[00:03] Join: |caesar joined #corewars [00:06] <|caesar> morning,everybody [00:09] morning [00:11] <|caesar> aha,how r u,brx? [00:13] I am angry. [00:14] <|caesar> oh no,whats wrong, brx? [00:16] haskell's typesystem is going on my nerves [00:17] apparently I have to rewrite lots of code to make up for a design mistake [00:28] <|caesar> that's bad,repair it is a funny i thing [01:29] MSG: [02:02] MSG: Remote host closed the connection [04:26] Join: Mizcu joined #corewars [04:43] Join: datagram joined #corewars [04:58] Evening [04:58] Or morning. [05:25] Join: fiveop joined #corewars [05:28] <|caesar> :) [05:28] <|caesar> that's funny,here's afternoon [05:29] Yes of course [05:29] <|caesar> but datagram is deep Evening [05:30] <|caesar> hey,datagram do u wana sleeping? [05:31] <|caesar> or maybe i wake up u sleeping time [05:31] I have a different schedule [05:31] I sleep during the day [05:32] <|caesar> ...is it well? [05:32] <|caesar> iwana sleeping in day2 [05:32] ? [05:32] So do it [05:32] <|caesar> yeay,i'll try to do it [06:12] Join: caesar joined #corewars [06:14] MSG: <|caesar> Quit: GoodBye , My Friends... [06:14] Part: caesar left #corewars [06:14] Join: caesar joined #corewars [06:18] Part: caesar left #corewars [06:22] Join: |caesar joined #corewars [08:05] Join: willvarfa joined #corewars [08:37] Join: brx joined #corewars [08:56] Join: Roy joined #corewars [08:57] Hello everyone [08:57] hi [08:58] willvarfa, The hacker buddies who I'm getting into corewar mentioned that we should code a php interface for hills : ) We could probably crank it out in a day if we worked on it together....so maybe your dreams aren't far away [08:58] Hello Roy [08:59] What's up [08:59] wow, that would be cool, do you have scripts for a hill? [08:59] then we could have a copy&paste hill ;-) [08:59] No [08:59] a nice interface to an existing hill would be nice [08:59] It wouldn't be hard [08:59] sal comes to mind [08:59] Or maybe you could include it into SAL [08:59] * Roy thinks the same [08:59] it would be nice as an XUL interface too [09:00] In the third week of november I give a presentation about Corewars to a group of hackers [09:00] So after that more will be interested [09:00] Nice, good luck [09:00] Yes, I have to try to not suck by then [09:01] Hehe, would be nice to say things like: "Ohw, I'm currently on the top 20 hill, best of the world, come beat me etc etc" [09:01] I challenged them by saying I'm currently KOTH :-) when I held a presentation at college [09:03] Heh [09:03] Well make a hill where only I have warriors [09:03] I'll for sure kick ass on that one [09:06] I will run the powerpoint to you guys before I give the presentation so that you can scrutinize it [09:09] Join: Nenad joined #corewars [09:09] Hi [09:09] hi nenad [09:10] :) [09:10] did you get my mail? [09:10] Hi Nenad [09:10] Hi datagram [09:10] Did you read the log?:) [09:11] err... not yet [09:11] But the text is great :) [09:11] Read the logs at 8:56+ [09:11] Today [09:11] So that I don't have to spam : P [09:11] :) nice. [09:13] So yes. [09:14] Also: my warriors are catastrophes [09:15] they are not... [09:15] what did you make? [09:15] avamp? [09:17] A vampire that eventually turned into a stunner [09:17] And a very bad bomber [09:22] Not sure how to proceed [09:22] well, why is the bomber bad? [09:22] Heh, 0-10-0 against aeka rave and flashpaper [09:22] plain bomber cannot defeat aeka or flash [09:22] you should go for 40-50% against Rave first [09:23] Does it suicide..? [09:23] or stun itself? [09:23] It shouldnt [09:23] what bombs does it use? [09:23] The bomber? Dat [09:23] I think I messed up the stunner though [09:23] You could paste the code, then we could have a peek [09:24] One sec and I will [09:24] Working on multiple terminals [09:27] Yeah the code for the stunner is messed up [09:28] what bombs does it use? [09:28] MSG: Ping timeout: 252 seconds [09:29] Join: brx joined #corewars [09:29] so, show us the code... [09:31] mov.i 4, >3 [09:31] mov.i 4, >3 [09:31] djn.b -2, 1 [09:31] jmp 3, #15 [09:31] spl 0 [09:31] jmp -1 [09:31] mov 2, >2 [09:31] jmp -1 [09:31] dat #0, #5 [09:31] The jmp field doesnt work right [09:32] It keeps going 15/16 [09:32] Because of the djn [09:32] the second move overwrites the spl 0 [09:33] in the second line, you ment mov 4, >2? [09:33] Ah yes [09:33] try using labels [09:33] Wait [09:33] No [09:33] Wait yes [09:33] mov 5,>2 [09:33] Haha [09:34] They are both 4 [09:34] Because it is relative, no? [09:34] ah yes [09:34] but the djn.b decreases the pointer, you should try djn.b -2,>1 [09:34] (or something) now it just drops spl's [09:35] It only puts it at the one location [09:36] But now it does this [09:36] spl,jmp,dat [09:36] I guess that is good anyways? [09:36] MSG: Ping timeout: 252 seconds [09:36] The djn makes dat 0, -1 [09:37] If I am understanding correctly [09:37] Join: brx joined #corewars [09:37] No, it just decrements the jmp 3,#17 to jmp 3,#16 and then writes a spl again [09:37] What is causing every third instruction on the bombing path to be dat 0 -1? [09:37] djn.b -2,@1 [09:38] err djn.b -2,>1 [09:38] It's doign that [09:39] Rave hit me with spl 0 at the top of my code though [09:39] : ( [09:40] No, you did that yourself probebly [09:40] It didn't go all the way around [09:40] I understand, but how does you code stop you from doing that? [09:41] It is spl.a [09:41] My bombs are spl.b [09:41] Rave's bombs are spl.a [09:41] Hmm ok, but your code (in a self fight) stuns yourself [09:42] Yeah the loop to jump out isn't operational yet [09:42] you keep bombing spl 0 jmp -1 until you hit yourself again [09:42] It is supposed to jump to the coreclear [09:42] how? [09:42] Supposed to, but doesnt work because djn never fails [09:43] Indeed, let me make some adjustments to the code, one second [09:43] I know I also need better step sizes [09:43] But I just wanted to get this working first [09:43] Is it bad that I'm moving the jmp -1 also? I like that it makes things stuck [09:44] No its ok, some stunbombers do that [09:44] here, its a bit better this way: [09:44] mov.i bomb2 , mov.i bomb1 , djn.b -2 , ptr [09:44] mov datbomb , >datbomb [09:45] jmp -1 [09:45] datbomb dat #0, #5 [09:45] ptr dat #0, #-11 [09:45] bomb1 spl #0 [09:45] bomb2 jmp -1 [09:45] I don't like labels [09:45] They are in the code on the machine I'm running on, but I prefer numbers [09:45] ok, I'll remove them (makes it easier though) [09:45] datagram: make a spl 0 jmp -1 bomber instead [09:45] Well I figure it is easier to just look x lines down or up instead of having to fine the line the label is on : ) [09:46] Nenad, That is what my code is doing, da? [09:46] mov.i 8 , <6 [09:46] mov.i 6 , <5 [09:46] djn.b -2 , 4 [09:46] mov 2 , >2 [09:46] jmp -1 [09:46] dat #0 , #5 [09:46] dat #0 , #-11 [09:46] spl #0 [09:46] jmp -1 [09:46] datagram: well, no. [09:46] you're just pushing your bombs through the memory [09:47] instead of throwing them [09:47] you warrior is more like a coreclear [09:47] *your [09:47] Well, pretend they were spaced right [09:48] I realize they are sequential, though [09:48] Roy, it should be mov.i 6 for both, right? [09:50] Heh, that code bombs itself, I think [09:52] No it doesn't, because the <6/<5 is placed for away [09:53] I got it working [09:53] the djn.b -2,4 decrements the dat#0,#11 [09:53] Before it bombed itself [09:53] and then that becomes 1 and it jumps to the clear [09:54] When I ran it it overwrote the dat instruction with a spl [09:57] Woot [09:57] I like watching paper get caught [09:57] It makes me feel special [09:58] Does it kill it well now? [09:58] Roy, I changed the code to: [09:58] mov.i 6, <5 [09:58] mov.i 6, >4 [09:58] Err >5 [09:58] And it works [09:58] Before you move a dat instead of the spl [09:58] so it makes dat jmp -1 [09:59] Ah this is cool [09:59] I'm testing vs flashpaper [10:00] MSG: <|caesar> Ping timeout: 252 seconds [10:00] It caught quite a few and now the processes are shooting very high for it; it isn't bombing [10:00] However I did overwrite myself : ) [10:01] So the dat coreclear didn't fire off [10:01] But that was fancy enough [10:05] You want a fancy version? I went rewriting it a bit: [10:05] sub.ab #1953 , 3 [10:05] mov.i 6 , <2 [10:05] djn.f -2 , 1 [10:05] dat #0 , #-1 [10:05] mov 2 , }-4 [10:05] djn.f -1 , }-5 [10:05] dat #7 , #7 [10:05] spl #1 , 1 [10:07] it bombs with a interval trhough the core with spl #1,#1 [10:07] and when it bombs the dat #0,#-1 line the djn slips through [10:08] then it has a spl #1,#1 mov 2, }-4 and does a d-clear [10:08] Hmm [10:09] But I like the jmp, so that it gets their code stuck and slows them down very badly [10:09] the spl does the same thing too [10:09] maybe you should look at WW3 then... [10:09] But maybe I can change it back to a jmp too [10:09] WW3 = WinterWerewolf3 [10:09] Oh okay [10:10] Does the process that split stay where it is for it's next cycle? [10:10] I thought it moved down 1 [10:10] And only the split process stuck at 0 [10:10] yes, one is let through and the other stays on the spl line [10:11] Is taking an extra cycle to add jmp -1 behind it that much more inefficient? [10:12] the problem with the extra jmp is that we now use selfmutation, and placing a spl #1,jmp -1 in your code stuns yourself [10:12] WW3 is a bit evil, it is good example of self-mutation and getting different kinds of stuff packed in; it becomes even more remarkable once you notice it is in '88 [10:12] yes. :) [10:12] I like it a lot. [10:17] ok datagram, new version: [10:17] (now with spl #1/jmp -1 bombs) [10:17] mov.i 9 , @4 [10:17] s sub.ab #1954 , 3 [10:17] mov.i 6 , >2 [10:17] djn.a -3 , 1 [10:17] dat #0 , #-1 [10:17] mov 2 , }-4 [10:17] djn.f -1 , }-5 [10:17] dat #7 , #7 [10:17] spl #1 , 1 [10:17] jmp -1 [10:17] end s [10:19] One second I'm doign a tournament [10:21] hrm that warrior actualy beats some good warriors :-) [10:22] MSG: Ping timeout: 252 seconds [10:23] Roy: send it to 94b just for fun [10:23] I'd like to try to improve it and then send it [10:23] ok. :) [10:24] Try the last version I send you, but don't ruin it ;-) [10:24] No promises [10:24] add a huge decoy and boot away [10:24] it will help against oneshots [10:24] and a qscan and you're done ;-) (its bad against papers...ssd clear would do better) [10:24] I don't like scanners very much [10:24] Join: brx joined #corewars [10:24] Though I did want to try a vampire scanner [10:25] But I saw someone trying that, so I didn't feel as original [10:25] even in my dreams I don't allow myself to be lucky [10:25] ? [10:25] Having nightmares again brx? :P [10:26] Roy: not really nightmares. just *weird* dreams. [10:26] on speed defeating Maelstrom? [10:26] ssd clear? [10:26] datagram: spl - spl - dat clear [10:26] Ah [10:27] first wipe the core with spl [10:27] it ended with me being back in school and PE. teacher saying "from now on your life is 10 fps" [10:27] and we'd be forced to play fifa on playstation [10:27] I actualy have that same thing, my life is perfect, good a nice girlfriend and I'm happy, and then I dream that she does awfull things (like shout/fight/be a bitch etc) but she never does that in real life :-S [10:27] And then throw dat's every 3? [10:27] datagram: no [10:27] datagram: no [10:27] datagram: two complete wipes with spl [10:27] Ah [10:27] datagram: then dats [10:27] so, first clear the core with spl instructions, the again with spl and then dat [10:27] first wipe with spl, then wipe once again with spl, then wipe with dat [10:28] Roy: maybe that means you are actually missing all that [10:28] you need explanation from me too ?;) [10:28] datagram: look at Stargate [10:29] brx: Nah... can't miss that! i think its just my fair of her becomming that, and leaving me in the future somewhere [10:29] fair>fear [10:30] Roy: I dreamed about a girl I know. she fell in love with me and I just went with it. [10:30] I called her by her name. I knew it was her. she looked differently though [10:31] In a dream the concept is more important than things attached to it. But you of course know it already. [10:32] Mizcu: of course I was only daydreaming in my dream. [10:32] how else could it be! [10:32] for god's sake [10:33] Heh of course, I can't imagine suddenly my city is full of mountains and in the middle of the summer they build a snowramp out of woodenbeams with plastic snow just for fun, and after that my girfriend came to watch the new slope and she was evil :-) I know some things are just odd [10:35] once that boy whom I told about tree roots turned into an elephant with two giant poles, walking into the distance and sticking them into the ground before waving back to me, I kind of felt that whatever happened between me and her on the couch in the stables might not have been real [10:36] am I still connected? [10:36] seems so [10:36] what was the last thing I wrote? [10:37] it is just a dream, brx [10:37] to irc that is, to reallife no idea " once that boy wh" [10:37] you know that you can't be connected this long [10:37] Hrm, have to go, lessons [10:38] :( [10:38] will you be here tonight? [10:38] when are we going to do that 94b thing? [10:38] Well, probebly not, have to train :( [10:38] tomorrow? [10:38] Just copy datagrams warrior and what I made from it, done ;-) [10:39] :) [10:39] Maybe tomorrow.. no idea yet [10:39] ok. see you later. [10:39] * Nenad waves [10:39] Have to finish this document tomorrow, but that evening is probebly still free [10:39] MSG: [10:41] Heh [10:41] Did Roy write the last one correctly? [10:41] ? [10:41] Wait, I know what happened [10:41] It went 0% wins [10:41] But I messed one part up I see [10:42] Forgot to tell it to start on line 2 [10:42] :) [10:42] will you send it to 94b now? [10:42] I didn't modify it yet [10:43] I just ran another tournament [10:43] I wanted to see how it did compared ot the other version [10:43] and? [10:43] how did it do? [10:43] I messed up the code, as I said [10:43] So %0 wins [10:43] Heh [10:43] I try again [10:44] What are the standard hill fights set at? [10:44] set at? [10:44] 250 rounds, I think [10:44] if that was what you asked. [10:45] if i remember right, sal does 800 rounds, and as Nenad said, bit over 200 [10:45] I see [10:45] (koth over 200, that is) [10:45] Corewin is much slower than pmars on linux [10:45] Corewin IS slow... [10:45] Is there a built in tournament feature in pmars? [10:46] Well, it's written in Delphi [10:46] I'm not sure... there probably is. [10:46] Mizcu, do you perhaps know if there's some round robin mode for some MARS? [10:47] If not I can write a python frontend to it [10:47] I mostly use corewin to test at first, then optimax to benchmark, so... I don't know. [10:47] Hmm [10:47] Nenad: i dont remember any one having other than corewin [10:48] I think I saw once some patch to enable that.. [10:48] or something like that. [10:48] but I forgot where and what [10:48] MSG: Ping timeout: 252 seconds [10:48] datagram: if you decide to write something, don't do it for pmars. [10:49] Why not? [10:49] there are quicker MARS [10:49] fmars, optimars [10:49] Join: brx joined #corewars [10:49] Ah I thought pmars was the standard used [10:49] pmars is the old one [10:49] sal uses dmars [10:50] (pmars with load distribution) [10:50] Heh [10:50] Is there that much activity that it needs that? [10:50] check few lines backwards, that 800 rounds per match [10:50] 20 match per challenge [10:50] But even still [10:51] You can do that very fast with pmars [10:51] No, it doesnt need that, but fast response rate is nice [10:51] With -v 000 the rounds go by very fast [10:51] How many challenges per day? [10:51] depends [10:52] Is there an interval it goes at? [10:52] Ah yes [10:52] This version he gave me does much better [10:52] using server-version of pmars you get even faster than normal and -v 000 [10:52] exmars is very promising, speed-wise [10:53] From 91.3 pts | 20.7 W | 50 L | 29.3 (old working) to 126 pts | 34 W | 42 L | 24 [10:53] Join: Neogryzor joined #corewars [10:53] hello [10:53] Hello [10:54] hello german [10:54] But when you are brute-force optimizing warriors, thats a lot of battles (say, 100 * number of opponents you test against) times amount of numbers you test (thats 50-7950 times numbers you optimize) [10:54] hi all [10:54] Hi Neo [10:54] MSG: Ping timeout: 252 seconds [10:54] * Neogryzor is trying to repair his web's disaster [10:55] :) [10:55] now, no-one wants to test 7000*7000*7000 numbers, but 1000 tests is a good start. Thats still quite a lot of stuff to crunch. [10:55] Neo: do you know something about doing round-robin tournaments with some mars (without using corewin)? [10:56] isn't there a tournamet scheduler on koth? [10:56] ? [10:57] Is there any magic number for bootstrapping? [10:57] no [10:57] Good deal [10:57] i don't remember hte name. Did you see koth.org tools? [10:57] MTS? [10:57] you can start by trying +1000 and -1000 [10:57] could be [10:58] Corewar Tournament. Allows roundrobin tournaments of multiwarrior battles, created by Phil Knight. [10:58] That looks like it's going to be a windows binary [10:58] oh. yes. I've found it. :). I've downloaded it several months ago, but forgot all about it. :) [10:59] Is it a windows binary? [10:59] I have mts's source here.. [10:59] And Dos-binary [11:00] Boo urns [11:01] Join: brx joined #corewars [11:05] MSG: Ping timeout: 252 seconds [11:08] anyone knows what was and in htm [11:09] anchor [11:09] thanks [11:09] koth [11:12] I have to go. [11:12] * Nenad waves [11:12] MSG: [11:12] * willvarfa waves [11:12] * Neogryzor waves twice [11:17] Join: brx joined #corewars [11:21] MSG: Ping timeout: 252 seconds [11:26] time to go for me [11:26] * Neogryzor waves [11:26] * willvarfa waves three times [11:26] MSG: Quit: Physical Wreck IRC webscript: http://www.pswclan.com/cgi-bin/irc/irc.cgi [11:32] I got my bootstrap working [11:34] Join: |caesar joined #corewars [11:34] MSG: <|caesar> Client Quit [11:35] Join: |caesar joined #corewars [11:35] Join: brx joined #corewars [11:38] As well as the decoys [11:40] MSG: Ping timeout: 252 seconds [11:41] Join: brx joined #corewars [11:41] And it seems to be doing very well against flashpaper [11:41] cool :-) [11:41] Woot [11:42] 151.5 pts | 43.0 | 24.5 | 22.5 vs flashpaper over 200 rounds [11:42]