result in more Process discussions?
14:20
<annevk>
zcorpan: what?
14:21
<zewt>
i think that was a joke :)
14:21
<annevk>
oh :)
14:21
annevk
is looking at http://en.wikipedia.org/wiki/Parser for a better word
14:22
<zewt>
(it would be nice for things that affect algorithm flow to be a part of the algorithm, instead of redefining things like "emit an encoder algorithm" externally, so the code paths of the spec are self-contained when possible)
14:22
<annevk>
maybe "interpret"?
14:22
<zewt>
annevk: it doesn't look like you'd ever (or very rarely) want to invoke the "parse" url algorithm from outside of the spec (you'd only use create), so that doesn't necessarily need to be short and sweet
14:23
<zewt>
eg. "update the URL state from the string..." or whatever
14:23
<zewt>
off to work, later
14:24
<annevk>
hmm, not sure if longer makes it better, but feel free to do a pull request :)
18:21
<Yuhong>
On legacy color parsing: https://news.ycombinator.com/item?id=4581362
18:21
<Yuhong>
http://stackoverflow.com/questions/8318911/why-does-html-think-chucknorris-is-a-color
18:22
<annevk>
even on stackoverflow and ycombinator people don't bother to read the spec
18:23
<miketaylr>
s/even/especially/
21:26
<krijn>
annevk: y u no in #fronteers ? ;(
22:42
<Hixie>
anyone know if there's an animated gif spec, or something that passes for such a spec, anywhere?
22:50
<Hixie>
is http://www.ietf.org/rfc/rfc2818.txt the HTTPS spec?
22:54
<jsbell>
hixie: googling for "gif89a" suggests http://www.w3.org/Graphics/GIF/spec-gif89a.txt which seems reasonable (compuserve 1990). No idea if anyone has documented modern browser handling of the beasts, though.
22:54
<Yuhong>
Anyway, I am on StackOverflow and just posted a link to the spec as an answer.
22:54
<Yuhong>
http://stackoverflow.com/questions/8318911/why-does-html-think-chucknorris-is-a-color/12630675#12630675
22:55
<Hixie>
jsbell: doesn't look like it specs the animated pars
22:55
<Hixie>
parts
22:56
<Hixie>
jsbell: at least not all the netscape extensions
23:01
<jsbell>
Huh; I didn't realize that looping was an extension.
23:02
<jsbell>
hixie: knowing the term "application extension" leads me to http://www.matthewflickinger.com/lab/whatsinagif/
23:03
<jsbell>
which leads to http://odur.let.rug.nl/~kleiweg/gif/netscape.html - a reasonable spec-like description of the NETSCAPE2.0 extension...
23:03
<zewt>
. o ( still dreaming of a css style to stop gif animation )
23:04
<zewt>
also horrifying fact of my week: ios safari (seemingly randomly) caches xhr POSTs : |
23:08
<Yuhong>
And on document.all, it is unfortunate that DOM1 became a REC just as Netscape "Mariner" was being cancelled.
23:09
<Hixie>
jsbell: thanks
23:09
<Hixie>
jsbell: i guess i'll go with that, it's better than nothing!
23:09
<jsbell>
worth what you paid for it.
23:10
<Hixie>
:-)
23:10
<Yuhong>
Which caused IE to gain market share.
23:12
<Yuhong>
To the point where other browsers had to implement document.all
23:13
<Yuhong>
And leaving no browser that supports both document.layers and document.getElementById.