| 01:57 | <TabAtkins> | Can anyone explain this crazytime behavior? http://jsfiddle.net/leaverou/TwgMY/ |
| 01:58 | <TabAtkins> | It's exactly reproducible in webkit, ff, and opera. |
| 01:58 | <jamesr> | what's the problem? |
| 01:59 | <TabAtkins> | If you set fillStyle to an rgba() value and draw into the canvas, then read the color of a drawn pixel back out, you get a slightly different value. |
| 01:59 | <jamesr_> | TabAtkins, premultiplied alpha baby |
| 01:59 | <jamesr_> | unpremultiplying is lossy |
| 01:59 | <TabAtkins> | Oh! |
| 01:59 | <TabAtkins> | That makes sense. |
| 02:06 | Philip` | would rather say that premultiplying is lossy |
| 02:06 | <Philip`> | since that's where components get quantised more coarsely |
| 02:07 | <jcranmer> | well, either way you lose values in the ulp |
| 02:09 | <jamesr> | Philip`: correct |
| 02:09 | <jamesr> | unpremultiplying is something that happens after data loss occurs, so it's inexact |
| 04:29 | <smaug____> | has there been any discussion about Dart here? (and its evilness, and breaking the web etc.) |
| 04:30 | <TabAtkins> | I'd be happy to talk with you about it tomorrow morning. |
| 04:30 | TabAtkins | is going home now. |
| 09:11 | <annevk> | MikeSmith, you around? |
| 09:11 | <annevk> | or maybe shepazu? |
| 09:11 | <MikeSmith> | I'm here |
| 09:11 | <annevk> | I wonder whether I can get http://dvcs.w3.org/hg/url |
| 09:12 | <annevk> | I'll move abarth's Google Docs work there and hopefully inspire him to do some more work on it, but this time in HTML |
| 09:12 | <MikeSmith> | I can set it up now |
| 09:12 | <annevk> | And as something we can publish down the line. Sweet! |
| 09:14 | <MikeSmith> | I notice we already have http://dvcs.w3.org/hg/URLAPI/ |
| 09:15 | <annevk> | ah okay |
| 09:15 | <annevk> | do I have access rights to that? |
| 09:15 | <annevk> | and can we maybe lowercase it? |
| 09:26 | <annevk> | Is :nth-match() really something people need a lot? |
| 09:26 | <annevk> | Seems like feature creep to me... |
| 09:31 | <annevk> | MikeSmith, still there? |
| 09:32 | <annevk> | oh you created url I see |
| 09:32 | <MikeSmith> | yup |
| 09:33 | <MikeSmith> | and I just tested pushing a change to it and it worked |
| 09:33 | <MikeSmith> | so should work for you as well |
| 09:33 | <MikeSmith> | ACLs on it are set so that anybody who's a member of the WebApps WG can write to it |
| 09:33 | <annevk> | sweet |
| 09:33 | <MikeSmith> | I added a README file |
| 09:33 | <MikeSmith> | feel free to delete it |
| 09:34 | <annevk> | add one to your beers anne owes me counter :p |
| 09:34 | <MikeSmith> | heh |
| 10:50 | <MikeSmith> | great, I just got a notification from Google saying that I can use "Michael(tm) Smith" as my name in Google+ |
| 10:51 | <espadrine> | MikeSmith: I am +1-ing that in my mind right now. |
| 10:51 | <MikeSmith> | and now the UI is not letting me even edit it |
| 10:52 | <MikeSmith> | wonderful |
| 10:52 | <MikeSmith> | anyway, what I plan to do is to remove the (tm) and then just add it back again later |
| 10:55 | <MikeSmith> | "We understand that Google+ and its Names Policy may not be for everyone at this time." |
| 11:07 | <annevk> | boom |
| 11:07 | <annevk> | http://dvcs.w3.org/hg/url/raw-file/tip/Overview.html |
| 11:07 | <annevk> | I sort of think that specification should not just define the API, but also URLs in general... |
| 11:07 | <annevk> | I wonder if I can convince abarth of that |
| 11:08 | <annevk> | I mean everywhere else we have the model / syntax / API in one specification |
| 11:10 | <zcorpan> | annevk: why is the title "Progress Events"? |
| 11:13 | <annevk> | oops |
| 11:14 | <annevk> | copy pasta |
| 11:15 | <annevk> | fixed |
| 11:15 | <annevk> | guess I might clean it up a bit more at some point if abarth does not beat me to it |
| 11:18 | <hasather> | annevk: not sure I like the name "filename" since it's not necessarily a filename |
| 11:20 | <annevk> | I'm not the editor |
| 11:20 | <annevk> | email public-webapps |
| 12:32 | <Ms2ger> | annevk, right, I've dropped document.innerHTML for now; can add it back if there's interest |
| 13:01 | <Ms2ger> | # [14:30] <annevk> whoa, AryehGregor has made some crazy cool tests < Truth |
| 14:17 | <zcorpan> | Hixie: firefox's console says live dom viewer uses a few deprecated DOM features (like getAttributeNode) |
| 14:35 | <zcorpan> | annevk: hmm. http://www.google.com/codesearch#search/&q=hasFeature%5Cs*%5C(%5B%5E,%5D%2B,%5Cs*null%5Cs*%5C)&type=cs |
| 14:36 | <zcorpan> | annevk: maybe we need to change that one back |
| 14:45 | <annevk> | zcorpan, I see |
| 14:46 | <annevk> | Ms2ger, can I get access to Parsing and Serialization? |
| 14:46 | <Ms2ger> | Sure |
| 14:46 | <annevk> | Ms2ger, XMLHttpRequest relies on dom-innerHTML working for Document |
| 14:46 | <Ms2ger> | Oh, really? |
| 14:47 | <annevk> | See send() |
| 14:49 | <annevk> | Ms2ger, thanks, pushed my changesets |
| 14:50 | <zcorpan> | filed http://www.w3.org/Bugs/Public/show_bug.cgi?id=14184 |
| 14:51 | <annevk> | thanks zcorpan |
| 15:34 | <annevk> | Ms2ger, why does .hgtags only list two drafts? |
| 15:35 | <Ms2ger> | I was wondering that too |
| 16:03 | <annevk> | seems Native Client is out of the box |
| 16:04 | <annevk> | or should I say, ActiveG |
| 16:07 | <jgraham> | ? |
| 16:11 | <annevk> | a stable Chrome came out |
| 16:11 | <annevk> | http://chrome.blogspot.com/2011/09/new-stable-release-of-chrome-expanding.html?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+blogspot%2FEgta+%28Google+Chrome+Blog%29 |
| 16:11 | <annevk> | euh |
| 16:11 | <annevk> | http://chrome.blogspot.com/2011/09/new-stable-release-of-chrome-expanding.html |
| 16:11 | <annevk> | what is it with all this URL cruft these days |
| 16:11 | <annevk> | did people stop caring? |
| 16:13 | <Philip`> | Yay, now I can experience web applications that fail because they don't support my CPU |
| 16:13 | <Ms2ger> | Yet another reason to use Chrome |
| 16:13 | <jgraham> | As I non-chrome-user I can not experience web applications that fail because they don't support my CPU |
| 16:14 | <jgraham> | Is it possible to die of Google-demo starvation? |
| 16:14 | <jgraham> | If not I think I'll be OK |
| 16:15 | Philip` | wonders whether NaCl applications can rely on SSE/SSE2/etc instructions |
| 16:16 | <Philip`> | (If not then that sounds bad for performance which is basically the only point of NaCl; if so then it sounds bad for portability which is a significant point of the web) |
| 16:17 | Philip` | would like to rely on SSE2 for offline applications but collected some stats like http://zaynar.co.uk/0ad-pub/cpucaps.html and found that 1% of the application's users still don't support it 10 years after it was introduced :-( |
| 16:27 | <gsnedders> | Philip`: (No JS JIT in any shipped browser works on x86 without SSE2) |
| 16:28 | <gsnedders> | (But you probably knew that already) |
| 16:29 | <Philip`> | (SpiderMonkey's does) |
| 16:30 | <Philip`> | (at least as of some months ago) |
| 16:31 | <gsnedders> | Does it use x87 or just do all fp through C++? |
| 16:31 | <Philip`> | (and anyway you can run JS without a JIT, whereas you can't run NaCl without hardware support for whatever instructions the application uses) |
| 16:31 | <Philip`> | It uses x87 |
| 16:31 | <jcranmer> | shame |
| 16:32 | <gsnedders> | Philip`: But their regexp JIT won't, because the JSC stuff definitely doesn't support x87. |
| 16:33 | Philip` | knows since he encountered x87 JIT bugs in it a while ago |
| 16:33 | <gsnedders> | (Not that I can think of any reason for regexp to use floats at all) |
| 16:34 | <Ms2ger> | Well, we don't care about JITting regexps, so that's fine |
| 16:34 | <gsnedders> | Ms2ger: How so? |
| 16:34 | <Philip`> | Surely it'd be a better use of engineering effort to just hard-code the regexps that benchmarks use |
| 16:35 | <gsnedders> | (We have x87 support, though never shipped as it's a bit too buggy and fixing those bugs has never really been a priority. We also have support for JIT with no FPU.) |
| 16:36 | <Ms2ger> | Opera is weird like that :) |
| 16:37 | <gsnedders> | Ms2ger: What? Not bug-fixing it? Or the no-FPU support? |
| 16:37 | <Ms2ger> | Supporting obscure platforms |
| 16:38 | <gsnedders> | Eh, still plenty of TVs and the like which have no FPU. |
| 16:38 | <Ms2ger> | Exactly |
| 16:38 | <jcranmer> | many ARM procs don't have an FPU |
| 16:39 | <gsnedders> | jcranmer: ARM not so much nowadays, MIPS on the other hand… |
| 16:48 | <gsnedders> | jcranmer: Also, e.g., the HTC Wilffire doesn't have VFP. |
| 16:49 | <gsnedders> | jcranmer: So there are some fairly popular devices without a FPU. |
| 16:52 | <zewt> | keep those crappy devices from having a usable browser, so they become less popular :P |
| 16:55 | <gsnedders> | zewt: The default WebKit-based browser has JIT, I'm fairly certain. |
| 16:55 | <gsnedders> | (wait, it's V8, it must have JIT) |
| 17:21 | <annevk> | omg the Chuck Testa meme |
| 17:21 | <annevk> | Aslan returns http://i.imgur.com/eaGWI.jpg |
| 17:22 | <annevk> | original: http://www.youtube.com/watch?v=LJP1DphOWPs |
| 17:36 | <annevk> | haha |
| 17:36 | <annevk> | http://www.ojaitaxidermy.com/ |
| 17:37 | <annevk> | stuff like this why the web needs to be preserved |
| 17:51 | <miketaylr> | that video is priceless |
| 18:03 | AryehGregor | finally figures out that the only decent solution to his problem is just to run gnome-panel as a taskbar in Unity too |
| 18:03 | AryehGregor | is more or less happy now |
| 18:03 | <AryehGregor> | Hope no one pinged me when I couldn't tell. |
| 18:04 | <zewt> | as a choice of name it shows particularly ... questionable judgement. heh |
| 18:04 | <zewt> | there are only 290 things called "unity" |
| 18:05 | <AryehGregor> | It baffles me how they think that not allowing you to see the window names anywhere when switching windows makes any sense at all. |
| 18:05 | <AryehGregor> | I only have two or three terminals open, usually, but how the heck am I supposed to tell them apart from little pictures when I hit Alt-Tab? They're terminals, they look the same. |
| 18:05 | <zewt> | i don't know anything about it, but my first impression from their screenshot is: a column of meaningless icons, which is Novice UI Failure #1 |
| 18:05 | <AryehGregor> | And they're fullscreen, so just bringing them all to the foreground doesn't help. |
| 18:06 | <AryehGregor> | This is not even talking about my 21 XChat windows. |
| 18:06 | <zewt> | win7 has a dumb popover thing that shows thumbnail screenshots when you hover over grouped taskbar items--it's utterly worthless, you can't distinguish application windows from a thumbnail |
| 18:06 | <AryehGregor> | No, the UI I actually like a lot. There are just one or two really annoying flaws. |
| 18:06 | <AryehGregor> | Well, sometimes you can. Websites you often can, for instance, if there are only a few. |
| 18:06 | <AryehGregor> | Terminals or IRC chats, not so much. |
| 18:06 | <zewt> | "you have 5 notepads open, here are five thumbnails with white boxes and blurry unreadable text, which one do you want" |
| 18:07 | <zewt> | even distinguishing browser windows doesn't work very well; you have to actually *look* at the thumbnail and puzzle out what it is, since it'll look different every time (which browser tab is active, where you're scrolled on the page) |
| 18:07 | <AryehGregor> | Also, the sidebar thing is buggy when I have two monitors and it's on the left side of the right monitor. |
| 18:07 | <zewt> | even icons are better than that, since at least you can learn what the icons are and select them without too much though |
| 18:07 | <AryehGregor> | Yeah, thumbnails are way less helpful than names. |
| 18:08 | <AryehGregor> | It would help if hitting the Windows key on my keyboard actually did what it was supposed t odo. |
| 18:08 | <AryehGregor> | to do. |
| 18:12 | <AryehGregor> | On the plus side, nouveau's experimental 3D acceleration seems to work fine for me. |
| 18:34 | <Ms2ger> | annevk, so should HTMLElement.style be defined in CSSOM or HTML? |
| 18:34 | <shepazu> | TabAtkins, :) |
| 19:06 | <annevk> | Ms2ger, prolly CSSOM |
| 19:06 | <annevk> | Ms2ger, but I need to find time to do CSSOM work so for now don't bother I guess |
| 19:09 | <Ms2ger> | Want a patch? |
| 19:16 | <AryehGregor> | What license are the HTML WG tests under? |
| 19:16 | <AryehGregor> | Ah, I see. |
| 19:16 | <AryehGregor> | http://www.w3.org/html/wg/wiki/Testing/Submission/ |
| 19:26 | <annevk> | Ms2ger, feel free to commit directly |
| 19:35 | <annevk> | but tell me about it if you do |
| 19:35 | <annevk> | cvs conflicts are the worst |
| 19:37 | <Ms2ger> | Looks like I'd need Bert to give me access |
| 20:15 | <AryehGregor> | I hate fontSize argh argh argh stab. |
| 20:16 | <AryehGregor> | Why couldn't fonts on the web have just used pt to start with as the standard unit and saved us all this pain? |
| 20:18 | <aho> | i prefer px |
| 20:18 | <gsnedders> | AryehGregor: Because that would've meant knowing the dpi of every monitor? |
| 20:18 | <jarek> | AryehGregor: how would you define 1pt? |
| 20:19 | <AryehGregor> | gsnedders, no, it would have meant making pt a fake unit that doesn't really mean pt except for printing, which is what happened in real life anyway. |
| 20:19 | <jarek> | AryehGregor: this unit makes no sense on screen |
| 20:19 | <AryehGregor> | jarek, for printing, as expected. For screens, make something up. |
| 20:19 | <AryehGregor> | Sure it does, properly defined. |
| 20:19 | <AryehGregor> | It's currently defined by CSS as some fixed multiple of a px. |
| 20:19 | <gsnedders> | AryehGregor: Then "pt" as you've described it isn't the standard unit, it's another unit with the same name. |
| 20:19 | <AryehGregor> | Which is in turn defined as "you know, whatever you think makes sense for your monitor". |
| 20:19 | <AryehGregor> | gsnedders, correct. |
| 20:19 | <jarek> | AryehGregor: if you had assumed that 1pt = 1px then there would be no point in having pt unit at all :P |
| 20:20 | <gsnedders> | AryehGregor: Then what have you gained by using the standard unit? |
| 20:20 | <AryehGregor> | More specifically, I'm aggrieved that the editing command fontSize uses old-school values of 1-7 instead of sane values. |
| 20:20 | <AryehGregor> | gsnedders, it's familiar. And it will indeed be used as the standard unit for printing. |
| 20:20 | <AryehGregor> | Users know how big 10pt text is. |
| 20:21 | <AryehGregor> | Because it's what word processors use. |
| 20:21 | <aho> | i have absolutely no idea how big 10pt is :) |
| 20:21 | <aho> | pt = moonspeak |
| 20:21 | <gsnedders> | aho: Yeah, but that's what happens when you get a degree in music. |
| 20:22 | gsnedders | facepalms |
| 20:22 | <astearns> | 10pt is 0p10 picas. duh |
| 20:22 | gsnedders | mixes up aho and ato again |
| 20:23 | <jarek> | on screen 10pt could be anything (depending on which browser you use) |
| 20:23 | <jarek> | afair Firefox and Chrome use different functions for converting pt to px |
| 20:24 | <jarek> | shouldn't this be standarised by CSS? |
| 20:24 | <AryehGregor> | It is. |
| 20:24 | <astearns> | gsnedders: what's wrong with knowing what dpi you're drawing in? |
| 20:24 | <AryehGregor> | I think it's 1pt = 1.2px or something. |
| 20:24 | <AryehGregor> | It was decided a year or two ago, IIRC. |
| 20:25 | <astearns> | it's standardized for "normal" dpi, but allowed to be different for abnormal dpi |
| 20:25 | <gsnedders> | astearns: Ten years ago it wasn't massively easy to find out, AFAIK |
| 20:26 | <gsnedders> | astearns: It's allowed to be different? I thought it was defined to be a number of CSS pixels, though the relation between CSS pixels and device pixels is undefined. |
| 20:26 | astearns | digs for the reference |
| 20:26 | <aho> | speaking of which, i'd like to see media queries for sub pixel order & orientation :> |
| 20:26 | gsnedders | would be he's half-dead with a fever |
| 20:27 | <aho> | for subpixel rendered icons haha |
| 20:27 | <astearns> | the CSS reference pixel is defined as 96 dpi, and all the other length units (points, inches, mm) are defined by the reference pixel |
| 20:27 | <astearns> | http://www.w3.org/TR/CSS2/syndata.html#length-units |
| 20:28 | <astearns> | so making something 1 inch tall is also dependent on monitor resolution |
| 20:29 | <astearns> | if you're not at 96 dpi you can choose what number of device pixels best approximates the reference pixel |
| 20:29 | <aho> | px: pixel units — 1px is equal to 0.75pt |
| 20:29 | <aho> | ah... there it is |
| 20:29 | <jarek> | "The reference pixel is the visual angle of one pixel on a device with a pixel density of 96dpi and a distance from the reader of an arm's length. For a nominal arm's length of 28 inches, the visual angle is therefore about 0.0213 degrees. For reading at arm's length, 1px thus corresponds to about 0.26 mm (1/96 inch)." |
| 20:30 | <jarek> | I don't get it :/ |
| 20:31 | <astearns> | I understand why things were defined this way (for backwards compatibility) but making 1 inch not equal 1 inch is bad |
| 20:32 | <astearns> | jarek: it's a complicated way of saying "everyone's been assuming 96 dpi, so we're stuck with it" |
| 20:33 | <jarek> | btw, why rem unit is defined to be relative to the font size of top level element? |
| 20:34 | <jarek> | 99% of the websites specify base font size on body element, not on html element |
| 20:35 | <jarek> | no... actually this makes sense as SVG has no body element |
| 20:35 | <aho> | html doesn't necessarily have a body element |
| 20:36 | <aho> | or a head element for that matter |
| 20:36 | <aho> | going with the root node is the best bet |
| 20:38 | <jarek> | so specyfing font-size on body element is a bad practice, right? |
| 20:38 | <jarek> | it obviously makes rem unit useless |
| 20:38 | <aho> | rem is currently only supported in firefox, isn't it? |
| 20:39 | <jarek> | or is specyfing *any* base font size a bad practice? |
| 20:39 | <aho> | https://developer.mozilla.org/en/CSS/length |
| 20:39 | <aho> | so... rem does have some support these days |
| 20:40 | <jarek> | aho: it works on Chrome 14, though I haven't checked if it's accurate |
| 20:40 | <aho> | oh. ie9 also supports ch. nice :) |
| 20:41 | <aho> | <jarek> or is specyfing *any* base font size a bad practice? <- well, it's normalization. if everything needs to be as uniform as possible across different browsers, it's a thing you should do |
| 20:41 | <AryehGregor> | HTML will always have a <body> in the DOM if it was created by text/html. |
| 21:41 | <scheib> | Newbie question: I'm working on an API that will have a new event. It will be hookable via addEventListener("foo"). Should an implementation also support the older style of element.onfoo? |
| 21:47 | <AryehGregor> | scheib, this is for a spec? |
| 21:47 | <AryehGregor> | Generally we do support .onfoo for new events, AFAIK. |
| 21:49 | <scheib> | Well, the spec in question in "Mouse Lock", where I have not yet called out the 'onfoo' form of events specifically, but curious if I should and under what condition. http://goo.gl/9G8pd |
| 21:49 | <AryehGregor> | Ah, okay. |
| 23:00 | <AryehGregor> | scheib, try asking on a mailing list. It seems no one is around right now, probably because it's Friday afternoon. |
| 23:12 | <Hixie> | is anyone who cares about acid3 around? i have an update available for review. |
| 23:16 | <smaug____> | yes |
