[00:25] MSG: Ping timeout: 255 seconds [00:25] MSG: Quit: Chatzilla 0.9.77 [Firefox 2.0.0.1/2006120418] [00:28] Join: argiopewe joined #corewars [00:39] MSG: Read error: Connection reset by peer [00:44] Join: argiopewe joined #corewars [00:46] So tell me, why does djn run through the core and put all sorts of dat trash in it? [01:04] MSG: Quit: Trillian (http://www.ceruleanstudios.com [01:20] Join: johnkw joined #corewars [02:05] MSG: Ping timeout: 255 seconds [02:07] Join: argiopewe joined #corewars [02:29] MSG: Ping timeout: 255 seconds [02:35] Join: argiopewe joined #corewars [02:53] MSG: Read error: Connection reset by peer [02:53] Join: argiopewe joined #corewars [03:14] MSG: Quit: Leaving [03:49] Join: argiopewe joined #corewars [03:53] MSG: Read error: Operation timed out [04:51] MSG: Ping timeout: 255 seconds [05:08] Join: argiopewe joined #corewars [05:12] MSG: Ping timeout: 255 seconds [05:56] Join: Mizcu joined #corewars [06:01] Part: Mizcu left #corewars [06:01] Join: Mizcu joined #corewars [06:48] pak21, op [07:00] Join: jkw joined #corewars [07:00] MSG: Ping timeout: 255 seconds [07:06] there, 1 hour to do the labwork plus homework [07:06] and now i got 2 hours extra to spend time on [08:22] Join: ares joined #corewars [08:23] engfeh [09:29] Join: Core_old joined #corewars [09:39] MSG: Quit: Trillian (http://www.ceruleanstudios.com [11:22] Join: fiveop joined #corewars [12:26] Join: argiopewe joined #corewars [12:53] Join: Miczu joined #corewars [12:53] MSG: Read error: Connection reset by peer [12:54] MSG: Ping timeout: 255 seconds [12:54] Join: fiveop joined #corewars [12:56] MSG: Ping timeout: 255 seconds [12:56] MSG: Read error: Connection reset by peer [12:56] Join: argiopewe joined #corewars [13:05] hi argio [13:05] Nick Change: Miczu changed nick to Mizcu [13:08] Stop logging out :-) [13:08] its not my fault univ shell has needed reboot [13:09] Hi Mizcu. [13:09] and i most certainly wont be up 5.50 am to switch ops to another client [13:09] argio: still wondering about the djn-train? [13:09] How the hell did you know I was wondering about the djn-train? [13:09] <_< [13:10] everything on this channel is logged and publically available [13:10] * argiopewe doesn't ever remember asking about it... [13:10] Okay, maybe I did ask about it. [13:10] http://www.koth.org/irc-logs/ [13:10] Yeah, I've read them. [13:11] well, its about the b-decrementing redirection ( < ) which causes it [13:11] djn ofcourse if decreaseandjumpifnonzero [13:11] Right, that is obvious. [13:11] so, first, the < decreases target instruction by one, and then uses it as a pointer [13:11] then the djn itself's decrease occures at the pointer [13:11] I'm currently wondering what the A field does, and what the point of the whole instruction is. To overwrite and destroy? [13:12] Oh, so the a field is the decrease. [13:12] a field is where it will jump [13:12] b-field is where the decrement will be done [13:12] Oh, I see. [13:13] for example, a simple X -times loop can be done with djn.b -somewhere, #x [13:13] And if the djn turns up true (which it should never do,) then it goes on to the dat statement? [13:13] yes [13:13] Okay. [13:14] so you can replace normal jmp with djn to get the same effect plus some damaging effect [13:14] the reduction by one is usually enough to screw up a loop in a warrior [13:14] Another question: How does overwriting the add statement with a dat work? It's obvious that it kills the threads 1 by 1, but somehow they do a core clear at the same time. [13:15] even though dat does nothing, the decrements and increments still happen [13:15] And the threads start up again with no problem, creating an infinite loop. [13:16] and spl #something is equal of spl 0, which creates new processes by itself [13:16] Right. [13:16] so a dat {1, <1 when run will result in dat {1 <1, dat -1, -1 [13:16] Hold on a sec, I need to switch computers... I'll be back in 3-4 minutes. [13:21] MSG: Ping timeout: 255 seconds [13:42] Sorry about the invite only. Wasn't deliberate. [13:42] Join: argiopewe joined #corewars [13:43] Grrrr. Shame on you pak. [13:44] I blame Mizcu. Typo while giving him ops :-) [13:44] or you could just use /op mizcu? [13:44] Now, Mizcu, we were discussing addressing modes... [13:45] Yeah, whatever pak. [13:46] If you've got a problem with me occasionally making a mistake, sorry, but I'm not perfect. [13:46] I don't have problems with misteaks, provided they're well done. :p [13:46] Get Tuc to give people autoops and we'll be fine. Don't bitch at me because I happen to have a persistent connection. [13:46] mm.. steaks.. [13:47] Get Tuc to give people autoops and we'll be fine. Don't bitch at me because I happen to have a persistent connection. [13:47] Oh, this is all going wrong :-) [13:47] Whatever that is? [13:47] no idea [13:47] Is probably better (private). [13:48] What the heck... [13:49] heck, this channel could be moved over to quakenet or undernet anyway [13:49] Freenet... [13:49] My server of choice. [13:51] but, shall we return to regular discussion [13:51] So, back to the discussion at hand... [13:52] Yes, we shall. [13:53] We have now overwritten line 3 with a dat statement. The warrior is about to go haywire and do a core clear. [13:53] Unfortunately, it doesn't appear (to me) to have a reason to be writing data all over the place, especially with the large skip about 40 sectors out. [13:54] what if your opponent was a stone too, with a spl #0 in the start? [13:54] that means that even though single hit might cripple it, it would still survive [13:55] There are only three instructions currently operating in core. spl $0,$0 which duplicates and keeps dwarf from dying, mov which is moving an undetermined sector to another undetermined sector, and dat, which is killing processes spawned by spl. [13:55] Right [13:55] I've seen that with dwarf vs. dwarf. [13:57] So, what is that mov moving, and why is it constantly moving it further through core? [13:58] usually it moves either the same bomb it kept throwing, or some random piece of core (as mentioned before, highly likely to be dat 0) [13:59] Okay. [13:59] the constant moving is caused by the dat-bomb; the dat bomb's >-1 constantly manipulates the b-field of the mov [13:59] Ah, I see. And this is why you have to have the extra dat $0,$0. [14:00] Since it's not technically required in your final dwarf version. [14:00] At least not for the same reason that it was originally put in for... [14:01] And what purpose does the B statement of that dat serve? Just to confuse scanners? [14:02] in that dwarf, only to confuse scanners [14:02] Okay. Whereas in something like a impgate it's very important... [14:04] Now, is there a way to add an imp that runs in parallel with dwarf without a significant slowdown? [14:05] yes and no [14:05] there will be slowdown, but it is usually acceptable [14:05] I tried just a "spl imp" where imp was the regular mov 0, 1 imp. There was quite a bit of slowdown. [14:05] but the point is, that the stone is self-splitting, constantly creating more processes [14:05] and one process onto an imp means nearly nothing [14:05] Right. [14:06] if you take just the original dwarf, they both will have one process and that might be too heavy on the imp-side [14:06] Yeah. [14:07] And also, how would I make one add work for two movs your dwarf. I looked at it, but couldn't find a good way. [14:08] s/two movs your/two movs in your/ [14:08] below the coreclear-part, there are two possibilities shown [14:09] and more downwards, where some stones are represented, there are couple more [14:10] The two directly below the coreclear had me stumped... I didn't have time last night to look at the others. [14:11] And now I'm on a different computer. Could you throw me that link again? [14:11] http://koti.mbnet.fi/mizcu/stonearticle.txt [14:11] Thanks [14:11] thats a weird version i have there, its missing a piece of text [14:11] just noticed [14:11] Hehe [14:12] but i dont have the proper version on this puter, so i cant reupload [14:12] * argiopewe wonders what he missed out on [14:12] its just below where i gave the dwarf a spl $0, $0 [14:13] A) should have an explanation how the self-splitting spl 0 will slow down the rest of dwarf a little [14:13] Yeah, I see A) missing... [14:13] the amount of slowdown is 1/(current amount of processes) [14:13] I think it did mention that somewhere. [14:14] Yeah, it mentions it in B. [14:14] no wonder Metcalf hasnt given me "ok" on a final release [14:15] Is it worth the extra time to throw spl bombs before dat? [14:15] i dont trust spl/dat -throwing [14:15] No? [14:16] either youll throw dats, or spls. (or spl/mov -incendiaries) [14:16] Where would the move send the process? [14:17] That is, what would the mov move. [14:17] the idea of incendiary bomb is, that when hitting an opponent, it will by itself create a carpet of spl's [14:17] so you throw two instructions, but when hit, it will become larger [14:18] So basically you move spl $0 to address and mov -1,>1 to address +1 and hope it gets executed? [14:18] pretty much [14:19] A jmp back to the mov would probably speed it up/take up all the extra processes too. [14:19] And then you could coreclear and all those nice little processes would die. [14:19] * argiopewe thinks he's beginning to get the hang of this... [14:20] well, throwing triple-bombs is a bit of a waste of speed [14:20] (Behemoth throws triple-bombs, but it uses a little different trickery) [14:21] True, since you don't even know whether the bomb will be executed. And if it isn't, then it's a waste. [14:26] Are there any warriors that throw dat $1,$1s out in front to destroy djn trains? [14:30] Whoops, got to go start my day. See y'all later. [14:32] Part: argiopewe left #corewars [17:18] hrm. [17:18] sf has been busy this morning. [17:28] wow.. she even got one of zuls old warriors [17:53] * bvowk points out he's koth with a 2 point lead (this may never happen again, so he thought he better point it out quick!) [18:12] Join: argiopewe joined #corewars [18:12] 'grats on KoTH bvowk [18:13] greets argi [18:13] Howdy [18:15] Refresh my memory here... If I want to split to a command located at what was the command before's 101, I'd spl $100 (note $), right? [18:16] yes [18:17] Hmm, wonder why it isn't working... [18:17] you dont use debug features? [18:18] or view core/process queue in corewin? [18:18] I have a couple. Since this computer is a winsucks machine, I use corewin. It has some nice debugging stuff. [18:21] Okay, I got it. [18:21] Mizcu: Why does the core clear skip a large chunk inbetween about 50 and 90? [18:21] 41-79 to be exact. [18:23] Cool. Now I have an imp that doesn't run too fast, to begin with, but runs at the speed of the coreclear when it's happening. And it runs about 150 memory sectors in front of it. [18:23] i dont see any reason why it should skip [18:23] Of course, it's going to kill the coreclear before it finishes clearing... [18:23] unless you have modified the dwarf into two-add stone [18:23] two-by-one-add [18:24] in which case the skip is relative to the amount of (hop) [18:24] I didn't. It's possibly a bug in corewin. [18:25] show me the stone's source [18:25] i really doubt its a corewin-big [18:25] Doubtful. [18:27] http://pastebin.com/868030 [18:29] Hehe, it scores better against my benchmark list than the original at least... [18:29] 30 points better, in fact. [18:30] And I've got a relatively tough benchmark. [18:31] its a problem in the process-allocation [18:31] Yeah? [18:31] Cool. [18:31] What kind of problem? [18:31] you get a part in the loop where processes are run mov dat dat dat dat dat dat dat dat mov dat mov dat spl [18:32] Strange. How can I fix that? [18:38] it just doesnt die cleany [18:38] cleanly [18:38] but that it not a huge issue IMO [18:38] * bvowk doesn't intend to die cleanly either.. [18:38] I'm gunna make one hell of a mess ;) [18:38] No, it sort of runs into itself. But it should have killed everything by the time it does. [18:39] It will have core cleared everything but the last 200 or so, but the chances of anything important there not being killed by imp are small. [18:40] well, that one gets 100 points on wilkies -benchmark, which is not bad at all [18:40] Yeah, I know. [18:40] I'm running an even tougher benchmark, and it gets 70 something, as compared to 40 something for the original layed out at the end of your file. [18:41] you asked earlier if there are any warriors that throw 1,1 -bombs to harm djn-streams [18:41] the answer is yes [18:42] I'm running Behemot, CrazyShot2, Hazy Lazy..., Iron Gate, Mice, nPaper II, Piper, Reepicheep, Revenge of the Paper, RotF Copy, Silver Talon 1.2, Son of Vain, Thermite 1.0, Uninvited, Vanquisher II, Willow, and WingShot 0.66. It's Wilkies plus a few. It's pretty rough. [18:43] thats an interesting mix of stuff [18:43] Yeah, I like it. [18:44] Wilkies meets Wilfiz with thermite and wingshot (and Mice) [18:44] It really sucks when someone bombs the bomber... [18:45] yes [18:45] Iron gate seems to have a bad habit of doind that... [18:45] Iron Gate is a scanner [18:45] Damn scanner... [18:45] Now they're both frozen at 8000 procecesses. [18:45] Join: sf_ghoul joined #corewars [18:46] jr;;p [18:46] Oh [18:46] I mean hello! [18:46] heya sf [18:46] whats up? [18:46] ko. wpw are upi? [18:46] Have we swapped ops? [18:47] kinda-sorta [18:47] Hi argiopewe :) [18:47] nPaper is really rough against dwarf... [18:47] Hi sf. [18:47] Hey hey bvowkeee! [18:48] Even with the imp... [18:48] nPaper2 is really cruel against some stone/imps [18:48] how's it going sf? [18:48] you've been busy today [18:48] (and gets horribly trashed by certain types of oneshots) [18:48] w00t. I just scored a 129.4. :) [18:48] Oy, you've pushed some of my warriors off :( [18:48] 129.4? beginner hill? [18:48] *AND* I'm koth ;) [18:49] No, against my bookmark. [18:49] benchmark* [18:49] you added more imps? [18:49] Nope, just the one. [18:49] Maybe I just got lucky. ;) [18:50] Mizcu, Think 100 is a good offset? It's the minimum distance between processes on the beginner hill. [18:50] 100 is good enough [18:50] I was thinking a bit more. [18:50] Let's see what happens. ;) [18:51] it might help a little against smoe papers, but i doubt much [18:52] Possible. I'm personally quite impressed with the doubled score on my benchmark just from adding the imp... [18:52] you need heavier imps or spirals for better defence, if you are going to improve the stone further [18:52] Right. I'll be working on a spiral. [18:52] defence is important, just dont take it too zealously [18:53] Yeah. [18:53] I just need to find a good spiral that doesn't slow the bomber down much. [18:53] if you can find the vector launcher, try it [18:54] Vector launcher? [18:54] I'll look into it [18:54] I think I saw it somewhere.. [18:54] you get more points for murder than you do for mere survival. [18:55] True. [18:55] You've got a couple on the bg hill sf_ghoul? [18:56] your glow isn't all that firey SF [18:56] argiopewe: Just one, Bewitching [18:56] It's published [18:57] Oh, bg = beginner, I though bg = big! [18:57] Ah. [18:57] I was just wondering if you referring to me when you said "Oy, you've pushed some of my warriors off :(" [18:58] I guessed you weren't. [18:58] argi: sf and I are battling on nano [18:58] No I was referring to bvowky on the nano hill :) [18:59] Ah. Got it. [18:59] http://sal.math.ualberta.ca/hill.php?key=nano [19:00] Nice hill, very active [19:00] Mizcu: Well, I threw that one up on the beginner hill and it kicked the original down to #18 (from #16) and it's at #14. So, it was an improvement. [19:00] you killed your own warrior sf [19:00] 27th place isn't bad tho [19:00] It gave too many points to yours bvowk [19:00] argio: you can use ;kill oldwarrior at same time when sending a new one [19:00] yeah, nicely done ;) [19:00] I've got the top 3 now ;) [19:01] Mizcu, I'll have to do that. You can't specify more than one, can you? [19:01] Nice one bvowk [19:01] sure you can [19:02] Just put in multiple kill statements? [19:02] yes [19:02] Cool. [19:02] unless bvowk has been too lazy not to implement it [19:02] eh? [19:02] what am I implementing? [19:02] hehe [19:03] Mizcu, About what size is the average paper? [19:04] 8 instructions for the paper, 3 generate processes for it, and ~20 for the qscan [19:05] but thats an silk-style, not ye old-style paper [19:05] Damn. I had dwarf 2 (with imp) and npaper fight 20 times. 17 kills, 3 ties. :( [19:06] And that's npaper killing me, by the way. [19:06] not really surprising [19:06] Yeah, papers are hard to beat. [19:07] hrm. [19:07] sf.. you're not doing too bad.. stop it! [19:08] Does npaper throw dat $1,$1? My djn keeps getting screwed up. [19:08] almost [19:08] mov.i #1, {1 [19:08] Just as bad... [19:09] Nano hof is out of date again! [19:09] well nyaa nyaa [19:10] maybe I should get J to have sal keep the nano HoF itself [19:11] Mizcu, Does changing the djn.b initial value really change anything in terms of winning/losing? [19:11] Join: Core_old joined #corewars [19:12] yes, but not much [19:12] and you might want to try using djn.f [19:12] What will that change? [19:13] yaay me! [19:13] check argio: check the corewin ;) [19:13] Hmmm, yeah, I just showed a marked improvment by changing it. Still didn't win against nPaper, but I had a lot better tie ration. ;) [19:14] I just bought an 8 processor 8GB ram box :) [19:14] Go you! [19:14] ;) [19:15] Join: inversed joined #corewars [19:15] hi inversed [19:15] Hi Inversed [19:15] hi [19:16] aww. [19:16] sf, I *LIKED* 734ce6fe-73c5c938-380503f alot [19:16] You seem to have dropped off nano inversed :( [19:16] holy crap. [19:16] sf is koth [19:16] bvowk: how can you be attached to something evolved?! [19:17] I've been watching the 734ce6fe line for a long time.. [19:17] they're really cute [19:19] What's cute about it? [19:19] its a nice little HSA with a cute bomb [19:20] MSG: Read error: Connection reset by peer [19:20] Join: argiopewe joined #corewars [19:20] Scary how things just happen, eh? [19:20] Especially with those genetic warriors... [19:22] Wow. Nano list changes fast... [19:23] What kind of cute bomb? [19:23] like a mov 0, -1? [19:23] hrm. [19:23] * bvowk gives sf a dirty look. [19:24] :( [19:24] ;) [19:32] 0.4 ties seems low [19:35] Google mail down :( [20:22] hrm [20:22] damn sf.. you're a busy busy little cw'er [20:23] Not doing too well though :( [20:23] little too well I think :P [20:25] MSG: Quit: Lost terminal [20:46] fluffy is almost off completely.. [20:46] he's ranked well so far at least.. [20:47] #9 [20:47] on the beginners hill? [20:48] That's enough for one day on nano [20:48] That's fluffy's best on nano. [20:48] I've leave it to someone else to make Zul happy! [20:48] No, I'm still sitting at #14 on bg. [20:48] where's met these days anyways? [20:49] Stranded in some strange anti-matter universe trying to teach the spoon-creatures about corewar? [20:50] that sounds pretty plausible knowing met.. [20:55] Petro "I'm Old" Warrior [03] Zul Nadzri 151.7 999 [20:55] One more to go ;) [20:57] MSG: Quit: I was using TinyIRC! Visit http://www.tinyirc.net/ for more information. [21:03] Why is there no DJZ command? [21:04] dont know, dont really care [21:04] well, redcode was meant to be minimalistic in its function [21:12] DJN is usefull for loops and DJZ is not [21:15] Right. [21:15] Now, I have a djn loop. It should fall through when it reaches 0, right? [21:16] MSG: Quit: Trillian (http://www.ceruleanstudios.com [21:16] When it's B reaches 0, rather. [21:16] s/it's/its/ [21:17] you can have djn.a or djn.f [21:25] "djn -2,#400" will loop 400 times jumping back to -2 every time until it reaches 0, correct? [21:26] Then fall through? [21:26] it will jump back 399 times and then fall through [21:33] OKay. Have any idea why it's at -500 and it hasn't fallen through yet? [21:35] Join: gnik joined #corewars [21:35] It had something to do with the spl... [21:35] hello people! [21:36] hi gnik [21:36] argiopewe: Do you have spl #smth before djn ? [21:36] Join: fiveop joined #corewars [21:36] Yes [21:36] If so, only one process will fall [21:36] Hello gnik. [21:37] inversed, Okay. [21:41] goin to sleep, bye [21:41] MSG: [21:41] Bye [21:44] bye argio! [21:44] mhmmm, could some explain to me how (tiny) Zooom... works ? [21:44] I'm not going anywhere... [21:46] Arrg I should read more than one line... /tunnelvision ? [21:50] in really, really short: zoom has an add/jmz -scanloop, and when it finds something for the first time, the loop extends to mov/add/jmz [21:50] the mov will keep clearing the place found with spl's, and if something new is found, it is changed to clear the new found place [21:52] And in case of tiny zooom, before changing to clear, the loop expands to mov/mov/add/jmz ? [21:53] well, the problem is that the loop is hard to be made only to extend once [21:53] the top mov-instruction will reset the second extention [21:54] i havent actually seen nanozoom, atleast so i can remember [21:54] but i know the zooom well enough [21:55] do you think a nanozooom is possible ? [21:55] i think met has already done one [21:56] but dat-clear instead of spl [21:57] So what are the requirements for something to be called nano? [21:57] i would really like to see it [21:57] i hope he gets unbanned again [21:58] Why did he get banned? [21:58] nano is coresize 80, warriorsize 5, 800 cycles, 80 processes IIRC [21:58] no one knows [21:58] Mizcu, That's tiny. [21:58] tiny is 800 20 8000 800 [21:58] Got to love assembly. ;) [22:06] Nano Zoom is in the nano infinite [22:08] And i still dont understand what the metcalf-ban is about, no op has issued a ban and there is non on the list [22:09] maybe its the server which refuses connection [22:09] I think he's banned by his GF [22:09] thats my theory [22:10] possible [22:10] MSG: Quit: Lost terminal [22:11] me goes now zzz [22:19] Is it just me, or is crazyshot a vamp? [22:29] got to go, too ... [22:29] MSG: Quit: Chatzilla 0.9.77 [Firefox 2.0.0.1/2006120418] [22:49] MSG: Quit: Leaving