| 01:35 | <Hixie> | foolip: yt? |
| 01:44 | <roc> | oh dear |
| 02:11 | <Hixie> | oh dear? |
| 02:13 | <boogyman> | hello ian |
| 02:13 | <Hixie> | hi |
| 02:55 | roc | was reading Larry Masinter's blog |
| 08:00 | <tiglionabbit> | are there any good examples on how to use the schema versioning features in html5 databases? The spec is kind of confusing me |
| 08:22 | <hsivonen> | roc: It's interesting to see the overspecification argument coming from an Adobe guy when Adobe's Flash bloggers extoll the virtues of having the predictability of a single codebase runtime |
| 08:23 | <roc> | clearly fostering competitive implementations isn't a goal there |
| 08:24 | <tiglionabbit> | where are the arguments to openDatabase specified? |
| 08:26 | <foolip> | Hixie: I'm here |
| 08:27 | <Hixie> | tiglionabbit: http://dev.w3.org/html5/webdatabase/ |
| 08:27 | <Hixie> | foolip: hey |
| 08:27 | <Hixie> | foolip: consider: |
| 08:27 | <Hixie> | <div itemscope itemtype="http://example.com/a"; itemref="x"></div> |
| 08:27 | <Hixie> | <div itemscope itemtype="http://example.com/b"; itemref="x"></div> |
| 08:27 | <Hixie> | <div id="x"> <p itemprop="q" itemscope> <span itemprop="r">s</span> </p> </div> |
| 08:28 | <Hixie> | right now we generate four blank nodes in the RDF conversion |
| 08:28 | <Hixie> | seems like we'd want to generate three |
| 08:28 | <Hixie> | but what properties should the third one have? |
| 08:28 | <foolip> | let me try it |
| 08:30 | <foolip> | Hixie: is this about what to do with type in subproperties, or something else? |
| 08:32 | <Hixie> | that's how i came to notice the four/three issue, yeah |
| 08:33 | <foolip> | 4 blank nodes: 2 for the items, and a duplicated one from the itemref'd item, yes? |
| 08:33 | <foolip> | you're thinking that we should merge these? |
| 08:33 | <roc> | hsivonen: Larry Masinter and others seem to have a fundamentally different view of the world than we do |
| 08:34 | zcorpan | uses <fcaption> on a site |
| 08:34 | <roc> | can't we at least call it figcaption? |
| 08:34 | <tiglionabbit> | Hixie: thanks. But what if you have more than one migration in there? |
| 08:35 | <Hixie> | tiglionabbit: how do you mean? |
| 08:35 | <Hixie> | foolip: yeah |
| 08:35 | <Hixie> | roc: i intend to call it figcaption |
| 08:35 | <Hixie> | roc: the issue of the <details> legend element isn't resolved though |
| 08:35 | <foolip> | Hixie: since in the currently specced algorithm the itemprop="q" won't generate any triples (because it has no itemtype), it doesn't matter, but I guess you mean if we give an itemtype we shouldn't duplicate the data? |
| 08:35 | <roc> | cool |
| 08:36 | <roc> | hsivonen: I kinda wish I fully understood that view and why they hold it |
| 08:36 | <roc> | but I don't have the time |
| 08:36 | <Hixie> | foolip: assume that we'll generate triples for it (your argument is compelling, though i'm not sure i like using the type for it) |
| 08:36 | <tiglionabbit> | Hixie: do you just put in multiple calls to changeVersion in that callback there, and assume non-appliccable ones will not be run? |
| 08:37 | <Hixie> | tiglionabbit: you're, like, actually making me read the spec :-P |
| 08:37 | <foolip> | Hixie: OK, as long as you can get the data out, I'm quite fine with whatever monster predicate is used :) |
| 08:38 | <tiglionabbit> | Hixie: aren't you the editor? |
| 08:38 | <Hixie> | tiglionabbit: yup |
| 08:38 | <foolip> | Hixie: it would be quite easy to let the RDF converter hold a list of blank nodes and the DOM nodes they were generated for, to simply reuse them if the same node generated more triples later |
| 08:38 | <Hixie> | tiglionabbit: i edit hundreds of pages of spec, i don't know them all well :-) |
| 08:39 | <foolip> | Hixie: that would kind of break the current layering on top of the |
| 08:39 | zcorpan | switches to <figcaption> |
| 08:39 | <Hixie> | tiglionabbit: ok it looks like the expected way to do multiple version changes is just to chain the changes, so e.g.: |
| 08:39 | <foolip> | ... "properties of an item" definition, but OK |
| 08:40 | <Hixie> | tiglionabbit: if you have 1->5, 2->5, 3->4, 4->5, you would changeVersion 1 to 5 with the error handler trying 2 to 5 with that error handler doing 3 to 4 and its success _and_ error handlers doing 4 to 5, and then have the succcess handlers for all of those except 3->4 and the error handler for 4->5 all be the smae "ok we're ready" handler |
| 08:41 | <Hixie> | foolip: yeah the RDF algorithm might need to be reworked to handle this |
| 08:41 | <foolip> | Hixie: do you have a specific idea for what predicates should be used for itemtypeless subitems? |
| 08:41 | <Hixie> | foolip: ...(using a cache or something) |
| 08:42 | <Hixie> | foolip: so originally i figured i'd just stick the property name ('adr', etc) after the : and before a %20 before the property names, but then i realised there could be multiple property names (itemprop="a b") |
| 08:42 | <tiglionabbit> | Hixie: gosh that's complicated. I thought this was going to be more like rails. I suppose that's more flexible though, maybe |
| 08:42 | <Hixie> | foolip: and then i realised there could be multiple types too |
| 08:43 | <Hixie> | tiglionabbit: the whole database spec is basically on hold at this point anyway because mozilla and microsoft won't implement it |
| 08:43 | <Hixie> | tiglionabbit: so there is little interest in really fixing problems like that at this point |
| 08:43 | <Hixie> | tiglionabbit: i agree that the model is suboptimal |
| 08:43 | <hsivonen> | roc: the single runtime view or the view that overspecification is anticompetitive? |
| 08:44 | <Hixie> | foolip: so i don't know what to use |
| 08:44 | <roc> | the whole thing |
| 08:44 | <Hixie> | btw it's clear that _over_specification is anticompetitive |
| 08:44 | <Hixie> | e.g. if we specced the finer details of UI, or performance characteristics |
| 08:44 | <roc> | I disagree |
| 08:44 | <roc> | because vendors could simply ignore that |
| 08:44 | <tiglionabbit> | Hixie: I really want to write an application that manages a lot of client side data, and it would be disastrous to lose any of it. What should I do, serialize it and shove it in local storage every time it changes? |
| 08:44 | <foolip> | Hixie: an item can only have one type (since you changed it) |
| 08:44 | <Hixie> | roc: well then it's not specified |
| 08:44 | <roc> | if it's not needed for interop, vendors can simply ignore the spec and no real harm is done |
| 08:45 | <Hixie> | foolip: yeah but an item can have multiple parents with different types |
| 08:45 | <hsivonen> | Hixie: I think the DOM should have specced the big-O characteristics of firstChild and NodeList.item() |
| 08:45 | <Hixie> | roc: wasting my time is "real harm" :-) |
| 08:45 | <roc> | ah |
| 08:45 | <foolip> | Hixie: oh, shouldn't that be invalid? |
| 08:45 | <Hixie> | foolip: the example i gave above is an example of that |
| 08:45 | <roc> | I don't think Larry was thinking of you here :-) |
| 08:45 | <foolip> | Hixie: ah, I see |
| 08:45 | <Hixie> | foolip: i suppose we could make it invalid but catching that case would be a huge pain |
| 08:46 | <Hixie> | foolip: worse even than loops |
| 08:46 | <Hixie> | still need to make those less strict btw |
| 08:46 | <Hixie> | havne't looked at that yet |
| 08:46 | <Hixie> | roc: i assume that "overspecification" is only an issue if the specification is going to be followed |
| 08:46 | <hsivonen> | (or, more to the point, the DOM should have promised O(1) for nextSibling) |
| 08:46 | <Hixie> | roc: obviously a work of fiction is irrelevant |
| 08:47 | <Hixie> | roc: and so can't be harmful per se (well other than to the spec writer, but they're a minor concern) |
| 08:47 | <foolip> | Hixie: I've implemented a new "properties of an item" which basically throws away all properties which would lead to loops back to itself, and nothing else (the algorithm also detects other loops and ignores those though) |
| 08:47 | <Hixie> | foolip: did you mail that in? |
| 08:47 | <foolip> | Hixie: not yet |
| 08:47 | <hsivonen> | Hixie: speaking of overspecification, I'm currently assuming that the story of initial about:blank not firing onload isn't going to fly anyhow, so I'm working on making it fire |
| 08:47 | <Hixie> | foolip: please do |
| 08:47 | <foolip> | Hixie: I will, don't worry :) |
| 08:47 | <Hixie> | hsivonen: please send feedback on that so i'll remember to fix it :-) |
| 08:48 | <foolip> | now work (<video>) |
| 08:48 | <Hixie> | foolip: thanks |
| 08:55 | <webben> | roc: I think I agree with you and Hixie. Maybe it might be more persuasive to people like Larry to hear from some representatives of the smaller HTML consumers they seek to protect. |
| 08:56 | <roc> | that's a good idea |
| 08:56 | <webben> | a Netsurf developer maybe and/or a small HTML editor developer. |
| 08:56 | <Hixie> | webben: if you can find out where larry is coming from and explain it to me (or us, I believe roc and others are in the same boat here), that would be really helpful, IMHO |
| 08:56 | <roc> | I'd actually quite like to meet Larry and talk to him |
| 08:56 | <roc> | sometimes that helps |
| 08:56 | <Hixie> | direct communication with larry has so far failed to enlighten me, and not understanding his position is definitely a problem |
| 08:57 | <Hixie> | roc: i've tried several times, unfortunately |
| 08:57 | <Hixie> | er |
| 08:57 | <Hixie> | roc: i've tried several times, _to no effect_, unfortunately |
| 08:57 | <roc> | :-) |
| 08:57 | <webben> | Hixie: I find his argument hard to follow. |
| 08:58 | <Hixie> | that has often been the difficulty i have faced in understanding his positions |
| 08:58 | <roc> | I think one problem is that being a "small implementer" of the whole HTML5 spec is not really feasible |
| 08:58 | <webben> | Hixie: I was more thinking of getting some representatives of the "296 remaining implementations" to give their thoughts. |
| 08:58 | <roc> | and that's just the way the Web is |
| 08:59 | <roc> | unfortunately |
| 08:59 | <roc> | I don't think there ARE 296 other implementations of stuff like Web-compatible implementations of the <img> DOM |
| 08:59 | <roc> | and there never will be |
| 08:59 | <webben> | There are more than 4 though ;) |
| 09:00 | <roc> | it would be interesting to know how many there actually are |
| 09:00 | <roc> | that are actively maintained |
| 09:00 | <roc> | I guess 10 |
| 09:01 | <webben> | I guess htmlunit, lobo are two beyond the big 4 |
| 09:01 | <hsivonen> | WebKit is killing the RIM browser, right? |
| 09:02 | <hsivonen> | is NetFront suited for browsing the Web these days? |
| 09:02 | <webben> | netsurf doesn't have js yet iirc so can't be used for that particular example |
| 09:02 | <hsivonen> | IceBrowser looks EOLed |
| 09:03 | <webben> | You know ... if the devs can be tracked down for these /dead/ UAs, they might also have useful ideas about the costs of under-specification... ;) |
| 09:03 | <Hixie> | i love how the ietf ignored my feedback on the hybi charter, announced the hybi charter with no mention of the whatwg, didn't respond to my further feedback on the hybi charter, and is now e-mailing the whatwg list to direct people to the hybi list to discuss websockets |
| 09:03 | <hsivonen> | last I looked, the Nokia browser on S40 was mainly suited for downloading Opera Mini... |
| 09:03 | <Hixie> | i also like how their charter has websockets going to LC in March 2011 |
| 09:04 | <hsivonen> | Hixie: what's there left to do anyway? |
| 09:04 | <Hixie> | that's a very good question |
| 09:04 | <roc> | stuff it up |
| 09:04 | <Hixie> | i do have some pending feedback, but nothing that'll take over a year |
| 09:05 | <Hixie> | maybe a week, or two |
| 09:05 | <Hixie> | then a test suite of course |
| 09:05 | <ukai_> | Hixie: hi |
| 09:05 | <Hixie> | but that's not on their timetable at all |
| 09:05 | <Hixie> | ukai_: hey, i have websockets feedback scheduled for next week |
| 09:06 | <Hixie> | ukai_: re httpOnly, seems reasonable to send them |
| 09:06 | <ukai_> | Hixie: i see |
| 09:08 | <ukai_> | Hixie: thanks |
| 12:38 | <hsivonen> | how is the readystatechange event supposed to work with documents in iframes? |
| 12:43 | Dashiva | wonders if there's any given rationale for Java lacking unsigned types, apart from that one interview where Gosling thinks it's too complicated for programmers to understand |
| 12:44 | <hsivonen> | If that's too complicated, good luck with Enterprise Java |
| 12:45 | <Dashiva> | I've googled and read all I found, but there's no rationale to see |
| 12:45 | <Dashiva> | Well, except "But it's not _impossible to do things without unsigned" |
| 12:46 | hsivonen | notes that people don't seem to think that Apple's closed platforms not having Java applets is a big deal |
| 12:46 | <hsivonen> | I guess Java applets are almost dead already |
| 12:47 | hsivonen | waves to people who haven't switched away from clueless banks that require Java to do banking |
| 12:47 | <Dashiva> | With WebGL coming, you won't need them to do opengl in the browser either |
| 12:47 | <Philip`> | At least Java gives you unsigned shorts |
| 12:47 | <Philip`> | though it confusingly calls them "char" |
| 12:47 | <hsivonen> | Philip`: bytes being signed is very annoying |
| 12:47 | <Dashiva> | They're not really unsigned shorts, since they print characters if you print them :P |
| 12:48 | <Philip`> | Cast them to int before printing :-) |
| 12:48 | <hsivonen> | Dashiva: does OpenGL in the browser via Java work reliably anywhere? |
| 12:48 | <Dashiva> | Then I might as well use & 0xFFFF |
| 12:48 | <Dashiva> | hsivonen: Reliably how? It requires a graphics card |
| 12:49 | <Dashiva> | I don't believe it's any less reliable than a regular java app doing it |
| 12:49 | <hsivonen> | Dashiva: who uses OpenGL via Java applets? is Java to OpenGL glue something that requires a separate install (on top of the JVM)? |
| 12:49 | <Dashiva> | No, it's java libs all the way down to JNI |
| 12:49 | <Dashiva> | JOGL or LWJGL |
| 12:50 | <Dashiva> | Aion (the MMO) uses an LWJGL applet to render the characters on the profile pages |
| 12:50 | <hsivonen> | so the user needs to permit JNI? |
| 12:50 | <Dashiva> | Yes |
| 12:51 | <hsivonen> | just like banking Java |
| 14:29 | <payman> | jamesl: i remember you deleted _a lot_ of test results from one of my previous spartan test runs, do you do that still? i have 2 new test runs running now. |
| 14:29 | <jgraham> | payman: Wrong window? |
| 14:29 | <payman> | jgraham: yes! |
| 16:49 | <Dashiva> | ins-alt-s is a strange shortcut... I wouldn't consider ins a meta key at all. |
| 16:49 | <Lachy> | Dashiva, is that for a screen reader? |
| 16:52 | <Dashiva> | Yes, JAWS |
| 16:54 | <Lachy> | I suppose their aim is to avoid picking shortcut keys that would clash with other application keys. Or maybe it's just because they have so many shortcuts, and so few meta keys to work with |
| 16:55 | <Dashiva> | That's true |
| 16:57 | <Dashiva> | No indication for bold and italic is pretty surprising, though, especially when e.g. JAWS supports it just fine |
| 17:03 | <Dashiva> | Argh... "I'll blog separately about the "reverse engineering" myth." |
| 17:04 | <Dashiva> | I'm sure there are a few QA departments out there who would like a word with Larry about that |
| 17:30 | <Lachy> | Dashiva, where is that quote from? |
| 17:30 | <daedb> | Lachy: It's from http://masinter.blogspot.com/2010/01/over-specification-is-anti-competitive.html |
| 17:30 | <Dashiva> | http://masinter.blogspot.com/2010/01/over-specification-is-anti-competitive.html |
| 17:34 | <othermaciej> | we're a myth! |
| 17:38 | MikeSmith | wonders who's talking |
| 17:38 | <MikeSmith> | ooops |
| 17:41 | <Dashiva> | Well, that's part of being a cabal |
| 17:41 | <Dashiva> | Denying your own existence |
| 17:43 | <MikeSmith> | the Honored Society |
| 17:44 | <MikeSmith> | you don't deny you exist, you say, We're just simple people, we grow lemons, we make olive oil. |
| 17:57 | <Dashiva> | Has krijnh been really disconnecty lately? |
| 17:57 | <Dashiva> | The logs seem very sparse |
| 18:01 | <krijnh> | Yes |
| 18:01 | <krijnh> | Got a new modem and stuff |
| 18:01 | <krijnh> | Sorry about that :( |
| 18:01 | <Dashiva> | Just wondering |
| 18:01 | <krijnh> | I'm making it up by using <video> and Theora a lot lately :) |
| 18:02 | <Dashiva> | I'm here to see it all, I only use the logs for seeing what people highlight |
| 18:03 | <MikeSmith> | I have a hard time figure out what's interesting in some of that parts that people highlight |
| 18:04 | <Dashiva> | Maybe we need a slashdot-type tag system |
| 18:04 | <krijnh> | Perhaps I should reveal which people highlight which lines |
| 18:04 | <Dashiva> | "Interesting", "Funny", etc |
| 18:04 | <Dashiva> | "Evil" |
| 18:04 | <Philip`> | Maybe a lot of people just click randomly on the page and don't notice that it's accidentally highlighting lines |
| 18:04 | <krijnh> | Maybe it should just be removed |
| 18:04 | <Dashiva> | I like it |
| 18:04 | <krijnh> | Could serve the logs a lot faster if I did :/ |
| 18:05 | <Dashiva> | But you could hash the IP address or something to generate a unique color for each marker |
| 18:06 | <Dashiva> | krijnh: You could always make the markers optional, so the default loads fast and without them :) |
| 18:08 | <krijnh> | Sure :) |
| 18:10 | <hsivonen> | krijnh: noooo. please don't remove highlights |
| 18:14 | <krijnh> | I won't! |
| 18:15 | <Philip`> | You should make highlights a Premium Subscription feature |
| 18:15 | <krijnh> | Moneyz! |
| 18:16 | <krijnh> | If only money could buy me more time each day :( |
| 18:16 | <krijnh> | By the way, perhaps of interest to anybody in here: http://fronteers.nl/congres/2009/sessions/a-web-of-confusion |
| 18:33 | <JonathanNeal> | Good afternoon! |
| 18:42 | <zcorpan> | good evening |
| 19:49 | jgraham | is quite pleased to find out he doesn't exist |
| 21:05 | <AryehGregor> | Wait, so Slashdot has a story on the FSF's condemnation of the iPad, but no actual story about the iPad's release? That's pretty pathetic even for Slashdot. |
| 21:07 | <Steve^> | I can imagine a million cool uses for that thing, few of which work under the iphone's oppressive state. So maybe they have a point |
| 21:07 | <Philip`> | What would be the point in having a story that every single other news outlet in the entire world already has at least two stories on? |
| 21:07 | <Steve^> | its not exactly hard to find an article about the ipad on another site |
| 21:08 | <Dashiva> | http://apple.slashdot.org/story/10/01/27/1849207/Apples-iPad-Out-In-the-Open |
| 21:08 | <Dashiva> | Isn't that an article about the iPad's release? |
| 21:08 | <AryehGregor> | Hmph. Why did I miss that? |
| 21:08 | <Dashiva> | Because people love having their expectations fulfilled |
| 21:08 | <Dashiva> | :P |
| 21:08 | <AryehGregor> | . . . okay, I admit to being an idiot, I probably have Apple stories not displaying on the front page. |
| 21:09 | <jgraham> | Philip`: What is the point of having the FSF story that was inevitable? |
| 21:09 | <Steve^> | for dummies like me that don't think that far enough ahead |
| 21:09 | <Dashiva> | jgraham: To keep the choir's preach-quota satisfied? |
| 21:10 | <jgraham> | Dashiva: Judging by the number of comments it worked rather well |
| 21:10 | jgraham | wonders if any of the comments are along the lines of "FSF condemns apple device. News at 11" |
| 21:11 | AryehGregor | concludes Slashdot is just broken, as usual. |
| 21:11 | AryehGregor | wonders what other stories he's mysteriously missing. |
| 21:11 | <Dashiva> | The "AryehGregor is our lab rat user of the month" story, perhaps |
| 21:12 | <jgraham> | AryehGregor: If you're getting your news primarilly from /. you are missing things like "truth" |
| 21:12 | <AryehGregor> | Works with the classic index. I'll just use that from now on. |
| 21:12 | <AryehGregor> | jgraham, well, no, but I glance over the comments on some of the interesting stories. |
| 21:13 | <Dashiva> | The problem with slashdot is that you have to wait a few days before the comments stabilize |
| 21:13 | <AryehGregor> | Anyway, at least the iPad doesn't support Flash. |
| 21:13 | <Dashiva> | And by then it's old news, so you have to go way back in the archives |
| 21:14 | <Steve^> | "at least"? |
| 21:14 | <Steve^> | Purely so you can say it has failed? |
| 21:15 | <AryehGregor> | What? |
| 21:16 | <Steve^> | you said at least |
| 21:16 | <Steve^> | as if you are happy it has passed some threshold |
| 21:17 | <AryehGregor> | I mean, "Maybe it's an evil machine designed to take away users' freedom, but at least it will help contribute to the demise of Flash." |
| 21:17 | <Dashiva> | I wonder if it has hardware support for h264 |
| 21:17 | <Dashiva> | Because it had better support youtube |
| 21:18 | <Steve^> | AryehGregor, ah, that makes more sense :) |
| 21:18 | <AryehGregor> | Dashiva, it will be like the iPhone, you use the YouTube app for YouTube. |
| 21:18 | <Steve^> | Dashiva, it does support youtube |
| 21:18 | <AryehGregor> | So I assume, anyway. |
| 21:18 | <Steve^> | they are pushing HTML5, apparently |
| 21:18 | <Dashiva> | Isn't the youtube app just h264? |
| 21:18 | <Steve^> | think so |
| 21:19 | <Dashiva> | (With a secret entry point because it wasn't used elsewhere yet) |
| 21:21 | <AryehGregor> | What do you mean? H.264 is a codec, not an app. |
| 21:27 | <Dashiva> | That it just accessed the h264 streams from youtube |
| 21:34 | <AryehGregor> | Well, yeah, so does the Flash player. |
| 21:35 | <AryehGregor> | AFAIK you just have to enter the right URL and you can get the raw H.264. |
| 21:35 | <AryehGregor> | There's a "YouTube with <video>" Chromium extension that did just that, in fact. |
| 21:44 | <Dashiva> | Yes, but now that youtube has <video>, and safari supports <video>, surely they can bypass the separate app |
| 21:48 | <Steve^> | I think they probably do |
| 21:49 | <Hixie> | is anne stil on vacation? |
| 22:40 | <Lerc> | Making pictures out of canvas dropshadows. http://screamingduck.com/Lerc/jspic/ |
| 22:57 | <mpilgrim> | zcorpan: i have gone through my "Dive Into HTML5" inbox and addressed everyone's feedback except yours |
| 22:57 | <mpilgrim> | and now it's 6pm and i have to go play video games with my son |
| 22:58 | <mpilgrim> | maybe next week |
| 23:07 | <AryehGregor> | Is there a way in CSS to force table cells to have the heights you want, like table-layout: fixed does for widths? |
| 23:08 | <AryehGregor> | Also, any way to force something to be exactly one page tall, in print? |