00:00:02 Nobody knows shit about Larrabee, so nothing can be predicted. 00:00:12 I doubt it can raytrace in realtime. 00:00:18 ehird, there was some talk from intel about ray trace on it 00:00:29 AnMaster: Even so. 00:00:37 If Intel can do it at 10fps, fine. 00:00:40 But an FPGA does it at 5. 00:00:45 And that's still wicked performance. 00:01:04 ehird, I want it done at 40 FPS or so 00:01:10 then I may be interested 00:01:11 AnMaster: HAHAHAHAHAHA 00:01:13 buy a time machine 00:01:21 AnMaster: it's impossible. This is not for using in games. 00:01:26 This is for people working on e.g. car designs. 00:01:39 ehird, true. But 5 FPS isn't "real time" 00:01:49 AnMaster: It does in 5FPS what normally takes an hour. 00:01:51 That is real time. 00:02:02 ehird, and can't you do it in parallel 00:02:16 Gee, that's lovely and meaningless. Let's throw "parallel" after every word. 00:02:32 ehird, iirc you can. So just throw more FPGAs at it 00:02:40 AnMaster: Yes, with a gigantic render farm, you could match the 5 FPS that a single FPGA handles. 00:02:48 And you say that FPGAs are slow? 00:02:57 ehird, um... I'm not comparing it to CPUs 00:03:10 rather to an array of Nvidia Tesla 00:03:32 AnMaster: Woo, for way more cost you can get way more power-hungry equipment that can match an FPGA. 00:03:34 or whatever is their fastest one currently 00:03:35 FPGAs are so slow. 00:03:49 ehird, FPGAs are slow compared to ASIC 00:03:52 Yes. 00:03:52 says wikipedia 00:03:55 Compared to ASIC. 00:03:56 But they are not slow. 00:04:02 ehird, which is what I was comparing to 00:04:04 anyway 00:04:09 5FPS ray tracing is not slow by any stretch of the imagination. You could do sorting networks in FPGA just fine. 00:04:14 why can't you then use SEVERAL FPGAs to speed it up 00:04:31 Ask them 00:04:35 maybe each rendering part of the frame 00:04:41 or rendering different frames 00:07:19 ehird, is it this http://www.eecg.toronto.edu/~fender/pdfs/raytrace_fender.pdf ? 00:07:26 was the first hit I found for ray trace FPGA 00:07:33 AnMaster: http://www.caustic.com/ 00:15:34 fizzie: what res is that lg thing again? 00:18:19 the TV? 00:18:56 no 00:18:59 the monitor he uses 00:19:03 it had a tuner right 00:19:06 thus a tv 00:19:07 no 00:19:12 that one is a different one 00:19:14 which one had a tiner 00:19:16 tuner* 00:19:20 i don't know and I don't care 00:29:34 AnMaster: Yes, mowing. As in "lawn mower". 00:29:50 ah 00:29:57 why is it called mower 00:30:06 looks like a typo for "mover" 00:30:16 (I know it isn't but...) 00:30:23 Erm. context? 00:31:01 ehird, some hours before 00:31:03 Because it's a machine to mow. 00:31:05 ehird: I went out to mow. 00:31:08 ah. 00:31:11 An hour ago. 00:31:14 pikhq, why "mow" 00:31:18 why not "mov" 00:31:29 etymology 00:31:58 AnMaster: Because what Edwin Budding decided to call it. 00:32:02 Erm. 00:32:03 That's what. 00:32:17 who 00:32:19 Why "mov 00:32:20 "? 00:32:26 He invented the mower. 00:32:29 ah 00:32:33 Edwin Beard Budding (1795–1846), an engineer from Stroud, England, was the English inventor of the lawnmower (1830) and adjustable spanner. [1] 00:34:53 Before that, one could maintain maintain a lawn using a scythe... 00:47:09 playing with CUDA will be fun 00:47:16 i'll multiply matrices and stuff. 00:48:26 1920x1200. 00:48:53 Bought it a bit before the unsurprising proliferation of 1920x1080 screens, which seems to have happened recently. 00:49:10 Oh, and 94 dpi if you want actual resolution-resolution. 00:49:11 fizzie: Right. I assume 1080p media just gets two lil' bars? 00:49:23 Yes. I don't mind the bars. 00:49:27 Those probably aren't 1920x1080; most PC monitors are 16:10, not 16:9. 00:49:27 Mind, the gap. 00:49:32 pikhq: Nope. 00:49:37 Real 1080p monitors. 00:49:39 There are many 16:9 PC monitors nowadays. 00:49:45 ehird: Oh, they decided to have 16:9 monitors? 00:49:47 But I don't like screens *that* wide. 00:49:49 It's the latest craze. 00:49:51 pikhq: For the HD craze. 00:49:57 Good; seems dumb to have 16:10 monitors. 00:50:00 But yeah, I want 1920x1200; it's the more monitor-y resolution. Games, etc. 00:50:03 pikhq: No way. 00:50:10 Sure, the ultra-wideness is nice for movies and shit. 00:50:14 But for actual computing? 00:50:21 16:9 is gonna seriously cramp you vertically. 00:50:31 16:10 was chosen because it had little black bars for 16:9 content. 00:50:32 ... That's it. 00:50:45 pikhq: I don't care how it was chosen; it works well. 00:51:08 I don't have a firm opinion on how much the difference actually matters; I rather like the fact that I could put my 1920x1200 screen next to the 1600x1200 4:3 screen, and get a nice not-non-rectangular desktop out of it. 00:51:44 16:9 is very good for movies and stuff. Very realistic and whatnot. But I like the additional vertical headroom— I deserve height, dammit. 00:53:06 I haven't actually tried any 1080 pixels high screens, so can't really comment. But I don't really grok what sort of real disadvantages the 16:10 ratio should have. (I haven't noticed any.) 00:53:36 fizzie: It's just got dumb reasoning is all. 00:53:53 Who cares about reasoning? It's the results that matter 00:54:41 wide screens are good for viewing multiple documents or terminals side-by-side 00:54:49 The ratios are so arbitrary anyway. I guess there's "native" 16:9 content (esp. direct-to-TV stuff), but do they actually squeeze released movie-movies to that ratio? 00:55:05 RodgerTheGreat: they're good for just about anything— but 16:9 is just too wide vs height to be usable IMO. 00:55:11 for computer usage 00:55:14 hm 00:55:22 fizzie: Movies are 16:9, yes. 00:55:25 In general. 00:55:27 On DVD and such that is. 00:55:32 Widescreen TV> 00:55:33 *. 00:55:56 That sounds strange; how do they do it, since the movies-in-a-movie-watching-place are something absurdly wide? 00:56:05 Flagrantly cropping? 00:56:12 Two 136x90 terminals side-by-side is my most common content on the 1920x1200 screen. 00:56:50 For a single full-screen web browser it's a bit disturbingly wide. 00:56:51 On this dinky "1050p" 16:10 20" screen I have three Safari windows with tabs, only one of which is really visible, and an IRC client in the lower-right corner. 00:56:55 They are frolicking. 00:56:59 And I refuse to use full-screen windows. 00:57:22 Well, I'm one of those tiling-wm crazies. 00:57:26 Funny, most movies I see on DVD and Blu-Ray are letter-boxed. 00:57:43 fizzie: There's also native (roughly) 16:9 movies. 00:57:53 00:57 pikhq: Funny, most movies I see on DVD and Blu-Ray are letter-boxed. ← really? 00:57:54 lameo 00:57:58 And also native 4:3 movies. 00:58:01 I don't really buy digital content so I didn't know 00:58:15 * ehird checks a site of pirates to see the kind of resolutions 00:58:26 Original aspect ratio, bitch. ;) 00:58:35 night 00:58:51 pikhq: Here's an odd one: 1824x992. 00:58:59 I own a single movie on DVD: Miyazaki's Totoro, bought for 6 eur from a sales bin, mostly because CAT BUS. 00:59:05 It's a blu-ray rip, cropped since there were black bars on *all 4 sides*. 00:59:12 (4:3 was the original Academy ratio, then they moved to nearly 16:9, then they went wider. 16:9 was chosen by the ATSC group as a compromise). 00:59:13 THE FAILURE, OH THE FAILURE. 00:59:15 ehird: *What*? 00:59:18 pikhq: yep 00:59:18 That is failure. 01:00:05 Hmm. A whole 76 pixels on the sides and 88 pixels on the top and bottom. 01:00:10 Why the hell? 01:00:24 Here's an actual blue-ray rip. 01:00:27 1920x1040. 01:00:30 WTFomatic. 01:00:41 ehird: Original aspect ratio. 01:00:47 Also, why the hell do modern HDTVs have overscanning? 01:00:54 Yes, that's right. Overscanning. 01:00:56 pikhq: Yeah, but, black bars, man. Even on 1080p screens. 01:01:23 pikhq: Ahh, our shitty 21" CRT TV does overscanning badly, I think. The sides get chopped off the image. 01:01:28 Least I think that's it. 01:01:38 An LCD 1080p TV will (in some cases) do minor upscaling so it can chop off a few pixels. 01:01:58 Upscaling a picture it can display *natively*. 01:02:19 pikhq: 16:9 DVDs seem to be 704x480. 01:02:26 That is uncomfortably low. 01:02:44 ehird: Uh... That's completely bizarre, given that that's the resolution of 4:3 DVDs. 01:02:51 16:9 DVDs are supposed to be 720x480. 01:02:54 And yes, it is low. 01:02:55 Oh, hm. 01:03:02 SDTV is low. 01:03:30 "Video: h.264 (4:3)" 01:03:32 Real helpful that 01:03:40 720x576 or something for PAL, one would assume. 01:03:47 fizzie: Yes. 01:03:58 pikhq: These 704x480 widescreens appear to be anamorphic. 01:04:07 Meaning...blackbarred. 01:04:20 ehird: As are 720x480 16:9 DVDs. 01:04:24 Wait, no. Hmm. 01:04:28 And 704x480 4:3 DVDs. 01:04:29 Ahhhhhh. 01:04:33 SDTV is anamorhpic. 01:04:38 pikhq: 704x480 16:9 anamorphic means "stretched". 01:04:46 As in, you have to squish it back to the proper 16:9 to play it. 01:04:54 But it's stored as squashed/stretched 704x480. 01:04:57 Why? No. Fucking. Clue! 01:05:10 You do realise that the same applies for 4:3, right? 01:05:16 AAAAAAAAAAAAAAAAAAAAAA 01:05:18 -!- inurinternet has quit (Success). 01:05:20 I HATE VIDEO FORMATS 01:05:22 SDTV with square pixels would be 640x480. 01:05:25 THEY ARE RETARDED 01:05:51 HDTV is an attempt to make it non-retarded. 01:05:56 Square pixels are so square. 01:05:58 ... And people are still being dumb with it. 01:06:23 We should just have square fucking pixels with unsquashed video in the resolution and aspect ratio it's actually going to be played back on. 01:06:27 GOD. 01:06:35 ehird: Yeah. 01:07:00 I'll give the DVD spec some slack, because it still has to deal with the standards set in the 50s. 01:07:44 But everything that's not SDTV? Come on, just use fucking square pixels and encode it in the right aspect ratio. 01:07:50 There is *no reason* to encode freaking black bars. 01:08:06 pikhq: If I recall correctly, my father was present at the consortium-meeting-things that defined DVD. 01:08:11 Representing some sort of company or another. 01:08:20 So I will take my anger out on him. :-P 01:08:30 ehird: I wouldn't. 01:08:39 He is clearly an agent of the devil. 01:08:43 Evidence: DVD. 01:08:49 The DVD has to deal with legacy formats. 01:08:54 Phooey. 01:09:27 720x480 is the only time I will accept anything but square pixels in the right aspect ratio. 01:10:18 I seem to have a video file that has the resolution 660x362, but I think that's trying to match original 1.82 aspect ratio in the content, scaled down to some rather random number. 01:11:17 fizzie: does mplayer automatically handel anamorphic shit? 01:11:18 it should 01:11:29 As far as I know it does. 01:11:44 Assuming the container format is intelligent enough to encode the aspect ratio that it should be played at. 01:11:45 ehird: It's the first one to handle it properly. 01:11:47 that's nice 01:11:59 It's the only one that handles it right on AVI. 01:12:03 mplayer seems to be very good at "give me a file, I'll play it properly" 01:12:08 Well, other ffmpeg-based player might also do it right. 01:12:15 Yeah, that's what mplayer's great at. 01:12:30 It's also a rather good encoder, if you're willing to mess with the command line. 01:12:54 It seems less good at "five me an URL, I'll play it properly"; at least there's been a couple of strange asf/mms/whatever-streams only VLC could play for me. 01:13:01 s/five/give/ 01:13:34 pikhq: you mean mencoder? 01:13:34 fizzie: You have to add the -playlist argument for those, I think. 01:13:40 ehird: Yuh. 01:13:44 I've always wondered how to rip a DVD without reencoding it. 01:13:52 As in, just gimme the damn data in the standard MPEG2 format! 01:14:11 "-oac copy -ovc copy" could be suitable mencoder magic. 01:14:24 mm 01:14:41 ehird: mplayer -dumpstream -dumpfile foo.vob 01:14:48 Doesn't even remux it. 01:14:49 pikhq: that's a .vob 01:14:52 not an .mpeg thing 01:14:56 Yes. 01:15:00 I want an .mpeg :-P 01:15:05 That's the DVD on-disk format. 01:15:07 But WITHOUT reconverting. 01:15:11 Just stuff-it-into-the-container. 01:15:14 Technically, .mpeg would also be a valid extension. 01:15:18 Oh? 01:15:21 A .vob is a valid .mpeg? 01:15:33 pikhq: is dumpstream fast? 01:15:35 A .vob is a MPEG transport stream. 01:15:36 it would seem like it'd go in realtime 01:15:45 also, I want a regular video-style .mpeg 01:15:48 like downloaded from the interwebs 01:16:02 Dumpstream says "instead of playing it, dump that shit to disk as fast as you can." 01:16:06 Ah. 01:16:15 But what about .mpeggggg 01:16:18 The web says: "Program streams are used on DVD video discs and HD DVD video discs. The file extensions are VOB and EVO respectively. Blu-ray Discs use a transport stream (TS) format with an additional 4 byte time code added to the beginning of each TS packet." 01:16:31 And it claims that a .vob file is actually an MPEG-PS. 01:17:00 Not my area of expertise, though. I do know that the DVB broadcasts we get from digital TV are MPEG transport streams. 01:17:12 "mplayer -dumpstream -dumpfile foo.mpeg dvd://1" might work then. 01:17:51 At least mplayer'll play the result. :p 01:17:55 Seriously have to sleeps now. 01:18:01 Bai. 01:18:06 Baa-baa. 01:19:12 Erm, right. MPEG-PS, not TS. 01:19:26 pikhq: wat 01:19:38 Just aggreing with fizzie. 01:20:09 ah. 01:30:09 -!- kerlo has joined. 01:30:15 Well, I guess I have to find this out at some point. 01:30:19 What do I do with a .cpp file? 01:30:39 g++? 01:33:09 kerlo: You always surprise me with your lack of programming knowledge... 01:33:23 how can you not know how to compile a .cpp file and yet own and administrate a virtual server? 01:34:49 pikhq: gracias. 01:35:00 kerlo: Windows guy? 01:35:03 ehird: by paying $25 a month and yet not having much experience. 01:35:11 pikhq: si. 01:35:25 kerlo: Fix that. 01:35:37 * pikhq tosses kerlo a nickle for the blank media 01:35:38 Uh oh, now we have to listen to a windows guy justify his choice 01:35:44 Take cover! 01:36:19 ehird: Almost as bad and archaic as hearing a SLS guy justify his choice. 01:36:22 :p 01:36:31 pikhq: No, no— Yggdrasil. 01:36:33 SLS? 01:36:40 Sgeo: softlanding linux system 01:36:42 ancient linux distro 01:36:43 1992 01:36:44 Sgeo: The original distro. 01:36:56 "Yggdrasil was the first company to create a Live CD Linux distribution" 01:36:57 It's very fucking old. 01:36:58 Before 1995. 01:37:10 By "old", I mean "Slackware is a *fork* of it". 01:37:11 They charged for it 01:37:14 :-) 01:37:22 linux boot cd in 1992 01:38:03 ehird: I believe that last time, I justified my choice because you asked me to. 01:38:12 So no, you don't have to listen to me justify my choice again. 01:38:13 did you? 01:38:14 I don't recall 01:38:17 I don't have much of a memory 01:38:22 oh 01:38:28 kerlo: It was "Windows does everything I want", right? 01:38:30 kerlo: You still use Windows, so it can't have been a valid justification. 01:38:34 Only the worst argument for using Windows ever. 01:38:47 Actually, you listed... about 2 things Windows can't do that Linux could. 01:38:52 Which makes it an even worse argument. 01:39:11 I think it was that. 01:39:40 pikhq: Let's woe together about Windows users. 01:39:42 Yes, woe is a verb. 01:40:04 ehird: Let's. 01:40:08 * ehird woes. 01:40:19 * pikhq laments 01:40:40 lament: let's woe. 01:41:19 * pikhq woes 01:41:29 Should I find this funny or be sympathetic? 01:41:38 kerlo: You should woe. 01:41:46 * kerlo woes. 01:41:53 kerlo: Woe at yourself. 01:42:04 Then fix cause-of woe-from us. 01:42:53 * kerlo woes, and does not switch to Linux. 01:43:09 * pikhq beats kerlo with a real operating system. 01:43:16 Pick one, they're all more real than Windows. 01:43:25 Yes, including DOS. 01:43:46 * ehird punches kerlo and stomps on him. Drastic measures for (a) coloured people (as long as the colour isn't white), (b) immigrants and (c) Windows users. 01:44:02 The scummery and villanry of modern society! 01:44:30 Also, (d) Muslins. Death to fabric. 01:46:10 Ahem. 01:48:03 * Sgeo switched from Linux to Windows. What does that make me? 01:48:14 Sgeo: Benjamin Button. 01:48:16 Sgeo: Masochist. 01:48:30 pikhq: I prefer mine. 01:48:32 Sigh. I hate web sites where the password you have to use to log in is different from what you set your password to when you register. 01:48:48 what 01:49:08 I picked the password foo\bar. The account activation email told me my password was foo\\\\bar. My actual password ended up being foo\\bar. 01:59:04 -!- Gracenotes has joined. 02:00:46 bye 02:00:48 -!- Corun has quit. 02:16:49 -!- Gracenotes has quit (Read error: 104 (Connection reset by peer)). 02:18:55 -!- Gracenotes has joined. 02:30:29 -!- coppro has joined. 02:45:59 -!- bsmntbombdood_ has quit (Read error: 60 (Operation timed out)). 02:52:26 http://icehot.files.wordpress.com/2009/05/humor-linux.jpg 02:57:17 -!- bsmntbombdood_ has joined. 03:27:02 -!- inurinternet has joined. 03:45:49 -!- GregorR-L has quit (Read error: 60 (Operation timed out)). 04:14:12 mplayer will play _anything_ 04:16:34 I thought that was VLC? 04:17:23 -!- GregorR-L has joined. 04:23:45 -!- RodgerTheGreat has quit. 04:30:49 meh. mplayer's scaletempo has failed for me 04:31:22 compiling... different flags... repository needs updates... nonfree packages... o_o 04:31:24 why would you want to do that? 04:31:54 so I can listen to audio and watch video at a rate I find more engaging 04:32:09 that sounds awful 04:32:26 better to spend 30-something minutes on a lecture than an hour 04:32:47 unless you are listening to speech manybe 04:32:50 it is wonderful 04:33:18 for books on tape and video or audio lectures for computer science, not movies or anything. 04:33:27 or math 04:43:38 -!- GregorR-L has quit (Read error: 60 (Operation timed out)). 04:47:58 but, alas, no such option exists 04:48:14 so I have to listen with the pitches screwed up, like a sixth or seventh up. 04:55:24 how does it work otherwise? 05:35:18 Gracenotes, some MIDI players do what you want >.> 05:35:37 bsmntbombdood_, speeding it up and lowering the pitch in the opposite direction? 05:58:44 Sgeo: how do you lower pitch? 05:59:34 bsmntbombdood_, I'm assuming it's possible. Maybe it isn't without slowing things down 06:16:32 one can lower pitch by transposing one pitch to another 06:16:38 (just got out of shower) 06:17:02 with MIDI files, this is very easy, since the internal representation is notes, not the sounds themselves 06:17:32 for wav/mp3/etc. there are various properties of sound you can exploit to do it... 06:18:03 obviously it's possible, since mplayer has a feature, just not here >:| 06:18:36 i don't get it 06:20:31 -!- kar8nga has joined. 06:26:12 bsmntbombdood_: about the pitch? 06:26:19 yeah 06:27:54 afaik, you can either speedup+raise pitch or slow down+lower it. And you can calculate how much you raised/lowered the pitch mathematically and reverse that amount. 06:28:49 there is possibly an algorithm that combines these steps.. otherwise the quality might be lowered. I am not terribly familiar with sound science >_> 06:29:26 MIDI files are different entirely: instead of storing WAVs, they store discrete notes, noting exactly where they start and end and splitting them up into separate tracks 06:29:34 *waves 06:30:24 for example, a note might have the pitch C4. So to transpose it down an octave, just change it to C3, and likewise for all other notes. 06:31:25 or to speed up, just change the tempos used throughout the song accordingly. 06:31:37 which is an explicit variable, for tempo. 06:32:18 well obviously midi is easy 06:33:45 I know more than a few people who speed up lectures this way 06:34:22 are able to absorb content faster. It works for me too, but without the pitch correction it can get annoying. 06:34:50 well, sure 06:41:36 i just don't understand the mechanics 06:42:44 I can't say I do entirely either. From a musical perspective at least (playing piano), pitch can be modeled nicely mathematically 06:50:18 Should I try M*U*S*H? 06:55:09 http://www.somethingawful.com/d/news/wolfram-alpha.php 06:56:47 -!- Sgeo has quit ("Leaving"). 06:59:38 -!- kar8nga has quit (Remote closed the connection). 07:38:13 Yes, you can do both the speed-up-without-pitch-changes and alter-pitch-without-tempo-changes operations to generic digital audio, though you're always going to get some artifacts. 07:41:16 http://en.wikipedia.org/wiki/Audio_timescale-pitch_modification seems to be a (rather variable-quality) article about it. 07:59:59 -!- clog has quit (ended). 08:00:00 -!- clog has joined. 08:05:22 Speaking of memory cards (though the discussion ended a long ago, I guess), it seems that they've started to make 16 GB MicroSD(HC) cards since the last time I looked (when they went up to 8). That's the crazy. 08:11:27 -!- Slereah has joined. 08:17:05 -!- tombom has joined. 08:43:14 -!- bsmntbombdood has joined. 08:44:02 -!- bsmntbombdood_ has quit (Read error: 104 (Connection reset by peer)). 08:44:09 ....my computer just crashed 09:03:25 -!- GregorR-L has joined. 09:21:04 -!- oklodok has joined. 09:21:35 -!- oklodok has set topic: international tub http://tunes.org/~nef/logs/esoteric/. 09:22:28 The international tub of esoteric language developers. 09:24:29 * bsmntbombdood soaps of oklodok 09:24:48 i wonder what all of his suffixes mean 09:24:55 -!- oklodok has set topic: The international tub of esoteric language developers. Consider feasting on http://tunes.org/~nef/logs/esoteric/. 09:25:17 most of my suffices are just english words 09:25:24 this one is a duck. 09:25:37 i also have other birds like kok 09:27:57 fok is a hawk, f and h are the same character in finnish, for some values of lying my ass off 09:35:42 What's oklowob, then? 09:38:07 i'm not nearly as good with birds as my subconscious, no idea. 09:38:24 warbler or something? 09:55:30 -!- oerjan has joined. 09:56:36 hi oerjan 09:56:56 was the food you ate last bitter 09:56:56 hi oklodok 09:57:17 * oerjan just finished a slice with strawberry jam, so no 09:57:24 mmm 09:57:36 ima eat like noodles and meat today. 09:57:47 * oerjan now starts on the liver paté 09:57:57 i don't consider that bitter either 10:00:32 oh noes we're a tub now 10:01:55 well would you rather be in a cub or a rub? perv. 10:02:32 shit, the ai wins me in two rounds. 10:03:02 and here i thought i was special 10:03:21 rub a dub dub 10:20:57 How would you like to rub-a-dub-dub in my tub 11:15:51 oklodok, it wasn't bitter. It was butter (well, mostly bread, but that isn't as funny) 11:37:20 -!- oklodok has quit (Read error: 110 (Connection timed out)). 12:00:36 -!- KingOfKarlsruhe has joined. 12:11:54 -!- oerjan has quit (anthony.freenode.net irc.freenode.net). 12:11:54 -!- kerlo has quit (anthony.freenode.net irc.freenode.net). 12:12:33 -!- oerjan has joined. 12:12:33 -!- kerlo has joined. 12:16:38 -!- oerjan has quit (anthony.freenode.net irc.freenode.net). 12:16:38 -!- kerlo has quit (anthony.freenode.net irc.freenode.net). 12:18:36 -!- oerjan has joined. 12:18:36 -!- kerlo has joined. 12:26:10 -!- tombom has quit ("Peace and Protection 4.22.2"). 12:51:18 -!- ais523 has joined. 13:12:35 -!- MizardX has quit ("What are you sinking about?"). 13:31:05 hm 13:31:09 hi ais523 13:31:15 p[-43]=1; 13:31:15 p[-36]=11; 13:31:15 p[1]=6; 13:31:15 -do { 13:31:15 +{int i;for(i=0;i<1;i++){ 13:31:15 p[-43]+=255; 13:31:17 p[-36]+=16; 13:31:19 -} while (p[-43]); 13:31:21 +}} 13:31:29 I wonder why lostking contains a loop around that at all 13:31:30 ... 13:31:48 hmm... isn't that a repeat-once loop? 13:31:57 which is not really a loop at all 13:32:10 it could be an artifact of the way BFBASIC works 13:32:12 ais523, hm an if you mean? 13:32:41 ais523, however see the index cell is set to a constant just above 13:32:49 yes 13:32:52 so in that case it is a "dead if, once" 13:32:55 which is why the loop always iterates exactly once 13:32:57 agreed 13:32:59 dead if 13:33:09 hm 13:33:29 and getting rid of the loop bit in it will help a lot 13:33:30 since 13:33:34 p[0]=p[-36]; 13:33:34 p[-36]=0; 13:33:36 right afterwards 13:35:44 ok, this is weird 13:35:48 ais523, what is 13:35:53 the Windows computer right in front of me, which I'm not logged onto 13:35:58 is being logged into remotely by the admins 13:35:59 to do something 13:36:02 heh 13:36:08 ais523, spyware? 13:36:12 and I saw them type their username, and a concealed password 13:36:14 and why does it show up on the screen 13:36:22 and it's not spyware; it's remote keyboard/mouse control, but legitimate 13:36:29 because they installed the software to do that deliberately 13:36:29 ais523, wait... can you interact with it too? 13:36:41 AnMaster: apparently so, I just moved the mouse a few pixels 13:36:47 -!- Corun has joined. 13:36:48 but I'm not going to 13:36:51 ais523, ah 13:36:59 they logged in on a non-admin account, I think 13:37:03 oh well 13:37:17 but even if it was an admin account, I wouldn't want to get in trouble by grabbing control of the logged-in session 13:37:24 ais523, right 13:37:32 ais523, but it seems insecure 13:37:43 it is, rather 13:37:57 blame Windows for not really implementing remote sessions with a GUI 13:38:15 I imagine Windows admins would find the idea of ssh with X forwarding really alien 13:38:25 even though it would be a much easier way to accomplish what they're trying 13:38:37 hmm... the mouse pointer is teleporting, rather than moving smoothly 13:38:49 so it's not a direct interface 13:38:54 -!- lifthras1ir has joined. 13:38:59 ?TRACE({"STUPID PROGRAMMER^W^WREPEAT ONCE LOOP FOUND.",H}), 13:39:03 that seems fitting 13:39:50 -!- lifthrasiir has quit (Read error: 60 (Operation timed out)). 13:42:59 AnMaster: it seems that they loaded Excel, looked at the tools menu (presumably to verify a particular item was there), then logged off again 13:43:07 well, not quite, they're at the logoff confirm screen atm 13:43:14 mhm 13:43:22 ais523, that sounds extremely strange 13:47:45 hm handling those dead repeat once/if loops had a large cascading effect on lostking 13:48:03 from 74710 lines of output to 73376 lines. 13:48:08 that's not really that large 13:48:12 given how big it is 13:48:22 Misread "^W^WREPEAT ONCE LOOP FOUND" as "WEAR HATS [something unintelligible]" 13:48:33 ais523, 733K -> 719K 13:48:39 yes, exactly 13:48:43 ais523, but then the repeat loops are rather large 13:48:48 ais523, yes exactly to who? 13:49:03 fizzie, crazy misread :D 13:49:32 AnMaster: to you 13:49:37 ah 13:49:38 as in, 719 isn't much smaller than 733 13:51:05 next up is (if no IO, not deep and so on) flattern repeat loops. Polynom style. 13:51:20 oh and being able to convert more types to repeat loops 13:51:25 this is far from finished 13:51:46 currently it only handles index_diff = {add,255}, not even {add,1}. 13:57:32 -!- oerjan has quit ("Bus-iness"). 14:01:48 -!- Slereah has quit (Remote closed the connection). 14:20:41 -!- FireFly has joined. 14:24:17 -!- BeholdMyGlory has joined. 15:12:51 -!- MizardX has joined. 15:28:55 -!- lifthras1ir has changed nick to lifthrasiir. 15:58:57 -!- coppro has quit (Read error: 110 (Connection timed out)). 15:59:01 lifthrasiir, did you see what I said about esotope-bfc being broken on linux? 15:59:17 you can't give more than #!/path/to/program OneParameter 15:59:25 so /usr/bin/env python -O didn't work 16:03:10 -!- inurinternet has quit (Connection timed out). 16:04:00 AnMaster: okay, got it 16:05:49 -!- Corun has quit ("Leaving..."). 16:06:37 -!- Corun has joined. 16:06:53 -!- Corun has quit (Remote closed the connection). 16:10:31 -!- inurinternet has joined. 16:13:16 lifthrasiir, are you able to detect some finite loops even when the constant isn't known btw? 16:13:54 AnMaster: first transform the loop into if (some_condition) while (1); actual_loop; and get rid of if(some_condition) ... codes later. 16:14:07 lifthrasiir, hm? 16:14:34 -!- Slereah_ has quit (Read error: 60 (Operation timed out)). 16:14:35 well 16:14:47 lifthrasiir, I mean, if you don't know constant loop can in general be either finite or infinite 16:14:54 AnMaster: not yet then. 16:14:58 lifthrasiir, see http://rafb.net/p/LeOL2E65.html 16:15:19 lifthrasiir, I brute force tested it for all pairs {1..255,1..255} :P 16:15:45 if the target cell is constant, esotope-bfc removes the redundant if statement; but that's all for now. 16:16:08 -!- Slereah has joined. 16:16:26 lifthrasiir, I can mark a loop like "input [add y to cell]" as finite now, if y is an odd number. 16:16:46 lifthrasiir, which means I can move IO nodes over it 16:17:05 like, loop output_constant -> output_constant loop 16:17:31 right, finite and pure (no-IO) loop is free to move 16:17:40 lifthrasiir, yep 16:18:46 lifthrasiir, and I track if loop has input and/or output (for some reason I decided to track them as two separate parameters thinking "might be useful to know it has input but no output", but I can't think of a reason for it any longer...) 16:20:33 lifthrasiir, oh and here is a really trivial checker for is finite/infinite, less elegant than your code. But I understand how it works. 16:20:34 is_inf(Const, Diff) -> is_inf(Const, Diff, (Const + Diff) rem 256). 16:20:34 is_inf(_Const, _Diff, 0) -> false; 16:20:34 is_inf(Const, _Diff, Const) -> true; 16:20:34 is_inf(Const, Diff, Cur) -> is_inf(Const, Diff, (Cur + Diff) rem 256). 16:22:30 probably slow. but when testing I detected no major slow down. about half a second extra for lostking, and since I'm already taking around 40 seconds for it... 16:25:56 -!- Dewio has joined. 16:26:37 -!- Dewi has quit (Read error: 60 (Operation timed out)). 16:32:09 -!- kar8nga has joined. 16:39:39 -!- KingOfKarlsruhe has quit (Remote closed the connection). 16:44:48 -!- ais523 has quit (Remote closed the connection). 17:22:30 -!- M0ny has joined. 17:23:00 plop 17:23:21 -!- ais523_ has joined. 17:25:55 -!- ais523_ has changed nick to ais523. 17:28:51 -!- kar8nga has quit (Remote closed the connection). 17:33:24 -!- ais523 has quit (Read error: 54 (Connection reset by peer)). 17:34:35 -!- ais523 has joined. 17:38:22 -!- KingOfKarlsruhe has joined. 17:41:36 -!- Corun has joined. 17:59:36 -!- Corun has changed nick to Corun|away. 18:03:25 ais523, there? IIRC you wondered what optimisations were broken by treating [>>[[-]++]] as balanced for example? 18:03:35 that specific one: probably none 18:03:35 yes, I did 18:03:37 but what about 18:03:59 [>+[>]++[[-]++]] 18:04:21 ais523, because "balance" for me means that I can simply move > ahead and offset the instruction 18:04:24 which will break in that cas 18:04:27 case* 18:04:28 like: 18:04:52 turning >> [>+[>]++[[-]++]] into [(offset 2) >+[>]++[[-]++]] >> isn't valid 18:05:06 AnMaster: [>[-]>[>]+[<]<[-]] is balanced 18:05:12 but doesn't fit your balance optimisatoin 18:05:20 *optimisation 18:05:23 where offset means that p+=1; p[0] turns into p[1] p+=1; 18:05:34 and actually, in that case it is, because you do no I/O between the time the opimisation messes up and the infinite loop 18:06:13 ais523, in general it breaks though. I managed to generate a test case where it ends up accessing p[-1] in that case due to shifting the "p+=" forward 18:06:40 ah, ok 18:06:53 ais523, well a[-1] in fact. Out of bounds. 18:06:57 where a is the actual array 18:07:13 +>+>+<<[>]>>>> ,[ <,[,.]>[-]+ [++>->>>>,[,]<<<<-<]< ,] ,. 18:07:20 well, ok 18:07:28 you probably need a different term than "balanced" 18:07:35 maybe "known-offset" 18:07:37 the first bit is there only to make the "initial memory is all zero and use that knowledge" pass give up before 18:07:42 so it is easier to see what happens 18:08:28 -!- Corun|away has changed nick to Corun. 18:08:44 AnMaster: in [-]+[>[-]] the second loop isn't balanced, but it still has a known offset 18:08:55 (although, of course, nobody would deliberately write that particular program like that) 18:09:03 hm 18:09:27 ais523, I don't think even lostking contains that! 18:09:35 and lostking is horrible code 18:09:40 I think that what you need for that optimisation is not that the current loop is balanced, but that all loops it contains are balanced 18:09:56 even [>+] could be optimised with that optimisation 18:09:59 in fact you could optimise lost king by running it through an optimiser and outputting it back to bf 18:10:01 :) 18:10:01 LostKng is horrible, but hey, it's auto-generated. 18:10:13 instead of p+=1; *p++; you could do p[1]++; p+=1; 18:10:23 of course, you don't gain anything directly in such a short loop, but you might if it were more complicated 18:10:24 AnMaster: A dead-code elimination pass alone does wonders. 18:10:25 ais523, I do that already inside the loop 18:10:27 say [>+>+>+] 18:10:40 AnMaster: well, that optimisation has nothing to do with the loop being balanced or not 18:10:49 just to do with the loops /inside/ that loop being balanced 18:11:37 -!- BeholdMyGlory_ has joined. 18:12:55 ais523, my shifter pass, which is peephole style (operates on pairs of instructions), does "don't ever shift across unbalanced loops, shift pointer moves forward across everything else, don't two IO instructions across each other, when possible sort by offset" 18:13:10 it also recurses into loops to try to shift internally in them 18:13:38 getting rid of created >< and such is up to the combiner pass next time it runs. 18:13:39 AnMaster: well, I don't think unbalanced loops with embedded infinite loops mess that algorithm up 18:13:52 and that reminds me of gcc-bf, actually 18:14:01 I'm not entirely sure if it cleans up >< created as a result of that sort of thing 18:14:04 -!- BeholdMyGlory__ has joined. 18:14:04 ais523, I have a almost-DSL for it 18:14:13 AnMaster: like OIL? 18:14:27 ais523, more like you would make a DSL in scheme. 18:14:34 -!- BeholdMyGlory has quit (Nick collision from services.). 18:14:58 -!- BeholdMyGlory__ has changed nick to BeholdMyGlory. 18:15:07 I did a table describing all possible combinations, then I seralised it into a function returning a set of actions given two instructions. 18:15:23 err... you serialised a table into a function? 18:15:28 I don't think that's what "serialised" means 18:15:37 ehird: if you're here, please now laugh at either me or AnMaster 18:15:37 ais523, well I don't know a better word 18:15:41 it is not the right one 18:15:45 possibly 18:16:02 you could try the general "compiled", I suppose 18:16:14 or "hardcoded", even, although that might be ambiguous 18:16:20 or just the even more general "converted" 18:16:26 anyway the function returns false | {true, nochange | TranslatorFunc} | {maybe, TranslatorOrFalseFunc} 18:16:44 it just gets instruction names, no details like offsets and such 18:16:56 ais523, and I hand compiled it I guess 18:17:06 by trying to create a minimal set of rules 18:18:00 ais523, http://rafb.net/p/zosmmI54.html was the table. Haven't updated it after I added if/repeat/loop_inf 18:19:24 and and example from the serialising: 18:19:26 %% Swap mov ahead always. 18:19:26 should_swap({_,_}, {mov, _}) -> false; 18:19:26 %% Offset shouldn't change for out_const. 18:19:26 should_swap({mov, _}, {out_const, _}) -> {true, nochange}; 18:19:26 should_swap({mov, _}, {_,_}) -> {true, fun shift_mov/2}; 18:20:09 where the second parameter is used for loops to say if they are finite or infinite (or unknown), if they contain IO and so on. 18:20:20 loops and other blocks* 18:21:58 ais523, you know that fingerprint specification file thing that cfunge uses? 18:22:04 would you call that a DSL or not 18:22:04 yes... 18:22:11 and not really, it's more a file format 18:22:17 hm 18:22:20 true 18:22:22 would you call PNG a domain-specific language? 18:22:32 ais523, actually: maybe 18:22:45 though not a plain text such then 18:22:47 I wouldn't 18:22:58 ais523, hm yeah you can't compute with it indeed 18:23:08 ais523, what about svg then 18:23:11 that is definitely DSL 18:23:23 it's more a markup system + a scripting language 18:23:28 and JavaScript isn't very domain-specific 18:23:45 as for whether XML's a language or not, that's probably flamewar-worthy, so I won't go into it 18:24:02 XSLT is a langauge though 18:24:26 you can't possibly disagree about XSLT being a DSL 18:24:49 XSLT, I agree with you 18:26:02 ok... this is strange how did an loop_inf end up with loop data saying it is unknown if it is finite or not... 18:28:43 ais523, that [-]+[>[-]] ... Have there been much work on optimising unbalanced loops in BF before? 18:28:52 I don't think so 18:28:57 hm... 18:29:02 I do it partly 18:29:04 although, I've been thinking about the problem a lot in relation to gcc-bf 18:29:06 -!- BeholdMyGlory_ has quit (Connection timed out). 18:29:10 where most of the loops are unbalanced 18:29:27 the problem is to be able to demonstrate that the IP is known most of the time anyway 18:29:34 like propagating "infinite loop" upwards for unbalanced loops that contain infinite that will always be run. 18:29:58 though actual infinite loops I can only detect directly for balanced loops otherwise 18:31:00 ais523, you could put a bound on [>] for example by building a tri-state map containing "known set to constant, possibly clobbered, unknown" 18:31:24 example: +>+>+><<<[>] can be solves that way 18:31:30 AnMaster: yes 18:31:47 that is assuming at start of program or the values of those cells known to not be 254 due to something else 18:31:52 err 18:31:53 255 18:31:56 not 254 18:31:57 in gcc-bf, knowing that every third cell is usually set to 1, for instance, is very useful to do various sorts of optimisations 18:32:08 ais523, "usually"? 18:32:16 that doesn't help a lot 18:32:22 -!- inurinternet has changed nick to johnnyfive. 18:32:27 AnMaster: if they were left at 1 forever, they wouldn't be useful at all 18:32:39 but they only have different values for short periods of time 18:32:41 and only one does at a time 18:32:43 ais523, how could compiler know when it was at 1 then 18:33:00 well, it always is in the outer loop 18:33:06 although actually proving that might be rather difficult 18:33:08 ais523, does the generated bf contain any dead code? 18:33:24 AnMaster: it could do, but that would be if there was dead code in the asm 18:33:26 -!- BeholdMyGlory has quit (Connection timed out). 18:33:38 ais523, so no possibly BfBASIC style dead code? 18:33:39 and you'd need to investigate every function pointer in the input C to make sure it never called into that code 18:33:53 ais523, -fwhole-program 18:33:59 AnMaster: again, possible if you have something like a while loop which never iterates 18:33:59 -!- johnnyfive has changed nick to inurinternet. 18:34:00 could possibly do that 18:34:14 ais523, well sure, if the input C contains dead code 18:34:31 but I meant, will you ever generate anything like [-]+[->+++<] 18:34:32 or such 18:34:35 why are you so bothered about this, anyway? 18:34:39 -!- oerjan has joined. 18:34:51 and it might at the moment, I'm not sure how much I optimise 18:35:00 it's quite likely you could even generate >< with the current code 18:35:19 ais523, when BF code is so bad it can be optimised at the pure BF code level then I get rather uncomfortable 18:35:34 AnMaster: well, I'm aiming for something that works first 18:35:47 and the optimisations are likely to be peephole rather than global 18:35:48 -!- inurinternet has changed nick to sighcomputer. 18:35:54 ais523, lostking contains lots of dead code, and I don't detect all of it yet. 18:35:54 as in, that gcc-bf is missing 18:36:22 -!- sighcomputer has changed nick to inurinternet. 18:36:29 ais523, You mean you need peep hole optimisers? 18:36:39 yep 18:36:40 May I recommend erlang, it is truly awesome for those 18:36:44 but they're trivial enough to hack onto the end 18:36:51 and may I recommend Perl, it's even more awesome for those 18:36:53 it fits the pattern matching of erlang perfectly 18:36:58 -!- inurinternet has quit. 18:37:12 AnMaster: does erlang have pattern matching as powerful as Perl regexes? 18:37:12 optimise([#bfn{ ins = add, off = Offset, val = V2 }|T], 18:37:12 [#bfn{ ins = add, off = Offset, val = V1 } = A|Result]) -> 18:37:12 optimise(T, [A#bfn{ val = V1 + V2 }|Result]); 18:37:16 -!- inurinternet has joined. 18:37:18 :P 18:37:56 ais523, no, but it does have PCRE. Don't think it works at language level currently. Though I heard some rumours about that being considered. 18:38:09 -!- tombom has joined. 18:38:28 07:44 bsmntbombdood: ....my computer just crashed ← did you do something wrong? 18:38:50 ehird, what OS did he run 18:39:55 ais523, btw I recently read an interesting article about using superoptimiser to generate better peep hole optimisers. 18:40:05 an optimiser optimiser? 18:40:27 ais523, http://cs.stanford.edu/~sbansal/pubs/asplos06.pdf (linked from http://en.wikipedia.org/wiki/Superoptimization) 18:40:28 and I've had a couple of kernel panics over here before too, I think the wireless driver was at fault 18:40:34 but none recently 18:41:03 I have rougly one or two kernel panics / year. 18:41:14 AnMaster: what causes them? 18:41:18 17:15 AnMaster: I did a table describing all possible combinations, then I seralised it into a function returning a set of actions given two instructions. 18:41:18 17:15 ais523: err... you serialised a table into a function? 18:41:20 17:15 ais523: I don't think that's what "serialised" means 18:41:22 17:15 ais523: ehird: if you're here, please now laugh at either me or AnMaster 18:41:24 /me laughs at AnMaster 18:41:32 AnMaster: I think he's using Debian, and I don't care how many kernel panics you have. 18:41:39 ais523, usually nvidia or vmware drivers. Once buggy USB in kernel (around 2.6.8 or so) 18:41:51 "Wolfram Alpha v Google: Which is better? 18:41:52 Wolfram Alpha is the new computational search engine that could revolutionise the way we use the web. But is it good enough to replace Google as the most popular online search engine?" 18:41:54 *smashes head against brick wall* 18:42:03 ehird, I was talking to ais 18:42:15 pj 18:42:16 oh 18:42:16 I know I once managed to compile a division by zero into a custom kernel I was compiling 18:42:21 ehird, about how many kernel panics I had 18:42:21 so that panicked every single time 18:42:24 well ais523 was talking about bsmntbombdood's panic too 18:42:25 I never directed it to him 18:42:30 GregorR: GregorR-L: here? 18:42:30 ais523, hehe 18:42:34 and I don't care about ais523's crashes either 18:42:38 :p 18:42:46 ehird: but AnMaster does, I think 18:43:01 ais523: only so he can comment on how much more stable his system is 18:43:02 I don't run custom kernels on my own hardware, but this was for a project where we needed a massively small kernel 18:43:03 ais523, oh and of course I had a "panic, boot drive not found" when I messed up in grub on new installs. 18:43:08 happened a few times 18:43:14 AnMaster: you mess around with grub by hand? 18:43:22 my package manager sorts that out for me 18:43:29 ais523, um... grub-install misdetected what hd0 and hd1 was 18:43:31 :P 18:43:45 ais523, so even if I had used a package manager to do it, it wouldn't have helped 18:43:52 http://i.somethingawful.com/u/dannymanic/wolfram/wolfram04.gif ← oh, SA. 18:43:54 it automatically generates a sane grub configuration based on which packages you have installed 18:44:09 * ais523 vaguely wonders what it would do if I tried to uninstall the kernel; maybe default to booting memtest86 18:44:13 ais523: AnMaster messes around with _everything_ by hand 18:44:24 ehird: you can mess around with the auto-grub thing by hand too 18:44:27 just most people don't bother 18:44:33 ais523, basically linux live cd saw /dev/sda (SATA) and /dev/hda (PATA), grub-install decided sda was hd0. When booting it had instead decided hda was hd0 18:44:42 ais523, was ages ago 18:44:50 ais523: but that requires he let something else do trivially automatable work! 18:45:36 -!- MizardX has quit ("What are you sinking about?"). 18:46:08 ais523, and this could not possibly have been solved by a package manager. However. After switching to a new mobo (the old one broke, on warranty), I haven't had issues with reversed interpretations in grub 18:46:17 different mobo model in case you wonder 18:46:48 Speaking of the somethingawful W|A test, here's one particular input that went around IRC too: 18:46:49 W|A "2 girls x 1 cup the meaning of life": Input interpretation "2 girls × 1 cup × answer to life, the universe, and everything", result "84 cup girls". 18:47:03 That's a... rather interesting unit. 18:47:08 :D 18:47:22 um 18:47:26 what is up with the google logo 18:47:33 fizzie: http://www53.wolframalpha.com/input/?i=2+girls+1+cup gives it too 18:47:35 AnMaster: hover over 18:47:38 it changes all the time... 18:47:41 clicking on it gives me "missing link found" 18:47:44 * ais523 looks 18:47:47 surely you've noticed google's holiday logso 18:47:49 It's about the new fossil discovery, I guess. 18:47:49 logos 18:47:52 fizzie: yes 18:47:59 (this is evidence about how rarely I use Google, and how even more rarely AnMaster uses Google) 18:48:01 ehird, yes.. but which one is this 18:48:04 total baloney of course, the world is 6000 years old and was created in 7 days 18:48:09 ais523: AnMaster uses google all the time 18:48:11 AnMaster: hover ove 18:48:12 r 18:48:15 ehird: ah, ok 18:48:18 AnMaster: also 18:48:23 click the links on the missing link found? 18:48:27 it always links toa relevant search 18:48:28 ehird, Oh about darwin 18:48:37 ehird, clicking on it -> http://www.google.com/search?q=missing+link+found&ct=missinglink&oi=ddle 18:48:42 yes? 18:48:45 look at the articles 18:48:46 they're relevant 18:48:49 ehird: that "missing link" fossil had a similar release to Wolfram Alpha 18:48:49 oh 18:48:50 I was 18:48:52 hype-wise 18:48:53 thinking 18:48:55 so I don't entirely trust it 18:48:57 missing link == 404 18:48:58 :P 18:49:05 AnMaster: heh 18:49:10 Well, the "news results" articles are reasonably relevant, at least some of them. 18:49:15 ais523: I don't trust it because 18:48 ehird: total baloney of course, the world is 6000 years old and was created in 7 days 18:49:28 ehird: you don't believe that 18:49:28 -!- kar8nga has joined. 18:49:31 One of them is http://www.thespoof.com/news/spoof.cfm?headline=s5i53486 which is not quite what it's about. 18:49:37 ehird, "404 Not Fou.. Wait sorry, here it is. Err. 200!" 18:49:39 ais523: Correct! I know it! 18:50:26 sadly the HTTP protocol doesn't let you say you did a mistake halfway through the transmission 18:50:29 Also, we're crafted out of clay. 18:50:42 AnMaster: are you sure? 18:50:47 TCP does, at least 18:50:47 ais523: yes 18:50:51 so you could do it at the lower level 18:50:51 ais523, oh? 18:50:53 after the headers, it's just a raw stream 18:50:57 ais523, really? 18:51:04 ehird: I was thinking, halfway through the headers 18:51:14 AnMaster: if you send packets out of order, you can fill in the missing ones later 18:51:15 ais523: impossible too 18:51:19 there's no header for it 18:51:19 in fact, the other computer will bug you for them 18:51:22 ehird: pity 18:51:25 ais523, heh 18:51:41 AnMaster: useful because sometimes the packets end up out of order even though you sent them in the right order 18:51:49 oh yes, if HTML you could add in, then put the broken part inside 18:51:55 -!- BeholdMyGlory has joined. 18:52:08 however, hm 18:52:19 ais523, you had to make space for that extra packet right? 18:52:25 in the numbering 18:52:29 yes 18:52:41 ais523, and what if you decided you didn't need an out of order one there... 18:52:46 tricky 18:52:48 just send a packet full of comments 18:52:58 ais523, or zero length? 18:53:06 if that is valid 18:53:07 no idea 18:56:04 ais523: I am now. 18:56:14 GregorR-L: I wrote a BF Joust hg bundle 18:56:17 which is moderately tested 18:56:19 where should I send it? 18:57:50 sent 18:57:53 -!- MizardX has joined. 18:58:01 it requires Perl 5.10; I may have to produce a 5.8 version, depending on what you're running atm 18:59:07 ais523: debian testings' 18:59:11 hopefully perl 5.—— 18:59:13 Wait a second. 18:59:18 ais523? Depend on modern Perl features? 18:59:22 Ubuntu have been on Perl 5.10 for ages 18:59:24 ais523, did you read that about super-optimisation to produce better peep hole optimisers? 18:59:27 modern *5.10* features? 18:59:30 AnMaster: not the full article 18:59:35 ehird: and why not? 18:59:35 ais523: Who are you, and what did you do with ais523? 18:59:40 ais523, ah, see some examples near the end in a table 18:59:49 ehird: my aim for backwards compatibility is specifically about C-INTERCAL 18:59:55 it isn't so appropriate elsewhere 19:00:04 ais523: you've always seemed quite disdainful to the new fangled 5.10 stuff 19:00:09 I also believe in portability; but 5.10 is portable 19:00:18 and I don't remember being disdainful towards the 5.10 stuff 19:00:23 it makes the programs easier to read, at least 19:00:29 which is useful when showing them to someone else 19:17:10 -!- KingOfKarlsruhe has quit (Remote closed the connection). 19:18:13 hmm... does anyone else here view multiple short files by piping more into less? 19:20:01 * AnMaster makes a pass the scans once forward, then once backwards on the tree... 19:20:16 ais523, err what 19:20:19 why use more at all 19:20:47 oh I see. Pretty headers. 19:20:55 rather than having to do :n to see next fime 19:20:56 file* 19:21:21 ais523, I would never had thought of that... 19:22:42 ais523: that doesn't seem to work :< 19:22:58 ehird: as in, more *.c | less? 19:23:00 it works for me 19:23:11 I just get one of the files 19:23:24 no wait 19:23:27 I get all of the files, concatenated 19:23:32 from wikitech-l: How does one access the toolserver to use commands like that? with an axe. 19:24:03 Domas is in charge of server maintenance, etc, he's basically the person who most often uses root access on Wikimedia's servers 19:24:15 !bfjoust 19:24:15 Use: !bfjoust 19:24:40 sounds good 19:24:42 !bfjoust nop . 19:24:47 Score for nop: -2 (maximum 4) 19:24:55 Wooh, -2 rulz :P 19:25:00 I get all of the files, concatenated 19:25:02 here I get 19:25:05 hmm... that maximum's one higher than it ought to be 19:25:10 :::::::::::::: 19:25:10 Makefile 19:25:10 :::::::::::::: 19:25:12 stuff 19:25:15 :::::::::::::: 19:25:15 Emakefile 19:25:15 :::::::::::::: 19:25:17 other stuff 19:25:19 and so on 19:25:20 because in BF Joust, a program can never beat a copy of itself 19:25:25 I should have noticed that in my testing 19:25:26 not plain concatenation 19:25:49 GregorR-L: it puts a results table in a directory the same way the fyb one does 19:25:59 ais523: I see, I'll need to make that accessible some way :P 19:26:16 in fact, it's mostly the fyb code I'm using 19:26:19 mostly unmodified 19:27:28 http://codu.org/eso/bfjoust/report.txt 19:27:39 That looks a wee bit off ... 19:27:47 Oh, and I see why. 19:28:11 to do with tabs 19:28:13 I don't see a solution immediately, but I do see my looming dental appointment, so byeeeee :P 19:28:19 bye 19:28:38 anyway, attack/defend/fool are the scissors/rock/paper of BF Joust, I think 19:28:50 although writing good defence programs is /hard/ 19:28:58 at least, as far as I could tell 19:29:04 and a good attack can normally break through them 19:29:12 -!- asiekierka has joined. 19:29:13 hi ais523 19:29:16 hi asiekierka 19:29:19 ais523: How's BF Joust? Did anyone beat me? 19:29:21 !bfjoust nop . 19:29:26 Score for nop: -2 (maximum 4) 19:29:33 asiekierka: you can try out your programs on egobot now 19:29:36 yay! 19:29:46 ais523: Do you still have my app? 19:29:49 If yes, could you write it here? 19:29:51 I think I lost it 19:29:56 your program, you mean/ 19:30:02 as in, BF Joust program? 19:30:06 it should be in my logs, and in clog's logs 19:30:18 didn't you save it as "asiekierka.bf" 19:30:20 or someth 19:30:24 yes, but in /tmp 19:30:26 and I've rebooted since 19:30:26 oh 19:30:28 :/ 19:30:39 but it's in my logs, because you told me in the first place 19:31:10 [>+[-]+.] 19:31:17 !bfjoust asie1 [>+[-]+.] 19:31:18 Score for asie1: -1 (maximum 5) 19:31:21 ...eeh? 19:31:58 !bfjoust asie2 >>>>>>>>>>[>+[-]+.] 19:32:00 Score for asie2: -2 (maximum 6) 19:32:04 ... 19:32:07 !bfjoust asie2 >>>>>>>>>[>+[-]+.] 19:32:11 Score for asie2: -2 (maximum 6) 19:32:16 what's the scoring? 19:32:22 I thought I was 2nd place! 19:32:26 +1 for each program you beat, -1 for each program that beats you 19:32:36 and the test programs in there are different from the ones I was running against on my own computer 19:32:36 How many programs are there? 19:32:46 asiekierka: http://codu.org/eso/bfjoust/report.txt 19:32:59 it runs each program against all previous programs 19:33:19 !bfjoust asie1 [>+[-]+.] 19:33:19 Score for asie1: -1 (maximum 6) 19:33:25 !bfjoust ehirdomatic [[>[-]]>+] 19:33:28 Score for ehirdomatic: -1 (maximum 7) 19:33:40 the report runs in the background, so it won't update immediately 19:33:40 ais523: http://codu.org/eso/bfjoust/report.txt is empty now. 19:33:42 I broked it. 19:33:47 :( 19:33:48 see, it's non-empty now 19:33:50 ais523: it wipes it while running, it seems 19:33:54 yes 19:33:56 shouldn't it leave the previous content until then 19:33:58 that's silly 19:34:03 I just copied the FYB stuff 19:34:11 erm 19:34:13 ais523: 19:34:13 19:33 EgoBot: Score for ehirdomatic: -1 (maximum 7) 19:34:15 4 | - - - 0 - 0 | 15.3| -4| ehirdomatic.bfjoust 19:34:25 it says my score is 15, but correcting the obvious copypasta to pts, I get -4 there 19:34:27 and -1 from the bot 19:34:33 ehird: well, it's random tape lengths 19:34:37 although I'm not sure if that matters 19:34:46 ais523: it should average it out... 19:34:54 only one run for each pair of programs 19:34:59 otherwise it would take /ages/ 19:35:09 hm 19:35:28 ehird: although I will point out that your program is probably incapable of winning if the tape length is even 19:35:32 ais523, it might be interesting to have four states for the balance of a loop 19:35:34 which would explain the variety in results 19:35:39 ais523: argh 19:36:29 ais523, unknown, true and false are what I have currently. But it could be useful to have "sometimes" as well. Like: [>+,[>>[-]+]<] 19:36:33 Hmm, so I have 4 points? 19:36:34 that might be unbalanced 19:36:40 So I'm 1st place or what? 19:36:47 yep 19:36:51 ais523, also infinite though 19:36:52 maybe I should input one of my good programs 19:36:59 Well 19:36:59 I just put the basic program for each strategy in there to start with 19:37:07 I'm good with defense but bad at attack 19:37:30 !bfjoust ais523_defend5 >+>+([{>[(.)*20-]+}]<..........-[++[[]<(-..-.)*300>[>[-]+]]]<(+..+.)*300>[>[-]+])%2000 19:37:52 ()? 19:37:57 Score for ais523_defend5: -1 (maximum 8) 19:38:09 !bfjoust ais523_attack5 [>[-]-.-.-.-.-.-] 19:38:11 ais523, is high or low score bad 19:38:12 Score for ais523_attack5: -2 (maximum 9) 19:38:15 or best 19:38:15 AnMaster: high is good 19:38:17 a 19:38:32 attack5 does badly against other attack programs, but well against defense programs 19:38:32 ais523, so negative values indicate sucky programs? 19:38:35 which is why it's doing badly here 19:38:36 can't you auto-refresh report? 19:38:40 AnMaster: it means they lost more than they won 19:38:49 So I'm an attack program it seems 19:38:51 ais523, hm, which one won most 19:38:52 asiekierka: the report will take a whle to generate 19:38:54 which is wrong? 19:38:56 AnMaster: sam 19:38:57 *same 19:39:04 ais523, err? 19:39:09 same what 19:39:20 hmm... the "score for" seems wrong 19:39:29 !help 19:39:30 Supported commands: addinterp bf_txtgen bfjoust daemon daemons delinterp fyb help info kill mush userinterps 1l 2l adjust asm axo bch bct befunge befunge98 bf bf16 bf32 bf8 bfbignum boolfuck c chiqrsx9p choo cintercal clcintercal cxx dimensifuck echo forth glass glypho google hello kipple lambda lazyk linguine malbolge ook pbrain perl qbf rail rhotor rot13 sadol sceql sh show slashes test trigger udage01 underload unlambda whirl yodawg 19:39:31 as in, ais523_attack5 won every single game on the report, which I tested 19:39:35 but is low on the score-for, which is wrong 19:39:43 um 19:39:50 that sorting is off 19:39:56 GregorR, ^ 19:40:09 how is it off 19:40:14 GregorR, it would be more useful if the list was sorted 19:40:14 I'm not reading the whole damn thing 19:40:16 AnMaster: the help? it isn't, it lists special commands first then interpreters 19:40:21 ais523, ah 19:40:29 fyb and bfjoust both need special handling 19:40:38 http://www.youtube.com/watch?v=SYRLTF71Sow - check this out xDD 19:40:47 ais523, I see... what about the text gen one though 19:41:01 AnMaster: I suspect that's programmed weirdly too 19:41:04 !bf_txtgen BF Joust 19:41:06 102 +++++++++++[>+++>++++++>+><<<<-]>>.++++.<-.>++++.+++++++++++++++++++++++++++++++++++++.++++++.--.+.>-. [83] 19:41:13 heh, >< 19:41:18 ais523, Old news 19:41:24 and that's a lot of +s in a row in the middle... 19:41:27 !bf_txtgen BF Joust 19:41:27 it has been discussed several times the last few days 19:41:29 102 +++++++++++[>+++>++++++>+><<<<-]>>.++++.<-.>++++.+++++++++++++++++++++++++++++++++++++.++++++.--.+.>-. [577] 19:41:52 I reckon a human could probably beat that easily 19:41:59 although I can't be bothered to right now 19:42:08 ais523, I think it could beat itself by running for more generations 19:42:10 !bf_txtgen BF Joust 19:42:13 77 +++++++++++[>+++>++++++>++++++++++>+<<<<-]>>.++++.<-.>++++.>+.++++++.--.+.>-. [222] 19:42:16 see 19:42:17 ehird: i dun think so 19:42:23 well i oomed but then i fixed it 19:42:28 ais523, it is genetic algo after all 19:42:31 yes 19:42:35 bsmntbombdood: what was it an? 19:42:55 bsmntbombdood, you have 12 GB ram and you OOMed? 19:42:58 w 19:42:59 t 19:43:01 h 19:43:02 did you do 19:43:13 i was running a few too many "sort -S 5G"s 19:43:20 err 19:43:25 why would you be sorting 5 gig of data? 19:43:25 why did you do that 19:43:43 it was actually more than 5 gigs of data 19:43:43 AnMaster: to use up RAM? 19:43:49 i wanted to see how fast it was! 19:43:49 ehird, why 19:43:52 hm 19:44:05 AnMaster: cuz he has no other use for 12g 19:44:10 and how well i could parallelize sort(1) with bash 19:44:27 err what 19:44:36 AnMaster: what 19:44:43 and how well i could parallelize sort(1) with bash 19:44:46 what 19:44:50 what what? 19:44:52 ... 19:45:26 sort -m <(sort first_chunk) <(sort second_chunk) 19:45:28 how can you use sort(1) + bash to sort a file in parallel.. 19:45:33 hm 19:45:41 ah ok 19:45:51 brilliant 19:45:52 to chunk up the file, use split -l and fifos 19:46:01 ais523, it is merge sort kind of I guess 19:46:06 why did you need to test it on gigabytes of data? 19:46:08 "kind of"? 19:46:10 AnMaster: "kind of"? 19:46:13 it /is/ merge sort 19:46:16 that is exactly merge sort 19:46:27 ais523, yes, but sort internally doesn't use that for all parts 19:46:32 so it is merge for the outermost layer 19:46:36 you can merge sort a quicksort 19:46:37 but not for the inner ones 19:47:12 the "inner" sort that works on the separate chunks are not using merge sort afaik 19:47:26 it does when the data is bigger than memory 19:47:29 AnMaster: pure-mergesort is a relatively common sorting technique nowadays 19:47:36 btw, what happens if you try to use a buffer size smaller than the file 19:47:38 because unlike quicksort, it isn't slowed by pathological data 19:47:38 to sort 19:47:47 mergesort is very elegant, i'm a big fan of it 19:47:54 AnMaster: you can write temporary stages of mergesort on disk 19:48:03 Mergesort is pretty spiffy, indeed. 19:48:06 ais523, sort(1) does that? 19:48:06 slows it down slightly due to disk I/O time, but mergesort streams very well 19:48:06 i am preempting all of AnMaster questions by a couple of seconds 19:48:18 AnMaster: I don't know, I'd have to look at the source 19:48:37 I'm a bit more fond of bucket sort for hand-sorting stuff, though. :p 19:48:54 bsmntbombdood: that'll fall down when he asks a really stupid question 19:48:56 I prefer sorting networks 19:49:08 AnMaster: why? 19:49:17 ais523: he said that yesterday do 19:49:19 pikhq: I generally use insertion sort when hand-sorting 19:49:26 probably they're all ~*PARALLEL*~ like <3Erlang<3 19:49:30 *too 19:49:33 because when sorting by hand, I'm not normally sorting much 19:49:35 i have to sort 100s of items every day at work 19:49:40 and normally sorting physical objects 19:49:46 and they insertion-sort well 19:49:50 ais523: That's what I do if I'm hand-sorting something that's partially sorted. 19:49:55 bsmntbombdood: if only 100, it shouldn't matter what algorithm you're using 19:50:03 unless you can't automate it 19:50:09 (say, my Magic cards got a bit unsorted by someone) 19:50:14 *by hand 19:50:14 (well, excluding silly ones like stooge-sort 19:50:16 ) 19:50:24 Bucket sort is much nicer for hand-sorting things that aren't very well-sorted. 19:50:24 ais523, isn't it O(log n) iirc? 19:50:32 !bf_txtgen BF Joust 19:50:34 102 +++++++++++[>++++++>+++>+><<<<-]>.++++.>-.<++++.+++++++++++++++++++++++++++++++++++++.++++++.--.+.>>-. [99] 19:50:37 ais523, in size 19:50:46 AnMaster: what is? 19:50:47 or is it O(n log n) in size 19:50:50 ais523, sorting networks 19:50:53 all sorts are at least O(n) in size 19:50:57 because otherwise you couldn't store all the data 19:51:17 and mergesort is O(n log n) in time, O(n) in size, IIRC 19:51:31 'Tis. 19:51:33 AnMaster: ok, I looked up sorting networks, and you're full of shit, because they're just a way to represent conventional sorting algorithms: http://en.wikipedia.org/wiki/Sorting_network#Insertion_and_Selection_networks 19:51:41 Unless you've got inefficient allocation, that is. 19:51:52 :p 19:51:58 ehird, http://en.wikipedia.org/wiki/Sorting_network#Optimal_sorting 19:52:10 mergesort's nice because it hits theoretical optimums for both time and size, /and/ its worst-case performance is the same as its best-case performance 19:52:13 AnMaster: how's that relevant at all? 19:52:28 ais523: wait, mergesort always uses the same amount of time for a given length? 19:52:46 ehird: yes for some implementations 19:52:52 awesome 19:52:59 I assume that it isn't exactly the same in practice due to things like branch prediction 19:53:03 it uses exactly the same amount of comparisons iirc 19:53:07 and people not writing the program symmetrically 19:53:10 but it's the same in comparison count 19:54:41 an abacus can sort in O(sqrt(n)) says wikipedia 19:55:01 http://en.wikipedia.org/wiki/Bead_sort#Complexity 19:55:20 AnMaster: Define n 19:55:30 oh, gravity-based. 19:55:36 ehird, read the link 19:59:14 i've done buck, quick, merge sort by hand 19:59:34 i just do crapsort 19:59:51 find something that looks too big/small for position, throw it in right direction 19:59:54 repeat until sorted 20:00:19 merge sort is actually rather adaptive when you do it by hand 20:04:40 * pikhq <3 bead sort. 20:04:47 -!- asiekierka has quit. 20:05:12 build a machine to do it 20:05:27 bsmntbombdood: it's called an abacus 20:05:38 a machine with digital i/o 20:05:55 with a scale large enough that it's actually faster than a general purpose cpu 20:05:55 I'm pretty sure you could build a sorting coprocessor to do bead sort. 20:05:59 bsmntbombdood: it's useless without an infinitely long abacus 20:06:07 Say, an FPGA. 20:06:14 pikhq: he means analogue 20:06:16 with actual beads 20:06:17 but digital io 20:06:21 robot to move them 20:06:29 gravity to sor 20:06:29 t 20:06:40 Both digital and analog hardware implementations of bead sort can achieve a sorting time of O(n) 20:06:41 That'd be pretty cool, too. 20:06:42 eh, how boring 20:07:03 how does a solid state version work? 20:07:17 bsmntbombdood: just simulates an abacus? 20:07:22 and basic gravity 20:07:34 find something that looks too big/small for position, throw it in right direction <-- one way to optimise it would be by when it is too big and you move it down, then go back one step and compare, and so on 20:07:46 AnMaster: that's gnome sort 20:07:50 anyway, if it has to be O(n), you might as well use your fpga to do a counting sort 20:07:53 but it's too boring to remember all that stuff 20:07:53 ehird, oh? it has a name? 20:08:05 it's so much easier just to throw stuff around 20:08:30 ehird, it seemed like the best way for the instruction shifter in bf. Since there are so many restrictions on the top of the basic "sort by offset" 20:08:34 like IO can not cross 20:08:37 and so on 20:09:04 can sse help with sorting at all? 20:09:09 I'm rather fond of non-comparison sorts, really. 20:09:18 i know there are compare/permute instructions 20:09:18 bsmntbombdood: maybe 20:09:18 and of course it has to be a stable sort... "[-]>,<+" isn't same as "+[-]>,<" 20:09:19 :P 20:09:29 bsmntbombdood: you have sse4.2; go play with it 20:09:43 i have 0 asm knowledge 20:09:52 bsmntbombdood: just mix the sse instructions in with C 20:09:58 it's just registers 20:10:08 yeah 20:10:24 bsmntbombdood: Depends; are you doing any vector operations? 20:10:41 pikhq: CUDA! 20:10:46 Say, comparing 4 values to 4 other values at the same time? ;p 20:11:10 I'm going to try some CUDA stuff 20:11:14 right 20:11:15 Hope my shitty radeon can run it 20:11:17 aitw 20:11:19 wait 20:11:21 radeon is ati 20:11:23 * ehird slaps self 20:11:29 ehird: my shitty geforce can do it 20:11:42 bsmntbombdood: but my radeon absolutely cannot, being that CUDA is a nvidia technology :) 20:12:02 what's your card? the 8600? 20:12:16 um 20:12:23 9400 20:12:24 ATI have something similiar 20:12:25 iirc 20:12:30 AnMaster: yes, but it's not CUDA 20:12:32 and CUDA seems nicer 20:12:48 ehird, of course it isn't CUDA. But the one ATI uses is an open format iirc. 20:12:56 while CUDA is proprietary 20:13:01 bsmntbombdood: right, it's a mid-low-end gfx card 20:13:04 as in, mid of the low enda 20:13:06 end 20:13:09 AnMaster: cuda's specs are open, I think 20:13:13 and I don't really care 20:13:16 CUDA seems nicer to program in 20:13:33 d'you mean OpenCL, AnMaster? 20:13:42 ehird, maybe, I don't remember the name 20:13:48 "OpenCL was initially developed by Apple Inc., which holds trademark rights, and refined into an initial proposal in collaboration with technical teams at AMD, Intel and Nvidia." 20:13:49 "NVIDIA announced on December 9, 2008 to add full support for the OpenCL 1.0 specification to its GPU Computing Toolkit.[8]" 20:13:59 so I could use it if I wanted, but CUDA looks nicer 20:14:06 kay 20:14:11 easier 20:15:12 ehird, anyway Gnome sort is trivial to implement. And works fairly well for sorting BF instructions. 20:15:12 get an nvidia tesla 20:15:23 ehird, and it is peep hole style too 20:15:28 bsmntbombdood: why, they're just GTXes without the gaming stuff 20:15:59 ehird, trivial when working on linked two linked lists (one input and one output) as well 20:16:09 I'm considering getting a gtx, though, and putting a quiet fan on it 20:16:18 but I'm not sure I need the extra performance :p 20:16:23 vs 9800gtx+ 20:16:45 ehird, tesla doesn't have any VGA or DVI connectors. iirc 20:16:52 But yeah you only use monitors for gaming 20:16:57 AnMaster: heh 20:16:57 not for anything else 20:16:57 :D 20:16:59 I didn't know 20:17:03 but that's even more of a reason not to buy them 20:17:07 ehird, I may misremember 20:17:09 I bet they're more expensive than GTXes too 20:17:11 they are for HPC stuff 20:17:16 yeah, I know 20:17:21 but I'd just buy a GTX for that anyway :p 20:17:36 tesla has like 4gb of ram too 20:17:42 ah... 20:17:57 bsmntbombdood: meh 20:18:01 just SLI a bunch of gtx's up 20:18:05 should be enough for most anything 20:18:33 -!- BeholdMyGlory_ has joined. 20:19:05 -!- BeholdMyGlory has quit (Nick collision from services.). 20:19:08 -!- BeholdMyGlory_ has changed nick to BeholdMyGlory. 20:19:39 jesus 20:19:50 the high-end tesla uses 550watts 20:20:10 They should make some sort of combined sauna stove / GPU. 20:20:13 bsmntbombdood: good lord 20:20:24 how the hell do you cool that?!?! 20:21:01 bsmntbombdood: a fuckload of fans 20:21:02 I wonder how much you could quiet a gtx 275 20:21:20 Huge heatsink would probably get it down to a just-too-high temp, you could probably add a nice quiet fan on to that 20:21:22 but what about fan adjustment? 20:21:23 the puny little fan and heatsink i see on them doesn't look very capable 20:22:05 bsmntbombdood: it probably runs at 5k rpm or something daft like that 20:22:12 bsmntbombdood: or has 3 fans under the cover 20:22:22 the heatsink is still tiny 20:22:45 Maybe a fan with tiny fans mounted on the blades. 20:23:03 :D 20:23:07 infinite fans 20:23:12 unfortunately, infinite dB 20:23:27 but infinite cooling power! 20:24:06 ehird: actually, log-infinity dB 20:24:13 oh? 20:24:18 which means that you can have aleph-one cooling power, but only aleph-zero noise 20:24:23 decibels are a logarithmic scale 20:24:29 heh, true enough 20:24:34 ais523: that's some good cooling/noise ratio 20:24:43 unfortunately, I'm not sure my ears are clued up on aleph-one vs aleph-zero 20:24:49 they might just break anyway 20:26:13 They should make some sort of combined sauna stove / GPU. <-- that would definitely be a hit in Finland. Not anywhere else though... 20:26:37 AnMaster: whyever not? just have a sauna next to your house 20:26:38 profit and games 20:26:48 ehird, hm? 20:27:00 your gpu doubles as a sauna stove 20:27:05 sure 20:27:05 AnMaster: It'd also be a hit in much of the northern US and Canada. 20:27:07 so make a sauna next to your house and use it as one 20:27:12 pikhq, ok 20:27:14 and start a sauna business 20:27:17 you can profit while you game 20:27:21 and have an excuse to game all day 20:27:22 but what about California and such 20:27:27 too hot there already 20:27:33 And it'd be ubiquitous in Alaska. 20:27:38 and there is where computers are designed after all 20:27:42 mostly 20:27:49 AnMaster: what a great generalization 20:27:52 Yeah, in California, you sell a combined air conditioner and CPU cooler. 20:28:00 pikhq, more like that yes 20:28:07 ehird, sure! 20:28:15 anyway 20:28:19 I recommend a tank of liquid nitrogen for the computer and a fan to distribute the cold air. 20:28:23 my point was, they wouldn't come up with the idea 20:28:45 pikhq: people have overclocked CPUs to ~6.5ghz by just pouring liquid nitrogen on it 20:28:46 it wouldn't occur to them. Too hot. 20:28:50 willy-nilly 20:28:53 pretty fun 20:28:56 ehird: submerging in liquid nitrogen, I thought 20:29:06 given that it doesn't conduct electricity, it's not all that bad an idea 20:29:12 ais523: well, the video just showed a huge amount of smoke and someone tipping a canister 20:29:16 so it'd probably end up submerging it 20:29:20 um 20:29:21 but they tipped it continuously 20:29:24 that's steam, not smoke 20:29:26 ehird: 10GHz with Phenom II. 20:29:28 :p 20:29:29 err, steam 20:29:29 right 20:29:32 pikhq: nah 20:29:35 6.5ghz is world record 20:29:40 on some AMD chip, maybe phenom ii 20:29:40 well, sort of condensed liquid-phase steam 20:29:46 I'm not sure if that technically counts as steam or not 20:29:53 it's water, anyway 20:29:54 the photo I have seen showed them pouring it into a pipe mounted on the CPU 20:30:01 AnMaster: that's possible too 20:30:06 i don't really recall 20:30:16 it was on wikipedia iirc 20:30:27 might have been a different madman 20:30:59 i think you would have to get all your components completely anhydrous 20:31:15 what 20:31:26 hm 20:31:29 AnMaster: AMD and Gigabyte do it at open events 20:31:31 to promote their hardware 20:31:33 ehird, heh 20:31:38 so not exactly madman 20:31:54 ehird, http://en.wikipedia.org/wiki/File:2007TaipeiITMonth_IntelOCLiveTest_Overclocking-6.jpg 20:32:13 ehird, I'm quite sure they won't give support on those products if any customer did it! 20:32:27 AnMaster: that's from an Intel event, it seems 20:32:29 so different one 20:32:32 ehird, yes 20:32:37 didn't claim it was same 20:32:50 AnMaster: and, well— I read an AMD Athlon X2 manual today. it said that replacing heatsink/fan voided the warranty 20:33:00 so... it's barely supported anyway! 20:33:09 heh 20:33:29 i played with the overclocking settings on it a bit 20:33:33 Amusing, since they sell it without a heatsink as well. 20:33:36 -!- Sgeo has joined. 20:33:40 it was 2ghz and went to 2.67ghz quite painlessly 20:33:44 pikhq, haha 20:33:48 although sometimes it'd try to boot, not post, give a click, then reboot again and work 20:33:50 (automatically) 20:33:52 Hm? 20:33:58 Sgeo: read the logs if you want to know 20:34:11 but yeah, it ran just fine although I didn't do any intensive tests 20:34:12 was quite fun. 20:34:23 ehird, you have such a computer? 20:34:32 Reading logs means waiting for Firefox to start up 20:34:32 an athlon x2? 20:34:37 Which can be a few minutes 20:34:51 it's my mom's but I was toying with it earlier 20:35:44 Wow 20:35:49 the cray-2 was submerge-cooled 20:35:53 *submersion 20:36:07 http://en.wikipedia.org/wiki/Cray-2 20:37:32 http://en.wikipedia.org/wiki/Computer_cooling#Integrated_Chip_Cooling_Techniques 20:37:34 that looks cool 20:39:35 when i pushed the cpu past 2.67ghz it refused to boot, which made me sad :p 20:40:44 that's a pointless overclock, really 20:40:52 why not run it at regular speed if that's all the improvement you can get? 20:40:57 ais523: I did 20:41:03 I put it back 20:41:08 but I wanted to see how much it'd let me do 20:41:33 my computer is faster than early crays, right? 20:41:36 why overclock at all 20:41:38 bsmntbombdood: of course 20:41:47 AnMaster: to spend less money on a cheaper processor to get more performance 20:41:53 and for fun 20:42:19 but this didn't really work too well... the base clock thing was 200 to start with, and more than 207 made it more unstable than the sometimes-boots-twice-automatically-before-POSTing (i.e. refused to boot). multiplier max was 13x but that worked fine. 20:42:22 Hm. I never heard of anyone overclocking a laptop 20:42:25 so meh 20:42:37 AnMaster: http://www.google.com/search?client=safari&rls=en-us&q=overclocking+laptop&ie=UTF-8&oe=UTF-8 20:42:41 15 Nov 2002 ... Overclocking a laptop can be a mistake, as processor heat is not easily dissipated when using a heat sink with a fan pushing less air than I ... 20:42:47 fairly obvious 20:42:58 yeah 20:43:16 the fan on the athlon usually runs at 1600 rpm or so 20:43:28 with the small .67ghz overclock it ran at 1900 rpm or so 20:43:32 went a few degrees up too 20:43:39 kind of rubbish 20:43:41 -!- Slereah_ has joined. 20:46:15 well 20:46:25 ais523: .67ghz isn't that small; it's a 33.5% increase 20:46:35 but performance-wise, 2 vs 2.67ghz probably isn't gigantic, for a low-endp orc 20:46:35 proc 20:46:43 that is pretty small, only 33% 20:46:45 laptops are so fail 20:46:48 in terms of actual usefulness 20:47:00 ais523: err, over 20% overclock is considered large, I believe 20:47:13 hm no 20:47:14 over 30% 20:47:30 so why do people even bother with it? 20:47:44 ais523: because some CPUs overclock better 20:47:50 the 2.6ghz Core i7 920 20:47:52 can overclock to 4ghz 20:47:59 which is higher than an i7 you can buy 20:48:02 and: 20:48:10 920 $279.99 20:48:17 965 (3.2ghz) $999.99 20:48:30 so you save $720 and get .8ghz more 20:49:04 New OOTS :D 20:50:50 -!- kar8nga has quit (Remote closed the connection). 20:51:39 OOTS? 20:52:04 Sgeo: I'll wait a few days before visting it, I don't want to overload the servers 20:52:07 *server 20:52:13 in the meantime, don't spoil it 20:52:27 ais523: what kind of server can't handle a few web hits? 20:52:33 Maybe in the next few days, there'll be a new new one 20:52:39 a single server that's very popular 20:52:41 What are you up to? 20:52:44 ehird: Maybe a C64? 20:53:00 apparently it's message board traffic that puts the most strain on it 20:53:02 AnMaster: Order of the Stick. 20:53:06 ais523: stress someone else's single popular server: http://imgur.com/8h4xz.gif 20:53:08 The server was recently upgraded, and in a few months there will be two servers - one for the comic, one for the boards 20:53:22 ah 20:53:23 http://en.wikipedia.org/wiki/File:Alaska_Pipeline_Closeup_Underneath.jpg 20:53:24 ehird: not from this connection, and that's almost certainly copyvio if it's a copy of OotS 20:53:35 rehosting free material is not a copyright violation 20:53:39 It's a copy of OotS 20:53:40 -!- Slereah has quit (Read error: 110 (Connection timed out)). 20:54:06 (and if it's ruled so, I shall have to build an asylum for the world. Mark two.) 20:54:25 ehird: err, I don't see why it wouldn't be 20:54:26 Mark one being in the HHGG "trilogy"? 20:54:37 Sgeo: Yes. 20:54:39 because "free" != "free for all purposes" 20:54:43 ais523: *facepalms, builds underground bunker* 20:54:47 he might want the ad revenue, for instance 20:54:52 I know OotS has no direct ads 20:55:00 if he wants to turn a profit he should get a server that can handle hits :) 20:55:00 but it certainly advertises its own peripheral products a lot 20:57:09 ehird: I've already done that. This was after I discovered that most car owners were incapable of changing their oil or other such basic tasks. 20:57:15 ;) 20:57:31 pikhq: can I leave the Asylum? 20:57:45 I think you will find that I am incredibly sane, and thus can apply for entainment. 20:57:58 ehird: You are in the asylum, but not as an inmate. 20:58:03 (being the opposite of detainment) 20:58:07 So, yes, you are free to leave it at any time. 20:58:16 pikhq: where does it end? 20:58:47 I declare your home to be Outside of the Asylum. 20:58:59 Hooray. 21:00:40 heatsink reviewers need a more objective methodology 21:01:24 bsmntbombdood: like how 21:01:35 Infinite inversion snowflakes turn into MC Escher's dream about butterflies. 21:01:38 build a heater with a thermocouple and a processor shaped heatspreader, attach your heatsink, and measure how much heat you can put into the base until the temperature reaches some amount 21:02:14 something like 60 degrees 21:02:25 and then you can compare any coolers based on their wattage rating 21:02:36 yeah but that doesn't measure real-world processor performanc 21:02:37 e 21:02:50 pretty much 21:03:26 http://www.google.com/logos/missinglink.gif 21:03:34 Some Christians are upset by this 21:03:42 I'm not surprised. 21:03:48 Who gives a damn? 21:03:55 Sgeo: as in 404? 21:04:04 heh 21:04:06 ais523: err, what? 21:04:14 ehird, ais523 was making a joke 21:04:21 Sgeo: They can believe the world is 6000 years old and God makes each and every one of us out of clay, and we can advance humanity. 21:04:24 Sgeo: it's actually a metajoke 21:04:24 Sgeo: was he? 21:04:27 alright :P 21:04:41 given what happened in this channel a few hours ago 21:04:50 arctic cooling are pretty cool 21:04:51 ↑geddit? 21:04:55 ...I'm curious now, what happened? 21:05:07 Sgeo: AnMaster thought the link in the logo was an error 21:05:11 ehird: I'll tell you who gives a damn. 21:05:11 as in, link-is-missing 21:05:21 The Christians that think the world is older than 6,000 years. ;) 21:05:35 And most of them are caring enough to face-palm. 21:05:40 pikhq: Hey, they do associate with the other ones by choice... :-P 21:05:43 (↑joke) 21:05:54 * pikhq doesn't. 21:09:29 http://www.frostytech.com/articleview.cfm?articleid=2258 21:10:01 the orochi is so large as to be impractical 21:10:07 "less than 30 dBA noise" 21:10:08 bahahahaha 21:10:15 you mean like any decently silent heatsink/fan combination ever? 21:11:19 The heatpipes get bonus points for looking like some sort of tentacle monster. 21:13:59 I would like to move the innards of my secondary desktop computer from the current "boring beige box" into the not-in-use-since-the-smoke-went-out Lian Li "itty-bitty black box", but I think the monstrous hunk of a cooling device (which isn't even all that big by modern standards) is a bit too high to actually fit in. 21:16:33 fizzie: CPU cases have magic smoke now? 21:16:42 or did the smoke escape from the CPU inside it? 21:17:02 The smoke escaped from the parts that were formerly occupying the box, yes. 21:17:30 Well, I guess they still are occupying it in a physical sense, since I just stuck the thing in a corner so I won't stumble on it. 21:24:49 http://74.125.47.132/search?q=cache:gzD5srbF0_sJ:www.cs.ualberta.ca/~niewiado/TR07-02.pdf+sorting+with+sse&cd=4&hl=en&ct=clnk&gl=us&client=iceweasel-a 21:28:05 -!- BeholdMyGlory_ has joined. 21:28:35 -!- BeholdMyGlory has quit (Nick collision from services.). 21:28:38 -!- BeholdMyGlory_ has changed nick to BeholdMyGlory. 21:28:38 [ehird:/TheVolumeSettingsFolder/HFSExtentTables/MNT6835442416PHRKF87P115I8] % ls 21:28:39 6NGSHGFTH6UJ237QGNESSF403KDLU5C1N7V6OVSOMP4UC1JL3RM0 21:28:41 wtf. 21:29:40 ? 21:29:50 bsmntbombdood: poking around / 21:51:46 the problem with using sse for sorting is that if you are able to use it, you are also able to use a non-comparison sort 21:52:04 bsmntbombdood: heh 21:53:33 there is CMPLEPS 21:55:28 or MAXPS 21:56:08 SO LIEK OMG 21:57:24 LIEK OMG ON THE 21:58:27 hah, I have to try this out: http://web.archive.org/web/20001217124300/www.cacr.caltech.edu/~roy/upi/index.html 21:58:42 -!- nooga has joined. 21:58:44 the probability of two large random integers being coprime is 6/pi^2 21:58:57 according to Riemann zeta, anyway. 21:59:18 great way to calculate pi, amirite? 21:59:25 i wonder if there is sense in writing SADOL compiler that employs partial evaluation for different types 21:59:29 I must put all the computing power of my GHC to this task 21:59:38 I'll tell you when I get 5 decimal places. >_> 21:59:45 Gracenotes: ask bsmntbombdood 22:00:01 he'll throw 4 cores/8 threads, 2.9ghz and 12GB of DDR3 RAM at it 22:00:32 -!- BeholdMyGlory has quit (Remote closed the connection). 22:00:33 ehird: ooooh so now hyperthreading is a good thing 22:00:37 Gracenotes: that's brilliant, a bit like monte carlo evaluation of pi, but less efficient 22:00:41 bsmntbombdood: nope 22:00:44 bsmntbombdood: I was just stating the facts 22:00:46 that actually wouldn't be a bad shootout problem 22:00:53 Gracenotes: it'd be a bloody slow one 22:01:38 implement a 19937 mersenne twister and, given a seed via command line, calculate the number of iterations until you get pi to a certain accuracy 22:01:55 given consecutive pairs of high numbers from the twister 22:02:09 okay, maybe a little too specific for a shootout question 22:02:23 but still, lemme at least try implementing it first... mmmm. 22:02:44 easy to parallelize! 22:02:45 ghc --make -O2 -funbox-strict, here I come 22:03:03 i O2'd your mom's funbox 22:03:24 is a funbox like a funroll? 22:03:30 bsmntbombdood: let's sign up for a BOINC project!!! 22:03:49 !!1!1! 22:04:10 actually, the full name is -funbox-strict-fields 22:04:19 ...to calculate pi via the most inefficient algorithm possible? 22:04:42 I vote that you calculate pi via an evolutionary algorithm. 22:04:54 ais523: it unboxes stuff. 22:04:57 !sadol !!1!1!1 22:04:58 1111 22:05:01 ;D 22:05:01 strict fields in specific 22:05:03 Compare with a circle with known area and radius. 22:05:04 ;) 22:05:08 pikhq: eh. There's a different between that an hill-climbing 22:05:18 *difference 22:05:30 oh, interesting 22:05:37 but still more like hill-climbing >_> 22:05:44 woot 22:05:47 Gracenotes: Did I mention that this was meant to be implemented using floating-point emulation? 22:05:50 where's my llvm-gcc 22:06:14 linked to from here: http://www.reddit.com/r/programming/comments/8ltxo/absolutely_without_a_doubt_the_most_inefficient/c09pfc8 22:06:30 who calculated pi by emulating the pin drop method 22:06:50 ehird: does llvm-gcc exist on macports? 22:06:53 which is possibly more inefficient, but only the mathematicians can tell us! 22:07:07 nooga: try "port search". but use clang, foo 22:07:19 llvm-gcc42 @2.5 (lang) 22:07:23 port install clang? 22:07:30 nope 22:07:32 ais523: oh, a funbox isn't much like a funroll, but maybe a freschedule-modulo-scheduled-loops is 22:07:34 you need to check it out from svn 22:07:41 I love that the PRNG seed is Pi. 22:07:46 or maybe -fsched-stalled-insns-dep= 22:07:47 :D 22:07:58 ehird: i got llvm package from ports (2.5) 22:08:05 so what nao 22:08:09 * Gracenotes looks for the weirdest gcc flag he can find 22:08:25 nooga: http://clang.llvm.org/get_started.html 22:08:31 nooga: before you do those instructions: 22:08:33 export CC=gcc-4.2 22:08:36 otherwise it breaks 22:08:37 finline-functions-called-once 22:08:41 -!- M0ny has quit ("Read error: 182 (Connection reset by beer)"). 22:08:43 then follow those instructions exacterly 22:08:55 ehird: but it won't break my classic gcc? i need it for gayPhone 22:09:04 nooga: it won't, clang != gcc. 22:09:05 anyway. Time to get implementing 22:09:07 clang is a new compiler 22:09:07 ehird: Only works with gcc-4.2? 22:09:10 nooga: also, just don't 'make install; 22:09:14 pikhq: it just doesn't work with 4.1 22:09:16 finline-functions-called-once 22:09:18 bug in gcc 4.1 22:09:19 how is that weird 22:09:20 Oh. 22:09:24 but setting CC might damage something 22:09:26 nooga: and you can just use it from the source tree 22:09:27 and no it won't 22:09:28 nooga: it's temporary 22:09:30 for that session 22:09:31 since it'll goa s you close your terminal 22:09:35 once you close the terminal, it's forgotten 22:09:35 a 22:09:36 nooga: No. 22:09:36 so just close the terminal after that 22:09:47 Or just change CC back. 22:09:51 AnMaster: not really... it's somewhat specific, that's all. 22:10:06 I suppose if you're interested in a trade off between performance and binary size... 22:10:15 Gracenotes, um... ok 22:10:29 * pikhq should set up clang and LLVM... 22:10:32 *eyes AnMaster* o_O 22:10:53 Gracenotes, and what are you implementing 22:11:02 a scrollback searcher 22:11:03 for you! 22:11:16 for me? :D 22:11:20 ehird, sure. Can I saw it with my current headache 22:11:23 wwwwwww 22:11:24 eeee 22:11:26 aw. 22:11:32 eee 22:11:33 AnMaster: you can saw it but you might chop your finger off 22:11:34 kkkkkkkkk 22:11:39 always be careful with saws 22:11:41 AnMaster: http://web.archive.org/web/20001217124300/www.cacr.caltech.edu/~roy/upi/index.html 22:11:54 ehird, it is a migrane (sp?) 22:11:55 svn co http://llvm.org/svn/llvm-project/cfe/trunk clang ; make 22:11:56 ? 22:11:57 and also looking at random gcc options, cuz I feel like it 22:12:00 I know what the weirdest GCC opt is. 22:12:05 -mint16. 22:12:11 pikhq, for what platform 22:12:12 ewww. 22:12:12 AnMaster: why are you on irc with a migrane? 22:12:16 PDP-11. 22:12:21 nooga: follow the instructions directly. 22:12:23 without differage.. 22:12:31 oh yes, I only considered the ones beginning with -f 22:12:35 ehird, not being on it doesn't make a difference 22:12:38 but i've already got llvm 22:12:46 just turn the contrast down and it works quite ok 22:12:49 nooga: tough 22:12:51 -### is pretty weird, name-wise at least 22:12:51 it won't work properly 22:12:54 needs svn llvm 22:12:59 nooga: so just do what it says, it won't overwrite 22:13:00 uugh 22:13:01 just don't "make install' ffs 22:13:25 dup 22:13:34 DUPA, I MEAN 22:13:44 ais523, yeah 22:13:58 Gracenotes, anyway what are you implementing... 22:14:04 ehird: what is the difference between clang and llvm-gcc? 22:14:11 calculating PI by the above-linked method 22:14:11 AnMaster: http://web.archive.org/web/20001217124300/www.cacr.caltech.edu/~roy/upi/index.html <-- that? 22:14:13 clang is a new c compiler 22:14:16 yeah 22:14:16 llvm-gcc is gcc compiling to llvm 22:14:19 nooga: llvm-gcc is based on gcc 22:14:20 clang > llvm-gcc 22:14:22 I don't think I have anything to do 22:14:23 whereas clang is different 22:14:31 other than port my C++ ray tracer to Haskell 22:14:37 and figure out Haskell's SDL interface 22:14:42 also, in terms of what I got C-INTERCAL working for, llvm-gcc compiles to native code, clang to bytecode 22:14:46 but I suspect that isn't necessary 22:14:58 is speed of code generated by clang worth doing all that weird installation procedure? 22:14:58 ais523: ...? 22:15:05 nooga: 'weird'? 22:15:06 it's regular 22:15:10 and figuring out whether I need existentials to make a heterogenous list of shapes in the scene 22:15:13 you're just not any good at it :) 22:15:29 regular installation is: sudo apt-get install clang 22:15:38 nooga: haha 22:15:57 yeah it's not as if anyone ever compiles anything themselves 22:16:02 how unregular 22:16:17 okay 22:16:30 -!- pikhq has quit (Read error: 104 (Connection reset by peer)). 22:16:30 ais523, then you are doing it wrong 22:16:30 but when i get this llvm with clang compiled... 22:16:33 where to put it? 22:16:42 nooga: anywhere 22:16:48 ~/clang? 22:16:51 AnMaster: there's a packaging bug in clang over here 22:16:53 ais523, use -emit-llvm to llvm-gcc 22:16:58 which is presumably my fault as I compiled it 22:17:05 uh 22:17:10 fix it then+ 22:17:14 s/+/?/ 22:17:17 i don't want any software in my ~ 22:17:27 nooga: figure out something yourself then 22:17:36 too late, too lazy 22:17:39 too mac 22:17:49 don't touch my ~, pervert! 22:18:07 (yes, i am being annoying on purpose) 22:18:08 :D 22:18:12 >:| 22:18:35 nooga: *help service terminated* 22:19:16 nooga, I use ~/local/llvm/2.5/ 22:19:35 its clang+llvm 22:19:35 since I don't want non-package manager managed software elsewhere on the system 22:19:39 not llvm 2.5 22:20:05 ehird, yeah the system with llvm svn uses ~/local/llvm/svn/ for both clang and llvm 22:20:11 oh 22:20:13 since you place clang inside the llvm build tree 22:20:58 ehird, you can hack the main llvm build system so it builds clang too 22:21:13 ehird, just modify the list of subdirs of the tools directory 22:21:17 to add "clang" 22:21:35 this is (was?) needed for out of tree llvm/clang builds 22:25:36 uhm 22:25:38 sounds reasonable 22:25:43 but i'm too lazy 22:28:31 -!- oerjan has quit ("leaving"). 22:29:21 lol 22:29:43 PS1=... in .bashrc does not work :C 22:29:50 export foo 22:31:14 prompt like: macbook-xxxxx-yyyyyyyyyy:~ xxxxxxyyyyyyyyyyy$ is ... yuck 22:31:28 export foo 22:32:51 -!- pikhq has joined. 22:36:55 Atomic.cpp:28: error: ‘__sync_synchronize’ was not declared in this scope :C 22:37:06 and that what comes out from llvm ;p 22:40:59 :f 22:41:56 -!- tombom has quit ("Peace and Protection 4.22.2"). 22:45:04 id how to build that shit on os x 22:45:10 idk* 22:46:59 ehird: you should get a fusion io io drive for your new computer 22:48:27 the head does not build :C 22:52:24 baaww 22:57:13 -!- FireFly has quit ("Later"). 22:59:35 -!- ais523 has quit (Remote closed the connection). 23:05:32 Just pointed out a long-standing flaw in the M*U*S*H tutorial 23:13:51 my applw quacks 23:17:10 Atomic.cpp:28: error: ‘__sync_synchronize’ was not declared in this scope :C 23:17:18 that is a intrinsic in GCC 23:17:19 iirc 23:17:27 needs recent GCC probably 23:17:34 did you properly export CC 23:18:10 nooga, out of interest.. What esolang programming have you done 23:18:43 erm 23:18:46 why do you ask? 23:18:54 nooga, out of interest 23:19:07 out of answer 23:19:10 >:D 23:19:14 you haven't? 23:19:33 nooga, plus, that matches "out of question", not "out of interest" 23:19:48 uhh 23:20:08 long time ago i did some befunge and bf 23:20:17 + mangling sadol all the time 23:20:25 eg. raytracer in sadol 23:24:01 heh 23:24:15 nooga, where is the ray tracer in it 23:24:34 i wrote raytracer in sadol 23:24:38 nooga, LINK 23:24:44 ughhh 23:27:10 ffffffuuuuu 23:27:12 lost it 23:29:02 -!- coppro has joined. 23:30:47 -!- upyr[emacs] has joined. 23:33:18 -!- lament has joined. 23:33:31 GIGANTIC COCKS 23:33:33 -!- lament has left (?). 23:35:01 AnMaster: 23:35:05 ? 23:35:08 CC=gcc-4.2 23:35:11 DON'T STOP 23:35:12 BELIEVING 23:35:13 export 23:35:21 export CC=gcc-4.2 23:35:22 duh 23:35:23 oh gawd, Glee was rather good 23:35:24 ... 23:35:30 I hope they continue it in the fall 23:35:35 AnMaster: did that =.= 23:35:37 and wth was up with lament above... 23:35:38 Ok, this is the second time I accidentally spammed M*U*S*H 23:35:45 nooga, link to the ray tracer? 23:36:03 $ echo $CC 23:36:03 gcc-4.2 23:36:10 export CC=gcc-1.2 23:36:10 env | grep CC 23:36:13 lost the code ;p 23:36:19 nooga, backup fail 23:36:28 it exists 23:36:33 oh? 23:36:34 Incidentally, nobody told me that clang can act as a C++->C compiler :P 23:36:51 GregorR, maybe because we didn't know 23:37:03 but my laptop is far away 23:37:06 clang compiles to LLVM, LLVM (llc) has a C target. 23:37:12 can't access my data today 23:38:07 23:33 lament has joined (n=lament@S010600110999ad06.vc.shawcable.net) 23:38:07 23:33 lament: GIGANTIC COCKS 23:38:08 23:33 lament has left ("Paradise is exactly like where you are right now, only much better.") 23:38:10 ↑ <3 lament 23:38:20 AnMaster: nooga invented SADOL 23:38:35 ehird, he said that 23:38:46 sec 23:39:17 AnMaster: but if you want to be passive-aggressive, there are less obvious ways 23:39:30 ehird, what? 23:39:34 uhm 23:39:44 ehird, what are you talking about... 23:40:00 Hey nooga. I think you're being an idiot. I'm going to imply you're off-topic all the time by asking: what exactly have you done for esoteric programming? I seeee. 23:40:14 it wasn't particularly subtle. 23:40:16 ehird, I have no idea why you misinterpreted me like that. 23:40:17 i feel this way too 23:40:24 but nvm, i am an idiot 23:40:33 ehird, suggest a better way to ask it in English? 23:40:46 that doesn't imply that 23:40:57 to me it sounds perfectly non-agressive 23:41:00 it's the context, not the question 23:41:12 * ehird shrug. If you didn't mean that I apologize. 23:41:13 uhm, yea it sounded a bit 23:41:16 ehird, uh what do you mean 23:41:24 there wasn't much of a context 23:41:30 just a random thought 23:41:45 context: my lame bawwing about something not compiling 23:42:04 (as a hacker i should shut up and figure it out) 23:42:12 oh, that was separate. and well I don't know OS X. I know compiling on windows is a pain though 23:42:18 no idea how hard it is on OS X 23:42:29 compiling on os x is just like compiling on bsd 23:42:30 ever tried to compile anything using SDL on windows? 23:42:36 except your gcc is old unless you append -4.2 23:42:41 (even then it's still old, but :)) 23:43:51 then i don't know what to do 23:44:02 configure runs smoothly and then kaboom 23:44:12 nooga: do: 23:44:13 make clean 23:44:17 well the sync one is a GCC builtin 23:44:17 CC=gcc-4.2 ./configure 23:44:18 make 23:44:23 ehird, no 23:44:26 make distclean 23:44:26 AnMaster: Yes 23:44:30 Er, no. 23:44:30 not make clean 23:44:33 That removes ./configure. 23:44:34 ehird, what 23:44:42 ehird, no that would be make maintainer-clean 23:44:49 Are you sure? 23:45:08 ehird, make distclean remove generated Makefiles 23:45:09 and such 23:45:16 ehird, and yeah, if it uses autotools I'm sure 23:45:28 AnMaster: http://pastebin.ca/419385 23:45:31 oooold paste ;p 23:45:41 ehird, it might be make maintainerclean, not sure if it is a dash there or not 23:45:55 nooga, nice 23:46:05 nooga, how is it used 23:46:06 AnMaster: scene is hardcoded 23:46:10 ah 23:46:14 nooga, output file format? 23:46:33 echo 2 | ./BDSM2 ray.sad > image.ppm 23:46:39 ppm ok 23:46:44 where 2 is scale ;p 23:46:49 http://pastebin.ca/419385 ← is the output of this any good, i mean most toy raytracers are ~200 lines 23:46:51 well 23:46:52 and what is the unit of the scale 23:46:53 more like 500 23:47:11 AnMaster: don't remember 23:47:14 just try it 23:47:24 * AnMaster looks for the interpreter 23:47:30 sec 23:47:45 my interwebs are going slow 23:47:47 found it 23:47:51 http://regedit.gamedev.pl/BDSM/ 23:47:55 says the link on the wiki 23:48:05 yep 23:48:08 http shit isn't loading but irc works 23:48:09 maybe dns 23:48:20 oh, it loaded. 23:48:26 http's just going very slow then 23:48:33 NICE TAR BOMB 23:48:35 err 23:48:36 ZIP 23:48:40 * AnMaster stabs nooga 23:48:47 no. 23:48:47 STAB STAB STAB 23:48:52 Zip bombs is the most common way of doing zip. 23:48:56 Zips are compressed folders. 23:48:58 ehird, it is evil still 23:49:01 No it's not. 23:49:05 It's how zips are designed. 23:49:12 ehird, then unzip is broken 23:49:19 Yep. 23:49:30 g++ pch.cpp -c -o pch.obj 23:49:32 .obj 23:49:35 wth? 23:49:55 AnMaster: .obj is .o on windows 23:50:02 woot? 23:50:03 presumably the makefile was designed by a windowser 23:50:04 and why not a Makefile, why make.sh 23:50:11 AnMaster: because windows 23:50:15 ehird, *.sh 23:50:19 and? 23:50:24 windows would be *.bat 23:50:29 i'm sure it's a perfectly valid bat file too 23:50:50 Copyright (C) 2005 Adam Sawicki 23:50:50 All rights reserved 23:50:52 oh? 23:50:56 what? :D 23:51:00 why is there a gpl2.txt in there 23:51:04 err 23:51:05 gpl.txt 23:51:08 AnMaster: sounds like misplaced boilerplate 23:51:12 the page says it's GPL 23:51:14 ehird, sure does 23:51:15 probably this is standard 23:51:16 so 2-1, GPL wins 23:51:20 nooga: no, it is not 23:51:24 all rights reserved != gpl, very strongly 23:51:37 nooga, didn't you say you coded it? 23:51:37 all rights reserved = just watch the blinkenlights, hands off 23:51:41 sdandard header generated by his MSVC 23:51:43 nooga: Adam Sawicki 23:51:46 Martin Gasperowicz 23:51:57 He didn't code the interpreter, evidently. 23:51:57 Adam Sawicki implemented SADOL 23:52:04 This is free software. You may redistribute copies of it under the terms of 23:52:04 the GNU General Public License . 23:52:04 There is NO WARRANTY, to the extent permitted by law. 23:52:06 i coded it, in ruby 23:52:07 ;d 23:52:10 that is the standard GPL one 23:52:15 and now i'm coding a compiler 23:52:18 AnMaster: 23:51 nooga: sdandard header generated by his MSVC 23:52:20 MS Visual C 23:52:27 his ide generated the all-rights-reserved thing 23:52:28 ouch ok 23:52:29 and he added the gpl manually 23:52:31 a simple mistake 23:52:36 ehird, in the output of the help? 23:52:42 AnMaster: ok, that I don't know 23:52:45 $ ./BDSM2 23:52:45 Badly Developed SADOL Machine 23:52:45 Version: 2.0, 21 Dec 2005 23:52:45 Copyright (C) 2005 Adam Sawicki 23:52:45 All rights reserved 23:52:47 that is what I meant 23:52:50 "curse you meddling penguins" 23:52:51 irc ate newline 23:52:54 blooper then 23:53:17 I ate your newline 23:53:25 where is the damn raw line on pastebin... 23:53:31 there used to be a link 23:53:33 it seems gone 23:53:44 oh they moved it 23:54:02 AnMaster: did you run it? 23:54:44 nooga, running it 23:54:46 how long does it take 23:54:57 4 billion years 23:55:06 on what machine? 23:55:22 2 GHz Sempron, 64-bit 23:55:46 everything takes 4 billion years on a sempron 23:56:15 well.. the fact that the interpreter is compiled without -O doesn't help... 23:56:46 * AnMaster recompiles 23:57:32 done it says 23:57:42 ok that is quite nice 23:57:53 AnMaster: upload the output? 23:57:56 converted to png, preferably 23:58:06 ehird, um. It is trivial on your intel :P 23:58:14 beh fine 23:58:16 i'm so the lazy though. 23:58:17 so clearly uploading it will take longer 23:58:27 than you running it 23:58:27 :P 23:58:37 's one thing my new i7 won't be able to do: read my mind and act accordingly 23:58:39 ehird, since it took less than a minute on a sempron 23:58:54 in fact it'll be even more tedious, since I won't be waiting for any computations— so all the time spent will be drudgery ;-) 23:58:56 ehird, I would be scared if it could 23:59:02 Powerful computers ruin lives! 23:59:12 -!- oerjan has joined. 23:59:12 AnMaster: eh, I'd like a mind-controlled computer 23:59:15 ehird, because interpreting thoughts would be Strong AI 23:59:15 nice and efficient 23:59:16 I bet 23:59:19 no shit 23:59:26 that's like, the strongest AI you can get 23:59:29 ehird, dangerous.. It will be an EVIL i7 23:59:32 taking over the world 23:59:42 clearly yudkowsky was not properly involved 23:59:54 AnMaster: ah, that make.sh is the linux compilation file for unix, but written by a unixer