| 02:07 | <zewt> | http://lists.w3.org/Archives/Public/public-texttracks/2011Dec/0073.html gah, now gmail is apparently screwing up plaintext quote formatting, badly, when you send mails in the rich text editor |
| 03:52 | <Yuhong> | "hsivonen wonders if document.write was considered a simple feature in the Netscape 2 design phase" |
| 03:53 | <Yuhong> | I think it has a lot to do with tags as commands (which older browsers like Netscape 2 used) vs actually creating a parse tree (which later browsers like for example IE4 and later used). |
| 05:03 | <MikeSmith> | http://trac.webkit.org/changeset/103115 |
| 05:04 | <MikeSmith> | "Cache and reuse HTMLCollections exposed by Document." ... "This reduces memory consumption by ~800 kB (on 64-bit) when loading |
| 05:04 | <MikeSmith> | the full HTML5 spec." |
| 05:17 | <zewt> | that seems somewhat minor, for that case, heh |
| 05:20 | <roc> | We've been doing that for a long time, I'm surprised they weren't |
| 06:45 | <MikeSmith> | http://msujaws.wordpress.com/2011/12/16/more-enhancements-for-html5-video-and-audio-in-firefox/ |
| 06:45 | <MikeSmith> | "The loop attribute has been implemented" |
| 06:46 | <MikeSmith> | Features planned: "Add detection of H.264 videos and custom error UI" sounds interesting |
| 06:49 | <JonathanNeal> | hello |
| 06:53 | <MikeSmith> | JonathanNeal: hola |
| 06:54 | <JonathanNeal> | What's happenin'? |
| 06:54 | <JonathanNeal> | Everybody have a good week? |
| 07:05 | <MikeSmith> | Saturday afternoon here for me |
| 07:05 | <MikeSmith> | W3C published drafts of Audio APIs this week |
| 07:05 | <MikeSmith> | also, news this week that IE will start doing user auto-updates next year is pretty cool |
| 07:06 | <JonathanNeal> | Yea, I thought it was next month, must have been too happy to notice. |
| 07:06 | <MikeSmith> | and the RFC for the WebSockets protocol was finally published |
| 07:07 | <MikeSmith> | and news that responseType=json for XHR will ship Firefox 10 |
| 07:08 | <JonathanNeal> | woot. |
| 07:08 | <JonathanNeal> | responseType json allows for what? I can just google it I guess. |
| 07:09 | <MikeSmith> | http://dvcs.w3.org/hg/xhr/raw-file/tip/Overview.html#json-response-entity-body |
| 07:10 | <MikeSmith> | and in "dog bites man" non-news, the MPEG group has issued one of their periodic announcements suggesting they are on the verge of releasing some royalty-free MPEG codec |
| 07:11 | <MikeSmith> | that was a couple of weeks ago |
| 07:11 | <MikeSmith> | "It seems every few months a patent pool or some one manages to spin up wishful thinking that there will soon be a “new deal” that will answer the needs of royalty free communities, only to cleverly leave the hopeful twisting in the wind. " |
| 07:11 | <MikeSmith> | http://www.robglidden.com/2011/12/mpeg-plus-or-patent-pool-lite-mpeg-mulls-royalty-free-proposals/ |
| 07:12 | <JonathanNeal> | I'm sorry, I have trouble reading that description from the w3 |
| 07:13 | <JonathanNeal> | So, would you weigh in your thoughts on mpeg? Are you hoping webm takes off? |
| 07:35 | <MikeSmith> | oh my |
| 07:36 | <JonathanNeal> | oh your |
| 07:36 | <MikeSmith> | Y U not let us make own web browser?? |
| 07:37 | <JonathanNeal> | MikeSmith: I'm sorry for stopping you. Stopping redacted. |
| 07:37 | <MikeSmith> | heh |
| 07:39 | <JonathanNeal> | One of my co-workers is writing a javascript library that he hopes will make writing webapps more like writing iphone apps. His previous job was writing iPhone apps. I saw the code. It looked mighty scary. |
| 07:39 | <MikeSmith> | seems like there's a few people been doing that for a while now.. |
| 07:40 | <JonathanNeal> | I can't say if it was good or bad, but there was a lot of prototyping going on, and it looked awfully complicated. |
| 07:40 | <MikeSmith> | but I'm sure there are better ways to do it |
| 07:40 | <MikeSmith> | (better than what we already have, I mean) |
| 07:42 | <JonathanNeal> | For sure, I've been obsessing over a better events delegator all evening. |
| 07:45 | <JonathanNeal> | i want an easy way to listen for an event with multiple parts, like mousedown+mousemove, so it only fires on a mousemove after a mousedown. |
| 07:46 | <JonathanNeal> | i want an easy way to delegate by a selector contained by the currentTarget or containing the currentTarget. |
| 07:49 | <JonathanNeal> | Well, I'm glad to hear the internet is doing well. From my court, the new html5shim is being tested and will be released very soon. I hope it convinces more people to use html5 which ultimately rid us of ie<9. |
| 11:41 | <annevk> | Why is publishing stuff like http://allthis.com/u10838 okay and why is there no clear opt out? |
| 11:41 | <annevk> | Nobody needs to buy my time, I'm on IRC |
| 12:12 | <annevk> | http://mxr.mozilla.org/mozilla/source/intl/uconv/src/charsetalias.properties#515 :) |
| 12:28 | <annevk> | I guess nobody makes errors in shift_jis |
| 12:29 | <annevk> | 3C 6D 65 74 61 20 63 68 61 72 73 65 74 3D 73 68 69 66 74 5F 6A 69 73 3E E0 3C 74 65 73 74 3E gives different results all over |
| 12:29 | <annevk> | (they key being the E0 byte) |
| 12:29 | <annevk> | in Gecko you get a replacement character, Chrome displays nothing, Opera gives �test> |
| 12:30 | <annevk> | in Gecko you also get the <test> element in the DOM |
| 12:31 | <annevk> | if however the E0 is followed by an EOF, all browsers display nothing |
| 12:33 | <annevk> | actually, Chrome inserts an empty Text node followed by a an element node <test> (like Gecko) |
| 12:33 | <annevk> | so I guess they are a little closer to each other than Opera is to either |
| 12:37 | <annevk> | does that mean that in Opera shift_jis is not actually ASCII-compatible? |
| 12:40 | <annevk> | btw, for those interested in ECMAScript versioning: https://lists.webkit.org/pipermail/webkit-dev/2011-December/018924.html |
| 12:54 | <Philip`> | So you get to write either |
| 12:54 | <Philip`> | "use strict"; |
| 12:54 | <Philip`> | or |
| 12:54 | <Philip`> | use version 6; |
| 12:55 | <Philip`> | Isn't that going to be a little confusing? |
| 12:56 | <annevk> | trying out random shift_jis stuff is fun... |
| 12:56 | <bga> | es3 was good |
| 12:56 | <annevk> | 81 B0 undefined per 932, gives ・in Gecko |
| 12:57 | <annevk> | replacement character in Chrome/Presto |
| 12:58 | Philip` | wonders if there has been an attempt to enumerate every 2-byte string and see what character comes out |
| 12:59 | <annevk> | there will have been one soonish |
| 13:02 | <annevk> | it does seem like putting multi-octet encodings directly in the specification is going to be a bad idea |
| 13:02 | <annevk> | as for shift_jis that's a little over 16k potential code points |
| 13:03 | <annevk> | (~64 * 256) |
| 13:04 | <annevk> | heh, even simple things like 80 |
| 13:05 | <annevk> | U+FFFD in Opera, U+0080 in Gecko, nothing in Chrome |
| 13:05 | <Philip`> | Do implementations just have giant byte-sequence tables, or do they do the arithmetic instead? |
| 13:05 | <annevk> | you need a table to map it to Unicode |
| 13:05 | <annevk> | see http://msdn.microsoft.com/en-us/goglobal/cc305152.aspx for a pretty nice overview |
| 13:06 | <annevk> | with math you can go from shift_jis to some other Japanese standard, but then you still don't have code points (but you could do that to reduce memory footprint I guess) |
| 13:28 | <annevk> | Ms2ger: fixing xref |
| 13:30 | <annevk> | fixed |
| 15:08 | <annevk> | oh yes |
| 15:09 | <annevk> | inbox < 500 |
| 15:09 | <Ms2ger> | Wow :) |
| 15:12 | <Ms2ger> | Inbox < 70 |
| 15:17 | <gsnedders> | annevk: Can you make your encoding testsuite output codepoints in base 16? seing them in base 10 is weird. |
| 15:18 | <annevk> | what's the easiest way to do that? |
| 15:18 | <annevk> | I can figure it out, but if you know |
| 15:19 | <Ms2ger> | Number.prototype.toString ( [ radix ] )? |
| 15:21 | <gsnedders> | yeah, exactly |
| 15:23 | <MikeSmith> | annevk: I'm inclined to just turn on the notifications to www-dom now and ask for forgiveness later |
| 15:23 | <MikeSmith> | if anybody actually complains |
| 15:24 | <Ms2ger> | So, did we make WebIDL forbid overloading methods in subclasses? |
| 15:25 | <annevk> | MikeSmith: go ahead |
| 15:26 | <annevk> | MikeSmith: we'll find out quickly enough if it's problematic |
| 15:26 | <MikeSmith> | yup |
| 15:26 | <annevk> | and www-dom is relatively low traffic |
| 15:26 | <MikeSmith> | sou da na |
| 15:29 | <annevk> | Ms2ger: gsnedders, thanks and done |
| 15:29 | <Ms2ger> | Np |
| 15:33 | <annevk> | also put the source code in https://bitbucket.org/annevk/webencodings in case anyone wants to use it |
| 15:34 | <annevk> | it seems Opera treats unrecognized as only bytes 00-7F are okay |
| 15:35 | <annevk> | not sure that's a good fallback |
| 15:49 | <zewt> | Ms2ger: isn't that used? (HTMLAllCollection, etc) |
| 15:49 | <zewt> | annevk: sorry, parse error on that last |
| 15:51 | <annevk> | oops |
| 15:52 | <annevk> | with unrecognized labels only 00-7F work, the rest gets FFFD'd |
| 15:52 | <annevk> | rather than falling back to say windows-1252 |
| 15:52 | <zewt> | so unrecognized = true ASCII |
| 15:52 | <zewt> | would be nice if everyone did that to begin with, but since they don't ... |
| 15:52 | <zewt> | though |
| 15:53 | <zewt> | everyone else i'm sure falls back on locale-dependencies, right? |
| 15:53 | <annevk> | "true ascii" |
| 15:53 | <annevk> | "ascii" is windows-1252 as it should be :) |
| 15:53 | <zewt> | yeah I hate that I have to distinguish between "no, really, ASCII" and other crap :P |
| 15:54 | <annevk> | should just talk about 00-7F and U+0000-U+007F rather than ASCII |
| 15:54 | <annevk> | ASCII is ambiguous |
| 15:55 | <zewt> | "The results for windows-1252 (and friends) is confusing because this is a common fallback encoding." in windows, at least, change your system language to something else to distinguish |
| 15:55 | <zewt> | also, opera uses heuristics for unspecified encodings, right? |
| 15:55 | <zewt> | (according to recent discussions on webvtt) |
| 15:56 | <zewt> | not sure what other weird variables there are surrounding fallbacks |
| 15:58 | <annevk> | I think everyone uses heuristics |
| 15:58 | <annevk> | but webvtt should be utf-8 only |
| 15:58 | <annevk> | are they changing that? |
| 15:58 | <zewt> | er wait |
| 15:58 | <zewt> | those discussions were about language detection |
| 15:59 | <zewt> | (another big mess, with a lot of the same problems, but separate) |
| 15:59 | <annevk> | that sounds like the unicode layer |
| 15:59 | <annevk> | someone should fix that |
| 15:59 | <Ms2ger> | I read that as "the unicorn layer" |
| 15:59 | <annevk> | that's above all, not quite there yet :) |
| 16:01 | <zewt> | (the question was that since Firefox always, apparently, defaults to Japanese for CJK text unless @lang or a language-specific encoding says otherwise--since it does that, could Opera remove its special heuristics and do the same) |
| 16:02 | <zewt> | (among other things) |
| 16:03 | <annevk> | "for CJK text" is defined? |
| 16:03 | <zewt> | for text using chinese characters, where you need to pick the right font for the actual language |
| 16:03 | <annevk> | yeah but does that mean one such code point? |
| 16:03 | <annevk> | and within how many code points does it need to occur from the start? |
| 16:04 | <zewt> | not sure what you mean |
| 16:04 | <zewt> | do you know about han unification? |
| 16:04 | <annevk> | yes |
| 16:04 | <zewt> | well that's why you need to pick the font--can you rephrase the question? |
| 16:04 | <annevk> | the algorithm for determining CJK text is not defined |
| 16:04 | <zewt> | right |
| 16:05 | <zewt> | it's opera-specific (I don't know if any other browsers have anything like it, I'm guessing not) |
| 16:05 | <annevk> | you just said |
| 16:05 | <annevk> | Firefox defaults to Japanese for CJK text |
| 16:05 | <zewt> | right |
| 16:05 | <annevk> | presumably it somehow knows it's dealing with CJK text |
| 16:05 | <annevk> | based on some algorithm |
| 16:05 | <zewt> | for text in codepoint ranges representing chinese characters |
| 16:05 | <EvanR> | if you are ok with proportional fonts pango will layout and pick fonts for you when rendering |
| 16:05 | <zewt> | (how *exactly* it defines that I don't know offhand) |
| 16:06 | <annevk> | I don't think that's defined |
| 16:06 | <EvanR> | which is what many browsers use |
| 16:06 | <zewt> | (might be "if the default font doesn't have it and we go looking for a fallback") |
| 16:06 | <zewt> | (or rather "the current font") |
| 16:06 | <annevk> | zewt: but if you first have say 1024 western code points and then your CJK text starts? |
| 16:07 | <annevk> | oh well |
| 16:07 | <annevk> | got to go |
| 16:07 | <zewt> | heh later |
| 16:07 | <zewt> | and no I don't know the fine details |
| 16:07 | <zewt> | (or all of the coarse ones, for that matter) |
| 17:10 | <bga> | http://groups.google.com/group/jsmentors/browse_thread/thread/390477bfc89f9058 |
| 18:09 | <zewt> | annevk5: do you know whether all of the aliases in this table actually really map to the same thing? |
| 18:11 | <zewt> | eg. x-user-defined and iso8859-1 seem to be separate items in IE9 (they may have the same table, but they show up separately in the menu) |
| 18:16 | <Philip`> | Maybe x-user-defined is only equivalent to iso8859-1 when the user defines it to be? |
| 18:16 | <zewt> | i define it to be unicorns |
| 18:23 | <Philip`> | (http://msdn.microsoft.com/en-us/library/aa752010(v=vs.85).aspx indicates x-user-defined is code page 50000, and http://blogs.msdn.com/b/shawnste/archive/2007/03/17/hacking-code-pages-or-how-to-totally-hose-your-machine-and-your-data.aspx indicates you can write a DLL to provide a custom implementation of code pages >50000 (which might really mean ">=50000")) |
| 22:23 | <WeirdAl> | annevk: spec suggestion for XHR2, the open method, section 4.7.1. Step 11 (the timeout check) comes after the URL resolving. I think it might be better to move that to before resolving the URL (step 5, 6). |
| 22:24 | <WeirdAl> | resolving the URL would be more expensive than checking for the timeout not equalling zero |
| 22:24 | <WeirdAl> | and we'd throw an exception anyway |
| 22:53 | <smaug____> | WeirdAl: does that matter? |
| 22:57 | <WeirdAl> | eh, I'd like to save a few operations :) |
| 23:01 | <zewt> | note that either failing is usually a programming error, so in practice both should always succeed--so i doubt it makes a difference |
| 23:02 | <zewt> | (on a quick look) |
| 23:04 | <smaug____> | yeah, doesn't make sense to optimize the usually non-failing cases |
| 23:06 | <zewt> | not in deployed apis, anyway |
| 23:09 | WeirdAl | shrugs and removes a snippy comment |
| 23:21 | <zewt> | uh, okay, heh |
| 23:45 | <WeirdAl> | dammit, it ate my commit message |