[00:37] MSG: Quit: humhum [06:53] Join: Mizcu joined #corewars [07:43] Join: willvarfa joined #corewars [08:02] MSG: [10:32] Join: fiveop joined #corewars [14:27] Join: Mizcu joined #corewars [14:56] * datagram begins a new warrior [15:16] MSG: [16:43] hrm. [16:43] * bvowk pokes wils eye to see if he's still blinking [16:49] Hi bvowk [16:50] I'll be in Canada saturday night : ) [16:50] what part? [16:50] van right? [16:51] Yeah [16:51] And I don't think I'll be able to go to cansec [16:52] Its too close to LayerOne...don't wanna risk problems with all my stuff getting shipped back from canada and not be one time for L1 conference [16:52] on time* [16:54] understandable [16:57] But I must shower [16:57] And hopefully today I can actually finish a warrior for 94b (and get 1st!) [17:34] Join: Judodudo joined #corewars [17:34] hello [17:35] Part: Judodudo left #corewars [17:55] Part: willvarfa left #corewars [18:18] Join: Roy joined #corewars [18:25] hey roy [18:25] hi bvowk [18:25] wassup? [18:25] not much [18:25] you? [18:26] well, its a bit of a bad week :( [18:26] Join: Core29 joined #corewars [18:26] yeah? [18:26] that sucks.. [18:26] The mother of a friend of mine heard she had a disease (infection near the brain) so she has a lot of medicine [18:27] Another friend of mine almost lost his grandfather, heart and brain infarct, but he survived [18:27] And yesterday, in the volleyball-hall, somebody of our (little) club suddenly collapsed and died.. [18:27] yikes [18:27] (during his game) [18:28] And of course my feet are stil hurting from my own little operation, but that just stands in the shadows compared to all tha drama [18:31] So I'm a little bit down, and hoping all the bad stuff stops now :) [18:34] SOMEONE NEEDS A HUG [18:39] * Roy ducks [18:42] :P [18:46] hmm 159...a new record :) [18:46] (against optimax) [18:46] Not so many rounds, so it doesn't say much yet [18:52] Join: Fluffy joined #corewars [18:53] :) [18:53] hi [18:54] Hey, HFB is still alive :) [18:56] That is interesting. 2by4k is a paper without a qscan [18:56] hmm ... probably not that interesting though, because with a qscan it would only fill in the gap to the 19th [18:57] Hrm yeah, its a bit behind [18:58] But it has aged the hill :) [19:01] To anybody, who is interested: What functions should a library have, that implements a MARS? I mean apart from the obvious answers like "run(warriors, numrounds, ..)" [19:03] Tournament scheduling [19:04] Compile(warrior) (for labels and loops) [19:04] datagram: Could you be more specific? [19:05] Set it up so that I can just add a bunch of warriors and it will run a tournament based on SE/DE/RR etc [19:05] Roy: Yes, you probably mean, that it should be able to parse everything, that pMARS can [19:05] SE/DE/RR? [19:06] Single elim, double elim, round robin [19:06] datagram: good idea [19:07] What do you think about including everything to run a hill (except the user interface)? [19:08] Evoler :) Optimizer [19:08] That would be nice [19:11] Roy: Evolving seems to be difficult, because there are so many methods for mutating/procreating/... [19:11] You don't want to be finished with it this decade do ya? :P [19:11] Actually that would be nice [19:12] Optimizing seems to be easier: Adjust the parser to recognize what to optimize, choose random values, benchmark, repeat until aborted [19:13] random values > kinda hard in cw.. [19:13] inside of the core, but not for the program to do [19:13] You don't have a time or whatever to get a good seed... you have to hard-cide it in [19:13] *code [19:14] Roy, Not in your redcode [19:14] rand(); [19:14] .. rand.. [19:16] Fluffis, I want to send a little test to the koth.org hill... but it'll maybe kill HFB, so I won't :) [19:16] Roy: Send it [19:16] I have faith in HFB. [19:16] Nah, its not fair, I'll wait until a empty spot pops up [19:17] If it's going to die it's going to die [19:17] I want to test the new Hulla, but not kill the other [19:17] Not much point in prolonging it [19:17] Hey, I still need a good xmas present. You could make HFB +200. That would be nice [19:17] hrm, thats a bit much aging... [19:17] I'll try ;-) [19:18] Roy, datagram: Thanks for the comments. Now I know, what to do next with PyCorewar [19:18] ...? I said something usefull? [19:19] :P [19:19] Sorry I can't be of more help [19:19] crazy busy... [19:19] and about to leave on vacation! [19:19] Lucky you ;) [19:19] Where are you going? I'm going to Paris next week! [19:20] datagram: No problem. I couldn't decide what to do next [19:20] Now I know it [19:21] Canada [19:21] Spending a month with an adooooooorable little asian girl (I'm white). but he parent's aren't going to like me! haha [19:21] his parents? :) [19:22] her* [19:22] ;P [19:22] Roy: Where is your warrior? [19:22] Well, just make sure you bring a nice present, like that movie I saw a couple of weeks ago, about a black family with a white boy friend :) (damn whats the name..) [19:23] Well they hate me already no matter what [19:23] Fluffy, making a nicer one for you currently [19:23] How come? [19:23] So it's quite an uphill battle [19:23] white + sleeping with their daughter [19:23] Parents don't seem to like that [19:23] datagram: Then you have to accept, that you are doomed ;-) [19:24] datagram: But I hope, that you'll have fun anyway [19:24] I am, and yes I will hehe [19:25] :) [19:25] Roy: If you to make a nicer warrior for me, be sure to include lots of a-imps in it :) [19:26] Too bad, it only has a b-imp... [19:27] datagram: Its already too late anyway for their parents to safe anything [19:27] *the [19:27] I know [19:27] They'll get over it in time [19:28] They'll like you from the start, just be sure to talk a lot about CW and bring some CoreWarriors as a present [19:28] datagram: You don't know how parents can be. Try to make the "mistake" of marrying her ;-) [19:28] (poor data, he won't survive) [19:29] Then you are trying to be nice: [19:29] "Sir, can I please marry you daughter?" [19:29] "No..." [19:29] "Tough luck bastard! :)" [19:29] datagram: ... and try to make them play corewar manually (each parent executes his own instructions ;-) [19:30] Haha, we don't want to get married (that's probably what bothers them, too) [19:30] Her dad was a comp sci major in school, so I guess we can talk about that or something... [19:30] "We don't like you, but we hate you for not marrying her!" (??) [19:30] datagram: I don't know how easy it is to become divorced [19:31] well, they would be more okay that we were together if it was long term [19:31] Does that make more sense? [19:32] But we're not really concerned about that. Just sex and having fun together right now [19:32] datagram: It will probably never make sense, but only you have to understand it [19:32] I'll all make sense once you have a daughter who starts dating 'just for sex and fun' ;-) [19:32] Nah, no kids for me! [19:33] (yet) [19:33] (ever!) [19:33] bvowk: You need to censor the logs again or Google/Yahoo/whatever will ban koth.org for talking about sex ;-) [19:33] haha [19:33] Boobies! [19:33] No, we might (for the first time) get visitors!! [19:34] haha [19:34] Yeah, thats it, I know how we can make this game populair! "Corewar, the most sensual assembly language of all" [19:34] We probably would get more visitor, if we'd use better names for our warriors on the hill [19:34] I'll name my new warriors after porn stars [19:34] That will help [19:34] "The mother of programming games, mother...but more MILF" [19:35] datagram: Then koth.org will be shut down, because the names are probably trademarked ;-) [19:35] hehe [19:35] We just put *all names and situations are completely coincidental* at the bottom [19:36] Roy: How can we get new beginners, if they are being shut out from koth.org due to the restrictions of their nanny-programs [19:36] Err... [19:36] Thats where out Fluffy Bunny Mirror kicks in [19:36] *our [19:36] datagram: Maybe we should start to add standard disclaimers to all of our warriors [19:37] I don't think it's going to make a difference either way [19:37] http://fluffybunnyanimalsandothersweetstuff.org [19:37] (its still available!) [19:39] Roy: where warrior? [19:40] I'll send it... [19:40] now? [19:40] But don't complain when I push of hfb! [19:40] I still believe in harmless fluffy bunnies ;-) [19:40] I started on a new 94b entry earlier [19:40] Anyway, who would want to kill them. It would be cruel!! [19:42] Yes, I didn't see it on the hill, whats its status data? [19:42] datagram: already sent to hill [19:44] hmm, only third :( [19:45] hmm ... try to DAT the hill and use a better name [19:45] This was just the over-optimised Hulla, my real warrior just started optimising [19:46] what do you mean by "over"-optimized? [19:46] Probably not over-optimised, but its near its limits [19:46] Time for something new [19:48] Roy, I haven't finished writing it yet... [19:49] What is it going to be? [19:52] SI, of course [19:52] maybe a pswitcher (I can fit the code in with quite a bit of room to spare) [19:52] Join: Fizmo joined #corewars [19:52] hi [19:52] Hi Fizmo [19:53] Hi Fizmo! [19:54] Hi fiz [19:54] Join: inversed joined #corewars [19:54] Hi inversed! [19:54] Hi all (and especially pHluff) [19:54] You are koth on 94nop :-) Congrats! [19:55] Grats :) [19:55] That is because of that 20th warrior scoring ~50 [19:56] In any case, Rust has done well lately [19:56] Yes, it even beats some other papers (it wasn't designed to !) [19:58] In your strategy line you have written "New qscan". Does that mean new qscan as in "added an existing qscan" or "added a newly developed qscan"? [19:58] Just replaced q3 with q4 [19:59] ah, ok [19:59] Flu1515: I have an excellent evolving idea and you have fast Mars; do you want to try writing evolver? [20:00] papers which can beat papers seems a good task [20:01] if they resist scanners they'll be perfect [20:01] inversed: I'm already evolving for '88 with my new Python library :-) [20:01] Fiz: I've done huge work on 'paper vs paper' theme, I'll publish it soon [20:02] :) [20:02] Fizmo: Some '88 papers can beat silk-style papers easily, but unfortunately they are easily killed by scanners :-( [20:02] yeah, 88. Try to add a stone [20:02] inversed: Could you please give a short overview of the article? [20:03] Fizmo: Then it would no longer be a paper :) [20:03] Some results: Moore's paper isn't the best paper to kill other papers [20:03] and which is? [20:03] woot?!? [20:03] really!?! [20:03] I thought it would be the best [20:03] but silk-dwarfs can also beat some kind of papers [20:03] And Moore's paper kills other papers not because it cannot be executed by other papers [20:04] especially paper with silk-imp ala Fixed [20:04] it works best against coreclear paper [20:05] and other papers which have a loopback at the end [20:05] I have a very important conclusion: main reason why paper CAN'T beat paper is silk-imp degradation [20:05] That's it: after long time, almost every instruction is a kind of imp [20:05] silk-imp degradation? [20:06] probably means: everything becomes an imp [20:06] But it only applies to silk-style paper; Moore-style doesn't degrade so it can beat some silks [20:07] Hmm.. intresting [20:07] inversed: I'm already looking forward to read your article [20:07] well, it's because of the trash the paper leaves in the core [20:08] dat 0, 0 will be after a certain time a minor instruction in the core [20:08] No, it's because movs in silk pairs. [20:09] I've tuned almost any paper to beat other papers and now I'll post scores against my testset [20:09] hmm, if the a-field where the mov points to has a larger number [20:11] so, is there a practical solution to beat those silk-imps? [20:13] Testset: papers from: benj's revenge, my first paper, 2b0r!2b, reepecheep, cheephalfoff, soldier of silkland, venomous philtre, tyrant, disincentive, wallpaper [20:13] as I see in your strategy line a paper with separate d-clear. If the d-clear survives until the very end could this prevent silk-imp degradation? [20:14] Yes, dclear takes over after some time, so I can beat stone'imps without having anti-imp paper [20:16] inversed: good collection, but I miss nPaper or the similar Teenage Metal Isis [20:17] They are not included because they are not 'weak papers': they are timescape-style papers and they can't be beaten by other papers [20:17] and nice to see four of my papers out of 10 :) [20:17] ok for comparision: Hullabaloo scores 180. [20:18] ok, I see [20:18] there are also some few 'open-looped' papers like Head or Tail [20:18] my CCPaper scores 160 (here and further scores quantized to 5 points) [20:19] Silk 145 [20:19] Kline-style 190 [20:20] Paperone-style 200 [20:20] wow... [20:20] I think Sascha mentioned that some time ago [20:20] Moore 6-line 210 [20:21] Moore 8-line 225 [20:21] LP_1 190 [20:22] LP_2 235 (!!!) [20:22] Muffineer? [20:22] I know that it can kill some few 'normal' papers [20:22] So best paper to beat other papers is Heavily-modified LP paper. [20:23] Hmm, how cool [20:23] Not muffineer, but BioMech. I uses very cool evolved bombing technique [20:23] *It [20:24] But of course it cannot be used in 94nop, so real choices to combine with stones are Kline-Style and Paperone-style papers [20:26] I've tryed using evolved Moore paper + Stone (Digeridoo), but it wasn't able to kill imps so it was pushed off the hill soon [20:27] Important thing: all of these papers are bombing with 'ADD' to kill those silk-imps and also mov >a, {b [20:28] FluFF: You used this bombing in Yatima, how you guessed ? [20:29] I was unsuccesfull working on a paper which bombs with spl. Now I understand, that I increased the effect of paper-imps [20:30] Now I'm downloading mp3s of russian electronica band called Fizzarum ;) [20:31] lol [20:31] I hope they make good music :) [20:31] Near-idmish experimental stuff (http://www.fizzarum.com/) [20:32] inversed: Yatima used "mov > -bstep, { bstep/mov >-2*bstep, { 2*bstep/mov >-3*bstep, { 3*bstep" [20:33] If you're interested in my music, check out my page on mp3.de [20:33] http://www.mp3.de/musik/genre/band/060800/232508/1 [20:33] I have to download something while chatting because I have a dialup and I don't want to just waste my inet time [20:35] And if you are interested, you can check out http://dcsnake.narod.ru/ - it's game by me and my friend + soundtrack by me [20:35] FluFF: why you used namely mov > , { ? [20:36] inversed: I've used it only by pure luck. At this time I simply tried every possible variation of a paper and optimize the better ones [20:37] inversed: by the way, at this time I only optimized and tested against WilFiz [20:37] inversed: still downloading the soundtrack [20:37] Now it's scientifically proven that it's the best mov [20:37] against your collection of papers [20:38] The question is, what is the usual distribution of the different paper-variants [20:38] (if such a thing exists) [20:39] Ok, time for me to take the things we've just learned in a real warrior [20:39] Long time since I've created a real real paper [20:39] Roy: Is it already ready for the hill? [20:39] I'm not THAT quick.. [20:40] is it ready now? [20:40] Almost dammit :) [20:40] PhuFF: no, it's better against almost any paper: second adressing mode MUST be { to kill silk-imps (mov }0, ...) [20:41] inversed: You really need to hurry and publish your article :-) [20:41] Roy: already sent your new paper? [20:42] didn't see it appearing on 94nop so far [20:42] Just have to write a nice boot and think of some bootdistances!! [20:42] ;-) [20:43] inversed: did you test also silk-dwarfs? [20:43] Fiz: Soundtrack is not very good (SoundForge only), but I wanted it to match the game (16 colors, funny gameplay, ...) [20:44] a c64 sounding stuff would be funny [20:44] Fiz: No. There are other paper engines that I want to test, but my testing is actually evolving against testset [20:44] Ok, running benchmark now, to see if itw worth sending [20:44] there are a freeware softsynth which can nicely do that [20:45] MSG: Quit: Trillian (http://www.ceruleanstudios.com [20:45] Fiz: can you give me a link? [20:45] download is unfortunately very slow [20:45] Roy: you are really to slow [20:45] ;-) [20:45] Fiz: sweetheartofaberzerk is your mp3 page ? [20:46] not yet. I have to check on my music pc the name of the synth [20:46] yes. That's my stuff [20:46] check this page: [20:46] http://www.kvraudio.com/ [20:46] there you find all kind of free vst plug-ins [20:47] even cubase like programms as freeware [20:47] Unfortunately I can't listen to any stuff with vocal [20:47] send! [20:48] Roy: thanks, but really too slow ;-) [20:48] Sorry :-( and it sucks too ;) [20:49] Its a bit scannerfood.. [20:49] It only has to beat P. Kline's paper [20:49] Even thats hard [20:49] Anyway I'm already nearer to +200 :) [20:49] inversed: I have also plenty of pure electronical stuff, but actually not downloadable on the net [20:52] yippie, now 172 :) [20:52] hmm, download always stops at 92 kB :-/ [20:52] ok, gotta go now [20:53] * Fluffy waves [20:53] want to play some cs:s before going to bed [20:53] Aww.. not waiting for my paper, really slow :( [20:53] bye fiz ;) [20:53] bye all [20:53] * Fizmo waves [20:53] * Roy waves [20:53] MSG: [20:53] inversed: About your idea for evolving... what would you like to do? [20:55] bye fiz (oh that stupid free hosting) [20:55] Roy: :-( [20:56] ? [20:56] I've expected HFB to become 173 [20:56] Noticed a bug, new version HAS to be stronger ;) [20:56] but it didn't [20:56] Its not dead yet [20:57] and it won't die [20:57] never [20:58] Maybe I should switch the "DON'T SEND NEW WARRIORS TO 94NOP! NEVER!"-sign on again [21:01] Flu1515: My idea is Markov chains taken one step further. I already have a program to generate initial population [21:02] In which regard "further"? [21:03] Read: mod of Markov chains. Some comparision: [21:04] Valid programs (progs that can kill dummy) % [21:04] YACE 6.5 [21:05] 1st order 17.0 [21:05] MyMod 28.5 [21:06] what is your dummy? a "jmp 0"? [21:06] Y [21:06] Oh wait no, it's 99 nops and jmp -99 [21:07] hmm, it can alread be killed by a simply djn.f-stream, have you tried a "jmp # 0"? [21:08] Yes, I'll try it but anyway my mod seems to be very usefull. I have an extra characteristic for instruction: [21:08] altough it is definitly overkill, a score against a benchmark would be nice for comparison, too :) [21:09] Where it a and b fields point (what instruction) [21:09] I can send you this prog if you want [21:10] I'm definitely interested :) [21:10] C? [21:10] Perl? [21:10] Delphi. Need source? [21:10] delpi? [21:11] err ... Delphi, that is some kind of Pascal, right? [21:12] You can read + translate it I think... Delhpi isn't that complicated [21:12] bvowk: your nick [bvowk] looks like it was evolved ;) Delphi/pascal = VB/basic [21:13] eww.. [21:13] don't say things like that.. I'll get insulted. [21:14] inversed: I would like to have the source, altough I'll probably "translate" it into Python anyway [21:14] I thought it's a compliment for volver-guy [21:14] *evolver [21:14] bvowk: As you seem to be alive ... just in case for future fun ... [21:14] * Fluffy pokes bvowk into his left eye ... again :-) [21:16] Now everybody has already two reasons to ban koth.org: (needless) violence and sex ;-) [21:16] ow! [21:16] can I help you fluffy? [21:16] quite a long time to react [21:16] * bvowk sears fluffy over an open fire [21:17] hmm ... I hope, I'll taste well [21:18] bvowk: you could help? What features should a corewar-library have? [21:18] err .... * you could help. [21:19] a corewar lib? [21:19] well, you should have parsers [21:19] yes [21:19] thats a big help.. [21:19] FluFF: sending you my generator with all stuff. No error checking. [21:19] parse the code into simplified pmars format, read the code and divide it into its components.. [21:19] inversed: Thanks :-) [21:20] How do you define components? sth. like qscan or paper? [21:21] ? [21:22] inversed: Why the "?" [21:22] ? equ 'What your last sentence means?' [21:23] it was a reply to bvowk [21:23] (22:16:54) bvowk: parse the code into simplified pmars format, read the code and divide it into its components.. [21:23] components like instruction, modifier, amod, a, bmod, b :) [21:23] ok. I hope you'll understand how to use NRG [21:24] bvowk: ah, ok ... the current version has only a simplified parser, but I'm already working on a parser, which supports all of pmars' features [21:24] inversed: NRG? [21:25] Maybe I should make the first release of PyCorewar soon ;-) [21:26] My generator: NRG | eNeRGy | NewRedGen [21:26] stupid me [21:26] Downloaded? [21:27] not yet ... at first I have to reassemble my "big" computer [21:28] I'm usually sitting in front of my old laptop [21:29] I'll mail you complete description later [21:30] :) [21:31] big? [21:31] did you see the pictures of my big computer fluffy? [21:33] bvowk: URL? [21:33] bvowk: is 'Machines will rule' really evolved? It looks like man-made paper optimized as hell [21:33] bvowk: small = laptop, big = desktop computer [21:34] inversed: yeah.. over a bunch of months, with a huge wad of machines [21:34] But it boots two exact copies of paper [21:34] since we seed the chains with human coded warriors, its not entirely surprising that it looks like a man-made paper.. [21:35] yeah.. but it had to use two blocks, so I guess it just used the best paper it had twice :) [21:35] er.. three blocks I think.. [21:35] in retrospect, that was likely too big [21:36] since most of the successful ones evolved broken 2nd or 3rd blocks [21:36] At least they should have different constants [21:36] I didn't make that a condition.. so no reason it had to follow that.. [21:36] that was the one that scored the best... soo. [21:36] Thingamabob was basicly that paper with a good qscan and other boot :) [21:37] :P [21:37] no wonder it looks man-made, the men keep stealing it :) [21:37]