| 00:57 | <AryehGregor> | Does Firefox 4 do hardware acceleration on Windows XP? I'm having trouble finding sources. |
| 00:58 | <AryehGregor> | I assume it does, but I don't see it stated anywhere. |
| 00:59 | <zewt> | i've seen rendering corruption in canvas that looks reminiscent of texture corruption, anyway; don't know the real cause (too sporadic) |
| 00:59 | <zewt> | (xp64) |
| 01:00 | <kinetik> | AryehGregor: it has compositing acceleration via D3D9, if your video card/drivers are sufficient. |
| 01:00 | <AryehGregor> | So I assumed. |
| 01:00 | <kinetik> | AryehGregor: check about:support's graphics section to see if it's enabled on a given machine. |
| 01:01 | <AryehGregor> | kinetik, I was wondering because the Ars Technica review of IE9 repeated Microsoft's lie that they don't support XP because it doesn't support hardware acceleration. |
| 01:01 | <AryehGregor> | Would be nice if I had a good source to counter that. |
| 01:02 | <zewt> | hate how I have to use win7 to test IE9; win7 VMs are gigantic compared to XP VMs |
| 01:03 | <zewt> | also, it's not like you generally get hardware acceleration in VMware--yet IE9 works fine |
| 01:04 | <TabAtkins> | AryehGregor: I know that Chrome does hardware acceleration on XP. |
| 01:04 | <AryehGregor> | zewt, well, I assume it uses APIs like D2D and DirectWrite, and they presumably fall back to some kind of software rendering if acceleration is unavailable. |
| 01:05 | <zewt> | right |
| 01:05 | <AryehGregor> | That doesn't mean you could run it on XP without major changes, you'd have to add a compatibility layer. |
| 01:05 | <AryehGregor> | All the other browsers have to have such a compatibility layer anyway because they work on Mac and/or Linux. |
| 01:05 | <AryehGregor> | So it's not really much extra effort for them to support XP too. |
| 01:05 | <AryehGregor> | Does D2D actually do anything you can't do with DirectX 9? I mean, it's the same hardware, right? |
| 01:08 | <Philip`> | It apparently needs DX 10.1 for acceleration, which includes features that DX9 hardware lacks |
| 01:08 | <Philip`> | (I have no idea which of those features it uses, though) |
| 01:10 | <AryehGregor> | So does it only work on DX 10.1 hardware? |
| 01:14 | <Philip`> | Oh, actually, apparently it also works with DX9 hardware via Direct3D 10Level9 |
| 01:15 | <AryehGregor> | Which does, D2D? |
| 01:15 | <Philip`> | D2D |
| 01:15 | <Philip`> | like in http://www.techarp.com/showarticle.aspx?artno=637&pgno=2 |
| 01:16 | <Philip`> | assuming sufficient drivers |
| 01:17 | <AryehGregor> | It's sad how a reputable site like Ars is repeating Microsoft's lies about stuff like this. |
| 01:18 | <AryehGregor> | They also repeated Microsoft's lies about how much they contribute to the standards process, and how many tests they contribute. |
| 01:18 | <zewt> | they contribute to the standards process by silently editing out any mention of whateg :P |
| 01:18 | <AryehGregor> | I'm pretty sure that to date, you and I are the only people who contributed comprehensive test suites for any HTML5 feature at all. |
| 01:18 | <Philip`> | I don't know if that means the DX9 API and/or drivers are too limited to support D2D (despite the hardware being capable), or just that they didn't feel like implementing D2D on the DX9 API when they could use a more modern API and upgrade people's drivers |
| 01:18 | <AryehGregor> | At least last I checked. |
| 01:27 | <AryehGregor> | I need a blog so that I can pontificate more easily. |
| 01:27 | <AryehGregor> | The problem is, if I wanted to have a blog I'd have to write the software for it, just for the principle of the thing. |
| 01:28 | <Hixie> | i hear ya |
| 01:28 | <AryehGregor> | Hmm, I could do a WYSIWYG editor using my JavaScript implementation of my execCommand() spec. |
| 01:28 | <Hixie> | luckily blog software is amongst one of the easiest things to write |
| 01:28 | <Philip`> | Just use a .plan file |
| 01:28 | <AryehGregor> | Although I should probably hold off on that until I actually spec things like, e.g., block formatting. |
| 01:28 | <AryehGregor> | What's a .plan file? |
| 01:29 | <Philip`> | http://www.catb.org/jargon/html/P/plan-file.html |
| 01:30 | <zewt> | blogs before there were blogs |
| 01:32 | <Philip`> | (John Carmack's one (http://www.bluesnews.com/cgi-bin/finger.pl?id=1&time=19991226003141 etc) seems to be the most famous recent one) |
| 01:32 | <zewt> | carmack's is probably the only famous one, heh |
| 01:33 | <erlehmann> | >What needs to be done is to create two closed source programs that act as executable loaders / verifiers and communication proxies for the client and server. |
| 01:33 | <erlehmann> | my fail-sense is tingling |
| 01:35 | <Philip`> | Having an external verification program seems to be how modern anti-cheat systems work, as far as I'm aware |
| 01:36 | <zewt> | it's sort of dimly amusing that he's saying 1: the problem is the GPL won't let you close source it, so 2: we need to use a mechanism that deliberately thwarts the GPL |
| 01:36 | <AryehGregor> | In practice, is this strategy moderately effective, or totally ineffective? |
| 01:36 | <AryehGregor> | Obviously anyone willing to spend enough time with a debugger could subvert the system. |
| 01:36 | <AryehGregor> | It really depends on whether you make it annoying enough, and change it often enough, that no one with the necessary skills will bother spending the time. |
| 01:37 | <Philip`> | It seems to work well enough for Valve |
| 01:38 | <AryehGregor> | Speaking of which, someone should really get their act together and write good DRM for games. |
| 01:38 | <jcranmer> | you mean like Steam? |
| 01:38 | <AryehGregor> | If you're willing to require an Internet connection, it should be very feasible to make it impractical to crack at all for at least a moderate period after the game's release. |
| 01:38 | <AryehGregor> | What, Steam games aren't cracked? Since when? |
| 01:38 | <jcranmer> | well, they are cracked |
| 01:38 | <jcranmer> | but Steam provides a useful DRM service |
| 01:38 | <Philip`> | It helps if there's a strong penalty for getting caught (e.g. getting your account banned and losing your paid-for games), because then you don't have to prevent cheats as long as you detect them soon enough to disincentivise players (which could be weeks) |
| 01:39 | <jcranmer> | you need an internet connection once every 10 days or so |
| 01:39 | <AryehGregor> | Anyway, all you'd have to do is make sure that there are lots and lots of small essential pieces of info that the server holds, only a subset of which will ever be legitimately needed on a given playthrough. |
| 01:39 | <jcranmer> | and you also get the ability to download the game on any computer without needing the disk |
| 01:39 | <Philip`> | Players aren't willing to accept a required internet connection |
| 01:39 | <jcranmer> | that may no longer be true |
| 01:40 | <jcranmer> | well |
| 01:40 | <jcranmer> | hmm |
| 01:40 | <Philip`> | or at least they seem to complain vocally when you require that |
| 01:40 | <jcranmer> | I would say they're willing to accept an intermittent requirement |
| 01:40 | <AryehGregor> | Then you can just have the game request the pieces of info from the server as needed. Have the server keep track and throttle accounts that ask for too many. |
| 01:41 | <Philip`> | AryehGregor: How do you avoid suffering from latency when requesting information? |
| 01:41 | <AryehGregor> | Just make sure the pieces of information are only needed occasionally, and are fetched well in advance. |
| 01:41 | <AryehGregor> | So even if you lose your connection you could still play for, I don't know, a couple of hours. |
| 01:41 | <AryehGregor> | This approach would fit naturally for RPGs, for instance. |
| 01:41 | <jcranmer> | but not for, say, FPSs |
| 01:41 | <Philip`> | So you need lots of critical pieces of information, all of which can be predicted seconds or hours in advance of demand? |
| 01:42 | <AryehGregor> | Well, if their main selling point is the single-player plot, then it could work for FPSes too. |
| 01:42 | <Philip`> | including when players load saved games |
| 01:42 | <AryehGregor> | Philip`, yes. I don't think it would be hard. |
| 01:42 | <AryehGregor> | Lots of RPGs already have substantial amounts of random content and events. |
| 01:42 | <Philip`> | or use the in-game console to teleport to a random location in a random map |
| 01:42 | <Philip`> | Seems like it'd be a nightmare for QA :-p |
| 01:43 | <AryehGregor> | That would be handily masked by the level load delay. |
| 01:43 | <AryehGregor> | Network delays would be negligible. |
| 01:43 | <jcranmer> | it works better for RPGs than for FPSes |
| 01:43 | <AryehGregor> | Yeah, I was really thinking it would be best for RPGs. |
| 01:43 | <jcranmer> | except it might not work so well for non-linear RPGs, if those were ever created |
| 01:44 | <AryehGregor> | One simplistic way to do it would just be to separately encrypt every single texture, model, line of dialog, and so on. But that requires a reliable network connection. |
| 01:44 | <Philip`> | It doesn't seem like something you could feasibly retrofit onto an existing game design - you'd probably have to start with the same basic assumptions as an MMORPG |
| 01:44 | <AryehGregor> | If you could rely on the network connection, you could do it really straightforwardly in a typical RPG by just encrypting every line of dialog with a separate key. |
| 01:45 | <AryehGregor> | Yes, hackers would get a basic version out within a few days, but if you happen to choose weird conversation options . . . |
| 01:45 | <Philip`> | If you've got a reliable network you don't even need to run the game logic on the client - just do everything on the server |
| 01:45 | <AryehGregor> | Well, "reliable" is different from "low-latency and high-throughput". |
| 01:45 | <AryehGregor> | Although yes, you could also do it by just putting most of the game logic on the server. |
| 01:46 | <AryehGregor> | But you could minimize network involvement by putting only small, critical, hard-to-reverse-engineer parts of game logic on the server. |
| 01:46 | <Philip`> | (where "everything" means "what MMORPGs do", i.e. all resources on the client and lots of client-side prediction to smooth out lag etc) |
| 01:46 | <AryehGregor> | MMORPGs are designed for an entirely different purpose. |
| 01:46 | <AryehGregor> | The server exists to coordinate different clients. |
| 01:46 | <AryehGregor> | So I don't think the design would be comparable in practice. |
| 01:48 | <AryehGregor> | One thing you could do is have some decision-making pushed off to the server that's complicated and random, thus not hard to reverse-engineer approximately but impossible to reverse-engineer exactly. |
| 01:49 | <AryehGregor> | . . . to be honest, it would be a lot simpler and more foolproof to either put everything on the server or require trusted clients. |
| 01:49 | <AryehGregor> | But it's interesting to think about anyway. |
| 01:51 | Philip` | likes things to be simple and foolproof, since they're usually less of a pain to debug than clever schemes |
| 01:51 | <AryehGregor> | But much less fun to think about. |
| 01:54 | <AryehGregor> | Man, why does Google keep Chrome partly closed-source? It's annoying. |
| 01:54 | AryehGregor | pokes TabAtkins for unofficial comment |
| 01:56 | <Philip`> | I suppose you could make the game be like Demon's Souls, where you get hints and warnings from other players via the server |
| 01:56 | <Philip`> | so if you're offline or don't have a valid account then you get a degraded version of the game |
| 01:57 | <Philip`> | (though still playable, so players don't get annoyed and drive up tech support costs when they have a flakey connection) |
| 01:58 | <AryehGregor> | Are you suggesting that games have tech support? |
| 01:58 | <AryehGregor> | I don't think I ever even tried looking for tech support for a game. |
| 01:58 | <AryehGregor> | Also, lol: http://seclists.org/nmap-dev/2011/q1/767 |
| 02:00 | <Philip`> | I believe they always do, probably because otherwise disgruntled customers would call whatever phone number they could find for the company and annoy more highly-paid employees |
| 02:00 | <AryehGregor> | Sounds likely. |
| 02:01 | Philip` | likes getting feedback from users, as long as he doesn't feel obliged to reply to it |
| 02:02 | <Philip`> | (I added a text box on the menu screen of a game that sends a message anonymously to a server where we can read it, and got a hundred messages in a few days, and since it's anonymous there's no expectation of a response) |
| 02:02 | <Philip`> | (and more people use it for quick messages than use IRC or forums) |
| 02:03 | <Philip`> | (and some of the messages are actually useful) |
| 02:03 | <AryehGregor> | Yeah, that sort of thing is nice. |
| 02:04 | <AryehGregor> | It's the wiki principle: something that requires zero effort will be much, much more popular than something that requires slight effort. |
| 02:04 | <AryehGregor> | Although most wikis seem to not allow anonymous editing, which shows that this point is hard to grasp. |
| 02:05 | <Philip`> | (although some messages are of the form "pointer mouse is too sloww" or "Hello... How can i have the full game???Because the computer don't play..." and not terribly informative) |
| 02:05 | <AryehGregor> | Sure, but those are mostly harmless. |
| 02:06 | <Philip`> | Yeah, I think the important thing is for useless feedback and/or spam to have very low cost |
| 02:07 | <AryehGregor> | Thus on wikis, it's easier to undo changes than to make them in the first place. |
| 02:07 | <Philip`> | Spam on wikis is still fairly costly, in terms of the effect it has on readers and the requirement on editors to regularly check for it |
| 02:07 | <aho> | unless those changes were made by a botnet |
| 02:09 | <Philip`> | (especially on small wikis with few editors) |
| 02:09 | <Philip`> | (where spam can go unnoticed for hours or days or years) |
| 02:16 | <AryehGregor> | Yes, there's definitely a size threshold below which anonymous editing isn't useful. |
| 02:16 | <AryehGregor> | Having anonymous edits go into a moderation queue would be one obvious thing to do. |
| 02:17 | <AryehGregor> | MediaWiki should be able to do that these days with the FlaggedRevs extension, I assume. |
| 02:17 | <AryehGregor> | On Wikimedia wikis it's not being configured that way, obviously, but I imagine it could be. |
| 02:47 | <erlehmann> | AryehGregor, Philip`, it is interesting how you web engineers put the principles that make the web thriving away when talking about games. What makes them different? |
| 08:46 | <Creap> | Hi, how do I specify an input pattern to be case insensitive? |
| 08:52 | <zcorpan> | Creap: you can't currently |
| 08:53 | <Creap> | ok |
| 08:53 | <zcorpan> | but you can do [Hh][Ee][Ll][Ll][Oo] |
| 08:53 | <jacobrask> | yeah, I'll have to do a-zA-Z, not a big problem, but /i would be easier |
| 08:55 | <hsivonen> | is my recollection bad or did Opera 11.10 beta regress layout on https://demos.mozilla.org/en-US/ compared to 11.01? (I'm too lazy to go back to the old version to check) |
| 08:55 | <hsivonen> | the Planetarium demo crashes for me in Opera 11.10 on Linux 64 |
| 08:57 | <zcorpan> | hsivonen: it looks pretty much like in firefox in 11.01 for me |
| 08:57 | zcorpan | hasn't got the beta yet |
| 08:58 | <zcorpan> | planetarium doesn't crash here |
| 08:59 | <hsivonen> | zcorpan: thanks. I'll count the layout regressions as Opera beta bugs then, and won't file against the site |
| 09:00 | <hsivonen> | I submitted the crash report using the automatic reporter (I lost the id, but I put my @iki.fi address in the email field) |
| 09:01 | <jacobrask> | I believe it's known |
| 09:03 | <jacobrask> | zcorpan: Should I report the modifier issue somewhere? mailing list, or has it been discussed already? |
| 09:05 | <hsivonen> | jacobrask: what's known? |
| 09:06 | <jacobrask> | that input pattern does not support modifiers like case insensitivity |
| 09:06 | <jacobrask> | or oh |
| 09:06 | <jacobrask> | sorry |
| 09:07 | <jacobrask> | that Planetarium crashes in 11.10b is known |
| 09:07 | <zcorpan> | jacobolus: i think there's a bug on it (case insensitivity) |
| 09:08 | <zcorpan> | jacobolus: but you could whine about your use case in that bug to increase the likelihood that Hixie will add it to the spec |
| 09:08 | <jacobolus> | zcorpan: I think you mean jacobrask? |
| 09:08 | <jacobolus> | :) |
| 09:09 | <zcorpan> | yes, too similar names! |
| 09:09 | <jacobrask> | :P |
| 09:13 | <jacobrask> | where's the issue tracker? |
| 09:19 | <zcorpan> | you mean bugzilla? |
| 09:19 | <jacobrask> | I found w3c.org/Bugs |
| 09:19 | <jacobrask> | but no related issue |
| 09:19 | <jacobrask> | w3.org/Bugs ofcourse |
| 09:20 | <jacobrask> | I'll mail the list |
| 09:23 | <mrmil> | Hi, I noticed a "s" element in the spec and I am interested what it stands for (like "a" stands for "anchor", "s" stands for "?"), anybody happends to know? |
| 09:24 | <jacobrask> | strike |
| 09:25 | <mrmil> | ah, right, thanks :) |
| 10:17 | <jgraham> | AryehGregor: FWIW the one sensible argument I have heard against eagerly approving tests is that they could be picked up by some third party who want to use them in some sort of approval process |
| 10:18 | <jgraham> | Then having lots of buggy tests could be a problem |
| 10:19 | <jgraham> | Another reason I just invented is that it either encourages people to publish test results early (which I think is bad) or pushes back finding bugs until it is very late |
| 10:20 | <jgraham> | Neither of those reasons may — as a working group chair would no doubt phrase it — be decisive |
| 10:21 | <jgraham> | There are certianly advantages to removing the, sadly somewhat dysfunctional, review stage |
| 10:22 | <jgraham> | But the CSS2.1 situation where there is a big churn of the tests right at the end as people suddenly notice that they are full of problems isn't wonderful either |
| 10:24 | <jgraham> | (of course, taking a wider viewpoint, the whole idea of an "end" is an artificial construction that could well be gone by the time it becomes relevant to HTML) |
| 10:31 | <hsivonen> | It's quite sad that there is a prefix mapping error on the BustBuy page cited as evidence of prefixes not being especially complicated in the RDFa prefix ISSUE CP |
| 10:31 | <hsivonen> | *BestBuy |
| 10:32 | <hsivonen> | Today's the last day to object to http://www.w3.org/2002/09/wbs/40318/issue-120-objection-poll/results |
| 10:47 | jgraham | wonders if it worth pointing out that Yahoo shut down SearchMonkey |
| 10:48 | <jgraham> | Also that Julian's analogy with CSS is wrong |
| 11:21 | <hsivonen> | jgraham: please do point out that SearchMonkey has been shut down. (if it has. I was unaware.) |
| 11:21 | <hsivonen> | or had forgotten. or something |
| 11:25 | <jgraham> | Can someone confirm my findings that the facebook open graph api linter fails to extract any information if you change the prefix from "og" but works fine if the prefix is "og" but undeclared? |
| 11:25 | <jgraham> | My example documents are http://hoppipolla.co.uk/410/rdfa.html and http://hoppipolla.co.uk/410/rdfa-1.html |
| 11:26 | <jgraham> | http://developers.facebook.com/tools/lint/ |
| 11:26 | <hsivonen> | jgraham: your "Lint it" link is broken |
| 11:28 | <jgraham> | hsivonen: Yes,m I just copied the example file |
| 11:28 | <jgraham> | You need to copy the URI |
| 11:28 | <jgraham> | (I will fix) |
| 11:28 | <hsivonen> | jgraham: yeah, it seems to be that your finding is correct. |
| 11:28 | <hsivonen> | jgraham: please document it on the poll |
| 11:28 | <jgraham> | Also, the Java, Python, ruby and one PHP libraries all look for "og:" |
| 11:29 | <jgraham> | There is a PHP library that is actaully doing real RDFa processing and one perl one that I assume is |
| 11:29 | <hsivonen> | This kind of name dropping companies whose implementations don't implement the spec *really* annoys me. |
| 11:30 | <jgraham> | I think this is quite reasonable evidence that using prefixes actually already doesn't work |
| 11:31 | <hsivonen> | jgraham: indeed |
| 12:00 | <jgraham> | hsivonen, anyone: could you read http://hoppipolla.co.uk/410/120.txt and see if it makes sense, please? |
| 12:01 | <hsivonen> | jgraham: looking |
| 12:02 | <zcorpan> | "... it’s important for site developers to test their site in the Compatibility View Browser Mode." http://blogs.msdn.com/b/ie/archive/2010/10/19/testing-sites-with-browser-mode-vs-doc-mode.aspx |
| 12:03 | <hsivonen> | jgraham: looks good |
| 12:03 | <hsivonen> | zcorpan: aargh! :-( |
| 12:05 | <hsivonen> | zcorpan: on the bright side, it appears that IE9 itself doesn't (outside dev tools) provide UI for running the browser in the IE8 browser mode or the IE7 browser mode |
| 12:05 | <hsivonen> | zcorpan: though I gather that apps that embed Trident can do that |
| 12:05 | <kennyluck> | I wonder why no one ever mentions the possibility of moving RDFa out of HTML WG in the objections. The proposed change in Hixie's proposal is too large that it can't be acceptable. |
| 12:06 | <zcorpan> | since ie9 uses the same mode as the parent frame, i guess it's also "important" to test the site in the IE 5.5 mode if you care about being framed on arbitrary sites |
| 12:07 | <zcorpan> | hsivonen: the context of the sentence was the compatibility mode button |
| 12:07 | <zcorpan> | s/mode/view/ |
| 12:08 | <zcorpan> | i thought the x-ua-compatible would override compatibility view? |
| 12:08 | <hsivonen> | zcorpan: yeah, but I think it's very sad to put effort into making the site work if the user presses a "break design" button |
| 12:08 | <hsivonen> | zcorpan: X-UA-Compatible overrides the button and hides the button |
| 12:09 | <hsivonen> | zcorpan: also overrides the pref for displaying all sites in compat mode |
| 12:09 | <zcorpan> | i wonder why microsoft now recommends wasting even more time over using x-ua-compatible |
| 12:11 | <hsivonen> | there's no way an average user is going to last even a day without Flash Player using IE9 :-( |
| 12:11 | <hsivonen> | the browser is so eager to prompt for drive-by installs of Flash Player |
| 12:13 | <zcorpan> | "As long as developers need document modes to run their sites, we’re going to support them. We want to remove document modes as soon as the web transitions to run in the latest standards mode. The timing is really up to developers." |
| 12:13 | <hsivonen> | zcorpan: whoa! looks like I need to retest the framing bit. :-( |
| 12:13 | <zcorpan> | Translation: We're never going to remove any document modes. |
| 12:15 | <zcorpan> | hsivonen: i haven't tested it myself, but somebody on sitepoint forums said ie9 would use quirks mode if the parent frame uses quirks mode |
| 12:15 | <zcorpan> | regardless of doctype or x-ua-compatible, iirc |
| 12:15 | <hsivonen> | note to self: always retest *everything* |
| 12:19 | <hsivonen> | well, *this* is weird |
| 12:20 | <hsivonen> | when framed by a quirky page, it seems that X-UA-Compatible: IE=9 only elevates to mode to 8 |
| 12:21 | <hsivonen> | I need to draw some new decision diamonds for my chart |
| 12:22 | <hsivonen> | if my framing testing so far is right, this is gonna be bad news for HTML5 video embedding iframes |
| 12:23 | <hsivonen> | since <video> requires IE9 mode |
| 13:36 | <zcorpan> | hsivonen: http://www.sitepoint.com/forums/html-xhtml-52/ie9-iframes-doctypes-you-743000.html was the ie9 iframe thing |
| 16:06 | <TabAtkins> | AryehGregor: As far as I know, it's because we have some libraries that can't be open-sourced. |
| 16:06 | <TabAtkins> | Like our implementation of Flash and PDF. |
| 16:07 | <jgraham> | death_to_kittens.cpp |
| 16:10 | <TabAtkins> | Well, we do hate kittens. |
| 16:58 | <zewt> | what the |
| 16:58 | <zewt> | apparently checking out the webkit source gives you ... webkit.org? thanks, I guess... :) |
| 17:03 | <TabAtkins_> | ?_? |
| 17:07 | <Peter-> | The source of webkit.org is on svn, he's probably poining that out |
| 17:07 | <Peter-> | (the site) |
| 17:07 | <Peter-> | as are bugs. and planet. |
| 17:11 | <AryehGregor> | erlehmann, just because openness and standards work for the web doesn't mean they're needed everywhere. I tend to broadly agree with Eric S. Raymond's analysis of open-source software, where he suggests that closed-source software will always work better in some markets. The same logic applies to open standards. |
| 17:12 | <erlehmann> | AryehGregor, please elaborate. don't work “closed” products better even with open standards? |
| 17:13 | <AryehGregor> | Well, if interoperability is useful, yes. What do games need to interoperate on? |
| 17:13 | <erlehmann> | There may be more people than h4kon advocating open standards while being perfectly fine with closed software. |
| 17:13 | <AryehGregor> | A typical high-end game is a standalone product that interacts with no other program (beyond the OS) and is basically obsolete in two years. |
| 17:14 | <AryehGregor> | What exactly is there to standardize? |
| 17:14 | <erlehmann> | Special clients? I always hated it when strategy games did not give me the ability to write my own macros. |
| 17:15 | <erlehmann> | (The only one that included it was Earth2150 in 1999, but the current rights holder is opposed to releasing the source code.) |
| 17:15 | <AryehGregor> | What sort of API would they expose? Different games work very differently, that's one of the problems. |
| 17:16 | <erlehmann> | I just hate it when people lay bricks into my way of customizing things. |
| 17:16 | <AryehGregor> | I don't think standardization would help with that. |
| 17:16 | <AryehGregor> | Standardization is needed for inter-program APIs, document formats, things like that. But any two independent games are different enough that there's nothing I can see that they could usefully share, in general. |
| 17:16 | <AryehGregor> | Well, they share things like graphics engines, clearly. |
| 17:17 | <AryehGregor> | But not along nice clean standardizable lines that you could make a stable API out of. |
| 17:17 | <Philip`> | They can be standardised at the OpenGL/DirectX level |
| 17:17 | <AryehGregor> | At least it's my impression that when games use a third-party engine, they typically customize it to fit their needs. |
| 17:17 | <AryehGregor> | Well, yes. Also file access can be standardized. |
| 17:17 | <AryehGregor> | Novel idea right there. |
| 17:17 | <AryehGregor> | But that's standardizing the OS, not the game. |
| 17:18 | <Philip`> | Anything higher-level will be obsolete within a year so everyone will want to write their own instead, unless the shared engine is continually updated and no longer standardised |
| 17:18 | <AryehGregor> | Essentially, yes. |
| 17:19 | <AryehGregor> | (I've given thought to the possibility that an open-source graphics engine for games could be successful, if it used a dual-licensing scheme in the vein of MySQL.) |
| 17:19 | <AryehGregor> | (It's really ridiculous how companies rewrite their graphics engines from scratch for most games they make.) |
| 17:19 | <erlehmann> | Possible strawman: I neither do nor did advocate a common API or ABI for all games. |
| 17:20 | <Philip`> | (How would dual-licensing make it more successful than e.g. Ogre, which used to be dual-licensed and now is MIT?) |
| 17:21 | <AryehGregor> | erlehmann, even similar games aren't similar enough that I see any useful API you could make. |
| 17:21 | <AryehGregor> | Maybe adopting a common scripting language or something like that, sure. |
| 17:21 | <AryehGregor> | Would be possible. |
| 17:21 | <erlehmann> | LUA called, it wants its fame back ;) |
| 17:22 | <AryehGregor> | But high-end games are usually written from scratch in a very short period and then thrown out. That makes it unattractive to invest in features that aren't broadly useful. |
| 17:22 | <AryehGregor> | Philip`, I've never heard of Ogre. I'm just saying it could work, it could also fail for lots of reasons. |
| 17:22 | <erlehmann> | Philip`, re: dual-licensing “commercially successful” is different from “broad adoption successful” |
| 17:23 | <AryehGregor> | My strategy would be to distribute it for free and let small players use it for free, and figure that if any big games stole it we'd notice. |
| 17:23 | <AryehGregor> | Try to get universities to use it in their courses, try to get a development community going. |
| 17:24 | <AryehGregor> | Anyway, I'm not actually going to try this, just an idle thought. |
| 17:25 | <Hixie> | hober: is othermaciej on vacation or something? |
| 17:25 | <AryehGregor> | hsivonen, you are aware that the chairs will consider objections even if they don't use the words "I object" every two sentences, right? |
| 17:25 | <Philip`> | AryehGregor: Ogre's only the most well-known widespread open source graphics engine, which has been going seriously for ~6 years and has been used in several commercial games (e.g. Torchlight) :-) |
| 17:25 | <AryehGregor> | Ah, interesting. |
| 17:26 | <AryehGregor> | Where do they get their development money from, if they're MIT-licensed? Contracts, support? I thought good 3D engines required very large ongoing development costs. |
| 17:27 | <AryehGregor> | (As opposed to most open-source projects, which can be more or less completed and then add features at their leisure without necessarily falling far behind) |
| 17:27 | <Philip`> | (I thought vaguely about using it recently, but found it's a pain to install on Linux because it now relies on a library that interacts badly with system libraries so most Linux distros won't touch it, so it ends up being a bit of a mess getting everything installed) |
| 17:27 | <jgraham> | AryehGregor: It seems to be profitable to be very explicit when writing objections |
| 17:28 | <Philip`> | (so it's easier to just hack stuff onto an existing inflexible GL-based engine) |
| 17:28 | <AryehGregor> | jgraham, that I agree with. |
| 17:29 | <Philip`> | AryehGregor: As far as I'm aware they don't have development money, it's just volunteers who want to work on the engine or want to use it for their own open source games |
| 17:29 | <zewt> | i've worked on a commercial game that MIT-licenses most of its code for many years ... it works "okay" |
| 17:29 | <AryehGregor> | And it's still competitive? Or it's not competitive with the commercial engines, but it's good enough for the cost? |
| 17:29 | <Philip`> | and the result is okay enough that some commercial developers choose it instead of writing their own or buying one |
| 17:31 | <Philip`> | It's competitive if you're making a PC game that should run easily on 5-year-old hardware |
| 17:31 | <Philip`> | (from what I can see) |
| 17:31 | <Philip`> | (e.g. Torchlight lists GeForce 2 as a minimum requirement) |
| 17:32 | <Philip`> | (or Intel GMA 950, more relevantly) |
| 17:32 | <zewt> | that's more like 10-year-old |
| 17:32 | <Philip`> | 950 is more like 5, I think |
| 17:32 | <Philip`> | and is very widespread |
| 17:33 | <jgraham> | AryehGregor: I rather suspect I was insufficently explicit |
| 17:33 | <AryehGregor> | jgraham, on which issue? |
| 17:33 | <Philip`> | It's not exactly a graphics engine you're going to use for an Xbox 360 game |
| 17:34 | <Philip`> | (even if it supported consoles, which it doesn't) |
| 17:34 | <zewt> | well, no open source game is actually going to really support consoles, since the APIs are largely proprietary and it's probably quasi-legal to even release code using them, heh |
| 17:34 | <jgraham> | AryehGregor: 120 |
| 17:35 | <AryehGregor> | jgraham, I hate when people refer to issues by numbers. Can't you at least use the short description? |
| 17:35 | <zewt> | (we did a PS2 port of our game years back; none of that code was ever released) |
| 17:35 | <AryehGregor> | Is that the RDFa prefix one? |
| 17:36 | <AryehGregor> | I suspect that one's a lost cause politically, but the chairs may surprise me. |
| 17:36 | <jgraham> | AryehGregor: Yes, it's the RDFa one |
| 17:36 | <jgraham> | I have no idea what the shortname is |
| 17:36 | <jgraham> | Oh rdfa-prefixes |
| 17:37 | <Philip`> | Fortunately open source people seem very easily pleased with simple graphics - I see lots of compliments for the graphics of the game I work on, which has a 2004-era renderer that supports diffuse textures and shadows and a single light source and that's pretty much all |
| 17:38 | <hober> | yeah, I need to get some survey responses in today |
| 17:38 | <AryehGregor> | What does "a single light source" mean, in practice? |
| 17:38 | <Philip`> | The sun |
| 17:38 | <AryehGregor> | Ah. |
| 17:38 | <Philip`> | (No dynamic lighting or anything) |
| 17:38 | <AryehGregor> | So constant illumination from a fixed direction. |
| 17:38 | <AryehGregor> | I find graphics vaguely interesting. |
| 17:38 | <Philip`> | Indeed |
| 17:39 | <AryehGregor> | I remember being amused when I saw in my brother's 3D graphics program (3DS Max or whatever) that he could change the rate at which the intensity of light dropped off. |
| 17:39 | <AryehGregor> | Constant, inverse-square, inverse, inverse-cube, exponential . . . |
| 17:39 | <jgraham> | AryehGregor: Anyway, clearly the solution is to write a RFC defining terms like OBJECT and SUPPORT so people would know to look at the sentences with the defined terms and filter out the rest. |
| 17:40 | <AryehGregor> | Granted, constant makes sense for very far light sources like the sun, but exponential dropoff for light? I guess you're not always aiming for realism, in graphics. |
| 17:40 | <AryehGregor> | The software also supported indices of refraction less than 1. |
| 17:40 | <AryehGregor> | Although negative indices of refraction can be possible. |
| 17:41 | <AryehGregor> | jgraham, I don't think the chairs have dropped any objections lately just because of how they were phrased, did they? |
| 17:41 | Philip` | recently considered doing something as extreme as requiring GLSL shaders, which are only 7 years old, so he set up http://feedback.wildfiregames.com/report/opengl/ a few weeks ago to figure out what hardware support was available |
| 17:41 | <Philip`> | but it turns out quite a few people still have terrible hardware or drivers :-( |
| 17:42 | <AryehGregor> | Philip`, do you know of any book I could read that would teach me more about the principles of graphics programming, without expecting that I'd be willing to write any actual code except maybe really trivial snippets? |
| 17:42 | <zewt> | i'd imagine typical systems with onboard Intel video chipsets and similar laptops |
| 17:43 | <AryehGregor> | I would like to get it, read the first twenty pages, put it aside somewhere, and forget about it for months until I clean up my room, at which point I'll realize I have no time to read it and put it on my shelf, with a forlorn bookmark forever remaining in it as a testament to my lack of persistence. |
| 17:43 | <Philip`> | AryehGregor: No - I don't think I've read any particularly relevant books, and I don't know enough about the subject to guess which ones might be good |
| 17:43 | <jgraham> | AryehGregor: It's hard to tell. Based on http://www.w3.org/mid/20110310180425.M45789⊙hn it seems that some people don't feel their points have been considered in the way they would like due to the choice of wording |
| 17:44 | <AryehGregor> | jgraham, that objection was ignored because it suggested a change that wasn't suggested by any change proposal. |
| 17:44 | <AryehGregor> | I.e., it didn't actually support either change proposal, and they were only trying to decide between the change proposals, so it was useless to them. |
| 17:45 | <Philip`> | zewt: Yeah, lots of those, though I was surprised to see someone apparently still had a GeForce2 MX |
| 17:47 | <Philip`> | (which is even slower than an Intel 945GM, seemingly) |
| 17:47 | <AryehGregor> | I think I have a GeForce 1 somewhere. |
| 17:47 | <AryehGregor> | Not in an actual working computer, needless to say. |
| 17:48 | <AryehGregor> | Wasn't GeForce 1 a PCI card, though? If so, maybe it would still work. I bet the Linux kernel still supports it. |
| 17:50 | <Philip`> | AGP, apparently |
| 17:50 | <Philip`> | (http://en.wikipedia.org/wiki/Comparison_of_Nvidia_graphics_processing_units#GeForce_series) |
| 17:52 | <AryehGregor> | I probably have an AGP motherboard somewhere. Whether I have a CPU and RAM to go with it is a separate question. |
| 17:56 | Philip` | likes measuring things |
| 17:56 | <Philip`> | http://zaynar.co.uk/0ad-pub/performance-20110317.png - probably wholly inaccurate, but I think it looks pretty |
| 17:57 | <jgraham> | What did you plot it with? |
| 17:57 | <Philip`> | matplotlib |
| 18:02 | <AryehGregor> | I don't suppose anyone here knows how to get execCommand() to work in IE? It doesn't seem to work on my page: http://aryeh.name/spec/editcommands/autoimplementation.html |
| 18:02 | <AryehGregor> | The "browser" column remains stubbornly unstyled in IE9, with no errors logged or anything. |
| 18:03 | <AryehGregor> | It's worked before . . . |
| 18:03 | <AryehGregor> | Oh well, I don't need to know IE's behavior for this particular thing anyway. |
| 18:03 | AryehGregor | goes back to rewriting his spec for about the fourth time |
| 18:03 | <AryehGregor> | No, wait, I think it's only the third. |
| 18:19 | <zewt> | <Peter-> The source of webkit.org is on svn <- it's not having them in svn that's surprising, it's having them in the WebKit source tree |
| 18:22 | <nessy> | Hixie: are you there? |
| 18:22 | <Hixie> | here |
| 18:23 | <nessy> | issue-147 about playbackrate |
| 18:23 | <zewt> | webkit doesn't send onprogress reliably on chunked XHR responses :( |
| 18:23 | <nessy> | are you planning on writing a change proposal? |
| 18:24 | <zewt> | probably due to buffering |
| 18:24 | <zewt> | yeah, definitely |
| 18:26 | <Hixie> | nessy: not if i can help it |
| 18:26 | <nessy> | Hixie: if you like the idea of requestedPlaybackRate/actualPlaybackRate, I'd be happy to quickly write it |
| 18:26 | <nessy> | and request another week of discussion extension, I guess |
| 18:27 | <Hixie> | based on what frank said, he doesn't much care to do anything but what they already shipped |
| 18:27 | <Hixie> | so... |
| 18:28 | <nessy> | hmm, I care about a good spec and I would like to hear other browser vendor opinions |
| 18:28 | <Hixie> | i care about a good spec too, but it seems the w3c html wg is not the place we'll be able to achieve that |
| 18:28 | <Hixie> | (this is yet another example of how having microsoft implement cutting-edge stuff is a mistake) |
| 18:29 | <nessy> | I've not lost faith :) |
| 18:29 | <nessy> | other browsers have implemented stuff, too, that got changed later - that's not new |
| 18:30 | <nessy> | no worries, I'll go ahead then |
| 18:33 | <Hixie> | other browsers don't have 3 year ship cycles |
| 18:33 | <Hixie> | (my faith is mostly lost from sam's many decisions recently, not just this issue) |
| 18:36 | <Hixie> | (and the recent polls, which have no good solutions proposed) |
| 18:40 | <AryehGregor> | IE9 was shipped only two years after IE8, to be fair, not three. :) |
| 19:07 | <nessy> | OK, change proposal on issue-147 playbackrate sent - we'll see what the discussion will further turn up :) |
| 19:18 | <Hixie> | nessy: thanks |
| 19:18 | <Hixie> | nessy: there's a good chance that whatever happens we'll end up forced to spec whatever IE did though |
| 19:20 | <Hixie> | nessy: also "requestedPlaybackRate" makes it sound like its ok for the browser to be playing it back at another rate |
| 19:20 | <Hixie> | nessy: (if i was doing this myself rather than jumping through sam's hoops, i think we'd probably want to put this into a separate object hanging off the HTMLMediaElement that exposes all kinds of metrics, not just the actualPlaybackRate) |
| 19:21 | <nessy> | you should write that as a change proposal then :) |
| 19:21 | <Hixie> | *shrug* |
| 19:22 | <nessy> | I'm not sure I would regard the actualPlaybackRate as a statistic though - it's not quite in the same realm... |
| 19:22 | <Hixie> | how so? |
| 19:22 | <nessy> | other stats are about what the browser has tried to achieve and wasn't able to - this is about the feature being not available on some devices |
| 19:23 | <Hixie> | i don't understand the difference |
| 19:23 | <Hixie> | those sound like different ways of describing exactly the same thing |
| 19:23 | <nessy> | basically it confirms what you say above: it's ok for the feature not to be implemented |
| 19:23 | <Hixie> | it's not ok for the feature not to be implemented |
| 19:24 | <Hixie> | if it was ok for the feature not to be implemented, we should just remove the feature |
| 19:25 | <nessy> | no, I think people have different expectations of different devices |
| 19:25 | <nessy> | and their capabilities - what is ok on some is not ok on others |
| 19:25 | <Hixie> | we have an explicit hardware limitations clause |
| 19:25 | <Hixie> | which covers this kind of thing for everything |
| 19:25 | <Hixie> | why is playback rate special? |
| 19:25 | <nessy> | yeah, and that's fine - but then how do you determine as an author when something has been hw limited? |
| 19:28 | <Hixie> | you don't, because if we have to expose something like that for every feature, that won't scale |
| 19:28 | <AryehGregor> | The hardware limitation clause is invoked too much. We should specify how to behave in the event of hardware limitations where possible, same as with all error handling. |
| 19:28 | <AryehGregor> | Although obviously it could get out of hand if you go too far down that route, yeah. |
| 19:29 | AryehGregor | has no comment on the specific issue, hasn't looked at it |
| 19:29 | <Hixie> | it's not realistic to do that. for example browsers assume infinite ram, and will fail in unpredictable ways when ram runs out, and there's nothing we can do about that really. |
| 19:29 | <AryehGregor> | Generally it fails in a quite predictable way, viz., the browser exits or is killed. |
| 19:30 | <Hixie> | that is one of a number of different eventualities |
| 19:30 | <Hixie> | it's by far not the only one |
| 19:31 | <Hixie> | and it's not really one we can spec |
| 19:31 | <Hixie> | since it's very reasonable for a browser to try to do better |
| 19:39 | <nessy> | it's well possible that a metrics or statistics element could contain this information in a better way, so it might be time to get on top of that - what's your plan on that? (I know you're swamped with issues) |
| 19:40 | <nessy> | my change proposal basically bought us some more time... |
| 19:41 | <nessy> | I know that some browsers are already implementing statistics elements, so it might be good to move that forward |
| 19:41 | <Hixie> | for implementation issues like this all that matters is what browsers implement, not what the wg decides |
| 19:41 | <Hixie> | so it doesn't really matter what the chairs decide, we can always change it later if we find a better solution that gets implemented |
| 19:41 | <Hixie> | hence my lack of caring about the change proposal stuff |
| 19:42 | <Hixie> | re video statistics, is there any documentation anywhere about what browsers have implemented for that? |
| 19:42 | <Hixie> | i don't really know what authors want, stats-wise |
| 19:43 | <Hixie> | i have the same problem with PeerConnection -- we're gonna need to expose some stats there, but i dunno what authors want |
| 20:00 | <nessy> | brb |
| 20:09 | <nessy> | stats info is a bit all over the place actually |
| 20:09 | <nessy> | in mozilla bugs, webkit bugs, http adaptive streaming discussions |
| 20:10 | <nessy> | maybe I should start pulling it all together in a web page |
| 20:10 | <nessy> | wiki page |
| 20:10 | <Hixie> | that would be awesome |
| 20:10 | <Hixie> | i can definitely commit to speccing out such a feature sooner if you do that :-) |
| 20:10 | <nessy> | happy to help where I can! |
| 20:11 | <nessy> | I will be in MTV all of next week, btw |
| 20:11 | <Hixie> | cool |
| 20:11 | <Hixie> | i'm doing a lot of WFH recently for family reasons |
| 20:11 | <Hixie> | dunno where i'll be next week |
| 20:11 | <Hixie> | we should try to meet up though if i'm around |
| 20:11 | <nessy> | if you want to chat on Monday about the multitrack stuff, for example, just before the change proposal is due, I'd be there |
| 20:12 | <Hixie> | when is that due again? |
| 20:12 | <nessy> | Tuesday |
| 20:12 | <Hixie> | aw man |
| 20:12 | <nessy> | yeah… I know ... |
| 20:12 | <Hixie> | i guess i'd better cook something up over the weekend |
| 20:12 | <Hixie> | these deadlines are asinine |
| 20:12 | <nessy> | that would be awesome |
| 20:13 | <nessy> | (wow, had to look up that word - sometimes it sucks not being a native speaker ;-) |
| 20:14 | <nessy> | I'll be around Naomi and Ken most of the time, so just ping me when you're in the office and I'll come over (when I'm not in meetings and stuff) |
| 20:14 | <nessy> | we can grab a coffee |
| 20:14 | <Hixie> | (it wouldn't be so bad if the issues were being dealt with (a) in something resembling a realistic priority order and (b) were getting decisions from the chairs made promptly |
| 20:14 | <Hixie> | but anyway) |
| 20:15 | <hober> | It's particularly irritating when there's active discussion and possible convergence (as in this case), but that active discussion just happened to not happen before the decision process kicked in |
| 20:16 | <AryehGregor> | The decision process can always be stopped if there's an amicable resolution. |
| 20:29 | <Hixie> | AryehGregor: the problem is with something like the rel="" issue, where i think we'll never get amicable resolution, but where there's likely to be a better idea than the options currently on the table eventually, but as the decision process will be invoked before then (even though there's no rush on the issue), the htmlwg will not be able to change its decision, and if the whatwg tries the other option, we'll have a forked registration mechanism, which is bad for e |
| 20:29 | <AryehGregor> | Hixie, you got cut of at "which is bad for". |
| 20:29 | <Hixie> | was just loading the logs to find out where it cut out :-) |
| 20:30 | <AryehGregor> | If a new idea comes up that's better, the issue can probably be reopened. |
| 20:30 | <Hixie> | "which is bad for everyone." |
| 20:30 | <Hixie> | the chairs only reopen for "new information", not new ideas |
| 20:31 | <AryehGregor> | I expect they'd be willing to reopen if there were a good solution that hadn't previously been suggested. No way to tell until we try. |
| 20:32 | <Hixie> | your optimism is heart-warming. :-) |
| 20:32 | <AryehGregor> | Their goal is to get as much consensus as possible, so they're unlikely to refuse to reopen the issue if it looks like a new solution will get more people to agree. |
| 20:32 | <Hixie> | their goal is walk through the process, not get consensus, as far as i can tell |
| 20:33 | <AryehGregor> | They rely heavily on the process because they want to appear as impartial as possible. |
| 20:33 | <AryehGregor> | Thus they try to leave as little as possible up to their own discretion. |
| 20:33 | <Hixie> | i mean the lc/cr/rec process, not the htmlwg process |
| 20:33 | <AryehGregor> | Oh, well, that's obviously their goal too. |
| 20:33 | <Hixie> | their goal SHOULD be making a good web |
| 20:33 | <Hixie> | but anyway |
| 20:33 | <Hixie> | brb |
| 20:33 | <AryehGregor> | Well, yeah, but that's life. |
| 20:34 | <AryehGregor> | Only way you're going to fix it is if you get everyone important to jump ship at the W3C, or credibly threaten to. |
| 20:51 | <AryehGregor> | Is there a less verbose way to get the last element of an array than list[list.length - 1]? |
| 20:52 | <Hixie> | in js? |
| 20:52 | <jgraham> | AryehGregor: No |
| 20:52 | <AryehGregor> | Hixie, yes. |
| 20:52 | <AryehGregor> | jgraham, <3 |
| 20:53 | <Hixie> | if you don't mind changing the array you can call .pop :-) |
| 20:53 | <AryehGregor> | $ python -c 'print [1, 2, 3][-1]' |
| 20:53 | <AryehGregor> | 3 |
| 20:53 | <AryehGregor> | Yeah, I figured that out. |
| 20:53 | AryehGregor | grumbles |
| 20:53 | <Hixie> | in perl you can use $#, as in $array[$#array] |
| 20:54 | <AryehGregor> | Python's still shorter. :) |
| 20:54 | jgraham | assumes that perl makes sense to someone |
| 20:54 | <AryehGregor> | And it lets you access anything starting from the end of the list with about equal ease. |
| 20:54 | <AryehGregor> | jgraham, I assume $#array is just the length of the array, or maybe that minus one. |
| 20:54 | <AryehGregor> | #, after all. |
| 20:54 | <heycam> | AryehGregor, list.slice(-1) |
| 20:54 | <zewt> | python? more convenient than javascript? : O |
| 20:54 | <AryehGregor> | heycam, nice, thanks. |
| 20:54 | <heycam> | well that's a list actually |
| 20:54 | <Hixie> | you might be able to do array.slice(-1)[0] in js, but i expect the .length thing is more optimal |
| 20:54 | <AryehGregor> | Oh, that returns -- yeah. |
| 20:54 | <heycam> | so .slice(-1)[0] :) |
| 20:54 | <AryehGregor> | list.slice(-1)[0]. |
| 20:54 | <AryehGregor> | Hmm. |
| 20:55 | <Hixie> | oh heycam beat me to it |
| 20:55 | <AryehGregor> | Well, that's shorter if the variable name is long. |
| 20:55 | <jgraham> | That is slower though |
| 20:55 | <jgraham> | It has to create a new temporary array |
| 20:55 | <AryehGregor> | It's also more inscrutable. |
| 20:55 | <AryehGregor> | We can't just make Python the new official web programming language? :( |
| 20:55 | <jgraham> | Well unless the javascript engine is really clever I suppose |
| 20:55 | <heycam> | Object.defineProperty(Array.prototype, "-1", { get: function() { return this[this.length - 1] }, enumerable: false }) |
| 20:55 | <heycam> | :) |
| 20:55 | AryehGregor | really wishes he had more opportunity to write Python |
| 20:56 | <AryehGregor> | heycam, that's the most inscrutable sort of thing to do of all . . . |
| 20:56 | <jgraham> | heycam: He said shorter :p |
| 20:56 | <Hixie> | another option is to store your array in the opposite order |
| 20:56 | <AryehGregor> | jgraham, no, the point is then I could do array[-1]. |
| 20:56 | <Hixie> | and use [0] |
| 20:56 | <jgraham> | AryehGregor: Yeah, I understand |
| 20:56 | <AryehGregor> | Then I can't use push() and pop(), so my overall length in this case goes up. |
| 20:56 | <jgraham> | But if you only do it once... |
| 20:58 | <Hixie> | AryehGregor: you can use .shift and .unshift instead |
| 20:58 | <Hixie> | though i can never remember which adds and which removes |
| 20:58 | <AryehGregor> | Aha, that will do it. |
| 20:58 | <AryehGregor> | Not enough effort to rewrite my code now, but I'll keep it in mind for the future. |
| 20:59 | <jgraham> | It seems like they could be horribly inefficient too |
| 20:59 | <jgraham> | Depending on the implementation and the size of the array |
| 21:00 | <AryehGregor> | I'm ignoring efficiency here. |
| 21:00 | <Hixie> | no reason they should be any worse than pop and push |
| 21:00 | <AryehGregor> | Hixie, sure they should, if you're using something like std::vector to implement it, and that works by allocating an array. |
| 21:00 | <AryehGregor> | You'd have to shift all the elements up every time you unshifted. |
| 21:01 | <AryehGregor> | And down if you shifted, unless you wanted to just stop using the beginning of the array. |
| 21:01 | <Hixie> | i'd be shocked if browsers don't all have their own implementations of Array designed to handle all these things |
| 21:02 | <AryehGregor> | I said "something like". |
| 21:02 | <Hixie> | my point stands :-) |
| 21:02 | <AryehGregor> | I also really doubt they're actually using std::vector. |
| 21:02 | <Hixie> | i doubt it's anything like std::vector |
| 21:03 | <jgraham> | If you are representing an array as something a bit like a C array you can pop rather easilly (assuming you overallocate memory) |
| 21:03 | <jgraham> | But doing the same thing which shift seems more complex |
| 21:03 | <jgraham> | Or, at least, you could end up with memory fragmentation |
| 21:04 | <jgraham> | I don't know what actual browsers do though |
| 21:04 | <jgraham> | It seems like one could time it to find out :) |
| 21:07 | <zewt> | seems O(n^2) in chrome |
| 21:08 | <AryehGregor> | JavaScript is really obnoxiously verbose. It's the one language where I feel the need to do: autocmd FileType javascript setlocal fo-=t |
| 21:09 | <zewt> | or it did a minute ago--same test went from 6s to 5ms, scratching my head at it for a minute |
| 21:09 | <AryehGregor> | (which says "don't automatically break code at 79 characters) |
| 21:10 | <zewt> | okay, chrome is just acting strangely |
| 21:10 | <Hixie> | zewt: i wouldn't be at all surprised if it did something like adapt the implementation to match the usage |
| 21:11 | <zewt> | just refreshed the test, and it froze for around 15 seconds, then came back saying it took 13ms, suggesting it got stuck somewhere else |
| 21:12 | <Hixie> | i see that sometimes with the spec |
| 21:13 | <Hixie> | reloading the spec seems to make the tab go weird |
| 21:13 | <Hixie> | could be a GC thing |
| 21:14 | <gsnedders> | Hixie: I think it's still O(n) in most JS engines |
| 21:15 | <gsnedders> | Hixie: Also varies for some engines depending upon whether it's an actual Array or not, and what array representation it is using (AFAIK all have multiple) |
| 21:37 | <Philip`> | The trick to getting good performance is to use JS features in the same way that artificial benchmarks use them |
| 21:38 | <Philip`> | If you try do anything unusually clever you'll probably fall into a slow path |
| 21:39 | <Hixie> | i thought the trick was to write benchmarks that used js the way you do :-) |
| 21:40 | <Philip`> | That works but only if you're willing to wait a year or two |
| 21:40 | <gsnedders> | And manage to get anyone to pay attention to your benchmarks |
| 21:41 | <zewt> | Hixie: mailed a summary of how I think the PeerConnection crypto should be handled ... down to the bottom of the queue I go :) |
| 21:46 | <erlehmann> | <http://saveie6.com/>; I pooped a little. |
| 21:46 | <zewt> | tmi |
| 21:46 | <erlehmann> | i use reverse while loops in Js for performance :3 |
| 21:48 | <erlehmann> | <http://blogs.sun.com/greimer/entry/best_way_to_code_a>; |
| 21:48 | <zewt> | prediction, and only slightly exaggerated: in the future, android's browser becomes the new ie6 :| |
| 21:49 | <AryehGregor> | I just finished rewriting my spec so that it's slightly shorter, but produces dramatically better output. |
| 21:49 | <zewt> | (in older versions it's really bad--still is in some ways--and zillions of phones have been shipped with no mechanism to update the browser, short of carrier updates that never happen) |
| 21:49 | <AryehGregor> | Yay. |
| 21:50 | <Hixie> | zewt: thanks. I am replying to all the mail so far, started just after Lachlan's first e-mail. The rest of the mail predates the recent changes. |
| 21:50 | <erlehmann> | “Get the W3C standard changed to fit IE6” is listed as a goal of the site. Good job, Hixie ;) |
| 21:51 | <erlehmann> | zewt, i thought safari is the new IE6! |
| 21:51 | <zewt> | i don't know; i never use safari ever, except in its other webkit incarnations (such as ... android's browser :) |
| 21:52 | <zewt> | (but my annoyance with androidbrowser isn't really WebKit's fault) |
| 21:53 | <erlehmann> | my annoyance with safari isn't also webkit's fault. but you are right, android browser is an annoying UA, from a dev perspective. |
| 21:55 | <zewt> | android's browser is *really* bad at what I'd think would be a huge top priority: full-page web apps (eg. GMaps-style) |
| 21:55 | <zewt> | due to the magic-address-bar weirdness |
| 21:59 | <AryehGregor> | Google Maps has an app which works much better than the web page. |
| 21:59 | <jgraham> | erlehmann: (that fastest way to... article appears to be hopelessly out of date) |
| 22:00 | <zewt> | that's just what's so obnoxious--even Google making apps for things, instead of improving their browser so they can be done properly, as portable web apps |
| 22:02 | <zewt> | (at least on iOS's browser there's a mechanism to get rid of the address bar, even if it sort of sucks; on android you have to do all kinds of brittle magic tricks to convince it to scroll it off) |
| 22:03 | <erlehmann> | jgraham, thanks, i'll re-check it when i need it again. |
| 23:40 | <AryehGregor> | How do QR codes work? I tried taking a picture of one today, with my Nexus One, but apparently that doesn't work. |
| 23:40 | <AryehGregor> | I guess I need a separate app. |
| 23:40 | <gsnedders> | AryehGregor: Search the Android Market for barcode |
| 23:40 | <AryehGregor> | Feh. Should be built-in. |
| 23:41 | <AryehGregor> | I hate third-party apps. |
| 23:41 | <AryehGregor> | They tend to a) be poorly designed, b) request unreasonably many permissions. |
| 23:46 | <AryehGregor> | That's one reason I like Linux, I can get pretty much all the programs I need from the package repos. |
| 23:46 | <AryehGregor> | Some of those are poorly designed, but at least I know they're almost certainly not malicious or overly obnoxious. |
| 23:48 | <gsnedders> | AryehGregor: You have seen the perms that the built-in apps want when upgrading? /That/ disturbs me more than most third party stuff |
| 23:49 | <AryehGregor> | Yeah, but I don't care, because if Google wants to root my phone they can do it anyway. |
| 23:51 | <zewt> | well, largely a side-effect of the unfortunate permissions model: you have to request everything at once--it's impossible to request permission when features are used, so users can actually tell what each is for |
| 23:51 | <zewt> | so if you have a net-accessing feature that 1% use, everyone has to agree to it (or not use the app at all) |
| 23:55 | <AryehGregor> | I wish I could just disable the permissions selectively. |
| 23:55 | <AryehGregor> | That really annoys me. |
| 23:55 | <AryehGregor> | Anyway, this is interesting: http://blogs.msdn.com/b/ie/archive/2011/03/17/internet-explorer-9-network-performance-improvements.aspx |
| 23:55 | <AryehGregor> | Some of these are things I've never heard of. Do other browsers do all of that? |
| 23:56 | <AryehGregor> | Like caching which domains some domain includes resources for, and resolving them speculatively before you even send the HTTP request for the page content? |
| 23:56 | <AryehGregor> | I know Chrome does pre-resolution for links. |
