[03:25] MSG: Quit: Leaving [11:55] MSG: Remote host closed the connection [14:09] Join: Metcalf joined #corewars [14:09] Hi [14:12] Join: OoS joined #corewars [14:14] MSG: Ping timeout: 245 seconds [14:17] Join: Lazareth joined #corewars [14:26] Hi Lazareth [14:29] Hiya [14:29] Messing around with redcode again, trying to build a scanner from stretch, just to look at the problems I'll face and learn from trying to work around them [14:30] Thus far I have it scanning doing this: JMZ 0, }2 [14:30] and only that [14:30] at 1c [14:30] That's a good start [14:30] Though linear scanning isn't efficient [14:30] with a JMP pointing to itself afterwards to send it forward to something I haven't thought of yet =P [14:31] A typical opponent is between 4 and 15 lines long [14:31] Thus far I'm thinking of doing it in larger increments, overlooking the slight problem of hitting myself with the scan, but it is hard to fit into as few instructions (preferably one) as possible [14:32] Have you looked at the scanner article on corewar.co.uk? [14:33] not really, I'm mostly trying to figure things out on my own - to iron out any errors of wrong thinking. [14:33] but maybe I should give it a look =P [14:34] ...can't find it ^^ [14:34] oh, wrong site *doh* [14:34] If the scan finds something, you can easily check if it's part of your own code [14:34] how? [14:35] by checking the distance? [14:36] found it, reading :) [14:36] Yes [14:53] MSG: Ping timeout: 245 seconds [14:54] Join: OoS joined #corewars [15:08] MSG: Ping timeout: 245 seconds [15:08] Join: OoS joined #corewars [15:24] Join: Neogryzor joined #corewars [15:24] heya! [15:24] Hiya [15:25] Hi. New here? [15:26] si [15:27] Begun yesterday evening [15:28] still can't make anything resembling useful =P [15:28] Welcome to the corewar comunity. :P [15:28] thank you very much. I won't be very active until after the 24th of next month. Exams. [15:29] Don't hesitate to ask [15:29] will do. But it is not so much as to how it works, but as to how make it work in a useful way :) [15:30] heh [15:30] there are a few interesting guides for beginners [15:31] noticed. Not much in just creating simple imps and dwarves, though :D [15:33] Thus far I've managed to stomach simple "routines" or what you would call it. Linear scanning, gates, imps, stones, etc =P not anything useful in itself. [15:34] Did you see Steven Morell's guides at http://www.koth.org/info.html? (they are about imps an bombers though) [15:35] Read it once at the beginning... could be okay for a reread. I'm mostly reading right now, did some experiments yesterday [15:41] Take your time. Corewar needs patience but it can become addictive too :) [15:42] already is addictive, can just be frustrating to not be able to do the more complicated stuff - it becomes kinda convoluted when you try to press an idea into a fully optimized warrior in one go [15:44] okay, what do you want to do? [15:46] Hi Neo [15:46] Everything :D actually, it's the small steps that become hard to master. I start out with an idea of fx. bombing with "imp" instructions (MOV 0, 1) and at the same time having a perfect gate behind me, catching the "impified" enemies. Silly concept, but rather hard to do. Oh, and beyond that I would like not to hit myself, etc. It's when the simple idea is expanded that it becomes hard to fit it together =P [15:46] Hi John [15:49] mov imp,50 / add #10,-1 / djn -2,#790 /jmp 0,<-5 /imp: mov 0,1 [15:49] just as an example, it is not competitive :) [15:51] I'll send people out the door, back a while later :) [16:00] MSG: Ping timeout: 245 seconds [16:00] Join: OoS joined #corewars [16:07] Hey John. Would you like to try the paper generator i'm working on and give me you oppinion? (mucho work to do yet though) [16:10] Yes, I'd be interested to see it [16:11] http://labarga.atspace.com/redcode_web.htm [16:11] Not compatible with Firefox for some reason :( [16:15] I know it's not a wonder :-) [16:25] MSG: Ping timeout: 245 seconds [16:26] Join: OoS joined #corewars [16:27] How long does it usually take for some beginner to make anything competitive ^^; a month or two? A year? [16:28] "competitive" in the light sense ofc =P [16:29] depends [16:29] give or take [16:29] I was playing on the beginner hill for 8 months before I entered the pro hill. [16:29] beginner hill? [16:29] sure, you can just copy some paper, and then force-optimize its steps and get on a hill in a day [16:29] Depends on the amount of time available. There is a hill for beginner where you can start with [16:30] he probably refers to the death Pizza beginner-hill [16:30] sal.math.ualberta.ca has working beginner hill [16:30] death -> dead [16:30] "force-optimize"? Oh, and I'm mostly thinking of how long to making something that isn't just a quick rework or smash together of somebody else's code. [16:31] I prefer to not reuse code unless some very innovative reapplication or rework shows itself =P Am I doing it wrong? [16:31] making a program to fight a warrior against a benchmark, with randomly changing some numbers, and taking one that scores best [16:32] Neo: In firefox I get a config is not defined error [16:32] making something new and creative is a challenge, not a process [16:32] o.o thus far what I've made required quite the hand-crafting. Wouldn't randomly changing numbers result in more missteps followed by self-termination than any actual useful results? [16:33] Lazareth: I's Ok. We all like to use our own code but when an important improvement is achieved it is usually published to make it publicy available. [16:33] OoS: Does it work to you in Firefox? [16:33] Lazareth: when you fight hundreds of battles per minute, it doesnt matter if there are 'more than average' missteps [16:33] Lazareth: it took me 8 months to make it onto the pro hill because I didn't want to reuse someone else's code ;-) [16:34] Neo: no [16:34] I can live with 8 months of crappy, self-made code if I make it to the pro hill after that while still using my own stuff [16:36] Lazareth: You just have to read others' code and the strategies [16:37] that's the stage I'm at [16:37] But meanwhile I also have to pay attention to my exams starting tomorrow =P The first one is my native language, so it is not much of a worry (and I couldn't care less for the result since I'm going to the university no matter what) [16:38] there also exist a Corewar newspaper called "The Corewarrior" with plenty of code and tips: http://corewar.co.uk/cw/index.htm [16:38] Okay, I've registered at sal.math.ualberta.ca and I'm looking at the beginner's hill, but I can't find the "submit" option or whatever [16:38] Neo: you could set the paper generate to give output which can be used with optimax [16:39] Lazareth: send an email to with your warrior [16:39] OoS: Hey, good idea! :D [16:39] Neo: it works with Safari [16:40] OoS: won't that land me in the "pro" hill, which would wipe out such warriors as my puny, self-made experiments? [16:42] MSG: Quit: http://www.mibbit.com ajax IRC Client [16:42] Join: Neogryzor joined #corewars [16:44] Lazareth: You can write in the code in wich hill you want to submit your warrior [16:44] Lazareth: put the hill key at the top of your warrior - e.g. ;redcode-tiny for the tiny hill, ;redcode-94b for the beginner hill, etc [16:45] oh [16:45] I'll try and submit my "Format C" linear (backwards) bomber. I do not expect it to fare well =P [16:47] nothing goes in the subject line of the mail? [16:48] no [16:50] Yay, it says assembly succeeded. Is there a place where I can monitor it? [16:51] found it! [16:53] 21st :-) [16:53] woah [16:54] That's higher than I expected even for the beginner hill.. Well, I didn't expect it to get on the list at all [16:55] Lazareth: how far away do you begin your clear (linear bombing)? [16:56] Since the thing was by concept (I just submitted it to test the submitting), it begins bombing just before itself and backwards from that. Meaning at -1 from the first line of the warrior [16:57] and it spends a cycle jumping over itself just before the bombing reaches itself [16:58] well, resetting the bombing sequence. [16:58] The minimum distance between the start point of 2 warriors is 100 [16:58] See this: http://www.flickr.com/photos/avi_abrams/1097553646/sizes/o/ [16:59] Since most warriors are less than 100 lines, even if your opponent is right behind you, you'll bomb 90 or more empty cells before you hit it [16:59] I know, so it should logically start there [16:59] OoS: you didnt find any refrerence on the scoring of Swing/Pswing? [16:59] But hey, at least the "bombs" are colored and it bombs backwards, which I'm hoping will mean that the scanner finds the bomb line before it finds me =P [16:59] Yes, about -100 would be a good place to start a backward clear [17:00] *a scanner [17:00] Mizcu: no [17:01] i gotta reread the '94 warrior:s, i remember there having been a warrior with huge score, but lasting only a week [17:02] (i dont remember how long Trident stayed on the hill) [17:05] Oh, my... 5 52.5 9.1 38.4 Ke-te-tum-bo Neo 195.9 1 lol! [17:05] MSG: Ping timeout: 245 seconds [17:06] Join: OoS joined #corewars [17:07] huh? [17:07] ah [17:08] Is there anything else new on your webpage neo? [17:09] Back later [17:09] no, just that script. I want to include all the Corewarrior issues. [17:10] That generator is dangerous in newbie hands... heh [17:11] wouldn't use it, can't see what there is to learn from pregenerating >.> [17:15] I think the interesting thing is to see the many ways to implement the "paper" strategy and tricks [17:16] uh, if I use a unary statement (like ! - I assume this means NOT) in a CMP OpCode, can I then compare the number 1 for true or 0 for false with the result? Such that CMP 1, @3!2 would be seeing if the B-field of of address 3 is anything but 2, if true skip next instruction? [17:16] no... [17:17] but i think you can use it to define labels/values [17:19] value equ 100+( !(CORESIZE==8000)*100). I haven't tested it though [17:19] ...I will play around with it in the debugger after I've done some "homework" for the exams =P I'll be lurking around [17:19] i have to go [17:20] good luck with your exams [17:20] back later [17:20] Part: Neogryzor left #corewars [17:22] MSG: Ping timeout: 245 seconds [17:23] Join: OoS joined #corewars [17:42] MSG: Ping timeout: 245 seconds [17:43] Join: OoS joined #corewars [17:51] MSG: Ping timeout: 245 seconds [17:52] Join: OoS joined #corewars [17:57] MSG: Ping timeout: 245 seconds [18:58] Yay, I try to make an improvement, thinking "now it'll be faster thus more vicious" and the improved version actually scores lower than the previous xD [19:01] thats why people have benchmarks of their own to 'dry-test' [19:02] and dont be afraid of stoney #3 (actually #4..), it dies just like everyone else [19:04] I just after-tested the new one to see what went wrong... Turns out the scan didn't do exactly as predicted, despite falling through a debug test. I tried setting up against cannonfodder, finding out that it didn't actually scan anything but ran a full faux scan of the core and then proceeded to act like Format C >.> so it was kinda a delayed Format C, explaining why it scored so horridly. [19:07] Well, I kinda benchmark using the beginner's hill... seing as how I don't have any "intentionally weak" warriors laying around and that benchmarking against my own stuff would most of the time just involve two bugged warriors running against each other. [19:09] thats kinda understandable problem, yes, because even weakest of published benchmarks, wilkies, is one that can cause depression in beginners [19:11] I don't care if it beats me, but I don't want it to beat me just as much if I present it with a warrior thrice the strength of what I previously gave it. Then I won't be able to tell if I'm upgrading or downgrading. [19:12] which is also a problem, because against a balanced benchmark, beginner's warriors are able to give a fight against only a half of it [19:13] And as such, I like the Beginner's Hill, since I immediately spotted that "Scan and Clean" performed unexpectedly [19:23] I'll be off till tomorrow [19:23] MSG: Quit: Leaving. [19:51] Metcalf, this one is for you: http://en.wikipedia.org/wiki/Halt_and_Catch_Fire [22:29] Join: gamerdonk joined #corewars