[02:43] * bvowk submits all his reasonable scoring warriors that contain 'dead' in the name.. [02:46] hrm.. [02:46] a few beefs [02:46] no deadbeefs [02:46] no coffees [02:48] er.. c0ffee [11:49] MSG: Read error: Operation timed out [15:18] Join: MixolAtWo joined #corewars [15:19] Nick Change: MixolAtWo changed nick to Mixol [15:20] MSG: Client Quit [15:21] Join: Mixol joined #corewars [15:21] Morning. [15:24] Join: bvowk joined #corewars [15:27] hrm. [15:28] I guess the biggest question I have as a newbie to this is...where do I find the time to write and test assembly language code warriors [15:28] Cause...wow... [15:28] actually.. most of the players, it doesn't seem to take them very long. [15:28] and where are some handy tools besides just the basic pMars [15:28] metcalf for example seems to be able to wing them out from cocktail napkins a dozen at a time.. [15:29] you know any scripting languages? [15:29] wow, craziness [15:29] perl, php, learning groovy [15:29] meeting at work, bbiab [15:30] perls good.. I've got some perl utilities I used for benchmarking and the like if you want them. [15:45] might come in handy, tho i don't think i have a linux box running at home atm [15:46] need to make myself a VM or something [15:47] I'm a java developer by trade, with a BS in comp sci. I was probably one of 30 students or so that kind of enjoyed the assembly class [15:47] but that's mostly cause i learned to use the debugger and tons of paper well [15:48] perl/python should both run on windows... [15:57] yeah, i can get active state or something, it's just not as friendly without a real shell and command line interface [15:57] what do you use for editing the actual warriors? any old plain text? [15:58] I don't write my own warriors.. [15:58] I'm an evolver.. [15:58] I write programs to write my warriors [15:58] but I use emacs to edit them by hand [15:58] its just text. [15:58] genetic algorithms? [15:58] yeah [15:59] http://www.math.ualberta.ca/~bvowk/corewar.html [15:59] yeah, i was just curious if anyone had written any nice syntax highlighting/debugging/editing tool [15:59] I've never used it.. but ares has written something like that.. [16:00] http://harald.ist.org/ares/ [16:00] I would be interested to know what you thought of it.. [16:00] I'm sure he would too [16:00] whoa, looks pretty sweet [16:00] the screenshots look scary awesome. [16:01] indeed [16:04] i wish i could actually grab some of this while i'm at work instead of occasionally looking at the web sites [16:13] looks like you've had a lot of nano success with the evolved guys [16:14] he just spams us with having up to 100 times the calculation power than us [16:14] (well, only about 5 times, if he wants to keep hes job) [16:18] heh, yeah. if you submit 80% of the warriors on the hill, some of them have to stick to the wall, right? [16:18] right [16:20] but spamming hills is considered bad behaviour [16:20] of course [16:21] i take it the nano hill isn't that popular, so noone really minds vowk's experiments on it? [16:23] how many active code warrior writers are there these days? [16:26] actually nano hill is the most active currently [16:26] theres about a dozen of corewarriors currently [16:26] and another dozen lurking [16:27] not bad [16:29] i guess SAL's beginner hill is the best place to get started, eh? [16:30] yes [16:31] do they mind if I start with some "open source" warriors and edit from there? [16:31] the nano is hill is pretty active, and spamming is largely tolerated because 5 lines aren't a major time investment, and the hill is big and fluid.. [16:31] Mixol: just dont choose too new warrior to experiment with [16:32] right, is there a good aged archive of beginner hill appropriate warriors? [16:32] the 94nop hill is totally different however.. the warriors there are largely polished gems [16:32] with serious time invested in the good ones.. [16:32] i thought nop would be simpler since the lack of pspace is just one less thing to worry about? [16:33] Mixol: www.koth.org/planar/ [16:33] when using pspace, you can prepare for all types of opponents, but on 'Nop you have to be strong again all types of [16:33] well, 5 lines isn't really enough space for pspace.. so there's really no pspace on nano. [16:33] and its small size means you can run huge numbers of battles.. [16:34] 94nop is also more difficult because the warriors have room to be really solid. [16:34] nano warriors tend to be wickedly violent trashers of core.. [16:34] 94nop warriors tend to be a little less mess [16:34] y [16:35] gotcha [16:35] are there nano samples out there? besides dwarf and imp and MICE? [16:35] although, MICE is probably too big actually [16:39] if you want nano, there is the nano-infinite (search if www.corewar.co.uk ) or koeningstuhl ( http://www.ociw.edu/~birk/COREWAR/ ) [16:41] neat [16:42] just not really sure how to get started and how to find appropriate beginner warriors to study and work with [16:42] the archive page helps, but it's pretty big [16:44] wow, jEdit seems to recognize the redcode syntax [16:48] www.koth.org/wilkies/ [16:48] yep, found them. pretty handy [17:10] Join: Fluffy joined #corewars [17:10] :) [17:10] Topic Change: Fluffy sets topic: http://corewar.co.uk - http://www.corewar.info - Official "Hug bvowk" day today [17:10] * Fluffy hugs bvowk ;) [17:12] Hi Mixol [17:13] Mixol: I've just skimmed the logs. If you have any questions, don't hesitate to ask. [17:16] Mixol: I've found, that Neverland ( http://www.ociw.edu/~birk/COREWAR/94/HILL/neverland.red ) and Neverland II ( http://www.ociw.edu/~birk/COREWAR/94/HILL/neverland2.red ) are nice warriors, if you want to study some easy scanners. [17:26] Topic Change: bvowk sets topic: http://corewar.co.uk - http://www.corewar.info - Official "Worship bvowk" day today [17:26] :P [17:27] I'm not a damn teddy bear. [17:27] bvowk: I usually kill anybody, who I worship. [17:27] foolish human ;) [17:28] I still wait for a good nano-warrior from you :) [17:28] uh huh. [17:29] hrm. [17:29] I'm not getting any computrons on the one group of machines right now.. [17:29] they're doing *REAL WORK* [17:29] *sigh* [17:30] Aren't ned and 3jane enough? [17:30] fluffy, you've known me long enough to know its *NEVER* enough. [17:30] and not to forget mitzi [17:30] I could have a solid wad of computronium the size of texas, and I'd still want more. [17:31] (but I'd likely be doing pretty good on 94nop with that) [17:31] Wouldn't you first want to find a 16-hint-sudoku by enumerating (and checking) all possible sudokus? [17:31] heh [17:32] it would take a fair number of stor-a-trons to hold that.. [17:32] * bvowk eyes delaware. [17:33] You only need to check them and not store each and every possible sudoku. [17:34] whats the point if you're not going to keep all the sudoku!? [17:34] HMMMM ... ketchup with pizza! [17:34] you might get bored waiting for the end of everythign. [17:35] end? There won't be an end [17:35] sure there will [17:36] it might be a big crunch, it might be a big sigh, it might just be a slow heat death.. [17:36] but it'll end [17:37] The fun thing is that you can't tell :) [17:37] I still have the tiny hope that somebody will invent a Perpetuum mobile [17:38] * bvowk will stick around long enough to make sure fluffy doesn't escape destruction in some silly pocket universe or something. [17:38] I know that current physics tell us that you can't, but fortunately they don't work with real laws but only nicely fitting suggestions [17:39] I think physics is going to change quite a bit between now and then [17:40] As long as the hurry up and explain how beaming works, I'm satisfied [17:41] * bvowk stuffs fluffy in a transporter buffer, and then uses a klingon disruptor to wipe the buffer [17:41] How do you know that I don't continue to exist. [17:41] ? [17:42] its irc.. [17:43] sadly, you cannot be destroyed through textual means. [17:43] * Fluffy places a strangelet near bvowk [17:43] it takes a plane ticket, a duffel bag, duct tape, a chainsaw, and a bag of concrete mix to do that. [17:44] * bvowk hurls the strangelet at fluffy and yells "catch!" [17:44] bvowk: You forget the water! [17:44] *forgot [17:44] you don't have water over there? [17:45] Depends on your location "over here". [17:46] Strange, but I have the urge to hum "Blame Canada" [17:49] hey fluffy.. [17:49] www.math.ualberta.ca/~bvowk/files/jnsl-1.1.tgz [17:50] what did you change? [17:50] now mit PyCorwar! [17:50] ah, ok :) [17:50] and some docs [17:50] You don't have permission to access /~bvowk/files/jnsl-1.1.tgz on this server. [17:51] eh? [17:51] try again? [17:51] ok now [17:53] What does jnsl is for? [17:53] what does it stand for? [17:53] yes [17:54] nothing at all. [17:54] just hit my hand on the keyboard to get the name [17:54] you've hit that Kinesis keyboard? [17:54] Bad bvowk! Very bad, bvowk! [17:55] lol [18:04] Hrm ... only 3rd on nano :( [18:04] I should kill it [18:05] yes, you really better.. [18:05] nobody wants to see that much of a sad ass warrior.. [18:08] hmm ... SAL didn't execute the kill [18:08] sending again [18:10] ok, now it did work [18:11] bvowk: Just got a mail from HWMNBM. He said it would be ok, if I'd lay my hands on SAL's code. He'll mail it in a couple of days [18:14] heya Fluffy, thanks for the offer to help. [18:14] ;) [18:15] sadly I'm at work right now and just kinda stumbled upon the wiki page and related resources for core wars, so i can't really mess around with it right now [18:15] Take your time. [18:15] Yeah, it's pretty interesting stuff, tho. [18:16] A lot easier to work with than Brainfuck, that's for sure. [18:17] I prefer Malbolge to Brainfuck ;) [18:17] which one is that? [18:18] http://en.wikipedia.org/wiki/Malbolge [18:24] holy crap [18:24] Yes, I still haven't managed to write my first own program [18:26] that's just pointlessly esoteric, tho, at least brainfuck makes some sense and resembles real CPUs in some sense [18:28] Computers are fast enough (although bvowk does claim otherwise ;). So there's no need to use languages, which make sense or resemble anything. [18:29] ... or are fast [18:30] bah. [18:30] don't listen to fluffy.. his head is made of 60% polyester filling. [18:31] Would he puff up in a microwave? [18:31] No, I'd merely scream and die. [18:32] (not necessarily in that order though) [18:32] mixol: there's only really one way to find out for sure.. [18:32] * bvowk eyes fluffy evilly! [18:33] he only puffs up near things he is allergic though [18:33] cant remember one right away [18:34] I've allergic to beer. [18:34] Hehe. Indeed. Computers really aren't that fast, tho. There's still a couple 200+ digit RSA numbers to crack. [18:34] Join: rnd0 joined #corewars [18:34] * Mixol would commit suicide if he were allergic to beer. [18:34] Nick Change: rnd0 changed nick to elkauka [18:34] you just need a more intelligent algorithm [18:35] you can't really be allergic to beer.. [18:35] aren't you german? [18:35] Hi elkauka [18:35] wouldn't that violate something fundamental about your germaness? [18:35] bvowk: Yes, I am, but fortunately it doesn't require me liking beer [18:35] ocktoberfest! [18:36] Yeah, but writing in that language isn't going to help you learn to write more efficient algorithms like BF or even redcode might. [18:36] germaness? stupid and obese? [18:36] hi@all [18:36] Mixol: I use C for that :) [18:36] I would have thought they'd escort you to the nearest border and then you to never return or something.. [18:36] fluffy leaks memory tho.. [18:36] you have to watch him like a hawk ;) [18:37] memory? I only have one brian cell, which continously make me repeat my name in order not to forget it [18:37] *brain [18:37] *makes [18:37] heh, my name is Brian, what an amusing coincidence. [18:38] feels like ages the last time i been here -> off checking the hills [18:38] i have a friend that leaks electricity, everything electronical near him has 10 times the normal to blow up [18:38] elk: it *HAS* been a long while [18:39] you missed the crushing of nano [18:39] ? [18:39] and hate bvowk day [18:39] elkauka: Don't believe him a word. He can't even kill one of my weaker warriors :) [18:40] well fernandes leads with a gap off 5 points... [18:41] yes :) [18:41] http://sal.math.ualberta.ca/hill.php?key=nano&t=6684&how=late&sortby=author&dir=asc [18:41] the humans with only 7 warriors. [18:41] ... which you can't kill [18:41] I should say "which you aren't able to kill" :) [18:42] fluffy: thats not true. petro died. [18:42] Which proves my point. I've killed it :( [18:42] did mirco gp evolve their warrios too? [18:42] yeah [18:42] yes, I think so [18:43] they're got some seriously slick software [18:45] who/what is micro gp? [18:45] couple Italians [18:45] made an evolver quite a while ago [18:45] they're a research group that uses a GP/GA system to write code [18:45] some kind of evolver, which tries to identify useful pieces of code, which then are used to evolve new warriors [18:46] http://www.cad.polito.it/research/microgp.html [18:46] i dont remember reading about microgp's evolver identifying pieces of god [18:46] how's it different from yours bv? [18:46] the current one is painfully stupid.. [18:47] no, that's you, bvowk ;) [18:47] I have no idea why it works as well as it does. [18:47] the previous one was a big distributed computing monster [18:47] this ones like 50 useful lines of python and some cruft [18:47] Will's evolver was capable of identifying pieces and using them, while still socking [18:48] while still what? [18:48] http://sal.math.ualberta.ca/hill.php?key=nano [18:48] Fluffy: being crap [18:48] damn.. only third too fluffer. [18:48] * Fluffy kicks bvowk [18:49] well thats not very nice. [18:51] what's the difference between scan and qscan? [18:51] qscan has unrolled loop [18:51] qsacan basicly runs ones [18:51] a qscan only scans a few locations, but is very quick [18:52] and is used before the real warrior start up [18:52] *starts [18:52] so you scan every nth location in memory for something other than data and then roll into your real warrior? [18:53] no, of course you hit everything, that you find, very hard [18:53] and a regular scan just scans every location sequentially? [18:53] because it has a very high chance of being your opponent's code [18:54] yes, a "normal" scanner scans and drops only one (or some few) bombs at positions, where is finds sth. [18:54] hey w00t. [18:54] evolved nano-paper [18:54] o_O [18:54] qscan goes through a lot of memory and remembers multiple locations to bomb then? [18:55] no, just the first [18:55] no, very little memory gone through [18:55] which is immediately attacked with lots of bombs [18:55] man.. [18:55] well not immediately, the is a small phase of decoding the position though [18:55] my nano-paper is *NOT* anywhere near getting on the hill :( [18:56] bvowk: nyaa nyaa! [18:56] During the first instructions of a battle your oppoents didn't have much time to copy itself around. So IF your quickscan finds sth., it is very likely to be your opponent. [18:57] That's why you only hit the first found location [18:57] heh fluffy, what was the url of your webiste again? [18:58] finds sth.? [18:58] something to hit? [18:59] just finds something [18:59] since, as said before, you find something non-dat, it is suspicious and thus deserves to be nuked [19:00] Mixol: http://groups.google.com/group/rec.games.corewar/browse_thread/thread/19dece9fe3a3dc37/20bd4a144c283a20?lnk=gst&q=Quickscanners&rnum=3#20bd4a144c283a20 might be interesting [19:13] Topic Change: Fluffy sets topic: http://corewar.co.uk - http://www.corewar.info - Official "Hug bvowk" day today [19:27] Join: pkhuong joined #corewars [19:27] * pkhuong hugs bvowk [19:27] Hi pk :) [19:27] * Fluffy hugs bvowk [19:27] MSG: Client Quit [19:28] lol [19:29] I wonder, whether pk could actually walk over to bvowk and hug him irl ;) [19:30] if he had big enough boots and a backpack full of food [19:31] it's a shame the lexicon on corewar.info isn't more filled in [19:33] Mixol: Do you want to know anything special? There some nice texts here and there, which aren't linked on corewar.info [19:34] in this particular case how an imp-gate actually works, but there's no one nice glossary that i've found, it's all kinda scattered [19:35] well ... imp-gates are easy [19:35] Let's do a battle of IMP (mov.i $ 0, $ 1) against GATE (JMP $ 0, < -10) [19:35] you probably already know how an imp works, right? [19:36] right [19:36] k [19:36] the crucial part of the imp is the ", $ 1" [19:36] if that is changed, the imp dies in the next cycle [19:36] oh, cause it copies the move to some other location in memory [19:37] yes [19:37] and hits a dat instead of the copied MOV [19:37] exactly [19:37] so an opponent "only" needs to change the B-field of an imp [19:37] the problem is, that the imp isn't stationary [19:37] so you have to hit it at the right position [19:38] ... or wait for it until it moves to the right position [19:38] and that is, what the gate does [19:38] an "JMP $ 0" does nothing, but wait [19:39] an "JMP $ 0, < - 10" does nothing, but wait AND continously change the B-field of the location -10 [19:39] wow...that's neat [19:39] and when an imp comes along, its B-field is changed [19:39] and is dies :) [19:40] so hopefully hit your gate more often than the imp coming up on you is moving, or you just get a little lucky [19:40] huh? [19:41] a "JMP $0, < -10" always wins against a simple B-imp (mov.i $ 0, $ 1) [19:41] hence it is called a perfect gate [19:41] but only if the JMP is executed every step [19:41] yes [19:41] if it's only executed every 4 steps or something, it's less effective [19:42] yes, but every 4 steps is still good [19:42] because you a) shouldn't aim to kill you opponent 100% of the time and b) don't specialize in only ONE opponent [19:43] So if you have a free B-field (say in a JMP or a SPL), i.e. a B-field, which you don't use, then add a gate :) [19:43] right [19:44] i imagine there are tons of these little optimization tricks that are hard to learn just from looking at code [19:45] yes, but you have to start somewhere [19:45] yeah [19:45] Once you have your first working warrior, you'll grasp those tricks really fast [19:48] and the lack of stuff on the lexicon is not that bad, if you have irc-access [19:49] considering we are too easy to bug with silly questions that way [19:51] heh, well i do appreciate all your niceness [19:51] i've got a ton more guides to go through too [19:56] read one and start to code. that's far more fun than reading all 10 guides [19:58] Hey, the code of Shamu is on Koenigstuhl, but it doesn't very well [19:58] probably its constants are changed [20:21] Shamu? [20:22] koth on '94nop ( http://www.koth.org/lcgi-bin/current.pl?hill94nop ) [20:28] not surprising [20:28] 'tis just a oneshot [20:28] what's a oneshot? [20:28] scan into multi-phase coreclear [20:28] it searches for something, and then starts the coreclear on the place where something is found [20:29] easy to make, and efficient, but either lack the survivability or the punch to be great [20:32] neat [20:33] not, if they beat your warrior ;) [20:35] not my issue ;) [20:35] Wait until my new paper is ready [20:36] stil not an issue to me, since technically that is more Inversed's work than mine ;) [20:37] hehe [20:37] i just pointed out something interesting to him [20:38] I guess, that you want me to ask what that is and then say, that you can tell, right? [20:38] yaay! [20:38] my new giant sun box is coming! [20:38] giant equals what? [20:38] Fluffy: the comment says enough about about the warrior, ill keep silent until it is Koen'ed [20:38] 8way 8GB [20:39] ebay? [20:39] yup [20:39] faster than mitzi? [20:39] no, parts for mitzi [20:39] k [20:39] bvowk still thinks bigger is better [20:40] bvowk: how many cpus can mitzi max. have? [20:41] 14? [20:41] 16? [20:41] 14 [20:42] with 14 GB, right? [20:42] its got 16 slots, but sadly, one has to be filled iwth an IO board for it to be of much use as a computer :) [20:42] yup [20:42] sorry.. its got 8 slots that could hold cpu boards.. [20:43] Join: Roy joined #corewars [20:43] Hi Roy [20:49] Hi [20:49] Just got a email from (translates into ) hehe.. I have a mystery to solve :-) [20:49] The first clue was '24' [20:51] it is probably a mistake and it was meant to say '42' ;) [20:51] That would be a good answer ;-) [20:51] The odd thing is that the originating IP comes from the city I work in currently, but I didn't suspect anything from them :-S [20:55] Hmm.. mysterious :( [20:57] How do I find the answer.. I tried login into that email account, even tried to answer the secret question [20:58] you just aint haxor enough [20:59] Probably :-) [21:00] Join: codgeriff joined #corewars [21:01] hi [21:01] Hi! [21:01] I saw miscu's note on Shamu and it is not right. [21:02] Shamu is a boot-picker. It does a coarse scan, assumes what it finds is your [21:02] boot code [21:02] and then uses your fields as pointers to your runnning code. If you [21:03] leave your boot pointers unprotected it will normally stun you fast then [21:03] kill you later. [21:03] Does it bomb 1 position indirectly? [21:04] Or is it more like Blurrish scanner that it bombs continues? [21:04] the code is here: http://www.ociw.edu/~birk/COREWAR/94/HILL/shamu.red [21:04] got to go [21:04] * elkauka waves [21:04] * Fluffy waves [21:04] Ah, lets have a peek then :) [21:05] Part: elkauka left #corewars [21:05] Ah its a oneshot..! [21:05] check the leach and lLoop parts [21:05] Actually, my idea isn't very bad.. a continues spl-clear that uses the ptr it finds :) [21:06] when it finds something it backs up 100+ locations and sequentially scans [21:06] forwards until it finds the start of it. [21:06] Nice, I like the jmz #0 line hehe :) [21:06] then it goes into leach mode where it bombs through your a-fields [21:06] with a spl-carpet [21:07] The slt lines check it points inside a warrior or something? why is that? [21:07] after 102 locations it goes into Geist [21:08] the slt lines prevent it from bombing ahead into your boot code and [21:08] maybe overwriting your boot pointer. [21:09] actually the second slt line keeps it from uselessly bombing back [21:09] at code it has already inspected. [21:09] Its nice, I tried something similair with a qscanner once, it searched the qscan code and went to the last line (which points to the warrior) [21:09] But it was too slow to be effective, nice excersise though [21:10] shamu is selectively effective, real good if you have unprotected a-field pointers [21:10] like inversed's grunts :-) [21:11] after that Geist is still a pretty good program [21:13] :) [21:13] Yeah :) You should try it with Myrmidon ;-) its a improved version of Geist [21:16] they are not that different after the oneshot... [21:18] Froth and Fizzle is also on Koenigstuhl, see http://www.ociw.edu/~birk/COREWAR/94/hill10.html [21:18] (...pats self on back...) [21:19]