| 00:03 | <Hixie> | othermaciej: man, the copy/cut/paste events are as screwed up as the drag-and-drop events! |
| 00:04 | <othermaciej> | Hixie: sadly I am not surprised :-( |
| 00:12 | <AryehGregor> | Why wouldn't Microsoft just ship the H.264 codecs with XP/Vista if IE10 supports it? It's not going to cost them any money, right? |
| 00:14 | <jonnybarnes> | maybe they have their own agenda and want a different codec being used. |
| 00:14 | <jcranmer> | get people to update to Win7, of course |
| 00:15 | <AryehGregor> | Well, if that were their idea, they could just release IE9 for Windows 7 only. But that doesn't seem to be what they're doing so far. |
| 00:15 | <jonnybarnes> | reading wiki, is canvas patent-free? or do apple still have any patents for it? |
| 00:15 | <jonnybarnes> | wiki = wikipedia |
| 00:15 | <AryehGregor> | jonnybarnes, if they have any patents, they'd have to have declared them when HTML5 reached FPWD at the W3C, as far as I understand it. |
| 00:16 | <jcranmer> | a lot more people will complain if IE 9 is only available on 7 than if codecs are only on 7 |
| 00:16 | <jonnybarnes> | so essentially no-one has to worry about patenting issues? |
| 00:17 | <jcranmer> | apple would probably be shooting themselves in the foot |
| 00:17 | <jcranmer> | but that hasn't stopped people |
| 00:19 | <TabAtkins> | jonnybarnes: That's one of the nicer things about working in the w3c - the patent policy at least helps protect people when implementing the specs. |
| 00:25 | <AryehGregor> | jonnybarnes, you don't have to worry about a member of the HTMLWG having undisclosed patents on anything covered by the spec, no. |
| 00:26 | <AryehGregor> | At least, not on anything necessary to implement it. |
| 00:26 | <AryehGregor> | Not sure how far that goes into related technologies. |
| 00:46 | <quuxman> | what governs the height of a div tag? |
| 00:47 | <TabAtkins> | quuxman: The content, or the css height property. |
| 00:47 | <quuxman> | why would a div be twice as big as it should be? The bottom is just empty |
| 00:47 | <quuxman> | and it has no height property |
| 00:48 | <quuxman> | I just double checked there's no height with Firebug |
| 00:50 | <AryehGregor> | quuxman, link? |
| 00:51 | <quuxman> | oh duh, I had another element expanding it inside of course... just had to figure out which one |
| 01:51 | <Hixie> | anyone remember what the status is of the "image analysis" thread is? |
| 01:51 | <Hixie> | was there some conclusion about what i should change the spec to? |
| 01:52 | <AryehGregor> | I think the conclusion was you should make it vaguer so it doesn't sound like anyone is doing actual image analysis in deployed AT. |
| 01:52 | <AryehGregor> | Like "UAs should think up clever stuff to do if there's no alt text, but any author who relies on this will be held to judgment and suffer in hell for all eternity". |
| 01:52 | <AryehGregor> | (the latter part is non-normative) |
| 01:53 | <AryehGregor> | (unless you think you can God to follow the spec) |
| 01:53 | <AryehGregor> | s/God/get God/ |
| 01:56 | <TabAtkins> | AryehGregor's summary is accurate. |
| 01:57 | <Hixie> | ok |
| 01:57 | <Hixie> | done |
| 01:58 | <Hixie> | i guess i should mail public-html |
| 02:08 | <Hixie> | should we make lang="" missing on <html> trigger a warning? |
| 02:12 | <AryehGregor> | I don't see why. Most UAs don't actually need to know the language, and if they do, they can probably more reliably sniff it from content than by reading the lang tag. |
| 02:12 | <AryehGregor> | Especially for common languages. |
| 02:13 | <Hixie> | hmm |
| 02:13 | <AryehGregor> | What's the lang attribute actually good for anyway, realistically? |
| 02:14 | <Hixie> | other than making the i18n group happy? dunno |
| 02:14 | <Hixie> | i like making the i18n group happy, they're really nice people |
| 02:14 | <Hixie> | they never bitch or anything |
| 02:14 | <AryehGregor> | Heh. |
| 02:15 | <AryehGregor> | It would mean you'd need another boilerplate tag in a minimal document to avoid warnings. |
| 02:15 | <TabAtkins> | Hixie: Would that warning trigger if you omitted <html>? |
| 02:16 | <AryehGregor> | Hmm, I guess you already need to declare your charset to avoid a warning. |
| 02:17 | <Hixie> | TabAtkins: yeah i guess so |
| 02:19 | <TabAtkins> | Eh, whatever then. I don't care much about warnings anyway. I'll just continue omitting it because it's cooler that way. |
| 02:28 | <KevinMarks> | doesn't requiring a lang imply that the page only has one langauge in it? |
| 02:29 | <MikeSmith> | KevinMarks: no, I don't think it does |
| 02:30 | <MikeSmith> | it specifies the default language for the page |
| 02:30 | <Hixie> | yeah this would just be the default |
| 02:30 | <MikeSmith> | at least as far as I understand it |
| 02:30 | <Hixie> | i guess i'll leave it as is |
| 02:31 | <KevinMarks> | ah right, so you can override it |
| 02:31 | <MikeSmith> | Hixie: fwiw, validator.nu already emits a warning for a similar case, which is if there's no character-encoding information given in the doc |
| 02:31 | <Hixie> | even if the doc is all-ascii? |
| 02:31 | <KevinMarks> | Like Stephanie does: http://climbtothestars.org/archives/2010/01/26/my-journey-out-of-procrastination-doing-things-now/ |
| 02:31 | <MikeSmith> | yes |
| 02:31 | <MikeSmith> | Hixie: yeah, even for all-ascii, I think |
| 02:32 | <KevinMarks> | character encoding is not similar to language |
| 02:32 | <KevinMarks> | encoding is a computational construct, language is a human one |
| 02:34 | <Hixie> | MikeSmith: iirc the spec has a MUST if there's no encoding and it's not ascii (and some other conditions aren't met, e.g. boms, http headers), and there's no conformance problem at all without one if it's pure ASCII |
| 02:35 | <MikeSmith> | Hixie: I'm not sure that it's possible to have validator.nu make a determination of whether a doc is pure ascii or not |
| 02:35 | <MikeSmith> | it would have to sniff the entire doc, would it not? |
| 02:35 | <AryehGregor> | Doesn't it have to do that anyway? |
| 02:35 | <MikeSmith> | anyway, my point was that we do have a precedent of emitting a warning for a case that's not a strict conformance error but that could lead to problems with portability of the document |
| 02:36 | <Hixie> | yeah |
| 02:37 | <MikeSmith> | the message it currently emits is something that it emits early in processing, by looking just at the first 512 bytes for a character-encoding declaration |
| 02:37 | <MikeSmith> | or for a lack of one |
| 02:37 | <Hixie> | yeah it would take more work to do the right thing there |
| 02:38 | <MikeSmith> | it's certainly do-able, but not sure it would be worth it or not |
| 02:38 | <MikeSmith> | that code is in the HTML parser |
| 02:38 | <AryehGregor> | Could leaving the lang unspecified actually "lead to problems with portability of the document"? |
| 02:38 | <MikeSmith> | AryehGregor: I dunno |
| 02:38 | <Sonjaaa> | ahhh http://www.theregister.co.uk/2010/02/04/mpeg_la_h_264_codec_licence/ ! |
| 02:39 | <Sonjaaa> | h264 will stay royalty free! |
| 02:39 | <Sonjaaa> | i don't understand why firefox won't have it |
| 02:39 | <jcranmer> | um |
| 02:40 | <jcranmer> | what that basically means is you don't have to pay per individual use |
| 02:40 | <Sonjaaa> | ohhh |
| 02:40 | <jcranmer> | you still have to pay licensing fees |
| 02:40 | <Sonjaaa> | i see. tahnks fo helping me understand what is going on |
| 02:40 | <Sonjaaa> | with the future of webs! |
| 02:41 | <Sonjaaa> | multimedia webs! |
| 02:42 | <AryehGregor> | It will also only stay free until 2016. |
| 02:42 | <AryehGregor> | At that point they could impose fees until well into the 2020s, when the last patent expires. |
| 02:42 | <AryehGregor> | 2028, I think? |
| 02:42 | <Sonjaaa> | is this similar to the unisys/gif patent thingy? |
| 02:42 | <KevinMarks> | can you really assert the page is ascii just by checking all chars are 7-bit? |
| 02:42 | <jcranmer> | kind of |
| 02:42 | <MikeSmith> | AryehGregor: I don't know what all problems might arise with a lang-less document, but if the advice of the i18n folks is that we should warn about it lacking, their advice is worth considering very carefully. As Hixie points out, the i18n group are pretty nice people to work with, and they don't ask for stuff often, and when they do, it's always with a very considered opinion and communicated with tact |
| 02:42 | <AryehGregor> | But Mozilla would have to pay $5 million/year anyway to implement it in its browser, it wouldn't fall under the license-free clause. |
| 02:43 | <NickYoung> | incidentally, I still haven't been able to get QtWebkit working with youtube/html5. I'm thinking they're checking the user agent of the browser and the media engine. |
| 02:43 | <AryehGregor> | MikeSmith, maybe we could ask them why? Where did they ask for this? |
| 02:43 | <Sonjaaa> | i think the gif patent has expired now? |
| 02:43 | <Sonjaaa> | i still use png |
| 02:43 | <AryehGregor> | Sonjaaa, yes, GIF patents expired a few years ago. |
| 02:43 | <jcranmer> | except that MPEG-LA is probably better able to enforce patents than Unisys |
| 02:43 | <AryehGregor> | Unfortunately, H.264 is patented for a long while to come. |
| 02:43 | <MikeSmith> | I think the IPTV industry might end up being the single biggest force to help unwind the H.264 monoply |
| 02:44 | <KevinMarks> | m'colleague at Apple had a millennium countdown clock modified to countdown to the GIF patent expiry. |
| 02:44 | <jcranmer> | especially because GIF was widely used before Unisys started enforcing the patents |
| 02:44 | <jcranmer> | and H.264 was widely used after enforcement |
| 02:44 | <KevinMarks> | when it expired, he checked in encoding into QuickTime |
| 02:44 | <Sonjaaa> | is there no "free" equivalent of h.264? |
| 02:44 | <AryehGregor> | Sure is, Theora. |
| 02:45 | <MikeSmith> | lacking a free codec, IPTV hardware and service vendors and all their customers are otherwise going to be paying mountains of money in royalties |
| 02:45 | <Sonjaaa> | but youtube and such don't want to switch to theora? |
| 02:45 | <AryehGregor> | But Apple refuses to support Theora, and Google isn't distributing Theora video, and Microsoft probably will try to avoid supporting it too. |
| 02:45 | <Sonjaaa> | all the big boys don't like theora |
| 02:45 | <AryehGregor> | For a variety of reasons. |
| 02:45 | <Sonjaaa> | i'm assuming that theora is like vorbis? it gets put in an ogg file? |
| 02:45 | <AryehGregor> | Yes, Theora is put in OGG files, generally. |
| 02:46 | <MikeSmith> | Sun was working on a royalty-free codec before they got bought up by Oracle. Maybe Oracle will take up work on it again |
| 02:46 | <Sonjaaa> | avi is also a container file... you could have theora in avi too? |
| 02:47 | <Sonjaaa> | so they are disputing what should be the standard for html5? |
| 02:47 | <NickYoung> | youtube's problem with theora is a supposed concern about quality-per-bit |
| 02:50 | <MikeSmith> | KevinMarks: I wrote a remarkably similar set of articles to that series by Stephanie Booth. except my first article was titled, "My journey out of procrastination, part 1: Preparing for the journey" and part two was, "Still working on getting ready, give me a little more time" and part 3 was "Quitting bugging me about it already", etc. |
| 02:51 | <MikeSmith> | and I wrote another series pretty much along the same lines, "My journey towards sobriety" |
| 02:51 | <quuxman> | what exactly does vertical-align do? I'm using it to center inline-block elements inside a block element, and it works fine, but in another spot it's not working |
| 02:53 | <Sonjaaa> | is there an ogg theora hardware decorder? |
| 02:54 | <quuxman> | I see... it actually determines the alignment of adjacent elements |
| 02:55 | <quuxman> | in reference to itself. So it only works if you have an element that spans the entire height of the parent |
| 02:55 | <annevk> | fyi: http://www.defectivebydesign.org/ipad |
| 02:55 | <NickYoung> | Sonjaaa: Not as yet, I believe |
| 02:56 | <kinetik> | Sonjaaa: depends how you define "hardware", but yes: http://www.schleef.org/blog/2009/11/11/theora-on-ti-c64x-dsp-and-omap3/ |
| 02:56 | <MikeSmith> | getting Theora supported on more hardware would be a great thing |
| 02:57 | <MikeSmith> | especially in chipsets for mobile devices |
| 02:57 | <doublec> | also theora decoder on fpga: http://www.youtube.com/watch?v=fPmMYG6_VgI |
| 02:58 | <MikeSmith> | kinetik, doublec - great stuff, thanks for the heads-up |
| 02:59 | <quuxman> | the old vertical alignment problem. No good way to do it. It still baffles me |
| 02:59 | <MikeSmith> | doublec: you guys are supporting this work? |
| 02:59 | <quuxman> | er, vertical centering |
| 02:59 | <doublec> | MikeSmith, we supported the c64x dsp/omap3 stuff |
| 02:59 | <doublec> | the fpga stuff is fairly old |
| 03:00 | <MikeSmith> | ok |
| 03:02 | <Sonjaaa> | thanks ! |
| 03:05 | <nessy> | I am a big believer - I believe Google will "be good" and eventually provide a Ogg Theora alternative to YouTube |
| 03:05 | <Sonjaaa> | is google still "not evil" ? |
| 03:06 | <nessy> | well … I think they're trying to |
| 03:06 | <Sonjaaa> | their android OS is a fork, eh |
| 03:07 | <nessy> | pretty difficult as a large corporate |
| 03:07 | <nessy> | I haven't lost hope :) |
| 03:08 | <MikeSmith> | nessy: hey, thanks for the reply about the accessibility media issues. Eric Carlson was on the telcon and gave a summary of what you all had been discussing thus far. I think he said you were hoping to find some time to have a phone discussion some time soon |
| 03:08 | <nessy> | well, I'm not organising the media subgroup |
| 03:09 | <nessy> | I am hoping that John Foliot and/or David Singer will organise such a phone hook up soon |
| 03:09 | <nessy> | it would be good to have in about 2 weeks, I think |
| 03:10 | <nessy> | I have three new discussion threads to start and then with that feedback we should be ready for the phone call and some agreement on moving forward |
| 03:11 | <MikeSmith> | nessy: so far it seems like both John and Dave have not be able to free up enough time to really drive the work and expedite it. If you were able to help with leading it more, maybe we could get it moved along faster |
| 03:11 | <MikeSmith> | if you had time |
| 03:11 | <nessy> | that's why I volunteered on the next steps |
| 03:11 | <MikeSmith> | OK |
| 03:12 | <nessy> | I cannot promise it - right now I'm basically out of contract |
| 03:12 | <nessy> | I do need an income ;) |
| 03:12 | <MikeSmith> | I see |
| 03:12 | <MikeSmith> | yeah, understood |
| 03:12 | <nessy> | but I am preparing another funding request so hopefully that will go through |
| 03:13 | MikeSmith | wishes he could get his hands on some Web Foundation money... |
| 03:13 | <nessy> | I'll do what I can - I'm definitely not giving up |
| 03:13 | <nessy> | but in Dec and Jan I wasn't able to do anything, so not much happened about media a11y |
| 03:14 | <nessy> | had other contracts to finish |
| 03:14 | <nessy> | right now it's all good, but I can't say until when |
| 03:14 | <MikeSmith> | nessy: well, going forward, whatever I can do to help with the media a11y stuff, let me know |
| 03:14 | <nessy> | awesome, thanks |
| 03:15 | <nessy> | maybe we can start setting up a phone conf meeting for in 2 weeks |
| 03:15 | <nessy> | I'm sure John and Dave would agree, but just haven't time themselves to organise it |
| 03:15 | <nessy> | if you could do that, that would really help, IMHO |
| 03:20 | <MikeSmith> | you mean just a one-time call, or do you want to set up a regular weekly call? |
| 03:22 | <nessy> | just a one-time call |
| 03:22 | <MikeSmith> | OK |
| 03:22 | <nessy> | we work very productively over email, so don't need weekly calls |
| 03:23 | <nessy> | but a one-time call is probably a good thing now |
| 03:23 | <nessy> | just to make sure we are all on the same page |
| 03:23 | <MikeSmith> | OK. any particular day of that week? (the week of the 15th to the 18th? or after?) |
| 03:25 | <nessy> | how about 23rd? |
| 03:25 | <nessy> | oh, we said a11y meeting on 19th will be focused on media? |
| 03:26 | <nessy> | so maybe do it a day or two before that |
| 03:26 | <nessy> | at a time that suits me ;) |
| 03:28 | <nessy> | how about http://www.timeanddate.com/worldclock/meetingdetails.html?year=2010&month=2&day=17&hour=21&min=0&sec=0&p1=224&p2=179&p3=240 |
| 03:28 | <nessy> | should pass it by Dave and John, I guess |
| 03:29 | <MikeSmith> | nessy: yeah, I will send a message to the list proposing that |
| 03:29 | <MikeSmith> | should make sure that Eric can make it too |
| 03:29 | <nessy> | indeed - and Philip! |
| 03:29 | <MikeSmith> | ah yeah |
| 03:29 | <nessy> | I might even involve Ken from Google, even if he otherwise loathes standards participation |
| 03:30 | <MikeSmith> | as far as the regular weekly a11y call, I suggested that we move it that week to a different, eastern-hemisphere-friendly time, so that you could be on that week |
| 03:30 | <MikeSmith> | wow, somebody who loathes standards participation, that's a rarity :) |
| 03:30 | <MikeSmith> | I can't imagine why anybody wouldn't like it.. the rest of us love it so much |
| 03:31 | <MikeSmith> | people just don't know what they're missing |
| 03:33 | <nessy> | all these misunderstandings, repeated explanations, heated arguments over nothing, and finally little moves towards correct technical solutions - really don't know what they are missing! |
| 03:33 | <nessy> | ;) |
| 03:35 | <MikeSmith> | yep |
| 03:35 | <MikeSmith> | the drama of it all |
| 03:36 | <MikeSmith> | the Thrill of Victory and the Agony of Defeat |
| 03:36 | <nessy> | heckling is fun |
| 03:36 | <nessy> | it's the modern oriental market |
| 03:36 | <nessy> | you have to know how to stand your ground to sell your carpet ;) |
| 03:37 | <MikeSmith> | it's an embarrassment of riches, as far as the opportunities for heckling go |
| 03:40 | nessy | chuckles |
| 03:41 | <MikeSmith> | nessy: is there anybody in the media subgroup who's on the US east coast? |
| 03:41 | <nessy> | are you? |
| 03:41 | <nessy> | I assume you might be interested in participating? |
| 03:42 | <nessy> | ah, you're right - we need to make sure that Philip can participate - New York is not that important |
| 03:43 | <MikeSmith> | nessy: I live in Tokyo |
| 03:43 | <MikeSmith> | so I'm roughly in the same timezone comfort range as you |
| 03:43 | <MikeSmith> | the time you suggested is 22:00 for Philipp, so not too heinous |
| 03:44 | <MikeSmith> | I will propose it and Cc him and JohnF and Dave and Eric and see what they say |
| 03:44 | <MikeSmith> | nessy: meet for 90 minutes? |
| 03:45 | <nessy> | nah, maximum 60 should be fine |
| 03:45 | <nessy> | don't want to drag it out |
| 03:45 | <MikeSmith> | OK |
| 03:47 | <nessy> | 6am for you is a bit early! |
| 03:47 | <nessy> | but with such requirements, it might be hard |
| 03:49 | <nessy> | so, basically I want to discuss three things: |
| 03:49 | <nessy> | * Multitrack JavaScript API |
| 03:49 | <nessy> | * external caption formats |
| 03:49 | <nessy> | * declarative syntax for associating external text resources |
| 03:58 | <MikeSmithX> | nessy: http://lists.w3.org/Archives/Public/public-html-a11y/2010Feb/0063.html |
| 03:58 | <MikeSmithX> | foolip: ↑ |
| 03:59 | <nessy> | cool - I will reply with agenda items :) |
| 04:01 | <MikeSmithX> | thanks |
| 04:03 | <nessy> | John Foliot just said he's fine with that time |
| 04:04 | <MikeSmith> | great |
| 04:15 | <nessy> | thanks for organising! will you do zakim and stuff, too? |
| 04:16 | <MikeSmith> | nessy: yep |
| 04:16 | <nessy> | you rock :) |
| 04:21 | <MikeSmith> | I'm just the cook.. I gets paid two bottles of wine a week, and I don't have to pick no cucumbers or tomatoes |
| 04:24 | <nessy> | lol |
| 04:24 | <nessy> | I am a passionate tomato picker! |
| 04:27 | <nessy> | I only sometimes need to be paid to buy other fruit and vegetables - so I can turn all my attention to the tomatoes. |
| 04:29 | <MikeSmith> | tomatoes is one of God's greatest creations |
| 04:30 | <quuxman> | does IE support the display:table, display:table-cell vertical centering hack yet? |
| 04:33 | <Hixie> | tomatoes are evil |
| 04:36 | <MikeSmith> | Hixie: tomatoes told me that they like you |
| 04:36 | <nessy> | hahahaha |
| 04:44 | <TabAtkins> | quuxman: IE8 does. |
| 04:46 | <TabAtkins> | Or rather, it supports the full table-* set of display values, and all the hacks that entails. |
| 04:46 | TabAtkins | may go ahead and change his company's websites over to using table-* rather than <table> for the two-column, and just use js to hack IE7 into shape. |
| 05:04 | <quuxman> | TabAtkins: what's the market share of IE8 these days? |
| 05:46 | <MikeSmith> | man, it would be really nice if when I were using a comma-separated list of css selector expressions, I could have a comma after the last one and it would still work as expected |
| 05:46 | <MikeSmith> | Hixie: is there a reason why that case should not be allowed? |
| 05:47 | <MikeSmith> | TabAtkins: ↑ |
| 05:48 | <MikeSmith> | in many programming languages when you have a structure with a comma-separated list like that, having a comma after the final one in the list is valid |
| 05:48 | <MikeSmith> | it's just annoying in css when you're doing a lot of copy and paste |
| 05:50 | <TabAtkins> | MikeSmith: No clue. Bring it up on the www-style list? |
| 05:51 | <MikeSmith> | hai |
| 07:13 | <quuxman> | is there a better way in CSS to say ".foo[class~=bar] { ... }" to match an element with class="foo bar"? |
| 07:14 | <MikeSmith> | quuxman: .foo.bar I think |
| 07:14 | <MikeSmith> | string em together like a choo-choo train |
| 08:09 | <MikeSmith> | Hixie: have you been noticing any increased delays when committing to dev.w3.org? |
| 08:10 | <MikeSmith> | I'm getting long delays between calling cvs and getting any messages back |
| 08:10 | <MikeSmith> | doesn't seem to be due to locking, but just.. slow |
| 08:29 | <Hixie> | MikeSmith: no, but i don't generally notice when things are slow at that stage |
| 08:29 | <MikeSmith> | ok |
| 11:36 | <annevk> | I initially thought http://blog.vlad1.com/2010/02/05/mjs-simple-vector-and-matrix-math-for-js/ was about othermaciej |
| 11:36 | <annevk> | o_O |
| 11:36 | <othermaciej> | that is indeed a funny thing to call a library |
| 11:45 | <MikeSmith> | hsivonen: http://bugzilla.validator.nu/show_bug.cgi?id=716 |
| 11:45 | <MikeSmith> | from tantek |
| 11:46 | <MikeSmith> | I can't reproduce it |
| 11:47 | <hsivonen> | that bug report reminds me that I should ask rubys if he has gotten requirements from tantek and zeldman |
| 11:48 | <hsivonen> | if the polyglot stuff isn't going anywhere, it should be taken out of the UI in order to stop confusing users |
| 11:49 | <hsivonen> | I see the bug in deployment |
| 11:49 | <hsivonen> | let's try locally... |
| 11:51 | <MikeSmith> | hsivonen: I suppose the polygot stuff could be kept and made a runtime option for others who want to run the own downstream installs |
| 11:52 | <MikeSmith> | hsivonen: I suspect the problem is caused by the changes I made to move the required-attributes reporting to the assertions-checking code |
| 11:52 | <MikeSmith> | that stuff is working as expected for me locally |
| 11:52 | <MikeSmith> | but maybe I neglected to check in some necessary part |
| 11:52 | <MikeSmith> | anyway, I need to just revert that change |
| 11:53 | <MikeSmith> | in anticipation of getting the required-attributes reporting into the jing v.nu branch |
| 11:54 | <MikeSmith> | maybe you can take a look at the generated embed-vnu.rnc file and see if that's correct |
| 11:54 | <MikeSmith> | I can't figure what else would cause this problem |
| 11:56 | MikeSmith | starts a clean install |
| 11:56 | <hsivonen> | sigh. looks like I have a failing assertion in the parser |
| 11:57 | <MikeSmith> | what a relief that it's not me that broke something this time :) |
| 11:57 | <hsivonen> | I suspect, you broke something, too |
| 11:57 | <MikeSmith> | damn |
| 11:58 | <MikeSmith> | just when I was getting ready to pat myself on the back |
| 11:59 | <MikeSmith> | hsivonen: on the serious/bright side, dunno if you saw the patch submitted from Carey Evans, to build.py |
| 11:59 | <MikeSmith> | to get it work as expected under Windows |
| 12:00 | <MikeSmith> | he mentions that it seems to run and work fine |
| 12:00 | <MikeSmith> | last I remember, you had said that you couldn't get v.nu to run correctly on Windows |
| 12:00 | <hsivonen> | I was unaware of a patch. |
| 12:00 | <hsivonen> | I'm near email bankruptcy again |
| 12:01 | <MikeSmith> | close your eyes and start deleting |
| 12:02 | <hsivonen> | sigh. Eclipse doesn't break on assertions if I tell it to break on AssertionError |
| 12:02 | <MikeSmith> | http://bugzilla.validator.nu/show_bug.cgi?id=715 |
| 12:02 | <MikeSmith> | http://bugzilla.validator.nu/attachment.cgi?id=147 |
| 12:02 | <hsivonen> | hmm. no, I'm again looking at and debugging different things in Eclipse |
| 12:02 | <MikeSmith> | I guess I should get my Windows VM working again so I can help test on Windows |
| 12:03 | <hsivonen> | oh. awesome. so there's a way to create child processes from python that works on Windows |
| 12:04 | <MikeSmith> | yeah, seems so |
| 12:04 | <MikeSmith> | Carey deserves at least a beer for that |
| 12:05 | <hsivonen> | now eclipse breaks correctly |
| 12:05 | <hsivonen> | or an non-alcoholic beverage :-) |
| 12:06 | <MikeSmith> | hsivonen: yeah, true, maybe some laudanum |
| 12:06 | <MikeSmith> | hmm, I guess maybe that has alcohol too |
| 12:07 | <MikeSmith> | maybe could make some alcohol-free laudanum |
| 12:09 | <hsivonen> | so it seems that the assertion is revealing a real parser bug... |
| 12:09 | <hsivonen> | hmm. |
| 12:09 | <hsivonen> | no |
| 12:09 | <hsivonen> | doh |
| 12:09 | <hsivonen> | the assertion is bogus |
| 12:09 | <hsivonen> | xml:lang strikes again! |
| 12:10 | <hsivonen> | I've been living in the Gecko land so long that I've forgotten about other parser configurations when writing my assertions |
| 12:14 | <hsivonen> | MikeSmith: I can reproduce locally the bug tantek reported. |
| 12:14 | <hsivonen> | MikeSmith: so chances are you haven't checked everything in |
| 12:17 | <hsivonen> | MikeSmith: btw, I *think* my review queue is empty. Do you have patches that I should review but that I've missed? |
| 12:17 | <MikeSmith> | hsivonen: I don't think I do have any patches that I've sent to you but that you've not reviewed |
| 12:17 | <MikeSmith> | I need to finish the jing patch and get that to you |
| 12:17 | <MikeSmith> | but I won't get around to that til next week |
| 12:17 | <MikeSmith> | I basically need to trim down what I ported over from George's patch |
| 12:17 | <MikeSmith> | his patch builds a string representation of the content model |
| 12:17 | <MikeSmith> | rnc syntax |
| 12:18 | <MikeSmith> | but since we don't need/want to emit the model, but instead just the names of the missing attributes/elements, it seems like I should trim that part out |
| 14:23 | <zcorpan> | Lachy: it'd be in the html namespace if you specify the html namespace |
| 14:23 | <Lachy> | as long as you do while(stripos($input, ($token = random))); then your token will be secure |
| 14:24 | jgraham | wishes brutal death from above on people who advertise their website as being at foo.example but actually only have a server at www.foo.example |
| 14:24 | <Lachy> | zcorpan, oh, oops. I misread it. Kornel was using that as the prefix, not the namespace URI. |
| 14:25 | <Philip`> | othermaciej: If people find collision attacks on SHA512, I think HTML sandboxes are likely to be the least of people's concerns |
| 14:25 | <Philip`> | (Salt is helpful to make dictionary attacks n times harder where n is the number of users you're trying to attack simultaneously, and I'm not sure it's relevant here) |
| 14:28 | <othermaciej> | it would mean that even if you computed one viable attack input by brute force, it would only work when the salt value you used in generating it comes up |
| 14:29 | <othermaciej> | (which leaves you with as strong a defense as just using a good quality random number |
| 14:30 | <annevk> | jgraham, shouldn't browsers fix that? |
| 14:30 | <Philip`> | If you pick an arbitrary $token and then manage to find an input that hashes to that (with </sandbox token=$token> inside it), then the hash is severely broken and lots of cryptographic systems would collapse |
| 14:31 | <Philip`> | and I don't see how the token thing could be broken in easier ways |
| 14:32 | <Philip`> | and I'm happy to believe SHA512 is not severely broken enough for this to be an issue in the next decade |
| 14:35 | <jgraham> | annevk: Browsers are not the only UAs, you know (and not all browsers fix it) |
| 14:35 | <Philip`> | (...and you're not going to compute one viable attack by brute force, because there's more hash outputs than there are atoms in the universe) |
| 14:39 | <annevk> | jgraham, I guess people should use Google first then o_O |
| 14:41 | <zcorpan> | "52framework is a CSS framework which aims to provide an easy way to build websites using HTML5 & CSS3 while still supporting all modern browsers (including ie6)." |
| 14:41 | <zcorpan> | wait, ie6 is a modern browser? |
| 14:42 | <othermaciej> | Philip`: anyone knowledgable enough to make a considered judgment about what hash algorithm to use is also knowledgable enough to know how to get good random numbers |
| 14:44 | <Lachy> | zcorpan, unfortunately, for some people, developing sites for compatibility with IE6 is a reality. |
| 14:44 | <Lachy> | Although, that is rapidly changing now that some major sites have finally dropped support for it |
| 14:45 | <hsivonen> | Lachy: that doesn't make IE6 modern, though |
| 14:45 | <hsivonen> | on the site of the framework, text overlaps a vertical rule |
| 14:45 | <Lachy> | based on the quote, I think the implication that IE6 is modern is a grammatical ambiguity |
| 14:47 | <Philip`> | othermaciej: I'm not sure I agree with that |
| 14:47 | <Philip`> | othermaciej: It's easy to tell someone to use SHA512 because that's a good hash function, and they can easily find how to access it in their programming language |
| 14:48 | <othermaciej> | it's just as easy to tell them to use, say, arc4random |
| 14:48 | <hsivonen> | my attention span timed out before I found out what license 52framework is under |
| 14:48 | <othermaciej> | (though I suppose it might be less likely their language already has it) |
| 14:50 | <Philip`> | othermaciej: If I try to search for how to generate secure random numbers in PHP, I get pointers to mt_rand and sha512(mt_rand) and lots of things that seed based on the time |
| 14:50 | <Lachy> | othermaciej, as long as $token is subsequently generated to be a sufficiently long randomised string, based on the random number, as opposed to an integer between 0 and 32768 |
| 14:50 | <Philip`> | Looks like the only way to do it in PHP is to read /dev/urandom |
| 14:50 | <Philip`> | or something else on other platforms |
| 14:50 | <Philip`> | as far as I can tell |
| 14:50 | <Lachy> | (funtions like rand() in PHP or, apparently, arc4random() just return integers) |
| 14:51 | <Philip`> | and it's very easy to get it wrong, e.g. by using a wrong RNG or not seeding sensibly |
| 14:52 | <annevk> | why do we want <sandbox>? |
| 14:52 | <annevk> | what's wrong with <iframe srcdoc>? |
| 14:52 | <Philip`> | We probably don't |
| 14:52 | <othermaciej> | sure, you'd want more than 32 bits of randomness, but it doesn't really matter if you convert numbers to strings as ASCII or in some other way |
| 14:52 | <TabAtkins> | I still like srcdoc. Nuts to randomness or hashing. Just a single, trivial, quickly-and-obviously-failing escaping step is better. |
| 14:53 | <othermaciej> | <sandbox> would have some potential advantages over <iframe srcdoc>: |
| 14:53 | <othermaciej> | 1) frames have high memory and performance overhead - a page with hundreds or thousands of iframes would likely be a bad user experience |
| 14:54 | <jgraham> | Oh I thought we wanted it because security is a fun new thing to argue about once we are bored with accessibility and Process |
| 14:54 | <Lachy> | othermaciej, if your token is just a number convered to ascii, like <sandbox-18489>, where that number is randomly generated each time the page is generated, then an attacker could insert </sandbox-25930> and get a successful attack on approximatley 1/32768 visitors |
| 14:55 | <othermaciej> | 2) if your desired fallback is content that sizes naturally in the document, then it's hard to use <iframe> at all until all browsers support seamless iframes |
| 14:55 | <othermaciej> | (i.e if you are using sandboxing in combination with server-side sanitization for defense in depth) |
| 14:55 | <TabAtkins> | othermaciej: By the time @sandbox and @srcdoc are supported properly, @seamless should be too. |
| 14:55 | <annevk> | it seems it would be even harder to use <sandbox> othermaciej |
| 14:56 | <othermaciej> | I'm not saying it's better overall, just listing the possible advantages |
| 14:56 | <othermaciej> | it has some disadvantages too |
| 14:56 | <othermaciej> | there's no obvious way to do <sandbox allow-script> |
| 14:56 | <Philip`> | Lachy: If your token is based on a number between 0 and 32767 converted to a long randomised string by applying sha512() to it, you'll have the same problem |
| 14:57 | <othermaciej> | Lachy: you need more than 32 bits, but how you convert it to a string doesn't really matter |
| 14:58 | <othermaciej> | (incidentally, 2^32 is 4,294,967,296, not 32,768) |
| 14:58 | <othermaciej> | one in 4 billion is probably too close for comfort but even 64 bits of randomness would be sufficient |
| 14:58 | <Philip`> | (PHP's rand() apparently goes up to 2^16 on some platforms (Windows)) |
| 14:59 | <TabAtkins> | Yeah, rand() has a platform-specific max |
| 14:59 | <othermaciej> | I'm sure using PHP's rand() is terrible |
| 14:59 | <TabAtkins> | Believe me, that's what will be used. |
| 14:59 | <jgraham> | I think that is a killer |
| 14:59 | <zcorpan> | you could use 16 bit rand several times and add them together |
| 15:00 | <othermaciej> | bad random number generators are an attractive nuisance |
| 15:00 | <meledin> | If someone is inclined to use rand() anyway I doubt they'd take that extra step? |
| 15:00 | <TabAtkins> | zcorpan: And get less than a bit per extra? |
| 15:00 | <jgraham> | It's bad enough if part of the solution is people typing S-H-A into their code, let alone saying "you need a random number generator (but not that one) |
| 15:00 | <othermaciej> | Window's system random number generators are also really weak in addition to not using enough bits |
| 15:00 | <Philip`> | (With 32 bits of randomness you just need to put 64K sandboxes on a page and there's a good chance one of them will collide) |
| 15:00 | <Philip`> | (I think) |
| 15:00 | gsnedders | needs to decide whether to go to uni in Edinburgh or Glasgow |
| 15:01 | <othermaciej> | we need a birthday paradox calculator |
| 15:01 | <Philip`> | (Oh, actually that's rubbish) |
| 15:01 | <jgraham> | gsnedders: You have an offer frm E.? |
| 15:01 | <Philip`> | (because it's not any-vs-any) |
| 15:01 | <gsnedders> | jgraham: yes. |
| 15:01 | <jgraham> | gsnedders: Go there |
| 15:01 | <MikeSmith> | hsivonen: if/when you sync up to the build r91 commit and still see any problems, let me know.. but I think that should fix the problem tantek found (for now at least, until we get the jing missing-attributes enhancement in ) |
| 15:01 | <gsnedders> | jgraham: That's what I'll probably do. |
| 15:01 | <Philip`> | gsnedders: Go to both |
| 15:01 | <gsnedders> | Philip`: :P |
| 15:02 | <gsnedders> | But Glasgow has joint-honours CS and Eng. lang. which is tempting |
| 15:02 | <jgraham> | He's not doing social studies with a gap year in a university in an underprivilidged region, you know |
| 15:03 | <gsnedders> | Like the east-side of Glasgow? |
| 15:03 | <jgraham> | or the west side |
| 15:04 | <gsnedders> | (Calton has a life expectancy of 54) |
| 15:04 | annevk | reads the remainder of the "integration of HTM" |
| 15:04 | <othermaciej> | Philip`: ah, right, birthday paradox doesn't apply |
| 15:04 | <Philip`> | Does joint CS and Eng. lang. mean you only do half as much of each? |
| 15:04 | <othermaciej> | I wonder if the L was accidentally or intentionally omitted in that thread title |
| 15:04 | <Philip`> | That sounds bad in terms of learning anything particularly useful in either one |
| 15:04 | <gsnedders> | Philip`: Yes. |
| 15:05 | <jgraham> | Philip`: Yes (to the second point) |
| 15:05 | <jgraham> | (so that was really directed at gsnedders) |
| 15:05 | annevk | doesn't like how Richard just seems to try to push things through without listening to feedback on complexity |
| 15:06 | <TabAtkins> | If anyone has an infinite-precision float library sitting around, the problem is simply n = log(.5) / log( (2^32)-1 / 2^32 ) |
| 15:06 | <othermaciej> | annevk: I am getting the impression that he's trying to strong-arm people into going with his preferred design |
| 15:06 | <TabAtkins> | Unfortunately, I just get 0 back from the latter part. |
| 15:07 | <meledin> | -1/32 |
| 15:07 | <othermaciej> | annevk: I'm not sure if he is being intentionally misleading about the level of support his proposals have from Mozilla and Apple (and, in one case, specifically me) or if he just hears what he wants to hear |
| 15:07 | <gsnedders> | Also: it is somewhat ironic that Edinburgh teaches more Haskell than Glasgow. |
| 15:08 | <othermaciej> | TabAtkins: the birthday paradox number for 2^32 possibilities would be only 30084 for a 10% chance of collision |
| 15:08 | <othermaciej> | but, birthday paradox does not actually apply to this problem |
| 15:08 | <TabAtkins> | Yeah, but we're not hitting birthday. |
| 15:08 | <TabAtkins> | This is different. |
| 15:08 | <Philip`> | gsnedders: (I don't know about other subjects, but in CS there's usually a load of introductory stuff in the first year because some people don't know programming or basic algorithms etc, and if you're someone who's good at computers then it'll be a while before the more challenging stuff gets introduced) |
| 15:08 | <TabAtkins> | It's a much larger number. |
| 15:08 | <jgraham> | TabAtkins: Wolfram Alpha |
| 15:08 | <othermaciej> | it's simply linear |
| 15:09 | <othermaciej> | to have a 10% chance of collision you need 0.1 * 2^23 tries |
| 15:09 | <gsnedders> | Philip`: Yeah, I know |
| 15:09 | <othermaciej> | since each is independent |
| 15:09 | <gsnedders> | Philip`: Edinburgh starts with Haskell and other related things, whereas Glasgow starts with Python |
| 15:10 | <othermaciej> | maybe somewhat more, that's actually the number needed for your expected number of collisions to be .1 |
| 15:10 | <Philip`> | gsnedders: (and I guess if you only spend half as much time on CS then you'll be almost at the end of the course before it starts doing anything exciting) |
| 15:10 | <othermaciej> | but since you can have more than 1, that's not the same as the probability of having at least one |
| 15:10 | <gsnedders> | Philip`: In the first two years you get all the CS stuff. |
| 15:10 | <gsnedders> | Philip`: As in Scotland you normally do three subjects in first year, and two in second. |
| 15:10 | <TabAtkins> | Ah, yes, I'm getting that you need a 31-bit number of tries. |
| 15:11 | <gsnedders> | So it's only in honours (i.e., the final two years) where things really differ. |
| 15:11 | <TabAtkins> | (I realized that you could simplify the expression to a single log in a different base.) |
| 15:12 | <Philip`> | gsnedders: Ah, sounds all weird and crazy |
| 15:17 | <Dashiva> | If you have two prefixes (a and b) identifying the same namespace, can you do <a:e></b:e> ? |
| 15:18 | <Philip`> | No |
| 15:19 | <gsnedders> | No, that's no longer well formed XML |
| 15:20 | <Philip`> | "To conform to this specification, a document MUST be well-formed according to the XML 1.0 specification" |
| 15:20 | <Philip`> | (says XMLNS) |
| 15:20 | <gsnedders> | It needs to be both well-formed XML (per /TR/xml/) and namespace-well-formed (per /TR/xml-names/) |
| 15:20 | <Philip`> | (and "A document is namespace-well-formed if it conforms to this specification.") |
| 15:20 | <Dashiva> | So it's not well-formed, but is it namespace-well-formed? |
| 15:21 | <Philip`> | No |
| 15:21 | <gsnedders> | Namespace-well-formedness is only defined for things that are well-formed |
| 15:21 | <Dashiva> | kk |
| 15:21 | <Philip`> | If it's not well-formed XML, it doesn't conform to XMLNS, so it's not namespace-well-formed |
| 15:22 | <Philip`> | gsnedders: I don't see why that's the case |
| 15:22 | <Philip`> | It's defined in terms of conformingness to the XMLNS spec |
| 15:22 | <gsnedders> | Peh. |
| 15:22 | <gsnedders> | I'm wrong. |
| 15:22 | <Philip`> | and conformingness to the XMLNS spec is defined in terms of whether something is well-formed or not |
| 15:22 | <gsnedders> | I'm a terrible browser QA guy. |
| 15:22 | <gsnedders> | Philip`: You should replace me. |
| 15:23 | <Philip`> | s/gsnedders/???/ |
| 15:23 | Philip` | can't work out what to replace him with |
| 15:24 | <Dashiva> | Philip`: I think yourself |
| 15:31 | <jgraham> | Er, I hate to tell you this, but gsnedders just disappeared, and there are now three question marks asking me how to set up their computer |
| 15:49 | <TabAtkins> | Argh, how do I keep typing </fielset> over and over and over again? |
| 15:49 | <TabAtkins> | That should be a valid closing tag for <fieldset> just to fix my screwups. |
| 15:54 | <Philip`> | TabAtkins: That would be even more evil than <image> |
| 15:55 | <TabAtkins> | But it would mean I wouldn't have to make a 2-second fix every time I write a page with a form! |
| 15:56 | <Dashiva> | Or you could just not use fieldsets, use divs instead! |
| 15:56 | <TabAtkins> | But then I'd need to put in extra styles! This is clearly suboptimal. |
| 15:57 | <Dashiva> | Nobody uses default fieldset styling, surely |
| 15:57 | <TabAtkins> | Well, no, not in public-facing sites. |
| 15:57 | <TabAtkins> | Or most of my private pages. |
| 15:57 | <TabAtkins> | But my throwaway scripts do! |
| 15:57 | <Dashiva> | I don't think you should care about the look of your throwaway scripts, that will make you less likely to actually throw them away |
| 16:05 | <nesta> | does anony know if there is documentation on the javascript API to the <video> element (in safari/iPhone) ? -- e.g. how to get buffer status, play/pause, position, video duration, etc. etc. |
| 16:06 | <annevk> | the spec? |
| 16:06 | <TabAtkins> | The spec. |
| 16:06 | <nesta> | the spec |
| 16:07 | <nesta> | I haven't seen any dev documentation for controlling video |
| 16:07 | <TabAtkins> | The spec explains all the functions involved in the <video> API. |
| 16:07 | <annevk> | look for the green boxes |
| 16:07 | <hsivonen> | the spec doesn't say which parts of the spec safari implements, though |
| 16:07 | <annevk> | they give dev documentation |
| 16:07 | <TabAtkins> | Ok, that's true. |
| 16:08 | <nesta> | aye -- but they're probably following it |
| 16:08 | <nesta> | ? :) |
| 16:08 | <annevk> | yes |
| 16:08 | <nesta> | can you link to the URL? |
| 16:09 | <nesta> | http://www.whatwg.org/specs/web-apps/current-work/multipage/ ? |
| 16:09 | <nesta> | thankfully it works in Chrome |
| 16:10 | <TabAtkins> | Yup. |
| 16:10 | <TabAtkins> | Specifically, start here: http://www.whatwg.org/specs/web-apps/current-work/multipage/video.html#media-elements |
| 16:10 | <TabAtkins> | Everything below that point is talking about the apis. |
| 16:10 | <nesta> | TabAtkins: thanks -- it's making sense now |
| 16:11 | <TabAtkins> | Cool. |
| 18:08 | Philip` | likes how GStreamer splits its plugin packages into "good", "bad" and "ugly" |
| 18:11 | <Philip`> | (It'd be nice if they explained what the difference was, though) |
| 18:12 | <Philip`> | ((Looks like ugly is like good but with patent issues, and bad is like good+ugly but with quality issues)) |
| 18:13 | <Philip`> | http://gstreamer.freedesktop.org/documentation/splitup.html - oh, they do explain it |
| 20:17 | <annevk> | Hixie, fyi, document.head is in the spec |
| 20:23 | <Hixie> | it is? |
| 20:23 | <Hixie> | woo |
| 20:23 | <Hixie> | when did i do that |
| 20:23 | <Hixie> | what was the bug # again? |
| 20:25 | <Hixie> | got it |
| 20:25 | <Hixie> | commented |
| 20:35 | <Hixie> | hsivonen: yt? |
| 20:35 | <Hixie> | hsivonen: http://www.w3.org/Bugs/Public/show_bug.cgi?id=8375 |
| 20:35 | <Hixie> | hsivonen: is the last paragraph the only request at this point? |
| 20:44 | <TabAtkins> | Is Silverlight standard in IE now? |
| 20:45 | <TabAtkins> | Wondering about that new post in whatwg about the theora streamer leveraging silverlight3. |
| 20:45 | <TabAtkins> | If that's fairly dependable, then fuck yeah. |
| 20:45 | <AryehGregor> | Did David Gerard post that one too? |
| 20:46 | <AryehGregor> | He posted one to wikitech-l. |
| 20:46 | <TabAtkins> | Yes. |
| 20:46 | <AryehGregor> | Cortado is already established and will work much more reliably cross-platform. |
| 20:46 | <TabAtkins> | That's a java applet, though. |
| 20:47 | <AryehGregor> | Yes, and? |
| 20:47 | <AryehGregor> | That's why it works much more reliably cross-platform. |
| 20:47 | <TabAtkins> | So, I don't know what the penetration numbers are on silverlight vs java runtime. |
| 20:47 | <TabAtkins> | Within the IE-using community, specifically. |
| 20:50 | <TabAtkins> | I mean, if I can hit IE with this, then I can just deploy <video> with theora and ignore Safari until they shape up. |
| 20:50 | <Philip`> | I'm fairly sure it's not installed by default |
| 20:51 | <AryehGregor> | I'm pretty sure Java is more widely installed than Silverlight even on IE/Windows, but I haven't checked. |
| 20:52 | <Philip`> | (It's shown as an optional update whenever you run Windows Update, but I've never seen it get installed without me performing some explicit action) |
| 20:52 | <TabAtkins> | I wonder how many people have installed it already as a plugin prompt on major websites, though? |
| 20:52 | <TabAtkins> | I know that, frex, Netflix runs on Silverlight. |
| 20:54 | Philip` | installed it to try watching videos on itv.com |
| 20:55 | <Philip`> | before discovering that get_iplayer could download its videos as nice simple .asf files that work in mplayer and can run at fullscreen without dropping frames |
| 20:55 | <Philip`> | and without the unskippable adverts |
| 20:55 | <Steve^> | Did someone say silverlight3? |
| 20:55 | <Steve^> | I think Mono barely supports 1 |
| 20:55 | <TabAtkins> | Indeed. |
| 20:56 | <Philip`> | and before deciding that the video quality was barely watchable |
| 20:56 | <TabAtkins> | Shrug. Use FF or Opera and you don't need Silverlight. |
| 20:56 | <Philip`> | and before realising that they didn't have any content I wanted to watch anyway |
| 20:56 | <TabAtkins> | I'd just be using it to serve IE people. |