| 03:26 | <sephr> | roc: please don't tell me that that SVG in img patch is in firefox 8 stable |
| 03:28 | <roc> | the patch that breaks blob URIs? |
| 03:28 | <roc> | it's being backed out |
| 03:29 | <sephr> | I haven't even looked at the patch; I was just under the impression that it broke them from what you said |
| 03:29 | <sephr> | alright thanks |
| 03:29 | <sephr> | roc: speaking of ff8 changes, what's with the ugly 'no favicon' favicon? |
| 03:30 | <roc> | no idea |
| 03:40 | <jarek> | why W3C validator marks "@media all and (orientation: portrait) {}" as invalid? |
| 03:41 | <jarek> | I'm getting "portrait is not a orientation value ) {}" error |
| 03:41 | <jarek> | is this a bug in the validator? |
| 06:47 | <asmodai> | *sigh* that's awesome. 7.0.1 FF browser suddenly hangs on opening all my tabs on startup after upgrading to either 7.0.2/8.0 |
| 06:47 | <asmodai> | Time to get debugging again |
| 06:48 | <asmodai> | Consistent hang though, 13% CPU (Core i7) and ~320 MB |
| 08:31 | <asmodai> | Oh wow, http://www.zdnet.com/blog/perlow/exclusive-adobe-ceases-development-on-mobile-browser-flash-refocuses-efforts-on-html5/19226 |
| 08:35 | <Ms2ger> | Philip`, feel like doing some explaining in http://www.w3.org/Bugs/Public/show_bug.cgi?id=14421 ? |
| 08:59 | <hsivonen> | I wonder what the current thinking about X-OperaMini-Phone* headers is in the Opera Mini team these days. Good idea in retrospect? Bad idea in retrospect? |
| 10:09 | <Areks> | hello all |
| 10:09 | <Areks> | does ie8 support canvas or svg? |
| 10:19 | <jarek__> | Areks: http://caniuse.com/#search=canvas http://caniuse.com/#search=svg |
| 10:50 | zcorpan | removes his mpilgrim feeds |
| 10:51 | zcorpan | also removes ln.hixie.ch |
| 13:27 | <asmodai> | What a news day, both Flash and Silverlight seemingly going the way of the dodo. |
| 13:29 | <smaug____> | oh, also silverlight |
| 13:29 | <smaug____> | good good |
| 13:32 | <asmodai> | http://www.theverge.com/2011/11/9/2548975/microsoft-may-halt-development-work-on-silverlight-after-next-release |
| 13:32 | <smaug____> | if we could add NaCl to that list, then the day would be really good |
| 13:32 | <asmodai> | Was not that unanticipated after the WinRT announcement |
| 13:33 | <zcorpan> | can someone explain "entry script" and give illustrative examples? :) |
| 13:39 | <asmodai> | having no context I could imagine entry script being the first script that runs and acts like a bootstrapper? :) |
| 13:40 | jgraham | thinks illustrated examples would be more fun than illustrative examples |
| 13:40 | <Philip`> | smaug____: Do you know if anyone has started using NaCl in real life yet? |
| 13:41 | <Philip`> | (Anyone other than Google, in particular) |
| 13:41 | <asmodai> | Philip`: I know some gamedevs investigating it |
| 13:41 | <asmodai> | Philip`: and working on code for it |
| 13:42 | <jgraham> | zcorpan: Isn't it just the script associated with the topmost item in the callstack? |
| 13:43 | <jgraham> | (assuming a downward growing stack) |
| 13:43 | <asmodai> | Heh, acquaintance complaining about canvas performance: http://starlon.lyrical.net/canvas_spiney.php |
| 13:45 | <nlogax> | i get ~1 fps on a quad-core i7. |
| 13:46 | <asmodai> | Heh |
| 13:46 | <asmodai> | that averages what some of us see |
| 13:46 | <asmodai> | 0.3 fps to 2 fps depending on browser |
| 13:47 | <Philip`> | That doesn't exactly look like an efficient implementation of that effect |
| 13:48 | <asmodai> | Suggestions welcome, i'll let him know ^^ |
| 13:51 | <zcorpan> | jgraham: what's the entry script when script A document.writes script B, for instance? |
| 13:52 | <Philip`> | asmodai: For a start, don't blindly port code that's optimised for C :-) |
| 13:53 | <jgraham> | zcorpan: When A is running it is A. When B is running it is B |
| 13:53 | <Philip`> | There's a load of stuff like of[ofptr + 0]=DIV_2(f[fptr + 0]) + DIV_8(f[fptr + 0]) + DIV_8(f[fptr + 1]) + DIV_8(f[fptr + -1]) + DIV_8(f2[f2ptr + 0]) + adj_tl2; |
| 13:53 | <Philip`> | which is assuming array lookups are free, and assuming DIV_2 is a macro and free, etc |
| 13:54 | <Philip`> | Better to understand what it's doing algorithmically and then work out how to implement that as sensible efficient JS |
| 13:54 | <jgraham> | zcorpan: At least AIUI. Because the script ends up in "creating scripts" which ends up in "calling scripts" which always changes the entry point |
| 13:55 | <zcorpan> | jgraham: ok. thanks |
| 13:58 | <asmodai> | Philip`: *nod* |
| 13:58 | <asmodai> | he did mention: |
| 13:59 | <asmodai> | << starlon>> Should be accessing the image data directly for one. |
| 13:59 | <smaug____> | asmodai: that code could perhaps use typed arrays (at least when those are available) |
| 14:01 | <asmodai> | It's funny because people, not this guy, are whining about the state of HTML5 and taking over Flash features while it's not ready for it, but a lot of it is a bit catch-22, the moment Flash drops and people look at HTML5 it will drive HTML5 forward faster |
| 14:01 | <Philip`> | asmodai: I'd hypothesise (with no attempt at profiling the code or looking in any real detail) that the part that interacts with the canvas is fast (since it seems to be basically just putImageData), and it's all the JS code doing computations that is slow |
| 14:01 | <asmodai> | Philip`: That could very well be. |
| 14:01 | <jgraham> | I would hpothesise, without looking at anything, that there is no reason that the javascript should be slow for this |
| 14:02 | <asmodai> | jgraham: Haha |
| 14:02 | <jgraham> | if it is written in a way that works well |
| 14:02 | <jgraham> | (which it isn't at present) |
| 14:04 | <smaug____> | jgraham: btw, does Opera support typed arrays? |
| 14:05 | <Philip`> | I'd assume there's no fundamental reason it should necessarily be more than like 10x slower than C, if it's optimised for JS |
| 14:06 | <smaug____> | 10x sounds quite a lot |
| 14:06 | <jgraham> | smaug____: Yes, in the snapshots |
| 14:06 | <jgraham> | Although I am not 100% sure how optimised they are yet |
| 14:07 | <asmodai> | http://blog.illyriad.co.uk/index.php/2011/11/webgl-experiments-illyriads-3d-town/ |
| 14:23 | <hsivonen> | Looks like there are Android fanbois who became Flash fanbois because Android had Flash but iOS didn't |
| 14:24 | <hsivonen> | I'm making inferences from Android Central comments on Firefox 8 and the news about Adobe no longer developing Flash Player for mobile |
| 14:29 | <asmodai> | That's weird. |
| 14:29 | <asmodai> | I am a huge Android enjoyer, but I would not mind to see Flash go bye bye |
| 14:34 | <zcorpan> | i haven't had a single case of "hmm i need to install flash to make this work" on my phone |
| 14:36 | <asmodai> | Watching Khan Academy videos on my tablet does :( |
| 14:38 | <hsivonen> | I occasionally have to copy the page URL from Firefox to Opera Mobile on Android, because there's a video wrapped in swf |
| 14:39 | <jgraham> | zcorpan: I have needed it to view a resturant website on someone else's phone, I think |
| 14:40 | <hsivonen> | restaurant sites... what's possessing them to use swf and PDF? |
| 14:43 | <Philip`> | Maybe it's just seen as an extension of printed leaflets produced by their graphic designers, and they don't think they need to bother finding a web developer to get involved |
| 14:44 | <asmodai> | Happens a lot on websites. They convert their weekly product bonus folder to a Flash clickable version >_< |
| 14:47 | <smaug____> | hsivonen: although it was an email attachment, I was very surprised when I got that one pdf news letter yesterday to my moco email |
| 15:01 | <AryehGregor> | I don't suppose there's any free crippleware Word suitable for testing but not actual use, is there? |
| 15:03 | <jgraham> | Libreoffice? :p |
| 15:03 | <AryehGregor> | I have that, but it would be nice if I could test on Word too. |
| 15:03 | <jgraham> | (yeah, I know that is not sutiable for you :) |
| 15:11 | <AryehGregor> | I can test in Google Docs, though. |
| 15:12 | <AryehGregor> | That's another data point. |
| 15:17 | <hsivonen> | AryehGregor: doesn't Google Docs depend on what contenteditable does? |
| 15:17 | <AryehGregor> | hsivonen, no, it doesn't use contenteditable at all. |
| 15:18 | <AryehGregor> | No one in their right mind uses contenteditable for anything, it's a complete trainwreck. |
| 15:18 | <AryehGregor> | Well, not exactly. |
| 15:18 | <AryehGregor> | People without the resources of Google Docs are forced to use contenteditable, they just don't use execCommand(). |
| 15:18 | <AryehGregor> | Some poor suckers do use execCommand(), but anyone sensible uses a library like TinyMCE or such. |
| 15:19 | <AryehGregor> | But anyway, no, currently Google Docs doesn't use contenteditable at all, and its data model isn't even really HTML. |
| 15:19 | <AryehGregor> | But in any event, what I'm testing right now (auto-linking when you type a URL) is independent of contenteditable, except in IE, which I'm testing by itself. |
| 15:19 | <hsivonen> | AryehGregor: what does Google Docs use? |
| 15:20 | <AryehGregor> | Div soup and lots and lots of JavaScript. |
| 15:20 | <AryehGregor> | I mean, they do use HTML for rendering, not canvas or something. |
| 15:21 | <jgraham> | (Docs does actually use contenteditable to get IME support) |
| 15:21 | <hsivonen> | jgraham: I was about to ask about that next |
| 15:21 | <jgraham> | (but they create a contenteditable div and then delete it again, or something) |
| 15:21 | <AryehGregor> | Probably only because they haven't finished writing their own IME yet, anyway. |
| 15:21 | <jgraham> | Well the platform doesn't really support it yet |
| 15:21 | <jgraham> | They are pushing for it though :) |
| 15:21 | AryehGregor | doesn't have much of any inside info about this, was just given a brief summary by Annie Sullivan at a Toronto meetup a while back |
| 15:22 | <AryehGregor> | jgraham, why can't you make one up from scratch in JS? |
| 15:22 | jgraham | only knows what was said at TPAC |
| 15:22 | <jgraham> | AryehGregor: I'm not sure |
| 15:22 | <jgraham> | Maybe the idea is that you should use the system one? |
| 15:23 | <jgraham> | Because different users will have different configurations |
| 15:23 | <jgraham> | and having it work differently on one website would be a problem |
| 15:24 | <AryehGregor> | One might think that's true for everything else, too. |
| 15:24 | <AryehGregor> | Google Doc constantly annoys me by behaving differently from normal programs when I do things like Ctrl-Backspace. |
| 15:25 | <jgraham> | Well I would be pretty surprised if I typed "hello" and docs inserted "world" |
| 15:25 | <jgraham> | (rather than "hello") |
| 15:26 | <jgraham> | Moreso than if editing commands worked differently |
| 15:27 | <AryehGregor> | I don't think IMEs quite work that way. At least for CJK, they show you what they're going to input and you have to hit space to let it. Although perhaps for common words, users remember the right sequence and don't look because they know what they have to type for the first choice to be correct. |
| 15:27 | <AryehGregor> | I've seen them used, but never tried it myself. (Not that I know any CJK language.) |
| 15:28 | <jgraham> | My imagination is that learning a new IME is roughly like learning a new keyboard layout |
| 15:28 | <jgraham> | But I haven't used one either |
| 15:31 | <AryehGregor> | I picture it as more like smartphone autosuggest, except there's no option to just keep what you typed. |
| 15:35 | <AryehGregor> | To be fair, LibreOffice also often ignores platform conventions. |
| 15:39 | <hsivonen> | document.body.outerHTML = "<body>foo</body>"; |
| 15:39 | <hsivonen> | only pre-Ragnarök Opera works the way authors might expect |
| 15:40 | <hsivonen> | Chrome and Ragnarök work per spec |
| 15:40 | <hsivonen> | IE9 works yet differently |
| 15:40 | <hsivonen> | I guess I'll just make Firefox work per spec like Ragnarök and Chrome |
| 15:41 | <AryehGregor> | What does spec say? |
| 15:41 | <hsivonen> | AryehGregor: you get two heads |
| 15:41 | <AryehGregor> | o_O |
| 15:41 | <hsivonen> | AryehGregor: zero heads in IE9 |
| 15:42 | <AryehGregor> | Why should it affect the head at all? |
| 15:42 | <hsivonen> | one head in pre-Ragnarök Opera |
| 15:42 | <hsivonen> | AryehGregor: parsing stuff with html as the context node always generates a head and a body |
| 15:43 | <hsivonen> | correction: zero bodies in IE9 (one head) |
| 15:45 | <jgraham> | Two heads are better than one! |
| 15:48 | <annevk5> | talking about body and head elements as heads and bodies is weird |
| 15:48 | <annevk5> | we're not hitman |
| 15:49 | <dglazkov> | good morning, Whatwg! |
| 15:50 | <jgraham> | dglazkov: Damn, I obviously needed to spend more time with you to dull the edge of your optimism |
| 15:55 | <hsivonen> | jgraham, annevk5: do you happen to know why Opera Mobile says "Opera Mobi" in its UA string? |
| 15:55 | <hsivonen> | Is it avoiding the substring "Mobile" or the substring "Opera Mobile"? |
| 15:55 | <jgraham> | hsivonen: No idea |
| 15:55 | <jgraham> | I might be able to ask someone |
| 15:56 | <jgraham> | If you rally care |
| 15:56 | <hsivonen> | jgraham: I do really care, so it would be very nice if you could ask |
| 15:59 | annevk5 | leaves it to jgraham |
| 15:59 | <annevk5> | UA strings make me sad |
| 18:33 | <AryehGregor> | So Gecko throws on node.childNodes[-1], it seems. |
| 18:36 | <AryehGregor> | document.childNodes[-1] is undefined in IE9, Chrome 16 dev, Opera 12.00. Throws in Firefox 9.0a2. |
| 18:36 | <AryehGregor> | Spec seems to say it should be null. |
| 18:37 | <AryehGregor> | document.childNodes.item(-1) is null in IE9, Firefox 9.0a2, and Opera 12.00. Throws in Chrome 16 dev. Spec seems to say it should be null. |
| 18:37 | <AryehGregor> | Sigh. |
| 18:38 | <AryehGregor> | WebIDL doesn't permit a getter method to return something different from actually indexing, right? |
| 18:38 | <erlehmann> | argl |
| 18:44 | <AryehGregor> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=14743 |
| 19:05 | <AryehGregor> | Is there a way in Firebug to break on all exceptions, like in WebKit's Web Inspector? |
| 19:14 | <AryehGregor> | Blech, we really need to standardize HTML serialization properly. Gecko and WebKit behave differently in all sorts of ways. |
| 19:14 | <AryehGregor> | hsivonen, do you know what the story is with HTML serialization? Has anyone tried implementing the spec? |
| 19:14 | AryehGregor | notices that Gecko URL-encodes <a href=""> in serialization |
| 19:19 | <annevk5> | you have to pay for Growl now? |
| 19:32 | <wilhelm> | Do any of the search engines actively use the metadata described on schema.org yet? |
| 19:32 | <AryehGregor> | Almost certainly. |
| 19:32 | <AryehGregor> | Why would they make the project in the first place if they didn't intend to use it? |
| 19:35 | <wilhelm> | That's known to happen in this industry. (c: |
| 19:43 | <hober> | annevk5: made it to the airport ok? |
| 19:49 | <Hixie> | AryehGregor: document.childNodes[-1] is undefined in in the spec |
| 19:49 | <AryehGregor> | Hixie, really? How so? |
| 19:50 | <Hixie> | how would it be anything else? |
| 19:50 | <AryehGregor> | Hixie, in DOM4, NodeList has a getter, item(). item() is defined to return null for out-of-range indices. |
| 19:50 | <Hixie> | yes, item(-1) definitely returns null |
| 19:50 | <Hixie> | that's a method call |
| 19:51 | <Hixie> | but there's no -1 property |
| 19:51 | <Hixie> | on the object |
| 19:52 | <Hixie> | so you never even get to webidl, you return undefined all the way back in the JS spec |
| 19:52 | <Hixie> | certainly you don't get as far as invoking item() |
| 19:52 | <AryehGregor> | Hmm. |
| 19:52 | AryehGregor | rereads the relevant bit of WebIDL |
| 19:53 | <AryehGregor> | "If an indexed property getter was specified using an operation with an identifier, then the value returned when indexing the object with a given index is the value that would be returned by invoking the operation, passing the index as its only argument." |
| 19:53 | <AryehGregor> | It doesn't say that only holds for supported property indices. |
| 19:53 | <AryehGregor> | Maybe it should. |
| 19:53 | <AryehGregor> | Otherwise, I don't see where the concept is used. |
| 19:54 | <AryehGregor> | Maybe I should look at the ES binding? |
| 19:54 | <AryehGregor> | (Maybe we should stop pretending WebIDL is language-neutral?) |
| 19:55 | <Hixie> | well you should definitely also be looking at the ES binding, certainly |
| 19:55 | <Hixie> | but also the JS spec |
| 19:56 | <AryehGregor> | I don't see where ES comes into play. If WebIDL says what happens when you index the object, it overrides ES. |
| 19:57 | <Hixie> | if WebIDL overrides JS then it overrides the parts of JS it overrides, sure. But it doesn't override everything. |
| 19:57 | <AryehGregor> | Hmm, I think this bit implies that every object supporting indexed properties will appear to have all numeric properties from 0 to 2^32 - 2: http://dev.w3.org/2006/webapi/WebIDL/#dfn-array-index-property-name |
| 19:59 | <Hixie> | 4.7.2 is the algorithm JS calls |
| 20:00 | <Hixie> | and that algorithm, if the index isn't supported, defers to JS |
| 20:00 | <Hixie> | which then returns undefined |
| 20:00 | <AryehGregor> | Hmm, okay. |
| 20:00 | <AryehGregor> | That's confusing. |
| 20:00 | <AryehGregor> | I'll file some WebIDL bugs. |
| 20:00 | <Hixie> | WebIDL doesn't really override JS at all here, it just uses the hooks JS provides |
| 20:08 | <AryehGregor> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=14746 |
| 20:47 | <Ms2ger> | AryehGregor, does document.childNodes[-1] still throw in Nightly? |
| 20:47 | <AryehGregor> | Ms2ger, didn't test, only tested in Aurora. |
| 20:47 | AryehGregor | looks for Nightly |
| 20:47 | <Ms2ger> | You're probably still on the old bindings, then |
| 20:48 | <AryehGregor> | "old bindings"? |
| 20:48 | <smaug____> | I thought Aurora has new bindings |
| 20:48 | <smaug____> | for nodelists |
| 20:48 | <Ms2ger> | smaug____, but has there been an update already? |
| 20:48 | AryehGregor | finds https://launchpad.net/~ubuntu-mozilla-daily/+archive/ppa |
| 20:49 | <smaug____> | Ms2ger: ah, true |
| 20:49 | <smaug____> | yeah, AryehGregor, try Nightly |
| 20:49 | <Ms2ger> | AryehGregor, DOM bindings, we're experimenting with a new approach on nodelists |
| 20:49 | <Ms2ger> | See https://bugzilla.mozilla.org/show_bug.cgi?id=648801 |
| 20:50 | <AryehGregor> | I see, I had the repo installed but needed to install the firefox-trunk package. |
| 20:53 | <AryehGregor> | Ah, now it works. |
| 20:53 | <AryehGregor> | Nice. |
| 20:53 | <AryehGregor> | Thanks. |
| 20:53 | <Ms2ger> | Np :) |
| 21:13 | <annevk5> | hober: yeah all good |
| 21:13 | <annevk5> | hober: fly in an hour or so, was way on time :) |
| 21:18 | <wilhelm> | I thought I was running late when flying out on Sunday, but the airport had just forgot to adjust their clocks. :P |
| 21:22 | <annevk5> | ooh, that would be really annoying |
| 21:26 | <hober> | cool |
| 21:40 | <annevk5> | TabAtkins: fwiw, TR/html5/ uses the script thingie to close the notice |
| 21:40 | <annevk5> | TabAtkins: seems like the path of least resistance |