[00:03] Join: ares_ joined #corewars [00:04] MSG: Ping timeout: 255 seconds [00:09] Join: Waknak joined #corewars [00:14] Nick Change: Waknak changed nick to waknuk [00:15] Did Nenad pass through? [00:16] you mean the tournament? i think not: [00:16] well, I guess I'll go, as well :) I'll post the final results tomorrow night, and also the rules for round 3 [00:16] (nenads message) [00:17] * waknuk thanks ares_ [00:18] is it my trillian, why i often duplicate in here? [00:18] or is it just my multiple personae [00:18] seems to come with "automatic reconnect" [00:18] Not sure ... the client I'm using now has some peculiarities too [00:20] Perhaps this is a high traffic site (when I'm not around) [00:57] MSG: [03:53] MSG: Quit: CGI:IRC 0.5.7 (2005/06/19) [13:50] Join: Nenad joined #corewars [13:52] hi [13:53] hi [13:53] * Nenad is dealing with the "time travel" phenomenon [13:53] successfully? [13:53] yes [13:54] :) [13:54] i do so, too... every second i move one second forward in time [13:54] John had sent me his entry for round 2 9pm yesterday, but I only received it half an hour ago [13:54] Join: inv joined #corewars [13:54] which means that he sent it before the deadline [13:54] ah, some time thieves around [13:54] sigh... [13:54] hi inv :) [13:55] uhoh [13:55] Nick Change: inv changed nick to inversed [13:55] hi inv, 2 [13:55] okay. hi vni [13:56] inversed: I'm forced to rerun round2 once again... the final results will be finished within half an hour and I'll post them to rgc [13:59] Can you tell me round 3 rules now? [14:01] oh. well, haven't posted them yet, will do it once I post the r2 results. but I could mail them to you :) you'll have a 15min advantage to other players :)) [14:01] the deadline will be Jan 5th [14:01] a little bit more time.. [14:04] Mail them [14:04] I did [14:06] so? what do you think? [14:06] I will think about r3 while getting home from this idiotic institute [14:07] But I like the rules [14:07] I already have sucessfull double scanner, factorized convolution [14:08] I made only one good - float_double [14:09] But rules are somewhat subjective [14:09] in what way? [14:09] There is no objective way to tell if warrior type is CDS, SCN or CLR [14:11] mwahahahaha (playing MTG at the background.. , just won a bad battle with a nice card for the situation) [14:12] inversed: that's true [14:12] but generally, all scanners are allowed - so... [14:12] stone/scanners as well [14:12] so there is not much room for misinterpretation [14:13] Miz: which card? :) [14:13] So how will you decide if you recive something weird, like scan->paper? [14:14] if someone wants to waste time scanning and then launch a paper, he is free to do so [14:14] but I know what you're trying to say [14:14] Nenad: Drain power to Fireball [14:14] the warrior has to be a scanner... now the question is: how do you define a scanner? [14:14] or coreclear... those are allowed as well [14:15] Miz: :) [14:15] Miz: I'm reruning r2 due to John's time travel experiment [14:15] What about scanning paper? [14:15] well, has a working one which scores well ever been made? [14:16] LP scanning paper, "Paper With Eyes" [14:16] Scores quite good [14:16] don't forget -> you will be fighting mostly other double scanners [14:16] so a paper which loses time for scanning... [14:16] probably isn't such a good idea [14:16] So they're permitted? [14:16] yes [14:18] Got2go, bye [14:18] MSG: Quit: . [14:18] * Nenad waves [14:32] MSG: [15:41] Join: Nenad joined #corewars [15:46] nyah nyah nyah [15:46] computer opponent (highest difficulty) screwed up [15:46] results posted to rgc [15:46] round 3 rules posted to rgc [15:46] :) [16:05] Join: fiveop joined #corewars [16:06] hi fiveop [16:06] Join: Fluffy joined #corewars [16:06] :) [16:06] hi fluff [16:06] r2 res and r3 rules are at rgc [16:06] * Fluffy kicks Nenad for forgetting the 'y' [16:06] :) [16:07] * Fluffy runs away a cries for a while ... [16:07] well, there is always the next round [16:08] :) [16:08] ... and I already has one or two points ;-) [16:08] :) [16:09] inversed leads with 14, followed by Chip and Miz with 13 pts [16:09] John has 9 and Phil 8 [16:12] Nenad: Did you say that I won't be able to make a paper? [16:12] *Didn't [16:13] and who decides what a scanner is? [16:15] hmm, i have a bad feeling about knowing exactly what kind the winner will be =/ [16:15] no, that doesnt mean knowing how it is coded [16:15] well, it's rather simple really [16:15] it's a scanner if I say it is :)) [16:16] there is no objective way to distinguish warrior types, anyway [16:17] fluffy: inversed already inquired if it were possible to make a scanning paper. I said it was allowed. :) So you can still make a paper... if you want to... [16:17] but I wouldn't recommend it [16:18] uhm.. [16:18] And do I get that right ... my warrior is copies to 3643, then the warrior is started at wstart and then the copy at 3643, right? [16:18] *copied [16:19] if a scanning paper is allowed then it would be very good candidate for a winner.. [16:19] fluffy: yes [16:19] Miz: I really have to disagree... other scanners would annihilate it [16:20] it's not fighting only the mentioned bench warriors [16:20] all scanners fight each other as well [16:20] Nenad: I my final score is the averaged score of against all of my opponents (include Hulla, ...) [16:20] but you can always try proving me wrong [16:20] *including [16:20] yes [16:20] hmm [16:21] the goal is to find a good balance [16:21] and I can kill the original warrior at wstart after 300 cycles, if I want [16:22] if you want to, yes [16:22] hmm [16:23] if you feel like winning the lottery 20 times in a row.. [16:23] that's completely up to you... it would help against scanners and be disadvantegous against others... but since 300 cycles have to pass, it wouldn't help much against scanner anyway [16:23] but that's just my view [16:28] ... and I can kill the warrior at wstart as soon as possible if the copy runs with >= 2 processes all the time [16:29] lol [16:29] yes [16:29] of course, it was impossible to make a set rules that can't be abused [16:29] hmm ... now that I understand the rules they might not be as bad as I thought [16:30] so you thought they were bad [16:31] yes, yes, I know ... wait for the next round :) [16:31] * Nenad throws a pineapple at fluffy [16:31] Don't throw it like a girl [16:31] hrm [16:31] lol [16:44] Maybe it is time to learn how to write a scanner [16:44] mayhaps [16:44] MSG: Quit: Lost terminal [16:45] Join: fiveop joined #corewars [16:45] the good idea is worth more than anything else... but a litlle optimization might help as well [16:48] Yes, at the moment I think, that it is quite hard to get points from Hulla, ... [16:48] they will probably have crushed you soon after you've booted [16:49] well, float_double did quite well against pws (it was a double scanner) [16:50] but the wdist wasn't as ugly as this one, so it was easier to synchronize :) [16:50] Fizmo also had a good double scanner [16:50] I forgot the name, though [16:51] did you choose wdist as random != 4000 or did you use something else? [16:52] well, I "randomely" calculated wdist so that it is hard to combine it with frequent stepsizes :)) [16:52] hmm [16:52] random != 4000 wouldn't be good enough to complicate things [16:52] if it were a multiple of 10 or 8, it would be in vain [16:53] I wonder whether you already know the rules for round 4 [16:55] yes [16:55] :)) [16:59] even more convoluted than round 3? [16:59] more complicated by a factor of at least 5 [16:59] imho [17:00] r4 will be the most complicated one, most unconventional, and most interesting [17:03] ... and how many rounds will there be? [17:03] 8 [17:03] if players request a few more, I will oblige, but 8 were planned [17:03] * Nenad has to go [17:04] * Nenad waves [17:04] MSG: [17:20] Join: gareththe joined #corewars [17:21] hello [17:35] hi [17:36] :-) [18:13] re [18:14] Join: Core_old joined #corewars [19:00] Part: Fluffy left #corewars [19:16] Join: Core_old_ joined #corewars [19:19] MSG: Ping timeout: 255 seconds [19:45] any koth.org hill issues in the last few days? [19:54] found a bug in pmars' asm [19:54] a EQU #1, #2 [19:54] EQU JMP [19:54] MOV a, +1 [19:54] result strange [19:55] without the "," [19:55] mov a +1 [19:55] mov a +1 ---> mov #1, #2+1 ? [19:56] oh missed the jmp [19:56] mov #1, #2 \n jmp +1 [19:56] if it really were a "simple replacement" i'd expect this... but i get...: [19:57] JMP.B $ 1, $ 0 [19:57] ADD.AB -1629501199, )221318112 [19:57] ? [19:57] sic! [19:59] why the closing brace in the bVal of add?? [20:00] i am confused... and i'm wondering how my asm should handle multiple EQUs [20:02] ORG START [20:02] START JMP.B $ 1, $ 0 [20:02] MOV.A # 0, # 0 [20:02] I got that? [20:02] pmars v 0.9.2 under win [20:03] pmars v0.9.2 under linux [20:03] hm... [20:03] maybe i should reboot my pc ? *ggg* [20:03] hehe [20:04] did you try with a EQU #1 #2 or equ $1 $2 [20:04] with immediate (#) [20:04] a EQU #1, #2 [20:04] EQU JMP [20:04] MOV a +1 [20:05] so should be mov #1, #2 \n jmp +1 [20:06] I think that doing a move with two immediate operands doesn't make sense [20:06] so it turns them into direct addressing [20:06] it wouldnt dare, i hope [20:06] like when u have an imp: mov.i #666, 1 [20:06] that is the same as mov 0, 1 but it doesn't use the a field (i think) [20:07] right [20:07] but any combination is allowed [20:08] well, i suppose noone would try using multiline equ like that, but... pffff [20:08] Join: Fluffy joined #corewars [20:08] :) [20:08] hi fluffy [20:08] hi -~waving~~ [20:08] eeek! I get some attention [20:09] * Fluffy runs away [20:09] nonono stay [20:09] * ares_ ties fluffy [20:09] i'd like to hear your expertise [20:09] ares_: No! I'm not THAT kind of guy [20:09] lol [20:09] lol [20:10] fy: i have the impression, that you know much about the pmars internals, right? [20:10] Who is fy? [20:11] i meant you [20:11] :) [20:11] and who's that? Don't confuse me with new names [20:12] ares_: yes [20:12] i need information about mutliline EQU handling, (please, please dont tell me to decipher pmars source) [20:12] Don't try to decipher the pmars sources [20:13] even me wouldnt dare that kind a hack [20:13] what information? [20:13] waaah [20:13] it is easy [20:13] parse the multiline equ [20:13] and replace the label in the following text [20:13] i got weird results on: a EQU #1, #2 \n EQU JMP \n MOV a +1 [20:13] I don't get why the jmp was first [20:14] just a test case [20:14] i try to figure out pmars' behaivour [20:14] that does not work [20:15] i get no error, it crashes after printing this: [20:15] ORG START [20:15] START JMP.B $ 1, $ 0 [20:15] ADD.AB -1629501199, )221318112 [20:15] I get ORG START [20:15] START JMP.B $ 1, $ 0 [20:15] MOV.A # 0, # 0 [20:16] which version do you use? [20:16] 0.9.2 on win [20:17] could you please copy&paste the source here [20:17] (did you send me the binary?) [20:17] yes, but I didn't compile it [20:17] it might have been full of other stuff :) [20:17] ;redcode-94b [20:17] ;assert 1 [20:17] ;name [20:17] ;kill [20:17] ;author anonymous [20:17] ;strategy kill_the_opponent [20:17] ;date 2006-Dec-13 [20:17] ;version 1 [20:17] a equ #1, #2 [20:17] equ JMP [20:17] MOV a +1 [20:18] one moment .. [20:20] you say it so easily "parse the multiline EQU and replace the label in the following text"... [20:20] is it meant to be used like i tried? [20:23] yes, it should be a simple replacement and I think it is yet another parser bug of pmars [20:24] btw. i took a look to exhaust 1.9.2, and found no string "EQU" in the source [20:24] me2 [20:24] or FOR [20:24] (EQU might be there) [20:24] and by the way, I get jmp.b $ 1, $0 / mov.a # 0, # 0 as the result [20:24] same as gareth gets [20:24] me2, that makes sense except why is the jmp first? [20:25] yes, that's why I think, it is a parser bug [20:25] no wait it doesn't *g* [20:26] I guess no one uses multiline equs much [20:26] there are some rare case on Koenigstuhl [20:26] but yes, no one uses them [20:26] i will get into troubles, if i would try to implement it like in the example [20:27] you should implement it they it was thought to be done [20:27] especially if i try to do it in 2 passes [20:27] *the way [20:27] (i didnt even see your typing error) [20:27] but I did [20:27] (until you said *the way) [20:27] *ggg* [20:28] and before I forget: watch out for recursion [20:28] sth like: [20:28] a EQU b [20:28] yeah i know [20:28] b EQU a [20:28] nasty [20:29] no, easy to recognize [20:29] in the current ares version i used sth. like a "replacement label path" [20:29] you can for instance build a dependance-tree for every EQU [20:29] when the text-to-be-replaced was found in the path, it was a circular reference [20:29] yup [20:31] argl... the pmars parser is too good... it calculates things, before it even knows, that they need to be calculated, it seems [20:31] Join: John joined #corewars [20:31] Hi :-) [20:31] hi john [20:31] ares_: No, it is more or less a normal parse [20:31] Hi JM [20:31] *r [20:31] i dont know much about parsers [20:31] Hi ares, Fluffi [20:32] hi [20:32] * John ignores the conversation about parsing! [20:32] Hi Gareth [20:32] ares_: http://www.cs.vu.nl/~dick/PTAPG.html [20:32] Are you still trying to get on the nano hill? [20:32] no :-) too tough for a newbie like me [20:33] gareththe: there's any easy (but lengthy) way to get onto nano (without using your brain) [20:33] Some of the newbies are doing well on nano [20:33] i tried to use yacc, but i didnt get far with it [20:33] gareththe: 1) Create a warrior, which is good, but doesn't enter nano [20:34] :-) I wanna use my brain (that's why I haven't got on any hills hehe) [20:34] gareththe: 2) randomly adjust one opcode/modifier/value/whatever [20:34] gareththe: 3) send to the hill and see if the new version is better [20:34] 4) if it is better, keep it and start again with 2) [20:34] hehe [20:34] (otherwise keep the old warrior and go to 2) [20:35] do that until you are onto the hill [20:35] it really works [20:35] It's true, it really works [20:35] i could put an assistant for this in ares :-)) [20:35] no, you couldn't [20:36] mad [20:36] this method is already patented and trademarked [20:36] gareth: we should start a brain hill... [20:36] i am willing to put a (c) sign somewhere *ggg* [20:36] (at the back of the window? *g*) [20:37] ares_: No, I'd need to be paid! [20:37] you get a free copy of ares with asm v2 [20:37] eeek! [20:37] ever tried it? [20:37] (runs with wine) [20:38] I've seen a screenshot ;-) [20:40] John: Got some time? I need a good benchmark for my generator :) [20:40] Are the round 2 warriors published anywhere yet? [20:40] another question: how does one use the = operator (with registers) [20:40] there are 26 internal registers (a ... z) [20:40] example? [20:40] you do sth. like [20:40] Fluffy: what do you need? [20:41] test EQU (a = 42 + 1) [20:41] then test = 42 [20:41] and a = 42 [20:41] and you can later use [20:41] mov.i # a, 1 [20:41] strange, but it works [20:41] you mean a=43, dont you? [20:41] test=43 sorry [20:41] So far I've only seen it to calculate the inverse a value mod 8000 [20:42] Did anyone use anything like "org 840000" = org 0 in coresize 1400, or org 100 in coresize 8399 [20:42] yes, I've forgotten to tell you, that I usually work with very large values of 42 [20:42] no [20:43] I used jmp startsmall*(CORESIZE==1400)+startmedium*(CORESIZE==8399) [20:43] I've used: jmz. b code, # CORESIZE - 8399 [20:43] since I didn't have a mars handy to test whether an "org 840000" would work as expected [20:43] hehe, nice solutions so far [20:44] yes, it works [20:46] In pmars and corewin? [20:46] p [20:55] coffee.sys not found - sysop halted x-| [20:55] good night! [20:55] * John waves [20:55] * Fluffy reboots universe [20:55] core dumped [20:55] *ggg* [20:58] * Fluffy waves [20:58] MSG: Quit: fluffy.i < 1, # 42 [21:15] time for me to go to [21:16] MSG: Quit: mov.i #1,1 [21:30] MSG: Quit: Lost terminal [21:57] MSG: Quit: Trillian (http://www.ceruleanstudios.com [22:59] MSG: Quit: Ex-Chat [23:03] Join: Roy joined #corewars [23:06] * Roy just posted some fresh code on r.g.c! [23:06] It only needs somebody to have a good look at it and making it work :-D [23:09] MSG: Quit: Ik ga weg