| 05:14 | <Hixie> | deep in the middle of a page's markup, suddenly, a wild namespace appears!: |
| 05:14 | <Hixie> | <?xml version="1.0" encoding="utf-16"?><table xmlns="http://www.w3.org/TR/REC-html40"; xmlns:msxsl="urn:schemas-microsoft-com:xslt" xmlns:dt="urn:schemas-microsoft-com:datatypes" xmlns:user="urn:my-scripts"><tr>... |
| 05:28 | <tantek> | xmlns - worst attribute ever? |
| 09:31 | <matjas> | I just realized that XHTML handles character references differently |
| 09:31 | <matjas> | i.e. `…` → U+0085 in XHTML, while in HTML it’s U+2026 |
| 09:31 | <matjas> | but AFAICT this isn’t mentioned here: http://www.whatwg.org/specs/web-apps/current-work/multipage/tokenization.html#tokenizing-character-references what am I missing? |
| 09:32 | <matjas> | I assume the parsing section only applies to non-XHTML HTML, as XML is defined elsewhere, correct? |
| 09:36 | <Ms2ger> | That's in the HTML syntax section, no? |
| 09:37 | <matjas> | http://www.w3.org/TR/xml/#d0e3895 |
| 09:38 | <matjas> | Ms2ger: yeah, I guess it all makes sense |
| 09:38 | <matjas> | just never realized this difference before — kinda mind-blowing |
| 09:39 | <Ms2ger> | Yay for stupid stuff |
| 10:10 | <zcorpan> | matjas: check out http://software.hixie.ch/utilities/js/live-dom-viewer/saved/2696 |
| 10:24 | <hsivonen> | zcorpan: I can't remember what Gecko does with non-JS <script src> |
| 10:46 | <zcorpan> | hsivonen: in my testing it doesn't fetch anything |
| 10:49 | <hsivonen> | zcorpan: than it doesn't :-) |
| 10:49 | <hsivonen> | *then |
| 10:52 | <matjas> | zcorpan: oh wow |
| 10:56 | <hsivonen> | matjas: It's kinda annoying that Unicode starts with ISO-8859-1 instead of starting with windows-1252 |
| 11:06 | <annevk-cloud> | So much |
| 11:24 | <darobin_> | Domenic_: I read in the back log that you felt bad about being flippant and unpolitic about streams — don't. It can get a lot worse without being a problem. |
| 11:24 | <darobin_> | If you feel you should somehow apologise do so, but in any case just move on |
| 11:25 | <darobin_> | and give us a fucking streams API today :) |
| 11:26 | <hsivonen> | annevk: did you check the source code for the GBK and GB18030 decoders in Gecko? |
| 11:26 | <hsivonen> | the latter is a subclass of the former |
| 11:27 | <annevk> | hsivonen: but they are not using the same table I think |
| 11:27 | <annevk> | hsivonen: I find Gecko's code for encodings hard to follow |
| 11:27 | <annevk> | hsivonen: it has all kinds of abstractions that could be removed |
| 11:30 | <hsivonen> | annevk: not the same table, right |
| 11:30 | <hsivonen> | annevk: the GBK table has one entry! for euro |
| 11:30 | <annevk> | hsivonen: whoa |
| 11:30 | <hsivonen> | I can't figure out from the GB18030 tables if 0x80 maps to euro in the larger table |
| 11:32 | <hsivonen> | euro sign is kinda like time zones: the politician who come up with this stuff should be required to implement this stuff. *Correctly.* |
| 11:33 | <annevk> | hsivonen: in gb18030 0x80 maps to 0x20AC |
| 11:33 | <annevk> | hsivonen: so if Gecko does that for gbk... |
| 11:33 | <annevk> | and http://mxr.mozilla.org/mozilla-central/source/intl/uconv/ucvcn/gbkuniq2b.ut seems to indicate it does |
| 11:35 | <annevk> | hsivonen: actually, in Chrome gb18030 0x80 does not map to 0x20AC |
| 11:35 | <hsivonen> | I regret putting gbk in the new charset menu |
| 11:36 | <hsivonen> | since now it's more of a hassle to change the UI strings |
| 11:39 | <annevk> | hsivonen: my bad, I should have sorted this out long ago |
| 11:39 | <annevk> | hsivonen: seems that in IE byte 80 is also not mapped to the Euro sign, but it's not mapped to anything sensible either so... |
| 11:43 | <annevk> | hsivonen: compare http://dump.testsuite.org/encoding/gbk/byte-80-gbk.html and http://dump.testsuite.org/encoding/gbk/byte-80-gb18030.html btw |
| 11:43 | <annevk> | hsivonen: however, it seems that treating 0x80 as 0x20AC is harmless |
| 11:45 | <annevk> | hsivonen: http://en.wikipedia.org/wiki/GB_18030#GB18030_as_a_code_page has all the details it seems in the third paragraph with respect to the incompatibility |
| 11:45 | <hsivonen> | annevk: the latter shows a question mark for 0x80 in IE11 |
| 11:45 | <hsivonen> | why not U+FFFD? |
| 11:45 | <annevk> | hsivonen: same in IE10 |
| 11:46 | <annevk> | hsivonen: I suspect they just have weird error handling |
| 11:47 | <annevk> | I could test what they do for FF |
| 11:47 | <annevk> | or 8100 |
| 11:47 | annevk | makes a test |
| 11:50 | <annevk> | hsivonen: see also http://dump.testsuite.org/encoding/gbk/byte-FF-gb18030.html and http://dump.testsuite.org/encoding/gbk/byte-8100-gb18030.html |
| 11:50 | <annevk> | hsivonen: IE10 is a mess |
| 11:51 | <annevk> | hsivonen: not sure what Chrome is doing either :/ |
| 11:51 | <annevk> | Or I suppose, what ICU is doing, that seems broken |
| 11:52 | <hsivonen> | annevk: does IE just drop 0xFF? |
| 11:52 | <annevk> | hsivonen: no it renders it as |
| 11:53 | <annevk> | hsivonen: whoa, that's a round dot in Windows, but on Mac it renders totally different... |
| 11:54 | <annevk> | hsivonen: oh, it's PUA |
| 11:55 | <annevk> | hsivonen: U+F8F5 |
| 12:13 | <zcorpan> | wonder if i should test url query encoding in the prescanner |
| 12:18 | <annevk> | hsivonen: the IE gbk table is identical to the Chrome gbk table |
| 12:18 | <annevk> | hsivonen: I find one difference in the gb18030 table weirdly enough |
| 12:18 | <annevk> | hsivonen: index 6555 maps to 3000 in Gecko and E5E5 in IE |
| 12:24 | <annevk> | hsivonen: seems like we should just alias gbk and gb18030 and not worry about the 81/82 PUAs |
| 12:30 | <hsivonen> | annevk: aliasing will change how the euro sign gets submitted in forms, right? |
| 12:31 | hsivonen | hopes the euro sign doesn't get submitted to Chinese sites often |
| 12:31 | <annevk> | hsivonen: yeah, it would use the gb18030 two byte sequence, unless we special case 20AC in the encoder |
| 12:32 | <annevk> | hsivonen: special casing it in the encoder however would be incompatible with non-gbk compatible gb18030 implementations |
| 12:32 | <hsivonen> | special-casing the encoder might break sites that really are already using de jure GB18030 |
| 12:32 | <annevk> | right |
| 12:32 | <hsivonen> | currency signs are such a bad idea |
| 12:33 | <annevk> | Surprising how little code the x-user-def is |
| 12:33 | <hsivonen> | both this whole euro thing. and the sheqel sign difference between 8859-8 and windows-1255 |
| 12:34 | <hsivonen> | annevk: yeah |
| 12:34 | <annevk> | Was it Unicode 6.2 or 6.3 that did the Turkish thing? |
| 12:34 | <hsivonen> | oh and *that* |
| 12:35 | <hsivonen> | at least bitcoin uses an existing sign |
| 12:38 | <annevk> | hsivonen: well, "B⃦ has been the standard currency sign for BTC for a long time. Some existing Unicode symbols have been proposed but also serious work is being done on creating a custom Bitcoin sign with its own official Unicode that is recognized by the Unicode Consortium." |
| 12:38 | <annevk> | hsivonen: from https://en.bitcoin.it/wiki/Bitcoin_symbol |
| 12:42 | <hsivonen> | annevk: sadness |
| 12:47 | <MikeSmith> | ah cool x-webkit-speech |
| 12:47 | <MikeSmith> | https://www.w3.org/Bugs/Public/show_bug.cgi?id=24126 |
| 14:49 | <Ms2ger> | odinho++ |
| 15:07 | <annevk> | hsivonen: just to be clear, current gbk sites with <form> might run into issues with more than U+20AC |
| 15:07 | <annevk> | hsivonen: as every code point will map to a byte sequence |
| 15:11 | <SimonSapin> | jgraham: what’s the fix when Critic gets upset by an amended commit? |
| 15:11 | <jgraham> | SimonSapin: There isn't one |
| 15:12 | <jgraham> | SimonSapin: I have instructions from jl on how to go about fixing that though |
| 15:12 | <jgraham> | So, happy christams to me, I guess |
| 15:13 | <SimonSapin> | I knew that present would make you happy :) |
| 15:13 | <jgraham> | SimonSapin: Unless you want to make a patch ;) |
| 15:13 | <SimonSapin> | I could submit a new pull request |
| 15:13 | <SimonSapin> | for the same changes |
| 15:14 | <SimonSapin> | https://github.com/html5lib/html5lib-python/pull/123 |
| 15:15 | <hsivonen> | annevk: are *all* of those sequences now inconsistent between different browsers' gbk encoders? |
| 15:16 | <annevk> | hsivonen: gbk is a subset |
| 15:16 | <annevk> | hsivonen: so if you go outside the subset, you'd get &#....; for the code point or ?, depending on the error handling |
| 15:16 | <hsivonen> | annevk: so is it inconsistent between "&#....;" and "?"? |
| 15:17 | <annevk> | hsivonen: gb18030 would however never trigger error handling |
| 15:17 | <annevk> | hsivonen: it's inconsistent between generating a gb18030 byte sequence for the code point in the merger scenario and &#...; in the gbk <form> submission scenario |
| 15:18 | <annevk> | (for code points outside the two byte range) |
| 15:18 | <annevk> | (and 0x80) |
| 15:18 | <odinho> | Ms2ger: I'm incremented! :D (Not too long ago also, I went from 26 to 27 on the 6th) |
| 15:18 | <Ms2ger> | Happy incrementation :) |
| 15:20 | <jgraham> | odinho: Wait, what, you are tracking it? |
| 15:21 | <Ms2ger> | I guess some people would avoid tracking it, at their level :) |
| 15:21 | <jgraham> | Ms2ger-- |
| 15:21 | <jgraham> | :p |
| 15:22 | <Ms2ger> | :D |
| 15:24 | <odinho> | deferred incrementation :) I save up the function and only flush once per year. And then only run one cycle. |
| 15:25 | <odinho> | I wonder what happens if I start getting more decrements than all increments I've had in total. What will happen then :-O |
| 15:26 | <jgraham> | odinho: We can test if you want : |
| 15:26 | <jgraham> | p |
| 15:26 | <odinho> | ^_^ |
| 15:39 | <hsivonen> | annevk: so Gecko's GB18030 impl already matches the spec and we only need to change the label stuff, right? |
| 15:41 | <annevk> | hsivonen: afaict, yes |
| 15:41 | <annevk> | hsivonen: Gecko has the 0x80 mapping which is the only "weird" thing |
| 15:46 | <jgraham> | SimonSapin: urllib[2].urlopen doesn't return a HTTPResponse object. What am I missing? |
| 15:46 | <SimonSapin> | it does on Python 3 |
| 15:46 | <SimonSapin> | well, urllib.request.urlopen() |
| 15:46 | <jgraham> | And the bug doesn't happen on python2? |
| 15:47 | <hsivonen> | annevk: ok. bug filed: https://bugzilla.mozilla.org/show_bug.cgi?id=951691 |
| 15:48 | <SimonSapin> | jgraham: apparently urllib[2].urlopen does not use httplib on Python 2, but httplib is still affected |
| 15:48 | <GPHemsley> | Gonna be updating the wiki software today. Nobody panic. :) |
| 15:48 | <jgraham> | SimonSapin: I see |
| 15:48 | <annevk> | hsivonen: should we have a separate bug on removing GBK from the menu? |
| 15:49 | <annevk> | hsivonen: thinking about it, that has the same risk (in theory) so grouping them seems fine |
| 15:49 | <jgraham> | SimonSapin: Thanks |
| 15:49 | <hsivonen> | annevk: I think we can do that in the same bug unless the label bug gets stalled somehow |
| 15:49 | <hsivonen> | annevk: I think the menu is lower risk |
| 15:50 | <hsivonen> | annevk: since users probably already can't make an informed choice and can end up submitting whatever |
| 15:50 | <annevk> | hsivonen: have you thought about "Character Encoding" -> "Text Encoding" btw? |
| 15:50 | <hsivonen> | annevk: I haven't |
| 15:50 | <hsivonen> | annevk: not my bikeshed |
| 15:50 | <annevk> | hsivonen: ait |
| 15:50 | <annevk> | Chrome uses Encoding |
| 15:52 | <annevk> | IE names it Encoding as well |
| 15:52 | <annevk> | hsivonen: would that be a bug against Firefox or some other component? |
| 15:53 | <gsnedders> | jgraham: Are you fixing the bug, then? |
| 15:53 | <gsnedders> | jgraham: Or dealing with it at least? |
| 15:53 | <hsivonen> | annevk: Firefox |
| 15:53 | <annevk> | ta |
| 15:54 | <jgraham> | gsnedders: What bug? |
| 15:54 | <gsnedders> | jgraham: the html5lib one |
| 15:54 | <gsnedders> | jgraham: that you were discussing with SimonSapin above |
| 15:56 | <jgraham> | gsnedders: He fixed it |
| 15:56 | <annevk> | https://bugzilla.mozilla.org/show_bug.cgi?id=951695 |
| 16:03 | <rwaldron> | annevk ping for quick Q |
| 16:03 | <annevk> | rwaldron: go ahead |
| 16:03 | <rwaldron> | In Boston, you mentioned an "Elements" API |
| 16:03 | <rwaldron> | IIRC, this was an Array subclass |
| 16:03 | <rwaldron> | "from the future" ;) |
| 16:03 | <annevk> | rwaldron: see links in red box http://dom.spec.whatwg.org/#elements |
| 16:04 | <rwaldron> | <3 |
| 16:04 | <rwaldron> | you read my mind |
| 16:04 | <rwaldron> | thanks! |
| 16:04 | <rwaldron> | (ie. the next question was "i can has link?") |
| 16:04 | <rwaldron> | thanks again |
| 16:04 | <annevk> | heh |
| 17:01 | <GPHemsley> | Wiki upgrade in progress. Nobody panic. |
| 17:04 | <Hixie> | woot |
| 17:04 | Ms2ger | grasps his towel |
| 17:18 | <GPHemsley> | OK, I think we're good. Let me know if you see any issues. |
| 17:57 | <SimonSapin> | jgraham: what happens after approval in Critic? |
| 17:58 | <SimonSapin> | also, should I resubmit #123 to get Critic unstuck? |
| 18:03 | <jgraham> | SimonSapin: Someone has to merge |
| 18:03 | <jgraham> | or rebbase |
| 18:04 | <jgraham> | Although I think that html5lib has a non-linear history in any case so merging might not be so bad |
| 18:04 | <jgraham> | And yes, resubmitting is probably a good idea |
| 18:22 | <gsnedders> | jgraham: I aim for history from git onwards to be linear, FWIW |
| 18:23 | <jgraham> | gsnedders: Then you have a rebase to do :) |
| 18:24 | <gsnedders> | jgraham: If you can find the one number I need for this HMRC form. |
| 18:26 | <Hixie> | when you do frames[0].location.reload(), what's the source browsing context? the parent, or the subframe? |
| 18:27 | <jgraham> | gsnedders: Isn't the answer in the question? The number you need is 1. |
| 18:27 | <gsnedders> | jgraham: I suspect I paid more tax than £1. :) |
| 18:31 | <MikeSmith> | Hixie: the subframe...? isn't that the case of frames[0].location.assign and .replace? |
| 18:32 | <Hixie> | Navigation for the assign() and replace() methods must be done with the responsible browsing context specified by the incumbent settings object as the source browsing context. |
| 18:33 | <MikeSmith> | HAL^WHixie: you forgot to add ,Dave at the end of that |
| 18:34 | <jgraham> | gsnedders: Oh, just tell them the answer is in GTU (Gsnedders Tax Units) |
| 18:34 | <gsnedders> | jgraham: The form doesn't allow that. :( |
| 18:34 | <MikeSmith> | Hixie: so given that why should not that the same be true for .reload() ? |
| 18:35 | <gsnedders> | (I mean, HMRC should /already have this number/. Do I really need to tell them it!?) |
| 18:35 | <Hixie> | MikeSmith: yeah, that's my conclusion too. i've just checked that in. previously it was just undefined. |
| 18:35 | <Hixie> | MikeSmith: it's mostly academic, really. i think it only affects if you can do parent.location.reload() when you're sandboxed allow-same-origin allow-scripts, which is a very dumb situation to be in anyway. |
| 18:36 | <MikeSmith> | oh |
| 18:36 | <MikeSmith> | well I don't plan on the doing that at least |
| 18:36 | <Hixie> | heh |
| 18:37 | <Hixie> | oh actually location.reload() explicitly uses a different source browsing context in the situation of a non-overridden reload |
| 18:37 | <Hixie> | wonder why |
| 18:37 | gsnedders | kinda wants to prove the event loop is free from deadlock/livelock |
| 18:39 | <Hixie> | wow, http://software.hixie.ch/utilities/js/live-dom-viewer/?saved=141 actually shows that's true, too |
| 18:39 | <Hixie> | at least in chrome |
| 18:55 | <Hixie> | how the heck can i retroactively figure out the source browsing context of a document.open()'ed document |
| 18:57 | <MikeSmith> | Hixie: printf |
| 18:59 | <Hixie> | printf of what, though? |
| 19:03 | <MikeSmith> | some whatever object in the browser source code and you just dump it out |
| 19:03 | <MikeSmith> | after you hack the source and recompile it |
| 19:04 | jgraham | looks for a blunt instrument to apply to mq |
| 19:09 | <TabAtkins> | Sigh. I hate it when people in non-American timezones try to ping me repeatedly over multiple days during times when America is asleep, and sign off when they don't hear anything from me. |
| 19:09 | <TabAtkins> | I can't respond! I'm asleep! You're a dumbass! |
| 19:12 | <Hixie> | yeah, it seems like if you're not going to stick around, then send e-mail |
| 19:16 | <jgraham> | TabAtkins: You should schedule irc breaks into your sleep pattern |
| 19:16 | <tantek> | TabAtkins just ask them to summarize their question and see if they bother checking the logs. If they don't, it can't have been that important right? |
| 19:16 | <TabAtkins> | That seems like a legitimate solution to the problem, yes. |
| 19:17 | <TabAtkins> | tantek: I'm getting pinged privately, so there's no logs for them to check. |
| 19:17 | <tantek> | you're getting pinged privately about *standards* questions? |
| 19:17 | <Hixie> | /away i'm asleep, please leave a detailed message after the beep. BEEEP. |
| 19:18 | <tantek> | TabAtkins perhaps an irc/pm auto-responder? "If you have a question about the web platform, please ask it in #whatwg and feel free to cc my alias" |
| 19:19 | <jgraham> | or, apparently, /away If you ask me something now and sign off before I respond I will call you a dumbass in public |
| 19:19 | <tantek> | LOL |
| 19:19 | <TabAtkins> | tantek: Yup. I'd like to tell them to just ping me in #whatwg, but I can't since they dont' stick around in public places. |
| 19:20 | <Hixie> | oh this is in IM, not IRC? |
| 19:21 | <tantek> | TabAtkins - people are pinging you about *standards* exclusively in *private* places? Ignore them until they figure it out. Or send them this for some background reading: http://tantek.com/2011/168/b1/practices-good-open-web-standards-development |
| 19:21 | <TabAtkins> | tantek: Still can't send them anything. You keep missing the fundamental difficulty I'm running into here. ^_^ |
| 19:22 | <Hixie> | (fwiw, i get the same thing.) |
| 19:23 | <jgraham> | TabAtkins: /away allows you to do that, doesn't it? |
| 19:23 | <tantek> | You two are too patient. I've long since told people to go the (appropriate) IRC channel and ask there. |
| 19:23 | <Hixie> | tantek: how? |
| 19:23 | <TabAtkins> | jgraham: That requires me to actaully remember to set /away. |
| 19:23 | <tantek> | TabAtkins - how are you getting the pings? |
| 19:23 | <TabAtkins> | tantek: HOW DO I TELL ANYONE TO PING ME ELSEWHERE WHEN THEY LEAVE BEFORE I CAN TELL THEM ANYTHING?!? |
| 19:23 | <TabAtkins> | ^_^ |
| 19:23 | <Hixie> | yeah, what tab said |
| 19:23 | <jgraham> | TabAtkins: Well, sure, that's a problem |
| 19:24 | <jgraham> | You could possibly set up your client to do it automatically after an inactivity timeout |
| 19:24 | <tantek> | if they're able to ping you then you are able to setup an /away auto-response |
| 19:24 | <tantek> | I think some clients auto-set /away after an inactivity timeout |
| 19:24 | <jgraham> | Although assuming you use IRCCloud that might not be super-easy |
| 19:24 | <TabAtkins> | tantek: Not a specialized one, and I don't want to spam people who are legit PMing me during the day with a mesage telling them to buzz off. |
| 19:24 | <TabAtkins> | jgraham: Yes, that's what I use, and to the best of my knowledge I cant' set such a thing. |
| 19:25 | <jgraham> | Well you can |
| 19:25 | <jgraham> | Just need to write an extension or bookmarket, or something |
| 19:25 | <tantek> | FWIW Colloquy has a "Sleep Message" in the settings |
| 19:26 | <TabAtkins> | tantek: Not helpful for me on a Linux box, also IRCCloud is awesome and now free through work. |
| 19:27 | tantek | is trying the Sleep Message setting |
| 19:28 | <TabAtkins> | So IRCCloud has an auto-away, but it triggers only when you leave the page in all clients. |
| 19:29 | <Hixie> | most. ridiculous. test. ever. http://damowmow.com/playground/demos/sandbox/001.html |
| 19:29 | <Hixie> | short of bidi tests. |
| 19:30 | <tantek> | oh NM - the Colloquy sleep message is triggered like a quit message when you put your computer to sleep (as opposed to actually quitting the IRC client) |
| 19:30 | <tantek> | TabAtkins - perhaps file a feature request with the IRCCloud folks to have an "after hours" setting that auto-sets you /away with a message. |
| 19:31 | <TabAtkins> | tantek: Yeah, gonna do that. |
| 22:24 | <zcorpan> | TabAtkins: just change your nick to TabAtkins_i'm_asleep_dumbass |
| 22:24 | <TabAtkins> | zcorpan: That requires the same level of forethought as setting /away. |
| 22:26 | <zcorpan> | not if you have that nick when you're awake, too :-P |
| 22:27 | <Ms2ger_i_m_aslee> | Good idea |