[00:59] MSG: Quit: Lost terminal [02:22] MSG: Ping timeout: 255 seconds [10:04] Join: fiveop joined #corewars [12:01] Join: Core_old joined #corewars [13:06] Join: Fluffy joined #corewars [13:06] :) [13:34] MSG: Quit: Lost terminal [14:08] Join: niaatzeon joined #corewars [14:12] Part: niaatzeon left #corewars [14:58] MSG: Quit: Leaving. [14:58] Join: Fluffy joined #corewars [14:58] grmpf [15:15] frumple [15:15] Yes, I agree. [16:09] Mizcu: It seems as if you've infected me with your voodoo bugs :( [16:24] nyahnyah [17:42] Yippie! I've found the bug. [17:42] all in all I only have to change three lines to make it go away :) [17:57] Join: fff joined #corewars [17:57] Part: fff left #corewars [17:58] Join: fff joined #corewars [17:58] Part: fff left #corewars [17:58] Join: fff joined #corewars [17:58] mm.. [17:58] cant decide if to join or not? [17:59] Hi fff [17:59] Hi. [17:59] ye [17:59] yea* [17:59] so, how may we be of assitance? [18:00] You may not. [18:00] I joined for no reason. [18:00] hmm [18:31] It is time to generate a new nano warrior and abuse bvowk's computer a little bit more :) [18:51] Join: lookout|R joined #corewars [18:51] hey [18:51] Nick Change: lookout|R changed nick to BradGarre [18:51] Hi lookout|R [18:51] hey [18:51] Nick Change: BradGarre changed nick to Brad [18:51] that works [18:51] hmm [18:51] Are you new to corewar? [18:51] depends on what you mean [18:51] i played this game a while ago [18:51] :) [18:52] then quit because i couldnt find a decent windows pmars [18:52] then i found corwin [18:52] a few secs ago [18:52] I'd define that as not "new" [18:52] so now im back [18:52] and im wondering if you could point me towards some good warriors to test my new ones against [18:52] no problem [18:52] one moment [18:53] I guess you are writing warrior for '94nop at the moment, right? [18:53] There some kind of warrior archive at http://www.ociw.edu/~birk/COREWAR/koenigstuhl.html [18:54] At http://www.corewar.info/optimax/dload/ you can find the fsh-94nop benchmark (version 0.3), which contains a very good benchmark [18:54] for '94nop [18:55] its warriors are sorted according to strategies [18:55] kk [18:55] thankyou [18:55] :) [18:55] i may ask you a few questions as long as your here if that's ok [18:55] That's, what this channel is for :) [18:56] is there a way to get corewin to keep going even after a battle is won? [18:56] No idea, because I don't use Corewin [18:56] * Fluffy kicks Mizcu [18:56] kk [18:56] Mizcu: do you know? [18:56] what do you use? [18:57] my own MARS :) [18:57] cool [18:57] PyCorewar, which is more or less a Python wrapper for a fast MARS [18:57] but it isn't useful for debugging, but benchmarking and evolving works quite well [18:58] In any case you should consider using a faster MARS for benchmarking your warriors in the long run [18:58] bah [18:58] evolving [18:58] who needs it [18:58] Me? [18:58] hehe [18:58] bvowk? [18:59] i wont hold that against you [18:59] no problem [18:59] i will hold it against bvowk though... [18:59] because he overuses it [18:59] bvowk won't care [18:59] i dont even think hes a real person... [18:59] There's no way to prove that [19:00] ... at least here in the channel [19:00] you might no exist, too [19:00] hehe [19:00] You probably already know, that there's a beginner's hill at SAL, where you can test your warriors, too [19:01] http://sal.math.ualberta.ca/hill.php?key=94b [19:04] Is there any reason, why you despise evolving so much? [19:04] i think that its cheap [19:04] i dont think it shows real skill [19:04] Have you ever tried it? [19:05] no [19:05] Then how do you know, that writing a good evolver isn't a hard task? [19:06] oh, i know writing an evolver is hard [19:06] but what about people who use others evolvers? [19:06] don't evolver programs essentially optimize step sizes and stuff? [19:06] (i mean, they don't actually change the program's code) [19:06] They still have the problem to seed the evolver properly [19:07] not that i'd know, i only came in here to watch for a little while :) [19:07] malady: no, you probably mean optimizers, which try to find optimal "steps" against a given benchmark [19:07] i just mean numbers [19:07] hey jens.. [19:07] thanks for the email.. [19:07] malady: Evolver chan change the whole warriors ... even instructions [19:07] interesting [19:08] bvowk: No problem. I really wanted to fix that stupid ref-count bug [19:08] i wonder how that even works to be effective [19:08] bvowk: Does it work now? [19:08] I was just popping up to check my email before I left for a couple days.. [19:08] you're totally made my day ;) [19:08] i was thinking it sounded like it'd only "evolve" the constants that affect how the program behaves or something [19:08] * Fluffy chops bvowk in small bits [19:08] bvowk: I didn't intend to do that ;-) [19:09] heh [19:09] bvowk: In any case I wish you good luck with you evolver. In the mean time I'll try to rule the nano hill, ok? [19:09] malady: no, evolving attempts to make viable warriors from nothing [19:09] craziness ;p [19:09] fluffy: well, I've got the one on 3jane running 4 cores to the wind.. [19:09] malady: But it works [19:10] well it must i suppose, otherwise people wouldn't be writing programs to do it [19:10] fluffy: maybe I'll get lucky and get something cool before your crap code runs it out of memory ;) [19:10] bvowk: Try to beat my generated warriors :) [19:10] what kind of warrior should i try to make? [19:11] Brad: For '94nop? Whatever you like most. that's probably the best strategy :) [19:11] im gonna try a replicator [19:11] ive never made one befor [19:11] Brad: Good luck [19:11] damn [19:12] ive forgotten most of the commands [19:12] whats a good command refference? [19:12] is there perhaps a site or article you could point me at that talks about mutating random code to be effective? [19:12] Brad: http://www.corewar.info/lexicon/paper.htm might help [19:12] malady: http://corewar.co.uk/links.htm [19:13] Brad: Did you mean sth. like http://vyznev.net/corewar/guide.html ? [19:13] corewin question: no [19:13] alright.. [19:13] cya in a few days [19:14] bb bvowk [19:14] * Fluffy waves to bvowk [19:14] thanks fluffy [19:14] * Fluffy hugs bvowk to put a "kick me" on his back [19:14] Brad: http://www.koth.org/info/pmars-redcode-94.txt [19:15] what does { } do? [19:15] I am playing Civ4, so apologies for the sleeping moment [19:15] Mizcu: Evolver ... you want to write one ... evolver ... [19:15] same as <>, but with a-field instead of b-field [19:16] kk [19:16] Fluffy: keep your shoes on, it wont be running yet for a long time [19:16] I had to try [19:17] hmm [19:17] if a paper overwrites another program wouldnt that just turn that program into a paper as well? [19:17] no, it would need the right amount of processes [19:18] ... which execute at the right time [19:18] oh [19:18] which is not always given [19:18] this is gonna be tougher then i though [19:18] that's what makes it so interesting [19:19] what does .a do again? [19:20] It will turn it into paper, yes. Nearly all new papers have some sort of method for noticing abnormal processes and either kill them or suicide [19:20] .a means that the instruction will modify a-fields [19:20] so what does that mean for spl.a? [19:20] is is the same as every spl [19:20] nothing, spl ignores modifiers [19:20] hmm [19:20] thats funny [19:21] what is the equivelant of @ except for the a feild? [19:21] * [19:22] http://vyznev.net/corewar/guide.html might be a good idea :) [19:22] but you can use us for "quick lookups" nonetheless [19:27] damn, id need another screen for myself [19:28] but got no space at the table [19:28] Mizcu: How about buying one? [19:29] i already got a screen [19:29] but no space to put it on [19:31] There is always a wall, isn't it? [19:31] ... which should be enough for a flatscreen [19:34] You should see my workspace.. [19:34] Actually, i could take a picture [19:35] Then take it :) [19:35] malady: http://corewar.co.uk/thorsell/paper.htm and http://corewar.co.uk/evolving.txt should be interesting, too [19:35] when you spl which process goes first? [19:35] the one that split or the original? [19:36] original [19:36] kk [19:39] i was looking for what exactly was randomly chosen/mutated and how, but it looks like it's just .. really random :P [19:39] i thought maybe there was some method other than brute force for guiding it to at least producing valid programs [19:39] or increasing the liklihood by a significant amount, i guess i should say [19:40] that might not be possible, because changing a tiny bit of a warrior might result in big changes of its score against a benchmark [19:40] well, that's why i was curious :) [19:41] at http://corewars.jgutzeit.de/score_surfaces/index.en.html you can find some infos about that "roughness" [19:42] actually it is about optimizing a warrior, but the images show, that the score surfaces a too rough ... at least the chosen settings [19:42] *a -> are [19:43] http://koti.mbnet.fi/mizcu/13a.jpg (crap quality, but i dont have a real camera, dont whine) [19:44] * Fluffy starts to whine [19:45] There's enough space for another monitor above the LCD screen. Simply throw all the stuff away. It seems to have no use ;-) [19:46] the level above lcd cannot take a CRT [19:46] Then take a LCD. They aren't that expensive any more [19:47] ... or use an old laptop as your second screen [19:48] things to notice: Carmageddon cdcase, can of Jolt, Christmastree, [19:48] ... and that drink too much [19:48] ... and it seems to be some strange stuff [19:49] like what [19:49] all those tin cans above you LCD [19:49] *your [19:50] and on the left side of it [19:50] spl 1, 0 [19:50] spl 10, 0 [19:50] mov -2, 9 [19:50] jmp >-1, }-1 [19:50] ok [19:50] i understand why this dosnt work [19:51] but i wanna know how it dies when there are no other warriors in the core [19:51] from left to right: Can of Croc Tail, Dr. Pepper, Cherry Cola, Battery, Dutch Candycan, Shark energydrink, Mad-Croc ED [19:51] the leftmost white thing is a Chinese statue [19:51] Mizcu: See ... that's what I call strange stuff [19:51] Brad: one moment [19:52] because i thought that with the jmp it should at least loop until it overwrites itself with dats [19:53] jmn.f -1, }-1 will work [19:53] what does .f do? [19:53] jmn.f checks both a and b-field to see if either is non-zore [19:54] so it loops as long as the a-field of mov points to something non-zero [19:54] ok [19:54] Brad: Your jump seems to jump to the wrong location, probably because the > increments AFTER the instruction is executed [19:55] Fluff: you want weird stuff? That reddish can left from the LCD is smellpaste for catching fish [19:55] I have forgotten to take it to my fishing equipment for the last 5 years [19:55] But you don't eat that, do you? [19:55] of which 4.5 it has been hidden in closet [19:55] it smells horrible [19:55] and i wouldnt eat that [19:56] so it doesn't count [19:57] MSG: [19:58] Mizcu: You room seems to be a little bit small [19:58] no shit? [19:59] Probably because you don't have that much space in Finland ;-) [19:59] this corner takes 1/4th [20:00] send 4th goes to bed, 3rd to my bookshelves and closets [20:00] don't tell me, that you sleep in that room, too [20:00] send -> second [20:01] ... and Mizcu needs the fourth part ;-) [20:01] no, my sanity does [20:02] :) [20:04] Do you have a photo of yourself? [20:04] no [20:05] well, http://koti.mbnet.fi/mizcu/kuvat/Koira11.jpg [20:05] thats 2.5 years old [20:05] does "kuvat" mean anything? [20:06] pictures [20:06] hehe [20:13] hmm 70.462 meters with that discus flash app [20:14] 6 tables out of.. 16? done [20:14] yawn [20:14] what tables? [20:14] for the evolver [20:15] hehe, so it finally got you [20:15] i am not keeping hurry with it [20:16] I'm considering to switch to tiny with my generator [20:39] good evening! [20:40] h ares [20:40] Hi ares [20:41] i finished the first draft version of my redcode turotial (but only german for now) [20:41] URL? [20:41] http://harald.ist.org/?site=ares [20:43] i maybe will restructure it a bit, and also may add some content... next i'm up to is write a real ares manual [20:43] looks nice [20:50] damn generator [20:51] It is one these days. Today it probably decided, that it is a good day to use on tiny score increments. [20:51] even better than decrements, eh? [20:51] So far I've tested more 120000 different values and every now and then I get an improvement of 0.01 or so [20:51] decrements aren't possible [20:52] they allways say impossible... got no visions anymore? [20:52] Who's programmed that generator? [20:52] If a modified warrior scores worse than the current one, it is simply discarded and another change is made [20:52] So I only seem improvements [20:53] *see [20:53] (i knew that - see i like beeing kicked) [20:53] hows you cold? [20:54] I'm still alife ;-) [20:56] but fortunately the worst is over [20:56] Now I can walk around and start to infect others :) [20:56] ah, lovely [20:57] "loading up" new bacteria? [20:57] soon [20:57] current score is 155.328007 [20:57] but still at stage2 [20:57] (out of 3) [20:59] and you, any new warriors? [20:59] nope... i want to get the work behind me first [20:59] work? [20:59] ares update, manual [21:00] I wouldn't call that work [21:00] *g* [21:01] i fear, i cant stop extending ares, until it can write warriors by itself [21:01] that wouldn't be any good. You have to extend it until it starts to kill people ;-) [21:19] 35th on nano :) [21:22] * Fluffy congrats John for tlotd :) [21:22] John: Consider this to be my christmas present ;-) [21:42] * Fluffy waves [21:42] MSG: Quit: fluffy.i < 1, # 42 [23:07] MSG: Quit: Trillian (http://www.ceruleanstudios.com [23:35] Join: Brad joined #corewars [23:35] hey [23:35] im back [23:36] hey brad [23:39] aye