00:43
<MikeSmith>
http://www.w3.org/2005/Incubator/htmlspeech/XGR-htmlspeech-20111206/ is actually a really interesting document
00:44
<smaug____>
MikeSmith: comments welcome :)
00:44
<MikeSmith>
smaug____: you participated in that XG?
00:45
<smaug____>
yup
00:45
<MikeSmith>
oh cool
00:45
<MikeSmith>
I see your name in the editors list
00:51
<MikeSmith>
smaug____: the intro at http://www.w3.org/2005/Incubator/htmlspeech/XGR-htmlspeech-20111206/#introduction is the clearest high-level overview as far as explaining to Web-app developers what the proposed features would enable
00:52
<MikeSmith>
I think I'll point people to that part to read first
00:53
<smaug____>
yeah
00:54
<smaug____>
Google wants to do a smaller API first
00:54
<smaug____>
probably in WebApps WG if possible
00:55
<smaug____>
and I agree, smaller API first would be good
00:55
<smaug____>
implementing everything from the XG report is a huge task
01:13
<MikeSmith>
smaug____: yeah, it's pretty ambitious
01:13
<MikeSmith>
I would like to see a proposal for that smaller API
09:37
<webben>
what spec should i be looking at for the replacement for mutation events?
09:42
<Ms2ger>
DOM4
09:45
<webben>
Ms2ger: Ah, cheers.
10:14
<MikeSmith>
Ms2ger: on http://platform.html5.org/ yesterday, I added links to test suites/test cases where I could find any
10:14
<MikeSmith>
including to "submitted" tests, not just to approved ones
10:14
<MikeSmith>
the ⓣ links
10:15
<MikeSmith>
if you see anything missing that I should add links for, lemme know
10:15
<MikeSmith>
or fork https://github.com/sideshowbarker/platform.html5.org and make a pull request
10:18
<MikeSmith>
annevk: all links after Server-Sent Events ones still not working in my Opera
10:19
<MikeSmith>
and still I have no clue why
10:19
<MikeSmith>
looked at DOM and it's as expected
10:19
<MikeSmith>
and no errors logged to console
10:20
<zcorpan>
wfm
10:24
<MikeSmith>
zcorpan: weird
10:25
<MikeSmith>
I see the same problem in both Opera.next and 11.60
10:41
<Ms2ger>
Does HTML5 still have classList?
10:41
<MikeSmith>
did last time I checked
10:42
<MikeSmith>
http://www.whatwg.org/specs/web-apps/current-work/multipage/elements.html#dom-classlist
10:42
Ms2ger
files a bug
10:45
<MikeSmith>
Ms2ger: was it meant to be dropped?
10:45
<Ms2ger>
smaug____, were you working on a ProgressEvent constructor already?
10:45
<Ms2ger>
MikeSmith, it's moved to DOM
10:45
<MikeSmith>
ah
10:45
<smaug____>
Ms2ger: there is a bug open for that
10:45
<Ms2ger>
Good
10:45
<smaug____>
Ms2ger: bug 710882
10:47
<Ms2ger>
MikeSmith, http://w3c-test.org/html/, http://test.csswg.org/harness/
10:47
<Ms2ger>
http://hg.ecmascript.org/tests/test262/
10:48
<Ms2ger>
SVG also has tests somewhere
10:48
<MikeSmith>
yeah, I'll ask Doug about those
10:50
<MikeSmith>
thanks for the other links -- added
10:50
<Ms2ger>
https://dvcs.w3.org/hg/webevents/raw-file/6c149e83cd5d/test/touchevents/single-touch.html is apparently the only touch events test
10:51
<smaug____>
there is another for multitouch
10:51
<smaug____>
a simple one
10:51
<smaug____>
somewhere in pettay.fi. Art should move it to w3.org
10:51
MikeSmith
adds that one for now
10:52
MikeSmith
finds http://w3c.pettay.fi/webevents/tests/touchevents/
10:55
<smaug____>
MikeSmith: yeah, the single-touch.html should be the same as what is now in w3.org
10:55
<MikeSmith>
ah, OK
11:29
<Velmont>
zcorpan, MikeSmith: I have the same problem with links after SSE not working on http://platform.html5.org/
11:30
<MikeSmith>
Velmont: oh, cool
11:30
<MikeSmith>
well, not cool
11:30
<MikeSmith>
but at least we know it's reproducible
11:30
<MikeSmith>
and not a figment of my imagination
11:30
<MikeSmith>
oh
11:30
<MikeSmith>
it's Velmont
11:30
<Velmont>
MikeSmith: lol :P -- Yes, some bug with links and columns then I guess. I'll see if I can make a reduced TC.
11:30
<MikeSmith>
file a but please man
11:31
<MikeSmith>
thanks (about making reduced test case)
11:33
<annevk>
ah shit
11:33
<annevk>
I need a bigger harddisk
11:33
<annevk>
I wonder how much a 256 goes for these days
11:34
<MikeSmith>
Velmont: hmm, yeah, I see now if I switch from 3-column to 2-column or single-column, the problem goes away
11:34
<Ms2ger>
TB?
11:34
<zcorpan>
SSD?
11:34
<MikeSmith>
Velmont: oh, actually I see it doesn't -- it just moves
11:36
<MikeSmith>
Ms2ger: thanks for moving those tests
11:36
<Ms2ger>
Np
11:36
<MikeSmith>
I had actually been planning to do that but you are far lazier than me
11:36
<annevk>
yeah SSD
11:36
<MikeSmith>
Ms2ger: I should send you a copy of my TODO list
11:37
<annevk>
I'm on 128 now and with this new Windows 7 installation within VMWare it's beginning to crack down
11:37
<Ms2ger>
I'd been planning to do that for months as well :)
11:38
<MikeSmith>
heh
11:38
MikeSmith
adds some more test links
11:39
<MikeSmith>
sweet
11:39
<MikeSmith>
those tests even use testharness.js
11:42
<annevk>
I guess I'll call the shop that ordered my MacBook, see if they can do something
11:42
<Velmont>
It's seemingly affected by the height of the browser window...
11:42
<annevk>
hah, I do have 8GiB of memory
11:43
<MikeSmith>
Velmont: I see
11:43
zcorpan
also has 128 and 8
11:43
<Velmont>
Less height = less linky goodness, more height = more linky goodness.
11:43
<MikeSmith>
aha
11:44
<Velmont>
Smells optimization.
11:44
<MikeSmith>
devs getting too smart for their own good
11:45
<MikeSmith>
it is great to have proper column-break support, though (outside of the link bug)
11:46
<MikeSmith>
things don't look so great if a browser breaks the column right after a <dt> instead of before it
11:48
<MikeSmith>
anybody know if Hallvord has tests for Clipboard API and events?
11:48
<MikeSmith>
also, any tests somewhere for requestAnimationFrame
11:48
<annevk>
zcorpan: does platform.html5.org also work for you on the second column?
11:48
<zcorpan>
yes
11:48
<annevk>
Ms2ger: Gecko did remove initCustomEvent, right?
11:48
<annevk>
zcorpan: oh
11:49
<Velmont>
zcorpan: What screen size?
11:50
<Ms2ger>
Not sure
11:51
<Ms2ger>
Doesn't look like it
11:52
<MikeSmith>
and still haven't found any Web Messaging tests
11:59
<smaug____>
annevk: I'm not removing any init*Event methods from gecko
12:00
<smaug____>
I mean, not right now
12:00
<zcorpan>
Velmont: worked with various zoom levels
12:01
<MikeSmith>
heh "I still can't believe javascript - the f**ing backbone-language of the web - doesn't offer an API for mutating URLs."
12:01
<Ms2ger>
It does now, doesn't it?
12:12
<annevk>
I wonder what Marat wants the DOM4 editors to say
12:12
<annevk>
smaug____: kk
12:12
<Ms2ger>
"bz is right"
12:12
<smaug____>
:)
12:17
<MikeSmith>
is the URL API actually supported already?
12:18
<Ms2ger>
smaug____, do you remember if we landed the Gecko patch?
12:18
<MikeSmith>
oh, it is in Gecko
12:19
<annevk>
meh
12:20
<annevk>
not exactly cheap
12:20
<MikeSmith>
how do I get an actual console with completion in Dragonfly?
12:21
<bga>
lol http://dvcs.w3.org/hg/url/raw-file/tip/Overview.html
12:21
<smaug____>
Ms2ger: which gecko patch?
12:21
<smaug____>
Ms2ger: for URLs?
12:21
<Ms2ger>
Yeah
12:21
<smaug____>
I don't think so
12:22
Ms2ger
checks
12:23
<annevk>
bga: what?
12:23
<bga>
do you want standardize all things in the world?
12:24
<Ms2ger>
Nope
12:24
<Ms2ger>
bga, yes.
12:24
<bga>
anyway api is bad
12:24
<annevk>
send feedback!
12:24
<bga>
sec
12:33
<Velmont>
MikeSmith: ...? Go to the console tab? It should have tab completion.
12:34
<MikeSmith>
Velmont: don't seem to do completion for me the way it used to.. Maybe I need to upgrade my Dragonfly
12:35
<MikeSmith>
oh
12:35
<MikeSmith>
working as expected in my Opera.next
12:35
<MikeSmith>
weird
12:36
<Velmont>
It works here in 11.60. But then again, -- you can run any other DragonFly in the browser, -- Opera doesn't really bundle DragonFly but downloads it from the web and uses appcache to store it. AFAIR.
12:42
<MikeSmith>
Velmont: yeah, I knew that and thought it also auto-updated
12:43
<MikeSmith>
how do I tell what version of Dragonfly I have?
12:46
<MikeSmith>
oh, found it
12:46
<MikeSmith>
Settings > About in Dragonfly
12:59
<kennyluck>
webben, (re. what spec should i be looking at for the replacement for mutation events?) and the undomanager spec maybe?
13:01
<bga>
annevk beta version :) http://pastie.org/3079269
13:02
<bga>
full decomposition
13:04
<annevk>
bga: URIs are called URLs in APIs
13:05
<annevk>
no opinion other than that really, emailing whatwg⊙wo seems best
13:05
<bga>
fixed
13:08
<annevk>
for people with opinions on encodings and conformance checkers: https://www.w3.org/Bugs/Public/show_bug.cgi?id=15332
13:08
<annevk>
(already added hsivonen)
13:31
AryehGregor
gloomily contemplates 427 unread conversations in his spec inbox
13:33
<annevk>
I was gonna say that it's not a lot of email, but I guess it's a bit more email than 427 if that's conversations
13:34
<annevk>
MikeSmith: what would r12a want?
13:34
<MikeSmith>
dunno
13:34
<Ms2ger>
Is wwrw the new wwjd?
13:35
<MikeSmith>
annevk: he's already not happy about the reporting that v.nu does for UTF-16
13:35
<gsnedders>
Oh, I guess it's Merry Hixieday today!
13:36
<MikeSmith>
annevk: actually, I think he's not happy about what the spec says about use of UTF-16
13:36
<Ms2ger>
Indeed it is
13:36
<annevk>
wwrw?
13:37
<annevk>
AryehGregor: I guess you're married now, so congratulations again!
13:37
<AryehGregor>
Thanks.
13:37
<gsnedders>
AryehGregor: Congrats!
13:38
<Ms2ger>
^What they said
13:38
<annevk>
MikeSmith: we could make an exception for UTF-16, though everyone tells me it's a world of hurt
13:38
<annevk>
MikeSmith: making an exception for tis-620 and such though... dunno
13:38
<MikeSmith>
yeah
13:40
<MikeSmith>
about conformance-checking tools, Richard has an online tool that he's put a lot of work into and that represents what he thinks the proper type/level of reporting should be, so he kind of uses that a benchmark
13:41
<MikeSmith>
http://validator.w3.org/i18n-checker/
13:44
<MikeSmith>
oh hey
13:44
<MikeSmith>
for non-UTF8 docs, it does actually report "Non-UTF-8 character encoding declared"
13:44
<MikeSmith>
so maybe I need to STFU
13:45
<Ms2ger>
That certainly would make life less interesting
13:45
<MikeSmith>
heh
13:45
<annevk>
so if serve "7A 00" as utf-16be only IE outputs "z"
13:45
<MikeSmith>
and expanding that message shows among others things "UTF-16 is also a character encoding based on Unicode, but is little used on the Web, and generally best avoided."
13:45
<annevk>
everyone else gives "稀"
13:46
<annevk>
if I serve it up as utf-16 (without BOM) everyone defaults to utf-16le
13:47
<MikeSmith>
man, Richard puts a lot of work into this stuff -- he's added more since the last time I checked: "Using non-UTF-8 encodings can also have unexpected results on form submission and URL encodings... It is not a requirement to use UTF-8, but the HTML5 specification recommends its use, and you should consider it."
13:48
<MikeSmith>
and also "What to do: Replace the http-equiv and content attributes in your meta tag with a charset attribute"
13:48
<annevk>
utf-16le with FE FF 7A 00 results in "稀" in WebKit/IE
13:48
<annevk>
Gecko gives "�竿" and Opera "�z"...
13:49
<gsnedders>
Oh nice. :\
13:51
<annevk>
for utf-16be with le BOM it's kind of the same
13:51
<annevk>
conclusions: BOM is more important
13:51
<annevk>
utf-16le == utf-16 in WebKit/Trident
13:51
<annevk>
(afaict)
13:56
<annevk>
gsnedders: just to be sure, PHP just looks at octets right?
13:56
<gsnedders>
annevk: yup
13:56
<annevk>
gsnedders: that is I can write PHP using ASCII octets and follow the ?> with utf-16 octets
13:57
annevk
uses PHP to set a header
13:57
<gsnedders>
annevk: PHP has no concept of anything except ASCII
13:57
<annevk>
and it also doesn't mangle anything right?
13:57
<annevk>
in that sense it's kind of neat
13:57
<annevk>
maybe also vulnerable and what not, but neat
13:58
<gsnedders>
annevk: indeed
13:59
<annevk>
thanks
14:51
<annevk>
emailed utf-16 research to the WHATWG list
14:51
<annevk>
took a while to write it all out
14:56
<zewt>
annevk: using html mail set to a fixed-width font for tables makes them more readable for a lot of people, fyi
14:58
<annevk>
wait I'm using html email?
14:58
<divya>
ouch
14:59
<annevk>
I'm writing plain text as far as I know
14:59
<annevk>
http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2011-December/034260.html confirms
14:59
<zewt>
no, i mean if you have stuff that needs to be viewed as fixed-width, it's best to include an html version so you can set that
14:59
<zewt>
else you get the default of "whatever the user's viewer is set to", which will often be proportional
15:00
<zewt>
setting fonts in email has few legitimate use cases, but this is one of them
15:00
<annevk>
AryehGregor: please read IDL before filing a bug
15:00
<annevk>
AryehGregor: in particular (Node or DOMString) is valid
15:00
<annevk>
afaik
15:00
<AryehGregor>
You mean WebIDL?
15:00
AryehGregor
looks
15:01
<AryehGregor>
Is that a recent change?
15:01
<annevk>
yes
15:01
<AryehGregor>
Oh, I see.
15:01
<annevk>
in any event, I disagree with the premise of your bug
15:02
<annevk>
I'm not going to define experimental IDL in prose if I expect IDL to define it at some point in the near future
15:02
<AryehGregor>
Well, as long as there's at least a concrete proposal for the syntax that's expected to be looked at soon.
15:03
<AryehGregor>
Then I don't have an issue with it, if it's going to be resolved soon. No need to have strict ordering in that case.
17:55
<AryehGregor>
Is there any JS-visible difference between "interface Foo { void foo(); }; interface Bar { void bar(); }; Foo implements Bar;" and "interface Foo { void foo(); void bar(); };"?
17:56
<AryehGregor>
I mean, does "implements" just mean "pretend all the interface members are really on the other interface too"?
17:57
<bga>
depends how you implement OOP
17:58
<AryehGregor>
How so?
17:58
<bga>
var Bar = { bar: }; var Foo = { __proto__: Bar, foo: }
17:58
<bga>
1st case
17:59
<AryehGregor>
I'm asking what the WebIDL spec requires.
18:01
<smaug____>
AryehGregor: there is certainly difference. In the latter case "Bar" in window would be false
18:02
<smaug____>
(but I don't recall what WebIDL says about prototypes)
18:02
<Ms2ger>
AryehGregor, if you add [NoInterfaceObject], I think not
18:02
<AryehGregor>
Alternatively, is it the same as "interface Foo { void foo(); void bar(); }; interface Bar { void bar(); };"?
18:03
<smaug____>
hmm, what does WebIDL say about instanceof
18:04
<Ms2ger>
Read it :)
18:04
<AryehGregor>
I don't think it says anything.
18:04
<AryehGregor>
It just uses the ES5 definition, doesn't it?
18:04
<Ms2ger>
Anyway, you probably shouldn't use implements without [NoInterfaceObject]
18:15
<AryehGregor>
The IDLs for CSSOM and HTML have circular dependencies. :(
18:15
<AryehGregor>
(well, not exactly, but an IDL for CSSOM depends on one from HTML and vice versa)
18:18
<Philip`>
Solution: put all web technologies into a single spec
18:18
<GPHemsley>
Is there a reason that WHATWG specs exclude <head> and <body> in their source?
18:19
<GPHemsley>
oh, wait, they
18:19
<GPHemsley>
're not WHATWG
18:19
<GPHemsley>
I'm looking specifically at this: http://dvcs.w3.org/hg/url/raw-file/tip/Overview.html
18:19
<AryehGregor>
New WebIDL question: is there any difference between "interface Foo {}; partial interface Foo { void foo(); };" and "interface Foo {}; [NoInterfaceObject] interface Bar { void foo(); }; Foo implements Bar;"? Except that the former doesn't work if you want multiple interfaces to get the member?
18:20
<AryehGregor>
GPHemsley, the <head> and <body> tags are optional in text/html. No reason to include them unless you want attributes on them, usually (there are some other possible reasons).
18:20
GPHemsley
never likes the "it's optional, so I leave it out" <s>excuse</s> explanation
18:21
<GPHemsley>
also, why not take advantage of <header>?
18:21
<zewt>
typically the burden is to explain why to include something that's optional, not why to exclude it
18:21
<GPHemsley>
(etc.)
18:22
<GPHemsley>
zewt: Something like the difference between <head> and <body> seems pretty fundamental to me
18:22
<GPHemsley>
structurally/mentally-speaking
18:22
<webben>
GPHemsley: Wot?
18:23
<GPHemsley>
"just because you CAN do something doesn't mean you SHOULD"
18:23
<webben>
GPHemsley: Can you describe what the advantage of including redundant syntax would be in this case?
18:23
<TabAtkins>
I assume we're talking about nested-microdata-in-head?
18:23
<GPHemsley>
webben: legibility, at least
18:24
<GPHemsley>
TabAtkins: No, something much less complex.
18:24
<webben>
GPHemsley: Legibility to whom?
18:24
<GPHemsley>
me :)
18:24
<webben>
GPHemsley: Why do you need to be able to read the spec's source?
18:25
<GPHemsley>
Shouldn't one have the ability to read any page's source...? Otherwise, why aren't the elements just named <1> <2> <3> etc.?
18:25
<Ms2ger>
GPHemsley, that file is generated, there's no point in optimizing its legibility
18:25
<zewt>
seems more like "i'm used to doing this so I think it's weird not to simply due to habit"
18:26
<GPHemsley>
zewt: I'm sure that's part of it, but I think it's more than that.
18:26
<webben>
GPHemsley: 1) No, why? 2) Because good names makes it easier to know how to use them.
18:27
<webben>
GPHemsley: Particularly, with respect to 1) if WHATWG should be optimising for anything you could argue it should be performance.
18:27
<GPHemsley>
Well, thinking out loud, what is the burden on the parser to have determine where the <head> stuff ends?
18:27
<GPHemsley>
heh
18:27
<webben>
GPHemsley: There's no special burden on the parser.
18:28
<GPHemsley>
that's interesting
18:28
<webben>
GPHemsley: It knows it's left <head> as soon it see a start tag that cannot exist in <head>.
18:29
<Ms2ger>
The parser has to handle this anyway, so that's not a particularly compelling argument
18:29
<GPHemsley>
So then the <head> and <body> tags are vestigial and irrelevant now?
18:29
<TabAtkins>
Basically, yeah.
18:30
<GPHemsley>
interesting
18:30
<Ms2ger>
Well, now
18:30
<TabAtkins>
If by "now" you mean "since at least the last decade, and probably the decade before that".
18:30
<Ms2ger>
What TabAtkins said
18:30
<webben>
GPHemsley: I agree if you're trying to optimize for least surprise for fellow developers, then you might want to be explicit about implied elements.
18:30
<zewt>
since afaik no new tags can ever be added to <head> (because it would cause incompat between browsers that know the tag and those that don't), it sort of dead-ends <head> as a concept anyway
18:30
<webben>
GPHemsley: But that's not the only possible dimension around which to optimize (esp. production) source code.
18:31
<GPHemsley>
zewt: Yeah, I'd just read something to that effect.
18:31
<zewt>
leading to weirdness like people trying to wedge <intent> into <link>
18:31
<webben>
GPHemsley: (Also, in practice people generally don't remember all the implicit elements. Few people add implied TBODY for example.
18:32
<zewt>
also, having less boilerplate for simple documents is an overall win
18:32
<webben>
GPHemsley: Also, people generally don't complain when JS and CSS get minified.
18:32
<GPHemsley>
ack, ok
18:32
<GPHemsley>
sheesh
18:32
<zewt>
webben: i despise minifiers
18:32
<webben>
(This does have an effect on the web of course - less learning by C+P. But everything involves a tradeoff.)
18:32
<zewt>
they need to die in a ditch (on fire)
18:32
<webben>
zewt: "generally" ;)
18:33
<zewt>
(nothing's worse than trying to greasemonkey around a dumb site bug, to find that everything's a mess because somebody doesn't believe in deflate)
18:33
<TabAtkins>
Heh, I hate minifiers too. Squeezing a few extra bytes out (most of which will be removed by compression anyway) at the near-total cost of readability is a bad tradeoff to me.
18:33
<GPHemsley>
yeah... I think the discussion really comes down to who you're optimizing for: a human or a parser
18:33
<webben>
zewt: People are right not to believe in deflate - because of misconfigured proxies etc a lot of people won't get compressed content.
18:33
<zewt>
it's basically tossing the benefits of a textual format down the toilet
18:34
<webben>
Not really. GM has to interact with the DOM, and the DOM is implied.
18:35
<webben>
(I'm not in favour of minification that alters class names/ids or whatever.)
18:35
<webben>
i.e. I'm not in favour of minification that breaks the API.
18:35
<zewt>
i have to interact with the page source, in order to write a script in the first place
18:35
<webben>
zewt: GM interacts with the DOM not the raw markup, doesn't it?
18:36
<zewt>
yes, but a human (me) has to write the greasemonkey script in the first place, which requires understanding what the page is doing
18:36
<webben>
yes, but that requires looking at the DOM not the source.
18:36
<zewt>
eg. minification doubles as obfuscation (not always unintentionally)
18:36
<webben>
if you're talking about JS, that's a different story.
18:37
<zewt>
looking at the DOM doesn't help me read minifuscated javascript source
18:37
<TabAtkins>
I super-abhore packed JS, though luckily that's much less common these days.
18:37
<Ms2ger>
Is it?
18:37
<webben>
there's a case for preserving names in JS minification so when you re-beautify it's easier to read.
18:38
<webben>
I'm not sure whether the JS constitutes part of the public API of the page or not.
18:38
<zewt>
... most pages don't have a "public API"
18:38
<TabAtkins>
Ms2ger: I used to see it a lot when viewing people's source. Now I very rarely see it.
18:38
<webben>
zewt: Depends what you mean.
18:38
<TabAtkins>
Ms2ger: Also, it doesn't seem to figure into people's tutorials these days.
18:39
<webben>
zewt: I think a lot (most? probably not) of pages do have a public API.
18:39
<zewt>
if a page has an obnoxious javascript tooltip fade-in, and I want to remove it or make it not fade, i can't say I've ever seen a single page where that would be considered "public API"
18:40
GPHemsley
silently backs away
18:41
<webben>
e.g. use of odd bits of semantic HTML, attempts to make pages search-engine/AT friendly
18:41
<webben>
semantic naming
18:42
<webben>
zewt: Yeah. I don't think people generally see JS effects as part of the API.
18:42
<webben>
rather than an implementation detail
18:42
<zewt>
most of the JS implementation of pages is just considered internals
18:42
<webben>
exactly
18:42
<zewt>
and that's usually the sort of thing I end up GM'ing
18:43
<zewt>
i need to write a script to make twitter stop swallowing browser-owned keystrokes. heh
18:43
<webben>
I think the solution there is better HTML and CSS features so more of this is declarative rather than imperative.
18:43
<webben>
declarative and overrideable
18:43
<zewt>
it's not really okay to preventDefault on alt-t
18:43
<webben>
And more fine-grained UA control over what sites can do
18:44
<zewt>
webben: not holding my breath
18:44
<webben>
Sure.
18:44
<webben>
Probably wise ;)
18:44
<zewt>
at least twitter isn't eating alt-d anymore, which iirc it used to
18:45
<webben>
We do sometimes discuss better features for binding keyboard shortcuts to commands so that UA features can remap user actions to commands.
18:45
<webben>
I think TV Raman filed some bugs along these lines
18:46
<webben>
although I have feeling Hixie is punting to HTML.Next
18:46
<zewt>
twitter's oddly been the only notable site I've ever seen that breaks basic browser hotkeys
18:46
<webben>
Twitter's web frontend leaves much to be desired.
18:46
<zewt>
it's a lot better than most high-profile sites
18:46
<webben>
e.g. hash-bang URL nonsense
18:46
<zewt>
nothing wrong with that (get over it :)
18:46
<zewt>
well, caveat
18:46
<webben>
(No! ;) )
18:47
<zewt>
last I looked at that closely, it wasn't particularly optimized and regularly took many extra round-trips for a simple single-message page; that's dumb
18:47
<webben>
the history api allows you to set arbitrary URLs
18:48
<zewt>
sure, once it's widely-deployed, which it isn't quite yet
18:48
<webben>
Yeah, I think it's better to use it where available and use normal URLs where it's not.
18:48
<zewt>
(not in IE9)
18:48
<webben>
yeah it being left out of IE9 is regrettable
18:51
<zewt>
personally i find the ability to control history entries (replace vs. push) in history api to be much more interesting
18:51
<zewt>
allowing storing more fine-grained state in the URL without spamming history
18:52
<zewt>
(as well as things that can't be stored in the URL, like open files, though nobody implements that yet)
18:52
<zewt>
(afaik)
19:20
<zewt>
annevk: does encoding *to* legacy encodings need to be specified? i'm not even sure how that works for POST data
19:20
<annevk>
yes
19:20
<annevk>
for <form>
19:20
<annevk>
for URL
19:20
<zewt>
if I POST CJK through a shift-jis page the data is urlescaped Shift-JIS; if I POST CJK through iso-8859-1, the data is UTF-8 (wtf?)
19:21
<annevk>
POST CJK how?
19:21
<annevk>
using accept-charset?
19:21
<zewt>
enter CJK into a form in a page that was served as shift-jis
19:21
<zewt>
nothing special
19:22
<annevk>
that's some interesting behavior for windows-1252
19:22
<annevk>
I don't think that's correct per HTML
19:22
<zewt>
er
19:22
<annevk>
however I don't plan to define how to deal with out of range characters
19:23
<zewt>
sorry, apache is fucking with me
19:23
<annevk>
I just plan to define how to go from in-range characters to octets
19:23
<annevk>
if you have out-of-range characters you will need to do some pre-processing
19:23
<zewt>
it's taking what I'm giving to AddType and ... changing it, apparently
19:23
<annevk>
I think <form> does turns them into ? or some such
19:24
<zewt>
yeah, i've never heard of a standard placeholder like U-FFFD in legacy encodings
19:24
<zewt>
afk
19:56
<gsnedders>
annevk: I thought HTML5 already defined it for form, at lesat
19:56
<annevk>
yeah
19:58
<annevk>
"For each character in the entry's name and value that cannot be expressed using the selected character encoding, replace the character by a string consisting of a U+0026 AMPERSAND character (&), a U+0023 NUMBER SIGN character (#), one or more characters in the range U+0030 DIGIT ZERO (0) to U+0039 DIGIT NINE (9) representing the Unicode code point of the character in base ten, and finally a U+003B SEMICOLON character (;)."
19:58
<annevk>
and after that it encodes
19:58
<annevk>
that's exactly the kind of pre-processing I meant :)
20:31
<annevk>
AryehGregor: will look in IE tomorrow unless someone beats me to it
20:31
<AryehGregor>
annevk, thanks.
20:31
AryehGregor
has an AWS VM somewhere he can use, but is too tired to remember how right now
21:32
<TabAtkins>
Dammit, I can't find my list of necessary commands for webkit hacking. ;_;
21:33
<TabAtkins>
That thing took a few hours of wiki-diving and messing around (and two people helping me out) to accumulate.
22:10
<zewt>
<div style="display: block;" hidden>foo</div> is rendered? :|
22:10
<Ms2ger>
Yeah
22:10
<zewt>
lame
22:10
<Ms2ger>
style attributes override the UA style sheet
22:11
<TabAtkins>
Everything overrides the UA style sheet.
22:11
<zewt>
hidden should override display; that's the entire point, as far as I've ever used it
22:11
<TabAtkins>
Solution: put "[hidden] { display: none !important; }" in your stylesheet.
22:11
<zewt>
(to show/hide elements without having to stash and restore the original display style)
22:13
<zewt>
yeah, but doing that manually sort of partially defeats the idea of having it as a standard feature to begin with
22:13
<TabAtkins>
Kinda, yeah. Unfortunately, CSS doesn't define a notion of UA-important.
22:14
<TabAtkins>
Which we *should* do, because everyone that uses an explicit UA stylesheet implements it, and in the same way.
22:14
<zewt>
and it'd be especially annoying if some bits of a page ended up depending on the behavior of hidden *without* that CSS
23:22
<bencc>
is it possible that a browser will send only part of a websocket package even if the fin bit is 1?
23:22
<bencc>
maybe if the user closed the browser window