2012-01-01 [01:12:00.0000] happy 2012 everyone [03:50:00.0000] not a bad start: https://mail.mozilla.org/pipermail/es-discuss/2011-December/019112.html [03:55:00.0000] I wonder why ES people are so desperate to change typeof [04:00:00.0000] not everyone is, but Brendan Eich says people are being bitten by just checking for object and then assuming it has properties [04:01:00.0000] (typeof null == 'object') [04:11:00.0000] On Twitter, "Who to follow" basically became a spam list? [04:11:01.0000] E.g. for the @WHATWG account it suggests all kinds of Dutch crap to me [04:45:00.0000] Ms2ger: Because it's the most common complaint about the language by web developers. [04:52:00.0000] /me can't remember having ever used typeof in any code he's written [04:56:00.0000] /me can remember Philip` using typeof [04:57:00.0000] I see two in canvas-tests.js and three in tests2d.yaml [05:00:00.0000] Pfft, tests don't count [05:25:00.0000] Is someone maintaining a non-PDF Unicode specification? [07:11:00.0000] typeof null == "null" actually would be pretty nice. The fact that functions have typeof equal to "function" is annoying too. [07:14:00.0000] or just remove typeof operator because its text protocol/string typing, and use ===/instanceof instead [07:15:00.0000] anyway es6 will nonbackward compatible w/ es5- [07:15:01.0000] and we can do radical changes [07:20:00.0000] . . . according to Wikimedia traffic logs, WebKit almost has more market share than IE now. IE is down to 34.16% as of October. Wow. [07:20:01.0000] And Firefox is down from 30.71% in April 2009 to 23.56% in October 2011. [07:20:02.0000] /me clearly hasn't been paying attention [07:21:00.0000] thats good [07:24:00.0000] Dunno if the news about Firefox is good. [07:25:00.0000] A WebKit monoculture would sure be a lot better than an IE monoculture, but I don't think we actually want it. [07:25:01.0000] I think Firefox will increase again: the latest release has been acclaimed for its speed and they are starting to have a real web inspector [07:26:00.0000] Hope so. [07:26:01.0000] (this sentiment is not related to the fact that I started working for Mozilla today instead of Google) [07:26:02.0000] Good for you ;) [07:27:00.0000] what are you working on? [07:33:00.0000] oh congrats AryehGregor [07:45:00.0000] would be more interesting to compare webkit !mobile vs. IE than all webkit [07:50:00.0000] espadrine: i wouldn't expect that--only developers can tell one FF release from another right now, and only if they're looking for it [07:54:00.0000] congrats AryehGregor [08:22:00.0000] bga: The intention now is it will be backwards compatible with ES5-strict [08:23:00.0000] /me has certainly used typeof plenty... in test harnesses. [08:23:01.0000] I've written too many. :\ [08:25:00.0000] gsnedders Object(foo) instanceof String much better imho [08:25:01.0000] also it will be good if instanceof will work w/o Object() wrapper [08:25:02.0000] bga: But how do we get rid of string/number/boolean primitives in a backwards compatible way? [08:28:00.0000] before i should collect examples when implict Object() to any primitives will break code [08:29:00.0000] of cource you can emulate typeof via instanceof [08:30:00.0000] bga: Things rely upon the different semantics of the primitives v. their object wrappers. [08:30:01.0000] So emulating typeof doesn't work. [08:31:00.0000] but rid of text protocols from js will make CS culture of js community much better [08:32:00.0000] gsnedders i understand, thats why modern langs dont support pritives, everything is object really [08:51:00.0000] bga: Oh, I totally agree it would be a lovely change. [08:51:01.0000] There never should have been the primitive/object duality of the String/Number/Boolean. [08:52:00.0000] The situation would be much better if there was only one of the two --- regardless of which. Of course, I think most people would rather the object. [09:04:00.0000] AryehGregor, welcome :) [09:05:00.0000] espadrine, I'm working on specs. Same as before. Number one duty is still the editing spec, but that doesn't need much of my attention right now. The first thing I'm starting on other than that is writing tests for CSS transforms/animations. [09:06:00.0000] \o/ [09:06:01.0000] (for which I've never read the spec, but will soon, I guess) [09:07:00.0000] Also I'm apparently wanted to figure out exactly what we want [TreatNullAs=EmptyString] for. [09:07:01.0000] I think I'll do the CSS test stuff first. [09:13:00.0000] Ms2ger, do you have an opinion on the XXX here? http://dvcs.w3.org/hg/domcore/raw-file/tip/dom-core.html#dom-range-intersectsnode [09:14:00.0000] Or annevk or whoever. [09:15:00.0000] AryehGregor: all implementations is WebKit/Presto? [09:15:01.0000] annevk, yes. [09:16:00.0000] (Apparently Gecko used to have support, but it dropped it.) [09:16:01.0000] AryehGregor: I'd suggest asking someone from WebKit [09:16:02.0000] generally I'm in favor of less exceptions [09:16:03.0000] /me will try sometime other than New Year's Day [09:17:00.0000] fewer exceptions, damn it! [09:17:01.0000] Sorry, I must not be a grasmmar nazi. [09:49:00.0000] oh [09:50:00.0000] the sleeper spam accounts have awoken again on the wiki [09:51:00.0000] guess we-spam-wikis took a week off or so [09:59:00.0000] just delete any old account that's never done anything? [10:03:00.0000] Happy New Year, Whatwg! [10:04:00.0000] Eh, so there are two dates that dglazkov has special messages for. [10:05:00.0000] irc etiquette #1: no pointless timed messages [10:05:01.0000] zewt: It varies too much in time to be that. [11:12:00.0000] heloo every body [11:54:00.0000] ebay implementing fake scrollbars makes me sad [11:56:00.0000] doom of web [12:27:00.0000] I wonder how the incomplete CSS1 test suite used to prove that IE6 was CSS1 "compliant" was designed in the first place. [12:37:00.0000] Most of the test suites before very very recently were not much to write home about. [12:37:01.0000] In fact, most of them today also aren't. [13:20:00.0000] I believe the design of the testsuite was mostly "Hixie wrote a bunch of tests". 2012-01-02 [22:27:00.0000] AryehGregor: i did some research into [TreatNullAs=EmptyString] for opera. for methods and constructors, i think we should match the specs, and that should be mostly compatible. we don't match the spec for attributes because there was some problem with some attribute on i think [22:28:00.0000] but we've had "null" for attributes earlier and i think we can change back to that, possibly with a few exceptions if compat requires [23:24:00.0000] I'm getting an "argument to reversed() must be a sequence" when trying to run anolis [23:24:01.0000] sound familiar to anybody? [23:26:00.0000] full error output: https://gist.github.com/1549710 [00:26:00.0000] well, this is going to be a success https://www.w3.org/Bugs/Public/show_bug.cgi?id=15380 [01:21:00.0000] I've slowly come to realize multi-octet encodings can only be tackled one at a time :/ [01:21:01.0000] Today it's shift_jis [01:21:02.0000] well, cp932 [01:26:00.0000] By the end of the day you might have figured out what to call it ;) [01:29:00.0000] I suggest x-obsoleted-shift_jis [01:32:00.0000] AryehGregor: Congratulations on the job [01:33:00.0000] /me continues to work backwards through time [01:33:01.0000] i suggest you go by which has the most hits on google code search [01:34:00.0000] the name? [01:34:01.0000] the name is shift_jis [01:34:02.0000] the implementation is cp932 plus some details that are not documented currently [03:55:00.0000] what does the anolis --dump-xrefs option do? [03:56:00.0000] MikeSmith: The behaviour you complained about earlier looks like a bug to me. What did you use as input? [03:56:01.0000] It writes out a JSON file with the xrefs for other specs to use [03:57:00.0000] jgraham: that turned out to be pilot error [03:57:01.0000] Ah, not the kind that causes hundreds of people to plunge to their death I hope [03:58:00.0000] What did you do? The error message wasn't clear… [03:59:00.0000] jgraham: I was trying to run anolis from a symlink but I think it was expecting to find stuff in a relative path [03:59:01.0000] or something [04:00:00.0000] I fixed it by not running it from a symlink [04:00:01.0000] Ms2ger, so if I have "anolis --dump-xrefs=data/xrefs/dom/url.json Overview.src.html /tmp/spec", it's supposed to writ out a url.json file? [04:01:00.0000] Yes [04:01:01.0000] hmm [04:01:02.0000] if I run that, I get an error message: "[Errno 2] No such file or directory: 'data/xrefs/dom/url.json'" [04:01:03.0000] jgraham: any update on that multipage validator bug? [04:01:04.0000] Ah [04:02:00.0000] It does require a file with { "url": "..." } to be present already [04:02:01.0000] ah, OK [04:03:00.0000] ooh, did I never check in url.json? [04:03:01.0000] Don't think so [04:03:02.0000] my bad [04:03:03.0000] MikeSmith: shall I check mine in so you can run update? [04:03:04.0000] annevk: yes please [04:04:00.0000] Documentation for this stuff has been on my todo list for ages :) [04:04:01.0000] was trying to run the build for the URL spec [04:07:00.0000] annevk: I changed stuff. Did the problem get fixed? [04:08:00.0000] MikeSmith: updated Makefile and data/ [04:08:01.0000] thanks [04:08:02.0000] /me pulls [04:11:00.0000] jgraham: there was an open bug on the issue but I cannot find it [04:13:00.0000] jgraham: oh, I guess Hixie has not updated the spec either [04:13:01.0000] fyi: I'm on vacation starting next week Wednesday [04:14:00.0000] if anyone can take over WHATWG Weekly until February that would be nice [04:18:00.0000] jgraham: why again could it be a problem because of your script? because it is not a problem in single-page [04:18:01.0000] the bug is https://www.w3.org/Bugs/Public/show_bug.cgi?id=12539 btw [04:21:00.0000] Ms2ger: thanks for teh docs update [04:21:01.0000] annevk: iirc the problem was lxml didn't support the full set of named entities or something [04:21:02.0000] 'twas long overdue [04:21:03.0000] and lxml is not used for single page? [04:22:00.0000] It is used to parse it, but not serialise it. I guess Hixie doesn't use the named entities in the source [04:23:00.0000] The flow is Hixie -> single page -> multipage where each -> implies an application of the lxml parser [04:23:01.0000] ah, so with the second invocation it goes wrong [04:23:02.0000] Yeah [04:23:03.0000] thanks [05:48:00.0000] hm [05:49:00.0000] its not possible to apply many -webkit-mask to one element? [05:49:01.0000] only last is applied [05:49:02.0000] :( [06:12:00.0000] OK, my anolis now seems to be turning U+2019 into ’ [06:12:01.0000] Blame html5lib [06:13:00.0000] I should update my html5lib? [06:20:00.0000] When was it changed in the spec? [06:23:00.0000] gsnedders: I don't know that it was changed [06:24:00.0000] does the spec say which should be preferred? [06:25:00.0000] or does the spec even say anything at all about converting code points to named character references? [06:27:00.0000] MikeSmith: Well, is U+2019 the definition of ’? [06:27:01.0000] MikeSmith: What gets encoded to character references depends on the charset used. [06:28:00.0000] html5lib prefers named character references to numeric ones. [06:29:00.0000] http://www.whatwg.org/specs/web-apps/current-work/multipage/named-character-references.html#named-character-references lists both ’ and ’ as referring to U+2019 [06:31:00.0000] gsnedders, would you take a patch to prefer only those named character references that were in HTML4? [06:32:00.0000] can't we just use numeric throughout or actual references throughout? [06:32:01.0000] actual code ponits* [06:32:02.0000] points* [06:32:03.0000] How about ponies? [06:33:00.0000] I vote for preferring numeric throughout [06:38:00.0000] MikeSmith: You can do that [06:38:01.0000] oh? [06:38:02.0000] can I do it from anolis? [06:41:00.0000] …maybe [06:41:01.0000] ask gsnedders or Ms2ger :p [06:41:02.0000] Looking [06:43:00.0000] jgraham, resolve_entities? [06:44:00.0000] Yes, I think so [06:44:01.0000] That's an awful name [06:44:02.0000] Unless you have version control evidence to the contrary, I didn't invent it [06:46:00.0000] changeset: 1516:3f6b82fe7e8e [06:46:01.0000] user: jgraham [06:46:02.0000] date: Sat Mar 13 21:48:00 2010 +0100 [06:46:03.0000] summary: Work with lxml entities [06:46:04.0000] (Carefully omitting "based on patch by fantasai") [06:46:05.0000] Heh [09:40:00.0000] Opera has something similar to https://bug631751.bugzilla.mozilla.org/attachment.cgi?id=511429 [09:41:00.0000] though I don't think we check over 1024 octets [09:42:00.0000] in any event, that is content without a label [09:42:01.0000] not sure how much of that is in scope for the encoding spec [09:56:00.0000] FYI: For the encoding standard I'm going to flip the default of algorithms [09:56:01.0000] FYI: They terminate when something returns [09:56:02.0000] (minus second FYI) [09:57:00.0000] Great [09:57:01.0000] and I guess I will slowly do it elsewhere too [09:57:02.0000] e.g. DOM [09:58:00.0000] in the cases where an algorithm does not terminate when returning, we always say it explicitly anyway [09:59:00.0000] Yeah, because that is pretty odd behaviour [10:30:00.0000] here's a start with shift_jis http://dvcs.w3.org/hg/encoding/raw-file/tip/Overview.html#shift_jis [10:30:01.0000] anyone a good idea how to represent an array consisting of digits in a standard? ordered list made compact somehow by CSS? [10:30:02.0000] also ideas for external data tables are welcome too [13:51:00.0000] is there a spec for webrtc server? 2012-01-03 [23:26:00.0000] annevk: it looks weird to mix decimal and hex [23:28:00.0000] annevk: also, editorial, − looks better than - [23:34:00.0000] do we need to have serializers for all encodings? [23:56:00.0000] zcorpan: is there a better +? [23:56:01.0000] zcorpan: and mixing decimal and hex makes sense there [00:05:00.0000] + is fine [00:06:00.0000] how does it make sense? [00:06:01.0000] one is a code point and the other is an index [00:06:02.0000] in an array of 94n length [00:08:00.0000] ↪0x81 [00:08:01.0000] 8648 [00:08:02.0000] ↪0xA0 [00:08:03.0000] 0xF8F0 [00:08:04.0000] how do you explain this, andrew? [00:09:00.0000] the first is an index into a table, the second is a code point? [00:10:00.0000] 8648 = 92*94 [00:10:01.0000] though I think that has to be come 188 lower (all of them) [00:10:02.0000] and 0xF8F0 is not an index? [00:10:03.0000] no [00:11:00.0000] ok, then i'm confused [00:11:01.0000] "The shift_jis lead octet index for a given octet is:" [00:11:02.0000] I didn't want to name it the shift_jis lead octet index or code point for a given octet [00:12:00.0000] but I could I suppose [00:13:00.0000] ok, i see now in the to Unicode algorithm that 0xF8F0 and above will equal the code point [00:13:01.0000] and the second should prolly be named shift_jis offset rather than trail octet index [00:14:00.0000] but i don't see why that should mean that it's fine to mix decimal and hex [00:14:01.0000] also 0xE000 [00:15:00.0000] before the to Unicode algorithm, it's just an opaque number [00:16:00.0000] I think this is easier if you want to review what the algorithm does [00:34:00.0000] I wonder how bad an idea it would be to autodetect UTF-8 vs. locale-specific legacy encoding by seeing if the first non-ASCII byte and up to 3 bytes after it form a valid UTF-8 byte sequence [00:35:00.0000] that is, are there valid UTF-8 byte sequences that can reasonably occur in single-byte encodings? [00:35:01.0000] What is the problem? [00:35:02.0000] annevk: letting people publish UTF-8 content while being too inept to declare it [00:37:00.0000] there are some legacy encodings that have valid octets for the whole range [00:37:01.0000] e.g. windows-1252 [00:37:02.0000] or windows-1254 [00:37:03.0000] 1256 too [00:37:04.0000] etc. [00:38:00.0000] annevk: yes, but in the case of Windows-1252, UTF-8 byte sequences don't seem to result in sensible natural language when decoded as Windows-1252 [00:38:01.0000] dunno about 1254 and 1256 [00:38:02.0000] doesn't opera have heuristics to detect utf-8? [00:38:03.0000] zcorpan: interesting. from how many bytes? [00:39:00.0000] no idea [00:39:01.0000] hsivonen: What about pages that concatenate a small amount of UTF-8 content and a large amount of legacy-encoding content? [00:39:02.0000] They'll definitely look like UTF-8 at first, but that might not be the desired way to decode them [00:40:00.0000] Philip`: well, yeah, such pages are broken no matter what, but the degree of perceived brokenness could vary, sure [00:43:00.0000] anyway, adding UTF-8 autodetection would not advance the goal Faruk had at the start of the thread [00:44:00.0000] adding UTF-8 autodetection would only allow UTF-8-using authors to declare less [00:44:01.0000] and only if you happen to use it in the first 1024 octets? [00:48:00.0000] annevk: no, anywhere [00:49:00.0000] annevk: the bytes before the first-non-ASCII byte can be decoded as US-ASCII, so there's no need to stall processing except to wait for the 3 bytes after the first non-ASCII byte when the first non-ASCII byte is seen [00:50:00.0000] annevk: anyway, it's not at all clear that the problem this would solve is worth solving [00:52:00.0000] oh [00:52:01.0000] conceptually I guess it's not that complicated, but it's still kind of sucky [01:15:00.0000] annevk: (Re: old IRC logs) It's very likely that there are bugs in Gecko's charset alias list [01:50:00.0000] hsivonen: apparently we detect as utf-8 if there are 10 or more valid utf-8 byte sequences and no invalid utf-8 byte sequences [01:52:00.0000] zcorpan: At the start? [01:54:00.0000] zcorpan: I take it that valid byte sequences mean multibyte sequences so that 10 ASCII bytes don't count. [01:55:00.0000] zcorpan: can there be an arbitrary number of ASCII bytes between the UTF-8 byte sequences? [01:55:01.0000] right [01:55:02.0000] zcorpan: how does incremental parsing and rendering work before the sniffer has made its decision? [01:56:00.0000] *shrug* [01:57:00.0000] zcorpan: that's the most interesting part! [01:59:00.0000] if it's very interesting, i guess you could reverse engineer it :-P [02:01:00.0000] apparently we reparse if the detector makes a decision late [02:02:00.0000] zcorpan: ok. Thanks. I want to avoid introducing reparsing to locales that don't already have it as much as possible. [02:03:00.0000] I tried to remove reparsing from the locales that have it in the Firefox 4 cycle, but then the Japanese Planet Debian broke and people got worried [02:04:00.0000] the Japanese Planet Debian has fixed its encoding declaration since then [02:05:00.0000] would be really nice if reparsing could be nuked [02:05:01.0000] also, I want to avoid making the encoding determination timing-dependent [02:05:02.0000] people seem to believe that our feedback loop from Japan is broken, so decisions get made based on rumors and existence proof [02:19:00.0000] wow, an email from 2002 just made it to the list [02:19:01.0000] oh sorry, that was part of the email address [02:26:00.0000] But the content was about DTD so I can understand your confusion! [02:26:01.0000] do we have a FAQ entry about DTDs? [02:29:00.0000] http://wiki.whatwg.org/wiki/FAQ#Under_what_conditions_should_a_DOCTYPE_be_used_in_XHTML.3F - i guess not enough [02:30:00.0000] though it links to validator.nu faq, which may be enough [03:53:00.0000] are [TreatUndefinedAs=EmptyString] or [TreatUndefinedAs=Null] actually used? [03:54:00.0000] Yes [03:54:01.0000] In HTML/Obsolete, IIRC [03:54:02.0000] Oh [03:54:03.0000] Undefined? Not sure [03:55:00.0000] no hits [03:55:01.0000] in HTML anyway [03:56:00.0000] Mozilla does support it, IIRC [03:56:01.0000] annevk: why aren't UCS2 names defined as aliases of UTF-16? [03:57:00.0000] annevk: so is the plan to make BOM override HTTP so that UTF-16BE with little-endian BOM is decoded as little-endian? [03:58:00.0000] annevk: fwiw, in case there are bugs in how Gecko handles BOMs in HTML, it's because I was naive enough to assume that the decoder library worked per spec [03:58:01.0000] hsivonen: are they in any browser? UTF-16 labels are not really done yet, every browser has them done differently [03:58:02.0000] hsivonen: Basically yes, BOM "rules" [03:59:00.0000] annevk: by source inspection, it seems that Gecko supports crazy UCS2 aliases [03:59:01.0000] annevk: IIRC, a legacy SonyEricsson browser did, too [03:59:02.0000] not that it matters [04:00:00.0000] hsivonen: I think what put me off was that Gecko added them to BE [04:00:01.0000] UTF-16 is one of those things that must have seemed great ideas before implementation experience [04:00:02.0000] hsivonen: not UTF-16 [04:00:03.0000] annevk: UCS2 was BE in theory, no? [04:01:00.0000] oh okay [04:01:01.0000] I think per http://wiki.whatwg.org/wiki/Web_Encodings only Gecko had those labels [04:01:02.0000] but I'm fine with adding labels [04:02:00.0000] oh no, so Opera has those labels too, but added to UTF-16 [04:02:01.0000] annevk: if the Web doesn't need them and smontagu agrees, I'm OK with getting rid of those labels [04:02:02.0000] but I guess with UTF-16 now defaulting to LE that does not make sense [04:02:03.0000] annevk: the question is whether there's legacy content [04:02:04.0000] hsivonen: IE has this " iso-10646-ucs-2, ucs-2, unicode, utf-16, utf-16le" as labels for utf-16 [04:02:05.0000] annevk: not what makes sense [04:02:06.0000] agreed [04:03:00.0000] annevk: so iso-10646-ucs-2 is LE in IE? [04:03:01.0000] I would expect so based on that data [04:07:00.0000] inspired by the thread Faruk started, a console warning about form submissions using encodings that can't represent all of Unicode has landed on mozilla-central and will be in the next Nightly build [04:07:01.0000] I wonder if it has any impact on anything at all [04:07:02.0000] I hope it does [04:08:00.0000] and I hope people don't complain too much about Firefox whining about bad authoring practices [04:08:01.0000] And that Nightly also won't support geostd8 [04:08:02.0000] I thought those changes were great [04:09:00.0000] Ms2ger: is geostd8 an encoding? [04:09:01.0000] was :p [04:09:02.0000] Yes, something Georgian nobody supports [04:09:03.0000] hooray! [04:16:00.0000] I think the way the BOM stuff should work by the way is that you say "decode X using label Y" and that algorithm first checks if X has a BOM and then picks utf-8/utf-16/utf-16be if there is a BOM, and otherwise uses Y to pick an encoding [04:18:00.0000] annevk: what's your plan for integrating format-specific internal encoding declarations to the algorithms? [04:19:00.0000] well before HTML starts decoding it first does this search through a bunch of octets [04:19:01.0000] and if finds a label there it can then use that [04:19:02.0000] annevk: FWIW, even though I previously WONTFIXed a theory-based bug about the precedence of the BOM, I'm OK with changing the precedence now that there are examples of real pages in the wild where it matters [04:19:03.0000] cool [04:27:00.0000] heh, found an old page by Lachy http://wiki.whatwg.org/wiki/Character_Encoding_Detection [04:28:00.0000] I don't remember writing that. [04:29:00.0000] the Mozilla observations look weird, but I'm willing to believe the old parser did weird things [04:29:01.0000] hmm. Did it in December 2006. That's 5 years ago [04:31:00.0000] http://wiki.whatwg.org/wiki/StringEncoding seems kind of nonsensical to me to expose that many encodings to script authors [04:32:00.0000] especially with the addition of allowing user agents to allow arbitrary encodings [04:34:00.0000] annevk: that seems bad [04:34:01.0000] annevk: has it been proposed on a mailing list? [04:34:02.0000] dunno, I don't remember seeing it [04:34:03.0000] apart from on the wiki that is [04:35:00.0000] annevk: the API design is questionable, too [04:35:01.0000] it turns out that it's really hard to get the API right [04:36:00.0000] iconv gets the API wrong [04:36:01.0000] the Gecko-internal decoders get the API wrong [04:36:02.0000] Java gets the API right [04:37:00.0000] I hope the Encoding Standard gets it right, though it does not expose it to that level of detail currently [04:37:01.0000] And Java gets the performance wrong? :) [04:37:02.0000] so if you see a design that doesn't have the characteristics of the Java NIO API, it makes sense to suspect there's an API design error [04:37:03.0000] Ms2ger: not that I'm aware [04:37:04.0000] Then how do I get to whine about Java? :) [04:37:05.0000] pointer to that API? [04:42:00.0000] ICU4C seems to get the API right [04:43:00.0000] annevk: http://docs.oracle.com/javase/6/docs/api/java/nio/charset/CharsetDecoder.html [04:43:01.0000] annevk: the key being that it's necessary to be able to signal "no more input" to the decoder [04:44:00.0000] iconv and Gecko fail at that [04:44:01.0000] for example, if a file ends with an incomplete UTF-8 byte sequence, you don't see a REPLACEMENT CHARACTER for the trailing incomplete sequence in Firefox [04:45:00.0000] ah yes [04:45:01.0000] I introduced a "EOF octet" concept for that [05:08:00.0000] though I have not made shift_jis emit an explicit U+FFFD in such cases [05:08:01.0000] neither Gecko nor IE did that... [05:34:00.0000] gecko and presto don't emit U+FFFD for trailing half utf-8 byte sequence, but both consider that a bug [05:34:01.0000] howdy [05:34:02.0000] is it likely that the HTML5 specification will be finalized without specifying a standardized video and audio format? [05:34:03.0000] (except opera does it right in websockets because i tested for that and emoller fixed it for websockets) [05:35:00.0000] eightfold: Yes [05:35:01.0000] Well no [05:35:02.0000] The specification will never be finalized [05:35:03.0000] Except in some meaningless W3C way [05:36:00.0000] As long as people are still implementing it, the HTML standard will keep evolving along a continuum [05:36:01.0000] jgraham: yes, but it will be finalized the meaningless W3C way? [05:36:02.0000] annevk: how about we make the URL spec a WebApps deliverable instead? [05:36:03.0000] "finalized the" = "finalized in the" [05:37:00.0000] MikeSmith: what about the charter? [05:37:01.0000] The video codec won't be mandated in whatever W3C pushes to Rec. [05:37:02.0000] MikeSmith: maybe we should first write the spec and then see who wants to publish it... [05:38:00.0000] annevk: working on that now [05:38:01.0000] sweet [05:38:02.0000] about the charter, I guess we'd need to propose it [05:38:03.0000] zcorpan: interesting [05:38:04.0000] /me is playing with euc-jp decoding [05:38:05.0000] it's somewhat more sane than shift_jis, but more complex [05:38:06.0000] eightfold: Meaningless finalisation will require some impossible tasks to be completed (like a comprehensive test suite and comprehensively interoperable implementations), so it'll depend on whether those requirements get dropped or ignored or whether they block progress forever [05:38:07.0000] jgraham: what is the "Rec."? recommendation? [05:38:08.0000] Yes [05:39:00.0000] Ignored, I guess [05:40:00.0000] but does anyone here believe there will ever be a common video format among browser vendors? [05:40:01.0000] i hope there will be! [05:41:00.0000] but who doesn't [05:41:01.0000] yes, i sure do too. [05:41:02.0000] No point in letting unreasoble requirements get in the way of celebrating finishing a legacy technology [05:42:00.0000] microsoft and apple doesn't seem to be very eager to implement webm support. at least i realized the >ie9 solution seems to simplify the installation of a browser plugin. [05:42:01.0000] /me will brb [05:44:00.0000] WebM support for IE9 isn't really a browser plug-in [05:45:00.0000] it's an extension component for the system-wide media framework [05:45:01.0000] different layer of extension [05:45:02.0000] different API [05:53:00.0000] https://github.com/jterrace/js.js [05:53:01.0000] %) [06:07:00.0000] "It currently takes about 40 seconds to run print(1+1)." [06:16:00.0000] hsivonen: that was what i was trying to say, but didn't. :) what does the process look like to the end user? i don't have a windows installation handy. [06:27:00.0000] is there any chance w3c will recommend h.264 in the html5 spec, given it's proprietary nature? [06:27:01.0000] its [06:30:00.0000] eightfold: I guess only if h.264 becomes royalty-free [06:33:00.0000] smaug____: MPEG-LA has said that it will be forever, but only for end users, right? so software that implements h.264 still has to pay? [06:33:01.0000] "it will be" = "it will be free" [06:35:00.0000] that's the same thing, heh [06:36:00.0000] oops [06:37:00.0000] eightfold: it must be free for implementors [06:38:00.0000] smaug____: for w3c to recommend it, you mean? [06:39:00.0000] there are endless places they can drop in licensing (encoders, decoders, per type of media encoded, for streaming, etc) [06:40:00.0000] eightfold: there are w3c members which certainly won't accept recommendation which includes non-free parts [07:36:00.0000] eightfold: to get WebM for IE9, you go to a Web page, download an installer .exe and run it as administrator [07:36:01.0000] eightfold: the experience is the usual Windows software installation experience [07:37:00.0000] eightfold: MPEG-LA wants royalties from encoders, decoders and from a subset of people who distribute bits arranged with an H.264 encoder [07:38:00.0000] eightfold: for H.264 to be properly royalty-free, it needs to be royalty-free to decode, to encode and to transfer [07:38:01.0000] if you ever hear the MPEG-LA offering anything for "free", double-check--the entire *point* of the MPEG-LA is to license MPEG stuff for money [07:39:00.0000] they're not going to do anything just to be nice [07:39:01.0000] eightfold: where encode and decode are shorthand for making, distributing and running decoder/encoder [07:41:00.0000] (though I could just almost see them offering a hamstrung profile for free, just to try to cripple competing free codec development) [07:58:00.0000] zewt: There are plenty of members of MPEG-LA and the MPEG WG who want there be a RF standard. Apple is one of the most vocal. [07:59:00.0000] (And don't think Apple are just saying that after the Ogg/Theora stuff going everywhere: they were vocal about this at the time H.264 was being standardized a decade ago.) [07:59:01.0000] gsnedders: they were vocal about MPEG-4 (Visual) Simple Profile before H.264 [08:00:00.0000] And I believe about stuff before that too. [08:00:01.0000] But yeah, that there are patent holders pushing for a RF baseline isn't new, but it's no more likely to happen now than a decade ago. [08:01:00.0000] Also, as foolip has pointed out multiple times, without an audio codec and the container becoming RF, it doesn't matter much as you don't gain compat. with existing toolchains. [08:01:01.0000] and of course it would be a reduced profile [08:01:02.0000] gsnedders: so much so that the release of QuickTime 6 was delayed, because Apple tried to get the stuff made royalty-free [08:02:00.0000] (of course, when you have shown you have the software ready to ship, it's not particularly credible to withhold it for a short while. they know you are going to ship it) [08:02:01.0000] gsnedders, if apple cares about royalty free standards now, why haven't they supported vorbis when everyone else does? explain. [08:03:00.0000] "Apple agrees with paying a reasonable royalty for including MPEG-4 codecs in QuickTime, but does not believe that MPEG-4 can be successful in the marketplace if content owners must also pay royalties in order to deliver their content using MPEG-4." [08:03:01.0000] gsnedders: you'd gain compat with existing chips [08:03:02.0000] hsivonen: True, but I get the impression that stuff is starting to move on mobile to where it is on TVs: having DSPs reprogrammable. [08:04:00.0000] erlehmann: There are plenty of reasons why they might not have done so; it's hard to speculate. [08:04:01.0000] aren't they already? see the OMAP work. [08:04:02.0000] gsnedders: reprogrammable by whom [08:04:03.0000] gsnedders, i thought you might be more in the know than me. [08:04:04.0000] erlehmann: Most mobile devices ship with locked-down DSPs. Certainly OMAP devices in general have unlocked DSPs, not sure about mobile applications of them. [08:04:05.0000] hsivonen: Anyone whom the OS allows access to the chips to. [08:05:00.0000] gsnedders: if the chip vendor gives you enough documentation and toolchain [08:05:01.0000] hsivonen: Sure, but for most Android devices that's probably enough. [08:06:00.0000] hsivonen: Because WebM is now part of the standard Android impl [08:06:01.0000] gsnedders: does Android let apps push code to the DSP? [08:06:02.0000] hsivonen: I believe not. But I believe everyone who does WebM decoding uses the Android's native support. [08:06:03.0000] hardware on android tends to be hidden behind magic proprietary .so's or hidden away on the locked-down radio side of things [08:06:04.0000] it'd be the vendor who would have to implement it [08:07:00.0000] gsnedders: is it so much part of the standard Android impl that Opera is going to let it show in canPlayType? [08:07:01.0000] (Though our experience has been WebM support often interacts badly with hardware-accelerated drawing in a lot of cases.) [08:09:00.0000] (i.e., WebM's support in Android on many devices is buggy) [08:10:00.0000] Apparently it often draws in the wrong place, for example. [08:14:00.0000] heh, second time google search has wonked up the cursor position while i was typing a search [08:19:00.0000] hsivonen: Somehow Google needs to better ensure Android devices don't have horrible bugs like that :\ [08:20:00.0000] they can't even get android devices to use a standard button layout, so ... [08:43:00.0000] erlehmann: A few possible reasons: they don't feel their customers require it; they don't think another format that exists for the sake of being RF is benefical and would rather an existing format; they think the patent risk of Theora is too high, etc. [08:44:00.0000] gsnedders, i was more talking about vorbis. [08:44:01.0000] which is usable almost everywhere except when apple software comes into the picture [08:45:00.0000] erlehmann: Vorbis is a more interesting question, and less clear why MS/Apple's native frameworks don't support it. [08:47:00.0000] gsnedders, … says a person working for a company that would take a sizable hit if it had to pay royalties for media decoding, opposed to apple and ms. [08:49:00.0000] reducing competition is an incentive. [08:52:00.0000] good morning, Whatwg! [08:53:00.0000] gsnedders, or wouldn't it? how high are royalties? [08:53:01.0000] erlehmann: Apple and MS both pay more for licensing than they recieve from them, AFAIK [08:53:02.0000] erlehmann: Apple certainly does. [08:54:00.0000] Apple certainly has little financial interest in MPEG-4 being used, and have pushed for everything in MPEG-4 they hold patents on to be RF. [08:55:00.0000] as far as I'm concerned, anything MPEG being RF is a bad thing (for the rest of the world) [08:56:00.0000] zewt: Why? [08:56:01.0000] it will slow the adoption and development of truly free codecs [08:56:02.0000] gsnedders, that is not a point. but 81 million NOK being around 13.5 million US$ may be a point. [08:57:00.0000] erlehmann: huh? [08:57:01.0000] apple income is measured in *billions* of dollarites [08:57:02.0000] gsnedders, i just looked up how big the companies are and figured that licensing media codecs hits a small company more than a big. you tell me if i am correct. [08:58:00.0000] „big“ as in „operating income“ [08:58:01.0000] erlehmann: The cost of licensing it to ship encoders and decoders will stay for a while regardless; there's too much existing content out there, and the fact that DVDs and Blu-Ray require the decoders (and encoders, if you want to author content, which bundled software allows) [08:58:02.0000] erlehmann: Right, yeah. [08:59:00.0000] if the cost for h.264 stuff is capped at 6.5 million, that it as least the same ballpark. scary. [09:00:00.0000] The licensing fees aren't a major issue to Apple, but nor is the income they get at all notable. The only financal reasons not to ship more codecs is developer time and patent risk. [09:00:01.0000] hahaha, doctor evil http://obamapacman.com/wp-content/uploads/2010/02/Dr-Evil-H.264-5-million-dollars-licensing-myth.jpg [09:01:00.0000] gsnedders, are you assuming that it is not a strategic benefit to make it harder to enter a market? [09:01:01.0000] >obamapacman.com [09:01:02.0000] wat [09:01:03.0000] erlehmann: Which markert? [09:01:04.0000] harry potter. OBAMA. [09:01:05.0000] *market [09:02:00.0000] gsnedders: it's not hard to contrive other possible financial reasons (whether they're actual considerations or not, only they know) [09:02:01.0000] Practically, I can't think of any market for which it is a strategic benefit. [09:02:02.0000] gsnedders, any market. if company A can afford to pay for patents more easily than company B, a competitor, an open standard may not be in the best short-term-interest of the shareholders of company A. [09:02:03.0000] gsnedders: You seem to be missing the point. Irrespective of whether Apple profits directly from licensing fees, requiring browser implementors to support h.264 would be like regressive taxation [09:03:00.0000] In that it would disproportionately affect the poorer players [09:03:01.0000] for example, everyone using codecs that cost money could increase the cost--slightly--of creating competing phones [09:03:02.0000] jgraham: But Apple practically isn't competing with other browser vendors, esp. on iOS. [09:03:03.0000] gsnedders: Huh?! [09:03:04.0000] gsnedders, what? apple is hindering them for the benefit of its own ecosystem. see the crippling of the audio element API. [09:04:00.0000] I agree that they are trying to have an unlevel playing field [09:04:01.0000] which is a strategic benefit. gsnedders, i do not see where you are going here. [09:04:02.0000] But they clearly compete with mini on iOS and with other platforms on mobile [09:04:03.0000] erlehmann: I'd argue that was not wanting the web platform competing with native applications, not a fear about browser marketshare per-se. [09:06:00.0000] jgraham: And Mini is going for a sufficently different segment of the market when it comes to video support, for example, that it isn't really competition in this context. [09:08:00.0000] Basically, my point is that for iOS/iPod the market is so full of competitors that already support the same formats (excluding Apple's DRM) that they have no competitive advantage there, as allowing more competitors into the market likely would not be a massive threat to their marketshare; on the desktop it possibly makes F/OSS OSes more plausible, but most users there already install the patent-encumbered codecs anyway, it doesn't give them a comp [09:34:00.0000] I wonder when Opera is going to have the compute power to do video transcodes on the Mini servers [09:35:00.0000] hsivonen: A few customers have video transcoding done for custom Mini-based products, FWIW [09:36:00.0000] /me wonders what the average screen resolution of Mini users is [10:06:00.0000] gsnedders: I entirely disagree with the notion that Mini doesn't compete with Mobile Safari. [10:07:00.0000] jgraham: It does in general. It doesn't compete for people wanting to view YouTube, for example. [10:07:01.0000] That's all I mean. [10:08:00.0000] It doesn't compete because it doesn't support the one prerequisite thing: video support. [10:24:00.0000] gsnedders: do you think that is more or less likely to change if usable video implies paying tens of millions of NOK a year to MPEG-LA? [10:26:00.0000] jgraham: No real change. [11:08:00.0000] jgraham, lol. [11:08:01.0000] tens of millions of NOK sounds like EVE Online [12:13:00.0000] http://www.nytimes.com/2012/01/02/opinion/krugman-nobody-understands-debt.html "But it also revealed something else: when people in D.C. talk about deficits and debt, by and large they have no idea what they’re talking about — and the people who talk the most understand the least." [12:13:01.0000] sounds like your average standards meeting [12:57:00.0000] s/in D\.C\. //;s/deficits and debt/anything/ [12:58:00.0000] so the writer of that article has @nytimeskrugman as twitter account, but has not posted said article there :/ [12:58:01.0000] I've been spending about five minutes trying to research the most appropriate tweet to retweet [12:58:02.0000] AryehGregor: What are you doing for Moz now? [12:59:00.0000] (@nytimes has not posted it either as far as I can tell) [12:59:01.0000] TabAtkins, right this very second, I'm trying to figure out where I'm supposed to call for a one-on-one chat with someone. [12:59:02.0000] /me is about to just take a guess [12:59:03.0000] In general, I'm working on a bunch of stuff, starting with tests for CSS Transforms. [13:16:00.0000] annevk: Who's taken over MQ? [13:19:00.0000] Oh, media queries, not mercurial queues [13:19:01.0000] Florian I think [13:19:02.0000] and Sylvain [13:19:03.0000] Thanks! [13:30:00.0000] I like to mention how IE8 don't support 10 years old web technology: [13:30:01.0000] http://news.ycombinator.com/item?id=3420948 [14:44:00.0000] Hixie, heycam|away: I'm trying to figure out what the resolution to https://www.w3.org/Bugs/Public/show_bug.cgi?id=8241 is. [14:44:01.0000] that whole discussion is way over my head [14:45:00.0000] I'm just trying to figure out what the resolution is for the sake of fixing http://code.google.com/p/chromium/issues/detail?id=80591 [15:09:00.0000] ojan, the upshot is that the var statement will create a new (unconfigurable) own property on window, with value undefined, that shadows the named property [15:11:00.0000] ojan, that's due to the use of [ReplaceableNamedProperties] on the Window interface, and the order of resolution of named properties vs own and prototype properties in http://dev.w3.org/2006/webapi/WebIDL/#getownproperty [15:12:00.0000] ojan, also in the ecmascript spec it's defined that var statements in global code do a Object.defineProperty(globalObject, varName, { configurable: false, value: undefined }) [15:16:00.0000] heycam: ok, so
should alert undefined? [15:16:01.0000] ojan, yes [15:16:02.0000] heycam: great. i'll file the webkit bug. [15:16:03.0000] heycam: thx [15:16:04.0000] ojan, actually let me just confirm one thing first [15:17:00.0000] ojan, ah maybe that is not the case, but I think that is the preferred behaviour [15:18:00.0000] ojan, I think I might need to confirm some things with the ES group, since I'm having trouble understanding some aspects of the spec wrt variable declarations [15:19:00.0000] /me will mail public-script-coord and cc es-discuss [15:20:00.0000] heycam: thx. i would definitely like that to be the end result [15:20:01.0000] heycam: webkit is willing to change...we just want the behavior clearly specced to avoid changing multiple times [15:20:02.0000] ojan, sure [15:20:03.0000] I'll CC you on the mail [15:21:00.0000] heycam: the current webkit behavior is clearly suboptimal 2012-01-04 [16:16:00.0000] would anyone disagree with defining a "stable specification" as one which will not change in backwards-incompatible ways [16:17:00.0000] (compared with other metrics, like "not changing very fast") [16:17:01.0000] zewt: I don't think so. [16:18:00.0000] ojan: I'm pretty certain the spec says undefined, quite clearly. [16:18:01.0000] the context is when it's reasonable to deprefix implementations--I'd say when the spec is stable (by the above definition), the implementation implements a usable subset of that spec, and doesn't implement anything not in the spec (unless it's prefixed itself) [16:19:00.0000] But when is it stable by the above definition? That's hard to judge :) [16:19:01.0000] the judgement of the editor, i'd say [16:50:00.0000] gsnedders, see my mail that I just sent to public-script-coord -- I think "var" without an assignment won't shadow the named property, unfortunately [16:51:00.0000] heycam: No, it's almost 1am and I have better things to do :P [16:51:01.0000] gsnedders, as if! [16:52:00.0000] I have friends! [16:52:01.0000] Real ones, too! [16:53:00.0000] get out. [19:40:00.0000] other than the proposals for an inputmode-like attribute and translate attribute, are there any other proposals for new attributes that have much traction? [22:54:00.0000] wait, so chrome also detects utf-8? [22:59:00.0000] /me can't reproduce [22:59:01.0000] /me finds from reverse engineering that opera needs 11 valid byte sequences [23:01:00.0000] oh, i need to choose "auto detect" in chrome [23:11:00.0000] Hixie: http://lists.whatwg.org/listinfo.cgi doesn't work [00:12:00.0000] annevk: can you give me write access to the https://bitbucket.org/ms2ger/specification-data repo? [00:12:01.0000] or do I need to bug Ms2ger? [01:05:00.0000] despite what the HTML spec claims, it seems to me its current definition of "URL" does not violate anything in RFC 3986 [01:07:00.0000] what's your bitbucket account MikeSmith? [01:07:01.0000] MikeSmith: oh, it seems Ms2ger has to do it, I'm no admin for that repository [01:08:00.0000] Hmm? [01:08:01.0000] Ms2ger: add MikeSmith to specification-data or me as admin [01:08:02.0000] annevk: sideshowbarker [01:09:00.0000] context is that for the URL spec, I added some specs to references.json [01:09:01.0000] and some terms to xrefs/network/url.json [01:09:02.0000] Should all be admins now [01:09:03.0000] thx [01:11:00.0000] sweet [01:12:00.0000] in other news, Opera will be removing support for windows-sami-2, tcvn, and viettcvn [01:12:01.0000] not sure what release [01:12:02.0000] what's tcvn? [01:12:03.0000] a vietnamese encoding? [01:14:00.0000] /me attempts a push [01:18:00.0000] MikeSmith: yeah [01:20:00.0000] OK, just pushed https://bitbucket.org/ms2ger/specification-data/changeset/09fda5dc9e13 [01:21:00.0000] I noticed that there is a RFCs.json file, but I assumed that's just there for purposes of being copied from [01:21:01.0000] so I copied data for the needed RFCs over from that to references.json [01:22:00.0000] needed ones being RFC3987, RFC3986, and RFC2781 [01:22:01.0000] for IRIs, URIs, and UTF-16 [01:23:00.0000] which are needed for defining what a valid URL is - http://www.whatwg.org/specs/web-apps/current-work/multipage/urls.html#terminology-0 [01:23:01.0000] Yes, the RFC files is just the data I used to generate HTML's references [01:24:00.0000] MikeSmith: I think we might end up with a slightly more complicated definition [01:24:01.0000] MikeSmith: or simpler even, because that definition is quite specific to HTML [01:25:00.0000] MikeSmith: e.g. in XMLHttpRequest or CSS usage it does not matter what the resource is encoded as [01:25:01.0000] annevk: OK [01:26:00.0000] ... are two different URLs in a windows-1252 encoded resource... [01:26:01.0000] but for now I am really just at the point of copying over what we need from abarth URL-parsing draft and that Terminology part of the URL section of the HTML spec [01:27:00.0000] kk [01:27:01.0000] I suppose I ought to get this pushed to dvcs.w3.org at some point [01:28:00.0000] annevk: maybe you can do a sanity check on what I have at this point? [01:28:01.0000] before I push it [01:28:02.0000] MikeSmith: nah just go for it [01:29:00.0000] MikeSmith: I think we're at least a month and a half away from me having time [01:29:01.0000] (for working with you on it) [01:30:00.0000] btw, Opera is currently not removing the viscii encoding as there are email clients out there using it, including Gecko [01:30:01.0000] also single-octet, Vietnamese [01:32:00.0000] \o/ [01:34:00.0000] sorry not "using", supporting [01:35:00.0000] and if it was my call we would remove it, but Peter wants to tread carefully [01:45:00.0000] haha, via ojan http://www.astrodigital.org/space/stshorse.html [01:46:00.0000] comment along with it: "This accurately describes how many features of the web platform got where they are today." [01:52:00.0000] Ms2ger: aah [01:53:00.0000] Ms2ger: I guess I can take a stab at updating your branched copy [01:54:00.0000] I'm looking [01:54:01.0000] ooh, that's in a separate branch? [01:54:02.0000] kk [01:58:00.0000] is anything on the web platform actually derived from a horse's ass? [02:03:00.0000] Little known fact: The glue that holds the platform together is derived from boiled horse arse. [02:37:00.0000] Ms2ger: btw, Berjon tells be you were asking about write access on the dvcs testframework repos [02:37:01.0000] I will set that up for you as soon as i get back home (I'm on a train now getting close to Shinjuku station and need to drop off in 5 minutes) [02:45:00.0000] MikeSmith: I forgot to ask you about the book [02:45:01.0000] oh shit [02:46:00.0000] yeah [02:46:01.0000] MikeSmith: are you in the office tomorrow and can you send it then? [02:46:02.0000] probably not going to the office but I will ask somebody there to package it up and send it [02:46:03.0000] I think somebody will be there tomorrow [02:47:00.0000] today was actually still pretty much a holiday here -- most people won't be back to work til tomorrow [02:48:00.0000] more holidays than golden week in Japan? [02:48:01.0000] crazy [03:01:00.0000] MikeSmith: ta [03:39:00.0000] should I file https://bugzilla.mozilla.org/show_bug.cgi?id=690225#c30 as a new bug? [03:39:01.0000] or just be patient? [03:40:00.0000] No idea, I can't access it [04:06:00.0000] someone online who can explain HTTP auth in combination with XHR to me? sick... [04:06:01.0000] euh, meant to autocomplete sicking [04:36:00.0000] euhm why did Alexander Pechtold retweet Newt Gingrich? [04:36:01.0000] /me wonders whether he missed something important [06:11:00.0000] Chrome does not support iso-2022-jp? [06:15:00.0000] it does in the menu [06:16:00.0000] just not listed here: http://src.chromium.org/viewvc/chrome/trunk/deps/third_party/icu46/source/data/mappings/ucmlocal.mk [06:16:01.0000] confusing [06:22:00.0000] /me filed http://code.google.com/p/chromium/issues/detail?id=109120 [06:39:00.0000] hmm. document.write("foo"); location.href = "something"; document.write("bar"); [06:40:00.0000] No, I don't think I want to know :) [06:40:01.0000] /me laughs somewhat evily [06:42:00.0000] annevk: That doesn't include any purely algorithmic encodings [06:42:01.0000] annevk: That's just a list of those that have mapping tables [06:44:00.0000] sure [06:44:01.0000] but all other encodings are mentioned there [06:47:00.0000] GB18030 is probably supported and not mentioned. [06:48:00.0000] it is mentioned [06:48:01.0000] just with a space after GB [06:48:02.0000] whoa. it seems that under some circumstances, where doc.write() should imply doc.open(), it doesn't in chrome [06:48:03.0000] but does in Safari? [06:49:00.0000] I'm on Linux here. let's see [06:49:01.0000] willing to run a URL through Safari :) [06:49:02.0000] /me can't read [06:49:03.0000] fyi: http://tools.ietf.org/html/draft-reschke-http-status-308 [06:50:00.0000] beverloo: is there an open bug on WebKit being the only browser still using UTS22 alias/label matching? [06:51:00.0000] beverloo: everyone else does trim leading/trailing whitespace, ASCII case-insensitive [06:51:01.0000] huh. safari is even more broken [06:51:02.0000] No results for "UTS22" on either Chromium or WebKit [06:51:03.0000] http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1292 works in Chrome but not in Safari [06:51:04.0000] I honestly wouldn't know if that's been reported [06:52:00.0000] maybe there's a space [06:52:01.0000] odd. now the case that failed in Chrome started working [06:55:00.0000] to its credit, GB18030 is a rough ASCII superset and an UTF [06:55:01.0000] which makes it suck less than UTF-16 [06:55:02.0000] it's GBK plus 4 byte fun [06:55:03.0000] iirc [06:56:00.0000] (suck less for interchange that is; it sucks for in-memory use, of course) [06:56:01.0000] hsivonen: That's about all that's nice about it, though [06:57:00.0000] annevk: the 4 byte fun is really fun, though, because it isn't just a continuous algorithmic mapping from 4 byte sequences to codepoints [06:57:01.0000] annevk: it's continuous, starting from the first 4 byte sequence, but excluding anything in GBK. [07:00:00.0000] yeah [07:00:01.0000] beverloo: cannot find it either [07:03:00.0000] hmm. I see nothing in the spec saying that document.write() shouldn't imply document.open() when called on an aborted document [07:03:01.0000] man [07:04:00.0000] that we're still figuring out document.write() [07:04:01.0000] such a shame [07:05:00.0000] Hixie: could you, please, add the HTTP header to disable the IE XSS filter on the live DOM viewer? modifying IE's settings again and again is really annoying [07:12:00.0000] this page mentions there's a flag in the DOM to tell whether the browser is on/offline. http://diveintohtml5.info/offline.html -- what is that flag? [07:13:00.0000] crankharder: window.navigator.onLine [07:14:00.0000] ty [07:16:00.0000] hasather: dunno, can't get that to change. disconnecting wifi and making a request still shows true [07:17:00.0000] crankharder: it reflects the offline setting in your browser [07:25:00.0000] looks like Adobe, a font company, is still using bitmaps instead of real fonts for blog titles... [07:26:00.0000] link? [07:26:01.0000] astearns: e.g. http://blogs.adobe.com/accessibility/2012/01/wcag-2-0-techniques-for-pdf.html [07:27:00.0000] thanks, I'll raise some hell :) [07:27:01.0000] ah, cufon [07:30:00.0000] hasather: k, well there must be a different flag that detects whether to use the cache or not [07:31:00.0000] "by AWK" [07:31:01.0000] that's pretty close to AvK [07:37:00.0000] my app manifest looks like this: http://pastie.org/3125389 -- /offline contains a single alert, but when viewing / while offline that alert is executed 3 times. why? [08:00:00.0000] Ms2ger, what's the status on bug 647323 (translating testharness.js tests to mochitests)? [08:00:01.0000] I've started on it again this week [08:00:02.0000] I'm supposed to write testharness.js tests and also get them into Gecko, it seems. [08:00:03.0000] Excellent. [08:01:00.0000] (I hope nobody's going to take a close look at my python code) [08:01:01.0000] /me does an hg pull on his mozilla-central checkout that was last updated months ago [08:01:02.0000] Ooh, there's a sentence that sounds full of win [08:02:00.0000] What, about my python code? :) [08:02:01.0000] I was thinking of AryehGregor's sentence a couple of lines above :) [08:02:02.0000] Oh :) [08:03:00.0000] Well, I don't expect to do any of the actual work translating from testharness.js to mochitests. [08:03:01.0000] Which is why I'm poking Ms2ger. [08:03:02.0000] AryehGregor: so Mozilla hired you to write specs and tests? [08:04:00.0000] hsivonen, yes. [08:04:01.0000] AryehGregor: congrats [08:04:02.0000] Thanks. [08:04:03.0000] and hooray for whoever decided to hire spec/test writers [08:05:00.0000] ^That [08:05:01.0000] is Ms2ger getting a job too? [08:05:02.0000] I'm too lazy to get a job [08:06:00.0000] (Instead doing the same work in my free time) [08:06:01.0000] AryehGregor: congrats! that's awesome [08:06:02.0000] Ms2ger: you don't require money? :) [08:07:00.0000] Nah, I live in a hole in the ground [08:07:01.0000] He only needs to be connected to AC power [08:07:02.0000] /me knew Ms2ger was a hobbit [08:08:00.0000] Ms2ger, so your testharnessreport.js only registers known failures based on the test name, not the assert name too? [08:08:01.0000] hober: nice haha [08:08:02.0000] Yes, I think so [08:09:00.0000] Ms2ger, wouldn't it be better to check the test name plus the assert name, for the case of tests with many asserts? (Like, um, mine.) [08:10:00.0000] Or do we not care if it's a different assert that fails? [08:10:01.0000] Don't you have multiple test()s per file? [08:10:02.0000] Yes. [08:10:03.0000] Sometimes tens of thousands of test()s, but also sometimes tens of asserts per test. [08:11:00.0000] I think we can live with it if the failing assert changes [08:11:01.0000] If the failing assert changes, that probably indicates that either 1) an earlier assert than before fails (regression that should be flagged), or 2) a later assert than before fails (unexpected pass that should be flagged). [08:12:00.0000] Mm [08:12:01.0000] /me posts on the bug [08:12:02.0000] How would you identify the asserts? [08:12:03.0000] That's not exposed to the result callback? [08:12:04.0000] If not, it should be. [08:12:05.0000] /me pokes [08:12:06.0000] Probably just the message, no? [08:13:00.0000] Yeah, the assert can be identified just by message. [08:13:01.0000] I mean, by its name. [08:13:02.0000] If it has one. [08:13:03.0000] Really, you should be able to identify it by whatever info the default
spits out into the rightmost column. [08:14:00.0000] Well, most of my asserts, at least, don't have messages [08:14:01.0000] And I'd rather not have to write messages for all failing tests that we import [08:14:02.0000] test.message? [08:14:03.0000] The full message is supposed to come back. We have it reported at Opera. [08:14:04.0000] Well, if the asserts don't have messages, then the failing message will be the same regardless of which assert fails. [08:14:05.0000] Which is no worse than the status quo. [08:15:00.0000] I do that too if my test has only one or two asserts. [08:15:01.0000] It isn't autogenerated? [08:15:02.0000] Because I also don't feel like updating all the expectations when the default message's wording changes [08:16:00.0000] Some of my tests have messages, -- some don't. -- At least you get a somewhat usable message when something throws, so you most probably want the message anyway. [08:16:01.0000] We could adopt the approach of stabbing anyone who updates the default message's wording. [08:16:02.0000] Alternatively, regex. [08:16:03.0000] Hmm, the stabbing idea sounds interesting [08:17:00.0000] You're now talking about going from FAIL to FAIL, but identifying different fails? [08:17:01.0000] If it's failing in a different way, -- someone should probably look at it. Either because message changed, test changed, or code changed. Or regression, ofc. :] [08:17:02.0000] Changes in which assertion fails within a test() [08:18:00.0000] My problem is mostly someone being me :) [08:18:01.0000] Should be no different from any test, right? Whoever breaks it has to deal with it? [08:18:02.0000] If it turns out testharness.js changes are needed to make things more reliable, we can do those. [08:20:00.0000] I'm afraid of the work that needs to be done whenever testharness.js or sets of tests are updates [08:21:00.0000] As soon as anyone is relying on the format of test failure messages, testharness.js shouldn't be changing in ways that change those messages. [08:21:01.0000] Nor should tests, for that matter. [08:21:02.0000] I would be scared of getting lots of work every time a third party updated a test, if I were you [08:21:03.0000] I am [08:22:00.0000] For example if I update 10000 tests to have a different error message when they assert, that sounds bad under your proposal [08:22:01.0000] You have timestamp on the files, -- if they are new (updated), they are not "real" problems any more and should be marked in a special way. [08:22:02.0000] Velmont, doesn't work that way over here [08:23:00.0000] /me would just go with the sytem as it was designed to work i.e. a Test is the atomic thing [08:23:01.0000] Ms2ger: OK :-) I don't know moz system, -- I just thought it was somewhat similar to Opera's. [08:23:02.0000] jgraham, the various WGs should adopt policies that prohibit that kind of change without good reason. It's not in anyone's interest, after all. [08:23:03.0000] AryehGregor: It's in our interest if it gives us better error messages, for example [08:23:04.0000] So you think a difference in the failure message shouldn't be flagged? [08:24:00.0000] True, but reconfiguring the number of tests is also sometimes in our interest. [08:24:01.0000] Like breaking tests up by finer granularity so that they're more useful. [08:24:02.0000] Or coarser so that they run faster. [08:24:03.0000] No. I think it offers a small theoretical advantage at the cost of fighting the way the system is supposed to work [08:24:04.0000] It will also encourage Mozilla people to write tests that don't work well for others (many asserts per test) [08:25:00.0000] So it's bad from my point of view :) [08:25:01.0000] ^^ [08:25:02.0000] I already write tests like those, partly because it makes the tests run much faster given how long it takes to lay out the results (superlinear in the number of results). [08:25:03.0000] But they're also less useful, I gues. [08:25:04.0000] guess. [08:25:05.0000] If laying out the results is the bottleneck we should fix that [08:25:06.0000] So I can just write them at finer granularity if that's what people prefer. [08:26:00.0000] I tried, didn't you notice? :) [08:26:01.0000] Yeah, but there are other ways to try [08:26:02.0000] Suggestions? [08:26:03.0000] Yeah, and you keep breaking my XML tests :) [08:26:04.0000] Like having id=log-fail that only prints fails [08:26:05.0000] It doesn't *need* to lay out the tests when you're using testharnessreport to do it? [08:27:00.0000] Or having id=log-delayed that requires human interaction to actually generate the tests [08:27:01.0000] Are you speaking about slow as in a testrunner system, -- or when people run them themselves? [08:27:02.0000] *test results [08:27:03.0000] The latter [08:27:04.0000] Velmont, the latter. [08:27:05.0000] I guess it is both? [08:27:06.0000] jgraham, or automatically print only fails if there are a lot of passes. [08:27:07.0000] Right, there are a number of things that we could do to make it more efficient [08:28:00.0000] Another question: should testharnessreport.js as used by browser vendors remove
? As noted, for my tests it can sometimes take a really long time to lay out the results. Like 30+ seconds. [08:28:01.0000] Or should we just fix that? [08:28:02.0000] Oh, I had another issue with test descriptions that included characters XML didn't like [08:28:03.0000] Silly XML. [08:28:04.0000] Well, that's an open door [08:28:05.0000] AryehGregor: That seems like a reasonable hack, although I would expect it to break some tests [08:29:00.0000] Maybe better to just have those tests use delayed logging or whatever. [08:29:01.0000] Yup. Although it shouldn't start layout until after the results have been sent so it might be interupted by the next test loading (or something) [08:30:00.0000] jgraham, hmm, could you add a function to testharness that just disables logging? [08:30:01.0000] I was about to say that. [08:30:02.0000] I'd want the log to be filled in when tests are run outside the harness, though [08:31:00.0000] It should disable only the logging at finish, though, since the logging after each result might affect a later test's outcome. [08:31:01.0000] (the logging after each result shouldn't be a big deal since I updated it to happen only every 1000th test if there are lots) [08:31:02.0000] Right, the question is how one tells when it should be enabled or not [08:31:03.0000] I'm not sure I want the log filled in by default when tests are run outside the harness, because that's really slow in some cases. [08:31:04.0000] I can tell you that in testharnessreport :) [08:31:05.0000] AryehGregor: What logging after each result? [08:32:00.0000] jgraham, updating the number of tests run so far. [08:32:01.0000] Oh, Ok [08:32:02.0000] Ms2ger: That seems reasonable [08:32:03.0000] (a manually set off switch) [08:33:00.0000] I'm probably already going to get shouted at for making our test runs longer :) [08:33:01.0000] For tests not using testharnessreport.js, I also want a way to default some or all of the show/hide checkboxes to hide. [08:33:02.0000] So that to display the tests you need to check one of the boxes. [08:33:03.0000] Class names? [08:33:04.0000] Hmm? [08:34:00.0000]
[08:34:01.0000] Gecko has by far the most complicated Encoding UI, followed by Opera :/ [08:34:02.0000] oh, maybe IE is worse [08:34:03.0000] (I want a way to shut it up without touching the test's DOM, though) [08:35:00.0000] IE does not expose an encoding menu? [08:35:01.0000] that's kind of novel [08:36:00.0000] Ms2ger, a function to be called by the tests makes more sense to me. Like a parameter to pass to setup(), maybe. [08:36:01.0000] That would work too [08:36:02.0000] annevk: If only people listened to mpt in 2004 (or whenever) [08:36:03.0000] jgraham: pointer? [08:36:04.0000] For testharnessreport.js, we want a way to stop stuff from even being added to the DOM. [08:37:00.0000] It's just a waste of resources. [08:37:01.0000] Yeah [08:37:02.0000] annevk, Epiphany if you have it installed, otherwise it's on Google Groups somewhere [08:37:03.0000] For manual runs, it's probably fine to always add stuff to the DOM. IIRC, that's not a big deal even for really big result sets, given that you use my innerHTML optimization. [08:37:04.0000] annevk: http://groups.google.com/group/netscape.public.mozilla.ui/msg/60e186a20d5d69d0 [08:37:05.0000] (2000) [08:38:00.0000] Until recently, we had a bug open to expose encoding in the context menu :) [08:39:00.0000] Wow [08:40:00.0000] Safari calls it "Text Encoding", Chrome/Opera just call it "Encoding" [08:40:01.0000] Gecko still calls it "Character Encoding" [08:40:02.0000] pretty bad [08:46:00.0000] Does my W3C login work for hg.csswg.org, or do I need yet another login? [08:47:00.0000] /me wants somewhere to check in his work as he writes it [08:47:01.0000] Another one [08:48:00.0000] Ask plinss [08:48:01.0000] Via e-mail? [08:49:00.0000] Has anyone ever remarked that having most tests named things like test_bug382027.html is not very convenient for figuring out what they do? [08:49:01.0000] yep [08:51:00.0000] Well, the usual response is that those tests test one particular piece of code, and it isn't much use to call it test_getattribute-style-for-border-style [08:52:00.0000] It's not very useful for sharing them to name them based on internal bug ids :) [08:52:01.0000] Sharing? What's that? [08:52:02.0000] I don't actually see any explicit transform tests anywhere. [08:53:00.0000] (in Gecko) [08:53:01.0000] (based on where dbaron told me to look) [08:53:02.0000] (grepping for -moz-transform\|mozTransform) [08:54:00.0000] /layout/reftests/transform-3d/? [08:54:01.0000] And /layout/reftests/transform/ [08:55:00.0000] (http://mxr.mozilla.org/mozilla-central/search?string=moz-transform&find=%2F&findi=&filter=^[^\0]*%24&hitlimit=&tree=mozilla-central) [08:55:01.0000] good morning, Whatwg! [08:56:00.0000] Ms2ger, oh, I was looking for mochitests. [08:56:01.0000] Wasn't thinking about reftests just yet. [08:56:02.0000] (and I'm only doing 2d for now) [08:56:03.0000] There don't tend to be many mochitests for CSS stuff [08:57:00.0000] Why not? Obviously you need some reftests, but those must be a lot slower per assertion. [09:01:00.0000] /me expects to write a handful of reftests, and lots and lots of JS tests [09:03:00.0000] Ms2ger, I should just send an e-mail to plinss? [09:04:00.0000] Yeah [09:13:00.0000] Done. [09:16:00.0000] jgraham, want me to file a bug for disabling the logging or would you remember it without one? :) [09:16:01.0000] CC me if you file it. [09:17:00.0000] File it [09:18:00.0000] Alternate theory: turn logging off by default and then only turn it on in testharnessreport.js [09:18:01.0000] Advantage: encourages people to include testharnessreport.js [09:18:02.0000] Wait, what am I supposed to do when writing tests if there are vendor prefixes? Just write tests that will fail in all browsers? [09:18:03.0000] Really? [09:18:04.0000] jgraham, you mean make the default testharnessreport.js non-empty? [09:19:00.0000] I want testharnessreport.js to be empty, ta :) [09:19:01.0000] OK, well just a thought :) [09:19:02.0000] (it occurs to me that due to the peculiar way that Opera runs tests this solution isn't going to work well for us :( ) [09:19:03.0000] Which solution? [09:20:00.0000] Using a flag in testharnessreport.js to disable output [09:24:00.0000] . . . Okay, so is there actually any programmatic way to detect where a transformed element wound up? [09:24:01.0000] .offsetLeft and friends seem to report the untransformed position. [09:25:00.0000] If there's no way to check it programmatically, that sort of implies I can only do reftests. [09:26:00.0000] Hmm, so I doubt that I can rely on the relative order of testharness.js and testharnessreport.js loading [09:26:01.0000] Without retrospectively editing all existing tests [09:27:00.0000] It would be kind of bad if you can't rely on the order. [09:27:01.0000] Well it's kind of bad then [09:28:00.0000] Because of Opera's legacy tests, or? [09:29:00.0000] Okay, so getBoundingClientRect() works, although frankly I have no idea whether that's actually required by spec. [09:29:01.0000] Or if it works in IE. [09:29:02.0000] /me tests in IE [09:29:03.0000] For example in submissions/AryehGregor interfaces.html has testharnessreport.js first but reflection/reflection-embedded/html has testharness.js first [09:30:00.0000] Add a check to testharness.js to figure out somehow and complain. [09:30:01.0000] /me votes for setting a global variable in testharnessreport.js [09:30:02.0000] Ms2ger, why don't you want that? [09:30:03.0000] You know plh lives close enough that he could quite easilly track me down and kill me if I break all the legacy tests [09:31:00.0000] Which legacy tests? [09:31:01.0000] All the ones written up till now [09:31:02.0000] Setting a global works of course [09:31:03.0000] But it is really ugly [09:31:04.0000] testharness.js should be first [09:32:00.0000] It would be nice if in testharnessreport.js one could do setupHarness(options) [09:32:01.0000] or something [09:32:02.0000] And I'll probably have to fix up the tests I care about [09:32:03.0000] Ms2ger: And the tests you don't care about too, if you want it to work :) [09:32:04.0000] jgraham, if the goal is to get the legacy tests working with browsers, and browsers' use of testharnessreport.js requires a specific order, the legacy tests are already broken. [09:32:05.0000] Nothing wrong with making them more obviously broken. [09:32:06.0000] jgraham, I'm not going to import all tests :) [09:33:00.0000] Certainly not the widget ones [09:33:01.0000] Okay, so getBoundingClientRect() interoperably reflects transforms. [09:34:00.0000] I have no idea why, since AFAICT the spec doesn't say anything meaningful about what exactly the rectangle is supposed to be, unless I don't understand the secrets of CSS terminology. [09:35:00.0000] But it will do for now. [09:51:00.0000] AryehGregor: it does? meh [09:51:01.0000] oh well, not my spec anymore [09:51:02.0000] annevk, meh that it reflects transforms? [09:52:00.0000] Can anyone explain to me why IE and WebKit prefix with ms and webkit, but Gecko and Opera prefix with Moz and O? [09:52:01.0000] Uppercase vs. lowercase? [09:52:02.0000] It's confusing. [09:53:00.0000] In the DOM? [09:54:00.0000] Yes. [09:54:01.0000] I think Moz is a tanktekism [09:54:02.0000] We also have mozRequestAnimationFrame, mozMatchesSelector [09:54:03.0000] Also, I think the simplest solution for the testharness.js thing is to allow setup to take another parameter output that controls whether output is enabled. This has the effect that one could override it on a per-test basis which seems bad but not too bad [09:55:00.0000] AryehGregor: yeah, I thought we were going to have another API for transforms [09:55:01.0000] jgraham, can I call setup multiple times? [09:55:02.0000] AryehGregor: -o-x -> OX; dash -> next character uppercase [09:56:00.0000] AryehGregor: for non-CSS, I think we did it so the DOM could introduce methods that started with a lowercase o [09:56:01.0000] dunno how consistent we've been [09:56:02.0000] All I know is http://my.opera.com/emoller/blog/2011/12/20/requestanimationframe-for-smart-er-animating?cid=78181842#comment78181842 [09:57:00.0000] Ms2ger: Yes [09:57:01.0000] That works, then [09:57:02.0000] I think allowing tests to override it is fine [10:05:00.0000] Okay, so Gecko serializes matrices with px units for the last two values. [10:05:01.0000] Is that supposed to be defined somewhere? [10:05:02.0000] /me hates CSS specs with a passion [10:06:00.0000] dbaron would know about that [10:06:01.0000] about what? [10:06:02.0000] Okay, so Gecko serializes matrices with px units for the last two values. [10:06:03.0000] Is that supposed to be defined somewhere? [10:07:00.0000] dbaron, I started writing JS-based tests for CSS Transforms, as discussed. The first non-interoperable thing I found is Gecko serializes (e.g.) translate(10px) as matrix(1, 0, 0, 1, 10px, 0) in getComputedStyle(), and all other browsers serialize as matrix(1, 0, 0, 1, 10, 0). [10:08:00.0000] I think the group hasn't really resolved on it [10:08:01.0000] but I'm inclined to give in [10:08:02.0000] should i be entertained that the stupid nonsense I see on the web never slows; that people always find new, stupid things to do badly [10:09:00.0000] such as this address form, where I entered my zip code; tabbed; started entering my city; and halfway through entering the city, apparently some json request finished and autocompleted the city, leaving me with "onsONSET" entered in the field [10:12:00.0000] it's also discomfiting to be asked my mother's maiden name in 2012 [10:19:00.0000] it could be worse. you could be a woman, and wonder endlessly why all the damn forms ask you if you're "Miss" or "Mrs" [10:19:01.0000] as if your marital status mattered when you registered a domain name, for example [10:58:00.0000] assert_equals: getComputedStyle expected "matrix(1, -1, 0, 1, 0, 0)" but got "matrix(1, -0.9999999999999999, 0, 1, 0, 0)" [10:58:01.0000] Sigh. [10:58:02.0000] Why does CSS have to be so poorly defined? [10:58:03.0000] Is defining rounding for serialization too much to ask? [10:58:04.0000] Yes [10:58:05.0000] The WG doesn't want that kind of thing [10:59:00.0000] /me notes that Gecko doesn't have any such issues that he's discovered [11:00:00.0000] Go us! [11:08:00.0000] most test frameworks have fuzzy matchers for floating point values [11:08:01.0000] jamesr_, yes, I just switched to using that. [11:08:02.0000] I had to make up an epsilon (I picked 1.0e-5). [11:08:03.0000] At first I tried 1.0e-10, but IE was deviating from expected values by the order of 1.0e-8 sometimes. [11:08:04.0000] So I made up a number that happened to work. [11:09:00.0000] How in heaven's name does the CSSWG expect anyone to write any correct tests under these conditions? [11:09:01.0000] For API stuff? Not, I guess [11:10:00.0000] /me also notes that he didn't test getBoundingClientRect() results when the box he was using was transformed by a singular matrix, because again, IE happened to produce different results in that case and getBoundingClientRect() isn't actually defined in any precise way [11:10:01.0000] What's the border box of something transformed by (-1, 1, 1, -1, 0, 0)? [11:12:00.0000] Is there an easy way in python to merge two dicts? [11:14:00.0000] dbaron, thanks. [11:15:00.0000] dict.update? [11:15:01.0000] Thanks [11:46:00.0000] is there a way to determine whether or not the app cache has network connectivity? [12:07:00.0000] i'm having a real hard time with the order of events firing as it relates to the app cache [12:07:01.0000] the docs on the app cache say that the browser will request the manifest file as soon as it encounters the directive in the tag [12:07:02.0000] but that doesn't seem to be the case [12:07:03.0000] http://pastie.org/3126878 [12:08:00.0000] instead, it looks like the browser processes the whole document, runs all javascript, and only then requests the manifest file [12:31:00.0000] Public opinion as requested: We should kill DAP and if they do anything relevant to the web, webapps should take it on [12:33:00.0000] /me runs AryehGregor's Range tests in Mozilla's mochitest framework [12:33:01.0000] Yay. [12:33:02.0000] Does it work? [12:33:03.0000] /me waits patiently [12:38:00.0000] Hmm, test_interfaces probably needs some external file [12:46:00.0000] ? [12:47:00.0000] The IDL parser and stuff [12:47:01.0000] hmm [12:47:02.0000] even when not hungry [12:47:03.0000] you need more than 8 nigiri [12:48:00.0000] fortunately there's always yoghurt :) [12:57:00.0000] And bacon cheeseburgers :p [12:58:00.0000] Ms2ger, you just need the stuff in /resources. [12:58:01.0000] Yeah [12:58:02.0000] \o/ [12:59:00.0000] First finished mochitest run :) [12:59:01.0000] :) [13:00:00.0000] OTOH, so many test failures the actual count scrolled off my console [13:00:01.0000] 1231 INFO Passed: 676 [13:00:02.0000] 1232 INFO Failed: 551 [13:01:00.0000] jgraham: indeed, I need to get myself back to Oslo :) [13:01:01.0000] (That's interfaces.html) [13:03:00.0000] /me may assemble patches tomorrow [13:04:00.0000] There should be lots of low-hanging fruit to fix for interfaces.html. [13:36:00.0000] Does Mozilla Bugzilla have a clone bug feature? [13:36:01.0000] yes [13:36:02.0000] on bottom right corner [13:37:00.0000] thanks [14:23:00.0000] so far only indication that Gecko/Opera support iso-2022-cn and euc-tw [14:23:01.0000] the latter with incompatible labels [14:25:00.0000] Opera only recognizes euc-tw [14:25:01.0000] Gecko only x-euc-tw and some others [14:26:00.0000] Chrome does recognize iso-2022-cn, but to explicitly not support it [14:26:01.0000] prolly to avoid the issue Philip` pointed out earlier [15:00:00.0000] what's a good spec that defines terms like "directional navigation", "sequential navigation", "document navigation order", etc.? [15:01:00.0000] www.whatwg.org/specs/web-apps/current-work/multipage/editing.html#editing has some bits, but definitely not complete [15:02:00.0000] www.w3.org/TR/css3-ui/#keyboard has some other bits, but also not fully fleshed out [15:02:01.0000] what kind of navigation are we talking about? [15:03:00.0000] keyboard control/focus [15:03:01.0000] tabbing and spatial nav [15:03:02.0000] that's not really well defined anywhere [15:04:00.0000] ok, no problem. I'll stitch something together for shadow DOM. [15:15:00.0000] note for tomorrow's annevk: XML5 [15:15:01.0000] that sounds delicious [15:15:02.0000] Is tomorrow's annevk a different annevk from today's? [15:15:03.0000] Is there an army of annevk clones or something? [15:16:00.0000] Because that sounds kinda scary, but I guess it could improve productivity when it comes to writing specs. [15:17:00.0000] perhaps it's more of a larvae->pupa->butterfly situation [15:27:00.0000] dglazkov: tabindex defined here: http://www.whatwg.org/specs/web-apps/current-work/multipage/editing.html#focus [15:28:00.0000] not in a ton of detail [15:28:01.0000] yep [15:28:02.0000] I was just looking for a grand unified theory of keyboard navigation [15:29:00.0000] For anyone who fails to understand me: http://gothsuptrees.tumblr.com/ [15:43:00.0000] why would window.webkitIndexedDB be undefined in safari 5.1.2 ? [15:45:00.0000] crankharder: I believe you'll find that each webkit browser {en,dis}ables different features [15:47:00.0000] hober: indesedDB is enabled - a demo site is working 2012-01-05 [16:28:00.0000] I'm getting effectively the same error in chrome and ffx from this indexedb example [16:28:01.0000] http://html5-demos.appspot.com/static/html5storage/index.html#slide31 [16:28:02.0000] http://pastie.org/pastes/3128254 [19:08:00.0000] fyi, I just realized today that I had at some point borked the link-fixup behavior for the dev.w3.org version of the spec [19:08:01.0000] such that broken fragment refs didn't get redirected the way they are supposed to [19:09:00.0000] but anyway, fwiw, I just fixed it so that they should work again as expected [23:32:00.0000] http://lists.w3.org/Archives/Public/public-webapps/2011OctDec/att-1696/speechapi.html [00:08:00.0000] http://www.jwz.org/blog/2012/01/jon-mitchell-google-hates-the-internet/ "A far distant 99th priority is the person who actually created the thing in the first place." Isn't that the same with traditional media? [00:30:00.0000] annevk: if you look at Google's plans for search results, they seem to be going in the same direction [00:31:00.0000] Google will try to answer any direct question itself, downplaying the source link where it got its information [00:31:01.0000] https://www.google.com/search?ie=UTF-8&q=who+is+matt+damon+married+to [00:32:00.0000] You have to click "show sources" to actually see where Google got it from. [00:35:00.0000] weirdly, the results if you replace "angelina jolie" for "matt damon" are different [00:35:01.0000] Even worse I'd say, there are 7 sites they're not even mentioning. [00:38:00.0000] indeed [00:39:00.0000] A kick in the balls for the semantic web if you ask me. [00:39:01.0000] You go through the trouble of marking everything up with microdata, Google steals your data and doesn't even bother to credit you properly. [00:40:00.0000] woef: does that feature even use any semweb markup? [00:41:00.0000] woef: and since when was semantic web about attributing anything? [00:41:01.0000] woef: AFAICT, the SemWeb folks were never that concerned about provenance [00:41:02.0000] woef: you stick all facts and lies you find into a triple store, shake and out come conclusions [00:44:00.0000] hsivonen: Google's microdata effort are certainly behind it. [00:44:01.0000] The best way to "know" an answer. [00:45:00.0000] woef: nope. click show sources [00:45:01.0000] woef: and, among others, you get http://www.huffingtonpost.com/2010/12/15/matt-damons-wife-luciana-steps-out-first-time-since-giving-birth_n_796981.html [00:46:00.0000] woef: looks like they are extracting stuff from English phrases without any microdata [00:46:01.0000] I never said they did. [00:46:02.0000] *not do that too [00:47:00.0000] But if you have the chance between marked up content using microdata and English phrases ... what are they going to prefer? [00:47:01.0000] The data that machines can read 100% accurately, or the stuff where they still need to do some guessing? [00:47:02.0000] is "Google's microdata effort" involved in any of the sources claimed for figuring out who Matt Damon's wife is? [00:48:00.0000] You claim they won't use it? [00:48:01.0000] They'd be dumb not to. [00:49:00.0000] I wouldn't expect them to use microdata for this particular thing, no [00:49:01.0000] Is there any reason not to? [00:49:02.0000] hsivonen, regarding provenance, there are differences among semweb folks in its importance [00:50:00.0000] Beyond the fact that hardly any big sites these days are implemting it for relations between people? [00:50:01.0000] woef: there are lots of people providing user-visible simple English phases about celebrity marriages [00:50:02.0000] schema.org lists: spouse Person The person's spouse. [00:50:03.0000] woef, expressing relations between people is solved by XFN, indexed by Google, and available to anyone thru the Google Social Graph API [00:50:04.0000] woef: there are few sources motivated to provide correct explicit metadata declarations of celebrity marriages [00:51:00.0000] oh, and has nothing to do with microdata [00:51:01.0000] Microdata (/schema.org) has everything ready to start using microdata for things like this [00:52:00.0000] woef: so changes are that phrases grepped from text would give better confidence for this kind of thing that explicit metadata [00:52:01.0000] woef "ready to start using" == handwaving. whatever. [00:53:00.0000] lol [00:53:01.0000] tantek: does the capital Semantic Web stack have a provenance layer these days? [00:53:02.0000] Okay, so "they're not using it now so there is no problem at all" then. [00:53:03.0000] hsivonen - depends on whose version of SemWeb "Trust" you use [00:53:04.0000] I haven't been looking that carefully, but it seems to me that capital-S stack developed all kinds of machinery before provenance [00:54:00.0000] sure, it's a multilayer cake [00:54:01.0000] and provenance was in the "we'll figure it out later bucket" [00:54:02.0000] nah, that's a bit of a mischaracterization [00:54:03.0000] for example, a simple answer is to supply provenance as a triple annotation on any existing triple [00:55:00.0000] it can complicate queries sure, but it's not an unsolvable problem [00:55:01.0000] though last time I heard TimBL answer a question of SemWeb provenace was in 2004, so things might have progressed since then [00:55:02.0000] tantek: it's solv*able* sure [00:56:00.0000] to be clear - I'm just passing along bits I've picked up from discussions [00:56:01.0000] it's not how I would choose to solve provenance [00:56:02.0000] given that I don't choose to use uppercase SemWeb tech either (but it's easy enough to create simple data to interop with it) [00:58:00.0000] on a completely different topic: in the popup blocker is disengaged, Chrome, IE9 and Opera open a new window/tab and load the Mozilla site into it when loading https://bug98654.bugzilla.mozilla.org/attachment.cgi?id=77369 [00:58:01.0000] Firefox doesn't [00:58:02.0000] I don't see any mechanism in the spec that explains the Chrome/IE/Opera behavior [00:58:03.0000] AFAICT, Firefox complies with the spec [00:59:00.0000] and Firefox has had the Chrome/IE/Opera behavior before [00:59:01.0000] What's the mechanism Chrome/IE/Opera apply here? [00:59:02.0000] s/in the popup/if the popup/ [01:01:00.0000] Hixie: am I misreading the spec or do Chrome/IE/Opera have an unspecced mechanism at work here? [01:03:00.0000] on another completely unrelated topic: apparently kids in schools are now learning how to coordinate in-person group-trolling: http://www.takepart.com/article/2011/12/07/cyberbaiting-teens-find-high-tech-ways-humiliate-teachers [01:06:00.0000] tantek: that's sad. kids are cruel [01:07:00.0000] Well, we knew that, no? [01:08:00.0000] during my freshman year at the university, I was asked to come back to my old school as a substitute art teacher for a week, but I was too afraid that I couldn't keep order in the class, so I declined [01:08:01.0000] i find it a bit [01:09:00.0000] I have a feeling politics will get very different in ~10 years when that generation has reached voting / political participation age [01:09:01.0000] don't worry [01:09:02.0000] the Republicans will make sure they can't vote [01:09:03.0000] tantek, kids weren't cruel when you were at school? [01:10:00.0000] they weren't that efficiently coordinated about it [01:13:00.0000] oh and they focused coordinated/group attacks more on weaker individuals, in lower grades etc. not teachers. [01:15:00.0000] Parents became "our friends", so authority based on seniority is somewhat gone. [01:38:00.0000] /me doesn't think history bears "I have a feeling that politics will be very different when {some subset of today's youth} has reached participation age" [01:42:00.0000] I occasionally wish WebKit and Chrome had something like MXR [01:43:00.0000] /me wonders why the site map link on the www.chromium.org site is in Cyrillic letters [01:44:00.0000] is there a useful third-party online cross reference for WebKit's source that I should know about? [01:45:00.0000] Is that "occasionally" that maps 1:1 to the "occasionally" that you want to look up something in their source code? [01:46:00.0000] jgraham: yes [01:46:01.0000] http://codesearch.google.com/#&exact_package=chromium [01:47:00.0000] (Also, bug 674657) [01:47:01.0000] Ms2ger: thanks [01:48:00.0000] hmm. no non-spec magic in their document.write beyond a recursion limit [01:49:00.0000] hmm. they do have a non-spec check in thei document.open impl. [01:51:00.0000] though that check doesn't seem to explain this [02:06:00.0000] Ms2ger: you now have perms to the W3C testframework dvcs repo, as well as to add test suites through the Web UI [02:06:01.0000] Thanks [03:56:00.0000] whoa [03:56:01.0000] so I have [03:56:02.0000] ~ [03:56:03.0000] [03:56:04.0000] gives � %EF%BF%BD%20 in IE9 [03:57:00.0000] there's two 0x0A after the ~ [03:57:01.0000] that's kind of insane [03:57:02.0000] oh sweet [03:57:03.0000] if I add a DOCTYPE [03:57:04.0000] I get [03:57:05.0000] � %EF%BF%BD%0A%0A [03:58:00.0000] not sure I want to know [04:01:00.0000] annevk: probably an innerHTML difference between modes [04:02:00.0000] yeah seems likely [04:03:00.0000] the weird thing is mostly the BOM :) [04:42:00.0000] kind of funny how these encodings are from 95 and nobody ever bothered to take a proper look at them [04:42:01.0000] makes me wonder about the network stack [04:43:00.0000] consider also that HTML was from 94 or so (earlier?) and nobody took care of it until 2006 [04:46:00.0000] so I have this to Unicode algorithm [04:46:01.0000] when it says return failure, it's a fatal error [04:46:02.0000] when it says "emit code point" it is run again (unless the code point is the EOF code point) [04:46:03.0000] what should I use for running it again without emitting anything? [04:47:00.0000] just return? [04:47:01.0000] and explain it above? [04:47:02.0000] something like this is needed for the encodings that are basically based on state machines [04:48:00.0000] continue; [04:58:00.0000] yeah I considered that and I'm not sure why I rejected it, so I'll go with that [05:02:00.0000] e.g. "Set the hz-gb-2312 flag and continue." Ms2ger? [05:02:01.0000] Sure [05:05:00.0000] annevk: then you probably need to define "continue" [05:05:01.0000] annevk: Hixie uses goto step # [05:06:00.0000] annevk: of course, it's nicer if the spec uses continue/break instead of goto :-) [05:13:00.0000] no answers to my window.location/document.write question yet :-( [05:37:00.0000] hsivonen: yeah I have defined it for the to Unicode algorithm [05:37:01.0000] hsivonen: I haven't referenced it all over though [05:37:02.0000] (not yet checked in, still working on the last details of hz-gb-2312 to Unicode) [05:57:00.0000] now committed: http://dvcs.w3.org/hg/encoding/raw-file/tip/Overview.html#hz-gb-2312 [06:55:00.0000] bitbucket.org offline? [07:00:00.0000] Is for me [07:16:00.0000] INFO | runtests.py | Running tests: end. [07:16:01.0000] mochitest-plain passed [07:16:02.0000] \o/ [07:17:00.0000] With the testharness.js tests? [07:18:00.0000] With 5 of AryehGregor's tests [07:19:00.0000] And after multiplying your timeouts by ten :) [07:32:00.0000] Heh [07:33:00.0000] Well once I push the changes you can do that from testharnessreport.js. Although that is sort of very problematic. Hmm [07:35:00.0000] fwiw, mochitests use an API to multiply the timeout by a factor, rather than passing an absolute value [07:36:00.0000] Yes, maybe that is a better idea [07:36:01.0000] bitbucket back [07:38:00.0000] A and B are null [07:38:01.0000] A or B are/is??? null [07:38:02.0000] is [07:39:00.0000] either A or B is null [07:39:01.0000] k [07:39:02.0000] But ask someone with a different language background :) [07:39:03.0000] Is [07:40:00.0000] Because it is read as singular [07:40:01.0000] i'd say they're both right; i'd say are [07:41:00.0000] since singularity isn't implied (english "or' isn't "xor") [07:41:01.0000] It is commonly read as xor [07:41:02.0000] no, definitely not [07:41:03.0000] that would mean that if both A and B are null, then the condition is not satisfied [07:41:04.0000] Sure. "Shall we go left or right?" [07:41:05.0000] I mean in common usage [07:41:06.0000] Not in technical contexts [07:41:07.0000] that's a choice, not a condition [07:41:08.0000] left and right are mutually exclusive though [07:42:00.0000] What did you set off now, annevk... ;) [07:42:01.0000] if you say "are there any cats or dogs?", you don't say "no" if there are both cats and dogs :) [07:42:02.0000] zewt: that would be "are" even if there are only cats, since cats is plural [07:42:03.0000] not relevant [07:43:00.0000] the point is simply that "or" isn't "xor" [07:43:01.0000] zakim, who is here? [07:43:02.0000] Well it depends on context obviously [07:43:03.0000] sorry:-) [07:43:04.0000] Ms2ger: dunno I moved on long ago [07:43:05.0000] and more fundamentally, "if A or B are null" doesn't sound wrong to me as a native speaker (which is always the final measurement) [07:43:06.0000] david_carlisle, I am! :) [07:44:00.0000] zewt: I find it acceptable I guess, but less nice than "is" [07:44:01.0000] Which is also the final measurement :p [07:45:00.0000] just go with "if A is null, or if B is null" [07:45:01.0000] fffff [07:46:00.0000] don't do that :P [07:46:01.0000] if not A or not B? [07:47:00.0000] Ms2ger: doesn't it need to be and if you negate them? [07:47:01.0000] !a <==> a == nsnull [07:48:00.0000] touche [07:51:00.0000] if null ∈ {A,B} [07:53:00.0000] Bah [07:53:01.0000] {A,B} ∋ null [07:56:00.0000] Now you're just showing off [08:19:00.0000] annevk: Did you get your note? [08:25:00.0000] jgraham: yeah, but I didn't really work on it :( [08:28:00.0000] hi guys, curious as to whether or not there is a need for or an element similar to say - [08:32:00.0000] Echoes2: You tell us. Is there a need? I can't think of any good use cases [08:33:00.0000] Echoes2,

Company Name

seems good enough. [08:44:00.0000] Oh, bah [08:44:01.0000] AryehGregor, now I get to figure out something for your patches that fail on debug builds and pass on opt builds :/ [08:45:00.0000] Ms2ger, some of my tests fail on debug builds and pass on optimized builds? That sounds like a fairly serious problem. [08:45:01.0000] (in Gecko) [08:45:02.0000] Also, since Bitbucket is down, I'm pushing a DOM4 change to W3C only, all right? [08:45:03.0000] annevk, ^ [08:45:04.0000] Stringification differs [08:45:05.0000] So be sure to pull from there. [08:45:06.0000] Oh, really? That sounds like a bad idea. [08:45:07.0000] Debug builds add the pointer [08:45:08.0000] AryehGregor: Doesn't sound that surprising to me :) [08:46:00.0000] jgraham, what doesn't? [08:46:01.0000] AryehGregor: sure [08:46:02.0000] AryehGregor: guess we'll sync it later [08:46:03.0000] /me now has a desktop with much faster CPU than his server [08:46:04.0000] we should maybe consider moving all specs to github for better integration with the web community [08:46:05.0000] AryehGregor: Different behaviour in debug [08:47:00.0000] but git vs hg might be too much of a pain [08:47:01.0000] Like [object Window @ 0x7fe97b3ef320 (native @ 0x7fe97b112c78)] [08:47:02.0000] Almost twice the bogomips! [08:47:03.0000] annevk, it's not that bad at all, in fact. hg-git works quite well. [08:47:04.0000] The major nuisance is you have different commit id's. [08:47:05.0000] annevk, I'm not touching git if I can help it :) [08:47:06.0000] /me isn't touching hg if he can help it, but apparently can't :( [08:48:00.0000] It would actaully be easier to ahve the server on git and let people that want to use hg as the client do taht [08:48:01.0000] annevk, Ms2ger: Running make on DOM4 seems to have changed the output format of Overview.html to HTML5 instead of HTML 4.01. [08:48:02.0000] (I would slightly prefer that, even) [08:48:03.0000] Er, wait a second [08:48:04.0000] Is that expected? [08:48:05.0000] Have you committed already? [08:48:06.0000] No. [08:48:07.0000] I always look at the diff first. [08:48:08.0000] (and also after committing but before pushing, generally) [08:49:00.0000] (and by "always" I mean "usually, if I feel like it") [08:49:01.0000] hg update 4de5b048478e [08:49:02.0000] That'll get you back to the ED version [08:49:03.0000] Oh, I was on the WD version? [08:49:04.0000] The WD version was on tip [08:50:00.0000] bitbucket sort of seems up [08:51:00.0000] maybe only the site [08:51:01.0000] :/ [08:51:02.0000] Yeah, seems up-ish. [08:51:03.0000] Push just succeeded. [08:54:00.0000] http://science.slashdot.org/story/12/01/05/1453244/leap-second-coming-in-june-2012 [09:10:00.0000] good morning, Whatwg! [09:10:01.0000] Happy new year, dglazkov! [09:11:00.0000] Happy New Year, Ms2ger! [09:21:00.0000] /me grumbles about webkit [09:22:00.0000] Altogether now for one last pre-ephiphany chorus of "We Wish You an Alright Christmas and an OK New Year" [09:22:01.0000] jgraham, yeah, you too [09:22:02.0000] :) [09:26:00.0000] http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Cdiv%20style%3D%22height%3A50px%3Bwidth%3A50px%3Bbackground%3Agreen%3B-o-transform%3Arotate(180deg)%22%3E%3C%2Fdiv%3E%0D%0A%3Cscript%3Ew(document.querySelector(%22div%22).getBoundingClientRect().height)%3C%2Fscript%3E [09:26:01.0000] In Opera, that logs "52". [09:27:00.0000] That's the height of a 50px square div with -o-transform: rotate(180deg). [09:27:01.0000] lul? [09:28:00.0000] Note: scale(-1, -1) produces the same computed value of -o-transform, but a height of 50 as expected. [09:28:01.0000] Um, why isn't that loading anything? [09:28:02.0000] What do you mean? [09:29:00.0000] Works for me. [09:29:01.0000] Loads for me and logs 51 [09:29:02.0000] The HTML is: [09:29:03.0000]
[09:29:04.0000] [09:29:05.0000] I'm testing in Opera Next 12.00 alpha. [09:29:06.0000] Seems like a bug. I wonder if it is fixed internally [09:29:07.0000] I'm guessing there's too much rounding going on. [09:30:00.0000] gsnedders: You said you weren't doing anything; try it on mainline :) [09:30:01.0000] Like probably 180deg is converted to radians and then something gets rounded excessively and it winds up really rotating by 176deg or something. [09:31:00.0000] skewX and skewY with small values have similar problems: http://aryeh.name/tmp/css-test/contributors/aryehgregor/incoming/2d-transforms.html [09:31:01.0000] Throws URIError here [09:32:00.0000] Because I missed the final E of the URL [09:32:01.0000] AryehGregor: Probably an effect of AA [09:33:00.0000] Rotation by 180 degrees of a pixel-aligned square that's an integer number of pixels wide has what exactly to do with AA? [09:37:00.0000] AryehGregor: I don't know the details of our implementation, but I suspect it may be related to the fact that pi radians can't be expressed as a double [09:37:01.0000] I suspect so too. It looks like an egregious rounding error. [09:37:02.0000] My tests tolerate up to 1.5px; this is 2px. [09:38:00.0000] (depending on platform) [09:38:01.0000] (1px for me) [09:38:02.0000] 1px for me too [09:39:00.0000] Bah. [09:39:01.0000] I wish I could make the tolerances lower. [09:39:02.0000] You'd think we could at least require some precision for getClientBoundingRect(). [09:39:03.0000] Even if we don't for actual layout. [09:40:00.0000] Although I guess the two should match up. [09:40:01.0000] It makes no sense if it doesn't much the actual layout box [09:41:00.0000] I guess. [09:41:01.0000] Can't we just require precision in layout too? [09:41:02.0000] Bah on CSS. [09:42:00.0000] it's kind of funny how CSS once led the effort to more precision and is now the one that is behind [09:42:01.0000] kind of similar to the industrial revolution in England [10:55:00.0000] Yay, Gecko nightly crash. [10:55:01.0000] /me isolates [11:33:00.0000] forums thread on www-style is now up to 21 messages. "Do you know who also used forums? Nazis." [11:33:01.0000] True fact! [11:38:00.0000] from mattwilcox.net: "To argue effectively, don’t talk too much." [11:38:01.0000] Hahaha, that would be a totally jerk move to bring up. ^_^ [11:38:02.0000] dglazkov: :) [11:46:00.0000] TabAtkins, what format do CSS people like spec bug reports in? [11:47:00.0000] I've found a few in CSS 2D Transforms. [11:47:01.0000] Email? [11:47:02.0000] Should I make it one issue per e-mail, or does it not matter? [11:48:00.0000] Personally, I prefer large issues that might require discussion to be separate emails, and small issues (typos, small changes) to be collected together. [11:48:01.0000] I'll send one big e-mail and see what happens. [11:49:00.0000] That works too. When discussion starts happening with an email like that, I just split a new thread off myself. [11:49:01.0000] AryehGregor: we're using bugzilla to track issues in the transforms/transitions/animations specs [11:49:02.0000] hober, okay. [11:49:03.0000] AryehGregor: but emailing www-style is fine too [11:50:00.0000] Ah, you're one of the editors, good. [11:50:01.0000] AryehGregor: nice tests btw [11:50:02.0000] I'll file bugs. [11:50:03.0000] thanks [12:00:00.0000] AryehGregor: I'm out of CSS [12:01:00.0000] annevk, oh, you are? Maybe you should remove your name as editor from CSSOM stuff, then. [12:01:01.0000] I was kind of hoping the new editors would have done that by now :/ [12:01:02.0000] Most specs list former editors, with dates. [12:15:00.0000] hsivonen: do you think Gecko is willing to standardize it's "universal charset detector"? [12:15:01.0000] hsivonen: as another step towards deterministic encoding detection [12:20:00.0000] Hey, we've hired tantek and AryehGregor, we might be able to standardize stuff now :) [12:21:00.0000] :) [12:21:01.0000] annevk, looking for a job, btw? ^^ [12:21:02.0000] /me is in the depths of specing/fixing text-overflow details [12:22:00.0000] Ms2ger: if you're hiring, maybe [12:23:00.0000] Well, I don't know who'd actually have to decide that, but perhaps [12:23:01.0000] and I just learned iso-2022-jp-2 was added to Gecko for fun times https://bugzilla.mozilla.org/show_bug.cgi?id=72468 [12:24:00.0000] Ms2ger: you mean you're actually a team? [12:24:01.0000] I'm a volunteer [12:24:02.0000] I have no idea what MoCo people do [12:25:00.0000] with "you" I meant you, not MoCo ;) [12:25:01.0000] I wish I was a team [12:27:00.0000] annevk, you seem to be leaving messages for other annevks, so I should make it clear you'd only get one paycheck ;) [12:27:01.0000] the more I read about iso-2022-jp-2 the more I wish I didn't [12:27:02.0000] Ms2ger: hehe [12:28:00.0000] And yeah, if people threw patches at us, we tended to take them rather easily [12:29:00.0000] in this case the module owner actually agreed it would be nice to have [12:29:01.0000] but yeah, a couple of years later there was the XForms stuff [12:30:00.0000] :/ [12:32:00.0000] after that we got more organized as browsers I think, with SVG sort of in between [12:32:01.0000] though I suppose now you could argue about multimedia standards [12:32:02.0000] which are prolly some new kind of mess [12:37:00.0000] also [12:37:01.0000] the character encoding fatal stuff [12:37:02.0000] is not nearly implemented everywhere [12:37:03.0000] for all encodings that is [12:37:04.0000] and it's basically only for XML [12:38:00.0000] Go, XML! [12:38:01.0000] "XML tax" [12:38:02.0000] (which in turn somewhat compliantly implemented by browsers because of Acid3, which in turn had the test donated by annevk... o_O) [12:39:00.0000] was somewhat* [12:42:00.0000] Well, if you're going to blame yourself... [12:42:01.0000] Then I don't have to do it :) [12:45:00.0000] http://czyborra.com/utf/#UTF-9 (UTF-7,5) [12:46:00.0000] Ms2ger: I conveniently forgot if I did it to point out how silly XML was or because I actually believed it was a good idea [12:47:00.0000] http://czyborra.com/utf/#UTF-1 is that the same Glenn Adams mentioned in the footnote? [12:48:00.0000] not sure why I'm reading this, back to iso-2022-jp... [12:51:00.0000] http://unicode.org/mail-arch/unicode-ml/Archives-Old/UML009/0011.html hehe [12:51:01.0000] about UTF-8: "It shall be succeeded by UCS-2/4 in the long run." [12:51:02.0000] if I was John Gruber, I could have a field day [12:53:00.0000] John Gruber lives in Philly and is a giant d*ck. annevk, you're 0 for 2 on that. [12:54:00.0000] oh hello Android :p [12:55:00.0000] /me regrets calling anyone a d*ck. [12:55:01.0000] apologies, Internets. [12:56:00.0000] to be clear, I don't always agree with him either, I was just referring to his practice of calling people on silly things they've said in the past [12:59:00.0000] I finally understand why we have 17 planes in unicode. [12:59:01.0000] Oh? [13:00:00.0000] To be forward-compatible with string theory? [13:00:01.0000] lolol [13:00:02.0000] Becasue the surrogate range can encode 16 planes. [13:00:03.0000] And you obviously don't need surrogates to refer to the bmp, so 16+1=17 [13:20:00.0000] Man I missed dglazkov being grumpy [13:20:01.0000] :D [13:20:02.0000] That was like a once-in-a-lifetime oppertunity [13:27:00.0000] what's the easy way to parse out query parameters on a page? [13:27:01.0000] do we have a web platform API for that yet? [13:33:00.0000] abarth was working on a spec about a year ago: https://docs.google.com/a/lvp-media.com/document/edit?id=1r_VTFKApVOaNIkocrg0z-t7lZgzisTuGTXkdzAk4gLU&hl=en [13:34:00.0000] that's now here: http://dvcs.w3.org/hg/url/raw-file/tip/Overview.html [13:34:01.0000] With a WebKit bug here: https://bugs.webkit.org/show_bug.cgi?id=71968 [13:34:02.0000] jamesr_ ^ [13:34:03.0000] proper link for the docs.google.com page: https://docs.google.com/document/edit?id=1r_VTFKApVOaNIkocrg0z-t7lZgzisTuGTXkdzAk4gLU&hl=en [13:36:00.0000] looks like it's mostly ongoing [13:44:00.0000] it's not really being worked on at the moment [13:44:01.0000] hopefully later this year [13:48:00.0000] I must say, I really like G+'s concept of Pages. I guess Facebook has something similar, too. But man, it's so easy to build a blog-like status updatery thing that's also looped into all your social network: https://plus.sandbox.google.com/103330502635338602217/posts [13:51:00.0000] "Not Found" [13:52:00.0000] https://plus.google.com/103330502635338602217/posts [13:52:01.0000] jeez, what a doofus i am. [13:55:00.0000] dglazkov, can you control who has write access to a page? [13:55:01.0000] jamesr_: yep. you can define managers and an owner. [13:55:02.0000] I would really like for G+ to support embedded images though [14:01:00.0000] hmm http://mxr.mozilla.org/mozilla-central/source/intl/uconv/ucvja/nsJapaneseToUnicode.cpp#645 is the same as ASCII [14:02:00.0000] dglazkov: you need a logo [14:02:01.0000] dglazkov: maybe a shadowy figure hiding in the DOM [14:02:02.0000] :) [14:02:03.0000] Like the incognito mode guy! [14:06:00.0000] oh another one is secretly a duplicate too [14:06:01.0000] Ms2ger can remove some code again if he wants to [14:06:02.0000] a shadowy figure with well-documented output ports [14:07:00.0000] jamesr_: That sounds dirty. [14:31:00.0000] which spec defines dom ranges and selections nowadays? [14:35:00.0000] dglazkov: dom range is in DOM4 [14:35:01.0000] selection is ... somewhere else [14:36:00.0000] found it, editing APIs [14:36:01.0000] smaug____, thanks! [14:37:00.0000] I would assume selection handling is very much non-reviewed [15:15:00.0000] what on earth is beforeload event? [15:16:00.0000] dglazkov: rniwa: do you or someone else familiar with webkit know? [15:16:01.0000] smaug____: beforeload event? [15:16:02.0000] smaug____: do we support that? [15:16:03.0000] apparently yes [15:16:04.0000] safari does [15:16:05.0000] I have no idea what it is [15:16:06.0000] ah, is it safari only [15:16:07.0000] well [15:16:08.0000] they implemented it, i'm not sure if it's on in chromium or not [15:17:00.0000] when does it fire [15:17:01.0000] and why [15:17:02.0000] iirc it's used for safari extensions to do resource blocking stuff [15:17:03.0000] http://developer.apple.com/library/safari/#documentation/Tools/Conceptual/SafariExtensionGuide/MessagesandProxies/MessagesandProxies.html [15:17:04.0000] search for 'beforeload' [15:17:05.0000] ah, you can do preventDefault() [15:20:00.0000] terrible name for the event [15:20:01.0000] and without prefixes or anything [15:21:00.0000] *thumbsup* [15:21:01.0000] the safari extensions message-passing model special-cases that event, i believe [15:22:00.0000] all kinds of weird [15:23:00.0000] I wonder whether beforeload handles things like style="background-image" [15:24:00.0000] test it! [15:27:00.0000] smaug____: I don't think so [15:27:01.0000] so, I don't understand then the use case [15:27:02.0000] it is like for ad-block stuff... [15:28:00.0000] but doesn't quite work [15:28:01.0000] i think that's correct [15:29:00.0000] codesearch.google.com/codesearch#search/&exact_package=chromium&q=dispatchBeforeloadEvent%20file:WebCore&type=cs [15:32:00.0000] beforeload is like badly implemented contentpolicy (in gecko) 2012-01-06 [16:01:00.0000] jamesr, location.search? You can parse it with .split(). [16:02:00.0000] AryehGregor, and parse all the = things myself and escape/unescape [16:02:01.0000] smaug____, some parts of the selection spec are quite solid and interoperable. [16:03:00.0000] AryehGregor: I didn't say it wouldn't be. But is it reviewed :) [16:03:01.0000] jamesr_: Use split twice [16:04:00.0000] jamesr_, parsing & and = is just something like: var bits = location.search.replace(/^\?/, "").split("&").map(function(s){return s.split("=")}); [16:04:01.0000] Or something like that. [16:04:02.0000] As for encoding, have fun. IIRC, it's not interoperable. [16:04:03.0000] Nor am I sure it's even specified. [16:04:04.0000] AryehGregor: Yeah, exactly that. [16:04:05.0000] AryehGregor: Well, split("=", 2) [16:04:06.0000] Yeah, I'm sure it could be made more correct. [16:04:07.0000] and you want to decodeURIComponent() each of the parts of that [16:05:00.0000] But apart from that, that's exactly it [16:06:00.0000] location.search.replace(/^\?/, "").split("&").map(function(s){return s.split("=", 2).map(decodeURIComponent);} [16:06:01.0000] oh dear my semicolon has migrated to the left [16:06:02.0000] but otherwise does that look close? [16:06:03.0000] jamesr_: No, that's mapping dUIRC across 2-arrays. [16:06:04.0000] ah snap [16:06:05.0000] Or wait, I'm dumb. [16:07:00.0000] Never mind, you're good. [16:07:01.0000] no wait it's mapping both strings [16:07:02.0000] Yes. [16:07:03.0000] jamesr_: That's right. [16:08:00.0000] location.search.replace(/^\?/, "").split("&").map(function(s){return s.split("=", 2).map(decodeURIComponent)}); [16:08:01.0000] seems to work [16:08:02.0000] Yeah, that should be fine. [16:08:03.0000] btw, MDN is accessing event.layerX/layerY, which are deprecated in WebKit [16:08:04.0000] Kinda nice to get it into an object, but oh well. [16:08:05.0000] throws a warning in the inspector [16:08:06.0000] can some of y'all fix that? [16:09:00.0000] guess i could map that into an object.... [16:09:01.0000] or can you do that? [16:09:02.0000] gsnedders: That's easy. prepend "var query;", append ".map(function(x){query[x[0]]=x[1];})". [16:10:00.0000] Or .forEach? [16:10:01.0000] Alternately, define mapObject or something that expects you to return a 2-array. [16:10:02.0000] AryehGregor: Yeah, forEach would be better semantically. [16:15:00.0000] Note that forEach and map and the like are likely to be fairly slow for at least another year or two [16:16:00.0000] Really? Why? [16:16:01.0000] TabAtkins: In large part because of jumping between C++ and JS threads [16:16:02.0000] s/threads/stacks/ [16:17:00.0000] Interesting. Where is the stack jump? In the function invocation or something? [16:17:01.0000] /me knows nothing about the interface between c++ and js. [16:17:02.0000] heh, i doubt that'll matter at all in this case [16:17:03.0000] are you talking about in a specific JS engine? [16:17:04.0000] i thought map/forEach were implemented natively in v8 [16:17:05.0000] /me checks [16:18:00.0000] Calling into C++ in the first case, then into the function, then out of the function, and then after doing those two a number of times back to JS [16:18:01.0000] /me notes that parsing a single query string is not a situation where performance is of any concern [16:18:02.0000] jamesr_: I thought they originally were in V8 but aren't any more. [16:18:03.0000] http://google.com/codesearch#OAMlx_jo-ck/src/v8/src/array.js&q=forEach&ct=rc&exact_package=chromium&cd=8&sq=&l=1039 [16:18:04.0000] i'm not sure exactly what that is [16:18:05.0000] gsnedders: Is that because map is implemented in C++? [16:19:00.0000] Rather than as a JS for-loop? [16:19:01.0000] TabAtkins: Right. [16:19:02.0000] kk [16:25:00.0000] http://jsperf.com/map-builtin-v-js/5 is about as close as you can get with pure JS [16:26:00.0000] In Chrome the native function is the quickest map impl because it has all its magic functions to optimize it better (IIRC some of them have specific low-level impls) [16:26:01.0000] right, the built-ins aren't strictly speaking straight JS [16:27:00.0000] but they don't have to make a full JS/C++ stack jump [16:27:01.0000] jamesr_: Right [16:27:02.0000] at least that's my understanding. i'm not a JS VM engineer, i just play one on IRC [16:28:00.0000] jamesr_: That is the case, they don't have any stack jump, they're just another JS function from the engine's POV, it's just some things they call are magic [16:28:01.0000] JSC I don't really know why it's perf is as it is, and SpiderMonkey is just a matter of how slow the JS impl is mainly, AFAIK [16:30:00.0000] /me doesn't know enough about what's happened with JSC in the past year really, it's kinda sad [16:33:00.0000] /me will bear this in minf when he has to parse a million key vlue pairs out of the query string [16:33:01.0000] Hey, it's a viable DoS attack, as has been proven recently! :P [19:48:00.0000] hmm [19:48:01.0000] when parsing a URL, how do you canonicalize the port? [19:49:00.0000] On the basis of the scheme? [19:49:01.0000] Or am I missing something? [19:59:00.0000] gsnedders: dunno [19:59:01.0000] it's missing from abarth URL parsing draft [19:59:02.0000] http://tools.ietf.org/html/draft-abarth-url-01#section-5 [20:00:00.0000] he has "If the canonicalized-port is non-empty and is not the default port for the scheme" [20:00:01.0000] which is the part you mean, I think [20:00:02.0000] so that's clear [20:01:00.0000] but that step is done after the port is "canonicalized" [20:01:01.0000] port - one of the few pieces of a URL that people seem to agree on for the most part [20:01:02.0000] heh [20:01:03.0000] yeah [20:01:04.0000] port or port number [20:01:05.0000] or if you're a PHP person, SERVER_PORT [20:01:06.0000] MikeSmith, you've seen this right? http://tantek.com/2011/238/b1/many-ways-slice-url-name-pieces [20:02:00.0000] so not clear to me, given just a port number without knowledge of the scheme, what exactly do I do to canonicize it? [20:02:01.0000] /me looks [20:02:02.0000] tantek: ah yeah [20:02:03.0000] that's quite a diagram [20:03:00.0000] abarth introduces the term "after-scheme" for all the stuff collectively that is not the scheme [20:03:01.0000] I like that [20:03:02.0000] I have two more rows to add too [20:04:00.0000] oh? [20:04:01.0000] oh great [20:04:02.0000] make that 3 [20:04:03.0000] then [20:04:04.0000] is this an RFC track thing? draft-abarth-url-01 [20:04:05.0000] in which case, maybe I'll wait until there's a permalinkable spec for it [20:05:00.0000] since IETF doesn't seem to believe in permalinks for drafts [20:05:01.0000] :P [20:05:02.0000] the tools.ietf.org/html/ links are pretty much stable [20:05:03.0000] The actual drafts vanish after 90 days, but they stay around there forever :P [20:06:00.0000] GO LOGIC! [20:06:01.0000] that ID is being not being maintained any longer [20:06:02.0000] s [20:06:03.0000] so it won't move forward at the IETF [20:06:04.0000] huh [20:06:05.0000] our plan is to put all the URL stuff into a single spec [20:06:06.0000] and publish it at W3C [20:06:07.0000] /me will await a permalink for that spec then [20:07:00.0000] tantek: http://dvcs.w3.org/hg/url/raw-file/tip/Overview.html [20:07:01.0000] I am working on it as we speak [20:07:02.0000] a-ha ok [20:08:00.0000] just taking the work that abarth has already done and combining into it one doc, along with some bits from the HTML spec [20:09:00.0000] so the plan is for this doc to to have the definition of "URL" and "valid URL", then algorithms for dealing with URLs, then the URL API [20:09:01.0000] ah - looks like the terms are a re-use of 1996 DOM window.location [20:09:02.0000] so that's good, not a new terminology at least [20:10:00.0000] OK [20:10:01.0000] well, none of it is set in stone of course [20:11:00.0000] the main goal for now is to get it all in one place and then get it out for review [20:11:01.0000] and the high-level goal is to "get it right" [20:11:02.0000] including to have it actually reflect reality [20:12:00.0000] or realities plural [20:12:01.0000] in that the browser behavior for URL parsing it not completely interoperable [21:35:00.0000] http://lists.w3.org/Archives/Public/public-webapps/2012JanMar/0023.html was reported before and I think it was concluded to not be a security issue [21:35:01.0000] don't remember why though, abarth? [21:36:00.0000] dunno [21:36:01.0000] sounds like a server issue [21:37:00.0000] we can't account for all manner of crazy transformations servers might do [21:38:00.0000] yeah, and I sort of doubt even though would let Cotnent-Length be spoofed that way [21:38:01.0000] because it's a higher-level library [21:39:00.0000] you should see the proxies that permute letters in Referer :) [21:40:00.0000] heh, I'll stay semi-naive about HTTP for now [21:40:01.0000] encodings are enough fun as it is [21:41:00.0000] also crazy [21:44:00.0000] that we're still figuring out document.write() [21:45:00.0000] Someone mentioned before that it was invented in Netscape 2 [21:45:01.0000] And as I said before, Netscape 2 did not parse HTML into a tree, instead it treated tags as commands. [21:46:00.0000] I don't think async script loading was supported at all. [21:48:00.0000] That was why it was considered a simple feature in Netscape 2 [21:49:00.0000] And probably why Netscape 2 to 4 supported accessing only a limited type of elements. [21:49:01.0000] IE4 actually parsed HTML into a tree, that was why it was able to invent document.all. [21:50:00.0000] and then we had a lot of problems [21:50:01.0000] Which is yea when the problems start. [21:52:00.0000] The problem was made even worse with DOM Level 1 which allow full manipulation of the document tree. [21:52:01.0000] gmail is sure going to lengths to make everyone acutely aware of the major inherent flaws of web apps [21:52:02.0000] you like how things are? too bad, we're making everything worse and there's nothing you can do about it [21:54:00.0000] For example, here is a 0-day security vulnerability reported in Mozilla as late as 2010: https://bugzilla.mozilla.org/show_bug.cgi?id=607222 [23:50:00.0000] quite proud of http://dvcs.w3.org/hg/encoding/raw-file/tip/Overview.html#iso-2022-jp [23:50:01.0000] waking up early works [00:27:00.0000] heh [00:27:01.0000] annevk: I checked in changes to http://dvcs.w3.org/hg/url/raw-file/tip/Overview.html [00:27:02.0000] I think most of the necessary bits are all there now [00:28:00.0000] though lots and lots of TODOs [00:30:00.0000] annevk: also, the book is on the way to you [00:31:00.0000] sweet [00:39:00.0000] kind of a sad caption http://xkcd.com/1000/ [00:40:00.0000] but the math joke is fun :) [00:41:00.0000] MikeSmith: nice [00:41:01.0000] MikeSmith: also, you didn't set all the id attributes :p [00:42:00.0000] eh? which ones needs set? [00:42:01.0000] I rather you don't, but you let them generate automatically and I remember you didn't like that [00:43:00.0000] (I tried making a joke :)) [00:49:00.0000] there's a lot of RFCs btw attempting to define encodings [00:49:01.0000] people put some BNF in there, a couple of words, and put a sticker on it [02:38:00.0000] P { font-size: 12pt ! legal "IATA regulations" } [04:24:00.0000] Ms2ger: euh? [04:24:01.0000] http://www.w3.org/TR/WD-css1-951209.html#legal [04:34:00.0000] well thank you [04:34:01.0000] I never understood why CSS had such a silly !important thing and even less why it was exposed as "priority" in the OM [04:45:00.0000] AryehGregor: how is http://w3c-test.org/webapps/DOMCore/tests/approved/Range-surroundContents.html approved with test errors? [04:47:00.0000] AryehGregor: jgraham or Velmont might provide a patch I hear [04:49:00.0000] Because nobody actually reviewed [04:49:01.0000] AryehGregor: other comment "and it doesn't seem like the test follows step 10 in the DOMRange.insertNode spec" [06:59:00.0000] So, as a totally experimental thing, I tried cloning the html testsuite repository onto github [06:59:01.0000] https://github.com/jgraham/html-testsuite/ [06:59:02.0000] Mainly for the possibility of code review [07:00:00.0000] do you want @WHATWG to announce it somehow? [07:00:01.0000] and if so, how? [07:01:00.0000] I have no strong desire for it to be announced [07:02:00.0000] More than I just did by writing on public, logged, irc, I mean [07:02:01.0000] Let me work out if it is useful first :) [07:03:00.0000] k [07:09:00.0000] Is Hixie on vacation? [07:10:00.0000] well he's not here [07:16:00.0000] Does anyone know about IE's support for big5 / big5-hkscs? Ideally based on what Windows 7 ships [07:16:01.0000] http://longsonr.wordpress.com/2012/01/06/svgtests-interface-implemented/ <- sigh [07:16:02.0000] not interested in pre-Vista [07:17:00.0000] jgraham: aah [07:18:00.0000] jgraham: do we ship similar crap? [07:20:00.0000] annevk, what do you mean by "here"? [07:22:00.0000] standards world [07:22:01.0000] oh well [07:22:02.0000] a [07:22:03.0000] n [07:23:00.0000] hmm http://www.microsoft.com/hk/hkscs/ seems to suggest that via Unicode you get there, but not via big5-hkscs [07:24:00.0000] annevk, what specific questions are you asking with regard to big5 / big5-hkscs ? [07:25:00.0000] I think I found my answers already [07:25:01.0000] ok [07:26:00.0000] Windows has PUA-support for the characters HKSCS introduced and the label "big5-hkscs" means "big5" which uses PUA afaict [07:26:01.0000] but then other browsers support "big5-hkscs" independently from "big5" (different encoding) and do not map to PUA [07:26:02.0000] Yeah, that PUA story is quite famous. [07:27:00.0000] I see. [07:27:01.0000] so I guess if you are in HK you really want UTF-8 [07:28:00.0000] I've not seen a big5-hkscs page but I am not in HK either. These characters are out of BMP right? [07:30:00.0000] maybe? I've no idea [07:35:00.0000] PUA support. [07:35:01.0000] wait, what. [07:36:00.0000] annevk: I'm not sure. It wouldn't surprise me, but hopefully we're not *adding* that kind of crap years after it became obvious that it's a design anti-pattern [07:38:00.0000] It's obviously odd if PUA in DOM can get displayed as normal text. [07:41:00.0000] system fonts [07:44:00.0000] You doom us all to inhuman toil for the One whose Name cannot be expressed in the Basic Multilingual Plane. [07:49:00.0000] is the whole BMP full? [07:50:00.0000] No I don't think so. [07:50:01.0000] No [07:50:02.0000] at least Thai has some space I just found out [08:08:00.0000] "Our new years resolution should be at least 1920x1080" [08:11:00.0000] neh [08:12:00.0000] higher! [09:05:00.0000] good morning, Whatwg! [09:11:00.0000] hyvää iltaa [09:13:00.0000] I love umlauts. [09:20:00.0000] those aren't actually umlauts, but separate letters ;) [09:22:00.0000] letters with umlauts? [09:22:01.0000] ä is a letter itself [09:23:00.0000] what is the pronunciation effect of two of them? [09:24:00.0000] You mean having two ä-letters? It is just longer sound [09:24:01.0000] Finnish is easy to pronounce [09:25:00.0000] is it pretty much speak out each letter? [09:25:01.0000] like Russian [09:25:02.0000] yeah [09:26:00.0000] that's nice. What about grammar? [09:27:00.0000] well, you can put words almost in any order, so it is easy. [09:28:00.0000] but picking up the right form of some word can be tricky [09:29:00.0000] If I trust one source...it says there can be 2253 different forms of every noun [09:29:01.0000] sounds about right [09:29:02.0000] o_O [09:30:00.0000] but there's logic to that, right? You don't have to just memorize? [09:30:01.0000] yeah, there is quite simple logic [09:31:00.0000] http://en.wikipedia.org/wiki/Finnish_morphology#Noun_forms [09:35:00.0000] ah! very similar to Russian: en.wikipedia.org/wiki/Russian_grammar [09:36:00.0000] annevk, the test bugs in surroundContents are due to changes in testharness.js after the test was written, which are in turn due to spec changes. [09:36:01.0000] /me fixes [09:37:00.0000] dglazkov: your letters make that all hard :) [09:37:01.0000] :) [09:38:00.0000] annevk, hmm, so we throw InvalidStateError for two things in surroundContents? That's not good. [09:39:00.0000] AryehGregor: do you have somewhere spec'd how window.getSelection() reacts to selections in iframes? [09:40:00.0000] dglazkov, seems not right now, no. I think I have feedback asking me to look into it, which I plan to do. [09:41:00.0000] AryehGregor: cool. I'll just write up something that makes sense for shadow DOM (which is a similar situation). [09:41:01.0000] dglazkov, I think the current theory is there's one selection per window, so iframes have a different selection. [09:41:02.0000] ah, so it's not similar. [09:42:00.0000] annevk, pushed a fix to the test bugs. [09:42:01.0000] Or not? [09:42:02.0000] Oh, right, the fix isn't approved. [09:55:00.0000] /me is amazed how www-style can have such a long and silly discussion about using forums [09:57:00.0000] yikes, why is getSelection on both window and document? [09:57:01.0000] historical reasons? [09:58:00.0000] probably [09:58:01.0000] ""Deprecated method document.getSelection() called. Please use window.getSelection() instead."" was added to Gecko 2001 [09:59:00.0000] I think that warning has been removed [10:00:00.0000] maybe AryehGregor knows [10:00:01.0000] dglazkov, I don't know the history. But that's what implementations do. [10:01:00.0000] Until a few months ago, Gecko implemented document.getSelection() as returning a string, but now I think we have interop. [10:01:01.0000] Apparently document.getSelection has been in Gecko since 1998, so it could be Netscapeism [10:02:00.0000] Yes, I'm pretty sure it dates to Netscape. [10:02:01.0000] Maybe it was originally on Document only, but then people realized it made more sense on Window. [10:04:00.0000] fascinating [10:05:00.0000] window.getSelection was added to Gecko 2000 [10:46:00.0000] smaug____: does Gecko support multiple-range selections? Or is this all deprecated? [10:46:01.0000] Gecko does support multiple selections [10:46:02.0000] I mean, multi-range [10:47:00.0000] ok [10:48:00.0000] dglazkov, yes, it does. rniwa and I get ehsan to agree at some point that it shouldn't, but I dunno if the rest of Gecko is on board with that. [10:48:01.0000] why are multi-range selections sucky? [10:49:00.0000] that is not clear to me [10:50:00.0000] but IIRC other browser vendors don't want to implement them [10:54:00.0000] at least to me, multi-range selections are the thing that make selections work non-crazily with shadow DOM, so I am interested [10:55:00.0000] dglazkov, because 98% of the time there's only one Range, so authors do getSelection().getRangeAt(0) and expect it to be the only range in the selection, which breaks. [10:56:00.0000] If we need non-contiguous selections, a better API is required, like maybe one that returns all nodes contained in the selection instead of endpoints. One where the non-contiguous selection is not exposed as a special case in the API, because it's too rare for anyone to pay attention to. [10:56:01.0000] FWIW, Gecko itself internally has crazy bugs with multi-range selections. [10:56:02.0000] ehsan demoed some of them with execCommand(). It does pretty much random things when multi-range selections are involved. [10:56:03.0000] If Gecko engineers can't get multi-range selections right, do you expect web authors to? [10:57:00.0000] I see. [10:59:00.0000] AryehGregor: smaug____: yeah, I meant to start that discussion at some point, but I held off on it cause I never had the time to rip them out [10:59:01.0000] and I didn't want the discussion to end with no results [10:59:02.0000] AryehGregor: thanks, I need to think about this some more. [11:00:00.0000] Honestly, the Range API is way too complicated anyway. Like the fact that offsets are sometimes child numbers and sometimes offsets into CharacterData, and the way that you can have a zillion different boundary points that map to the same location. [11:00:01.0000] It's a PITA for authors to use. [11:01:00.0000] Author code is littered with simplifying assumptions like "there's only one range per selection" and "range endpoints are always in Element or Text nodes". [11:01:01.0000] Which they don't even realize are wrong. [11:01:02.0000] damn authors! [11:01:03.0000] why can't they be smarter! [11:01:04.0000] :D [11:02:00.0000] AryehGregor: ehsan: I think mats has fixed most of the bugs related to multi-range selections [11:03:00.0000] er, possibly not related to execCommand [11:03:01.0000] /me is not interested in execCommand :p [11:03:02.0000] smaug____: no, he hasn't :) [11:04:00.0000] multi-range selections is a nice way to implement spellchecking UI, IMO [11:06:00.0000] smaug____: that can be debated, but for the most part what we want is stop exposing them to the content [11:07:00.0000] I feel that the way we use multi-range selections to handle spell checking etc is not optimal [11:07:01.0000] but that's a different discussion [11:19:00.0000] annevk, for big5, are you inclined to spec Gecko's big5 table or CP590 of others? [11:26:00.0000] Is the idl from the html5 spec available somewhere without all the other verbiage? [11:41:00.0000] bz_moz, the source code of this has a copy-pasted version with a few tweaks: w3c-test.org/html/tests/submission/AryehGregor/interfaces.html I used this to generate it from the one-page WHATWG spec: var s = ""; [].forEach.call(document.getElementsByClassName("idl"), function(idl) { if (!idl.classList.contains("extract")) { s += idl.textContent + "\n\n"; } }); document.body.innerHTML = '
';     d

[11:41:01.0000] 
ocument.body.firstChild.textContent = s;

[11:41:02.0000] 
(see http://w3c-test.org/html/tests/resources/idlharness.js )

[11:43:00.0000] 
AryehGregor: looking

[11:44:00.0000] 
AryehGregor: thanks, that's perfect

[11:44:01.0000] 
:)

[11:44:02.0000] 
It might be slightly out-of-date.

[11:45:00.0000] 
I just need the interface hierarchy

[11:45:01.0000] 
not the members

[11:45:02.0000] 
I doubt that's changed

[11:50:00.0000] 
bz_moz, since transform-origin isn't inherited, the only way the computed value is detectable is through getComputedStyle(), no?  Having different resolved and computed values is only meaningful for inherited properties, because it's the computed value that's inherited?

[11:50:01.0000] 
Or is computed value used somewhere other than inheritance and getComputedStyle()?

[11:53:00.0000] 
AryehGregor: all properties can be inherited

[11:53:01.0000] 
AryehGregor: some are just not inherited by default

[11:53:02.0000] 
AryehGregor: anyone can always write "transform-origin: inherit" and boom!

[11:54:00.0000] 
AryehGregor: as long as your dimensions are different from your parent you can now tell whether the computed value was percent or pixels

[11:54:01.0000] 
AryehGregor: resolved value is a concept specific to getComputedStyle

[11:54:02.0000] 
AryehGregor: because UAs historically do weird shit there, because the CSS2 definition of computed value was .... odd

[12:46:00.0000] 
i have returned!

[12:47:00.0000] 
and won't be reading backlog, so if you asked me something, ask again or send mail :-)

[12:47:01.0000] 
...where did you go?

[12:47:02.0000] 
/me has literally 1000s of e-mails to deal with

[12:47:03.0000] 
...when did you leave?

[12:47:04.0000] 
TabAtkins: vacation

[12:49:00.0000] 
hixie: were you gone for more than 3 days?

[12:50:00.0000] 
more like 3 weeks

[12:50:01.0000] 
Wow.

[12:50:02.0000] 
hixie: then why do you need the "literally"?  ;)

[12:50:03.0000] 
I... guess it's a good thing I didn't notice?

[12:50:04.0000] 
bz_moz: fair point! :-)

[12:50:05.0000] 
TabAtkins: :-P

[12:50:06.0000] 
/me would figure that 2000/week is a good lower bound for email

[12:50:07.0000] 
actually this is after my filters get rid of 90% of my mail

[12:50:08.0000] 
bz_moz: Surely your silicon brain can handle more than that.

[12:51:00.0000] 
hixie: ah, heh

[12:51:01.0000] 
Until I see you at SXSW and satisfy myself that your'e an organic, I'm going to continue to assume you're a supercomputing cluster in an MIT basement.

[12:51:02.0000] 
TabAtkins: my laptop can probably handle all sorts of mail volumes

[12:52:00.0000] 
heh

[12:52:01.0000] 
I have no evidence you're organic either

[12:52:02.0000] 
Please.  My effiency and work ethic is clearly organic.

[12:52:03.0000] 
plus, how would you tell apart a human from R Daneel Olivaw anyway?

[12:52:04.0000] 
Invasive x-ray.

[12:53:00.0000] 
/me hopes sxsw passes on those

[12:53:01.0000] 
I'm hoping I get it through airport security.

[12:53:02.0000] 
you'd have to catch me at the airport

[12:53:03.0000] 
heh

[12:53:04.0000] 
/me has taken to traveling with his kids

[12:53:05.0000] 
that way I go through the sane line

[12:53:06.0000] 
Airport x-ray isn't invasive. It would be fooled by organic skin.

[12:53:07.0000] 
at least so far

[12:53:08.0000] 
TabAtkins: yeah, I know

[12:54:00.0000] 
I just opt-out every time.  10 minutes wasting their time with a patdown and I'm good.

[12:54:01.0000] 
and as far as that goes, the TSA actually rolled back a dumb policy

[12:54:02.0000] 
Unless, like last time, they detect explosives on me and call the cops.

[12:54:03.0000] 
TabAtkins: heh

[12:54:04.0000] 
TabAtkins: I've been under too much time pressure to do that the one time it came up.  :(

[12:55:00.0000] 
I've done a pornoscanner once in a similar timecrunch, but I usually ensure I have 1.5 hours so I can eat the time.

[12:57:00.0000] 
what I want is confirmation of ms2ger's humanity

[12:58:00.0000] 
but I may be out of luck.

[12:58:01.0000] 
Which humanity?

[12:58:02.0000] 
I want conformation he is non-human

[12:58:03.0000] 
It would be way more interesting

[12:58:04.0000] 
But less surprising

[12:59:00.0000] 
Still passing my Turing test?

[12:59:01.0000] 
bz_moz, ah, right.  Thanks.

[12:59:02.0000] 
Ms2ger: oh, sure

[12:59:03.0000] 
Ms2ger: but at this point chatbots do _that_

[12:59:04.0000] 
Ms2ger: (you do a bit better, because you comment on patches, which most chatbots can't)

[13:00:00.0000] 
True.  Cleverbot is pretty human most of the time.

[13:00:01.0000] 
tabatkins: more precisely the turing testers are not doing it right

[13:00:02.0000] 
tabatkins: imho

[13:00:03.0000] 
bz_moz, hmm, I want a bot to complain about style issues in random people's patches :)

[13:00:04.0000] 
Another 10 years, and the turing test will stop being useful for testing AI, because we'll just assume that making conversation is something that *of course* computers can do.

[13:01:00.0000] 
(10y is a conservative estimate)

[13:01:01.0000] 
Ms2ger: http://beaufour.dk/jst-review/

[13:01:02.0000] 
bz_moz, yeah, but I want it to comment in bmo

[13:01:03.0000] 
TabAtkins: depends on conversation and demographic.  random party conversation, sure

[13:01:04.0000] 
TabAtkins: which is why you have to find a common area of interest and specialize

[13:02:00.0000] 
TabAtkins: e.g. I'm pretty sure I could prove to myself that you're not a bot

[13:02:01.0000] 
And it's not useful for me, because I deliberately don't rev uuids in patches I put up for review

[13:02:02.0000] 
TabAtkins: we'd just talk css specs

[13:04:00.0000] 
Well, sure, because Texas doesn't have good AI research.

[13:04:01.0000] 
/me had no idea texas was involved

[13:04:02.0000] 
It's where I was raised.

[13:05:00.0000] 
if we're talking that....

[13:05:01.0000] 
On the other hand, you're somewhere up in the northeast, where all the fancy universities are.

[13:05:02.0000] 
neither does dc

[13:05:03.0000] 
(amount of I in general there could be debated)

[13:20:00.0000] 
zewt, thanks for your reply about encoding/XHR/json, that's why I wanted the guy to tell public-webapps about his bug :)

[14:31:00.0000] 
matjas, matijsb: I forget which of you is which, so I'll just say: the CSS commits feed is back to working.  It's not showing the diffs yet, but you don't need that for the Twitter feed.

[14:33:00.0000] 
TabAtkins: hehe, this is matjas'

[15:17:00.0000] 
"This is an editor's draft of a spec, it's not a recommendation, so it's hardly a violation of anything." <- losing a lot of respect for webkit here

[15:17:01.0000] 
timeless, "anyone have experience cajoling MXR into doing regexp searches?  I'm trying to get it to search for '#\d+[#=]' (in Perl-speak), and I'm getting nowhere"

[15:17:02.0000] 
cc jwalden

[15:18:00.0000] 
zewt: url?

[15:18:01.0000] 
public-webapps, the JSON thread

[15:18:02.0000] 
responseType "json" thread on webapps

[15:24:00.0000] 
ah, i didn't read that thread

[15:24:01.0000] 
The HTML5 parser defines a context-sensitive language, right? Has anyone tried to do a formal definition of it?

[15:24:02.0000] 
are you saying the HTML spec isn't formal? :-P

[15:25:00.0000] 
Yes. :P

[15:32:00.0000] 
bz_moz: you get too much email :)

[15:33:00.0000] 
bz_moz: having said that, I think in three months I had about 20000, so maybe you are right... depressing

[15:33:01.0000] 
three months away*

[15:33:02.0000] 
Hixie: Can you think of anything that stops it from being a CSG?

[15:34:00.0000] 
kennyluck: 950, which I thought is what Gecko implemented

[15:34:01.0000] 
AryehGregor: thanks

[15:35:00.0000] 
wb Hixie

[15:44:00.0000] 
gsnedders: CSG?

[15:44:01.0000] 
Hixie: Context sensitive grammar

[15:44:02.0000] 
/me will be scared if it isn't

[15:44:03.0000] 
document.write()?

[15:44:04.0000] 
Because if it isn't, you need a Turing machine to parse it.

[15:44:05.0000] 
you do

[15:44:06.0000] 
Hixie: Assume scripting disabled

[15:45:00.0000] 
oh

[15:45:01.0000] 
(Once you have JS involved, obviously a Turing machine is needed)

[15:45:02.0000] 
do you mean conforming syntax only too?

[15:45:03.0000] 
No.

[15:45:04.0000] 
gsnedders: take a few classes off and figure it out

[15:46:00.0000] 
gsnedders: I'm sure mankind will thank you, and university might hand you a few credits

[15:47:00.0000] 
gsnedders: i'm not familiar enough with the strict definition of CSG to say one way or the other

[15:47:01.0000] 
gsnedders: certainly in the common case -- scripting enabled -- parsing can be non-deterministic

[15:47:02.0000] 
My gut says it is a CSG, but I don't have proof for it.

[15:47:03.0000] 
Hixie: Certainly. document.write guarantees that.

[15:49:00.0000] 
gsnedders: does the fact that HTML streams can be unbounded make it not a CSG?

[15:50:00.0000] 
No.

[15:50:01.0000] 
wikipedia says a CSG can be parsed by an LBA, and an LBA assumes bounded input as far as i can tell

[15:50:02.0000] 
Hixie: No

[15:50:03.0000] 
but here we are outside my area of expertise

[15:50:04.0000] 
so...

[15:51:00.0000] 
Hixie: Even a regular grammar can handle unbounded input

[15:51:01.0000] 
it seems wikipedia's definitions also has the key sentence "The only restriction placed on grammars for such languages is that no production maps a string to a shorter string."

[15:51:02.0000] 
but i don't really understand what that means

[15:51:03.0000] 
e.g., a null-terminated string can be parsed with a regular grammar, but you don't know when, if ever, you'll reach the null byte.

[15:52:00.0000] 
So inevitably anything higher up in the hierarchy can be unbounded.

[15:52:01.0000] 
k

[15:53:00.0000] 
oh oh

[15:53:01.0000] 
the AAA

[15:53:02.0000] 
it can mutate previously-provided output

[15:53:03.0000] 
does that violate some rule of CSGs?

[15:53:04.0000] 
 and , too

[15:54:00.0000] 
and x

[15:54:01.0000] 
I'm somewhat concerned about the AAA.

[15:54:02.0000] 
aren't we all

[15:54:03.0000] 
What's the effect of ?

[15:54:04.0000] 
mutates the earlier  node

[15:54:05.0000] 
iirc

[15:54:06.0000] 
(adds attributes)

[15:55:00.0000] 
Oh, is this because the parser has some special-case for ?

[15:55:01.0000] 
That's not an issue. You're still creating the html node.

[15:55:02.0000] 
That adds the token first?

[15:55:03.0000] 
 parses the same as  iirc

[15:55:04.0000] 
where you can have any amount of content between each of those tags

[15:55:05.0000] 
But yeah, the fact that AAA can move content an arbitrary distance past arbitrary content inside the table would probably break a CSG.

[15:56:00.0000] 
Same with  attribute merging, yeah.

[15:57:00.0000] 
Yeah, you'd need a two-stack DFA, which is turing-equivalent.

[15:57:01.0000] 
Is a two-stack DFA Turing-equiv? I thought you could do that with a CSG.

[15:57:02.0000] 
Yes, it is.

[15:58:00.0000] 
If i'm remembering my automata theory properly

[15:58:01.0000] 
Shit, I meant PDA.

[15:58:02.0000] 
A two-stack PDA.

[15:59:00.0000] 
Well, a PDA is basically just a one-stack DFA

[15:59:01.0000] 
Hm. You may be able to get by with a nested-stack PDA, which is still a type 1 language.


2012-01-07
[16:00:00.0000] 
I don't think you can, given the fact you can invoke AAA at any point.

[16:00:01.0000] 
Just push the entire in-table contents into a stack until you hit 
, looking for AAA-stuff and parsing appropriately.  Then, emit  and start parsing the built-up in-table stack.

[16:01:00.0000] 
Yeah, that's the vague sort of approach I was thinking of for 

[16:01:01.0000] 
But I'm not sure you can share the stack with that of open elements.

[16:01:02.0000] 
I used to have a PDA. I had to put a couple of AAA's into it to make it work.

[16:02:00.0000] 
Yeah, I'm not certain either.  I'm not really familiar with nested-stack machines.

[16:03:00.0000] 
In fact, you almost certainly can't

[16:03:01.0000] 
Well, I'm fairly certain that HTML is parsable by a LBA, since all actual computers are that.

[16:04:00.0000] 
And we have bounds on the size that some stacks can grow.

[16:04:01.0000] 
I guess we'd need to make sure that there aren't any unbounded stacks.

[16:04:02.0000] 
Stack of open elements is unbounded.

[16:04:03.0000] 
Ah, right.  There you go, then.

[16:04:04.0000] 
All the way past CSG, I think.

[16:05:00.0000] 
Unbounded stacks can exist in an LBA

[16:05:01.0000] 
and in a PDA too

[16:05:02.0000] 
You only need a finite stack for a given finite input

[16:05:03.0000] 
(btw, a two-stack pda is turing-equiv because the two stacks are equivalent to a single tape with a bidirectional head)

[16:05:04.0000] 
(Oh, duh)

[16:06:00.0000] 
/me sulks because nobody liked his joke

[16:06:01.0000] 
Oh, wait.  Let's see.  Are there any stacks that grow faster than linear in the input size?

[16:06:02.0000] 
The stack of formatting elements did at some point, I think.

[16:06:03.0000] 
No.

[16:06:04.0000] 
Okay, then we're good.

[16:06:05.0000] 
It's likely an LBA then.

[16:06:06.0000] 
At some point, yes. But it has a limit now.

[16:06:07.0000] 
And thus a CSG.

[16:07:00.0000] 
It's going to get horrifically messy, but I think it's doable.

[16:07:01.0000] 
Yes, definitely.

[16:07:02.0000] 
Yeah, we're definitely an LBA.

[16:08:00.0000] 
Though, honestly, a proof of such just requires that it's (a) parsable by a turing machine, and (b) has no memory requirements that grow faster than linearly.

[16:09:00.0000] 
i have no idea what the heck you are all talking about, but let me know if you come to a decision that affects the spec :-D

[16:09:01.0000] 
dglazkov: if it helps, i thought it was amusing. :-P

[16:09:02.0000] 
yay!

[16:09:03.0000] 
Hixie: It mostly has implications on proving the correctness of implementations

[16:11:00.0000] 
TabAtkins: What I wonder is whether a nested stack will do

[16:12:00.0000] 
Yeah, I'm not actually familiar with nested-stack, and the wikipedia article is minimal.

[16:16:00.0000] 
Oh well, next year I get to learn about grammar from two POVs at uni.

[16:16:01.0000] 
(From both a CS POV and a theoretical linguistics one)

[16:18:00.0000] 
zewt: "This is the worst thing I've seen anyone say in here in a long time" sounds like a challenge.

[16:18:01.0000] 
heh

[16:18:02.0000] 
have at thee, archives

[16:19:00.0000] 
Oh, I was just going to say something involving hitler and tentacles.

[16:19:01.0000] 
the world is desensitized to tentacle hitlers

[16:20:00.0000] 
Does foreign content make conforming content context-sensitive?

[16:20:01.0000] 
Hmmm

[16:20:02.0000] 
Because it's allows arbitrary XML?

[16:20:03.0000] 
If so: yes, or at least prevents it from being context-free.

[16:20:04.0000] 
Nah, nested stack will do.

[16:20:05.0000] 
XML is *almost* context-free.

[16:21:00.0000] 
why does google docs always always break with multiple addresses

[16:21:01.0000] 
If only they did end-tags by reversing the name of the tag.

[16:21:02.0000] 
TabAtkins: Doesn't it work if you just push the name of the element to the stack?

[16:22:00.0000] 
Why would you need to do it char by char?

[16:22:01.0000] 
AIUI XML is context-free.

[16:22:02.0000] 
Because that's how I've typically seen grammars defined?

[16:22:03.0000] 
Specifically, {ww | w in {a,b}+} isn't context-free.

[16:23:00.0000] 
And thus, neither is {}

[16:23:01.0000] 
Right, okay

[16:23:02.0000] 
If you tokenize at a higher level, then sure.

[16:23:03.0000] 
Which is effectively what a nested stack parser does.

[16:24:00.0000] 
Ok, that makes sense.

[16:24:01.0000] 
Then... I suspect a nested stack parser would work.

[16:25:00.0000] 
Unless I'm misunderstanding.

[16:27:00.0000] 
No, that makes sense given the minimal wikipedia definition.

[16:32:00.0000] 
So given scripting enabled...

[16:33:00.0000] 
We're a decider, as we halt.

[16:33:01.0000] 
Provably, the HTML5 parser will halt.

[16:35:00.0000] 
(Of course, the system as a whole is Turing complete, because the JS might not)

[16:41:00.0000] 
(Sad reality: I'm starting to wonder whether AAA is the best solution. But I guess I'll get murdered if I suggest that without a good reason for it.)

[17:36:00.0000] 
Man, a 2-day delay in me getting webkit set up was caused by an e/i typo.

[17:36:01.0000] 
;_;

[21:46:00.0000] 
annevk5, use data:text/html;charset=big5,%C7%B1%C7%AB%C7%E5 as an example and you'll see the difference.

[21:47:00.0000] 
(re. Gecko implementing a superset of CP950)

[23:43:00.0000] 
kennyluck: ah I see what you mean; I guess we'll go with the bigger table if it's not too much of a hassle, dunno

[23:49:00.0000] 
annevk, but this is not implemented in browsers besides Firefox so this is sort of a proprietary extension, though I guess it would be nice if IE can adopt it.

[23:51:00.0000] 
hmm, if only Gecko has it, it might be something they can consider removing...

[23:51:01.0000] 
or is it used?

[23:51:02.0000] 
It is used. This is an example → http://www.ptt.cc/bbs/C_Chat/index.html

[23:54:00.0000] 
Not so frequently though, basically Japansese kanas in big5. People rarely complain about this problem in other browsers so removing it doesn't harm much either…

[23:58:00.0000] 
kennyluck: interesting

[00:02:00.0000] 
I don't really get the bar as to what "breaks the web". The encoding detection step can obviously be arbitrarily complex just to save more pages...

[00:05:00.0000] 
kennyluck: it's a judgment call between complexity and making more pages work

[00:44:00.0000] 
kennyluck: any ideas on how to present those data tables btw and what the best location is to retrieve them?

[00:44:01.0000] 
kennyluck: I've currently done them as tables, but it might be better to just use arrays

[00:45:00.0000] 
all single-octet encodings currently defined are effectively arrays of 128 items

[00:45:01.0000] 
multi-octet encodings have arrays that are longer, but there are not that many

[00:46:00.0000] 
there's only one for Japanese for instance

[00:49:00.0000] 
Google - total NIH :) replace all web related thechnologies to own, s/js/dart/ s/http/spdy/ ....

[00:53:00.0000] 
annevk, a button for copying into the clipboard I guess :) nobody reads the table itself anyway.

[00:53:01.0000] 
Is there a standard way to present such a table?

[00:55:00.0000] 
well, a space separated list would be the most reusable format I guess, like this → http://moztw.org/docs/big5/table/moz18-b2u.txt if IETF doesn't have a standard already.

[00:55:01.0000] 
oh good times

[00:56:00.0000] 
the default styling for 
     makes it so that if you go over 9000 (actually 9999) you can no longer see the first digit

[00:57:00.0000] 
lol

[01:00:00.0000] 
/me is looking at the iconv source to find the biggest table → http://git.savannah.gnu.org/cgit/libiconv.git/tree/lib?id=3a33986e1d2c819dc2b8a84684e067cf177d1815

[01:01:00.0000] 
Is cns11643_inv used on the Web?

[01:03:00.0000] 
euc_tw… I guess it's not

[01:03:01.0000] 
http://git.savannah.gnu.org/cgit/libiconv.git/tree/lib/utf8.h looks buggy; still handles six octets

[01:04:00.0000] 
kennyluck: both Opera and Gecko support it I believe in incompatible ways, but also through iso-2022-cn (not sure how compatible it is there)

[01:04:01.0000] 
9000 seems enough, though I am still not convinced why we want to format it in HTML...

[01:05:00.0000] 
Japanese has an index of multiples of 94

[01:05:01.0000] 
so if that could somehow be nicely represented in HTML it might be a nice way for people to look up stuff

[01:05:02.0000] 
on the other hand, a long list of index -> codepoint could also work

[01:06:00.0000] 
in text/plain that is

[01:07:00.0000] 
kennyluck: would you say the single-octet encodings should not be represented as HTML either?

[01:08:00.0000] 
annevk, no.

[01:08:01.0000] 
I mean it should be presented as HTML.

[01:09:00.0000] 
I guess you should use the 
     element :p

[01:11:00.0000] 
for multi-octet tables?

[01:11:01.0000] 
that could work

[01:12:00.0000] 
still need a way to present them

[01:12:01.0000] 
94 characters per row is quite long :)

[01:24:00.0000] 
Larry Masinter: fighting silly registry policies since '98: http://unicode.org/mail-arch/unicode-ml/Archives-Old/UML013/0007.html

[01:24:01.0000] 
(and prolly before)

[01:28:00.0000] 
annevk, by the way, how do you know big5-hkscs is needed for the Web?

[01:32:00.0000] 
there's issues in various bug databases suggesting it is

[01:33:00.0000] 
though I think the way it works is that users use manual override or an extension or some such

[01:33:01.0000] 
I'd love to get more information on that somehow

[01:38:00.0000] 
Hixie, fwiw, I'd also like to get div.impl highlighted, and I don't care about Workers/Websockets

[02:48:00.0000] 
there's various notes suggesting Gecko supports an eight octet sequence in euc-kr, but I cannot find how it's implemented :(

[02:48:01.0000] 
e.g. https://bugzilla.mozilla.org/show_bug.cgi?id=9962

[02:50:00.0000] 
another thing that makes this difficult is that Gecko patches generated files rather than source files: https://bugzilla.mozilla.org/show_bug.cgi?id=134749 :/

[02:53:00.0000] 
although I guess the process is documented so it should be okay

[03:10:00.0000] 
hmm, iso-2022-kr looks like quite the mess

[03:10:01.0000] 
so surprising

[03:47:00.0000] 
can someone explain the purpose of ESC $ ) C in iso-2022-kr to me?

[03:52:00.0000] 
looks like it's meaningless

[03:55:00.0000] 
but adds a ton of complexity

[04:17:00.0000] 
so it seems Gecko only supports ESC $ ) C if it's at the beginning of the file and it will have zero effect

[04:17:01.0000] 
the effect is that you don't get to see ESC $ ) C on the screen

[04:18:00.0000] 
maybe we should just drop it altogether

[06:10:00.0000] 
whoa that XHR discussion blew up a bit

[06:31:00.0000] 
dglazkov: in Web Components, do you plan on making templates have substitution slots as in http://infrequently.org/11/fronteers/fronteers.html#40 ?

[06:53:00.0000] 
kennyluck: http://w3techs.com/technologies/details/en-b5hkscs/all/all

[06:54:00.0000] 
"whose character encoding we know" there's a big caveat if ever I saw one. heh

[06:55:00.0000] 
http://i818.com/ and http://www.consumer.org.hk/website/ws_chi/ are still online and do indeed declare big5-hkscs

[06:55:01.0000] 
wonder how smart its charset guessing is (guessing big5 is probably fairly easy)

[06:56:00.0000] 
they also render fine in IE which reportedly treats big5-hk.. as label for big5

[06:58:00.0000] 
aah actually

[06:58:01.0000] 
IE does not render some characters correctly

[06:59:00.0000] 
I get boxes and UPA code points and in other browsers I get Unicode assigned characters

[06:59:01.0000] 
so I guess you need a HK Windows setup to render such a page correctly in IE

[06:59:02.0000] 
(but you'll still get into trouble if you copy data elsewhere)

[07:53:00.0000] 
okay, hkscs is for later

[07:54:00.0000] 
hkscs / gbk/gb18030 madness / and prolly nuking iso-2022-cn and euc-tw in a way similar to Chrome

[07:54:01.0000] 
oh and then utf-8 and utf-16

[07:55:00.0000] 
and the tables and the encoders

[07:55:01.0000] 
tralala

[07:55:02.0000] 
Job security :)

[07:55:03.0000] 
nah, just good times :)

[07:56:00.0000] 
death to modal encodings

[07:57:00.0000] 
s/modal/

[07:57:01.0000] 
/

[07:57:02.0000] 
be sort of hard to do anything without *any* encodings

[07:58:00.0000] 
could certainly do away with the 99%

[07:58:01.0000] 
#occupyiso-2022-cn

[07:58:02.0000] 
don't say that around F ms2 got there first

[07:59:00.0000] 
was that a joke in a different encoding?

[08:05:00.0000] 
btw Ms2ger / zewt, it's prolly good to stay nice to Jarred; he's fighting the good fight

[08:05:01.0000] 
a good idea*

[08:05:02.0000] 
gotta go now

[08:05:03.0000] 
See you

[08:31:00.0000] 
espadrine: I am thinking about it. Please file a bug here: https://www.w3.org/Bugs/Public/enter_bug.cgi?comment=&blocked=14949&short_desc=%5BExplainer%5D%3A%20&product=WebAppsWG&component=Component%20Model

[08:45:00.0000] 
bold use of "assume" in https://www.w3.org/Bugs/Public/show_bug.cgi?id=15447

[08:46:00.0000] 
"planning to plan to" sure sounds like an academic

[08:50:00.0000] 
"One of my design rules is that the agent may ignore parts of standards, if I deem those parts unnecessarily inelegant." - why bother trying to get the standard changed, then?

[08:51:00.0000] 
"i know better than the rest of the web, sit down everyone"

[08:52:00.0000] 
/me wonders what would happen if there was an opportunity to redesign HTML from scratch without caring at all about compatibility, based on the lessons currently learnt from it

[08:56:00.0000] 
(Also: "I am assuming that, ignoring GeoCities-era websites, usage in the wild is now limited to ad and tracking scripts." - maybe someone could point out that the page on which he is writing that comment uses document.write for useful functionality)

[08:57:00.0000] 
(Maybe Bugzilla counts as GeoCities-era technology, though)

[09:00:00.0000] 
annevk: not usre if there's interest in standardizing chardet, but there is a paper that explains what it does, IIRC

[14:53:00.0000] 
hsivonen: k, guess I'll take a look once I've done everything else

[14:53:01.0000] 
which may take a while

[15:04:00.0000] 
Notes:

[15:04:01.0000] 
* use encoder / decoder as terms

[15:05:00.0000] 
* write as "The gbk decoder (decoder for gbk) is:" for easier referencing

[15:06:00.0000] 
* keep the single-octet tables, but use an external file in Unicode.org-style as normative format (maybe stripping 00-7E)

[15:07:00.0000] 
* see about using Unicode.org-style for the jis table (e.g. cp950) rather than an array

[15:26:00.0000] 
so as far as I can tell the only encoding not compatible with ASCII that made it into the new standard is utf-16

[15:26:01.0000] 
utf-16 is truly unique and evil

[15:46:00.0000] 
hey anyone know if there's something that would cause