00:10:02 -!- skj3gg_ has joined. 00:10:02 -!- skj3gg_ has quit (Client Quit). 00:10:25 -!- shikhin_ has joined. 00:10:40 -!- ais523 has quit (Read error: Connection reset by peer). 00:10:56 -!- ais523 has joined. 00:12:02 http://www.spoj.com/SHORTEN/problems/MB2/ 00:12:13 Ah yes gmail the only email I only ever bother checking every other decade 00:12:29 next time I check it will be when I'm 20 00:12:51 can someone tell me how this can be solved with 12 chars when just to create the constant number 10 you need 10 chars? 00:12:53 -!- skj3gg has quit (Quit: ZZZzzz…). 00:13:01 am I missing something? :-S 00:13:46 -!- shikhin has quit (Ping timeout: 264 seconds). 00:14:42 Tritonio: yes 00:14:50 you don't need to care about 10 00:15:08 but the input ends with that char and if you "," after the end of input it just hangs. 00:15:41 I doubt it does 00:15:45 most bf implementations don't behave that way 00:16:01 they return chr(0)? i'll try again with that assumption. 00:16:09 or EOF? 00:16:14 0 or 255 or no change 00:16:17 it's probably 0. 00:16:27 I just solved it in 10 in vim. 00:16:36 you can look at the source of bff ( http://swapped.cc/#!/bff ), or there are other ways to find out too 00:16:59 thanks i'll check the source and see how it behaves. 00:17:20 also, ideone uses most of the same language versions, so you can do tests there 00:17:31 for spoj in general 00:17:44 actually maybe my solution is 11, but whatever. 00:18:18 if you want me to tell you the EOF behavior, i also don't mind just saying it :p 00:18:58 mitchs no thanks I'll figure it out. :-) Cheers elliott and mitchs. :-) 00:19:04 ok :) 00:19:26 mitchs: is that 10 byte solution yours? 00:19:58 mitchs: if yes, how do you deal with gur arrq gb fxvc gur ynfg vachg olgr, naq gur (cerfhzrq) arrq gb abg eha bss gur yrsg bs gur gncr? 00:20:38 i am confused, but pleasantly so 00:20:42 haven't used vim 00:20:43 rot13 :p 00:20:51 I didn't want to spoil Tritonio. 00:21:31 I'll get out for a moment elliott. brb in 10 min or so. ;-) 00:21:36 oh sorry 00:21:43 first of all there are pm's lol 00:21:44 -!- skj3gg has joined. 00:21:46 -!- paul2520 has quit (Quit: WeeChat 1.0). 00:21:50 rot13 is more fun 00:21:57 and secondly you said "didn't". 00:22:03 it is. 00:22:05 -!- paul2520 has joined. 00:22:05 aha, you meant you wrote it using vim, not that it's a vim solution like on anagol, of course 00:22:29 -!- paul2520 has changed nick to Guest16073. 00:22:41 well lots of poeple are tied for first place with 12 bytes, which is dictated by the EOF behavior 00:23:12 -!- Guest16073 has changed nick to paul2520. 00:23:19 -!- paul2520 has quit (Changing host). 00:23:19 -!- paul2520 has joined. 00:23:44 oh 00:23:54 I forgot that Tritonio said 12 chars. 00:26:44 also, the standard judge on spoj works like your_output.split() == expected_out.split() 00:30:05 put 00:31:53 -!- CrazyM4n has joined. 00:32:44 -!- Sprocklem has quit (Ping timeout: 245 seconds). 00:43:01 -!- tswett has joined. 00:43:02 -!- tswett has quit (Changing host). 00:43:02 -!- tswett has joined. 00:45:35 -!- tswett has quit (Client Quit). 00:47:09 -!- AndoDaan has joined. 00:49:51 -!- skj3gg has quit (Quit: ZZZzzz…). 00:51:10 -!- zzo38 has joined. 00:52:03 Creepy Suzan? 00:52:20 -!- augur has quit (Ping timeout: 272 seconds). 00:53:12 -!- mhi^ has joined. 00:54:17 is it using int sized cells??? so overflows don't happen at 256? https://github.com/apankrat/bff/blob/master/bff.c 00:55:15 -!- Sprocklem has joined. 00:56:49 Tritonio: yes 00:57:31 -!- augur has joined. 00:59:27 -!- jbkcc has joined. 00:59:49 -!- Sprocklem has quit (Ping timeout: 244 seconds). 00:59:53 aww... :-( 01:00:59 huh, why is it using -1 for eof in one case (getc for reading from stdin) and 0 in another (getc_ext for readinf from a file)? 01:01:47 -!- skj3gg has joined. 01:02:14 Tritonio: it may be enough to change the type of 'v' and the math in grow() 01:03:11 int-e nah I don't need an interpreter. I just wanted to see how the one in spoj.com behaves... 01:03:35 is that the one they're using? 01:03:56 yes 01:04:50 hmm. 01:14:14 solved it with 13. lets see how i can get to 12 01:16:11 -!- skj3gg has quit (Quit: ZZZzzz…). 01:22:35 ah, that's the trick. 01:23:29 -!- aloril has quit (Ping timeout: 246 seconds). 01:27:03 and spoj is horrible with all those banners. (I usually use Adblock but I tried that site with fresh profile...) 01:35:01 -!- boily has joined. 01:35:34 int-e just found the solution too. spoj was counting the emptyspace too apparently... :-) 01:36:53 -!- Patashu has quit (Remote host closed the connection). 01:37:03 s/int-e/int-e I 01:37:07 -!- aloril has joined. 01:37:28 -!- Phantom_Hoover has quit (Ping timeout: 245 seconds). 01:37:38 -!- Patashu has joined. 01:37:51 -!- jbkcc has quit (Quit: My Mac has gone to sleep. ZZZzzz…). 01:39:22 -!- callforjudgement has joined. 01:41:38 -!- ais523 has quit (Ping timeout: 245 seconds). 01:42:47 -!- callforjudgement has changed nick to ais523. 01:47:15 -!- arjanb has quit (Quit: zzz). 01:55:21 -!- ais523 has quit. 01:55:36 -!- adu has joined. 01:57:14 -!- aloril has quit (Read error: Connection reset by peer). 01:57:31 -!- aloril has joined. 01:57:49 Tritonio: I do indeed with one of the bots had a regex replacer 01:58:14 s/I do/cake 01:59:09 doesn't work anymore? 02:00:15 fungot: do you even replace? 02:00:16 boily: i mean, hell: making a really simple idea, but unpractical with the fnord 02:00:18 Have you written a program to compile SQL codes into native codes or any other VM codes? 02:00:27 Tritonio: too impratical. it has fnord. 02:01:20 SQLite can compile into its own VM codes if the schema (and possibly analysis data) is known, therefore other program could use the EXPLAIN command in SQLite to use a partially compiled code to compile into something else. 02:03:38 -!- augur has quit (Quit: Leaving...). 02:07:13 -!- adu has quit (Quit: adu). 02:07:44 -!- adu has joined. 02:07:55 -!- adu has quit (Client Quit). 02:08:28 This tickles my fancy 02:09:51 -!- adu has joined. 02:14:21 -!- augur has joined. 02:16:17 -!- jbkcc has joined. 02:41:54 -!- skj3gg has joined. 02:45:38 -!- boily has quit (Quit: REPRESENTATIVE CHICKEN). 03:13:56 -!- skj3gg has quit (Quit: ZZZzzz…). 03:59:53 -!- Tritonio_ has joined. 04:03:06 -!- Tritonio has quit (Ping timeout: 276 seconds). 04:03:16 -!- Tritonio_ has changed nick to Tritonio. 04:10:33 -!- Froox has quit (Quit: *bubbles away*). 04:12:54 -!- Frooxius has joined. 04:27:20 -!- chaosagent has joined. 04:41:57 -!- Zefphex has changed nick to Elizaveta. 04:42:11 -!- GeekDude has left ("{{{}}{{{}}{{}}}{{}}}"). 04:45:19 The chat... *dramatic pause* ... HAS DIED! 04:46:10 Are you sure? 04:46:22 I am not quite so sure. 04:46:33 * Elizaveta pets zzo38 04:48:37 Good Kitty. 04:50:26 I wanted to make up some Magic: the Gathering cards based on some stuff in my "level20.tex" story, and related things I have done; I already have "Kjugobe's Timer". 04:50:57 Never heard of this. 04:51:18 -!- Sgeo has joined. 04:51:24 Then you must learn. How many things have you failed to heard of? 04:52:19 I haven't the slightest idea. 04:53:52 -!- Sgeo_ has quit (Ping timeout: 244 seconds). 05:04:13 -!- Lymia has quit (Ping timeout: 256 seconds). 05:08:16 Imagine if the newtons laws of motion went like this every thing has an equal and unequal reaction So if you punched a wall it would push you back with the force of a speeding truck 05:08:28 oh man the world would be crazy 05:09:07 "Imagine punching somebody so hard that they turned into a door. Then you found out that’s where ALL doors come from, and you got initiated into a murder club that makes doors. The stronger you punch, the better the door. So there are like super strong murderers who punch people into Venetian doors and shit." 05:09:31 ... 05:09:55 Oh 05:09:57 my god 05:10:15 It makes So much sense 05:10:30 Nonsense. One does not simply punch into more doors. 05:11:02 Imagine if all light liquified 05:11:09 and it just splashed down 05:12:06 punny. 05:12:44 * Elizaveta pets AndoDaan 05:12:48 How was your day 05:13:09 Unproductive, but quick. 05:13:27 s/quick/quick to pass 05:13:52 And yours? 05:14:25 Good. 05:16:14 -!- nys has quit (Quit: quit). 05:23:42 -!- shikhin_ has quit (Ping timeout: 276 seconds). 05:28:59 -!- Elizaveta has changed nick to Lilax. 05:29:29 -!- CrazyM4n has quit (Quit: sleep). 05:47:00 Did you just explain the Door Lords from Adventure Time? 05:47:20 -!- Lymia has joined. 05:49:03 maybe that's another power they have 05:49:05 And since each door comes with a key, all the magic keys they have are from doors they made by punching people 05:50:26 -!- oerjan has joined. 05:54:50 -!- jbkcc has quit (Quit: My Mac has gone to sleep. ZZZzzz…). 06:00:49 There are probably really powerful ones out there with a huge bounty on their heads because of how many people they've taken down and how much they've stolen, and they have crazy, hard to open doors 06:01:32 just be sneaky by punching people into trap doors 06:01:37 also, what a great line for oerjan to join on 06:02:14 they're the real door lords 06:02:22 and nobody can catch them 06:02:31 Because they are door lords 06:02:33 beats pan lords, I suppose 06:03:18 I'm sorry for doing this to you oerjan 06:03:29 's ok i mostly blame elliott 06:03:52 * Lilax pets oerjan 06:03:53 thanks :( 06:03:55 Gud kitty 06:04:34 So like how was your day All persons available 06:07:37 -!- AndoDaan_ has joined. 06:09:24 -!- AndoDaan has quit (Ping timeout: 245 seconds). 06:19:58 -!- hjulle has joined. 06:39:06 -!- AndoDaan_ has changed nick to AndoDaan. 06:44:38 -!- AndoDaan has quit (Read error: Connection reset by peer). 06:59:23 @tell boily -!- boily has quit (Quit: TRIGOTILLECTOMIC CHICKEN). <-- that's not a word, now you're just cheating tdnh 06:59:23 Consider it noted. 07:06:37 J_Arcane_: is your nick registered if so Set enforce and the ghost anyone using it <-- usually it's not someone else using it, but themselves disconnecting and reconnecting, and the old nick not being released fast enough 07:06:52 so enforce won't help 07:07:09 tl;dr: some people have crappy connections 07:08:01 `` sed -i 's/M/Trigotillectomic M/' wisdom/boily 07:08:03 No output. 07:08:04 * oerjan suspect "themselves" is ungrammatical in that sentence but doesn't know a better way 07:08:12 `? boily 07:08:13 boily is monetizing a broterhood scheme with the Guardian of Lachine. He's also a NaniDispenser, a Trigotillectomic Man Eating Chicken and a METARologist. He is seriously lacking in the f-word department. 07:08:42 approved. 07:13:17 oerjan: it parses okay to me 07:13:21 a little awkward maybe but eh 07:13:29 okay I should sleep actually 07:13:30 bye 07:13:37 *suspects 07:14:19 it seems i and elliott are currently maximally out of phase 07:14:27 * oerjan just got up 07:15:23 which means my sleeping schedule is currently "normal" 07:19:18 -!- quintopia has quit (Ping timeout: 272 seconds). 07:20:19 -!- quintopia has joined. 07:22:15 -!- Lilax has changed nick to Lilax|poof. 07:22:55 u wot 07:23:00 I'm not gonna sleep 07:23:09 I thought elliott went to bed 07:24:49 ok now I'm just gonna 07:24:54 -!- Lilax|poof has changed nick to Lilax. 07:39:28 -!- adu has quit (Quit: adu). 07:42:00 -!- AndoDaan has joined. 08:01:27 * Lilax pets oerjan 08:01:39 How's your day goin' 08:02:47 you do not have clearance for that information hth 08:03:41 -!- ^v^v has quit (Read error: Connection reset by peer). 08:04:05 -!- ^v^v has joined. 08:06:25 Ah 08:06:29 well then 08:07:07 oerjan, when somebody uses the /// here, does that automatically corrected in the log? 08:07:25 that get* 08:07:56 probably not 08:08:04 or are you asking 08:08:24 `8ball 08:08:25 Signs point to yes. 08:08:42 hmm. 08:08:48 [/1] then [//2] for the 2 / that are displayed 08:08:52 eg; 08:08:56 // 08:09:09 I used three /'s to make 2 08:09:16 go see 08:09:20 I'm just wonder if there's any functionality to using on here ///, or if it's just visual. 08:09:24 okay. 08:10:16 ya AndoDaan 08:10:27 it shows // instead of /// 08:10:38 hmm. 08:10:45 the [/1] is the command / for irc 08:11:03 and its not listed in logs since its only pinging the server from your client 08:11:16 so nah it only shows // 08:11:44 [/]// makes the first one invisible to logs 08:12:36 Just to make sure, I mean the http://esolangs.org/wiki//// sense of /// 08:12:58 wellllll 08:13:07 WELL 08:13:11 let me check 08:13:16 Sorry, I didn't think. 08:14:29 Ah 08:14:32 I see 08:14:43 the s/word/replace/ 08:14:45 thing 08:14:52 Ye don't most bots have that 08:15:01 Except everybot here 08:15:15 Yeah. Your "probably not" probably still stands. 08:15:46 Well if we had something to test it 08:16:01 foobar 08:16:07 Doesn't fungot have one? 08:16:07 Lilax: obviously there's the business of writing out let but in a different language to allow it to render outside the box? totally awesome. 08:16:13 I don't think [/1] is a IRC command? 08:16:22 zzo38: 08:16:33 [/1] = / 08:16:39 first slash 08:16:44 then // 08:16:48 to make /// 08:16:56 which results in // 08:17:19 Pardon my Spam 08:17:44 Once I was on some IRC that did automatically correct people who typed s/whatever/whatever/ but it annoyed both me and whoever I was trying to communicate with; he had control over that bot so he disabled it. 08:18:46 -!- AndoDaan_ has joined. 08:19:32 s/example/example/r/ on the bot I used replaced everything and would litterally spam all replaced words for that month 08:19:43 I just thought maybe it could allow a user to make a quick ninja edit to mispelled sent chat line. 08:19:52 I used litterally and probably spekt it wrong 08:20:37 you mean a replacer bot? 08:20:44 I don't like the automatic fix (especially raw logs shouldn't automatic fix) it can cause problems if used improperly and that stuff 08:20:52 AndoDaan: no, the bots do not. 08:20:59 -!- AndoDaan has quit (Ping timeout: 264 seconds). 08:21:21 s/stuff/thing bot: Lilax meant to say Things and stuff 08:21:25 like that? 08:21:28 Alright. I don't know why these type of questions pop in my mind. 08:21:36 because magic 08:21:40 more liqe 08:21:43 oops 08:21:45 Lilax: we had such a bot here a while ago. 08:21:56 s/liqe/like 08:22:02 Ye 08:22:06 it wasn't related to the logs, though. 08:22:13 Is that what you meant? 08:22:33 Or do you want a bot that corrects misspellings in the logs 08:22:43 that would be improbable 08:22:56 Not want, but thought maybe. 08:23:01 atleast almost impossible 08:23:37 only the logging bot itself could do that. also we have (at least) two of those. 08:24:13 Since if I were to say sinfe the bot would have to figure out how many times I want that sinfe to be since then what would it do if I said sinfe two times but wanted it to correct it once 08:24:14 You could also download it and then use a local program to perform the corrections 08:24:21 or dat 08:24:29 but that takes tiiiime 08:24:59 and it would in any case not change the fact that everyone actually present will have _already seen the original_ 08:25:16 ye 08:25:26 Good work scoob! 08:26:12 It might be cool to write something at would tried to interpret any and all possible esoteric code snippits found in the chat. 08:27:25 Okay, thanks. 08:29:34 ye 08:32:06 D: 08:33:09 -!- AndoDaan_ has changed nick to AndoDaan. 08:34:29 I want to write IRC bot using C and SQLite. I don't have time right now but may do with in a week. Do you like this, or do you hate this, or both or neither? What is your opinion what you would want such thing doing? What question/complaint please? 08:35:41 * Lilax pats zzo38 08:35:46 Do what you want 08:36:02 Its your choice we Support ye 08:36:08 Without money 08:36:12 we are poor 08:36:16 Yes I know, but I also just wanted to know if other people have opinion about it. 08:36:45 gnight 08:36:48 The money is not required; I do it for free. 08:36:56 -!- Lilax has quit. 08:39:01 If the database contains no triggers then it would just send the NICK/USER/PASS/JOIN/MODE commands at the start and then make a log of everything it receives in the database (with one record per message received), but if you add triggers then you can make it to do different kind of stuff instead of or in addition to logging, such as to make up a chess game. 08:39:33 Does this seems like properly to you? 08:42:15 Anything other than logging would have to be implemented by you. The user couldn't do anything then set and send triggers? 08:46:43 -!- _AndoDaan has joined. 08:46:51 -!- AndoDaan has quit (Quit: bbl). 08:47:07 Triggers would be included in the database schema; that is how SQL works. 08:47:28 -!- _AndoDaan has changed nick to Andodaan. 08:53:39 Any risk of memory leakage, or unsanitary data with using C SQLite? 09:01:38 -!- Andodaan has quit (Ping timeout: 245 seconds). 09:22:48 -!- shikhin has joined. 09:24:25 zzo38: probably like, but I probably want to make an irc but myself eventually, to rewrite my ugly bots to a sane structure. 09:25:23 zzo38: i have a mostly sane proof of concept implementation that handles all that server extension stuff that lets me follow who is logged in as what nickserv account, 09:25:43 which I'll eventually need when I rewrite the cbstream bot (I use the codename "cbvapor" for the rewrite) 09:26:39 but I want to rewrite that initial implementation because it sucks, especially the way it's hardwired to work only with SASL login. 09:27:06 plus I want to make the framework work (with reduced functionality) on irc networks other than freenode. 09:35:47 -!- jameseb- has changed nick to jameseb. 09:45:39 -!- Tritonio has quit (Ping timeout: 276 seconds). 10:15:29 screw riscwars 10:15:33 long live brainwars 10:15:40 wot 10:15:52 http://codepad.org/i8YO5Bpq 10:16:20 oh I forget < and > 10:16:42 does ARM still count as a RISC cpu by the way? I knew it started as RISC, but since then they've added all kinds of nonsense extra instructions and modes that I'm not sure it still is one 10:17:08 mind you, it still seems to have fewer hindering historical cruft on it than x86 10:17:32 That's because it's accumulating the historical cruft right now 10:17:39 exactly 10:17:44 Be patient and give it another ten years 10:18:00 x86 had more of a head start (it started with z80 and whatever that other cpu is) 10:18:17 ARM certainly has at least _some_ historical cruft too 10:18:42 some of it quite similar to x86 actually: a flags register that has both arithmetic result status flags and processor mode control flags 10:19:17 why does that always have to be the same register? 10:20:16 Why would you use more than one register for that 10:20:23 (maybe so that it can be saved on the stack together in interrupt routines, but still, there's probably less need for that in ARM) 10:20:50 b_jonas: would you rather they kept the flags in the program counter like they used to in older versions of ARM? 10:20:56 Jafet: because it leads to nonsense like that writing (restoring) the flags register is a privilaged instruction even though you want to do it just to restore arithmetic flags 10:21:38 jameseb: I don't know really. 10:21:52 -!- oerjan has quit (Quit: leaving). 10:22:17 /0>\ is obviously probably the simplest program 10:22:22 I can see your point though, having access to arithmetic flags would make sense 10:22:30 Jafet: and you can't just write a new value to the status register usually, but only by modifying an existing value, because it has flags you mustn't modify 10:22:34 The ISA lets you set the arithmetic flags directly as well. 10:23:12 I'm not giving an opinion now on whether there should be an arithmetic status flags register in first place, I'm just saying, if there is one, it should be separate from control flags 10:23:47 just look at the MOS 6502 which does it right: 10:24:18 there's an interrupt disable flag, but it's not in the status flags register 10:25:32 It doesn't really matter since it's not a real register anyway, it's just a bunch of bits that happen to be listed in one register for convenience 10:25:33 (besides /\ which just idles around) 10:25:46 and given how on ARM, fewer instructions modify the flags register, and it has shadow registers for modes like the z80, it wouldn't even be essential to save it automatically when entering an interrupt 10:26:14 unlike in the x86 where most instructions mess up the flags 10:26:28 //>\0\ should scan until it reaches a non zero cell, set it to zero and continue 10:36:10 .oO( Ubuntu 2025 "Faulty Fly" for armeb64, now with Jazelle-accelerated systemd! ) 10:42:26 mroman: so wait, what's this? 10:42:50 Er, s/Faulty Fly/Perfidious Pinworm/ 10:43:05 mroman_: is this something like bfjoust but with a way more powerful instr set? 10:50:16 -!- chaosagent has quit (Ping timeout: 265 seconds). 10:51:33 b_jonas: yep 10:51:37 but there's no flag 10:51:46 you have to trick your opponent into executing NUL 10:51:51 (a null byte) 10:52:58 and programs are actually on the tape 10:53:18 so you can write self-replicating programs even 10:57:28 but I'm thinking of adding an extra cost to loops 10:57:35 i.e. if the jump is taken it costs you an extra cycle 10:57:40 mroman_: um, is there a more complete description? 10:57:47 b_jonas: Not yet, but soon :) 10:58:10 It's still a WIP :) 10:58:19 why aren't there proper goto/comefrom instructions, whether with immediate label or computed, rather than these stupid loops only? 10:59:02 because it's brainfuck like? 10:59:10 or supposed to be at least 10:59:50 suggestions are welcome If you have any 11:18:57 -!- boily has joined. 11:42:22 -!- Tritonio has joined. 11:57:33 b_jonas: technically because these just use the stack you can do computed jumps too 11:58:26 but it might mess up forward jumping :) 12:00:17 @word scrozzle 12:00:29 what was it again... 12:00:35 @massages-loud 12:00:35 oerjan said 5h 1m 11s ago: -!- boily has quit (Quit: TRIGOTILLECTOMIC CHICKEN). <-- that's not a word, now you're just cheating tdnh 12:01:18 @tell oerjan of couse I'm cheating! never let orthograph get in the way of a good sounding word, I say. 12:01:19 Consider it noted. 12:01:34 @wp scrozzle 12:01:36 No match for "scrozzle". 12:01:40 @wn scozzle 12:01:41 No match for "scozzle". 12:01:43 ... 12:02:14 http://www.catb.org/jargon/html/S/scrozzle.html 12:10:57 -!- shikhin_ has joined. 12:13:08 -!- ^v^v has quit (Read error: Connection reset by peer). 12:13:32 -!- ^v^v has joined. 12:14:02 -!- shikhin has quit (Ping timeout: 245 seconds). 12:16:36 -!- boily has quit (Quit: LEGITIMATE CHICKEN). 12:31:48 -!- ais523 has joined. 12:32:16 -!- j-bot has joined. 12:46:12 is there no fucking image viewer for linux that lets you display multiple images in the same window 12:46:15 I mean 12:46:23 Debian probably has 5 image viewers pre-installed and none seem to be able to do that 12:49:27 http://i.imgur.com/CSYstxV.jpg 12:49:32 I got inspired to play with image filters 12:49:47 but there no good image viewer has been found so far to make side-by-side comparisons 12:50:04 -!- kcm1700_ has joined. 12:50:47 -!- J_Arcane_ has joined. 12:51:17 -!- EgoBot has quit (Ping timeout: 252 seconds). 12:51:26 -!- kcm1700 has quit (Read error: Connection reset by peer). 12:52:08 -!- EgoBot has joined. 12:54:29 -!- J_Arcane has quit (Ping timeout: 252 seconds). 12:54:42 -!- J_Arcane_ has changed nick to J_Arcane. 12:55:14 -!- Patashu has quit (Ping timeout: 245 seconds). 13:04:19 although this filter here takes hours for 640x480pix images 13:14:47 jesus christ it's so slooow 13:16:44 mroman_: It's probably because you're supposed to do it the Unix way, and make up some sort of pipeline calling ImageMagick to put your images side-by-side. 13:17:01 -!- Tritonio_ has joined. 13:18:51 -!- Tritonio has quit (Ping timeout: 276 seconds). 13:23:17 -!- Tritonio_ has changed nick to Tritonio. 13:38:13 I should use threads 13:38:21 otherwise it's even taking too long for 256x128 13:47:57 Just create a HTML file with the layout you want. 13:49:16 so 13:49:20 now my filter runs on 8 cores 13:49:21 cool 13:49:43 Although, debian has at least five web browsers and probably none of them have proper colour handling (then again, do any of the image viewers?) 13:52:51 yeah it ran through 13:56:49 -!- vanila has joined. 13:56:52 is lambdabot ok 13:57:52 vanila: lambdabot works over (or did earlier) 13:58:04 s/over/over here/ 13:58:11 > 1 + 1 13:58:13 2 13:58:28 how strange 13:58:31 must be just #haskell that has lambdabot broken 13:58:32 it does not respond to #haskell 13:58:40 don't know why 14:00:04 Math.pow(x,2); is apparentely faster than Math.pow(x,1) 14:01:26 what 14:03:47 -!- shikhin_ has changed nick to shikhin. 14:05:57 http://i.imgur.com/BBN78m8.png <- anyway. After minutes that what my filter did to the image 14:06:03 (left is original) 14:06:44 so, it blurs 14:08:57 yep 14:09:30 but it's somewhere in O(N^4) 14:12:23 O(x^2*y^2) to be exact 14:12:38 > 640*640*480*480 14:12:40 94371840000 14:15:59 shouldn't it be O(wh) 14:16:20 no 14:16:40 for regular blur maybe, yes 14:25:38 but yes, Math.pow(x,2); is faster than Math.pow(x,n); for n = 1..10\2; 14:25:54 i.e. for Math.pow(x,n) for n = 1..10; n = 2; is the fastest one 14:26:34 -!- GeekDude has joined. 14:26:36 and by faster I mean *really faster* 14:26:44 like a half hour faster 14:28:07 lol 14:28:18 mroman_: what does the filter look like? I believe the standard approach for this is to do 2D FFTs, and then it should become O(log(wh)wh) 14:28:33 that is, if you can't decompose things into box filters 14:28:44 (which are O(wh) each) 14:29:19 but that assumes linear filters which are shift-invariant. maybe yours isn't 14:33:31 -!- skj3gg has joined. 14:35:13 vanila: http://codepad.org/uGxDOc8L 14:35:39 that's how much faster Math.pow(x,2) really is 14:36:08 i guess it's using base 2 14:39:27 well, using a for loop to pow is actually way faster :) 14:39:41 if n is an integer using a for loop is way faster than calling Math.pow 14:39:56 is that... java? 14:40:21 -!- perrier has joined. 14:40:42 yeah 14:40:54 It looks pretty clear there's a special-case for x^2. 14:43:02 http://codepad.org/5MZBlUjm 14:43:04 Math.pow vs for loop 14:43:08 -!- ais523 has quit (Read error: Connection reset by peer). 14:43:15 -!- ais523 has joined. 14:43:53 well, n=4 and n=8 also take 10s 14:43:57 for pow 14:44:17 so looks like a special case for x^2 14:44:51 perhaps there's a java benchmark that does stuff like sqrt(pow(x,2)+pow(y,2)) 14:45:22 Aw, "public static native double pow(double a, double b);" 14:45:27 could you try comparing pow(pow(x,2),2) against pow(x,4) ? 14:45:39 pleas mroman_ 14:46:02 Java_java_lang_StrictMath_pow(JNIEnv *env, jclass unused, jdouble d1, jdouble d2) { return (jdouble) jpow((double)d1, (double)d2); } /* hmm */ 14:46:29 Can't tell where "jpow" comes from, at least immediately. 14:46:49 try to make it call a real optimized hypot 14:48:43 Oh, there it is. 14:49:28 Calls __ieee754_pow. 14:50:20 http://codepad.org/6u1ZCdYg 14:50:28 Don't see the ^2 special case there, but it's quite possible I missed something while skipping through. 14:50:29 ^- this is the fastest way to do pow so far I have found 14:50:32 it even beats the for loop 14:50:40 mroman_, I have a rqeuest for compare 14:50:44 if you can benchmark it 14:50:57 btw I know a slightly faster algorithm for power 14:51:02 you can use the binary expansion 14:51:20 Math.pow is probably not optimized for int 14:51:25 -!- ais523 has quit. 14:51:54 Oh, right, there it is. 14:52:08 /* special value of y */ if(hy==0x40000000) return x*x; /* y is 2 */ 14:52:11 x^1011 = ((x^2)^2*x)^2*x 14:52:18 vanila: And what would that request be? 14:52:22 could you try comparing pow(pow(x,2),2) against pow(x,4) ? 14:52:36 rather than just pow(x,2) against pow(x,4) 14:52:51 so the result numbers will be equal 14:53:03 mroman_: There's also a special case for x^0.5 to do sqrt(x). And there *should* also be a special case for 1.0, I don't know why you didn't observe that. 14:53:46 /* y is +-1 */ 14:54:13 Yes. 14:54:46 horrible, horrible code. 14:55:21 It seems to be inherited from Netlib. 14:55:30 http://www.netlib.org/fdlibm/ 14:55:59 Or maybe it's the other way around, I don't know. That still has Sun's copyright notices in there. 14:58:02 vanila: I can 14:59:42 Math.pow(x,4) := 9904ms 15:00:08 -!- skj3gg has quit (Quit: ZZZzzz…). 15:00:22 Math.pow(Math.pow(x,2),2) := 33ms 15:00:36 -!- skj3gg has joined. 15:01:36 oh my... http://cgit.haiku-os.org/haiku/plain/src/system/libroot/posix/glibc/arch/x86/e_pow.S 15:02:03 (hmm, why did I find a haiku link first for a glibc file, but who cares) 15:02:34 what the :EEEEE 15:02:37 thats really weird 15:02:51 i thought it migth just be sfaster because of smaller numbers 15:03:47 using myPow it's 7ms 15:04:14 but yeah, generally instead of writing Math.pow(x,3); you might as well just write x*x*x instead 15:04:21 much, much faster :p 15:06:20 -!- skj3gg has quit (Quit: ZZZzzz…). 15:09:10 -!- `^_^v has joined. 15:11:42 -!- G33kDude has joined. 15:14:27 -!- GeekDude has quit (Ping timeout: 264 seconds). 15:14:34 -!- G33kDude has changed nick to GeekDude. 15:15:08 -!- Sgeo_ has joined. 15:17:29 -!- Sgeo has quit (Ping timeout: 244 seconds). 15:22:48 yeah well... 15:22:55 either use some fancy algo or at least for loops :) 15:23:01 that's what I've learnt today from Math.pow 15:26:22 -!- GeekDude has quit (Quit: {{{}}{{{}}{{}}}{{}}} (www.adiirc.com)). 15:37:41 -!- ais523 has joined. 15:46:54 http://eighty-twenty.org/2015/01/25/monads-in-dynamically-typed-languages/ 15:47:37 how can you use such a pure concept in such an impure environment. 15:48:00 -!- bb010g has quit. 15:49:08