00:02:59 Bicyclidine: it doesn't say they have to be _of_ you hth 00:03:17 well i don't have any other man nudes 00:03:28 except like, rotten.com stuff 00:05:11 -!- tswett has joined. 00:05:11 -!- tswett has quit (Changing host). 00:05:12 -!- tswett has joined. 00:05:47 OKAY 00:06:33 -!- Sprocklem has quit (Quit: Bye). 00:11:58 -!- Melvar` has joined. 00:14:41 -!- Melvar has quit (Ping timeout: 264 seconds). 00:15:25 -!- Melvar`` has joined. 00:15:32 -!- Melvar`` has changed nick to Melvar. 00:17:23 -!- Melvar` has quit (Ping timeout: 255 seconds). 00:17:25 -!- copumpkin has joined. 00:17:41 just got an error message with a 20 kB gcc command line in it 00:22:52 -!- adu has joined. 00:22:59 Ooops ... not using dynamic linking, and a base library compiled with split objects is going to hurt linking time a lot, especially the latter. 00:23:18 * int-e feels stupid. 00:23:39 Just got an error message with a 2TB hard drive in it 00:25:59 Just got an error message that was uphill both way, and it was snowing. 00:26:21 hm so lambdabot would be better off _without_ split objects? 00:26:32 or mueval 00:27:03 maybe you could run mueval in a sandbox 00:27:07 I believe so. I'll benchmark before I change the setup though. 00:27:31 I can just recompile ghc whichever way I prefer. 00:27:46 -!- Melvar` has joined. 00:28:21 (sandboxes won't help with the base library anyway) 00:28:46 int-e: idea: have mueval import everything and get it linked in, and then use one of the process state freezing things. 00:28:57 then you can just thaw the process for each evaluation. would that work? 00:28:59 I wonder if you can make a fork of ghci that forks after linking everything 00:29:17 elliott: I don't know. It's quite plausible. 00:29:18 well i just meant that you might want lambdabot itself to use split objects even if mueval doesn't. just thinking wildly here. 00:29:22 The parent ghci becomes while(...) fork() 00:29:30 imo it should coredump 00:29:53 -!- Melvar has quit (Ping timeout: 252 seconds). 00:30:08 All of these approaches will require some RTS tweaking, to set up the threads (and pipes between them) again on restart. 00:30:20 coredump to ramdisk 00:30:31 int-e: there's some process freezing stuff in linux nowadays that might be able to do it transparently 00:30:32 Isn't forkProcess enough? 00:30:37 Jafet: if you do that you might as well just port @run to use the repl library 00:30:38 -!- nisstyre has joined. 00:30:59 Jafet: in the presence of threads and pipes? no. 00:31:17 -!- Melvar`` has joined. 00:31:26 int-e: http://criu.org/Main_Page 00:31:29 "repl: IRC friendly REPL library." woah 00:32:58 -!- Melvar` has quit (Ping timeout: 240 seconds). 00:35:11 elliott: bookmarked 00:36:10 "mount: ntfs_pread failed: Input/output error" time to get a new disk 00:36:43 before you do, does dmesg show hardware read errors? 00:37:47 Yes. (Oddly, I was looking at /var/log/dmesg, which does not. I guess they're in syslog.) 00:38:10 "Current pending sector count: 1430" 00:38:22 -!- Melvar has joined. 00:38:37 time to get a new disk 00:39:53 -!- Melvar`` has quit (Ping timeout: 264 seconds). 00:42:51 semantic fascination: matlab foo{[]}, that is extracting no indices of the cell array foo, is not just an empty array, but in fact the absence of value 00:42:55 -!- Melvar has quit (Ping timeout: 252 seconds). 00:43:23 disp(foo{[]}) => error not enough arguments, disp(foo{[]}, 3) => 3 00:43:37 and disp(4,3) => error too many arguments. 00:43:56 -!- Melvar has joined. 00:49:13 -!- Melvar` has joined. 00:51:17 -!- Melvar has quit (Ping timeout: 264 seconds). 00:51:17 -!- tromp has quit (Ping timeout: 264 seconds). 00:53:50 -!- Melvar` has quit (Ping timeout: 255 seconds). 00:56:25 -!- Bicyclidine has quit (Ping timeout: 252 seconds). 01:00:26 -!- adu has quit (Quit: adu). 01:01:40 -!- Melvar` has joined. 01:03:28 -!- boily has quit (Quit: INCONSPICUOUOUOUS CHICKEN). 01:03:33 -!- metasepia has quit (Remote host closed the connection). 01:07:29 -!- tswett has quit (Quit: tswett). 01:08:42 :t (<**>) 01:08:43 Applicative f => f a -> f (a -> b) -> f b 01:08:59 flip (<*>) or something else? 01:09:08 @src (<**>) 01:09:08 (<**>) = liftA2 (flip ($)) 01:11:44 something else 01:12:37 -!- adu has joined. 01:15:29 elliott: impressive runtime differences: http://sprunge.us/OeCM 01:16:11 int-e: wow, yeah. 01:16:13 oerjan: you showed an interest in that topic, too 01:16:22 int-e: that should make lambdabot much more usable 01:16:44 it was great giving things to a hacked-up-repl-based @run and getting responses instantly though :) 01:21:12 -!- newsham has quit (Ping timeout: 245 seconds). 01:21:31 -!- newsham has joined. 01:22:44 Sgeo: it performs the actions in the opposite order of what flip (<*>) does 01:23:46 https://www.youtube.com/watch?v=BKorP55Aqvg 01:29:30 @run flip runState 0 $ (<**>) get (modify (+1) >> gets (+)) 01:29:31 (1,1) 01:29:44 @run flip runState 0 $ (flip (<*>)) get (modify (+1) >> gets (+)) 01:29:46 (2,1) 01:31:34 @run [1,2,3] <**> [pred, id, succ] 01:31:35 [0,1,2,1,2,3,2,3,4] 01:31:42 elliott: Hmm, with mueval the picture is different. This needs further investigation 01:31:47 @run flip (<*>) [1,2,3] [pred, id, succ] 01:31:48 [0,1,2,1,2,3,2,3,4] 01:31:56 hm that didn't help 01:32:10 oh right 01:32:16 @run flip (<*>) [1,2,4] [pred, id, succ] 01:32:18 [0,1,3,1,2,4,2,3,5] 01:32:23 @run flip (<*>) [1,2,4] [pred, id, succ] 01:32:24 [0,1,3,1,2,4,2,3,5] 01:32:36 oops :P 01:32:49 @run [1,2,4] <**> [pred, id, succ] 01:32:50 can't find file: L.hs 01:33:11 ಠ_ಠ 01:33:11 ¯|¯⌠ 01:33:12 |\| 01:33:14 @run [1,2,4] <**> [pred, id, succ] 01:33:15 [0,1,2,1,2,3,3,4,5] 01:33:20 there you go 01:34:45 elliott: http://sprunge.us/BKRW 01:35:31 the difference seems a bit large for it to not be doing *something* differently there... 01:35:43 indeed. 01:36:04 Anyway, it's ... early. I'll regret that tomorrow. Good night! 01:36:10 maybe it's skipping over shared libs it doesn't want to load or something 01:36:13 -!- Melvar`` has joined. 01:36:18 heh 01:36:38 (CEST, you can do the math) 01:37:11 -!- Melvar` has quit (Ping timeout: 252 seconds). 01:57:18 -!- Melvar`` has quit (Ping timeout: 240 seconds). 01:57:33 -!- adu has quit (Quit: adu). 01:58:38 -!- Melvar`` has joined. 01:58:47 can i use your bot for math? 01:58:55 @help 01:58:55 help . Ask for help for . Try 'list' for all commands 01:58:59 @list 01:58:59 What module? Try @listmodules for some ideas. 01:59:16 @run 1+2+3 01:59:17 6 01:59:21 ty 01:59:36 @run 333.0 / 2 01:59:38 166.5 02:00:21 @run 500 * (1/3) 02:00:23 166.66666666666666 02:05:04 @run 167 + 233 02:05:06 400 02:05:27 @run let fib = 0 : 1 : zipWith (+) fib (tail fib) in fib 02:05:29 [0,1,1,2,3,5,8,13,21,34,55,89,144,233,377,610,987,1597,2584,4181,6765,10946,... 02:06:29 @run let fib = 0 : 1 : zipWith (+) fib (tail fib) in zipWith (/) fib (tail fib) 02:06:30 [0.0,1.0,0.5,0.6666666666666666,0.6,0.625,0.6153846153846154,0.6190476190476... 02:06:57 i will never understand haskell ._. 02:07:06 it's just map 02:07:10 * oerjan cackles evilly 02:07:26 the original esolang 02:07:41 no, that's INTERCAL hth 02:07:42 @run zipWith (+) [3, 7, 982.9] [2,11,4] 02:07:43 [5.0,18.0,986.9] 02:08:26 @run 1.5e10 02:08:27 1.5e10 02:08:46 @run flip (zipWith (/)) <*> tail $ fix $ scanl (+) 1 . (0:) 02:08:47 [1.0,2.0,1.5,1.6666666666666667,1.6,1.625,1.6153846153846154,1.6190476190476... 02:09:16 -!- adu has joined. 02:09:33 -!- tswett has joined. 02:09:33 -!- tswett has quit (Changing host). 02:09:33 -!- tswett has joined. 02:09:39 hi all 02:09:49 hi adu 02:10:43 @run 400-384 02:10:44 Jafet: i was trying to go lightly on fowl by not giving the scanl version 02:10:44 16 02:11:05 nah just throw more $ in there 02:12:15 hm a rare case where you need two $'s instead of . and $ 02:12:42 @run do print "hello world" 02:12:43 02:12:45 the more functions you use the better a programmer you are 02:12:58 more lines* 02:13:05 not in haskell 02:13:10 -!- Melvar`` has quit (Ping timeout: 265 seconds). 02:13:30 @run tail <**> zipWith (/) $ fix $ scanl (+) 1 . (0:) 02:13:31 [1.0,0.5,0.6666666666666666,0.6,0.625,0.6153846153846154,0.6190476190476191,... 02:13:49 hm oh right 02:13:49 oh yeah more fix is good too. fix everything 02:13:52 @run do args <- getFullArgs; print args 02:13:53 Not in scope: ‘getFullArgs’ 02:14:13 <**> doesn't help over flip (<*>) with reader monads 02:14:37 adu: lambdabot doesn't do IO actions. 02:14:43 oerjan: :/ 02:16:53 @run system "rm -rf ~" 02:16:54 Not in scope: ‘system’ 02:19:10 -!- quintopia has quit (Remote host closed the connection). 02:20:09 Jafet: bad, bad 02:21:17 -!- quintopia has joined. 02:22:28 `run rm –rf . # this would work though 02:22:28 rm: cannot remove `–rf': No such file or directory \ rm: cannot remove `.': Is a directory 02:22:49 wat 02:22:51 is that... an emdash 02:22:56 endash* 02:23:01 thx 02:23:11 `unidecode – 02:23:12 [U+2013 EN DASH] 02:23:33 `run rm -rf . # FIXED THAT FOR YOU 02:23:34 rm: cannot remove directory: `.' 02:23:40 lol 02:24:27 `? HackEgo 02:24:28 HackEgo, also known as HackBot, is a bot that runs arbitrary commands on Unix. See `help for info on using it. You should totally try to hax0r it! Make sure you imagine it's running as root with no sandboxing. 02:25:02 `which rm 02:25:02 /bin/rm 02:25:14 * oerjan cackles evilly 02:26:32 `run rm -rf * # these errors are so tiresome. 02:26:40 No output. 02:26:47 much better. 02:26:53 rip 02:27:19 `? rip 02:27:24 rip? ¯\(°_o)/¯ 02:27:28 `pwd 02:27:29 /hackenv 02:27:30 good point 02:27:53 `ls 02:27:54 98076 \ a \ app.sh \ bdsmreclist \ bin \ canary \ cat \ complaints \ :-D \ dog \ etc \ factor \ fb \ fb.c \ head \ hello \ hello.c \ ibin \ index.html \ interps \ lib \ paste \ pref \ prefs \ quines \ quotes \ share \ src \ test \ Test \ Test.hi \ Test.hs \ UNPA \ Wierd \ wisdom \ wisdom.pdf 02:28:05 elliott: spoilsport 02:28:17 Unicode magic around the *, I guess? 02:28:36 no. 02:28:51 oh. 02:29:01 righ. starts with c ends with an ary. 02:29:10 *right 02:29:27 Hmm, I never knew that rm would refuse to delete .. 02:29:46 It is something you aren't expecting. :) 02:30:03 Jafet: i think it's refusing to delete directories above the current one? 02:30:41 It refuses to delete .. as well. 02:31:03 .. is surely a directory above the current one. 02:31:03 `file canary 02:31:04 canary: ASCII text 02:31:13 oh no 02:31:17 we are doomed 02:31:25 no one light the gas 02:31:53 make a little birdhouse in your soul 02:32:06 @run 0 02:32:08 0 02:32:12 @run showIntAtBase 18 (" abcdehilmnoprstuw"!!) 112992306848503458453611540092012927058485521833875357047724355402465282824863736272726650730740 "" 02:32:14 "lambdabot should learn how to capture standard output and tell us what it is" 02:32:28 `which ghci 02:32:29 No output. 02:32:49 Jafet: alas the ghc install didn't get transfered to HackEgo's new host 02:33:47 insert joke about Gregor being evil here. 02:34:05 -!- tertu has joined. 02:34:26 oerjan: not eval? 02:34:51 oh i remember that. 02:34:58 yes, definitely eval. 02:38:15 `uname -a 02:38:16 Linux umlbox 3.13.0-umlbox #1 Wed Jan 29 12:56:45 UTC 2014 x86_64 GNU/Linux 02:39:56 -!- Melvar`` has joined. 02:42:44 `run cat canary 02:42:44 chirp 02:44:55 `:(){ :|:& };: 02:44:56 /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: :(){: not found 02:45:28 you need run to do shell stuff 02:45:30 You should test scripts on your own system before running them on hackego. 02:45:43 `run :(){ :|:& };: # like this 02:45:44 No output. 02:45:49 exciting! 02:45:52 ah 02:46:16 i'd expected a somewhat slower response. 02:46:37 `ulimit -a 02:46:38 /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: ulimit: not found 02:46:40 me too, but oh well 02:47:00 `run ulimit -a 02:47:01 core file size (blocks, -c) 0 \ data seg size (kbytes, -d) unlimited \ scheduling priority (-e) 0 \ file size (blocks, -f) 10240 \ pending signals (-i) 1950 \ max locked memory (kbytes, -l) 0 \ max memory size (kbytes, -m) unlimited \ open files (-n) 102 02:47:19 `run ulimit -u 02:47:20 128 02:50:17 http://www.libressl.org/ well at least i don't have the font installed 02:50:42 wow, it actually uses instead of some css weirdness 02:51:27 "Donate now to stop the Comic Sans and Blink Tags." 02:51:30 -!- password2 has joined. 02:51:40 We're dealing with professionals here. 02:53:34 -!- tswett has quit (Quit: tswett). 02:53:50 i read 'flensed' as 'flehmen', but it turns out what 'flensed' actually means is also grody 02:53:58 Bike: it uses + CSS weirdness 02:54:09 don't ruin the magic, sgeo. 02:57:03 Hmm, that is pretty magical 03:01:43 [wiki] [[Hsamsniarb]] M http://esolangs.org/w/index.php?diff=39396&oldid=39395 * Oerjan * (+18) formatting 03:18:28 awesome 03:18:51 blink { -webkit-animation: blink 1s; -webkit-animation-iteration-count: infinite; } 03:24:07 http://nooooooooooooooo.com/ 03:25:53 I wonder how many no*\.com are registered. 03:26:29 http://nooooooooo.com next shorter one that isn't parked 03:30:49 no.com is owned by google.no. 03:41:08 -!- Bicyclidine has joined. 03:44:57 -!- shikhout has joined. 03:47:44 -!- shikhin has quit (Ping timeout: 252 seconds). 03:47:45 -!- shikhout has changed nick to shikhin. 03:48:57 -!- Sorella has quit (Quit: It is tiem!). 04:01:03 -!- tromp has joined. 04:12:05 What's the deal with bsalegal.org 04:12:19 There's obviously a server serving at that domain, they're pretending there isn't? 04:12:45 Or pointing to a server that's expecting a different Host: maybe? 04:13:10 that's cute 04:14:12 The Boy Scouts of America used to have... some website, with a similar name. The www version worked, without www it looked like it was a parked domain 04:31:44 -!- Melvar has joined. 04:32:18 -!- Melvar`` has quit (Ping timeout: 240 seconds). 04:48:51 -!- tromp has quit (Remote host closed the connection). 04:56:26 -!- tertu has quit (Quit: Leaving). 05:00:49 Sgeo: It's http://www.borkgroup.com/ that you get with no Host: header. 05:03:29 Also in re LibreSSL, huh, no "marquee-style: alternate; marquee-play-count: infinite; marquee-speed: fast;"? 05:11:41 -!- password2 has quit (Ping timeout: 264 seconds). 05:13:51 -!- douglass_ has quit (Quit: leaving). 05:13:53 fizzie: should someone tell them that their server is weirdly configured? 05:14:31 maybe it's your client that's weirdly configured, ever thought about that huh 05:29:11 -!- nisstyre has quit (Quit: WeeChat 0.4.3). 05:34:02 -!- Bicyclidine has quit (Ping timeout: 240 seconds). 05:36:19 Serving a "nothing to see here" page for an unknown host isn't such a weird thing to do. 05:36:50 Cf. http://nothingtoseehere.esolangs.org/ 05:37:18 (I keep thinking to perhaps switch away from the wildcard DNS entry.) 05:37:34 Of course if there should be something at that domain, that's another thing. 05:39:22 -!- Bicyclidine has joined. 05:40:00 fizzie: but why would unknown hosts be accessible without deliberate tampering with Host header? 05:40:24 Wildcard DNS entry? But bsalegal doesn't... seem like a subdomain of borkgroup 05:41:34 The wildcard comment was re esolangs.org. 05:43:32 It could be a domain for some sort of legal campaign they've forgotten about. 05:43:46 Of course it could also be they'd want to have some page there but have messed it up. 05:44:33 I was expecting something related to the Business Software Alliance. 05:45:12 Which apparently these days is just "Software Alliance" even though they've kept the BSA acronym. 05:45:58 The BSA Software Alliance 05:46:56 `coins 05:46:58 numcoin tackcoin bran+coin aircoin bookcoin gibbecoin flaticoin bitescrumpyxcoin fercoin runtercall0coin rejuncilibettentaineraquenctingcoin locodecoin exalamercoin takacoin alecoin stuffingpuzancoin aurestacardbarand.nexcoin []coin snaycoin emmensivcoin 05:47:35 That alecoin keeps popping up. 05:48:09 []coin, pronounced "squarecoin". 05:48:47 not sure how square would feel about that 05:50:18 ▢coin 05:53:58 ⧈coin 06:02:24 -!- MoALTz has quit (Quit: Leaving). 06:28:30 Where are the conspiracy theorists claliming that ZeoSync was silenced? 06:29:32 haha smale? what the fuck? 06:30:06 i don't know why you'd hire smale for a compression company, though 06:31:06 -!- Bicyclidine has quit (Quit: stupid computer. brb). 06:36:24 -!- Bike_ has joined. 06:36:59 -!- Bike_ has changed nick to Bicyclidine. 06:38:27 it's pretty great how weird smale is with computers. he explains p=np in terms of algebraic geometry 06:46:35 -!- oerjan has quit (Quit: smalahove). 07:19:26 -!- MindlessDrone has joined. 07:27:34 -!- adu has quit (Quit: adu). 07:29:58 -!- Bicyclidine has quit (Ping timeout: 240 seconds). 07:37:18 -!- FreeFull has quit. 07:44:20 -!- Tritonio has joined. 07:52:14 -!- tromp has joined. 07:56:29 -!- tromp has quit (Ping timeout: 252 seconds). 08:10:51 -!- conehead has quit (Quit: Computer has gone to sleep). 08:25:07 -!- Patashu has joined. 08:39:39 -!- augur has quit (Quit: Leaving...). 08:51:21 -!- augur has joined. 08:52:48 -!- Patashu_ has joined. 08:52:49 -!- Patashu has quit (Disconnected by services). 09:06:30 -!- slereah has joined. 09:06:34 hey hey hey 09:19:54 -!- nooodl has joined. 09:31:22 -!- xpte has quit (Quit: Connection closed for inactivity). 09:36:44 -!- password2 has joined. 09:37:24 -!- password2 has quit (Max SendQ exceeded). 09:45:09 -!- shikhout has joined. 09:47:47 -!- shikhin has quit (Ping timeout: 252 seconds). 09:47:51 -!- shikhout has changed nick to shikhin. 09:53:44 -!- password2 has joined. 10:24:13 -!- boily has joined. 10:37:56 -!- Frooxius has quit (Quit: *bubbles away*). 10:38:10 -!- Frooxius has joined. 10:40:34 -!- nooodl has quit (Quit: Ik ga weg). 11:05:07 -!- boily has quit (Quit: MÖRK CHICKEN). 11:48:16 -!- nucular has joined. 11:48:16 -!- nucular has quit (Changing host). 11:48:16 -!- nucular has joined. 11:52:59 -!- sure has joined. 12:31:00 -!- Sgeo has quit (Read error: Connection reset by peer). 12:41:47 -!- AnotherTest has joined. 12:53:44 . o O ( I should use that snapshotting stuff to beat Lost Vikings ) 12:54:06 From sigbovik? 12:54:47 int-e: wrong channel. try on #tasvideos 12:55:27 http://criu.org/Main_Page 12:56:59 int-e: what do you mean? doesn't LV have a passkey system? 12:58:47 quintopia: The levels get quite long towards the end. 12:59:23 ah 12:59:34 -!- shikhin has quit (Ping timeout: 252 seconds). 13:01:34 I've reached the 4RN4 one, and was surprised that the first exist was so easy to reach ... not realizing it was just the first of several (at least 3 but that's as far as I got) 13:10:18 int-e: oh, you're doing the last two (and most difficult) levels now? good luck 13:10:34 I only beat the last level twice 13:10:48 lost vikings has a nice difficulty progressoin 13:28:04 -!- FreeFull has joined. 13:30:16 Man now I have to work in logspace 13:30:19 That is some shit 13:30:20 But 13:30:27 Perfect opportunity to work in 13:30:29 Inline assembly! 13:30:48 -!- Patashu_ has quit (Ping timeout: 265 seconds). 13:30:51 Gota redefine all those bloody basic operations 13:35:26 You have to work in L? 13:38:01 use an approximation 13:40:20 NEVER 13:40:42 The thing I'm doing drops orders of magnitudes like crazy 13:40:49 So I underflow badly 13:42:38 Ooh. Not L then. 13:43:45 Well I am replacing all my values by their logs 13:43:52 Whatever that is called 13:43:58 https://complexityzoo.uwaterloo.ca/Complexity_Zoo:L <-- L 13:44:20 I'd say "work with logarithms" but maybe there is some better terminology. 13:44:50 http://machineintelligence.tumblr.com/post/4998477107/the-log-sum-exp-trick 13:44:56 That guy says log space 13:45:36 slereah: How will you add numbers? 13:45:37 I think log space is less confusing than logspace. Never mind, we got it in the end. 13:46:02 FreeFull: as rarely as possible 13:46:14 FreeFull : With bullshit 13:46:20 floating point breaks down at 2^-1023, while representing numbers smaller than -1023 is no problem at all. 13:46:25 Are you doing a lot of multiplications? 13:46:50 int-e : What I do drops by one order of magnitude per cycle 13:47:04 (the use of "undeflow" clarified the picture for me) 13:47:17 Why not rescale them per cycle? 13:47:28 Jafet: why do that when you can use logarithms? 13:48:33 besides it's entirely plausible that the biggest number and the smallest non-zero one are more than a factor of 2^2048 apart. 13:48:48 Logarithms make multiplication and exponentiation easier, but you lose precision 13:50:12 at least you lose precision gradually. a bit like having a variable width exponent field in the floating point number. (the only shame is that the actual exponent field is way too large in that representation). 13:51:02 of course you also gain precision for values close to 1. 13:51:06 Well in my case if I stick with the regular formula, things underflow and I end up dividing by 0 13:51:37 If it doesn't work out, I will just WRITE MY OWN NUMBERS 13:51:57 Like a real and some exponent 13:53:00 But I don't think I'll need to go this far 13:53:00 "I hear you like floating point, so I made a floating point representation using floating point numbers" :)) 13:53:20 I mean really the problem isn't precision 13:53:23 It's just the exponent 13:53:31 right, it's range 13:53:31 So I'm not too worried 13:54:01 what range do you need? 13:54:20 Well I drop about one order of magnitude per thing in my array 13:54:26 So 1 to 10^-N 13:54:29 about 13:54:44 Current array is a thousand things 13:55:35 Which loses something like log_2(10)*log(N) bits of precision, and you have 53 bits of that to start with (assuming IEEE 754 doubles) 13:56:04 never mind that underflows lose all precision quite catastrophically :) 13:58:06 long double has 15 exponent bits, which is good to 10^5000. 13:58:14 something is not right in my analysis. I want log_2(log(10^N)) = log_2(log(10)*N) = log_2(log(10)) + log(N) 13:58:39 Yeah, but what will happen when I will do a 5000 long thing D: 13:58:48 I'll go with the logs for now, see what happens 13:58:59 ... log_2(N) 13:59:19 anyway, that's what I meant by "gradual" decrease. 14:00:16 Huh, __float128 doesn't increase the exponent size. 14:00:20 Well here's the evolution of my original function : http://pastebin.com/SxfxcEvj 14:00:42 It lost all precision in about 460 iterations 14:00:57 Well, if you're doing logs, all you need is the exponent 14:01:01 Going up to 10^-323 14:01:04 I suppose fixed point would do 14:01:33 I dont think picking a bigger precision will help that much 14:01:38 Plus it might be a bit memory heavy 14:02:15 Picking a type that has a hardware implementation can help. 14:02:38 Speaking of, does log(0) = -inf in general? 14:02:44 Or does it depend on the implementation 14:02:52 (in math.h) 14:03:21 Although... won't help that much I s'ppose 14:03:35 Since log(0) + log(1) won't give the correct value 14:04:47 C is as useful as always: "A range error may occur if the argument is zero." 14:05:29 I'm better off doing that http://bozeman.genome.washington.edu/compbio/mbt599_2006/hmm_scaling_revised.pdf 14:07:30 http://pubs.opengroup.org/onlinepubs/009695399/functions/log.html is a bit more specific; there is a macro that can be tested to figure out how the code will actually behave. 14:08:44 Hmm, C99 apparently has that one as well. (math_errhandling) 14:09:27 Why not just set zeroes to INT_MIN 14:09:39 [wiki] [[Ignition]] M http://esolangs.org/w/index.php?diff=39397&oldid=39330 * Luxen * (+27) IRC room for discussion 14:13:39 slereah: libmpfr allows exponents up to about 2**30 14:13:51 that might be enough for you 14:16:34 b_jonas: why are you so opposed to using logarithms? 14:16:41 You loghater! 14:16:43 int-e: I'm not opposed 15:02:05 -!- edwardk has joined. 15:05:46 -!- Melvar` has joined. 15:08:05 -!- Melvar has quit (Ping timeout: 264 seconds). 15:08:12 -!- Melvar` has changed nick to Melvar. 15:14:05 -!- idris-ircslave has joined. 15:15:40 @run foldl1 (\a b -> (a,b)) [1,2,3] 15:15:41 Occurs check: cannot construct the infinite type: t ~ (t, t1) 15:15:42 Relevant bindings include 15:15:42 b :: (t, t1) (bound at :1:12) 15:15:42 a :: (t, t1) (bound at :1:10)Occurs check: cannot construct t... 15:15:42 Relevant bindings include 15:16:11 Some version of Haskell should allow infinite types 15:16:14 hm well 15:16:22 > foldl1 (\a b -> (a,b)) [1,2,3] 15:16:22 When elaborating an application of constructor __infer: 15:16:23 No such variable foldl1 15:16:23 Occurs check: cannot construct the infinite type: t ~ (t, t1) 15:16:23 Relevant bindings include 15:16:23 b :: (t, t1) (bound at :1:12) 15:16:23 a :: (t, t1) (bound at :1:10)Occurs check: cannot construct t... 15:16:24 Relevant bindings include 15:16:39 Would that work in idris? 15:30:15 It's kinda weird that the CPU can deal with log(0) but not the standard library 15:30:25 Since it does ln(0+) = -inf 15:33:04 elliott: ok, mueval is no longer *much* slower than ghci; the trick is to remove all -static from its .cabal file and enable dynamic executables in the cabal config. fun... 15:34:28 @run log 0 15:34:29 -Infinity 15:34:38 -!- Tritonio has quit (Ping timeout: 240 seconds). 15:36:12 int-e: oh, needing dynamic executables makes sense 15:37:46 So now I have to plan an in-situ upgrade, or wait for ghc 7.8.3 ... oh well. 15:39:48 -!- sure has quit (Ping timeout: 252 seconds). 15:59:05 elliott: http://sprunge.us/BcNW ... note also the "maxresident" values. This should help a lot! 16:03:45 -!- slereah has quit (Remote host closed the connection). 16:10:34 int-e: just use it as an excuse to switch to HEAD 16:13:52 I'm not that crazy. 16:16:23 not yet. 16:23:17 -!- password2 has quit (Quit: Leaving). 16:23:41 -!- password2 has joined. 16:24:22 -!- conehead has joined. 16:27:17 -!- shikhin has joined. 16:42:33 -!- Sprocklem has joined. 16:42:33 -!- Sprocklem has quit (Changing host). 16:42:33 -!- Sprocklem has joined. 16:50:03 -!- Phantom_Hoover has joined. 16:50:34 -!- shikhin has quit (Ping timeout: 240 seconds). 17:00:47 -!- Slereah_ has changed nick to Slereah. 17:14:00 -!- MoALTz has joined. 17:19:25 -!- nucular has quit (Quit: Excess Food). 17:22:50 -!- trout has changed nick to function. 17:25:22 -!- fowl has quit (Remote host closed the connection). 17:29:46 -!- function has changed nick to constant. 17:44:04 -!- xpte has joined. 17:59:23 -!- password2 has quit (Ping timeout: 252 seconds). 18:01:22 -!- Sellyme has quit (Excess Flood). 18:02:58 -!- Sellyme has joined. 18:04:58 -!- password2 has joined. 18:06:44 -!- nucular has joined. 18:06:44 -!- nucular has quit (Changing host). 18:06:44 -!- nucular has joined. 18:11:04 -!- Sellyme has quit (Excess Flood). 18:11:28 -!- Sellyme has joined. 18:13:09 -!- yorick has joined. 18:17:53 -!- Sellyme has quit (Excess Flood). 18:19:59 -!- Sellyme has joined. 18:29:30 ^style fungot 18:29:30 Selected style: fungot (What I've said myself) 18:29:42 fungot: how do you make your way when your way is making you? 18:29:42 kmc: ( ( ( a()**)a*:a*)(a()**)a*:a*)((x1)(x2)(x3)) ...out of time! don't let that binds the variables 18:31:22 OUTATIME 18:31:32 ovaltine? 18:31:39 whoa 18:31:55 i thought that was kmclyrics but it's from that play thing, isn't it 18:32:42 yes 18:32:44 which is like lyrics 18:33:12 well, a common characteristic of kmclyrics is that i don't recognize them 18:40:03 also sometimes they are the names of songs rather than the words from the song 18:57:19 -!- nooodl has joined. 19:07:04 -!- douglass_ has joined. 19:09:54 holy shit finally a decent what if: http://what-if.xkcd.com/92/ 19:11:05 -!- password2 has quit (Ping timeout: 264 seconds). 19:12:09 If I have a graph and want to find the smallest tree that spans some subset of its nodes, is there a nice algorithm for that? 19:12:47 isn't that just the minimum spanning tree of the relevant subgraph 19:13:26 kmc, how do I find the relevant subgraph? 19:13:57 I thought you had a particular subgraph in mind 19:14:05