| 00:01 | <TabAtkins> | Sigh. You have to clean up or something? |
| 00:01 | <TabAtkins> | And the answer is "someone attempting to commit suicide via alcohol poisoning". |
| 00:02 | <gsnedders> | TabAtkins: Get him sober enough so that a taxi would let him in to get him home |
| 00:02 | <gsnedders> | TabAtkins: And now I've done that, I've run to my room, and put on headphones to cut it out. Everyone else can deal with this now. |
| 00:03 | <gsnedders> | Guy in the flat above literally walked away saying something like "I'm fucking tired of dealing with his drunk fucking ass all night". If I hadn't been in the same flat, and knowing what fallout it would cause, I would've done the same. |
| 00:04 | <gsnedders> | I'd been trying to get him to stop drinking for a while. I'd try to convince people that we shouldn't let him come out, as he'll just collapse (which he did) as well as getting refused entry to where we were going (which he did). |
| 00:04 | <gsnedders> | So I have even less sympathy than I normally do in such cases. |
| 00:06 | <gsnedders> | So yeah, that ball was kinda shit, seeming we just got there, waited for the second taxi load of people to arrive, and then five minutes after it did had him collapse onto paving (though thankfully didn't crack his head open or anything like that) and then had to help with that. |
| 00:06 | gsnedders | sighs |
| 00:06 | gsnedders | should just shut up |
| 00:06 | gsnedders | is way too pissed off to say anything rational |
| 00:07 | <TabAtkins> | Heh, agreed. Take shameful photographs? |
| 00:07 | <gsnedders> | No, that's just taking pleasure in someone else's misery. |
| 00:08 | <TabAtkins> | Not necessarily. The shame is intended for him, in a "dude, you are seriously fucking yourself up, just look" kind of way. |
| 00:09 | <gsnedders> | Heh. I did that once. I've never spoken to them since. |
| 00:09 | <gsnedders> | They rather don't like me for that. |
| 00:09 | <TabAtkins> | Welp, looks like you don't like him anyway. ^_^ |
| 00:09 | <gsnedders> | But oh well, I don't regret doing it. |
| 00:09 | <gsnedders> | TabAtkins: I don't really know this guy. I've been in the same flat as him since Monday pretty much. |
| 00:10 | <gsnedders> | And there's only two people in the flat (of 12, inc. me) who I really get on with. |
| 00:46 | <MikeSmith> | gsnedders: is your online outliner currently busted? |
| 00:46 | <MikeSmith> | I'm trying to use it but getting a 500 response |
| 01:08 | <gsnedders> | MikeSmith: It is known to be busted with file upload |
| 01:08 | <gsnedders> | It is unknown to be busted with URLs |
| 01:21 | <danj> | Does anyone know with mobile safari / webkit if it's possible to restrict audio playback to the ear speaker on an iphone when playing something through an audio element? |
| 01:22 | <TabAtkins> | If there is, it's proprietary. |
| 01:22 | <danj> | sure |
| 01:22 | <danj> | I should have generalised that more |
| 01:22 | <danj> | It shouldn't be device specific |
| 01:23 | <TabAtkins> | Then no. |
| 01:23 | <danj> | thanks |
| 02:10 | <MikeSmith> | it's great to read Sylvain's tweets, because he has wit and he doesn't mind going toe-to-toe with whoever wants to step into the ring |
| 02:11 | <MikeSmith> | e.g., http://twitter.com/sgalineau/status/24713191647 |
| 02:11 | <paul_irish> | <3 |
| 02:12 | <MikeSmith> | I wonder if Alex Mogilevsky uses twitter |
| 02:15 | <TabAtkins> | MikeSmith: I don't think he does, but he has a facebook. |
| 02:15 | <MikeSmith> | ok |
| 02:15 | <TabAtkins> | (If he does have a twitter, I'm not following him.) |
| 02:15 | <MikeSmith> | I guess I must eventually do the facebook thing |
| 02:16 | <paul_irish> | he did. but he quit http://twitter.com/#!/alexmog |
| 02:22 | <MikeSmith> | oh |
| 02:22 | <MikeSmith> | a sad loss for twitter |
| 02:25 | <paul_irish> | what is a usecase for MessagePorts? |
| 04:25 | <mpilgrim> | paul_irish: older browsers returned "no" instead of "" if they didn't think they could play a video |
| 04:35 | <paul_irish> | mpilgrim: ah okay. do you know which? |
| 04:37 | <paul_irish> | looks like FF 3.5.1 and below. |
| 06:57 | <kennyluck> | MikeSmith: 白癡 |
| 06:58 | <MikeSmith> | kennyluck: 谢谢 |
| 07:48 | <zcorpan_> | http://www.w3.org/News/2010#entry-8894 - xml-stylesheet PER now announced on w3.org |
| 07:50 | <MikeSmith> | zcorpan_: congrats |
| 07:50 | <MikeSmith> | seems like that will move along pretty quickly |
| 07:52 | <zcorpan_> | yeah i hope we're pretty much done |
| 07:53 | <MikeSmith> | you can be certain you're done until Bjorn has reviewed it |
| 07:53 | <MikeSmith> | I trust that Bjorn has already commented? |
| 07:54 | <zcorpan_> | nope he hasn't |
| 07:54 | <MikeSmith> | uh-oh |
| 07:55 | <MikeSmith> | the phrase "behooves you" comes time mind here |
| 07:55 | <zcorpan_> | DoC: http://www.w3.org/XML/2010/01/disposition |
| 07:55 | <MikeSmith> | eh? |
| 07:55 | MikeSmith | gets a blank page |
| 07:56 | <MikeSmith> | well, not blank |
| 07:56 | <MikeSmith> | just mostily blank |
| 07:56 | <MikeSmith> | hmm, works in Opera |
| 07:57 | <MikeSmith> | works in Minefield... |
| 07:57 | <zcorpan_> | ah. chrome gets the xml version |
| 07:59 | <zcorpan_> | content negotiation-- |
| 08:15 | <annevk> | WAI-ARIA is in Last Call again |
| 08:25 | <annevk> | ah, the message from yesterday was related to some moving of the list: "[admin] ietf-types is moving" |
| 09:19 | <MikeSmith> | kennyluck: Colloquy |
| 09:19 | <hsivonen> | annevk: is there a disposition of comments for ARIA somewhere? |
| 09:20 | <hsivonen> | is one required for previous LC comments when re-entering LC? |
| 09:21 | <MikeSmith> | hsivonen: I think it's required only if it's followed by a transition request to the Director |
| 09:21 | <hsivonen> | MikeSmith: ok |
| 09:22 | <hsivonen> | I wonder if the current state of the PF comment status tracking system is viewable from outside the PF (before it presumably gets dumped into a Disposition of Comments later) |
| 09:22 | <hsivonen> | "dumped" in the technical sense |
| 09:22 | <hsivonen> | not in the "disposing of" sense |
| 09:23 | <hsivonen> | hmm. do you dispose of a comment by putting it in the disposition of commments? |
| 09:24 | <zcorpan_> | does ie9 have an ie8 mode? |
| 09:24 | <annevk> | no, disposition of comments is an overview of all comments |
| 09:24 | <annevk> | including indication of whether the comment was accepted or rejected and whether any formal objections were raised |
| 09:24 | <hsivonen> | zcorpan_: it has two of them! |
| 09:24 | <hsivonen> | zcorpan_: IE8 Standards and IE8 Almost Standards |
| 09:25 | <zcorpan_> | sigh |
| 09:25 | <hsivonen> | zcorpan_: I intend to update /doctype/ with a new flowchart once I've tested everything |
| 09:25 | <annevk> | can't have enough modes |
| 09:26 | <zcorpan_> | so now there's ie5.5, ie7, ie8 almost standards, ie8 standards, ie9 almost standards, ie9 standards and ie9 xhtml? |
| 09:26 | <hsivonen> | the non-permanent Editor's Draft is http://hsivonen.iki.fi/doctype/ie9.html |
| 09:26 | <hsivonen> | zcorpan_: I think there are seven modes, yes, but I haven't yet verified that the XHTML mode is distinct in terms of DOM and CSS |
| 09:27 | <hsivonen> | though I expect it to be |
| 09:27 | <Rik`> | and no XP mode |
| 09:27 | <hsivonen> | also of interest is if text/xml and application/xhtml+xml are distinct |
| 09:27 | <hsivonen> | I mean they have to have a plan for all of that legacy intranet XSLT |
| 09:29 | <zcorpan_> | what's with the legacy intranet XSLT? |
| 09:29 | <hsivonen> | zcorpan_: previously, the output was text/html |
| 09:29 | <hsivonen> | zcorpan_: I don't know if it works the same now or if there's something different |
| 09:30 | <hsivonen> | if different, the hypothesis should be that there are more modes to find in there... |
| 09:30 | <hsivonen> | so many hypotheses to test |
| 09:39 | <zcorpan_> | i wonder if it's possible to convince microsoft to not have ie9 modes in ie10 |
| 09:40 | <roc> | don't try |
| 09:40 | <roc> | eventually the browser will collapse under its own weight |
| 09:41 | <zcorpan_> | yeah i guess it's going to bite themselves in the end |
| 09:45 | <annevk> | I'm tempted to start misusing X-UA-Compatible on sites |
| 09:46 | <roc> | imagine what it's like to do site QA for IE12 |
| 09:47 | <zcorpan_> | hsivonen's flowchart is going to be 10x bigger than it is now |
| 09:47 | <jgraham> | roc: You just made my brain explode |
| 09:50 | <zcorpan_> | <meta http-equiv="x-ua-compatible" content="IE=<?php emitRandomVersion(); ?>"> |
| 09:51 | <annevk> | I like how people put it in HTTP instead so it "validates" |
| 09:52 | <annevk> | You'd hope that at one point people realize that hiding it from the validator does not actually make it better |
| 09:52 | <annevk> | But so far history is against that |
| 09:53 | <zcorpan_> | did shelley's server melt? |
| 09:53 | <jgraham> | Henri should declare anything with a x-ua-compatible header invalid :) |
| 09:54 | <annevk> | all X- headers really |
| 09:54 | <jgraham> | Of course then people would just send different headers to the validator |
| 10:01 | <Rik`> | jgraham: I could still use conditionnal comments |
| 10:03 | <Rik`> | (if that works) |
| 10:09 | <hsivonen> | jgraham: yeah, the validator really shouldn't allow itself to be gamed with a HTTP header |
| 10:09 | <hsivonen> | Rik`: it doesn't! |
| 10:10 | <hsivonen> | Rik`: if there's a script of conditonal comment before the meta, it gets ignored |
| 10:10 | <hsivonen> | yeah, yeah, I've been lazy since I haven't actually written about conditional comments on /doctype/ |
| 11:50 | <zcorpan_> | http://www.teknikveckan.se/2010/09/periodiska-systemet-for-html-5/ |
| 11:56 | <hsivonen> | zcorpan_: cool, but why <div>s instead of <td>s? |
| 11:57 | <hsivonen> | (a bit uncool to have links to W3Schools, though) |
| 11:57 | <zcorpan_> | yeah |
| 11:58 | <jgraham> | I wonder if people percieve that W# schools is somehow endorsed by W3C |
| 11:58 | <jgraham> | *W3 |
| 11:59 | <zcorpan_> | some do, yes |
| 12:01 | <brucel> | loads do, imo |
| 12:02 | <brucel> | I posted that it was often wrong, and many ppl said "it can't be, it's from the W3C" (which is never wrong, presumably) |
| 12:05 | <hsivonen> | brucel: the W3C is so right that it's an authoritative source about itself for Wikipedia |
| 12:08 | <zcorpan_> | W3C is like god |
| 12:09 | <zcorpan_> | just can't be wrong |
| 12:18 | <zcorpan_> | what's the plan with websockets in firefox 4? |
| 12:19 | <hsivonen> | smaug____: ^ |
| 12:23 | <jgraham> | zcorpan_: You saw the message from Patrick McManus on hybi? |
| 12:23 | <jgraham> | """For those that aren't aware, I'm working with the Mozilla Websockets |
| 12:23 | <jgraham> | team. And with that hat on I can say FF is unlikely to deploy -76 in FF |
| 12:23 | <jgraham> | 4 - except perhaps under a clearly for-experiment-only namespace which |
| 12:23 | <jgraham> | is subject to breakage at any time. |
| 12:23 | <jgraham> | """ |
| 12:23 | <zcorpan_> | ah, thanks |
| 12:24 | <jgraham> | Would be nice to get confirmation of that though |
| 12:24 | <jgraham> | Does it still work with -76 in the current nightlies? |
| 12:26 | <zcorpan_> | yes |
| 12:28 | <jgraham> | Hmm |
| 12:37 | <Workshiva> | jgraham: Are they more likely to implement -75 or -77? |
| 12:37 | smaug____ | missed some messages |
| 12:38 | <smaug____> | what is -77? |
| 12:38 | <smaug____> | there was -76 |
| 12:38 | <zcorpan_> | there's no -77 |
| 12:38 | <smaug____> | and then ietf-0 |
| 12:40 | <smaug____> | FF4 betas have -76 |
| 12:40 | <jgraham> | Workshiva: -77 is -01 |
| 12:40 | <jgraham> | Due to amazing IETF numbering |
| 12:40 | <smaug____> | after that the protocol has been way too unstable to implement anything |
| 12:42 | <zcorpan_> | smaug____: so what's the plan for firefox 4? |
| 12:42 | <smaug____> | I'm not quite sure about the plan |
| 12:42 | <zcorpan_> | ok |
| 12:42 | <smaug____> | *if* the protocol becomes more stable, we should probably update the implementation |
| 12:43 | <jgraham> | On an unrelated note, has minefield lost the ability for the user to override the page character encoding? |
| 12:43 | <smaug____> | otherwise I hope we just keep the API, and warn web app developers about using unstable/old protocol |
| 12:43 | <smaug____> | no |
| 12:44 | <jgraham> | Oh, where has it moved to? |
| 12:44 | <smaug____> | At least on linux the UI is still there |
| 12:44 | <jgraham> | Right |
| 12:44 | <jgraham> | We couldn't find it on Windows |
| 12:44 | <smaug____> | uh |
| 12:46 | <smaug____> | jgraham: it should be in the "developer" menu |
| 12:47 | <smaug____> | (that menu name will change) |
| 12:47 | <jgraham> | Oh |
| 12:47 | <jgraham> | Maybe this build was just missing it |
| 12:48 | <jgraham> | Not sure how old it was |
| 12:52 | <hsivonen> | jgraham: I can't see it either when the menu bar is hidden |
| 12:52 | <hsivonen> | I also can't figure out how to get the menu bar back once it is hidden |
| 12:53 | <hsivonen> | jgraham: to get the menu bar back, right click in the space between the forward button and whatever is to the right of the forward button |
| 12:58 | <hsivonen> | IE9 puts Encoding in the page context menu in the menubarless world |
| 12:58 | <jgraham> | Seriously? In the context menu? |
| 12:58 | <hsivonen> | jgraham: yes |
| 12:58 | <jgraham> | Wow |
| 12:58 | <jgraham> | (it turns out that the verion of Windows Minefield we were looking at was like a month out of date) |
| 12:59 | <jgraham> | I guess it was too optimistic to hope that character encoding issues have gone away :) |
| 13:02 | <hsivonen> | IE could make the Compat Mode button magic so that when you press it, it tries the new combination of mode and encoding :-) |
| 13:02 | <hsivonen> | that way, the Encoding context menu submenu wouldn't be necessary :-) |
| 13:17 | <annevk> | does any browser have stats on how often users use the encoding menu? |
| 13:18 | <annevk> | I've been thinking it should just be killed altogether, but maybe that fails |
| 13:18 | <jgraham> | annevk: there is https://testpilot.mozillalabs.com/testcases/menu-item-usage/aggregated-data.html |
| 13:19 | <jgraham> | but you might want to worry about selection bias |
| 13:21 | <annevk> | why are there no graphs? |
| 13:21 | <annevk> | would have been easy to make graphs, no? |
| 13:21 | <jgraham> | there are on a blog somewhere |
| 13:22 | <jgraham> | http://blog.mozilla.com/metrics/2010/04/14/menu-item-usage-study-an-update-to-the-initial-analysis/ |
| 13:27 | <jgraham> | In particular the selection bias I would worry about would be over-representation of western demographics where character encoding issues might be less prevalent |
| 14:33 | <annevk> | hmm |
| 14:33 | <annevk> | thinking about it some more throwing for starting with "xmlns" on setAttribute seems problematic |
| 14:37 | <annevk> | I suppose in XML you do not want <html xmlns="http://www.w3.org/1999/xhtml">; to be overwritten by a setAttribute call |
| 14:37 | <annevk> | hmm |
| 14:40 | <annevk> | mah, lets define it and see who implements it; leaving it as open issue doesn't really solve anything either |
| 14:41 | <jgraham> | Argh |
| 14:41 | <jgraham> | Why did plh blog the results table from the HTML5 testsuite |
| 14:41 | <jgraham> | I know he says not to pay attention to it |
| 14:42 | <jgraham> | But seriously, it can only be harnful right now |
| 14:42 | <jgraham> | *harmful |
| 14:43 | <annevk> | I'm still not sure how seriously to take the HTML testing effort |
| 14:43 | <annevk> | well, some tests are pretty good, but a lot of them... |
| 14:45 | <jgraham> | It is a serious effort |
| 14:45 | <jgraham> | But the volume and quality so far is not great |
| 14:45 | <jgraham> | We will get there |
| 14:46 | <bl4ckcomb> | evil Xpost: what would semantically be better: putting a 'post a comment' form in the footer of an article or putting it in a separate section outside the article? |
| 14:46 | <Philip`> | Is the test runner still entirely manual? |
| 14:47 | Philip` | guesses someone should automate it before approving a lot more tests |
| 14:47 | <zcorpan_> | bl4ckcomb: do you think of it as a footer or as a separate section? |
| 14:47 | <jgraham> | The test runner needs work |
| 14:47 | <jgraham> | Kris said he would do something, but I don't know exactly what his plans are |
| 14:47 | <bl4ckcomb> | zcorpan_, I still think it is relevant to the article, but it also isn't really footer stuff |
| 14:48 | <zcorpan_> | bl4ckcomb: how about a nested section? |
| 14:48 | <bl4ckcomb> | zcorpan_, you suggest a section in an article? |
| 14:48 | <zcorpan_> | yeah, why not |
| 14:48 | <bl4ckcomb> | ok |
| 14:49 | <bl4ckcomb> | I'm not saying its wrong :) thx for your help |
| 14:53 | <zcorpan_> | http://www.wired.com/threatlevel/2010/09/html5-safari-exploit/ |
| 15:01 | <zcorpan_> | hmm. http://www.html5rocks.com/tutorials/workers/basics/ advocates using postMessage() without any argument |
| 15:02 | <zcorpan_> | but the argument is required and it's not defined in webidl if it should throw or not |
| 15:02 | <zcorpan_> | (it throws in opera) |
| 15:08 | <zcorpan_> | wonder why chrome doesn't run workers from file:// |
| 15:10 | <gsnedders> | zcorpan_: For the same reason XHR doesn't work on file://, it considers it xorigin |
| 15:11 | <zcorpan_> | does opera also block xhr on file://? |
| 15:11 | <gsnedders> | No |
| 15:12 | <jgraham> | zcorpan_: The chrome people decided that all file:// access was x-origin iirc |
| 15:14 | <zcorpan_> | in order to pro-actively address security problems before they are discovered? |
| 15:15 | <gsnedders> | zcorpan_: Too many things have access to each origin on file |
| 15:20 | <AryehGregor> | So things like FFTs, Gaussian blurs, and AES are meant to be more real-world than what SunSpider or V8 test? |
| 15:20 | <AryehGregor> | Really? |
| 15:20 | <AryehGregor> | Those must be extremely non-real-world, then. |
| 15:21 | <AryehGregor> | Why doesn't someone just browse for a few minutes in Gmail or whatever, capture all the calls to browser functions, replace them with workalikes that will always return the same answer to avoid going down different code paths, and package that up as a test? |
| 15:22 | <AryehGregor> | Because I guess different browsers go down different code paths all over the place in an actual real-world app like Gmail due to sniffing. |
| 15:22 | <gsnedders> | AryehGregor: Because Gmail gives massively different JS to different browsers. |
| 15:22 | <AryehGregor> | And the same is probably true for every other real app. |
| 15:22 | <AryehGregor> | Sigh. |
| 15:22 | <gsnedders> | It's done on the server-side, so you only ever see your own JS. |
| 15:23 | <gsnedders> | (We have a bunch of non-standard Mozilla stuff in Opera to support some Google stuff where they only have Gecko/WebKit codepaths) |
| 15:23 | <AryehGregor> | Sigh. |
| 15:24 | <annevk> | not so much, I think |
| 15:25 | <annevk> | the utterly useless createContextualFragment (or whatever it was called) and something else that is pretty close to useless on Range() |
| 15:25 | <Philip`> | AryehGregor: Is the JS execution on Gmail (once you exclude all the DOM calls etc) a bottleneck? |
| 15:25 | <Philip`> | I'd have thought it wouldn't be doing much computation that would take any time |
| 15:25 | <AryehGregor> | Probably not. |
| 15:25 | <AryehGregor> | That's the point, isn't it? |
| 15:25 | <jgraham> | Wave caused some problems due to highly browser-specific code |
| 15:25 | <AryehGregor> | You want whole-browser benchmarks. |
| 15:25 | <AryehGregor> | jgraham, good thing no one used it. |
| 15:26 | <jgraham> | AryehGregor: It depends what you want to measure |
| 15:26 | <Philip`> | AryehGregor: Depends who "You" is |
| 15:26 | <gsnedders> | annevk: HTMLUnknownElement was origninally added for some Gecko codepath, Error.stack is in the same undocumented format as Gecko because GWT stuff relies upon the format(!)… |
| 15:26 | <Philip`> | People whose job is working on JS engines presumably want to make their JS engines really fast |
| 15:26 | <AryehGregor> | Well, whoever you are, you should want whole-browser benchmarks. Maybe you also want subsystem-specific benchmarks, but that doesn't remove the need for whole-browser benchmarks. |
| 15:26 | <jgraham> | AryehGregor: A benchmark on gmail isn't going to tell you how fast you are at rendering some js-limited game |
| 15:27 | <jgraham> | (Microsoft had a whole paper where they said basically what you just said) |
| 15:27 | <jgraham> | (and it ignored the same issue) |
| 15:27 | <AryehGregor> | No, but if you focus JS-only benchmarks then you'll optimize it way more than necessary. |
| 15:27 | <AryehGregor> | You could have a bunch of whole-browser benchmarks. Have some for games too |
| 15:27 | <AryehGregor> | . |
| 15:28 | <jgraham> | AryehGregor: http://benfirshman.com/projects/jsnes/ is pretty much javascript limited |
| 15:28 | <Philip`> | People will always write slower JS games - you'll never have optimised it enough |
| 15:28 | <gsnedders> | AryehGregor: Kraken does actually show up some interesting limitations of current non-Tracing engines |
| 15:28 | <annevk> | gsnedders, interesting, is that second one on Web ECMAScript? |
| 15:28 | <AryehGregor> | jgraham, that looks illegal. |
| 15:28 | <jgraham> | AryehGregor: (and has been used when assessing the performance of js engines) |
| 15:29 | <gsnedders> | AryehGregor: (namely, calling something like Math.abs in a tight loop is expensive doing a global lookup followed by a property lookup everytime) |
| 15:29 | <jgraham> | AryehGregor: Maybe. That's not really the point here though :) |
| 15:29 | <AryehGregor> | Wow, that runs horribly on Firefox 4b6. |
| 15:30 | <AryehGregor> | But really smoothly on Chrome dev. |
| 15:31 | <Philip`> | Once JSNES runs fast enough, someone needs to write JSPlayStation |
| 15:31 | <Philip`> | Actually they should call it JStation |
| 15:33 | <Workshiva> | And then JS2 |
| 15:33 | <AryehGregor> | Has anyone ported Super Mario RPG yet? I would totally want to play that. |
| 15:33 | <AryehGregor> | I think I even still own a legal copy somewhere. |
| 15:34 | <Workshiva> | I know someone was writing a gameboy emulator in JS too |
| 15:38 | <annevk> | allowing setting of xmlns attributes through the DOM at all seems kind of wrong |
| 15:39 | <AryehGregor> | I guess writing an XBox emulator would be easy, just write an x86 emulator. When you can write an efficient x86 emulator in JavaScript, I'll be impressed. |
| 15:39 | <AryehGregor> | (alternatively, cheat and use NaCl?) |
| 15:41 | <Philip`> | AryehGregor: And an NVIDIA GPU emulator |
| 15:41 | <AryehGregor> | Yes, that too. |
| 15:41 | <Philip`> | and a DirectX API emulator |
| 15:41 | <AryehGregor> | Well, that's just a software reimplementation, not an emulator. |
| 15:41 | <Philip`> | and probably a lot of the rest of the Win32 API |
| 15:42 | <Philip`> | Just port Wine to JS, I guess |
| 15:42 | <AryehGregor> | Well, if you have an x86 emulator already, you can just run |
| 15:42 | <AryehGregor> | Well, if you have an x86 emulator already, you can just run Wine on top of the emulator. |
| 15:42 | AryehGregor | waits for that to become fast. |
| 15:42 | <Philip`> | True |
| 15:42 | <Philip`> | though you'd have to run Linux on your emulator before running Wine on top of that |
| 15:43 | <Philip`> | Rather than emulating x86, you could emulate a much simpler instruction set and then run something like Qemu on that |
| 15:43 | <Philip`> | which'd save a lot of pain |
| 15:43 | <AryehGregor> | I guess you'd need to emulate kernel-mode x86 too, yeah. |
| 15:44 | <AryehGregor> | (I'm putting my money on NaCl being more practical here.) |
| 15:44 | <AryehGregor> | Although even that requires a recompile, right? |
| 15:44 | <AryehGregor> | Feh. |
| 15:44 | <gsnedders> | DirectX converted to WebGL? |
| 15:44 | <AryehGregor> | Well, you could still run Wine natively in NaCl, and then run an x86 emulator natively in NaCl . . . |
| 15:45 | <AryehGregor> | gsnedders, you can convert parts of Direct3D to WebGL, but not the rest of DirectX. |
| 15:45 | <AryehGregor> | Like DirectSound, and whatever else is part of it. |
| 15:45 | <gsnedders> | AryehGregor: Wlel, yeah |
| 15:45 | <Philip`> | As far as I'm aware, console games often work by sending raw data streams to the GPU and skipping the DirectX/OpenGL/etc layer |
| 15:45 | <gsnedders> | *Well |
| 15:45 | <gsnedders> | Oh. That makes it way more fun. |
| 15:45 | <Philip`> | so you need to emulate the lower level too |
| 15:45 | <AryehGregor> | Philip`, kind of like how games are often assembly-optimized? |
| 15:46 | <Philip`> | AryehGregor: I guess so |
| 15:46 | <Philip`> | i.e. only in the parts where it's really a bottleneck |
| 15:46 | <AryehGregor> | . . . okay, I think we can conclude at this point that we should look forward to JStation before JBox360. |
| 15:47 | <Philip`> | What about a Jii? |
| 15:48 | <AryehGregor> | J64 first, surely. |
| 15:55 | AryehGregor | begins reading the IE9 beta announcement carefully |
| 15:55 | <AryehGregor> | "Before IE9, browsers used perhaps 10% of the PC’s capability." Is that like how people only use 15% of their brains? |
| 15:55 | <AryehGregor> | (i.e., completely false?) |
| 15:58 | <AryehGregor> | I like the use of totally incongruous statistics. "About 40% of Windows 7 users have used Aero Snap. (Only 15% of users have control-click’d a link in a page.)" That tells us . . . what? |
| 15:58 | <annevk> | yay for Text nodes having a baseURI |
| 15:59 | <AryehGregor> | What's yay about that, either really or sarcastically? |
| 15:59 | <AryehGregor> | All nodes have them, don't they? |
| 15:59 | <annevk> | sarcastically |
| 16:00 | <AryehGregor> | So why is it bad? |
| 16:00 | <annevk> | bloat? |
| 16:00 | <AryehGregor> | Is it really more bloated than the rest of the web platform? :) |
| 16:00 | <annevk> | xml:base is of course very bad in general |
| 16:01 | <annevk> | AryehGregor, euh, with that line of reasoning you can add every feature you want |
| 16:01 | AryehGregor | notes that the blog post touts IE9's new tab page, which is a carbon copy of the functionality introduced by Opera (?) like five years ago, and copied by Chrome and Safari |
| 16:02 | <AryehGregor> | annevk, are we talking about adding new features, or speccing some stupid feature that no one really wants but some sites depend on it so we have to keep it? |
| 16:02 | <AryehGregor> | I was assuming the latter, given the usual tenor of this channel. |
| 16:02 | <Philip`> | AryehGregor: You could probably get a figure like 10% by computing utilisation of gigaflops from CPU+GPU, or something |
| 16:02 | <AryehGregor> | Philip`, it's probably more like 0%, since most of the time you're browsing, even the CPU is idle. |
| 16:02 | <annevk> | dunno if sites rely on baseURI |
| 16:03 | <Philip`> | AryehGregor: I assume they mean maximum usage of the PC's capability, not average |
| 16:03 | <AryehGregor> | I suggest that we make browsers use more of the PC's capability by replacing all blocking system calls with polling in a busy loop. |
| 16:03 | <AryehGregor> | Philip`, then it still uses very little of the memory or disk. |
| 16:03 | <AryehGregor> | I mean, it's just a really stupid way to pose the question. |
| 16:04 | <annevk> | hsivonen, so appendChild(script) acts as if async was set? |
| 16:04 | <Philip`> | Hmm, I'd be surprised if a browser only used 10% of the available RAM |
| 16:04 | <annevk> | hsivonen, if it has src= |
| 16:04 | <Philip`> | (even though I've got 4GB) |
| 16:04 | <AryehGregor> | I hope it uses a lot less than 10% of available disk in most cases. |
| 16:04 | <hsivonen> | annevk: if it has src, right |
| 16:05 | <hsivonen> | annevk: and runs sync if no src but child text |
| 16:05 | <annevk> | interesting |
| 16:05 | <Philip`> | AryehGregor: Yeah |
| 16:05 | <hsivonen> | annevk: now Presto is the only one left that doesn't do this like HTML5 says |
| 16:05 | <Philip`> | AryehGregor: Anyway, I think I generally agree it's a completely meaningless statistic |
| 16:05 | <annevk> | aah |
| 16:06 | <hsivonen> | interestingly enough, the HTML5 parser broke a couple of sites when this fix wasn't also present |
| 16:07 | <hsivonen> | I also made document.written inline scripts never block the parser |
| 16:07 | <hsivonen> | regardless of style sheet loads |
| 16:07 | <hsivonen> | but that didn't fit in the tweet |
| 16:08 | <jgraham> | Doesn't spotify use 10% disk space by default? |
| 16:16 | <AryehGregor> | People often get large hard disks specifically to store huge amounts of music or video, so that's reasonable. |
| 16:16 | <AryehGregor> | Browsers, less so. |
| 16:17 | <AryehGregor> | I have 1.5 TB of usable disk space, and I'd be kind of annoyed if my browser started using 150 GB. |
| 16:17 | <AryehGregor> | (actually it can only see 110G, due to partitioning, but that's beside the point) |
| 16:18 | <Philip`> | What could it use that space for? |
| 16:19 | <jgraham> | It's not obvious to me that a browser cache is a less valuble use of disk space than a spotify cache |
| 16:19 | <jgraham> | Given that the goal would be roughly the same |
| 16:19 | <jgraham> | i.e. to make content that you frequently access faster |
| 16:20 | <jgraham> | s/fater/appear faster/ |
| 16:20 | <jgraham> | and given that browsers are increasingly beiong used as media players |
| 16:20 | <AryehGregor> | Well, browser caches are way too small, for sure. |
| 16:21 | <AryehGregor> | They should be at least a few gigabytes. I doubt they'd be able to effectively use 100 GB, though. |
| 16:21 | <AryehGregor> | 10 GB, sure. |
| 16:21 | <jgraham> | Well I doube spotify can either |
| 16:22 | <jgraham> | Oh it is 10% of free disk space by default |
| 16:22 | AryehGregor | doesn't really know what Spotify is, so might have misunderstood the comparison |
| 16:22 | <jgraham> | which is somewhat different |
| 16:40 | <AryehGregor> | Do HTMLWG members get commit access to dvcs.w3.org somehow? Are there instructions somewhere on how to use thaT? |
| 16:40 | <AryehGregor> | that? |
| 16:47 | <jgraham> | Yes, you should just be able to use your w3c credentials |
| 16:48 | <jgraham> | http://www.w3.org/html/wg/wiki/Testing/Submission/ |
| 16:48 | <jgraham> | AryehGregor: ^ |
| 16:49 | <AryehGregor> | Oh, I missed the part where it said how to do it. |
| 16:49 | <AryehGregor> | I guess I should just use that instead of http://aryeh.name/gitweb.cgi?p=tests. |
| 16:49 | <AryehGregor> | I suppose it's not worth bikeshedding about git vs. hg. |
| 16:49 | <jgraham> | No |
| 16:49 | <AryehGregor> | (although git is better!) |
| 16:49 | <AryehGregor> | No? |
| 16:49 | <jgraham> | It's not worth bikeshedding |
| 16:49 | <AryehGregor> | Oh, okay then. |
| 16:50 | <jgraham> | Yes, you can just use that |
| 16:50 | <jgraham> | Although people might start trying to approve your tests or something |
| 16:50 | <AryehGregor> | That might be inconvenient. |
| 16:50 | <jgraham> | Did you see testharness.js in resources? |
| 16:50 | <AryehGregor> | Didn't look at it. |
| 16:50 | <jgraham> | AryehGregor: Well it is easy to stop them |
| 16:50 | <jgraham> | We seem to be using that for javascript tests |
| 16:52 | <AryehGregor> | I think the output format would be a little awkward, given that I've written >16,000 tests so far, >90% will pass on any remotely sane browser (i.e., not IE9), and failures tend to be extremely repetitive. |
| 16:53 | <jgraham> | AryehGregor: That isn't really a problem |
| 16:53 | <AryehGregor> | Although on the plus side, maybe the other framework won't crash on WebKit nightlies. |
| 16:53 | <AryehGregor> | No one has touched https://bugs.webkit.org/show_bug.cgi?id=45713 since I filed it four days ago. |
| 16:53 | <AryehGregor> | That never happens to me with Mozilla bugs. Sigh. |
| 16:54 | <AryehGregor> | jgraham, why isn't it really a problem? |
| 16:55 | <jgraham> | AryehGregor: Well, why would it be? |
| 16:55 | <AryehGregor> | Because it will have 16,000 output rows, instead of a couple hundred? |
| 16:55 | <jgraham> | If you adapt the tests you can more or less choose how fine-grained to make it |
| 16:56 | <jgraham> | according to how much becomes an assert() vs how much becomes a seperate test |
| 16:56 | <AryehGregor> | Hmm, so I could have one test with a lot of asserts, and it would just say which asserts failed? |
| 16:56 | <AryehGregor> | That sounds tenable. |
| 16:57 | <jgraham> | Well no it would fail at the first assert |
| 16:57 | <AryehGregor> | That seems undesirable. |
| 16:57 | <AryehGregor> | Well, I'll work with my own reporting framework for now. |
| 16:57 | <jgraham> | Well then you have 16,000 seperate tests and we just have to deal |
| 16:57 | <AryehGregor> | Which only reports failures. |
| 16:58 | <AryehGregor> | Actually, I've been thinking maybe I'll just remove the number and percentage of passed tests. It's somewhat meaningless anyway. |
| 16:58 | <jgraham> | It won't write anything to screen if you don't include an <div id="log"></div> |
| 16:58 | <AryehGregor> | The list of failures is the only important thing. |
| 16:59 | <jgraham> | AryehGregor: We can adapt the framework to optionally only report failures or something |
| 16:59 | <AryehGregor> | Yeah, in the medium term it's obviously right to use the same framework for everything. |
| 16:59 | <AryehGregor> | But for now I think it's easier for me to stick to my own, if only until I finish actually writing all the tests. |
| 17:06 | <hsivonen> | what's the address of zcorpan's single-pixel GIF site? |
| 17:07 | <hsivonen> | found it |
| 17:07 | <hsivonen> | http://spacergif.mobi/ |
| 17:09 | <hsivonen> | Philip`: http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/tag/spacer says Can't call method "selectrow_array" on an undefined value at /var/www/canvex/survey/2007-07-17/analyse.cgi line 293. |
| 17:10 | <Philip`> | hsivonen: I moved it to a new server and never set up the database for that |
| 17:12 | <AryehGregor> | Should I copy just the latest version of my tests into the W3C DVCS, or the whole history? |
| 17:16 | <Philip`> | AryehGregor: Will anybody ever care about the history? |
| 17:16 | <AryehGregor> | No. |
| 17:16 | <Philip`> | Seems like unnecessary effort and bandwidth, then |
| 17:16 | <AryehGregor> | However, it's fun to think about doing crazy stuff like that when you're using a DVCS. |
| 17:16 | <AryehGregor> | :) |
| 17:19 | <hsivonen> | Philip`: should I avoid linking to that URL? |
| 17:21 | <Philip`> | hsivonen: It hasn't worked for many months, and currently I don't care enough to make it work (since it's ancient and a tiny sample and doesn't seem very useful now), so it's probably best not to use it |
| 17:21 | <hsivonen> | Philip`: OK |
| 17:22 | <Philip`> | (It relied on Postgres, since that was less slow than SQLite, but that's not running on the new server) |
| 17:28 | <hsivonen> | blogged without the link: http://hsivonen.iki.fi/spacer/ |
| 17:31 | <miketaylr> | hsivonen: "Where spacer is still in use, it’s rarely makes any layout difference." s/it's/it/ |
| 17:31 | <hsivonen> | miketaylr: fixed. thanks |
| 17:32 | <miketaylr> | it's a fun article. :) |
| 17:36 | <Philip`> | hsivonen: "The spacer element made of the page wider than necessary" - s/of// |
| 17:39 | <hsivonen> | Philip`: fixed. thanks |
| 17:43 | <TabAtkins> | That was... interesting. I'd never before experienced webkit's "we won't show the text until we have the font" 'feature'. |
| 17:43 | <TabAtkins> | Here I got to see headers appear first, then monospace, then plain text. |
| 17:45 | <TabAtkins> | "without comparing the page into browsers at the same time." s/into/in two/ |
| 17:45 | <jgraham> | TabAtkins: You could fuck with people by making key words appear delayed delaying the font load |
| 17:46 | <jgraham> | Of course this is easy to do in other ways |
| 17:46 | <TabAtkins> | hsivonen: ^^^ (my previous line, not jgraham's) |
| 17:47 | <hsivonen> | TabAtkins: thanks, fixed. |
| 17:47 | <hsivonen> | Dictation still isn't perfect :-( |
| 17:47 | <TabAtkins> | Oh, cool, that was dicatated? |
| 17:48 | <hsivonen> | TabAtkins: most of it, yeah |
| 17:48 | <miketaylr> | whoa cool |
| 18:10 | <AryehGregor> | Okay, so now my WebKit bug includes three lines of script that cause a reliable crash in WebKit nightlies. Let's see if anyone notices. |
| 18:11 | Dashiva | sells as zero-day |
| 18:11 | <AryehGregor> | At least I know what the bug is now, so I can work around it. |
| 18:11 | <TabAtkins> | You could replace their coffee with Folgers crystals at the same time. |
| 18:13 | <AryehGregor> | Drat, it still crashes. |
| 18:13 | <AryehGregor> | Oh, wait. |
| 18:13 | <AryehGregor> | Yeah, it still does. |
| 18:13 | <AryehGregor> | Dangit. |
| 18:14 | <AryehGregor> | No more removeAttribute()s in sight. Blech. |
| 18:15 | <AryehGregor> | No, it doesn't crash? Hrm? |
| 18:16 | <AryehGregor> | Oh, caching. |
| 18:17 | <AryehGregor> | Now it works. |
| 18:53 | <annevk> | ap, I didn't reply to the EventSource bug anymore as it seemed like a repetition of a debate we had before on encodings |
| 18:53 | <annevk> | ap, I guess your sense of pragmatic here just does not align with mine :) |
| 18:55 | <ap> | annevk: I know, I don't expect to convince anyone at this point. But I want folks working on WebKit know more about the tradeoffs here than blindly follow whatever a spec says at the moment |
| 18:56 | <annevk> | that seems fair |
| 18:57 | <annevk> | questioning specs has not been done often enough in the past, lets not repeat that mistake :) |
| 19:25 | <annevk> | hsivonen, you wanted me to test CR and \0 in responseText right? |
| 19:36 | <hsivonen> | annevk: yes |
| 19:38 | <annevk> | http://tc.labs.opera.com/apis/XMLHttpRequest/responsetext-decoding.htm |
| 19:38 | <annevk> | seems to work fine, even in Opera! |
| 19:38 | <annevk> | I would have guessed we would have zero byte issues |
| 19:38 | <annevk> | the test is the one labeled "%0D%0A%00%0D" |
| 19:43 | <annevk> | god time zone math |
| 19:47 | <annevk> | maybe baseURI should not be readonly... |
| 19:55 | <cying> | is it possible to wrap text around a rectangle in CSS3? |
| 19:56 | <annevk> | no |
| 19:57 | <cying> | kinda like a float: left but the text that's right adjacent to the float gets indented but everything else (on top and bottom) flows around? |
| 19:58 | <cying> | hopefully? :D |
| 19:58 | <annevk> | still no :) |
| 19:58 | <cying> | right now i've been splitting the div in 2, but it's kinda less than ideal |
| 19:58 | <annevk> | but stuff like that is brought up every now and then |
| 19:59 | <annevk> | I've not really been involved much in the layout part of CSS, but I guess it's tricky |
| 19:59 | <cying> | yea, i was shocked it can't be done |
| 19:59 | <annevk> | welcome to the web |
| 20:01 | <TabAtkins> | Yeah, the algorithms are there to address that case, it's just that no one's done it yet. |
| 20:02 | TabAtkins | has all the layout specs on his plate for working on in the near future. |
| 20:05 | <cying> | *nod* |
| 20:08 | <annevk> | <META name="msapplication-task" content="name=Task 1;action-uri=http://host/Page1.html;icon-uri=http://host/icon1.ico"/>; |
| 20:08 | <annevk> | yay |
| 20:09 | <annevk> | <meta name="msapplication-navbutton-color" content="red"/> whoa |
| 20:09 | <annevk> | it's like we're back in IE3 era |
| 20:15 | <AryehGregor> | You have *got* to be kidding me. |
| 20:15 | <AryehGregor> | WTF? |
| 20:15 | <AryehGregor> | Are they smoking crack? |
| 20:16 | <AryehGregor> | I wouldn't object quite so much to the deliberately and completely proprietary part of it, I expect that from Microsoft, but until now I had some fond idea that maybe their browser team wasn't actually completely incompetent. |
| 20:16 | <AryehGregor> | Then again, the iPhone does some similar things, right? |
| 20:16 | <AryehGregor> | Sigh. |
| 20:16 | <daedb> | Well, they can't let Apple have all the <meta> fun. |
| 20:17 | <annevk> | true, <meta name=viewport> ugh |
| 20:17 | <daedb> | It's just so awesome to put half a dozen <meta> elements on your page just for Iphones and IE9! |
| 20:18 | <annevk> | but that is kind of lemmingy |
| 20:19 | <annevk> | no need to follow bad precedents |
| 20:19 | <Philip`> | At least they didn't make it look for those settings in a hard-coded /application.xml URL |
| 20:20 | <annevk> | you know there is this proposal for /.well-known/ or some such? |
| 20:20 | <annevk> | with a registry et al |
| 20:20 | <annevk> | bah |
| 20:25 | <AryehGregor> | I think Mozilla's Account Manager thing uses /.well-known. I don't see how it makes anything better. |
| 20:25 | <AryehGregor> | Just longer and uglier URLs. |
| 20:26 | <AryehGregor> | As long as it's something that won't be checked on every page view, a well-known location makes sense. |
| 20:27 | <hsivonen> | annevk: thanks for adding the test |
| 20:37 | <hsivonen> | Mike isn't the only one with a trademarked name: http://www.techdirt.com/articles/20100915/00102711022.shtml |
| 20:40 | <AryehGregor> | But he's sure to make much more money if he starts prosecuting. I mean, who's named "Ann de Wees Allen"? |
| 20:43 | <annevk> | lol |
| 20:43 | <annevk> | did we make krijnserver liable for prosecution now? |
| 20:43 | <annevk> | I guess she'd have to set up shop in the Netherlands first |
| 22:32 | <BenoitRen> | Hi everyone. I've been inserting <link>s into documents recently, and I ran into a backwards compatibility issue. |
| 22:34 | <BenoitRen> | HTML5 introduces the rel="up up" and siblings into the mix. The problem is that this is being interpreted as just "up", and takes precedence over the entry that is just rel="up". |
| 22:34 | <BenoitRen> | At least, it's that way on Gecko. |
| 22:34 | <BenoitRen> | This might be something worth mentioning. |
| 22:35 | <AryehGregor> | There's an issue filed in the HTMLWG, I think. |
| 22:35 | <AryehGregor> | Or a bug, at least. |
| 22:36 | <AryehGregor> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=7543 |
| 22:36 | <AryehGregor> | http://www.w3.org/html/wg/tracker/issues/119 |
| 22:36 | <AryehGregor> | Clients can change. |
| 22:36 | <AryehGregor> | Compatibility is not absolutely essential here, and other solutions are probably uglier. |
| 22:37 | <AryehGregor> | It might change back, though, who knows. |
| 22:44 | <BenoitRen> | Just read through it. Thanks for the links. |
| 22:55 | <BenoitRen> | I'm not sure I understand what alternative text to provide for images in galleries of fictional characters. |
| 22:55 | <AryehGregor> | What would you say in place of the pictures if you had to read the page to someone over the phone? |
| 22:56 | <BenoitRen> | "There are 12 thumbnails of Alisa on the page." (and yes, I just read through the guidelines again :( ) |
| 22:57 | <AryehGregor> | If that's all that the page contains, and the person you're talking to really wants to know what the page contains, you'd probably briefly describe each one. |
| 22:57 | <AryehGregor> | I mean, frankly it doesn't work well no matter what you do. |
| 22:57 | <AryehGregor> | Gallery pages are constructed specifically to display pictures, you can't replace them by textual descriptions. |
| 22:57 | <BenoitRen> | Yeah, that's the problem. |
| 22:58 | <BenoitRen> | And a description isn't alternative text. |
| 22:58 | <exp> | does it really need alternative text then? |
| 22:58 | <AryehGregor> | Between you and me (and whoever is reading the logs so as to quote this at an inopportune moment to discredit me), trying to come up with good alt text here is probably a waste of time, just leave it blank or give a few words for each ("Alisa talking with Bob"). |
| 22:59 | <AryehGregor> | A description can be alternative text. It depends. |
| 22:59 | <AryehGregor> | Some of the alt text given is just descriptions. |
| 23:00 | <BenoitRen> | Hmm, I see. |
| 23:05 | <BenoitRen> | Thanks AryehGregor. :) |
| 23:05 | <BenoitRen> | Bye. |