00:14
<GPHemsley>
Argh, Apache/DreamHost. When I say "unset Content-Header, I do NOT mean "send Content-Type: text/plain"
00:14
<GPHemsley>
+"
00:19
<GPHemsley>
Firefox, Chrome, and Opera sniff PDFs; Safari doesn't
00:19
<GPHemsley>
at least, not with just '%PDF-'
00:24
<GPHemsley>
ARGH.
00:24
<GPHemsley>
This testing is going to be impossible if the server insists on ignoring my directives.
00:28
<zewt>
sure would be great if firefox's options ui wasn't in the dark ages
00:28
<zewt>
spending more than a minute looking through every screen searching for the cookie display heh
00:30
<zewt>
what the hell? they changed the cookie button to something that's not a button at all, which makes it unfindable since i was looking at the buttons, not things that look like links
00:30
<zewt>
and serious question: how is it 2012, and popunders are still possible
00:31
<zewt>
they're easily the single most user-hostile abusive things that websites do, and i'm pretty sure there are literally 0 legitimate non-abusive uses for doing it
00:33
<zewt>
wow, seriously: firefox's UI for opening the cookies display is totally different depending on an unrelated option
00:34
<zewt>
(which is why I couldn't find it--I was in a clean profile, with that unrelated option set to something else)
00:34
GPHemsley
mumbles something about stupid .htaccess rules.
00:37
<GPHemsley>
ah, 500 error, even better
00:44
<TabAtkins>
Anyone know Hg well enough to help me unwedge my repo, before I just nuke it from orbit?
00:44
<TabAtkins>
I didn't pull before I started editting, and the commits hanging in the ether conflicted with the local changes I just made to a file. Merge conflict, obviously.
00:45
<TabAtkins>
I try to hg merge, then hg resolve -am. Now, if I try to hg merge, it tells me I can't because I have uncommited changes. If I try to hg commit, it tells me I can't partially commit a merge.
00:46
<TabAtkins>
Ah, got myself unwedged.
00:47
<zewt>
that's what i like about plain old svn: i just never hit stupid weird things like that
00:47
<TabAtkins>
I never hit stupid things like that in Git, either. Hg is just slightly worse at everything.
00:54
<roc>
it is possible to get confused in git
00:55
<TabAtkins>
Yo, roc, am I being dumb in the thread with Dirk, or is he just confused?
00:55
<roc>
of course git advocates will say "oh, you just do <MAGIC> to get out of that"
00:55
<roc>
TabAtkins: you're perfectly clear to me
00:56
<roc>
I think he's confused but it could just be monumental miscommunication
00:58
<TabAtkins>
Usually with Dirk it's the latter, sprinkled with a little bit of confusion.
00:58
<TabAtkins>
But I feel better now.
01:56
<TabAtkins>
Hixie: Got some initial feedback from Gregg.
01:57
<TabAtkins>
He says that WebGL people are moving to prefer contexts being separable from canvases, so that you can render to multiple canvases without having to duplicate all the state.
01:57
<Hixie>
yeah, i saw them talking about that on the gl list
01:57
<TabAtkins>
"var x = new WebGLRenderingContext(...); canvas.bindDrawingBuff(x); canvas2.bindDrawingBuff(x);"
01:57
<Hixie>
i think that if they do that the way it sounds like they are going to do that, it'll work fine with this too
01:58
<Hixie>
(they just need to get the CanvasProxy objects into th worker with the WebGLRenderingContext() object)
01:58
<TabAtkins>
Yeah, then the context can be constructed, either on the worker or in the document and sent to the worker, and then bound to the canvas by the document.
01:58
<Hixie>
the way i'm currently going you'd send the proxy to the worker and the context would be created in the worker
01:58
<Hixie>
since you can't transfer contexts
01:59
<TabAtkins>
Unfortunately, he didn't directly respond about context transfer across process boundaries. I'll press him on that.
02:03
<TabAtkins>
Hixie: Were you thinking that you'd always have to call commit(), or that the browser would normally automatically shuttle things back and forth, and commit() would be a helpful reminder for "I'm done for this frame, go ahead and take it".
02:03
<Hixie>
that you'd need commit(), i think
02:04
<Hixie>
we could imply commit if you spin the event loop in the worker
02:04
<TabAtkins>
Yeah, Greg was talking about putting rAF in a worker.
02:04
<TabAtkins>
Which would accomplish the event-loop spin *and* let you safely modulate your frame rate.
02:04
<Hixie>
definitely need to do that, yes
02:04
<TabAtkins>
Cool.
02:11
<TabAtkins>
Hixie: Did you mean for the CanvasProxy, once neutered, to throw when you read width/height?
02:12
<Hixie>
pretty much everything on the proxy will need to throw once neutered, yeah
02:15
<TabAtkins>
That's annoying, since it neuters as soon as you generate a context from it. That means you can no longer use the CanvasProxy to get at the width/height of your drawing surface.
02:15
<TabAtkins>
I don't see why the CanvasProxy should throw on reads of width/height.
02:16
<Hixie>
oh i'm planning on changing that
02:16
<Hixie>
creating a context shouldn't neuter it
02:17
<Hixie>
sorry that sketch is out of date
02:18
<TabAtkins>
Ah, kk.
02:23
<Hixie>
TabAtkins: (in particular, i'm using the Proxy now as a way to get the image from the canvas in the worker)
02:24
<TabAtkins>
Hm, kk. I'm interested to see what you end up with when you're finished.
02:26
<Hixie>
you and me both :-)
08:16
<annevk>
terrorists win? :p
08:22
<Ms2ger>
You win?
08:32
<MikeSmith>
after re-listening to that conversation I think now that Roy was referring to browser-engine projects collectively when he said that
08:32
<MikeSmith>
not to the whatwg
08:32
<MikeSmith>
but hey it works great either way
08:56
<hsivonen>
how healthy to see the producers of products that are complementary to your own products as “terrorists”
08:56
<hsivonen>
not even optionally complementary but necessary complements for your product to have demand
08:57
<hsivonen>
browser people aren’t particularly happy about how Apache has handled a couple of config issues, but at least we’ve managed to avoid terrorist comparisons
08:59
<Ms2ger>
... in public
09:04
<MikeSmith>
if it was the ruler of the nginx project and he called me a terrorist, that would hurt a lot more
09:09
<hsivonen>
looks like my Rust efforts broke frameset-ok.
09:09
<hsivonen>
oops.
09:09
<hsivonen>
good thing I didn’t land this stuff yet
09:10
<hsivonen>
refactoring is hard.
09:17
<annevk>
TabAtkins: is there an online tool for those railroad diagrams?
09:18
<annevk>
TabAtkins: URL wants to use
09:20
<MikeSmith>
annevk: which railroad diagrams*
09:20
<annevk>
MikeSmith: if you heard something the URL Standard can address please let me know
09:20
<annevk>
MikeSmith: http://dev.w3.org/csswg/css3-syntax/#token-diagrams
09:20
<MikeSmith>
didn't hear anything
09:20
<MikeSmith>
oh nice
09:21
<hsivonen>
good luck using diagrams like that in an RFC
09:21
<annevk>
I listened to it once and I missed a few words, but I did not get the impression there was anything actionable there
09:21
<annevk>
hsivonen: heh
09:26
<MikeSmith>
annevk: wait sorry I think I still made a mistake
09:26
<hsivonen>
the way they think about naming vs. versioning at the IETF is fascinating
09:28
<MikeSmith>
annevk: I listened to it again really carefully and now I'm pretty sure what he really said is, "Anne's a rhyme terrorist at the microphone, Hixie drops the beats like a predator drone."
09:29
<hsivonen>
It takes a correct observation that if you have an old API entry points and you want and you incompatible new API entry point, the solution is to give the new one a different name. But then it confuses this with non-operational naming and definitional incompatibility (as opposed to actual operational incompatibility).
09:30
<hsivonen>
s/and you and you/and you want to have/
09:32
<hsivonen>
also ignores that the operational names that take the thing annevk is speccing are already called “url()” and the like
09:32
<MikeSmith>
annevk: https://github.com/tabatkins/railroad-diagrams
09:34
<annevk>
ah cool
09:38
<hsivonen>
I feel like a blog post coming up, but must avoid 386.
09:39
<annevk>
hsivonen: I wrote http://lists.w3.org/Archives/Public/uri/2012Nov/0015.html but that's mostly technical
09:39
<annevk>
hsivonen: I'd be interested in a blog post on the manner though if it doesn't take up too much of your time
09:41
<annevk>
(posted a short TPAC note on text/css btw; might do another one later on CORS)
09:44
<hsivonen>
annevk: cool. I’ll update my Gecko patch.
09:45
<zcorpan>
GPHemsley: if you're tying to test invalid http responses with apache, you're gonna have a bad time
09:45
<zcorpan>
GPHemsley: don't use apache
09:46
<zcorpan>
GPHemsley: use a custom server where you can control the bytes
09:48
<Ms2ger>
GPHemsley, https://bitbucket.org/annevk/simpleserver perhaps
09:48
<hsivonen>
zcorpan: is your CSS BOM test suite available under a permissive license?
09:50
<zcorpan>
hmm, wonder if a license is applied when submitting to csswg
09:50
<jgraham>
I think that all of W3C uses testsuite license + BSD
09:51
<hsivonen>
ok.
09:51
<hsivonen>
zcorpan: did you submit it already? where do I find the submission?
09:51
<zcorpan>
https://test.csswg.org/source/contributors/opera/submitted/css3-syntax/charset/
09:52
<zcorpan>
i was told that's the proper home for css tests
09:52
<hsivonen>
zcorpan: thanks
09:54
<zcorpan>
"The tests are intended to be released under both the W3C Document License and the BSD 3-clause license, so unless you represent a W3C Member, you must give your explicit permission for us to use your contributions under these licenses." http://wiki.csswg.org/test/css2.1/contribute
09:54
<hsivonen>
zcorpan: what’s the expected header config for .bogus.css files?
09:54
<jgraham>
I thought it was 2 clause
09:55
<jgraham>
Oh, I was wrong
09:55
<zcorpan>
hsivonen: clone http://hg.csswg.org/test
09:56
<zcorpan>
or http://hg.csswg.org/test/file/7feb6727108c/contributors/opera/submitted/css3-syntax/charset/.htaccess
09:57
<zcorpan>
confusing that it shows a completely unrelated commit message
09:58
<hsivonen>
zcorpan: thanks
10:14
<MikeSmith>
hsivonen: is there any reason for not adding ITS validation support to the default preset at this point?
10:15
<MikeSmith>
if there is none, I'll go ahead and add it
10:28
<hsivonen>
MikeSmith: IIRC, Jirka had a reason to wait for a bit.
10:30
<hsivonen>
MikeSmith: as for making it default, I think ITS would be mostly harmless functionally as part of the default but making a non-browser-targeted extension part of the default is a bit of a slippery slope as far as setting a precedent goes
10:31
<MikeSmith>
I see
10:31
<MikeSmith>
i'll go back and re-read the thread with Jirka
10:32
<MikeSmith>
making it not part of the default has the downside of further complicating the list of presets
10:32
<hsivonen>
MikeSmith: yeah, the list of presets is not great
10:32
<hsivonen>
MikeSmith: identifying presets by URLs rather than some persistent symbols is not great in practice
10:33
<hsivonen>
but hey, the URL thing is right in principle!
10:33
<MikeSmith>
hah
10:34
<hsivonen>
I heard ITS described as an enterprise feature at TPAC.
10:48
<MikeSmith>
hsivonen: hmm that's the kiss of death
10:48
<MikeSmith>
I hope the ITS advocates don't use that term themselves if they want to have people view ITS positively
10:50
<hsivonen>
MikeSmith: it wasn’t ITS advocacy :-)
10:51
<MikeSmith>
ok
10:51
<MikeSmith>
"enterprise" is like "terrosist"
10:51
<MikeSmith>
except spelled correctly
11:03
<annevk>
hsivonen: only URIs are correct in principle
12:01
<annevk>
slightlyoff: meant to talk to you about the URL API last week, but for some reason it didn't happen
12:01
<annevk>
but twitter works
12:02
<jgraham>
API design via twitter? What could possibly go wrong?
12:02
<annevk>
@__proto__, right? :)
12:08
<smaug____>
s/API design//
12:08
<annevk>
but don't worry, it's going to a list ;)
12:15
<zcorpan>
jgraham: at least we don't have to worry about overly long names
12:18
<jgraham>
Heh. But I don't think that Hixie is so influential that we need to worry about annevk making a DeferredURLProxyFactoryFactory yet
12:20
odinho
h8 .stopPropagation .cancelDefault .stopImmediatePropagation and whatstheirnames
12:21
<odinho>
preventDefault it was I think, omg, always have to look that up
12:23
<annevk>
defaultPrevented?
12:23
<annevk>
oh the method
12:25
<smaug____>
naming is inconsistent. events can be cancelable, in which case preventDefault() isn't no-op
12:25
<smaug____>
but too late to change that all
12:26
<odinho>
Sugar Spec, -- just sayin'.
12:32
<vargadanis>
is a PHP version os the lib lagging so much behind the pythong version? 3 years?
12:34
<annevk>
vargadanis: the person working on PHP html5lib stopped working on it
13:27
<annevk>
http://www.sitepoint.com/have-you-considered-polyglot-markup/
13:27
<annevk>
"You want better quality. This goes along with the first item. Also, “application/xhtml+xml” can signify quality."
13:28
<annevk>
:/
13:29
<Ralt>
hello people
13:29
<Ralt>
should I trust WHATWG living DOM standard or W3C DOM3 standard?
13:30
<MikeSmith>
annevk: I think Sitepoint meant to publish that article in their Onion section
13:31
<annevk>
Ralt: DOM3 is obsolete
13:31
<annevk>
Ralt: apart from DOM3 Events I suppose, that lingers on
13:31
<Ralt>
can't find DOM4 :(
13:31
<annevk>
Ralt: it's somewhere, but also not as up to date as the DOM Standard (and it's a fork of that anyway)
13:32
<Ralt>
hm, got it
13:32
<Ralt>
there are some differences though
13:32
<annevk>
?
13:32
<Ralt>
ah no
13:32
<Ralt>
it's ok, my bad
13:33
<Ralt>
hm, there isn't supposed to be any live NodeList anymore then, right?
13:34
<Ralt>
since the only method returning NodeList is QSA
13:35
<zcorpan>
readonly attribute NodeList childNodes;
13:37
<annevk>
Ralt: there's some dispute over getElementsByClassName as to whether or not they should return HTMLCollection, though maybe that's settled by now
13:38
<Ralt>
annevk: in both specs, they return HTMLCollections (living & DOM4), so I guess it's settled?
13:38
<annevk>
Ralt: W3C DOM is just a copy; settled refers to potential ongoing debate
13:39
<annevk>
Ralt: I haven't looked into that particular issue in a while
13:39
annevk
is fixing mutation observers
13:39
<annevk>
(out of date copy*)
13:39
<Ralt>
so W3C DOM4 just merges DOM living standard from time to time?
13:40
<annevk>
afaik that's the plan
13:40
<Ralt>
alright, thanks :)
13:41
<Ralt>
zcorpan: ah! thanks
13:42
<hsivonen>
annevk: Polyglot will supply 386 for the next decade.
13:42
<annevk>
Ralt: if you're an implementor I suggest using the WHATWG copy, even Microsoft does so
13:43
<annevk>
Ralt: if you're a lawyer I suggest using the W3C copy
13:43
<Ralt>
haha ok
13:43
<annevk>
hsivonen: hahaha
13:43
<annevk>
hsivonen: also tragedy
13:53
<hsivonen>
unsurprisingly, implementing the new CSS charset stuff fails a test called test-charset-utf-16-le-no-bom.html
13:53
<hsivonen>
also be
13:54
<hsivonen>
uh oh. looks like those are part of an official test suite
13:55
<hsivonen>
when I seach for “test-charset-utf-16-le-no-bom.html” on DDG, I get a Microsoft ad saying that IE is better now.
13:56
<zcorpan>
is ie better at bomless utf-16 now?
13:56
<jgraham>
That's a super targetted ad ;)
13:58
<zcorpan>
microsoft went "gotta convince hsivonen that ie doesn't suck these days. what do we do?"
13:59
<hsivonen>
how does the CSS WG deal with Level 3 invalidating a test in the CSS 2.1 test suite?
13:59
<MikeSmith>
take 2.1 back to CR!
14:00
<jgraham>
You might hope that there was one testsuite to rule them all
14:00
<zcorpan>
http://w3cmemes.tumblr.com/post/31865121758/the-joker-shares-his-approach-on-css2-1-issues
14:03
<hsivonen>
sigh. I don’t like test cases that I have to look in a hex editor to see what they are doing.
14:03
<zcorpan>
hsivonen: maybe reply to http://lists.w3.org/Archives/Public/www-style/2012Oct/0870.html ? (or an earlier email in the thread)
14:03
<zcorpan>
hsivonen: are my tests like that?
14:04
<hsivonen>
zcorpan: I trust your tests do what they say on the tin.
14:04
<zcorpan>
hsivonen: i got little-endian and big-endian wrong at first, FYI :-)
14:05
<hsivonen>
now I’m unhappy about web-sniffer.net
14:06
<hsivonen>
the test case I’m looking at tests what it says but web-sniffer.net drops 0x00
14:08
<hsivonen>
aaargh. someone at Microsoft though about evil test cases for CSS 2.1 syndata but instead of concluding that the spec didn’t make sense, the WG put the tests in the test suite
14:09
<hsivonen>
once you have a test case like http://test.csswg.org/suites/css2.1/20110111/html4/at-charset-045.htm in hand, aren’t you supposed to conclude the spec is crazy?
14:13
<Lachy>
hsivonen, does the spec say that @charset takes precedence over the BOM?
14:13
<hsivonen>
interesting https://bugzilla.mozilla.org/show_bug.cgi?id=462458#c0
14:14
<hsivonen>
Lachy: yes! http://www.w3.org/TR/CSS21/syndata.html#charset
14:14
<hsivonen>
Lachy: Level 3 fixes this.
14:20
<Lachy>
wow, that's crazy.
14:26
GPHemsley
wonders if Google has any Hey GURL memes lying around.
14:30
<GPHemsley>
Interestingly, Chrome and Opera seem to also store the appropriate file extension when doing sniffing.
14:31
<GPHemsley>
If you load a resource with a file type that they don't support, they will offer to download it—with a file extension appended.
14:31
<GPHemsley>
(My tests don't use file extensions.)
14:31
<GPHemsley>
I wonder if that should be specced?
14:34
<annevk>
whereis smaug?
14:36
<zcorpan>
GPHemsley: that seems like a platform-specific thing and also something that doesn't get exposed to web content directly so we don't need interop
14:37
<GPHemsley>
zcorpan: Ah, OK.
14:40
<hsivonen>
zcorpan: can you explain https://test.csswg.org/source/contributors/opera/submitted/css3-syntax/charset/page-windows-1251-css-utf8-bom.html ?
14:40
<hsivonen>
the id is the REPLACEMENT CHARACTER but the selector is not bogus UTF-8
14:44
<zcorpan>
hsivonen: it was intended to be bogus utf-8. seems i forgot to remove the 80 byte when switching from 80 to c8 as the "interesting byte"
14:44
<zcorpan>
will fix
14:45
<hsivonen>
also, https://test.csswg.org/source/contributors/opera/submitted/css3-syntax/charset/page-windows-1251-charset-attribute-bogus.html seems to be off
14:45
<hsivonen>
it has a C8 byte but the test is written for E8 byte
14:47
<hsivonen>
zcorpan: https://test.csswg.org/source/contributors/opera/submitted/css3-syntax/charset/page-windows-1251-css-at-charset-bogus.html fails because the .css file is 404
14:47
<zcorpan>
fixed the first thing
14:48
<hsivonen>
zcorpan: https://test.csswg.org/source/contributors/opera/submitted/css3-syntax/charset/page-windows-1251-css-http-bogus.html also has the C8 vs. E8 thing
14:49
<hsivonen>
\o/ the UTF-8 one passes now
14:50
<zcorpan>
man, case folding :-(
14:52
<hsivonen>
https://test.csswg.org/source/contributors/opera/submitted/css3-syntax/charset/page-utf16-css-bomless-utf16be.html fails, but that’s because the decoder sniffs endianness when the BOM is missing and the test assumes that not to happen, right?
14:55
<GPHemsley>
Is there a convention for naming tests?
14:56
<zcorpan>
hsivonen: fixed c8 vs e8
14:59
<zcorpan>
note to self: use data:text/html;charset=windows-1251,%c8 and javascript:alert(document.body.textContent.charCodeAt(0).toString(16)) instead of looking up the character with a google search
15:01
<zcorpan>
hsivonen: fixed the 404
15:01
<zcorpan>
hsivonen: thanks
15:02
<hsivonen>
zcorpan: thanks
15:02
<hsivonen>
can you explain https://test.csswg.org/source/contributors/opera/submitted/css3-syntax/charset/page-windows-1251-css-at-charset-windows-1250-in-utf16.html ?
15:03
<hsivonen>
AFAICT, the windows-1251 interpretation of the style rule is the one that should match
15:03
<hsivonen>
but the test does not want it to match
15:04
<annevk>
no smaug today?
15:04
<hsivonen>
https://twitter.com/ronsman/status/266538463262801921
15:05
<zcorpan>
hsivonen: you're right
15:05
<annevk>
http://html5.org/temp/mo-queue.html was hard to write :(
15:08
<zcorpan>
hsivonen: fixed. (also utf16be)
15:08
<MikeSmith>
hsivonen: the guy wants you tell him in 500 to 1000 words how to acquire clue?
15:10
<zcorpan>
hsivonen: ask in #css-test on irc.w3.org if you want push access
15:11
<zcorpan>
hsivonen: i have to leave now
15:15
<annevk>
so the queue thing
15:15
<annevk>
each time you modify a DOM node
15:15
<annevk>
that runs
15:15
<annevk>
well, if you have some listeners set up
15:15
<annevk>
shit's expensive
15:16
<miketaylr>
http://dvcs.w3.org/hg/webevents/raw-file/default/gamepad.html :|
15:17
<miketaylr>
"error: gamepad.html@cde6c06f4579: not found in manifest"
15:18
hsivonen
reviews test cases by implementing the spec and then running the tests
15:24
<annevk>
IDNA disallows http://💩.la too
15:24
<annevk>
IDNA2008
15:25
<hsivonen>
huh. no Ms2ger
15:41
<GPHemsley>
annevk: Whoever's in charge of that site is bad for the Web.
15:42
<annevk>
yeah, better invalidate his domain name
15:42
<annevk>
o_O
15:42
<GPHemsley>
Independent of his domain name.
15:42
<GPHemsley>
He uses webkit-only CSS
15:42
<annevk>
they don't really seem to see it as a problem either, invalidating domain names
15:43
<annevk>
so sad
15:43
<gavinp>
GPHemsley: may I have wiki user rights?
15:43
<gavinp>
Yesterday annevk had to create an entry for me, which is fine, but it would be nice to do that myself.
15:43
<GPHemsley>
gavinp: What kind? I already added you to the autoconfirmed list.
15:43
<gavinp>
ok, great.
15:43
<GPHemsley>
gavinp: You should be able to create pages and all that good stuff. :)
15:44
<gavinp>
Right now I cannot create pages
15:44
<gavinp>
should i log out and back in or something?
15:44
<GPHemsley>
hmm
15:44
<GPHemsley>
couldn't hurt
15:44
<gavinp>
and that did it.
15:44
<gavinp>
thanks.
15:44
<GPHemsley>
Ah, good.
15:44
<gavinp>
next step was to reboot my windows machine.
15:44
<GPHemsley>
heh
15:44
<gavinp>
'k. You're awesome. Thank you.
15:44
<GPHemsley>
:)
15:45
<gavinp>
I am going to assume the ~18hrs of silence that my prerender events proposal has received is widespread unanimous consent from all implementers and developers.
15:45
<gavinp>
and proceed on that basis.
15:45
<GPHemsley>
gavinp: Thank you, too. Your request for rights uncovered a bug with permissions to give those rights. :)
15:46
<annevk>
gavinp: http://en.wikipedia.org/wiki/Warnock's_dilemma (though I suspect most of the other browsers are just not doing prerendering at all yet)
15:47
<gavinp>
afaik we are the only one, and partly for architectural reasons.
15:48
<gavinp>
And I think the first rev of the API wasn't as useful for sites as it could be. So hopefully this kind of incremental change is working.
15:48
<gavinp>
I was very happy to just use element removal as an API signal to kill prerenders; that feels very natural and doesn't increase the scripting language surface area at all.
15:55
<SimonSapin>
Is it necessary or useful to escape > in HTML text?
16:58
<dglazkov>
good morning, Whatwg!
17:00
<Ms2ger>
Good night
17:05
<TabAtkins>
annevk: I should definitely make an online tool for them, but as MikeSmith already pointed to the repo, you can probably do it yourself. ^_^
17:11
<TabAtkins>
hsivonen: If level 3 changes 2.1 behavior in a way that would invalidate a 2.1 test, that's fine. You can just claim conformance with the higher spec. If it bothers you, you can submit a revised version of the 2.1 test; if that's not possible, you can ask for it to be undefined in 2.1 with a pointer to level 3. But I wouldnt' care about it.
17:11
<TabAtkins>
hsivonen: As long as there's no BOM, if you're using sublime text you can just go File->Reopen with Encoding and choose Hexadecimal.
17:12
<TabAtkins>
hsivonen: For a hex editor, that is.
17:13
<TabAtkins>
SimonSapin: No need to escape > in raw text or in a quoted attribute. Escape it otherwise.
17:14
<TabAtkins>
(Of course, you shouldn't have > in an element or attribute name in the first place, and if your attribute value is outside alphanumeric range, you should be quoting it.)
17:15
<TabAtkins>
So in practice, no, never escape >. The only things you need to escape are < (always), & (most of the time, might as well do it for safety), and your quoting character inside an attribute value.
17:16
<jgraham>
M-x hexl-mode
17:16
<jgraham>
But it's still annoying
17:16
<SimonSapin>
TabAtkins: great, thanks
17:23
<bLh>
Ello fellas, i`ve few questions about wc3 web sql database. I`ve created mobile app with PhoneGap, which uses Web sql database, since it`s not in active maintance anymore, i`m worried what will happen with newer versions of browsers(webkit)
17:23
<bLh>
did i made a big mistake ;|
17:26
<smaug____>
bLh: don't use Websql
17:26
<Ms2ger>
Yeah, you don't want to use websql
17:26
<smaug____>
use IndexedDB
17:28
<bLh>
Phonegap only supports websql ;[
17:29
<bLh>
the only alternative i had was key/value storage, which didnt really satisfied me at the point i was developing it
17:30
<TabAtkins>
I kinda doubt that webkit will be able to remove websql (for reasons like yours), but nobody else is going to add it, so your "mobile app" won't work on other mobile browsers.
17:32
<bLh>
yeah, i guess was a big mistake i`ve done, didnt research it enough at that point
17:32
<bLh>
i guess i made myself some extra work ;[
17:36
GPHemsley
wonders why he's the only one who ever has to merge branches with xref
17:37
<GPHemsley>
blame would be cooler if the lines were color-coded by age
17:48
<GPHemsley>
annevk: When you move the web-apps-tracker to github, did you set the site up to automatically pick up changes?
17:48
<GPHemsley>
+d
17:50
<Ms2ger>
GPHemsley, doesn't look like it, afaict
17:50
<GPHemsley>
Ms2ger: So committing to that repo will do nothing? :P
17:51
<Ms2ger>
GPHemsley, maybe the password is in the private repo ;)
17:53
<GPHemsley>
Ms2ger: If it is, then I don't know what I'm looking for.
17:54
<GPHemsley>
ow, my eyes! http://html5.org/tools/
17:55
<Ms2ger>
Hah
17:56
<Ms2ger>
Looks like an anneism
17:56
<TabAtkins>
Argh dammit
17:56
<TabAtkins>
jeezus
17:56
<TabAtkins>
whoever decided that 'lime' should be a named color should be shot.
17:57
<Ms2ger>
lime is nice for tests
17:57
<TabAtkins>
True. You should need to set a debug flag in order to use lime. ^_^
18:06
<MikeSmith>
GPHemsley: hahaha
18:07
<MikeSmith>
maybe the purpose is to get you to never visit that URL again but instead bookmark the stuff it links to
18:11
<paul_irish>
<h1 style="text-transform:uppercase">Title Case</h1> <== If I select and copy/paste that, what case should it be in?
18:12
GPHemsley
guess Title Case
18:12
<GPHemsley>
+es
18:12
<TabAtkins>
Pretty sure it's DOM-based.
18:13
<paul_irish>
Right now WebKit's behavior is uppercase though Title Case seems much more reasonable to me
18:13
<paul_irish>
https://bugs.webkit.org/show_bug.cgi?id=43202#c5
18:14
<paul_irish>
Let me confirm, but iirc Mozilla/IE do Title Case
18:18
<paul_irish>
Confirmed. Opera/IE/FF behavior is to copy the DOM text. WebKit behavior is to copy the transformed text.
18:18
<paul_irish>
What's the best way to resolve this? Does this belong in a spec?
18:19
<smaug____>
yes, but which spec :)
18:19
<smaug____>
someone started a spec for selection handling
18:19
<smaug____>
Ms2ger: maybe you
18:19
<Ms2ger>
Eh?
18:19
<Ms2ger>
Maybe AryehGregor
18:19
<smaug____>
ah, yeah, perhaps Aryeh
18:20
<Ms2ger>
Or, hmm, copy/paste may be hallvors
18:20
<divya>
yeah thats hallvors1
18:25
<paul_irish>
pinging hallvord. thanks
18:25
karlcow
votes for Title Case too.
18:26
<karlcow>
or at least having the option in the browser as a "Copy" (default), "Copy with style", "Copy Markup"
18:28
<karlcow>
Though I have seen users who loves to have the style and the links copied from a Web page into an html email. So It might depends on the users
18:29
<Ms2ger>
karlcow, style or markup? :)
18:32
<karlcow>
Ms2ger: I have the impression it is a combination of both.
18:32
karlcow
is trying something
18:39
<karlcow>
Ms2ger: http://www.la-grange.net/2012/11/08/copy-paste
18:40
<karlcow>
copy on the Web page in Safari
18:40
<karlcow>
Paste in Adium top left
18:40
<karlcow>
Paste in Text Mail top right
18:40
<karlcow>
Paste in Rich Text Mail in bottom right
18:41
<karlcow>
it depends on the paste context, but at least the rich text information is in the clipboard.
18:49
mhausenblas
waves to paul_irish
18:49
<mhausenblas>
thanks again a million for connecting me with Monsur - we're making good progress, see https://github.com/mhausenblas/enable-cors.org/issues?state=open
18:50
<paul_irish>
:D awesome
18:50
<mhausenblas>
yeah. you are, indeed! ;)
19:32
<MikeSmith>
eleven sparql specifications published today
19:32
<TabAtkins>
woooo
19:33
<MikeSmith>
making it up for it in volume
19:47
GPHemsley
hates it when he forgets to resolve a bug when he makes a comment about resolving it.
19:47
<GPHemsley>
IOW: Yay bugspam!
20:36
<Hixie>
what's the snapshot state for, in the File API?
20:42
<Hixie>
i gotta say, the File API spec has reached a point where it's pretty well written
20:42
<Hixie>
nice work
20:42
<Hixie>
(sicking: ^)
20:43
<sicking>
Hixie: all credit should go to Arun
20:43
<sicking>
Hixie: now the only problem is that I think FileReader uses the wrong model :)
20:44
<sicking>
Hixie: we should have had blob.readAsX methods instead, which return promise-like objects
20:46
<Hixie>
sicking: oh i wasn't talking about the design, just the actual spec writing :-)
20:46
<sicking>
Hixie: yeah, i know
20:46
<Hixie>
sicking: as the editor of the html spec, i understand only too well the difficulty of designing these things :-P
20:47
<Hixie>
(i don't really know of anything in the html spec that i wouldn't change in some way or other if we could start over)
21:52
<gsnedders>
HTML6?
21:52
<gsnedders>
Oh, wait, that was already done!
22:02
<TabAtkins>
Fun times: Fixing a parsing quirk to be strict, like FF and IE are, and finding that it causes a test to fail. The test has a note saying that it's making sure we're loose, because FF and IE are.
22:02
<TabAtkins>
Races to the bottom are great.
22:02
<TabAtkins>
(Even better when we can arrest them and go back up.)
22:04
<smaug____>
TabAtkins: how old is the test?
22:04
<TabAtkins>
smaug____: Haven't checked, but clearly old.
22:05
<smaug____>
since hsivonen's parser is quite different from the old Gecko parser
22:05
<TabAtkins>
This is a CSS parser thing.
22:05
<smaug____>
ah
22:05
<TabAtkins>
How to parse @charset.
23:17
<TabAtkins>
Anyone have an opinion on whether a <style scoped> scoped to the root element should be more specific than an unscoped stylesheet?
23:20
<TabAtkins>
(Rules scoped to a descendant element are more specific than one scoped to an ancestor.)
23:21
<divya>
what does more specific mean
23:22
<divya>
override general html {} rule?
23:38
<TabAtkins>
divya: I mean like specificity. This sets at a higher level than selector-specificity - if you're scoped at a lower element, you *automatically* win against rules scoped at a higher element.
23:38
<TabAtkins>
Just like !important rules auto-win against normal rules, regardless of the selector.
23:40
<divya>
TabAtkins: srry i am rusty on the jargon how would it look like in an e.g