| 05:23 | <zcorpan> | annevk: hi. do you have any good ideas on how to check https://www.w3.org/Bugs/Public/show_bug.cgi?id=16773#c3 ? |
| 05:28 | <annevk> | implementing HTML's encoding detection algorithm presumably |
| 05:28 | <zcorpan> | what does that give you? |
| 05:29 | <annevk> | well then you can reliably find all labels |
| 05:29 | <annevk> | if there's conflicts you can do an additional check |
| 05:29 | <zcorpan> | what we want to know (with this data) is whether supporting the label is better or worse than not supporting it |
| 05:29 | <zcorpan> | getting more reliable data might be good but is not what i'm trying to do right now :-) |
| 05:30 | <Hixie> | ship an update to 10% of your users that doesn't recognise the label, and instrument the charset menu to see how many more people on that build look at the charset menu to fix it |
| 05:31 | <annevk> | or decode the text using the two alternatives (stated and what the fallback would have been) and see what generates less U+FFFD |
| 05:31 | <annevk> | oh hey |
| 05:32 | <zcorpan> | i don't think either encoding would produce more U+FFFDs generally |
| 05:32 | <annevk> | the chairs implicitly acknowledged they were doing something weird with the innerHTML issue |
| 05:32 | <annevk> | zcorpan: yeah for single-byte encodings it's not gonna go that well I reckon |
| 05:32 | <annevk> | zcorpan: though for the "sjis" label it should |
| 05:32 | <annevk> | "sjis" or "windows-1252" is a world of difference |
| 05:32 | <zcorpan> | yeah |
| 05:35 | <zcorpan> | <meta http-equiv="Content-Type" content="text/html; charset=windows-1250; no-cashe; no-cache"> |
| 05:36 | <annevk> | hahahaaha |
| 05:36 | <annevk> | no cash, no cache, deal |
| 05:39 | <Hixie> | annevk: innerHTML issue? |
| 05:39 | <zcorpan> | good news: all cp1250 pages also had a <meta> decl with windows-1250, so it makes no difference as to whether cp1250 is supported (in this data) |
| 05:45 | <annevk> | Hixie: publishing DOM Parsing & ... through the W3C |
| 05:45 | <Hixie> | ah, yeah, i see |
| 05:45 | <Hixie> | why don't you just make a CG and publish an FSA? |
| 05:45 | <annevk> | because I'm not editing and Ms2ger presumably cares little |
| 05:45 | <Hixie> | fair enough :-) |
| 05:57 | zcorpan | realizes he can make the script look for later decls and check what the label is there |
| 05:58 | <annevk> | I should really write a presentation for later today, but I sort of want to hack this HTML encoding detection algorithm in Python :) |
| 06:02 | <annevk> | http://dev.w3.org/csswg/cssom-view/ #css on Freenode with a link to wiki.whatwg.org... |
| 06:03 | <zcorpan> | annevk: what should 'euc-cn' map to? |
| 06:03 | <annevk> | sounds like gbk |
| 06:04 | <annevk> | yeah, per Wikipedia that's gb2312 which is gbk |
| 06:04 | <zcorpan> | thanks |
| 06:15 | <hsivonen> | is gbk not a subset of gb18030? gb2313 is a subset of gb18030, right? |
| 06:16 | <annevk> | gb2313 is a subset of gbk |
| 06:16 | <annevk> | but in practice they are now treated identically |
| 06:17 | <annevk> | gbk is in turn a subset of gb18030 and these are still treated distinct, though some people think we should merge them too |
| 06:17 | <annevk> | why does IE not load a data URL like data:text/html;charset=utf-8,%F8%80%80%80%80 |
| 06:19 | <annevk> | hmm |
| 06:19 | <annevk> | zcorpan: why did you make HTML handle five and six byte sequences of utf-8 per old utf-8 rather than per new utf-8 |
| 06:20 | <annevk> | zcorpan: both WebKit and Trident give 5 U+FFFD for F8 80 80 80 80 80 rather than 1 which is correct per Unicode |
| 06:22 | <hsivonen> | meanwhile on xml-dev: reinventing one of the most failed part of the XML ecosystem: http://lists.xml.org/archives/xml-dev/201204/msg00023.html |
| 06:22 | <hsivonen> | *parts |
| 06:24 | <annevk> | are you subscribed? |
| 06:31 | <hsivonen> | annevk: yes |
| 06:42 | <annevk> | Hixie: you just followed zcorpan for utf-8 error handling right? |
| 06:44 | <Hixie> | i do not recall |
| 06:45 | <annevk> | k |
| 06:47 | <Hixie> | hey are there any other specs available under free licenses that i can get local snapshots of? it would make it easier for me to grep for things... here's what i have so far: http://damowmow.com/temp/free-specs/ |
| 06:47 | <annevk> | found https://www.w3.org/Bugs/Public/show_bug.cgi?id=9663 |
| 06:47 | <annevk> | okay, so I'll file bugs on Opera/Gecko to fix their utf-8 decoder |
| 06:47 | <annevk> | and fix it in the Encoding standard |
| 06:48 | <Hixie> | anyone else use trello btw? i've been playing with it as a way to organise bugs in small projects and it's really quite neat |
| 06:49 | <Hixie> | couldn't handle a load like HTML or a browser, but for small games and stuff it's great |
| 06:57 | <annevk> | what's the component in Gecko Core for utf-8 bugs? |
| 06:58 | <annevk> | I'm going with i18n |
| 07:02 | <annevk> | https://bugzilla.mozilla.org/show_bug.cgi?id=746900 |
| 07:02 | <annevk> | finding utf-8 decoder bugs in 2012, teehee |
| 07:20 | <nesta_> | good day ;) |
| 07:23 | <annevk> | hsivonen: I have filed no such bug at least |
| 07:26 | <hsivonen> | annevk: maybe I should file one |
| 07:59 | <zcorpan> | annevk: dunno, maybe it matched opera when i tested |
| 08:00 | <jgraham> | hsivonen: What is the current state of the spec wrt about:blank? |
| 08:00 | <annevk> | zcorpan: k, filed a bug on Opera and Gecko to fix their utf-8 decoder |
| 08:01 | <jgraham> | i.e. does it match what you think needs to be implemented |
| 08:01 | <jgraham> | ? |
| 08:01 | <jgraham> | Or what you have|plan to implement in Gecko? |
| 08:10 | <hsivonen> | jgraham: my plan is still what I described at http://hsivonen.iki.fi/about-blank/ |
| 08:10 | <hsivonen> | jgraham: I don't expect the spec to match currently |
| 08:11 | <jgraham> | hsivonen: Is there a bug open? |
| 08:13 | <Von_Davidicus> | Hello, all. |
| 08:15 | <hsivonen> | jgraham: https://bugzilla.mozilla.org/show_bug.cgi?id=543435 |
| 08:18 | <jgraham> | hsivonen: Thanks. I was thinking of a W3C bug that I could ask Hixie to look at soon :) |
| 08:18 | <jgraham> | (but that is helpful too) |
| 08:23 | <hsivonen> | jgraham: my patches have rotted quite a bit since last summer, so lately, I've mainly been focusing on making readyState less broken in Gecko |
| 08:24 | <hsivonen> | because it's hard to get readyState right for about:blank if readyState in general is bogus |
| 08:30 | <jgraham> | Sounds reasonable |
| 08:31 | <annevk> | zcorpan: fwiw, not sure I'll get much more work done today on labels or anything really |
| 08:32 | <annevk> | zcorpan: I'm planning to write a presentation on bytes/code points, and how they fuck you up when using <form>, XMLHttpRequest, and URLs, unless you use utf-8 that is |
| 08:32 | <annevk> | and I guess I might mention 16-bitness of ECMAScript |
| 08:32 | <annevk> | given all the trendy emoji stuff |
| 08:37 | <zcorpan> | annevk: k |
| 08:37 | <jgraham> | Pretty sure that a trendy presentation these days would just have one emoji on each slide, and you would shout the name of the emoji at the audience over and over before moving on to the next slide |
| 08:37 | <zcorpan> | having old data is annoying because many pages have changed or disappeared :-( |
| 08:39 | <annevk> | jgraham: I can probably go on for half an hour explaining the story behind 💩 and why its string length is 2 |
| 08:39 | <annevk> | :p |
| 08:42 | <jgraham> | (which codepoint was that?) |
| 08:43 | <zcorpan> | http://www.fileformat.info/info/unicode/char/1f4a9/index.htm |
| 08:43 | <jgraham> | Oh, the obvious one |
| 08:44 | <kennyluck> | U+1F4A9 PILE OF POO lol, what's the story behind it? |
| 08:44 | <jgraham> | I just have visions of annevk standing at the front of a packed audience and saying "pile of poo" first in a calm voice and then in an increasingly irate voice |
| 08:44 | <jgraham> | For some minutes |
| 08:45 | <jgraham> | Possibly my internal sense of humor is wonky though :) |
| 08:46 | <tomasf> | http://xn--ls8h.la/ |
| 08:48 | <jgraham> | Or possibly it was just reading http://westcoastlogic.com/slides/debug-mobile/#/ that left me with the idea that being shouty is fashionable |
| 09:13 | <MikeSmith> | hsivonen: a while back you said you were planning to land the VerifierCommandLine code from the pfalcon bitbucket fork. Still planning on doing that? |
| 09:14 | <hsivonen> | MikeSmith: planning, yes, doing, evidently not quickly :-( |
| 09:14 | <MikeSmith> | OK |
| 09:15 | <MikeSmith> | I also have some changes to suggest for that |
| 09:15 | <MikeSmith> | after it's landed |
| 10:05 | <zcorpan> | jgraham: i've never seen slides so full of memes as that |
| 10:21 | <jgraham> | Yeah memes are the new meme |
| 10:21 | <AryehGregor> | Someone has finally granted review to my first batch of transforms tests. |
| 10:22 | <AryehGregor> | Unfortunately, that means I have to work on submitting all the rest. :( |
| 10:22 | <jgraham> | At W3C? |
| 10:22 | <AryehGregor> | Yes. |
| 10:22 | <jgraham> | Nice! |
| 10:23 | <hsivonen> | jgraham: looks like profanity is still OK on presentations slides |
| 10:28 | <jgraham> | hsivonen: I'm not sure I find profanity very offensive (when it isn't being used in a way intended to cause offense), so I'm not sure I am the best person to decide whether that should be socially unacceptable or not. I wouldn't do it myself for sure. Maybe one could argue it is associated with the sort of macho culture that I sometimes percieve exists in part of the web dev community. And if that culture exists I would argue it is a bad thing, so�� |
| 10:31 | <hsivonen> | hmm. I get a feeling that Gecko might be accidentally setting readyState to "complete" and firing the load event in cases where it wasn't originally supposed to |
| 10:32 | <hsivonen> | brittle code is brittle |
| 10:32 | <hsivonen> | especially without fatal assertions and proper test cases |
| 11:01 | <Velmont> | hsivonen: Where? There's thousand readyStates :P |
| 11:08 | <AryehGregor> | Hmm. I seem not to be able to access the US. I think Europe works . . . |
| 11:08 | <hsivonen> | Velmont: e.g. in the case where an XSLT transform fails because the XSLT transform fails to compile |
| 11:08 | <AryehGregor> | Yay for DVCSes, I gues. |
| 11:08 | <AryehGregor> | Yay for DVCSes, I guess. |
| 11:33 | <AryehGregor> | So, WebKit used to output class="Apple-style-span" everywhere. Spec dilemma: should we special-case the class given the amount of content that uses it? |
| 11:33 | <AryehGregor> | (probably substantial) |
| 11:33 | <AryehGregor> | If we didn't, it would leave <span class="Apple-style-span"> lying around . . . is that less bad? |
| 11:34 | <AryehGregor> | It doesn't affect rendering. |
| 11:34 | <AryehGregor> | I doubt WebKit will follow the spec if it doesn't special-case Apple-style-span . . . |
| 11:34 | AryehGregor | supposes it's better to special-case it, sigh |
| 11:36 | <AryehGregor> | (recent WebKit no longer emits that class, thankfully) |
| 11:36 | <AryehGregor> | (but still special-cases it for removal) |
| 11:36 | <gsnedders> | AryehGregor: removes it where/when? |
| 11:37 | <AryehGregor> | gsnedders, if you have <span class="x" style="color:red">foo</span>bar and make it green, WebKit produces <span style="color: green"><span class="x">foo</span>bar</span>. If the class is "Apple-style-span", instead you get just <span style="color: green">foobar</span>. |
| 11:37 | <AryehGregor> | The richtext2 test suite tests for this. |
| 11:38 | <AryehGregor> | And realistically, it is better to remove it in that case. |
| 11:38 | <AryehGregor> | It's just ugly to write that in a spec. |
| 11:38 | <AryehGregor> | Sadly, WebKit tends to be extremely reluctant to degrade their behavior slightly in order to match other browsers, even when it would be much uglier or more annoying for the other browsers to change. :( |
| 11:39 | <zcorpan> | that doesn't mean other browsers are willing to implement special cases for Apple-style-span just to be compatible with webkit :-) |
| 11:39 | <AryehGregor> | The question is what I should spec and implement. |
| 11:40 | <zcorpan> | does special-casing the class gain web compat? |
| 11:40 | <zcorpan> | if not, i see little reason to special-case it in opera |
| 11:40 | <AryehGregor> | In the sense of pages not breaking, no. |
| 11:40 | <AryehGregor> | Noted. |
| 11:40 | <AryehGregor> | I'm ambivalent. |
| 11:40 | <AryehGregor> | So I'll push the call off to ehsan. |
| 11:41 | <AryehGregor> | :) |
| 11:42 | <AryehGregor> | The only reason to do it is for richtext2, honestly. |
| 11:42 | <AryehGregor> | So I think I'll say to heck with richtext2, they should change. |
| 11:46 | <zcorpan> | hsivonen: doesn't the escape key abort document load in IE? |
| 11:49 | <hsivonen> | zcorpan: maybe, but that's not a Web-exposed API |
| 11:50 | <zcorpan> | use Watir :-) |
| 11:50 | <hsivonen> | looks like I need to write test cases in order to figure this stuff out. sigh. |
| 11:52 | <hsivonen> | and here I thought fixing readyState was a quick step that'd make fixing about:blank easier |
| 11:54 | <hsivonen> | there has to be a Laundry Room Viking meme in here somewhere |
| 11:56 | <Philip`> | jgraham: I find the bigger problem with the profanity is that it drives out any attempts to analyse the problems rationally - shouting disparaging remarks about some technology is seen as a viable substitute for understanding it, and images of angry people are a substitute for facts and data |
| 11:58 | <Philip`> | Even if the presenter fully understands what they're complaining about, the point of giving a presentation is that you're largely presenting to people who don't understand that area yet, and they'll end up remembering the emotion without gaining any of the understanding |
| 12:10 | <hsivonen> | aargh. why don't DOMContentLoaded and readystatechange propagate to the host iframe element like load? |
| 12:16 | <smaug____> | hsivonen: load doesn't propagate to the host iframe |
| 12:16 | <smaug____> | it is a separate load event for iframe |
| 12:21 | <hsivonen> | smaug____: ok. interesting |
| 12:23 | <hsivonen> | smaug____: is there a race condition-free way to observe the load, DOMContentLoaded and readychange events happening inside an iframe from the parent? |
| 12:24 | <hsivonen> | If I set them on .contentWindow and then navigate the iframe, it seems the navigation blows away the listeners |
| 12:24 | <smaug____> | if the docs are same origin, I guess the iframe could call some method in the parent |
| 12:24 | <hsivonen> | smaug____: ok, so the child needs to participate |
| 12:24 | <hsivonen> | that won't work for loading XSLT error documents into the iframe |
| 12:24 | <smaug____> | afaik, yes |
| 12:25 | <hsivonen> | the design of this platform is unfortunate |
| 12:26 | <smaug____> | hey, it is the Web. don't expect too much :) |
| 12:26 | <zcorpan> | Ms2ger: http://krijnhoetmer.nl/irc-logs/whatwg/20120418#l-1104 hmm, i had hoped we could get away with dropping one of the tables... |
| 12:46 | <annevk> | hello again, anyone got a visible newline character for source code? |
| 12:47 | <annevk> | I'm having bad luck googling one |
| 12:49 | <zcorpan> | v.nu uses ↩ |
| 12:49 | <annevk> | thanks |
| 12:52 | gsnedders | pushes another perf optimization to html5lib, 16% quicker than 24 hours ago parsing the spec. |
| 12:52 | <gsnedders> | (In CPython 2.7.2) |
| 12:54 | <annevk> | madman |
| 13:09 | <annevk> | gsnedders: do you have a few minutes? |
| 13:14 | <annevk> | seems gsnedders is not around, can anyone else review a few slides? |
| 13:14 | <annevk> | oh |
| 13:14 | <annevk> | maybe he is |
| 13:25 | <hsivonen> | hooray. we don't even have interop at the simplest baseline: http://hsivonen.iki.fi/test/moz/readystate/baseline.html |
| 13:25 | <AryehGregor> | Reformatting by shell script, the hard way! for FILE in ref-*/*.html; do echo -e ':set shiftwidth=2 expandtab\ngg=G:wq' | vim $FILE; done |
| 13:25 | <hsivonen> | Opera doesn't fire readystatechange for the transition to "interactive" |
| 13:25 | AryehGregor | thinks this will take a while |
| 13:27 | <hsivonen> | whoa! IE10 doesn't fire readystatechange to "interactive", either |
| 13:29 | <hsivonen> | hmm. IE9 won't start. should have snapshotted the VM more often... |
| 13:31 | <jgraham> | hsivonen: Seems like a clear bug in IE/Opera unless we know of a web-compat reason for the existing behaviour |
| 13:32 | <hsivonen> | wild. in IE9 and IE10, readyState is already "interactive" when the listeners get attached |
| 13:32 | <hsivonen> | so that's why IE9 and IE10 don't fire a readystatechange |
| 13:33 | <jgraham> | Isn't teh about:blank case something far away from the baseline? |
| 13:33 | <hsivonen> | in other news, WebKit and Trident don't support DOMFrameContentLoaded |
| 13:34 | <hsivonen> | jgraham: yes. that's why the baseline loads baseline-inner.html via HTTP |
| 13:34 | <jgraham> | What happens if you just do it in the top level browsing context? |
| 13:34 | <jgraham> | Oh, right I misread the test |
| 13:39 | <hsivonen> | ok. "interactive" in IE is weird |
| 13:39 | <hsivonen> | jgraham: without iframe: http://hsivonen.iki.fi/test/moz/readystate/without-iframe.html |
| 13:39 | <hsivonen> | says "interactive" in IE10 when attaching listeners |
| 13:40 | <hsivonen> | I wonder if IE sets readyState to interactive when it has consumed the network stream even though the parser is not supposed to have seen the whole stream yet |
| 13:49 | <hsivonen> | OK. IE is really weird about not doing incremental rendering: http://hsivonen.iki.fi/test/moz/slow.php |
| 13:49 | <hsivonen> | how slow does a page have to be to get incremental rendering in IE? |
| 13:50 | <hsivonen> | I feel I'm failing. I can't even establish the baseline in IE |
| 13:56 | <jgraham> | Opera also gets the order of events wrong |
| 13:56 | jgraham | has convereted the simple TC to a testharness.js test |
| 13:56 | <jgraham> | So I will get Opera fixed :) |
| 13:56 | <jgraham> | s/TC/demo/ |
| 13:59 | <hsivonen> | slow.php above does get an incremental treament in IE6 |
| 14:07 | <hsivonen> | I guess I have to write a shim around attachEvent if I want the tests to work in IE6... |
| 14:13 | <hsivonen> | jgraham: http://hsivonen.iki.fi/test/moz/readystate/without-iframe.html changes upon reload in Opera Next |
| 14:13 | <hsivonen> | jgraham: says DOMContentLoaded; readyState: complete on reload |
| 14:13 | <hsivonen> | I guess Gecko isn't the only engine with very sad and broken readyState |
| 14:15 | <hsivonen> | why isn't my code working in IE6? |
| 14:16 | <hsivonen> | no script errors |
| 14:17 | <jgraham> | hsivonen: Oh man, that's sad |
| 14:17 | <jgraham> | Thanks |
| 14:17 | <hsivonen> | You're welcome |
| 14:23 | <hsivonen> | well, *this* is bizarre. In IE8, http://hsivonen.iki.fi/test/moz/readystate/without-iframe-slow.php is parsed incrementally, but it still says "interactive" when attaching event handlers |
| 14:23 | <hsivonen> | so when does *any* version of IE say "loading"? |
| 14:27 | <hsivonen> | http://msdn.microsoft.com/en-us/library/ie/ms534359%28v=vs.85%29.aspx says there's a "loaded" state |
| 14:28 | <kennyluck> | foolip, do you have any idea about how to move forward with big5? Conclude that big5-hkscs is the best we can do? Add some big5-2003 mappings to big5-hkscs? |
| 15:07 | <miketaylr> | hsivonen: not sure how relevant this is to what you're working on, but it came up the other day in jQuery https://github.com/jquery/jquery/pull/736/files#L1R15 |
| 15:09 | <jgraham> | Yeah Opera seems to be Just Wrong there |
| 15:10 | <miketaylr> | â„¢ |
| 15:24 | <wilhelm> | This algorithm is wonderful: http://dvcs.w3.org/hg/webdriver/raw-file/tip/webdriver-spec.html#rendering-text |
| 15:24 | <jgraham> | wilhelm: Don't reference DOM3 Core! |
| 15:25 | <wilhelm> | jgraham: Use DOM4 instead? |
| 15:25 | <jgraham> | YEs |
| 15:25 | <wilhelm> | Oh, it's even in /tr/. |
| 15:25 | <jgraham> | s/E/e/ |
| 15:25 | <Hixie> | lordy, don't reference the TR/ version |
| 15:25 | <wilhelm> | (c: |
| 15:26 | <Hixie> | that's perennially out of date |
| 15:26 | <wilhelm> | Noted. |
| 15:26 | <jgraham> | It helps if you imagine that TR stands for "Toally Wrong" |
| 15:26 | <jgraham> | *Totally |
| 15:27 | <Hixie> | "Temporally Relegated" |
| 15:27 | <jgraham> | If you're going for that, Retarded would be better than Relegated, I think |
| 15:28 | <Hixie> | "retarded" unfortunately has implications that are offensive to some |
| 15:28 | <wilhelm> | jgraham: More input in the same genre would be very helpful. Feel free to skim the rest of the draft. (c: |
| 15:28 | <jgraham> | Yeah :( |
| 15:28 | <Hixie> | so i tend to avoid the word entirely |
| 15:28 | <Hixie> | relgated is actually pretty apt, though |
| 15:28 | <Hixie> | relegated |
| 15:29 | <hober> | rel=gated? |
| 15:29 | <jgraham> | Retarded is used a lot in physics though |
| 15:29 | <jgraham> | In a non-offensive way |
| 15:30 | <Hixie> | yeah |
| 15:31 | <Philip`> | "Tired Relics" |
| 15:31 | <jgraham> | (e.g. "retarded waves" in http://en.wikipedia.org/wiki/Wheeler–Feynman_absorber_theory ) |
| 15:31 | <Hixie> | the thing about TR/ is it's not necessarily so out of date that "Totally Wrong" or "Relic" would be completely fair |
| 15:32 | <Hixie> | it's usually only a few months out of date -- just enough to be a terrible choice for implementors, but not enough to be so obviously wrong that people naturally avoid it |
| 15:32 | <Hixie> | e.g. the way they typically avoid HTML4 |
| 15:36 | <Ms2ger> | zcorpan, that sure would be nice... |
| 15:36 | <Ms2ger> | Does Opera have mode-specific tables too? |
| 15:50 | <Hixie> | hsivonen: document.open() also calls the abort stuff according to the spec, fwiw |
| 15:51 | <Hixie> | hsivonen: also you may be able to test IE by hitting Escape to stop it |
| 15:52 | <jgraham> | hsivonen: Does window.stop() actually call abort? |
| 15:52 | <jgraham> | er, Hixie ^] |
| 15:52 | <Hixie> | not per spec, but who knows about reality |
| 15:52 | <Hixie> | i was surprised to find i'd specced it at all :-) |
| 15:53 | <jgraham> | I was kind of hoping that the spec might bear some resembelance to reality here |
| 15:53 | <jgraham> | :) |
| 15:53 | <Hixie> | i'm happy to assume the spec is reality until proved otherwise :-) |
| 15:54 | <jgraham> | FWIW it's very unclear to me what "cancel that navigation" entails |
| 15:54 | <jgraham> | Is that defined anywhere? |
| 15:54 | <jgraham> | (I thought I filed a bug about this but it seems to have got lost) |
| 15:54 | <Hixie> | not explicitly. i intended it to mean you just stop the algorithm wherever you are in it |
| 15:55 | <Hixie> | feel free to file another using the thing in the spec |
| 15:55 | <Hixie> | duplicates are very low-cost |
| 16:01 | <jgraham> | Hixie: So apart from not firing an abort event why wouldn't window.stop() be just like running the abort algorithm? |
| 16:01 | <Hixie> | it probably predates my writing the abort algorithm |
| 16:01 | <jgraham> | OK |
| 16:02 | <jgraham> | The abort algorithm looks reassuringly explicit about several things that window.stop ignores :) |
| 16:02 | <Hixie> | :-) |
| 16:05 | <Ms2ger> | jgraham, and is it correct? :) |
| 16:06 | <jgraham> | Ms2ger: No idea :) |
| 16:07 | <jgraham> | Something that is explicitly wrong is easier to work with than something that is vaugely right |
| 16:08 | <tantek> | jgraham - is that a justification for religion over fuzzy science? ;) |
| 16:09 | <Ms2ger> | tantek, or astrophysics over fuzzy science? :) |
| 16:09 | <tantek> | astrology over astronomy? |
| 16:09 | <Ms2ger> | jgraham, did you say something about a testharness.js test you were going to submit? :) |
| 16:09 | <jgraham> | tantek: I'm not sure that religious people regard themselves as explicitly wrong :) |
| 16:10 | <jgraham> | But it is a justification that a scientific theory that makes an explicit, but incorrect, prediction is more valuable than a fuzzy notion that can be made to fit multiple sets of empirical findings |
| 16:11 | <tantek> | jgraham - those of each religion regard the others as explicitly wrong ;) |
| 16:12 | <tantek> | I'm not sure how valuable explicit but wrong scientific theories are, as there is no shortage of them in the space of ideas. That being said, documenting them can help reduce wasting time re-discussing them. of course that doesn't stop it from happening (on nearly every standards mailing list). |
| 16:13 | <jgraham> | They are very valuable. Of course you typically don't know they are wrong until later |
| 16:14 | <jgraham> | But without a theory that makes precise predictions you can't design an experiemnt that will test them |
| 16:15 | <jgraham> | Ms2ger: All in good time :) |
| 16:17 | <dglazkov> | good morning, Whatwg! |
| 16:24 | <Ms2ger> | Good, dglazkov |
| 16:30 | <Hixie> | +1 to precise testable statements being more valuable, even if wrong, than vague statements that can be made to fit almost any data |
| 17:57 | <MikeSmith> | kennyluck: somebody needs to just make an executive decision. I vote for you |
| 17:58 | <kennyluck> | MikeSmith, huh? |
| 17:58 | <MikeSmith> | "do you have any idea about how to move forward with big5? Conclude that big5-hkscs is the best we can do? Add some big5-2003 mappings to big5-hkscs?" |
| 17:59 | <kennyluck> | MikeSmith, oh… I have no idea... In fact I care very little about this :p |
| 17:59 | <MikeSmith> | and I care far less than you do |
| 18:00 | <MikeSmith> | especially since most sites in china are about how to pirate software anway |
| 18:01 | <kennyluck> | MikeSmith, I believe most sites related to piracy have all changed to use utf-8 for business reasons ;) |
| 18:01 | <MikeSmith> | heh |
| 18:02 | <kennyluck> | only sites that don't make money and sites nobody care about still use legacy big5. |
| 18:02 | <MikeSmith> | yeah |
| 18:03 | <MikeSmith> | seriously I think it makes more sense to break content for mainland china than it does for taiwan |
| 18:03 | <MikeSmith> | since taiwan is open |
| 18:03 | <MikeSmith> | and PROC is not |
| 18:04 | <MikeSmith> | I mean, a few broken sites for mainland users is nothing compared to the thousands of sites they can't to at all at anyway |
| 18:05 | <MikeSmith> | and they are used to dealing with things |
| 18:05 | <MikeSmith> | the savvy users I mean |
| 18:06 | <MikeSmith> | plus I really wonder what kinds of sites are using the extended stuff anyway, and why |
| 18:06 | <MikeSmith> | kennyluck: btw, are you in Beijing? |
| 18:07 | <MikeSmith> | I was there last month you know |
| 18:07 | <MikeSmith> | and I expected I'd run into you |
| 18:07 | <MikeSmith> | but you were nowhere to be found |
| 18:09 | <MikeSmith> | in other news, Emotion Markup Language headed to CR |
| 18:10 | <Ms2ger> | \o_ |
| 18:10 | <TabAtkins> | Please mark up that emotion semantically, Ms2ger |
| 18:17 | <jgraham> | TabAtkins: That's an emotion? I thought it was "guiding a plane in to land" |
| 18:17 | <Ms2ger> | Beep |
| 18:18 | <TabAtkins> | jgraham: If you dont' feel emotions when you successfully guide a plane in to land, you're dead inside. |
| 18:20 | <jgraham> | Who said anything about successfully? |
| 18:20 | <TabAtkins> | Either way, really. |
| 18:20 | <Ms2ger> | If you don't feel emotions when you make a plane crash on landing, you're not dead inside? |
| 18:23 | <Philip`> | Ms2ger: Surely it's the passengers who are dead inside |
| 18:23 | Ms2ger | kicks Philip` |
| 19:49 | <Hixie> | sigh, so annoying that webvtt is in a separate document rather than part of the html spec |
| 19:50 | <Ms2ger> | Ah, the joys of CGs |
| 21:01 | <smaug____> | dglazkov: so inside decorators, <script> element would work in a different way than elsewhere in DOM ? |
| 21:01 | <smaug____> | (different 'this') |
| 21:13 | <dglazkov> | smaug____: yes. |
| 22:10 | <annevk> | gotta love this email http://mail.apps.ietf.org/ietf/charsets/msg02053.html |
| 22:11 | <annevk> | great timing too while I was away advocating usage of utf-8 at a local developer meetup http://annevankesteren.nl/presentations/1F4A9.html |
| 22:11 | <TabAtkins> | Argh, ietf, y u use <pre> without manual line-wrapping? |
| 22:13 | <TabAtkins> | I fail to see how "pages will be broken even with a common standard" is any different from "pages will be broken because there's no common standard". |
| 22:13 | <TabAtkins> | Or, more precisely, I fail to see how the former is an argument against a common standard, given the latter being true today. |
| 22:14 | <annevk> | I'm not going to reply today, but he's saying some great future Claim Chowder stuff and in general ignoring my point that interpreting existing content is hard |
| 22:14 | <TabAtkins> | Claim Chowder? |
| 22:14 | <annevk> | I mean "Use Unicode." is not an answer if you want to decode a page encoded in euc-kr or gbk |
| 22:15 | <TabAtkins> | Indeed. |
| 22:15 | <annevk> | "There’s a term called Claim Chowder that was, as far as I can tell, coined by Daring Fireball’s John Gruber. It refers to when someone makes a prediction with an aura of certainty and knowledge that turns out to be horribly wrong." |
| 22:15 | <annevk> | -- http://401stblow.wordpress.com/2010/01/20/claim-chowder/ |
| 22:16 | <zewt> | annevk: nothing validates microsoft-hatred quite as much as "this is broken, we'll never fix it, we don't care how much pain this causes the industry, go away" |
| 22:22 | <zewt> | (not that they're the only ones guilty of that, like the webgl guy saying that he won't change typed array views to little endian in the spec regardless of anything anyone says) |
| 22:26 | <annevk> | Microsoft also would not do CORS or XMLHttpRequest Level 2 |
| 22:26 | <annevk> | at some point we thought they would not do anything with their browser anymore |
| 22:27 | <annevk> | aligning quirks mode seemed out of the question too |
| 22:27 | <annevk> | it's like they can't see the game long term |
| 22:27 | <zewt> | (or else, the usual conspiracy theories) |
| 22:28 | <annevk> | it's no problem, we'll just lead the platform with Presto/WebKit/Gecko and they'll hobble along later |
| 22:34 | <smaug____> | dglazkov: what is setupEvent ? |
| 22:40 | <annevk> | heh, the other guy is questioning if I have the right to tighten up utf-8 |
| 22:42 | <Hixie> | annevk: the great thing about implementors who say they won't change at all is that it frees you to ignore their implementation |
| 22:43 | <Hixie> | annevk: it's equivalent to a (closed source) vendor having gone out of business |
| 22:43 | <zewt> | annevk: ... but you're not, you're just writing a parser for something that is already (as far as I know) tightly specified |
| 22:43 | <zewt> | it's not like you're making up new rules |
| 22:44 | <annevk> | Unicode has options for the error handling, with a should for the variant in my draft |
| 22:44 | <annevk> | Unicode also does not define all the utf-8 labels last I checked |
| 22:44 | <annevk> | and I'm not sure if BOM handling is identical |
| 22:44 | <annevk> | and it's in a silly PDF |
| 22:45 | <zewt> | that's sort of out of scope for the UTF-8 definition itself; you're just defining a bit more than that |
| 22:45 | <annevk> | Hixie: yeah, although we have to reverse engineer them mostly :) |
| 22:45 | <Hixie> | annevk: well that's about being compatible with content that relies on their implementation |
| 22:45 | <annevk> | fair enough |
| 22:45 | <Hixie> | annevk: the arguments you're getting here are the same as the arguments we got when speccing the html parser |
| 22:45 | <zewt> | (it's not like you're really saying "this is how utf-8 handles the BOM", you're saying "this is how UTF-8 works, and when used by browsers, we treat the UTF-8 BOM in this way") |
| 22:46 | <Hixie> | annevk: and the same as the arguments about how living standards are bad because they are "unstable" |
| 22:46 | <Hixie> | annevk: i.e. the assumption that the spec will not be of sufficient quality to actually beat implementations at the QoI game |
| 22:47 | <Hixie> | annevk: that any change to the spec is by definition one that makes the spec worse |
| 22:47 | <annevk> | Hixie: he omitted what I said in the previous email, but the HTML parser was my example, see http://mail.apps.ietf.org/ietf/charsets/msg02050.html |
| 22:47 | <Hixie> | annevk: i doubt this guy has any idea what we did with html, your e-mail probably way over his head |
| 22:48 | <Hixie> | annevk: you can't really fault people for thinking that what we're doing is dangerous, i mean, for the longest time specs didn't try to match reality |
| 22:48 | <Hixie> | annevk: people who haven't spent years seeing the recent changes to web spec quality are just assuming we're doign the same as everyone else, and changing things blindly |
| 22:49 | <annevk> | dangerous how though? to their job as expert about undocumented knowledge? |
| 22:49 | <zewt> | blindly assuming that other people are doing things blindly |
| 22:49 | <zewt> | there's no small irony in that |
| 22:50 | <Hixie> | annevk: to compatibility |
| 22:50 | <Hixie> | annevk: as in, if they implement it, their implementation will be of lower quality relative to existing content than if they did not |
| 22:50 | <annevk> | I also find it somewhat funny he thinks my effort would be better put into promoting utf-8; I've been promoting utf-8 since before I understood how it actually worked, but still in 2012 browsers deal with encoding bugs |
| 22:51 | <annevk> | Hixie: ah right |
| 22:51 | <Hixie> | q.v. people telling me i should promote content-type. or xhtml2. :-) |
| 22:51 | <Hixie> | or link headers |
| 22:51 | <Hixie> | or namespaces |
| 22:51 | <Hixie> | or... |
| 22:51 | <annevk> | SGML comments! |
| 22:51 | <annevk> | oh wait |
| 22:51 | <Hixie> | indeed! |
| 22:52 | <Hixie> | sgml in general |
| 22:52 | <Hixie> | all these things i used to promote before i realised that our efforts are better spent designing resilient technologies than trying to educate the world :-) |
| 22:57 | <annevk> | oh, I think I somewhat convinced :emk about the Encoding Standard |
| 22:58 | <annevk> | nice |
| 23:00 | <annevk> | oh, I should update my WebVTT validator |
| 23:00 | <annevk> | it still has options implemented as single letters |
| 23:00 | <annevk> | well, settings I guess they were called |
| 23:04 | <Hixie> | :emk? |
| 23:05 | <zewt> | vim? heh |
| 23:08 | <ojan> | anyone know the state of moving eventtarget from being a mixin to a superclass in opera, gecko, IE? |
| 23:08 | <ojan> | the webkit bug is stalled due to a resistance to moving forward without other browser vendors changing around the same time. :( |
| 23:09 | <smaug____> | ojan: slowly in progress in gecko |
| 23:09 | <smaug____> | XHR has that already |
| 23:09 | <ojan> | smaug____: is there a bug i can follow? |
| 23:09 | <smaug____> | and other objects should get the same behavior as we move to new dom bindings |
| 23:09 | <smaug____> | ojan: not really |
| 23:09 | <ojan> | smaug____: ok...thanks |
| 23:11 | <smaug____> | perhaps slowly was wrong |
| 23:11 | <smaug____> | but I think only XHR will have it in FF14 |
| 23:16 | <annevk> | Hixie: works on encodings among other things in Gecko |
| 23:21 | <Hixie> | annevk: cool |
| 23:21 | <annevk> | ojan: you guys should make it easier to experiment with changes to be able to convince the other guys |
| 23:21 | <annevk> | ojan: I think |
| 23:22 | <ojan> | annevk: that's true...we could put this behind a flag and just ship in in chromium for now |
| 23:22 | <annevk> | ojan: we don't really have the resources so it's mostly Gecko that's bearing the cost of slowly fixing up the platform |
| 23:22 | <annevk> | we = Opera |
| 23:22 | <ojan> | annevk: yup |
| 23:22 | <ojan> | annevk: chromium would be willing to ship it and deal with any fallout |
| 23:23 | <annevk> | sleepytimes here, nn |