2010-11-03T00:00:26 his routing code is broken 2010-11-03T00:00:34 he routed defense through his other planets which took longer 2010-11-03T00:01:54 *** davidjliu has quit IRC (Quit: Page closed) 2010-11-03T00:04:54 *** fmeyer has joined #aichallenge 2010-11-03T00:07:00 *** hellman has joined #aichallenge 2010-11-03T00:09:24 *** hellman has quit IRC (Remote host closed the connection) 2010-11-03T00:09:31 *** Rubicon-|-Cross has joined #aichallenge 2010-11-03T00:09:45 *** hellman has joined #aichallenge 2010-11-03T00:12:24 sigh: that was an or question, not a do you? :P 2010-11-03T00:12:36 well a do you do a or b, not do you do a 2010-11-03T00:13:27 Top 10 players: bocsimacko(4409), dmj111(4232), jimrogerz(4226), Hazard(4169), shangas(4140), george(4132), bix0r4ever(4126), Accoun(4091), FlameN(4075), felixcoto(4068) 2010-11-03T00:14:30 o_O 2010-11-03T00:14:34 augh my logger keeps crashing everything >:( 2010-11-03T00:14:41 how dumb is that 2010-11-03T00:36:58 *** McLeopold has joined #aichallenge 2010-11-03T00:50:25 raedwulf: do you have a trivial c logger or fprintf macro 2010-11-03T00:50:40 somehow vc++ has managed to make FILE ambiguous 2010-11-03T00:51:49 *** virdo has quit IRC (Ping timeout: 255 seconds) 2010-11-03T01:00:23 Has anyone used git and found it nice? 2010-11-03T01:02:12 demerzel_: yeqa 2010-11-03T01:02:17 two sec 2010-11-03T01:02:22 was asleep 2010-11-03T01:02:24 Azrathud: yea 2010-11-03T01:02:33 im using it for my bot :P 2010-11-03T01:02:37 lol np i am trying out a pseudo-boost one 2010-11-03T01:02:41 oh ok 2010-11-03T01:02:48 cxxtl 2010-11-03T01:02:55 ^^ 2010-11-03T01:02:57 yea go boosty 2010-11-03T01:03:01 thats better anyways 2010-11-03T01:03:08 if you do c++, do boost 2010-11-03T01:04:10 Do they have boost on the contest server? 2010-11-03T01:04:19 no 2010-11-03T01:04:28 yes, only the headers 2010-11-03T01:04:35 *** JamesGecko_ has quit IRC (Ping timeout: 265 seconds) 2010-11-03T01:04:37 they don;t link the libraries afaik 2010-11-03T01:04:39 ah, that's most of it anyways 2010-11-03T01:04:45 well you can bring the headers over manually 2010-11-03T01:05:01 fair enough 2010-11-03T01:05:05 i have nullstream for example 2010-11-03T01:05:08 TravisD: I removed all boost code once for the contest. It was a pain. 2010-11-03T01:05:11 a lot of boost is just clever macros hehe 2010-11-03T01:07:57 Frontier: why did you remove all boost code? 2010-11-03T01:08:28 well give it a test, just include one boost header and upload the code 2010-11-03T01:08:30 :P 2010-11-03T01:08:53 I'm using shared_ptr and foreach successfully 2010-11-03T01:09:00 for what it's worth 2010-11-03T01:09:30 you might be able to do a #pragma lib for linking libraries (i don;t know if it works with g++ tho) 2010-11-03T01:10:25 no there isn't 2010-11-03T01:10:26 sigh: I was only using foreach. Not any longer though. 2010-11-03T01:11:01 Frontier: but why did you remove it? 2010-11-03T01:12:20 *** yasith has joined #aichallenge 2010-11-03T01:13:27 Top 10 players: bocsimacko(4399), dmj111(4247), jimrogerz(4245), Hazard(4182), shangas(4146), george(4128), bix0r4ever(4113), Accoun(4083), felixcoto(4076), FlameN(4065) 2010-11-03T01:14:17 raedwulf:i'll take whatever you got this is bloated and doesn't do much useful shit for it 2010-11-03T01:15:56 Iirc I didn't want to run into problems when I upload the code. Someone said, it might not be supported. 2010-11-03T01:16:14 foreach in c++ the gods will not be pleased 2010-11-03T01:16:41 I looked through my installed packages and found boost was only a dependency of some exotic packages so I didn't bother trying on an ubuntu server 2010-11-03T01:17:00 all i wanted to do was have a global FILE* 2010-11-03T01:17:03 fair enough 2010-11-03T01:17:10 i can't even do that argh 2010-11-03T01:17:45 demerzel_: why not? 2010-11-03T01:17:56 vc++ has redefined it twice and yells about ambiguous 2010-11-03T01:18:05 :( 2010-11-03T01:18:07 then if i try to build while it's got the defs open it throw 200 other errors 2010-11-03T01:18:16 which is amusing and sad 2010-11-03T01:18:55 oh maybe i'll just use a simple ofstream can that be one-lined 2010-11-03T01:19:02 I just use ofstream for logging, wrapped up in some macros 2010-11-03T01:19:15 pastebin it i'll use it 2010-11-03T01:19:19 please :) 2010-11-03T01:19:27 i was looking for some macros earlier but it's a hard google term 2010-11-03T01:19:53 sure, when I get home... I haven't got the code with me atm 2010-11-03T01:20:20 2 or 3 hours time 2010-11-03T01:23:50 oh ok 2010-11-03T01:23:53 ty :) 2010-11-03T01:24:01 demerzel_: brb, ill get it to you soon 2010-11-03T01:24:19 i'll probably be gone here in a bit but will read logs 2010-11-03T01:24:25 dev on a laptop haha 2010-11-03T05:42:11 *** contestbot has joined #aichallenge 2010-11-03T05:42:12 *** ChanServ sets mode: +o contestbot 2010-11-03T05:42:25 *** jaspervdj has joined #aichallenge 2010-11-03T05:44:01 oh boy, i got to play george on my favourite map 2010-11-03T05:44:03 and i kicked his ass 2010-11-03T05:44:18 McLeopolds favourite map too 2010-11-03T05:44:33 contestbot: hello, you are back! 2010-11-03T05:44:34 sigh: I have no idea what you mean. 2010-11-03T05:44:52 that's the spirit 2010-11-03T05:45:06 antimatroid: what is your favourite map? 2010-11-03T05:45:34 7 2010-11-03T05:45:45 taking a comeback against flavor again, but he'll do it 2010-11-03T05:45:53 any reason why? 2010-11-03T05:45:56 http://72.44.46.68/canvas?game_id=834660 2010-11-03T05:46:09 it's a quick win for whoever gets it 2010-11-03T05:46:13 i test on that and 44 mainly 2010-11-03T05:46:16 79 if i want a long game 2010-11-03T05:46:40 ah, i see what you mean 2010-11-03T05:47:38 12th on tcp is a bit more satisfiable 2010-11-03T05:47:41 it's still not first :( 2010-11-03T05:48:10 bah, i just made a stupid move against george 2010-11-03T05:49:13 is george anyone in here? 2010-11-03T05:49:26 don't think so 2010-11-03T05:49:37 i think i'm 2 for 1 in the last hour against him 2010-11-03T05:49:43 :) 2010-11-03T05:50:31 oh god, i have exams saturday, monday and tuesday 2010-11-03T05:50:49 then undergrad is done and dusted 2010-11-03T05:51:50 Error323: :P 2010-11-03T05:52:34 =] 2010-11-03T05:53:25 this ones closer 2010-11-03T05:53:35 i think i have it under control though 2010-11-03T05:55:48 lol, my bot takes 0 growth planets 2010-11-03T05:59:47 here's a familiar pain, obligatoryshampoo 2010-11-03T06:00:55 *** Utkarsh has quit IRC (Ping timeout: 240 seconds) 2010-11-03T06:07:08 *** Utkarsh has joined #aichallenge 2010-11-03T06:10:31 3 loses to shampoo, i hate him so much 2010-11-03T06:14:45 Top 10 players: bocsimacko(4288), dmj111(4208), Hazard(4165), bix0r4ever(4153), shangas(4127), george(4094), Accoun(4071), felixcoto(4059), GreenTea(4038), rsergio(4015) 2010-11-03T06:15:29 antimatroid: fight back with some conditioner 2010-11-03T06:19:25 *** p4p4 has joined #aichallenge 2010-11-03T06:23:06 d0ugal: isn't that meant to reinforce the work of shampoo? 2010-11-03T06:24:07 antimatroid: hmm, true. we are all screwed then. 2010-11-03T06:26:31 So what language are people in here using to compete? 2010-11-03T06:27:14 just take a look at a leaderboard 2010-11-03T06:27:27 c++ 2010-11-03T06:28:01 *** Apophis has quit IRC (Read error: Connection reset by peer) 2010-11-03T06:28:07 *** Blkt has joined #aichallenge 2010-11-03T06:28:26 *** Apophis has joined #aichallenge 2010-11-03T06:28:42 good day everyone 2010-11-03T06:29:20 Naktibalda: True, but I don't know who of them are in here. 2010-11-03T06:31:21 'spose I was just wondering if anybody else here was using Python 2010-11-03T06:32:54 If you ask about 70 people who are here then an answer is certainly yes, but there is a chance that they wont respond :) 2010-11-03T06:49:25 *** wh1teside_ has joined #aichallenge 2010-11-03T06:52:43 *** wh1teside has quit IRC (Ping timeout: 252 seconds) 2010-11-03T06:54:04 *** Appleman1234 has joined #aichallenge 2010-11-03T07:00:29 *** JamesGecko_ has joined #aichallenge 2010-11-03T07:03:19 *** aerique has quit IRC (Quit: ...) 2010-11-03T07:05:15 *** JamesGecko_ has quit IRC (Ping timeout: 252 seconds) 2010-11-03T07:14:46 Top 10 players: bocsimacko(4266), dmj111(4189), Hazard(4157), bix0r4ever(4154), shangas(4115), george(4096), felixcoto(4057), Accoun(4054), GreenTea(4025), FlameN(4006) 2010-11-03T07:17:51 *** songoftheweave has joined #aichallenge 2010-11-03T07:18:05 * songoftheweave waves 2010-11-03T07:38:11 hah nice 2010-11-03T07:38:15 #31 on official 2010-11-03T07:42:21 bhasker: cool 2010-11-03T07:42:32 i'm doing muuuuuuch better on tcp aswell with 16.0 2010-11-03T07:42:33 :D 2010-11-03T07:42:53 so when i upload it i expect to do much better :) 2010-11-03T07:42:57 just under you somewhere :P 2010-11-03T07:43:23 that is, if malan*15 is ur current on ranking server 2010-11-03T07:43:47 no 2010-11-03T07:43:52 its 0.13 or 0.14 2010-11-03T07:43:58 though 0.15 is not much different 2010-11-03T07:44:18 so i would expect it to be close to the official 2010-11-03T07:44:52 cool 2010-11-03T07:45:11 cuz i've been running 16.0 all night and its _much_ better than 15.3 (current official) 2010-11-03T07:45:19 *** Blkt has quit IRC (Ping timeout: 260 seconds) 2010-11-03T07:45:46 nice 2010-11-03T07:46:43 that shampoo guy is a real pain btw :P 2010-11-03T07:46:57 just can't win consistently from him 2010-11-03T07:47:35 haha yea 2010-11-03T07:51:49 bam, i handed iouris ass to him in 50 moves 2010-11-03T07:52:25 mmm i'm kinda spamming that server with bot experiments lol 2010-11-03T07:52:40 * raedwulf needs to complete his first bot experiment xD 2010-11-03T07:52:57 i often use my second, so i can only run one bot at a time 2010-11-03T07:52:58 this is my first ever piece of AI 2010-11-03T07:52:59 iouri has so many versions running 2010-11-03T07:53:19 no, i lie, i did a connect 4 game some years ago 2010-11-03T07:53:19 the tron one was mine 2010-11-03T07:53:31 * antimatroid is just an undergrad 2010-11-03T07:53:32 but it had bugs 2010-11-03T07:53:35 same here 2010-11-03T07:53:47 3rd year undergrad 2010-11-03T07:53:53 i only learnt to program once i got to uni :P 2010-11-03T07:54:02 4th year undergrad, finish my last undergrad exam tuesday 2010-11-03T07:54:03 well I study AI 2010-11-03T07:54:09 so i should know a thing or two :P 2010-11-03T07:54:19 i plan to do a phd in game theory :P 2010-11-03T07:54:50 i would love to do it through a maths department, but i can't find one anywhere that actually does game theory 2010-11-03T07:55:04 i do artificial intelligence for games in the spring term 2010-11-03T07:55:06 Is anybody using a statistical approach? 2010-11-03T07:55:19 nope, too hard to define anything 2010-11-03T07:55:22 not me 2010-11-03T07:55:30 *** dmj111 has joined #aichallenge 2010-11-03T07:55:36 i'll try to fix up parameters towards the end, but not going to be using much in the way of stats to do it 2010-11-03T07:55:50 It'd be tough with the time constraints... but you might be able to amortize it across the opening turns 2010-11-03T07:55:54 what do you mean by satistical? 2010-11-03T07:56:03 Like and MCMC or HMM thing 2010-11-03T07:56:06 Error323: you got me :P 2010-11-03T07:56:07 *an MCMC 2010-11-03T07:56:14 ok, nope :P 2010-11-03T07:56:26 i recognise HMM from looking at voice recognition stuff 2010-11-03T07:56:37 but i have no idea how those things work 2010-11-03T07:56:45 markov chain / monte carlo? 2010-11-03T07:56:47 is that mcmc? 2010-11-03T07:56:48 Yeah 2010-11-03T07:57:05 they are pretty complex yes 2010-11-03T07:57:10 my approaches all involve game theory 2010-11-03T07:57:20 i've sort of gone through the maths for it 2010-11-03T07:57:26 but i've never actually tried implementing anything 2010-11-03T07:57:47 grrr antimatroid =) 2010-11-03T07:57:52 i did an or course on markov chains and linear programming last year and just finished a stats unit where we did a topic on mcmc 2010-11-03T07:57:53 I'm trying to do a heavily pruned adversarial search... but the state space is pretty darn big. 2010-11-03T07:58:27 i have tried decision trees, game trees (various kinds) and zero sum games 2010-11-03T07:58:29 its *far too big* 2010-11-03T07:58:50 i tried alpha beta 2010-11-03T07:58:50 i'm not just randomly checking lots of moves for them though, at each node i select just a few 2010-11-03T07:59:02 yeah, and that's why it fails 2010-11-03T07:59:12 i haven't implemented alpha beta, i need promising results before i care about throwing more in to make it change 2010-11-03T08:00:12 Well, the thing about this game is that a lot of the possible moves are very similar to other possible moves. 2010-11-03T08:01:17 that's a bold statement 2010-11-03T08:01:19 =) 2010-11-03T08:01:23 explain 2010-11-03T08:01:27 i'm going under the assumption that many are clearly dominated 2010-11-03T08:01:37 it's a matter of finding the undominated moves 2010-11-03T08:01:44 then playing those 2010-11-03T08:01:50 but atm i even need to pick the "best" of those 2010-11-03T08:02:03 well, some are identical. Issuing 1 order of n ships from x to y is identical to issuing n orders of 1 ship from x to y 2010-11-03T08:02:56 *** AJC has joined #aichallenge 2010-11-03T08:03:01 only if in the same turn 2010-11-03T08:03:02 *** sigh has quit IRC (Remote host closed the connection) 2010-11-03T08:03:06 Of course. 2010-11-03T08:03:15 AJC: hi Alex J. Champandard? 2010-11-03T08:03:18 hey, I'm a bit fuzzy on the spec... http://ai-contest.com/specification.php 2010-11-03T08:03:25 raedwulf: yes. 2010-11-03T08:03:27 But you only need to search the moves that aren't isomorphic. 2010-11-03T08:03:27 yes, but with say my decision trees, i will consider taking planet 1 and then 2 as well as 2 and then 1 2010-11-03T08:03:35 Alex? 2010-11-03T08:03:36 simply because the moves will quite often be made up differently 2010-11-03T08:03:43 But further, what I'm saying is that sending 7 ships from x to y is very similar to sending 8 ships from x to y 2010-11-03T08:03:46 ^^, mailed you about terravox some time ago :P 2010-11-03T08:04:08 songoftheweave: not true 2010-11-03T08:04:23 the spec. says advancement (adding planet growth) happens before the arrival, but in the Java PlayGame, the simulation seems to do the opposite. 2010-11-03T08:04:31 yeah, i wish there was more "cooked up" maps for the contest 2010-11-03T08:04:33 raedwulf: ah, that must have been a while back! 2010-11-03T08:04:38 such that decisions were a lot harder to make 2010-11-03T08:05:03 AJC: the specification is correct 2010-11-03T08:05:27 antimatroid: is the PlayGame.jar in the starter kits also correct? 2010-11-03T08:05:31 AJC: yea i noticed that in java playgame too 2010-11-03T08:05:35 there were quite a few bugs with the original engine that needed to be ironed out 2010-11-03T08:05:41 AJC: get jBotManager 2010-11-03T08:05:41 antimatroid: ah, shit. 2010-11-03T08:05:47 antimatroid: URL? 2010-11-03T08:05:53 look in the forums 2010-11-03T08:05:58 http://ai-contest.com/forum/viewtopic.php?f=18&t=751 2010-11-03T08:06:03 i bookmarked it, forgot 2010-11-03T08:06:09 the starter kit hasn't been updated though, that's scary. 2010-11-03T08:06:13 typically, java playgame isn't accurate 2010-11-03T08:06:14 biggest issue was neutral resolution 2010-11-03T08:06:16 AJC: you can also use this one: http://visualizer.naktibalda.lt/ its good 2010-11-03T08:06:22 it was surprisingly hard to pick between the feasible options 2010-11-03T08:06:30 It's more similar than sending 8 ships from x to z or 8 ships from x to y. I'm using the simplifying assumption that I can ignore quite a few of the moves and one that I search will be close enough to it. *shrugs* I am interested to see if it is a valid assumption. heh 2010-11-03T08:06:45 the starter packages are a guide, meh 2010-11-03T08:06:48 there's also a python playgame based off the server backend 2010-11-03T08:07:01 Error323: I have a new visualizer, just the old PlayGame 2010-11-03T08:07:12 this game is ridiculously hard anyway 2010-11-03T08:07:14 Zannick: its yours right? 2010-11-03T08:07:19 i still have no idea how to try and solve stuff 2010-11-03T08:07:20 ah k 2010-11-03T08:07:23 raedwulf: oh, that sounds nice. do you you have a link? 2010-11-03T08:07:29 trying to remember it 2010-11-03T08:07:29 *** yasith has quit IRC (Ping timeout: 260 seconds) 2010-11-03T08:07:32 i need it too lol 2010-11-03T08:07:39 its up in my logs somewhere 2010-11-03T08:07:43 optimal moves switch from the tiniest changes in numbers 2010-11-03T08:07:56 indeed 2010-11-03T08:08:24 i miss the comfort of tron, max 4 moves per turn 2010-11-03T08:08:38 saying 8 ships from x to y is very similar to 7 from x to y is a mistake as its consequences can be drastic over time 2010-11-03T08:09:11 most of the packages out there use the old buggy Java PlayGame... you should really get the word out, update the front page. 2010-11-03T08:09:15 it's fairly easy to cook up a 3 planet map where you should send everything you have and all growth into the future at the middle planet 2010-11-03T08:09:17 I only scan the forums. 2010-11-03T08:09:17 Sure it's a mistake. simplifying assumptions usually are. I'm exploring whether it's a small enough mistake. 2010-11-03T08:09:40 http://bitbucket.org/dmj111/planetwars_test_tools 2010-11-03T08:09:54 what i want is a magical function to resolve battles from both players making moves with any of their future ships 2010-11-03T08:10:16 AJC: under what name are you participating? 2010-11-03T08:10:19 i'm not sure anyone has that right 2010-11-03T08:10:27 Error323: aigamedev 2010-11-03T08:10:41 ah that makes sense :P 2010-11-03T08:10:51 are you the person from that website? 2010-11-03T08:11:24 antimatroid: I resolve battles, i was doing it according to the playgame.jar but I was occasionally off due to that bug :-) 2010-11-03T08:11:29 antimatroid: yes. 2010-11-03T08:11:46 AJC: but with both player making moves each turn both with their current planets and into the future? 2010-11-03T08:11:48 AJC: I don't see you on the leaderboard under france anyway =) 2010-11-03T08:11:55 what's ur rank? 2010-11-03T08:12:03 460 or something. 2010-11-03T08:12:13 the new bot suffers from those prediction bugs now 2010-11-03T08:12:16 hehe 2010-11-03T08:12:20 it gets messy when they start cancelling out the moves they planned to make 2010-11-03T08:12:28 i had that aswell at start :/ 2010-11-03T08:12:41 but the specs are correct indeed. 2010-11-03T08:13:07 AJC: have you got the correct amount for claiming neutrals and correct neutral resolution? 2010-11-03T08:13:56 antimatroid: I thought I had, then I realized the spec. was different, so updated it... and now it resolves differently in PlayGame. 2010-11-03T08:14:17 and here I am looking for a better implemented version (which IMHO, should be updated in all starter kits) 2010-11-03T08:14:17 antimatroid: I just wrote a battle resolver - seemed fairly straight forward - just step through the incoming fleets 2010-11-03T08:14:40 d0ugal: are you making moves from both players into the future? 2010-11-03T08:14:42 antimatroid: but yes, I have it. 2010-11-03T08:14:44 that's where it gets complicated 2010-11-03T08:14:45 Top 10 players: bocsimacko(4275), dmj111(4171), bix0r4ever(4148), Hazard(4141), shangas(4103), george(4081), Accoun(4059), felixcoto(4049), GreenTea(4018), rsergio(4005) 2010-11-03T08:14:52 antimatroid: Yeah 2010-11-03T08:14:59 antimatroid: I don't make moves for the enemy player, just predict what he could do. 2010-11-03T08:15:12 antimatroid: Only issue I've got is if they arrive at the same time 2010-11-03T08:15:23 i will let the enemy send ships from a planet they don't even own yet 2010-11-03T08:15:29 I only started it last night though and it mostly works so far 2010-11-03T08:15:35 for reference, this is my neutral resolution 2010-11-03T08:15:35 http://codepad.org/FrOJZ8NB 2010-11-03T08:15:39 for generating future states 2010-11-03T08:17:29 d0ugal: what are you doing when one of your "future fleets" gets rendered void from one of their newer "future fleets"? 2010-11-03T08:17:48 just not sending that anymore? the whole "move" or just that individual fleet? 2010-11-03T08:17:58 or throwing out that entire branch? 2010-11-03T08:19:23 *** yasith has joined #aichallenge 2010-11-03T08:20:26 hehe i call them "la" states :D 2010-11-03T08:20:32 la=lookahead 2010-11-03T08:20:49 what do you mean? 2010-11-03T08:21:08 instead of future, i call them lookahead 2010-11-03T08:21:11 i.e "la" :P 2010-11-03T08:21:27 yeah, i want to know how d0ugal is doing his future states thing 2010-11-03T08:22:58 *** Appleman1234 has quit IRC (Ping timeout: 240 seconds) 2010-11-03T08:24:39 raedwulf: that python test tool leaves my bot threads hanging with 100% CPU usage somehow... I have to kill it off. 2010-11-03T08:24:51 like tcp.c for that matter. 2010-11-03T08:25:19 AJC: sorry, i don't think that was the other one i was looking at 2010-11-03T08:25:35 hmmm :( 2010-11-03T08:25:43 ill have to look through my logs again 2010-11-03T08:28:54 antimatroid http://72.44.46.68/canvas?game_id=836240 2010-11-03T08:29:05 your routing code needs fixing 2010-11-03T08:29:17 you keep routing without checking if an enemy can take the planet if it sends forward 2010-11-03T08:29:22 i just sniped your edge planets 2010-11-03T08:29:31 before you could reinforce them 2010-11-03T08:29:52 bhasker: that's alright, it hasn't got anything for that yet 2010-11-03T08:30:22 i'm considering switching to making moves to planets without a specified final time 2010-11-03T08:30:32 does anyone have the link for the python playgame based off the engine? 2010-11-03T08:30:49 what do you mean specified final time? 2010-11-03T08:31:15 at my move has a single destination and "arrive time" then send ships from numerous source planets and at many time points 2010-11-03T08:31:41 but rather than pick a time i want my move to be executed i might just add times to send fleets and make moves for a planet into the future at once 2010-11-03T08:32:17 i sweep each planet to see what i can send at which point, then make moves based of those amounts 2010-11-03T08:32:28 http://bitbucket.org/zannick/planet-wars/src ? 2010-11-03T08:32:49 it's hard to get it all right, i think i still have a few bugs 2010-11-03T08:33:09 bhasker: wow, my bot sucked then 2010-11-03T08:33:34 ah k 2010-11-03T08:33:38 as soon as it starts losing it tries to expand like crazy, making it lose even faster 2010-11-03T08:33:49 my bot has the same problem 2010-11-03T08:33:57 bhasker: how do you prevent a situation like this: http://72.44.46.68/canvas?game_id=836289 2010-11-03T08:33:58 so i track a few parameters 2010-11-03T08:34:10 i was just looking at that game 2010-11-03T08:34:13 dmj111: thanks 2010-11-03T08:34:20 trying to understand why i didn't take any new neutrals 2010-11-03T08:34:22 AJC: look at dmj111 's link 2010-11-03T08:34:28 yeah 2010-11-03T08:34:35 dmj111: what happens to a bot that copies every move that it's opponent is doing ?will it always get a draw ? or will it always lose because of the turn advantage to the opponent 2010-11-03T08:34:35 i wont attack unless i plan to take it for at least one turn 2010-11-03T08:34:37 there were some problems on windows, but give it a go 2010-11-03T08:34:44 well i won this but i also often lose because i keep raging that planet 2010-11-03T08:34:47 and failing at it 2010-11-03T08:34:54 though i think i know why 2010-11-03T08:34:57 nullkuhl: copybot has been done, doesn't do very well 2010-11-03T08:34:58 i was already low on ships 2010-11-03T08:35:06 taking a neutral will probably have killed me 2010-11-03T08:35:17 i was down by 30 ships 2010-11-03T08:35:18 yeah, prolly 2010-11-03T08:35:32 though since most of your ships were in flight 2010-11-03T08:35:36 it shold have been relatively safe 2010-11-03T08:35:46 there must be a way of determining if enemy is mostly in flight 2010-11-03T08:35:52 then taking the neutral is safe 2010-11-03T08:35:52 yeah 2010-11-03T08:35:57 since enemy can't redirect them 2010-11-03T08:35:58 i've been experimenting with that aswell 2010-11-03T08:35:59 nullkuhl: someone (zannick I think) did that. It should lose, if it only copies, because of the turn advantage. 2010-11-03T08:36:03 but not successfully so far. 2010-11-03T08:36:24 antimatroid http://72.44.46.68/canvas?game_id=836321 2010-11-03T08:36:41 yeah, that's the one i said wow about 2010-11-03T08:36:53 i love watching tcp live 2010-11-03T08:37:17 ah 2010-11-03T08:37:24 lol, jambachili just expanded too much on the first turn against me 2010-11-03T08:37:27 i don't watch it live other than by following logs 2010-11-03T08:37:45 i just wait :P 2010-11-03T08:37:49 jBotManager lets you watch them live 2010-11-03T08:37:51 it's awesome 2010-11-03T08:38:27 http://72.44.46.68/canvas?game_id=836371 2010-11-03T08:38:33 he dc before i even finished him 2010-11-03T08:39:32 i saw jBotManager maybe i should use that 2010-11-03T08:39:57 thats a bug, no way he should have launched 58 ships with you like 3 turns away 2010-11-03T08:40:04 5 turns rather 2010-11-03T08:40:12 this is like a rage bot fail 2010-11-03T08:40:17 he probably loses to rage bot on this map 2010-11-03T08:40:40 yep 2010-11-03T08:40:53 i seem to beat shampoo more than not atm 2010-11-03T08:41:11 http://72.44.46.68/getplayer?player=antimatroid.heuristicbot recent history 2010-11-03T08:41:39 antimatroid: does the jBotManager allow you to keep running games on the TCP server in a loop? 2010-11-03T08:41:53 it lets you specify the number of reps 2010-11-03T08:42:05 not sure you can set it at infinity, but you should be able to find something suitably large 2010-11-03T08:42:08 you can save the games too 2010-11-03T08:43:07 AJC: most importantly, it lets you play your own bot 2010-11-03T08:43:21 antimatroid: oh, by hand? 2010-11-03T08:43:26 but you can only send ship amounts in x0% increments, which sucks 2010-11-03T08:43:37 like 10% of planets ships, 20, ..., 100 2010-11-03T08:44:21 ah, then it's a cosmetic feature :-) 2010-11-03T08:44:32 it's still great for debugging 2010-11-03T08:44:48 i just wish i could specify the exact number of ships to send 2010-11-03T08:44:56 yeah, 2010-11-03T08:45:09 you will want to increase the timeout limit when playing as a human, you can hit enter to "finish your turn" 2010-11-03T08:48:20 oh man, this is going to be close, shampoo has me on growth by one, i have more ships 2010-11-03T08:48:53 antimatroid: hehe, you need professional commentary, Starcraft 2 like. 2010-11-03T08:49:02 woohoo #27 on official! 2010-11-03T08:49:07 first time i broke top 30 2010-11-03T08:49:10 i haven't got anything on official atm 2010-11-03T08:49:17 it's under "youwillfail" when i do 2010-11-03T08:49:27 bhasker: gratz, I shall soon follow you =) 2010-11-03T08:49:32 hehe:) 2010-11-03T08:49:47 damn, he overtook me with less than 10 moves remaining 2010-11-03T08:50:07 who did? 2010-11-03T08:50:08 i'm pretty sure my bot should have been able to take him too :| 2010-11-03T08:50:10 shampoo 2010-11-03T08:50:13 ah 2010-11-03T08:50:18 yea he is an annoying bot 2010-11-03T08:50:25 http://72.44.46.68/canvas?game_id=836453 2010-11-03T08:50:30 though my bot does much better against him now 2010-11-03T08:50:55 my bot seems to have the upper hand, but he's still causing me pain 2010-11-03T08:51:05 why didnt you snipe the central 2010-11-03T08:51:08 when he launched the first time 2010-11-03T08:51:16 you waited till he captured it 2010-11-03T08:51:32 cause i had just sent 29 out 2010-11-03T08:51:37 i couldn't afford to send more from that planet 2010-11-03T08:51:47 oh wait, no i hadn't 2010-11-03T08:51:48 hmmm 2010-11-03T08:51:56 couldn't back it up? 2010-11-03T08:52:03 i check how much they could send that turn too 2010-11-03T08:52:26 yeah, i think that was it 2010-11-03T08:52:29 hmm he could have 2010-11-03T08:52:49 wonder why Shampoo is stockpiling on the left. 2010-11-03T08:53:25 not sure, planned for defense but not used yet, and not correctly sending ahead? 2010-11-03T08:53:38 my bot might for example might consider the other 2 planets to have already been taken by me 2010-11-03T08:53:48 and not use them for redistribution, but then the 9 is doin git 2010-11-03T08:54:09 what your bot should do, once they've reached a stable state, is stream ships from your big planet at the front into the middle. 2010-11-03T08:54:24 (you can't do an all out attack or he'll counter and take yours) 2010-11-03T08:54:30 yeah, i have troubles spamming the enemy 2010-11-03T08:54:32 heh my bot is climbing still #26 now 2010-11-03T08:54:33 he always does it too early 2010-11-03T08:54:41 if i make it into top 20 i will be happy 2010-11-03T08:54:48 bhasker: its on a massive killingspree 2010-11-03T08:54:56 bhasker: how much time have you spent on your bot? 2010-11-03T08:55:03 no idea lol 2010-11-03T08:55:04 lots 2010-11-03T08:55:08 its damn addictive 2010-11-03T08:55:17 lol, i've been programming with every spare minute since the contest was announced 2010-11-03T08:55:19 though most of it has been fixing bugs 2010-11-03T08:55:22 programming/thinking 2010-11-03T08:55:26 pretty much the same here 2010-11-03T08:55:34 i keep revisiting ideas 2010-11-03T08:55:41 i would put some heuristic in 2010-11-03T08:55:47 then 2 days later i would remove 2010-11-03T08:55:48 it 2010-11-03T08:55:51 back and forth 2010-11-03T08:55:54 yep 2010-11-03T08:55:56 hehe 2010-11-03T08:56:14 i can't get my actual ideas going without proper conflict resolution as specified before 2010-11-03T08:56:25 and my move selection sucks 2010-11-03T08:56:49 Errir323: what are you doing? 2010-11-03T08:56:56 yea it took me a long time to get conflict resolution right 2010-11-03T08:57:04 especially for neutrals where both me and enemy are attacking 2010-11-03T08:57:17 antimatroid: mmm? 2010-11-03T08:57:17 its quite complicated to get right with coordinated launches 2010-11-03T08:57:21 but are you doing future moves from both players? 2010-11-03T08:57:42 what do you do when one of their future moves cancels out one of yours? 2010-11-03T08:58:00 Error323: http://72.44.46.68/canvas?game_id=836546 2010-11-03T08:58:06 oef 2010-11-03T08:58:16 that map is just scary 2010-11-03T08:58:42 and your attack very good 2010-11-03T08:58:53 gg there =) 2010-11-03T08:59:17 like i said before, i wont make an offensive move unless i plan to take it 2010-11-03T08:59:33 mmm i have a different strat 2010-11-03T08:59:37 although sometimes i don't decide to finish executing one, but it was intended 2010-11-03T08:59:48 if i can attack an enemyplanet, do so with all my ships 2010-11-03T08:59:49 it's very costly to have them defend all your attacks 2010-11-03T08:59:56 it seems to be inferior :P 2010-11-03T09:00:38 attacking the enemy, you don't really lose anything ship wise, you just want to maxe growth ownership 2010-11-03T09:01:48 well yes and no 2010-11-03T09:02:05 attacking over long distances can be dangerous 2010-11-03T09:02:16 as in flight ships are "useless" 2010-11-03T09:02:30 *** Blkt has joined #aichallenge 2010-11-03T09:02:35 *** Utkarsh has quit IRC (Ping timeout: 240 seconds) 2010-11-03T09:02:36 that's why i make attacks iff i will win the planet for at least one turn 2010-11-03T09:02:52 and i'm considering changing my moves to go into the future, and not have a specified finish time 2010-11-03T09:02:54 antimatroid: how can youu possibly know 2010-11-03T09:03:07 since most bots have routing algorithms 2010-11-03T09:03:17 that make it incredibly complex to predit 2010-11-03T09:03:19 predict* 2010-11-03T09:03:40 i look at all the ships they have (in fleet and on planets including future growth) inside the time i plan to take it 2010-11-03T09:03:45 *** Utkarsh has joined #aichallenge 2010-11-03T09:04:40 *** perror has joined #aichallenge 2010-11-03T09:04:52 mja, i do the same 2010-11-03T09:04:57 but its not working well 2010-11-03T09:05:22 well, less good then just sending enemy fleets 2010-11-03T09:05:26 all fleets* 2010-11-03T09:07:31 antimatroid why did you decide to onl commit 16 shps on turn 1 2010-11-03T09:07:35 thats what killed Error 2010-11-03T09:07:40 he committed 68 2010-11-03T09:07:42 to the neutral 2010-11-03T09:07:44 and you sniped it 2010-11-03T09:07:59 because i suck when i make lots of moves first turn, so i do the suboptimal thing and make just one 2010-11-03T09:08:08 if the other bot makes too many, i take them out then though 2010-11-03T09:08:12 ah 2010-11-03T09:08:29 i have been having trouble with the same thing 2010-11-03T09:08:32 if i expand too much 2010-11-03T09:08:34 it kills me 2010-11-03T09:08:48 a lot of games i lose bcause i decided to acquire an expensive neutral far away 2010-11-03T09:08:49 yeah that works very counter productive for me, since I assume the enemy also maximizes growth first turn :P 2010-11-03T09:08:50 try doing the one move first turn trick, you only lose 1 move on expansion 2010-11-03T09:09:10 well on maps with a lot of low growth planets 2010-11-03T09:09:13 you lose the game 2010-11-03T09:09:13 and you get lots of information 2010-11-03T09:09:23 oops i mean plaents with low # of ship son them 2010-11-03T09:09:26 not always 2010-11-03T09:09:30 hmm 2010-11-03T09:09:37 you can analyse the map beforehand 2010-11-03T09:09:38 give it a go and see how it does 2010-11-03T09:09:40 and decide then 2010-11-03T09:09:40 let me try it 2010-11-03T09:09:47 *** vall has joined #aichallenge 2010-11-03T09:09:50 i intend to remove it once i smarten it up 2010-11-03T09:09:55 but for the moment it does a lot better 2010-11-03T09:10:42 its a RPS decision i thinkk 2010-11-03T09:11:00 until yyou can deduce a good heuristic :P 2010-11-03T09:11:49 i think most of my losses to shampoo are from him taking and managing to claim the middle planet 2010-11-03T09:11:50 antimatroid: why don't you submit this bot to the ranking server? 2010-11-03T09:11:55 it seems to be very good 2010-11-03T09:11:59 meh 2010-11-03T09:12:08 comp isn't till the end of the month 2010-11-03T09:12:11 antimatroid: Sorry, went for lunch. 2010-11-03T09:12:18 *** dmj111 has quit IRC (Ping timeout: 252 seconds) 2010-11-03T09:12:23 that's alright, it's 12 15am in australia :P 2010-11-03T09:12:25 yeah, so? :P 2010-11-03T09:12:47 last time i put it up it was crashing, i think the bugs should be fixed but meh 2010-11-03T09:12:53 antimatroid: I don't have the code here but basically I go through the incoming fleets - in order of the moves and work out what the outcome will be 2010-11-03T09:12:53 it's effort to clean out all my debugging stuff 2010-11-03T09:13:02 Sorry, in order of their arrival time 2010-11-03T09:13:08 d0ugal: what about fleets rendered unsendable? 2010-11-03T09:13:19 rendered unsendable? 2010-11-03T09:13:49 say i planned to send 15 fleets from planet 1 at time 10, but the send ships to planet 1 within that time leaving it with less than 15 shgips left 2010-11-03T09:13:54 i cannot send that fleet anymore 2010-11-03T09:14:01 I've setup my code to do logging and debugging only if the environment variable is set on my machine - so I can upload it without problems :D 2010-11-03T09:14:03 but they* send ships 2010-11-03T09:14:27 15 ships** (I am terrible at that) 2010-11-03T09:14:47 Top 10 players: bocsimacko(4254), dmj111(4157), bix0r4ever(4133), Hazard(4107), george(4071), shangas(4070), felixcoto(4027), GreenTea(4020), Accoun(4017), asavis(3975) 2010-11-03T09:14:52 to make matters even worse, that 15 fleet might have only been part of a move 2010-11-03T09:15:05 I didn't quite follow. You go to take planet A and then they send a fleet that will arrive first? 2010-11-03T09:15:10 haha my bot goes dumb 2010-11-03T09:15:10 or go to when the planet is changing ownership multiple times 2010-11-03T09:15:13 on turn 1 2010-11-03T09:15:15 bhasker: http://72.44.46.68/canvas?game_id=836708? 2010-11-03T09:15:18 i know 2010-11-03T09:15:20 o_L 2010-11-03T09:15:22 ah ok 2010-11-03T09:15:30 it doesn't account for neutrals that you may acquire 2010-11-03T09:15:31 in future 2010-11-03T09:15:33 thats why 2010-11-03T09:15:41 i am making moves from both players each turn, with any ships they own at any point in time (as present scenario) and trying to simulate that gamestate 2010-11-03T09:15:46 i have been trying to find a good way to do that 2010-11-03T09:15:51 but don't have one yet 2010-11-03T09:15:52 :-\ 2010-11-03T09:15:59 so it just assumed it could take your home planet 2010-11-03T09:16:09 and it's when i do that that problems arise 2010-11-03T09:16:20 generating future states is easy enough until one does that 2010-11-03T09:16:45 "does that"? 2010-11-03T09:16:47 lol 2010-11-03T09:16:56 my bot is geting dumber 2010-11-03T09:17:00 its doing dumb moves 2010-11-03T09:17:04 http://72.44.46.68/canvas?game_id=836719 2010-11-03T09:17:05 Perhaps I'm just over simplifying the problem too much 2010-11-03T09:17:10 it sends 72 ships from left corner of the map 2010-11-03T09:17:14 to the right corner 2010-11-03T09:17:21 to defend a planet and loses the original 2010-11-03T09:17:22 wow 2010-11-03T09:18:21 *** p4p4 has quit IRC (Ping timeout: 252 seconds) 2010-11-03T09:19:37 bhasker: did you change its behavior or sth? 2010-11-03T09:19:42 yea 2010-11-03T09:19:50 was that my bad? 2010-11-03T09:19:53 ytes 2010-11-03T09:19:55 haha 2010-11-03T09:19:59 works for me :p 2010-11-03T09:20:04 but the bug is not because of the launch 1 2010-11-03T09:20:09 ah, okay 2010-11-03T09:20:11 its because i am underestimating nearby ships 2010-11-03T09:20:21 it for some reason decided i could send 76 ships 2010-11-03T09:20:36 when enemy had a planet with 40 ships sitting at much shorter distance 2010-11-03T09:23:48 i wonder how often i lose of you 2010-11-03T09:23:55 i think more then 50% :/ 2010-11-03T09:24:14 i am guessing its closer to even 2010-11-03T09:24:23 our bots behave pretty similarly 2010-11-03T09:25:03 mmm 2010-11-03T09:25:53 i will let it run with this 1 acquisition at turn 1 for some time let me see how it does 2010-11-03T09:26:05 *** dmj111 has joined #aichallenge 2010-11-03T09:26:26 if the other bot is smart (bocsimacko) i think it hurts 2010-11-03T09:26:32 otherwise for the moment does alright 2010-11-03T09:27:01 apc just pounded me 2010-11-03T09:27:15 ajc** 2010-11-03T09:27:25 well bocismacko hurts either way 2010-11-03T09:27:30 yeah 2010-11-03T09:27:38 he does some reaally awesome stuff 2010-11-03T09:28:12 *** nullkuhl has quit IRC (Read error: Connection reset by peer) 2010-11-03T09:28:27 right, i'm taking my bot off tcp to try a change of move selection out 2010-11-03T09:28:45 *** nullkuhl has joined #aichallenge 2010-11-03T09:32:08 is it known that maps will be point symmetrical in the actual tournament? 2010-11-03T09:34:31 k i need to factor in neutrals enemy will acquire before i can do this one turn trick 2010-11-03T09:34:35 one move trick 2010-11-03T09:34:46 my bot just ends up launching everything at the enemy on turn 2 2010-11-03T09:34:51 and loses 2010-11-03T09:34:52 :-\ 2010-11-03T09:34:57 http://72.44.46.68/canvas?game_id=836849 2010-11-03T09:34:59 example 2010-11-03T09:37:25 *** Azrathud has quit IRC (Ping timeout: 255 seconds) 2010-11-03T09:38:20 lol 2010-11-03T09:38:27 barontrozo likes the fireworks 2010-11-03T09:44:37 hah mega1 started his bot 2010-11-03T09:44:42 yep 2010-11-03T09:44:56 and i lose 2010-11-03T09:44:58 .. 2010-11-03T09:45:04 *** songoftheweave has quit IRC (Quit: songoftheweave) 2010-11-03T09:45:08 yes, it should do better at stalemates 2010-11-03T09:45:57 mega1 's bot is on.. .RUN 2010-11-03T09:46:07 wow 2010-11-03T09:46:08 http://72.44.46.68/canvas?game_id=837017 2010-11-03T09:46:11 :-\ 2010-11-03T09:46:15 my bot did the dum thing 2010-11-03T09:46:21 sent my ships off to left corner:-\ 2010-11-03T09:46:41 our start was identical 2010-11-03T09:46:44 bhasker: my offense will consider every ship they could send at that planet in that time 2010-11-03T09:46:50 that way it usually goes for the closest enemy 2010-11-03T09:46:54 what's the performance of the official server? 2010-11-03T09:46:59 unless it really can snipe one further away from their stupidity 2010-11-03T09:47:06 AJC: no idea :P 2010-11-03T09:47:18 AJC: http://ai-contest.com/game_stats.php ? 2010-11-03T09:47:28 yea i need to change my code to be a little more smarter about calculating ships Required to take a planet 2010-11-03T09:47:30 do we have hardware specs? 2010-11-03T09:47:33 no 2010-11-03T09:47:33 especially accounting for neutrals 2010-11-03T09:47:35 not me 2010-11-03T09:48:11 i need to massively improve my sniping algorithm, it does appear to be shit 2010-11-03T09:48:16 AJC: i believe its variable, because some servers seem to be able to handle more games per minute than tohers 2010-11-03T09:48:20 others* 2010-11-03T09:48:22 also i need todo other things >_> 2010-11-03T09:48:33 makes it hard to figure out whether my code is going to be fast enough. 2010-11-03T09:48:44 takes 0.5s on my machine for the worst case in the worst map. 2010-11-03T09:49:06 its not an anytime algorithm? 2010-11-03T09:49:08 *** davidjliu has joined #aichallenge 2010-11-03T09:49:11 * mega1 will destroy his ISP 2010-11-03T09:49:42 Error323: it is, but then I need to know so I can test how it scales down. 2010-11-03T09:49:52 hah victory by timeout over bockismacko 2010-11-03T09:50:04 AJC: why not modify the time limit? 2010-11-03T09:50:09 mega1: ? 2010-11-03T09:50:13 e.g. 0.4, 0.3s 2010-11-03T09:50:27 http://72.44.46.68/canvas?game_id=837057 2010-11-03T09:50:28 lol 2010-11-03T09:50:29 raedwulf: it's easy to test, I just want to know what the server performance is so I know what value to set it at ;-) 2010-11-03T09:50:32 oh, i see 2010-11-03T09:50:43 yes, the bastards 2010-11-03T09:50:49 it happens every few hours 2010-11-03T09:50:58 meh, i don't care about timeout losses 2010-11-03T09:51:03 AJC: ah 2010-11-03T09:51:18 i'd have killed myself long ago if i did 2010-11-03T09:51:50 very impressive though, bastard :D 2010-11-03T09:51:55 bah the internet is bad here 2010-11-03T09:51:56 :/ 2010-11-03T09:52:11 i'm in tasmania and running off wireless through 3 brick walls, i can't complain 2010-11-03T09:52:25 lol 2010-11-03T09:52:33 AJC: there are multiple servers connecting via ssh through the main server if i understand correctly. And you don't know where it will be run. 2010-11-03T09:53:00 mega1: http://72.44.46.68/canvas?game_id=837038 what the hell were you doing at the end there? 2010-11-03T09:53:03 s/ssh/tcp prolly :P 2010-11-03T09:53:17 it made no difference, but still, i've seen some weird stuff with your redistribution other times 2010-11-03T09:53:23 one probably shouldn't question your ways though 2010-11-03T09:53:28 they work 2010-11-03T09:54:03 AJC: , you'll have to run sort of auto-tuner at the start of your bot 2010-11-03T09:54:07 i suspect 2010-11-03T09:55:24 *** phreeza has joined #aichallenge 2010-11-03T09:56:30 any java coders around ? 2010-11-03T09:57:14 hey, i am beating bocsimacko 2010-11-03T09:57:21 no, he got me 2010-11-03T09:57:23 muggle 2010-11-03T09:59:58 http://72.44.46.68/canvas?game_id=837144 2010-11-03T09:59:59 almost 2010-11-03T10:00:52 antimatroid: pwned :) 2010-11-03T10:01:18 i consider that a good effort against him :P 2010-11-03T10:04:31 antimatroid: the evaluation went wild seeing the end of the game 2010-11-03T10:05:31 *** Azrathud has joined #aichallenge 2010-11-03T10:06:02 antimatroid: so youre back to heuristics then? 2010-11-03T10:06:05 no more trees? 2010-11-03T10:07:47 no, he's using ents instead 2010-11-03T10:08:17 ents? 2010-11-03T10:08:29 poor LOTR reference :( 2010-11-03T10:08:37 oh :) 2010-11-03T10:08:53 to isengard we come! 2010-11-03T10:09:02 i don't know all the lotr terminology 2010-11-03T10:09:40 hehe :P 2010-11-03T10:12:08 *** nullkuhl has quit IRC (Read error: Connection reset by peer) 2010-11-03T10:12:15 *** antimatroid has quit IRC (*.net *.split) 2010-11-03T10:12:15 *** Mathnerd314 has quit IRC (*.net *.split) 2010-11-03T10:12:15 *** espes has quit IRC (*.net *.split) 2010-11-03T10:12:15 *** amstan has quit IRC (*.net *.split) 2010-11-03T10:12:16 *** mietek has quit IRC (*.net *.split) 2010-11-03T10:12:16 *** eregon has quit IRC (*.net *.split) 2010-11-03T10:12:16 *** drodil has quit IRC (*.net *.split) 2010-11-03T10:12:30 *** nullkuhl has joined #aichallenge 2010-11-03T10:13:47 netsplit kaboom 2010-11-03T10:14:43 doh 2010-11-03T10:14:46 Top 10 players: bocsimacko(4237), dmj111(4159), bix0r4ever(4146), Hazard(4105), george(4069), shangas(4068), GreenTea(4031), felixcoto(4018), Accoun(3985), asavis(3979) 2010-11-03T10:28:42 *** boegel has joined #aichallenge 2010-11-03T10:32:55 *** JamesMG has joined #aichallenge 2010-11-03T10:33:34 *** antimatroid has joined #aichallenge 2010-11-03T10:33:34 *** Mathnerd314 has joined #aichallenge 2010-11-03T10:33:34 *** espes has joined #aichallenge 2010-11-03T10:33:34 *** amstan has joined #aichallenge 2010-11-03T10:33:34 *** mietek has joined #aichallenge 2010-11-03T10:33:34 *** eregon has joined #aichallenge 2010-11-03T10:33:34 *** drodil has joined #aichallenge 2010-11-03T10:33:34 *** hubbard.freenode.net sets mode: +o amstan 2010-11-03T10:34:32 lol antimatroid, bhasker we are stacking =) 2010-11-03T10:34:48 my bot got worse i made some changs 2010-11-03T10:35:11 welcome back 2010-11-03T10:35:14 its lost like 30 games now 2010-11-03T10:35:17 in a row 2010-11-03T10:35:40 ow 2010-11-03T10:35:53 ouch 2010-11-03T10:35:54 put back in the old version then :P 2010-11-03T10:36:06 i am working on differnet moves entirely 2010-11-03T10:36:10 hopefully it's better 2010-11-03T10:38:30 *** hellman has joined #aichallenge 2010-11-03T10:40:46 antimatroid: where in tasmania are you? 2010-11-03T10:41:59 brx from his hostname i would guess melbourne? 2010-11-03T10:42:10 that's not in tasmania. :) 2010-11-03T10:42:16 i know 2010-11-03T10:42:26 victoria right? 2010-11-03T10:42:29 yes 2010-11-03T10:42:30 09:52 < antimatroid> i'm in tasmania and running off wireless through 3 brick walls, i can't complain 2010-11-03T10:42:44 ah :) 2010-11-03T10:43:44 stealing internet is so fun 2010-11-03T10:44:05 *** Florian95 has quit IRC (Read error: No route to host) 2010-11-03T10:44:21 http://72.44.46.68/getplayer?player=bocsimacko-b0.45 that's just scary man... omg 2010-11-03T10:44:42 geez 2010-11-03T10:44:45 :) 2010-11-03T10:44:45 *** demerzel_ has joined #aichallenge 2010-11-03T10:45:39 uh, wow 2010-11-03T10:46:01 ? 2010-11-03T10:46:05 dominating 2010-11-03T10:46:26 check log just before you came in demerzel_ 2010-11-03T10:46:36 *** Florian95 has joined #aichallenge 2010-11-03T10:46:42 i mean, you'd think the top bots would be some sort of tradeof with a winning from b, b from c, c from a etc 2010-11-03T10:46:44 are you all googly-eyed over boc 2010-11-03T10:46:45 but no 2010-11-03T10:46:48 this dude is dominating xD 2010-11-03T10:46:54 yeah 2010-11-03T10:47:07 hehe yea he's got an insanely good foundation 2010-11-03T10:47:13 and a month to tweak it 2010-11-03T10:48:24 i have no idea what concepts he's using but my only hope atm is to brute force a solution haha 2010-11-03T10:48:37 ofc that's been my plan since the beginning 2010-11-03T10:48:52 well 2010-11-03T10:49:13 i think its too fast to be brute force 2010-11-03T10:49:13 if you've got dominating heuristics, you can lookahead 2010-11-03T10:49:28 and be even better 2010-11-03T10:49:32 phreeza: oh i doubt it is - that's my hope lol 2010-11-03T10:49:47 phreeza: i'll never beat that with game theory because i don't understand it well enough 2010-11-03T10:49:48 i think its heuristics yeah 2010-11-03T10:50:43 maybe some pattern recognition, since its lisp? :D 2010-11-03T10:50:55 is the guy ever on irc? 2010-11-03T10:51:01 yes 2010-11-03T10:51:05 mega1: is him 2010-11-03T10:51:11 ah i see 2010-11-03T10:51:31 prolly the best programmer in all of hungary ^^ 2010-11-03T10:51:37 he is using a very good eval system 2010-11-03T10:51:40 and his state is perfect 2010-11-03T10:51:53 which is harder to do than most people give credit 2010-11-03T10:52:12 his state? 2010-11-03T10:52:15 or at least the advanced parts 2010-11-03T10:52:16 hungary? 2010-11-03T10:52:17 haha 2010-11-03T10:52:20 jk 2010-11-03T10:52:22 lol 2010-11-03T10:53:14 http://www.youtube.com/watch?v=fXVTa3UeQE0 2010-11-03T10:53:41 haha 2010-11-03T10:53:47 "You need to install Adobe Flash Player" 2010-11-03T10:53:49 do i 2010-11-03T10:54:38 with this kind of ISP it's hard to test 2010-11-03T10:54:51 I had 8 timeouts out of 184 games 2010-11-03T10:54:58 if i ever get a stable bot together i'm gonna branch off to try and find a heuristic bot to defeat you 2010-11-03T10:55:10 probably never happen but i'm curious what strategies work better than others 2010-11-03T10:55:24 i know a couple only by accident 2010-11-03T10:55:31 but not how to get there consistently lol 2010-11-03T10:56:31 mega1: will you release the src after the competition? 2010-11-03T10:56:42 playing strength is strange b0.36 vs b0.45 locally: 11/37/78 (win/draw/loss) 2010-11-03T10:56:57 will the tcp server go bye bye afterwards? what a waste of coding effort 2010-11-03T10:57:04 while they are practically the same on tcp 2010-11-03T10:57:14 maybe we can get that ai guy to make it permanent since he's collecting good money lol 2010-11-03T10:57:52 Error323: yes, I will if I'm not too lazy 2010-11-03T10:58:03 and it doesn't embarras me too much 2010-11-03T10:58:04 cool :) 2010-11-03T10:58:27 haha why would it embarrass you 2010-11-03T10:58:30 come oooooooooooooooon 2010-11-03T10:58:39 just win once of b0.45 2010-11-03T10:58:45 just win once of b0.46 even 2010-11-03T10:58:50 i originally coded with comments etc for releasing afterwards 2010-11-03T10:58:55 now it's barely commented hardcore c++ lol 2010-11-03T10:59:12 curses 2010-11-03T10:59:20 yeah i'm just gonna throw it in my git account 2010-11-03T10:59:58 Error323: 45 moves ^^ 2010-11-03T11:00:00 46 2010-11-03T11:00:31 yeah..... 2010-11-03T11:00:43 scary 2010-11-03T11:00:49 xD 2010-11-03T11:01:29 brx: new town 2010-11-03T11:01:35 are you tasmanian? 2010-11-03T11:02:20 my ip always shows up as though i'm in victoria 2010-11-03T11:02:56 antimatroid: no, but my ex flew to tasmania on monday. she's touring the island for two weeks before heading to alice springs. 2010-11-03T11:03:13 ah, well tassie is awesome ;) 2010-11-03T11:03:23 but i'm biased, i hate participating in queues 2010-11-03T11:04:56 i'm jealous, last time i was in australia i was ... 4 years old. 2010-11-03T11:05:11 i went to kindergarden in melbourne. ;) 2010-11-03T11:05:15 *** superflit has joined #aichallenge 2010-11-03T11:07:03 my bot was streaming as part of a bug 2010-11-03T11:07:12 but now i'm wondering if delayed launch really > streaming 2010-11-03T11:07:30 or staged launches 2010-11-03T11:09:54 i think launching against the enemy fast is good 2010-11-03T11:10:01 defending it's better to wait as long as possible 2010-11-03T11:10:11 and expanding it just plaing difficult to do without coordinating 2010-11-03T11:10:16 plain* 2010-11-03T11:12:12 hmm looks lke my official submission is on a rampage 2010-11-03T11:12:17 moved to #26 2010-11-03T11:12:58 nice, I need to fix my code and then resubmit finally 2010-11-03T11:13:06 my official one is from oct 13th still 2010-11-03T11:14:22 i constantly resubmit 2010-11-03T11:14:26 nothing official about it 2010-11-03T11:14:36 its just a pool aswell which is your testground ^^ 2010-11-03T11:14:47 Top 10 players: bocsimacko(4221), dmj111(4122), bix0r4ever(4104), Hazard(4058), george(4047), GreenTea(4018), shangas(4016), felixcoto(3996), Accoun(3959), joakim_mjardner(3945) 2010-11-03T11:15:19 yeah, I'm just too lazy to fight with the C# submission compiler 2010-11-03T11:16:00 *** nullkuhl has quit IRC (Read error: Connection reset by peer) 2010-11-03T11:16:22 *** nullkuhl has joined #aichallenge 2010-11-03T11:16:28 heh 2010-11-03T11:16:38 *** nullkuhl has quit IRC (Read error: Connection reset by peer) 2010-11-03T11:17:07 *** nullkuhl has joined #aichallenge 2010-11-03T11:18:09 *** rebelxt has joined #aichallenge 2010-11-03T11:18:18 i tried submitting before, i don't think it likes my sources being spread out across multiple files 2010-11-03T11:18:22 i really cbf cleaning that up 2010-11-03T11:18:28 it's retarded that i can't do that 2010-11-03T11:18:45 multiple folders****** 2010-11-03T11:20:21 *** Azrathud has quit IRC (Ping timeout: 265 seconds) 2010-11-03T11:24:35 i am using pragma region so i can just have 1 source file 2010-11-03T11:24:56 although i dislike that it then won't outline the content functions any more 2010-11-03T11:25:06 *** rebelxt has left #aichallenge 2010-11-03T11:25:39 1 source file would be unbearable 2010-11-03T11:25:47 yea it's pretty long haha 2010-11-03T11:25:54 i separated out tcp stuff 2010-11-03T11:25:59 so i can just remove from project 2010-11-03T11:26:01 what tcp stuff? 2010-11-03T11:26:10 and i do have my logic in another cpp atm but it's defined in the main header 2010-11-03T11:26:10 i don't have that in any source files 2010-11-03T11:26:18 oh i build tcp directly into the bot 2010-11-03T11:26:29 why? 2010-11-03T11:26:42 *** vall has quit IRC (Quit: Page closed) 2010-11-03T11:26:42 why bother compiling then linking through an exe 2010-11-03T11:26:47 i can just run it directly from the ide 2010-11-03T11:26:54 cost:75 lines of code 2010-11-03T11:27:00 fair enough 2010-11-03T11:28:32 i did originally have a proxy bot that read from file and fed to stdout 2010-11-03T11:28:37 but that was a pain in the ass 2010-11-03T11:28:42 once you write the socket code once you're done 2010-11-03T11:29:14 meh, i just do it through jbotmanager and watch the games live 2010-11-03T11:29:31 sucks when you know you're going to lose but game is going to take another 100 turns or so 2010-11-03T11:29:31 eh i can wait a minute - can watch it from the log lol 2010-11-03T11:30:47 is dimwit really slow or is my logger making me slow 2010-11-03T11:30:59 dimwit is slow 2010-11-03T11:31:02 ok good 2010-11-03T11:31:11 so is JAmesMG.fun, sorry about that :) 2010-11-03T11:31:25 you're fine actually - i used to hate playing flavor but i guess he fixed that 2010-11-03T11:32:09 I capped that bot to 750ms give or take a few to act on the info it churns out in that time 2010-11-03T11:32:34 could bump it up for TCP but that's just cheap 2010-11-03T11:34:10 what does it do 2010-11-03T11:34:14 *** bhasker has quit IRC (Quit: bhasker) 2010-11-03T11:34:17 yea tcp is a bit forgiving tho i understand why 2010-11-03T11:34:24 it's kinda annoying when it's taking 4 seconds per turn tho 2010-11-03T11:35:41 it's building as big a tree as it can demerzel_ 2010-11-03T11:35:45 in the 750ms 2010-11-03T11:35:52 ahh 2010-11-03T11:36:15 i take up to 0.9 seconds, but i'm only doing a decision tree and it often can't generate for that long before it's run out of feasible "moves" 2010-11-03T11:36:17 how long would it go if you didn't cap it 2010-11-03T11:36:40 mid game if it's winning, 10+ seconds 2010-11-03T11:36:48 dayam 2010-11-03T11:36:52 you need some c++ 2010-11-03T11:36:54 it's kinda dumb, exponential with owned planets 2010-11-03T11:37:03 I just haven't spent the time to clean it up 2010-11-03T11:37:24 which is also why it loses 85% of the time 2010-11-03T11:37:26 do you guys have jobs? 2010-11-03T11:37:33 do you find it hard to not think about the bot during the daytime ;-) 2010-11-03T11:37:48 AJC: i work as a research assistant, when i find time, but mostly just a student 2010-11-03T11:38:03 antimatroid: so you're doing this full time? :-) 2010-11-03T11:38:07 I'm a developer, I work on my bot at work... 2010-11-03T11:38:07 my uni work output has been dreadful since the start of the contest 2010-11-03T11:38:25 hehe 2010-11-03T11:38:29 i studied for graphics exam the day of it, didn't go to class since first week 2010-11-03T11:38:37 studied for stats exam the day before 2010-11-03T11:38:53 i have analysis saturday, adv topics in maths monday and pdes tuesday, haven't done much for them either :| 2010-11-03T11:39:24 today's been obsessive coding for me, like yesterday, but on the bright side I've finally found a way to unify attack and defense so all options are considered equally and "objectively" 2010-11-03T11:39:36 last 2 months have been obsessive coding for me 2010-11-03T11:39:50 hehe, that's what I'm worried about.. and it's why I asked. 2010-11-03T11:40:04 I will have to stop after today, I have a business to run :-) 2010-11-03T11:40:19 good thing is i have everything set up for game trees and zero sum games once i get all my bugs worked out and some more stuff written 2010-11-03T11:40:35 atm it's just using a decision tree 2010-11-03T11:40:54 decision tree? really? 2010-11-03T11:40:55 i need to make things more efficient too 2010-11-03T11:41:05 oh, I stopped worrying about performance. 2010-11-03T11:41:09 AJC: i am a developer too haha 2010-11-03T11:41:09 yeah, make one move, then plan based off that 2010-11-03T11:41:16 so did i, untill i started trying to make trees 2010-11-03T11:41:19 stopped worrying about performance? 2010-11-03T11:41:31 well that's cool until you want to look ahead more than 1 move 2010-11-03T11:41:50 most of my trees remain in the "same move" 2010-11-03T11:41:51 I'm predicting all planets ahead 40 turns, then re-evaluating everything once I've dispatched each fleet. 2010-11-03T11:41:59 but i am constantly making fleets from the future 2010-11-03T11:42:02 predicting state is diff 2010-11-03T11:42:16 that's quick even if you implement it half-assed 2010-11-03T11:42:20 *** McLeopold has quit IRC (Quit: Leaving) 2010-11-03T11:42:20 (performance-wise) 2010-11-03T11:42:34 demerzel_: are you doing some kind of game tree search then? 2010-11-03T11:42:45 demerzel_: you still need to predict the opponent, searching won't help much. 2010-11-03T11:42:46 most of my time building trees is wasted on the 9000 really dumb moves it looks at 2010-11-03T11:42:54 mine are approximations of game trees 2010-11-03T11:43:02 i can't possibly explore every possible move at each node 2010-11-03T11:43:07 AJC: i am doing jack atm - one move minmax 2010-11-03T11:43:08 you wouldn't make it past 3 turns 2010-11-03T11:43:12 my state still have bugs 2010-11-03T11:43:13 is that what all the top bots are doing? 2010-11-03T11:43:22 most of the top bots claim to have trees at least 2010-11-03T11:43:32 what that actually means probably varies hehe 2010-11-03T11:43:42 i think a lot are still mostly heuristic bsased 2010-11-03T11:43:58 people are trying and failing at bigger ideas 2010-11-03T11:44:05 i could be wrong 2010-11-03T11:44:06 yeah, I'm not convinced by planning in this case. 2010-11-03T11:44:27 i think they will work, but iff one gets the details perfectly right 2010-11-03T11:44:31 and that's extremely tricky 2010-11-03T11:44:39 yup 2010-11-03T11:44:45 that's what i'm banking on anyway 2010-11-03T11:44:46 state is extremely difficult to get exactly right 2010-11-03T11:44:51 it'll cost too much when you can get so much more out of a good heuristic based off looking ahead. 2010-11-03T11:44:55 most of my advantage thus far has been superior state calc 2010-11-03T11:45:06 but now that everybody is mostly caught up i have to go further lol 2010-11-03T11:45:19 demerzel_: what makes it superior? 2010-11-03T11:45:27 well just being accurate for one haha 2010-11-03T11:45:28 getting it correct 2010-11-03T11:45:35 the details are very messy 2010-11-03T11:45:40 people needed time to work it out 2010-11-03T11:45:41 but i have some numbers that most people wouldn't think of 2010-11-03T11:45:43 *** Sarin has joined #aichallenge 2010-11-03T11:45:47 for weeks, many bots were doing calculations wrong 2010-11-03T11:45:50 had them pretty early 2010-11-03T11:45:55 now they're pretty standard 2010-11-03T11:45:56 but I think we're mostly getting on the level with that now 2010-11-03T11:46:07 neutral resolution i think will be the biggest mistake 2010-11-03T11:46:16 and amount to claim neutral 2010-11-03T11:46:24 JamesMG: hehe, I'm not surprised. I came in here this morning to confirm something in the spec. which seemed to be odd in my Java player... the response I got didn't inspire me with confidence :-) 2010-11-03T11:46:35 AJC: the players are almost all bugged 2010-11-03T11:46:44 even iouri_'s caused me hours of debugging problems that didn't exist 2010-11-03T11:46:45 except bocksimacko 2010-11-03T11:46:47 demerzel_: the players or the game simulation? 2010-11-03T11:46:55 AJC: simulators 2010-11-03T11:46:57 antimatroid: which is that? 2010-11-03T11:47:01 that's why i build tcp directly into my bot 2010-11-03T11:47:02 demerzel_: how do you test then! 2010-11-03T11:47:08 ^^ 2010-11-03T11:47:15 i just wrap io 2010-11-03T11:47:21 can hook it into file/tcp/stdin 2010-11-03T11:47:24 AJC: http://72.44.46.68/getplayer?player=bocsimacko-b0.46 2010-11-03T11:47:33 that bot doesn't fuck around 2010-11-03T11:47:43 wow, it lost one 2010-11-03T11:47:49 AJC: I mean, he must use a version of PlayGame.jar locally that works? 2010-11-03T11:47:52 erm 2010-11-03T11:48:01 I mean "antimatroid", damn autocomplete 2010-11-03T11:48:03 http://72.44.46.68/canvas?game_id=838396 shampoo beat him without just clinching it on expansion 2010-11-03T11:48:18 *** lavalamp__ has joined #aichallenge 2010-11-03T11:48:18 impressive! 2010-11-03T11:48:27 how was that game decided? 2010-11-03T11:48:41 AJC: he's been untouchable for a while now 2010-11-03T11:48:43 * mega1 crowdsource debugging 2010-11-03T11:48:45 demerzel_: so you built TCP into your bot and you only test it online? 2010-11-03T11:48:46 +s 2010-11-03T11:48:56 demerzel_: no local simulation because it's potentially buggy? 2010-11-03T11:49:12 AJC: basically i go online, log all stdin to a file 2010-11-03T11:49:21 mega1: my problems with shampoo are from him taking the centre i think 2010-11-03T11:49:21 if the game goes bad then i hook into the file and step through in IDE 2010-11-03T11:49:24 can't we use the server's simulation then? 2010-11-03T11:49:32 he's a constant problem of mine :P 2010-11-03T11:49:47 don't help mega1 any more haha 2010-11-03T11:49:51 he seems to do it very well 2010-11-03T11:49:56 good to see somebody can beat him occasionally ;) 2010-11-03T11:50:16 meh, i'm interested in having lots of good bots at the end 2010-11-03T11:50:27 antimatroid: shampoo expands and defends well 2010-11-03T11:50:30 oh does mega1 still have a ways to go till he's good? 2010-11-03T11:50:42 mega1 is bocksimacko :P 2010-11-03T11:50:46 keep trying you'll get there 2010-11-03T11:50:51 hahaha 2010-11-03T11:50:56 if one loses the opening shampoo will punish him 2010-11-03T11:51:25 mega1: nup, i am terrible at my openings (1 move) and i beat him most of the time 2010-11-03T11:52:28 cool i got a mega1 stdin to test against lol 2010-11-03T11:52:38 something is horribly wrong :( 2010-11-03T11:52:44 then I don't know 2010-11-03T11:52:58 wow my bot is on an impressive losing streak: http://ai-contest.com/profile.php?user_id=10821 2010-11-03T11:53:11 have you considered it being him taking the middle? 2010-11-03T11:53:21 i've only recently started to think that's why he causes me so much trouble 2010-11-03T11:53:37 it doesn't seem like it was smart of him, but he always seems to do it and claim it 2010-11-03T11:53:53 AJC: there is a forum thread with updated an update playgame.jar file. http://ai-contest.com/forum/viewtopic.php?f=18&t=534 2010-11-03T11:54:17 dmj111: can someone suggest posting this on the frontpage and updating the SDK? 2010-11-03T11:54:34 i'm sure one of them works and isn't bugged 2010-11-03T11:54:45 i've just had 3 bad ones so far so decided <100 lines of code was worth the hassle 2010-11-03T11:55:03 I just use a local one to get rid of any obvious bugs and crashes 2010-11-03T11:55:09 then throw it on TCP for actual testing 2010-11-03T11:56:05 dmj111: ah, I think I have those already. have those been confirmed to be the same simulation as on the server?\ 2010-11-03T11:56:37 *** songoftheweave has joined #aichallenge 2010-11-03T11:59:50 AJC: I don't know. I used https://github.com/albertz/planet_wars-cpp for a couple of weeks until I wrote my own to use. 2010-11-03T12:00:05 I haven't used a java version for a month. 2010-11-03T12:00:45 dmj111: if it's a drop-in replacement can you run two games locally and compare the output? 2010-11-03T12:00:58 sorry to bother, but this stuff is rather crucial :-) 2010-11-03T12:01:17 the C++ runner is, yes. 2010-11-03T12:01:55 *** Blkt has quit IRC (Quit: Error: do not makunbound t please!) 2010-11-03T12:03:50 this is sad but i still have no idea why my bot streams 2010-11-03T12:03:52 i have no dist code 2010-11-03T12:04:23 i thought it was just stuck on a target but in this game i clearly am streaming from one planet to another http://72.44.46.68/canvas?game_id=838680 2010-11-03T12:04:27 and i only allow it 1 target per turn 2010-11-03T12:05:25 must have a nonnegative return for sending to another planet or something :( 2010-11-03T12:14:47 Top 10 players: bocsimacko(4224), dmj111(4125), bix0r4ever(4107), Hazard(4062), george(4053), GreenTea(4031), shangas(4020), asavis(3997), felixcoto(3997), Accoun(3962) 2010-11-03T12:15:39 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-11-03T12:30:07 *** boegel has quit IRC (Ping timeout: 245 seconds) 2010-11-03T12:35:27 *** songoftheweave has quit IRC (Quit: songoftheweave) 2010-11-03T12:36:35 *** fawek has joined #aichallenge 2010-11-03T12:43:21 *** flavor has joined #aichallenge 2010-11-03T12:43:52 *** flavor has quit IRC (Remote host closed the connection) 2010-11-03T12:44:11 *** hightime has joined #aichallenge 2010-11-03T12:44:22 hi all 2010-11-03T12:45:56 hi 2010-11-03T12:46:52 Man, shampoo is the bane of my bot's existance 2010-11-03T12:47:03 same for me 2010-11-03T12:47:04 his bot expands under pressure very well 2010-11-03T12:47:21 i don't know why he isn't ranked higher 2010-11-03T12:47:30 http://72.44.46.68/canvas?game_id=839187 2010-11-03T12:47:41 besides my weak defense code, check how his expands during the melee 2010-11-03T12:47:51 *** songoftheweave has joined #aichallenge 2010-11-03T12:49:22 yeah shampoo is a weird piece of work =) 2010-11-03T12:49:47 SO, I attack neutral planet with 35 ships. My enemy attacks the same neutral planet with 31 ships. The neutral planet has 40 ships. What will happen? 2010-11-03T12:50:04 neutral ends with 5 ships 2010-11-03T12:50:12 haha damn i finally found my def bug 2010-11-03T12:50:13 ok, thx 2010-11-03T12:50:37 it kept streaming to planets with defense needed until that turn cuz i had it thinking it was sending ships to itself not the planet it was defending 2010-11-03T12:50:53 I've got a weird defense problem now 2010-11-03T12:50:57 is shampoo ever on irc 2010-11-03T12:51:01 or is he another name like mega 2010-11-03T12:51:02 hightime: the specs are @ the website, read them carefully 2010-11-03T12:51:06 but the bot is doing better anyways 2010-11-03T12:51:22 never heard anybody claim shampoo demerzel_ 2010-11-03T12:51:30 and it's come up a few times 2010-11-03T12:53:08 I love it when this happens, it looks great: http://72.44.46.68/canvas?game_id=839257 2010-11-03T12:55:06 =) 2010-11-03T12:55:30 nice 2010-11-03T12:55:31 this is going to be a fun bug to track down: http://72.44.46.68/canvas?game_id=839263 2010-11-03T12:56:16 *** AJC has quit IRC (Ping timeout: 276 seconds) 2010-11-03T12:56:30 *** songoftheweave has quit IRC (Quit: songoftheweave) 2010-11-03T12:57:31 yeeeeehaw, i made my bot stronger again 2010-11-03T12:57:34 fixed sniping 2010-11-03T12:57:42 \o\ |o| /o/ 2010-11-03T12:59:15 quite a lot stronger actually 2010-11-03T12:59:18 holy crap 2010-11-03T12:59:29 i might beat bhasker now 2010-11-03T13:00:15 antimatroid: your heuristic bot seems to be doing well 2010-11-03T13:08:45 *** fluffluff1 has joined #aichallenge 2010-11-03T13:08:46 *** fluffluff has quit IRC (Ping timeout: 276 seconds) 2010-11-03T13:11:01 *** AJC has joined #aichallenge 2010-11-03T13:11:44 james you're supposed to attack me 2010-11-03T13:13:26 did we just play? 2010-11-03T13:13:45 your old version 2010-11-03T13:13:56 i am just messing with defense code but you never attacked me hehe 2010-11-03T13:14:23 haha, one of those games 2010-11-03T13:14:48 Top 10 players: bocsimacko(4227), dmj111(4108), bix0r4ever(4103), george(4031), Hazard(4024), GreenTea(4019), shangas(4010), asavis(3989), felixcoto(3980), Accoun(3943) 2010-11-03T13:18:10 *** Gudradain has quit IRC (Quit: Page closed) 2010-11-03T13:18:47 crap more state problems yay 2010-11-03T13:19:35 I'm having some of that myself 2010-11-03T13:21:56 my crappy defense code is killin me: http://72.44.46.68/canvas?game_id=839579 2010-11-03T13:22:19 had that game on turn 2, threw it away 2010-11-03T13:24:57 i actually made an invalid order :( thought i cleared all that crap up 2010-11-03T13:25:34 *** Bobng has joined #aichallenge 2010-11-03T13:28:39 *** TravisD has joined #aichallenge 2010-11-03T13:29:44 man i am getting really sick of trying to do stuff the c++ way and having it shit on me 2010-11-03T13:29:50 classes going out of scope, references disappearing 2010-11-03T13:30:08 hehe, what language do you normally use? 2010-11-03T13:30:23 i use c++ but use it like it's c 2010-11-03T13:30:29 with optional extras ;) 2010-11-03T13:30:34 my day job is C# 2010-11-03T13:32:06 class gameState -> struct gameState = problem solved 2010-11-03T13:32:19 despite having a global instance >:( 2010-11-03T13:34:06 *** bhasker has joined #aichallenge 2010-11-03T13:35:18 blargh i am going to just make this C compliant i am sick of tracking down disappearing references 2010-11-03T13:43:21 *** TravisD has quit IRC (Quit: Leaving.) 2010-11-03T13:50:27 demerzel_: the only difference between class and struct in c++ is the default access specifier 2010-11-03T13:50:35 sruct is public, class is private 2010-11-03T13:51:27 demerzel_: if you have to worry about references that disappear you're doing it wrong 2010-11-03T13:52:27 if you need resources with no cl/ear ownership, go for some kind of smart pointer, but I'd suggest clarifying the ownership and pre-allocating for performance. 2010-11-03T13:52:42 if I was to rewrite my bot in C++ I'd just pre-allocate all the arrays upfront 2010-11-03T13:52:53 and refer to things with identifiers 2010-11-03T13:52:59 that is exactly what i do 2010-11-03T13:53:20 well whatever your definitino of identifier is 2010-11-03T13:53:30 basically have structs with pre-allocated arrays before i ever hit turn 1 2010-11-03T13:53:31 index 2010-11-03T13:53:39 how come references are a problem? 2010-11-03T13:53:44 i have no idea 2010-11-03T13:53:49 semantically i'm not doing anything wrong 2010-11-03T13:53:53 it just fails half the time 2010-11-03T13:53:56 hmm 2010-11-03T13:53:57 probably because i'm using vs/vc 2010-11-03T13:54:19 GCC with full warnings catches some problems. 2010-11-03T13:55:12 *** fawek has quit IRC (Remote host closed the connection) 2010-11-03T13:56:40 i'm not worried about it i am just tired of thinking i have a calc error and finding a reference issue and/or a mysterious global instance descoping 2010-11-03T13:56:50 if i make it c-compliant i know i'll have zero issues lol 2010-11-03T13:57:16 did you overwrite your array accidentally? 2010-11-03T13:57:36 if your reference is invalid changing it to a pointer will probably not fix that... 2010-11-03T13:57:57 its a lot harder to screw up a ref than a pointer 2010-11-03T14:00:10 *** superflit has quit IRC (Quit: superflit) 2010-11-03T14:02:23 phreeza: you're probably right but i am just tired of dealing with classes etc when essentially i prealloc everything at the start 2010-11-03T14:02:44 was thinking about going pure C anyway just cuz there's only a dozen or so bots with it atm ;) 2010-11-03T14:03:16 hehe, so that would make you more esotheric? 2010-11-03T14:03:49 are bros "coding" in C more "authentic"? 2010-11-03T14:04:08 nah 2010-11-03T14:04:18 but yea to esoteric ;) 2010-11-03T14:04:50 *** narnach has joined #aichallenge 2010-11-03T14:10:17 *** superflit has joined #aichallenge 2010-11-03T14:10:17 *** Florian95 has quit IRC (Remote host closed the connection) 2010-11-03T14:14:49 Top 10 players: bocsimacko(4232), jimrogerz(4179), bix0r4ever(4119), dmj111(4097), Raschi(4038), george(4036), GreenTea(4029), Hazard(4026), shangas(4003), felixcoto(3978) 2010-11-03T14:25:06 *** hellman has quit IRC (Remote host closed the connection) 2010-11-03T14:27:36 bhasker: coudl you put your bot back in tcp please? 2010-11-03T14:27:48 my bot is in a crappy state 2010-11-03T14:27:50 i will start it up 2010-11-03T14:27:54 i am going to rewrite part of it 2010-11-03T14:28:04 its doing horribly right now 2010-11-03T14:28:07 de-crappyfy it? ^_^ 2010-11-03T14:28:21 nah i am just gonna rewrite those parts 2010-11-03T14:28:26 and i want to see how this change does 2010-11-03T14:28:30 it wasn't doing too bad either 2010-11-03T14:28:39 i made it a tad bit too conservative 2010-11-03T14:29:09 ok 2010-11-03T14:30:24 goddamnit i hate it when i upload my bot to the official server 2010-11-03T14:30:31 and it loses from a noob 2010-11-03T14:30:32 >_< 2010-11-03T14:30:53 that exploits my weakness :P 2010-11-03T14:31:00 which i don't know howto properly fix yet 2010-11-03T14:31:07 hehe 2010-11-03T14:31:10 RageBot? 2010-11-03T14:32:11 *** Migi32 has joined #aichallenge 2010-11-03T14:32:13 nah 2010-11-03T14:32:17 dunno 2010-11-03T14:32:18 some guy 2010-11-03T14:32:22 oh, improvements to the visualiser. Nice! 2010-11-03T14:33:10 oh? 2010-11-03T14:33:22 what improvements` 2010-11-03T14:33:26 http://ai-contest.com/visualizer.php?game_id=6573242 2010-11-03T14:33:35 ah official 2010-11-03T14:33:39 yea 2010-11-03T14:33:48 *** boegel has joined #aichallenge 2010-11-03T14:34:00 faster too 2010-11-03T14:34:08 the speed controls and growth graph, really awesome 2010-11-03T14:35:11 speed controls are not working here 2010-11-03T14:35:35 working for me 2010-11-03T14:35:37 chrome 2010-11-03T14:35:39 http://72.44.46.68/canvas?game_id=840416 fuuuuuuuuuuuuuuuuuuuuck yeah 2010-11-03T14:37:25 haha, Error323, good job :) 2010-11-03T14:37:31 graph looks nice too 2010-11-03T14:37:45 yea :) 2010-11-03T14:38:04 =] 2010-11-03T14:38:11 but seriously that shampoo guy 2010-11-03T14:38:30 omg he is a menace too all but bocsimacko :P 2010-11-03T14:39:05 is the gcc they use c99? i assume but you never know lol 2010-11-03T14:39:37 Error323, you appear to attack in this pattern: 2 big attacks, one small attack, then one really small attack, then repeat 2010-11-03T14:39:46 yeah 2010-11-03T14:39:59 basically i do this 2010-11-03T14:40:28 compute frontline 2010-11-03T14:40:32 *** boegel has quit IRC (Ping timeout: 245 seconds) 2010-11-03T14:40:48 underestimate enemyplanet 2010-11-03T14:40:53 and attack with all i got 2010-11-03T14:41:03 but when the distance is very big 2010-11-03T14:41:11 it fails horribly 2010-11-03T14:41:26 and i'm gonna fix that now 2010-11-03T14:41:30 cuz its driving me insane 2010-11-03T14:41:31 I just do this: if I have less total growth than the enemy, expand. Otherwise, pick a planet and go for it with everything you've got 2010-11-03T14:41:45 oh nice their compile_anything takes o haha 2010-11-03T14:42:06 demerzel_, ? 2010-11-03T14:42:08 o? 2010-11-03T14:42:14 haha 2010-11-03T14:42:16 compiled object files 2010-11-03T14:42:18 as in object files? 2010-11-03T14:42:20 oh cool 2010-11-03T14:42:29 so you don't even have to submit source files :P 2010-11-03T14:42:32 nope 2010-11-03T14:42:33 thats great 2010-11-03T14:42:34 Migi32: attacking when you are behind can give serious advantage 2010-11-03T14:42:45 since there is no cost of capping the neutral 2010-11-03T14:42:50 i guess theyll fix that though? 2010-11-03T14:42:57 they were talking about allowing executables 2010-11-03T14:42:57 um, nope 2010-11-03T14:42:59 i doubt it's an issue 2010-11-03T14:43:00 Error323, if I'm behind I assume they can hold though 2010-11-03T14:43:08 all the .o files are nuked before compilation 2010-11-03T14:43:22 probably an oversight since they don't include obj but i don't think they care 2010-11-03T14:43:27 I know it's not a great solution though, Error323 2010-11-03T14:43:37 that's why you don't see me on top of TCP :) 2010-11-03T14:44:38 well i'm getting there... 2010-11-03T14:44:44 except for that massive flaw 2010-11-03T14:44:46 Zannick: nuked? what does that mean specifically in this case 2010-11-03T14:44:54 if you actually manage to get a .o file without a corresponding source file to link, i would like to know 2010-11-03T14:45:11 demerzel_: the script performs the equivalent of rm -rf *.o MyBot 2010-11-03T14:45:13 yea 2010-11-03T14:45:31 iirc linux that's delete *.o? 2010-11-03T14:45:42 well if you really want you can include the executable then have your main bot read that executable to memory and call the main() function inside it 2010-11-03T14:45:50 yeah, rm = remove 2010-11-03T14:46:13 btw don't hit enter after "rm -rf *" 2010-11-03T14:46:15 oh so essentially they delete all the *.o then make new ones 2010-11-03T14:46:22 yeah 2010-11-03T14:46:28 -recursive force 2010-11-03T14:46:32 my fave 2010-11-03T14:46:41 i once did this : rm -rf . 2010-11-03T14:46:43 rm -rf ~ 2010-11-03T14:46:51 that messes ur system up pretty good 2010-11-03T14:46:55 touch -- * 2010-11-03T14:47:02 rm -- * 2010-11-03T14:47:12 ^ how to get rid of those tricky files 2010-11-03T14:47:15 lol 2010-11-03T14:47:15 Zannick: well thanks for making me look twice 2010-11-03T14:47:28 demerzel_: no problem 2010-11-03T14:47:56 while we're at it i assume gcc is c99? afa ansi-c 2010-11-03T14:47:59 i organized most of that script, so i'm familiar with how it works 2010-11-03T14:48:00 it's best to rtfm before you rm -rf anything... 2010-11-03T14:48:10 rtfm -rf 2010-11-03T14:48:40 should be c99, yeah 2010-11-03T14:48:54 if youreally care about not having your code read I'm sure there are obfuscaters out there for your language though 2010-11-03T14:48:59 it's the default in the gcc being used 2010-11-03T14:49:32 unless you're using Lisp, then you don't really need an obfuscater ;) 2010-11-03T14:49:46 oh whoa wtf inline = static? 2010-11-03T14:49:51 i didn't know that 2010-11-03T14:49:53 *** TravisD has joined #aichallenge 2010-11-03T14:50:06 inline most assuredly != static 2010-11-03T14:50:08 um, no 2010-11-03T14:50:56 but you have to have a declaration of the inline function to include it elsewhere, like anything else 2010-11-03T14:51:06 hmm some article says you have to do this for c89 # define inline static 2010-11-03T14:51:35 for a while i was using static inline functions defined in an h file until i said screw it and moved it into a c file 2010-11-03T14:51:39 i presumed that meant they were equivalent which was news to me 2010-11-03T14:52:18 if anything they are opposite 2010-11-03T14:52:49 yea that's what i thought 2010-11-03T14:53:04 i was like uh i thought static meant 1 instance and inline means 'make an instance anywhere it's used' 2010-11-03T14:53:29 substituting instance for a more appropriate word 2010-11-03T14:54:04 yeah that's the idea 2010-11-03T14:54:11 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-11-03T14:55:02 *** TravisD has left #aichallenge 2010-11-03T14:55:15 okey so that article was wrong or that's some mighty hack 2010-11-03T14:55:55 I can't imagine why you would do that, either there's some weird C thing or that guy was on something 2010-11-03T14:55:57 static is "no other file can use this" 2010-11-03T14:56:18 inline is "insert this everywhere it's used instead of calling it" 2010-11-03T14:56:25 essentially 2010-11-03T14:56:54 Zannick: thats my understanding of inline, but what do youmean by file in static? 2010-11-03T14:57:17 that's the meaning of static in front of global variables 2010-11-03T14:57:23 but i'm talking more static members 2010-11-03T14:57:33 http://72.44.46.68/canvas?game_id=840681 second win \o/ 2010-11-03T14:57:45 then i don't know 2010-11-03T14:57:52 that's my understanding of static 2010-11-03T14:57:52 here my flaw is my strength 2010-11-03T14:57:54 static members are on a per class basis instead of a per object basis 2010-11-03T14:58:09 *** mceier has joined #aichallenge 2010-11-03T14:58:12 oh, c++ 2010-11-03T14:58:17 idk why c89 would care about it tho 2010-11-03T14:58:24 since we're talking c and there are no member funcs 2010-11-03T14:58:30 yeah, i'm thinking only of global variables and functions, wrt c 2010-11-03T14:58:31 so you're right lol 2010-11-03T14:58:40 ah ok 2010-11-03T14:59:02 i guess inline in that case would make c89 expand it in only that obj file 2010-11-03T14:59:08 which essentially would be the same thing' 2010-11-03T14:59:22 *** RainCT has joined #aichallenge 2010-11-03T14:59:35 ugh i keep hitting shampoo guy 2010-11-03T15:00:03 is he here? 2010-11-03T15:00:13 he doesn't exist it seems 2010-11-03T15:00:15 :P 2010-11-03T15:00:18 never seen him 2010-11-03T15:00:30 btw AJC what are you doing outside of france? 2010-11-03T15:00:45 Error323: hehe, why do you ask? 2010-11-03T15:00:54 is leaving the country forbidden? 2010-11-03T15:01:03 ofc not, just curious :) 2010-11-03T15:01:10 you know who I am right? 2010-11-03T15:01:46 no 2010-11-03T15:01:49 maybe that's the problem :-) 2010-11-03T15:01:56 the intern @ guerilla games 2010-11-03T15:02:07 aha 2010-11-03T15:02:10 :-) 2010-11-03T15:02:16 I live in Austria. 2010-11-03T15:03:01 I see 2010-11-03T15:04:13 you using C++? 2010-11-03T15:04:18 yes 2010-11-03T15:05:43 previous version ranked around 60 @ official 2010-11-03T15:05:53 having much higher "hopes" for this one though 2010-11-03T15:05:56 *** hightime has quit IRC (Read error: Connection reset by peer) 2010-11-03T15:06:56 Error323: nice! 2010-11-03T15:07:17 ty 2010-11-03T15:07:51 *** perror has quit IRC (Quit: Bye all !) 2010-11-03T15:11:14 *** superflit has quit IRC (Quit: superflit) 2010-11-03T15:11:28 meen 2010-11-03T15:11:36 i keep winning from iouri, george etc 2010-11-03T15:11:42 but that shampoo guy 2010-11-03T15:11:45 and ofc bhasker :P 2010-11-03T15:11:51 :D 2010-11-03T15:11:58 though my current version is horrible 2010-11-03T15:12:06 good 2010-11-03T15:12:10 it lost like 8 games in a row to shampoo 2010-11-03T15:12:11 http://72.44.46.68/getplayer?player=malazanbot0.15 2010-11-03T15:12:16 yeah 2010-11-03T15:12:21 thats bad 2010-11-03T15:12:28 though shampoo is wel... 2010-11-03T15:12:30 weird 2010-11-03T15:12:32 ^^ 2010-11-03T15:13:35 but yeah my elo at tcp is hovering around 340 2010-11-03T15:14:08 *** superflit has joined #aichallenge 2010-11-03T15:14:18 *** Azrathud has joined #aichallenge 2010-11-03T15:14:48 Top 10 players: bocsimacko(4230), jimrogerz(4205), bix0r4ever(4119), dmj111(4095), george(4035), GreenTea(4028), Hazard(4022), shangas(4010), felixcoto(3975), asavis(3975) 2010-11-03T15:20:40 i'm beating iouri more consistently now \o/ 2010-11-03T15:20:42 i would really like to build a bot that can learn.... but i don't have a good idea how to do it 2010-11-03T15:21:07 if you have parameters to tweak 2010-11-03T15:21:11 almost everyone over 4k 2010-11-03T15:21:15 learning could be tweaking those parameters 2010-11-03T15:21:42 true... but i was thinking more explicit learning sonmehow 2010-11-03T15:21:52 like learning from mistakes 2010-11-03T15:21:58 ah 2010-11-03T15:21:58 case based 2010-11-03T15:22:10 something like that 2010-11-03T15:22:27 thats more complex 2010-11-03T15:22:32 well even if you manage to create sth like that 2010-11-03T15:22:58 you would need the tcp server, since the official doesn't allow file writing 2010-11-03T15:23:03 the trick is to define something useful from a series of states 2010-11-03T15:23:07 leaving you with less than a month 2010-11-03T15:23:11 yea or use the logs from official 2010-11-03T15:23:44 afaik, the only things i can think of could easily be calculated 2010-11-03T15:24:13 and most other things are pretty useless as they depend on the map more than anything 2010-11-03T15:24:22 18th @ tcp whoohoo 2010-11-03T15:24:26 ^^ 2010-11-03T15:24:40 you are probably right, raedwulf 2010-11-03T15:24:43 sorry i'm a bit excited :P 2010-11-03T15:24:59 Error323: damn it, i need to find time to write my bot 2010-11-03T15:25:12 but im busy with some uni work... also just started a relationship which doesn't help with time xD 2010-11-03T15:25:54 *** Azrathud has quit IRC (Ping timeout: 240 seconds) 2010-11-03T15:27:15 this map is really fail http://72.44.46.68/canvas?game_id=840978 2010-11-03T15:27:30 raedwulf: heh yeah I need to get back to my studies aswell 2010-11-03T15:28:07 raedwulf: could maybe use some kind of mesh distortion metric to determine how similar 2 maps are 2010-11-03T15:28:24 phreeza: yes, but is it worth it :P 2010-11-03T15:28:34 that is the question! 2010-11-03T15:28:36 :) 2010-11-03T15:28:36 ^^ 2010-11-03T15:28:41 lol 2010-11-03T15:28:50 well i would learn something along the way 2010-11-03T15:28:56 yep 2010-11-03T15:29:02 and i guess thats the real point of this whole excercise 2010-11-03T15:29:07 at least to me 2010-11-03T15:29:14 im going for simple parameter tweaking (actually lotsa parameter tweaking) 2010-11-03T15:29:23 and might try and learn that way 2010-11-03T15:29:25 hey a C# program in 2nd place 2010-11-03T15:29:38 nice 2010-11-03T15:29:51 yuck 2010-11-03T15:30:09 haha 2010-11-03T15:31:06 of course he is microsoft, they probably hired him to save C#'s rep :) 2010-11-03T15:31:18 haha 2010-11-03T15:32:30 haha 2010-11-03T15:34:58 *** boegel has joined #aichallenge 2010-11-03T15:36:46 any admins here who could look into this game? 2010-11-03T15:36:47 http://ai-contest.com/visualizer.php?game_id=6570812 2010-11-03T15:37:06 I would be highly surprised if my bot timed out over this game: http://ai-contest.com/visualizer.php?game_id=6570812 2010-11-03T15:39:20 boegel: unfortunately the information shown in the visualizer is really the full information we have from the game 2010-11-03T15:39:36 janzert: ah, k, thx 2010-11-03T15:39:52 janzert: why not differentiate between crash/time-out/did not start? 2010-11-03T15:40:02 janzert: that would make a whole lot of sense :) 2010-11-03T15:40:24 basically because they all look the same to the engine, it didn't get a move within 1 second 2010-11-03T15:40:57 of course the error message could not mention did not start when it is after the first turn :) 2010-11-03T15:48:14 hmmk 2010-11-03T15:48:22 I'm very puzzled what happened 2010-11-03T15:49:31 *** choas has joined #aichallenge 2010-11-03T15:49:35 I don't see any change in state compared to the previous turns for example 2010-11-03T15:49:38 weird :-/ 2010-11-03T15:49:56 could just have been some daemon starting, causing the system to slow down? 2010-11-03T15:50:07 * boegel joins phreeza 2010-11-03T15:50:30 I've seen (what I believe to be) freak time-outs before 2010-11-03T15:50:32 hm? :) 2010-11-03T15:50:39 although they used to be more frequent 2010-11-03T15:50:56 phreeza: I join you in suspecting some exterior cause, and my bot being blamed 2010-11-03T15:51:06 ah ok :) 2010-11-03T15:51:11 it should run in like 0.1s per turn 2010-11-03T15:51:15 yea they shouldn't be using wallclock time 2010-11-03T15:53:22 they're using wall clock time? 2010-11-03T15:53:24 * boegel gasps 2010-11-03T15:53:34 that really doesn't make sense people... 2010-11-03T15:53:36 yeap 2010-11-03T15:53:43 i think so... 2010-11-03T15:54:23 albert told me that was what it seemed like to him 2010-11-03T15:54:48 he has a searching bot andf it was timing out before he changed his abortion criterium to wallclock time 2010-11-03T15:54:49 yes, it is wall clock 2010-11-03T15:56:46 janzert: any particular reasoning behind using wall clock? 2010-11-03T15:57:39 that was done way before I was around (it was the same in the last contest) but I think it's mostly just because it was easiest to do 2010-11-03T15:58:16 trying to get accurate cpu time for a child process is not straightforward 2010-11-03T15:58:25 hmm 2010-11-03T15:58:44 I don't have a cut-n-clear answer for that, I must admit 2010-11-03T15:59:08 also of course you *still* must have a wall time limit 2010-11-03T15:59:27 so someone can't does the server with something like time.sleep(999) 2010-11-03T15:59:42 s/does/dos 2010-11-03T16:00:12 the other problem with cpu time is that it doesn't coun't I/O wait time 2010-11-03T16:00:34 so a bot that does lots of disk reading would have a relative advantage 2010-11-03T16:01:32 heh, my fingers are sticking apostrophes in between all nt pairs 2010-11-03T16:01:43 lol 2010-11-03T16:03:55 does anyone know where testingfrompiter is at officially? 2010-11-03T16:04:16 * boegel resubmits his bot to see if it occurs again 2010-11-03T16:04:50 janzert: why would IO matter? you're not allowed to write to file, right? 2010-11-03T16:05:03 not write but you can read 2010-11-03T16:05:11 say lot's of precomputed data 2010-11-03T16:06:00 janzert: where would you read it from? 2010-11-03T16:06:21 janzert: it would have to be embedded in your bot's binary 2010-11-03T16:06:35 you can upload data files along with your bot 2010-11-03T16:06:35 janzert: and the file size of your submission is limited 2010-11-03T16:06:43 yup 2010-11-03T16:06:44 janzert: I don't think IO would be an issue 2010-11-03T16:07:04 janzert: and these will be copied every tine your bot runs somewhere, alongside your binary? 2010-11-03T16:07:09 yes 2010-11-03T16:10:41 meen 2010-11-03T16:11:14 i won like 4 out of 12 from bocsimacko ::) 2010-11-03T16:11:49 http://72.44.46.68/canvas?game_id=841440 2010-11-03T16:11:50 :D 2010-11-03T16:11:55 pwnd 2010-11-03T16:14:10 *** sigh has joined #aichallenge 2010-11-03T16:14:50 Top 10 players: bocsimacko(4228), jimrogerz(4215), bix0r4ever(4137), dmj111(4100), george(4040), asavis(4040), Hazard(4038), GreenTea(4016), shangas(4004), felixcoto(3959) 2010-11-03T16:15:09 top9 over 4k 2010-11-03T16:16:38 * boegel would love to get into the top 100 2010-11-03T16:16:45 how hard is that nowadays? 2010-11-03T16:18:19 it can probably be done if you put in the time and build a heuristic bot, and iteratively refine it by analyzing mistakes it made 2010-11-03T16:18:23 how can we answer that :P 2010-11-03T16:18:49 phreeza: that's exactly what I'm doing :) 2010-11-03T16:18:58 phreeza: the top bots aren't heuristic? 2010-11-03T16:19:05 i think they all are 2010-11-03T16:19:11 pretty much 2010-11-03T16:19:14 then how in the world are you timing out boegel 2010-11-03T16:19:15 ah, that ives hope :) 2010-11-03T16:19:16 ? 2010-11-03T16:19:23 Error323: good question :) 2010-11-03T16:19:39 Error323: hence me blaming freak server time-outs 2010-11-03T16:19:57 Error323: I am analyzing fleets and such, which can get expensive if there are lot's of them 2010-11-03T16:20:26 uhhh 2010-11-03T16:20:27 Error323: but even then, timing out is rare 2010-11-03T16:20:35 *very rare* 2010-11-03T16:20:51 Error323: also, Haskell isn't the fastest language ever ;) 2010-11-03T16:20:56 ahh 2010-11-03T16:23:03 * phreeza is going to get pizza 2010-11-03T16:27:20 *** Accoun has quit IRC () 2010-11-03T16:33:19 *** Bobng has quit IRC (Read error: Connection reset by peer) 2010-11-03T16:41:41 *** Accoun has joined #aichallenge 2010-11-03T16:49:35 *** wdouglas has joined #aichallenge 2010-11-03T16:52:31 *** Bobng has joined #aichallenge 2010-11-03T16:53:24 *** sigh has quit IRC (Remote host closed the connection) 2010-11-03T16:53:27 bot unnerfed 2010-11-03T16:53:31 malazan0.16 is up 2010-11-03T16:53:36 lets see how it does 2010-11-03T17:03:22 *** Sarin has quit IRC (Remote host closed the connection) 2010-11-03T17:03:56 *** Cold-Phoenix has joined #aichallenge 2010-11-03T17:06:40 *** kersny has joined #aichallenge 2010-11-03T17:14:50 Top 10 players: bocsimacko(4230), bix0r4ever(4133), jimrogerz(4102), dmj111(4096), asavis(4061), george(4053), Hazard(4037), GreenTea(4018), shangas(4014), felixcoto(3960) 2010-11-03T17:22:35 *** boegel has quit IRC (Quit: Leaving) 2010-11-03T17:23:06 *** hornairs has joined #aichallenge 2010-11-03T17:25:23 *** choas has quit IRC (Quit: leaving) 2010-11-03T17:26:00 bhasker: good 2010-11-03T17:26:10 cuz i'm going through the roof without ur bot :P 2010-11-03T17:26:11 its not doing so well oh well 2010-11-03T17:26:15 losing to shampoo 2010-11-03T17:26:42 i guess i need to fix a few more things before it does better 2010-11-03T17:26:48 your bot seems to be doing much better 2010-11-03T17:26:50 so what did you change 2010-11-03T17:26:55 399 elo, playing for 3.5 hours now 2010-11-03T17:27:01 better enemy strength predictino 2010-11-03T17:27:10 and effectively enabled sniping 2010-11-03T17:27:23 *** Appleman1234 has joined #aichallenge 2010-11-03T17:27:36 nice 2010-11-03T17:28:06 *** kersny has quit IRC (Quit: I'm Out) 2010-11-03T17:28:14 i've won about 4 out of 10 vs bocsimaco 2010-11-03T17:28:20 nice 2010-11-03T17:28:28 not sure exactly but i think around those numbers 2010-11-03T17:28:34 i should count it heh 2010-11-03T17:33:48 ow and a new valuefunction 2010-11-03T17:34:06 44 double GetValue(Planet& p, int dist) { 2010-11-03T17:34:06 45 #define EPS 1.0e-10 2010-11-03T17:34:06 46 int smallestDist = std::numeric_limits::max(); 2010-11-03T17:34:06 47 for (unsigned int i = 0, n = bot::gAP->size(); i < n; i++) 2010-11-03T17:34:06 48 { 2010-11-03T17:34:08 49 const Planet& n = bot::gAP->at(i); 2010-11-03T17:34:11 50 if (n.Owner() == 0) 2010-11-03T17:34:13 51 { 2010-11-03T17:34:16 52 int dist = p.Distance(n); 2010-11-03T17:34:18 53 smallestDist = std::min(smallestDist, dist); 2010-11-03T17:34:21 54 } 2010-11-03T17:34:23 55 } 2010-11-03T17:34:26 56 return (pow(p.GrowthRate(), 2.0) / (p.NumShips() * dist + EPS)) - (EPS * pow(smallestDist, 2.0)); 2010-11-03T17:34:29 57 } 2010-11-03T17:34:32 there :P 2010-11-03T17:34:57 Error323: please use pastebin next time 2010-11-03T17:35:07 sorry amstan 2010-11-03T17:35:08 Error323: it's really poor form to paste stuff like thins on irc 2010-11-03T17:35:37 just wanted to be of some assistance 2010-11-03T17:35:41 but indd 2010-11-03T17:36:36 EPS== ? 2010-11-03T17:36:43 epsilon 2010-11-03T17:36:45 epsilon, I assume 2010-11-03T17:36:46 just some random value? 2010-11-03T17:36:47 epsilon 2010-11-03T17:36:51 small value i.e 2010-11-03T17:36:56 ye 2010-11-03T17:36:57 very small value 2010-11-03T17:38:55 is p your own planet or a neutral you're considering? 2010-11-03T17:39:42 *** sigh has joined #aichallenge 2010-11-03T17:39:42 *** Guest76203 has quit IRC (Read error: Connection reset by peer) 2010-11-03T17:39:54 IOW what does smallestDistance represent? 2010-11-03T17:40:15 it represents how "lonely" a planet is 2010-11-03T17:41:27 and what is the dist you are passing in to the function? 2010-11-03T17:41:28 and what is the dist parameter? 2010-11-03T17:41:33 hah same question 2010-11-03T17:42:17 distance from your planets? 2010-11-03T17:42:26 yes 2010-11-03T17:42:30 average 2010-11-03T17:44:22 interesting, I may steal it 2010-11-03T17:44:36 be my guest 2010-11-03T17:44:46 mine currently sucks 2010-11-03T17:44:59 *** rebelxt has joined #aichallenge 2010-11-03T17:46:05 hm, if I understand it right, won't this keep you from taking the last neutral planet? 2010-11-03T17:47:30 *** davidjliu has quit IRC (Ping timeout: 265 seconds) 2010-11-03T17:47:43 wait... 2010-11-03T17:47:57 I see, smallestDist always ends up 0 2010-11-03T17:48:33 because n might == p 2010-11-03T17:52:16 hm, no, if there are no more neutrals (EPS * pow(smallestDist, 2.0)) should evaluate to 461168601 2010-11-03T17:53:01 Error323, am I missing something? 2010-11-03T17:53:17 lavalamp__: no ur right, its a bug :$ 2010-11-03T17:53:26 haha ok 2010-11-03T17:53:48 *** fawek has joined #aichallenge 2010-11-03T17:53:55 lol 2010-11-03T17:54:02 how stupid 2010-11-03T17:54:08 :D 2010-11-03T17:54:29 I'm sure my bot has worse 2010-11-03T17:57:33 awesome, the one snippet of code i post has a bug :D 2010-11-03T17:57:45 -.- 2010-11-03T17:58:04 *** rebelxt has left #aichallenge 2010-11-03T17:59:04 *** Appleman1234 has quit IRC (Read error: Connection reset by peer) 2010-11-03T18:01:30 anyway 2010-11-03T18:01:45 golang lets me use unicode characters so I have "const ε = 1e-10" in a file and use ε instead of writing out epsilon :) 2010-11-03T18:01:58 given that it reduces to the first term 2010-11-03T18:02:04 cool 2010-11-03T18:02:08 yeah I still like the first term 2010-11-03T18:03:23 bhasker: why are we climbing so rapidly 2010-11-03T18:03:25 it scares me 2010-11-03T18:03:35 hmm not sure lol 2010-11-03T18:03:59 i think there are very few of the top bots running right now 2010-11-03T18:04:05 so our win rate is much higher 2010-11-03T18:04:05 yeah 2010-11-03T18:04:07 even my crappy tree bot went up by 100 ELO today on TCP 2010-11-03T18:04:25 very not cool 2010-11-03T18:04:28 and mega1 is not running now 2010-11-03T18:04:35 so the leveller is gone 2010-11-03T18:04:39 so everyone goes up a bit 2010-11-03T18:04:49 the almighty leveler xD 2010-11-03T18:04:53 kinda throwing my estimations of what changes are good out the window 2010-11-03T18:05:00 as far as i am concerned i am still losing to the same bots 2010-11-03T18:05:00 lol 2010-11-03T18:05:04 I'm testing locally, it's so noisy out there. 2010-11-03T18:05:06 so actual elo doesn't really matter 2010-11-03T18:05:28 ...that's why you should always run a control bot ;) 2010-11-03T18:05:34 mega1: could you still insert it plz? cuz this is fail ^^ 2010-11-03T18:05:49 just run your own bot 2010-11-03T18:05:51 again 2010-11-03T18:05:54 under another name 2010-11-03T18:06:01 then it will keep you from constantly climbing 2010-11-03T18:06:01 I can't, my poor laptop is stressed out already. 2010-11-03T18:06:08 ah ok 2010-11-03T18:06:17 it'd just time out 2010-11-03T18:06:31 i'll insert v 16 aswell then 2010-11-03T18:07:40 *** yasith has quit IRC (Ping timeout: 250 seconds) 2010-11-03T18:07:48 maan 2010-11-03T18:08:08 one loss in the beginning of ur submission @ official 2010-11-03T18:08:18 and you have to climb the entire freakin ladder 2010-11-03T18:08:20 >_< 2010-11-03T18:08:35 Error323, why are you doing pow(x, 2.0)? Can't you just do x*x? 2010-11-03T18:08:42 yes 2010-11-03T18:08:50 just now reading the snippet you posted earlier 2010-11-03T18:08:51 you can :P 2010-11-03T18:09:03 but it stems from tweaking 2010-11-03T18:09:03 Migi32: thats the first thing i though when i saw the code snippet ;) 2010-11-03T18:09:25 Error323: x*x ist faster,i'm quite sure? 2010-11-03T18:09:26 Error323, ah ok, then I understand :) 2010-11-03T18:09:32 *** JamesGecko_ has joined #aichallenge 2010-11-03T18:09:55 Error323, so after trying a whole bunch of different values 2.0 turned out to be the best? Interesting. 2010-11-03T18:10:03 oh 2010-11-03T18:10:12 well 2010-11-03T18:10:13 thats what you mean by tweaking 2010-11-03T18:10:17 i thought speed 2010-11-03T18:10:21 given my bot :P 2010-11-03T18:10:34 but yeah 2010-11-03T18:11:02 Error323, did you also try values like 1.7 or just integer values? 2010-11-03T18:11:32 because if you did, there may be a mathematical reason why 2.0 is the best 2010-11-03T18:12:16 i did a sort of binary search yeah 2010-11-03T18:14:51 Top 10 players: bocsimacko(4232), bix0r4ever(4151), jimrogerz(4103), dmj111(4098), george(4060), asavis(4036), Hazard(4032), GreenTea(4025), shangas(4015), felixcoto(3969) 2010-11-03T18:15:44 time for a beer now 2010-11-03T18:16:07 ever tried coding with alcohol? 2010-11-03T18:16:10 its interesting 2010-11-03T18:16:15 especially the next day 2010-11-03T18:16:23 yeah I have. 2010-11-03T18:16:42 I find this very true: http://xkcd.com/323/ 2010-11-03T18:17:03 :D yeah 2010-11-03T18:17:04 same 2010-11-03T18:17:06 there's this small peak and if you hit it just right you code like Linus Torvalds on crack :P 2010-11-03T18:17:10 its awesome 2010-11-03T18:17:12 hahaha 2010-11-03T18:17:32 or at least, you think you do 2010-11-03T18:17:42 as you said, the next day is interesting as well :P 2010-11-03T18:17:55 bhasker: >_< 2010-11-03T18:17:58 did you fix ur stuff? 2010-11-03T18:18:04 :D 2010-11-03T18:18:16 *** JamesMG has quit IRC (Quit: Leaving.) 2010-11-03T18:18:17 i hate you 2010-11-03T18:18:19 =) 2010-11-03T18:19:01 *** Appleman1234 has joined #aichallenge 2010-11-03T18:19:15 bhasker: what did you do? 2010-11-03T18:19:27 changed how i calculate available ships to be more accurate 2010-11-03T18:19:29 did you actually improve it, or are all the ratings screwed up badly 2010-11-03T18:19:37 ah ok 2010-11-03T18:19:38 so it doesn't do dumb shit anymore 2010-11-03T18:19:40 *** yasith has joined #aichallenge 2010-11-03T18:19:42 i still have one bug to fix 2010-11-03T18:19:44 which i will do later 2010-11-03T18:19:49 http://72.44.46.68/canvas?game_id=842686 2010-11-03T18:19:53 its with how i priortize planets 2010-11-03T18:20:03 stupid thing should have gone for the neutral 2010-11-03T18:20:04 and sniped you 2010-11-03T18:20:12 =] 2010-11-03T18:20:13 instead it went for your base planet and got owned 2010-11-03T18:20:19 indd 2010-11-03T18:20:37 the reason it did that is i look at enemy/my planets before i look at neutrals 2010-11-03T18:20:58 this one too: http://72.44.46.68/canvas?game_id=842690. Biggest reason why you lost it is because you took the 80 before the 30 2010-11-03T18:21:09 well it won't have launched if my shipsRequired function was accounting for neutrals 2010-11-03T18:21:13 correctly 2010-11-03T18:21:47 since it doesn't 2010-11-03T18:21:53 it calculated it had enough ships to take it 2010-11-03T18:22:01 and it launched 2010-11-03T18:22:03 :( 2010-11-03T18:22:15 i haven't figured how to account for ships in flight + neutrals 2010-11-03T18:22:18 which enemy will acquire 2010-11-03T18:22:34 i have :D 2010-11-03T18:22:40 but i have other issues:p 2010-11-03T18:22:58 like keeping attacking while you expand 2010-11-03T18:23:08 heh 2010-11-03T18:23:09 ps 2010-11-03T18:23:16 doh wrong window 2010-11-03T18:23:19 lol 2010-11-03T18:23:27 *** lavalamp__ has quit IRC (Quit: Page closed) 2010-11-03T18:24:53 *** bhasker_ has joined #aichallenge 2010-11-03T18:27:45 whois davina 2010-11-03T18:28:17 *** bhasker__ has joined #aichallenge 2010-11-03T18:28:34 *** bhasker has quit IRC (Ping timeout: 276 seconds) 2010-11-03T18:30:42 *** bhasker_ has quit IRC (Ping timeout: 240 seconds) 2010-11-03T18:32:21 *** nann has joined #aichallenge 2010-11-03T18:36:00 *** Apophis has quit IRC (Ping timeout: 264 seconds) 2010-11-03T18:43:43 *** Bobng has quit IRC (Ping timeout: 255 seconds) 2010-11-03T18:48:35 bhasker__: maan we are really dancing around eachother 2010-11-03T18:52:19 *** plediii has joined #aichallenge 2010-11-03T19:11:03 *** RainCT has quit IRC (Remote host closed the connection) 2010-11-03T19:14:53 Top 10 players: bocsimacko(4213), bix0r4ever(4137), jimrogerz(4135), dmj111(4087), george(4039), asavis(4022), GreenTea(4020), Hazard(4010), shangas(4007), Raschi(3965) 2010-11-03T19:15:36 *** cfaftw has joined #aichallenge 2010-11-03T19:27:16 *** AJC has left #aichallenge 2010-11-03T19:28:13 *** fawek has quit IRC (Ping timeout: 276 seconds) 2010-11-03T19:33:04 *** Cold-Phoenix has quit IRC () 2010-11-03T19:37:59 *** narnach has quit IRC (Quit: Leaving.) 2010-11-03T19:38:13 *** airhorns has joined #aichallenge 2010-11-03T19:38:26 *** bhasker has joined #aichallenge 2010-11-03T19:40:29 *** bhasker__ has quit IRC (Ping timeout: 260 seconds) 2010-11-03T19:40:43 *** mceier has quit IRC (Quit: leaving) 2010-11-03T19:40:43 *** Cold-Phoenix has joined #aichallenge 2010-11-03T19:42:06 *** superflit has quit IRC (Quit: superflit) 2010-11-03T19:42:12 *** hornairs has quit IRC (Ping timeout: 245 seconds) 2010-11-03T19:56:40 #22!! 2010-11-03T19:57:25 ok... I have tea, new ideas, and an awful lot of bugs, let's see what happens... 2010-11-03T19:57:42 *** hornairs has joined #aichallenge 2010-11-03T19:58:42 nice 2010-11-03T19:59:42 *** airhorns has quit IRC (Ping timeout: 245 seconds) 2010-11-03T20:01:24 i have no idea why its climbing though 2010-11-03T20:01:32 i never expected it to go above 40ish 2010-11-03T20:01:48 i think a lot of people are resubmitting newer versions and moving down the ladder 2010-11-03T20:01:54 i don't think my bot is that good 2010-11-03T20:02:07 meanwhile my submission bot has slipped down to 50-ish... and rightfully so 2010-11-03T20:04:04 *** flag has joined #aichallenge 2010-11-03T20:04:51 *** demerzel__ has joined #aichallenge 2010-11-03T20:06:14 *** Appleman1234 has quit IRC (Read error: Connection reset by peer) 2010-11-03T20:14:53 Top 10 players: bocsimacko(4220), bix0r4ever(4135), jimrogerz(4135), dmj111(4063), george(4036), GreenTea(4034), shangas(3990), Hazard(3989), Raschi(3967), felixcoto(3954) 2010-11-03T20:15:08 *** Migi32 has quit IRC (Quit: http://boinc.berkeley.edu) 2010-11-03T20:17:35 hmm this new version seems a little better 2010-11-03T20:17:40 i wonder if i should upload a new version 2010-11-03T20:18:11 bocsimacko lost a game! 2010-11-03T20:18:15 Oh, it was a timeout 2010-11-03T20:18:20 #21 now 2010-11-03T20:18:36 on tcp or main? 2010-11-03T20:18:38 main 2010-11-03T20:18:41 http://ai-contest.com/rankings.php 2010-11-03T20:18:56 #16 on tcp 2010-11-03T20:19:03 nice 2010-11-03T20:19:06 technically that should be 5-6 2010-11-03T20:19:11 malazan? 2010-11-03T20:19:13 yes 2010-11-03T20:19:23 ah, nice bot 2010-11-03T20:19:27 woop my bot is fully ANSI-C lol 2010-11-03T20:19:28 still does stupid things:( 2010-11-03T20:19:38 i need to fix my priority queue and start 2010-11-03T20:19:45 Not as stupid as mine :) 2010-11-03T20:19:49 which one is yours 2010-11-03T20:19:56 ObligatoryShampoo 2010-11-03T20:20:02 oooh 2010-11-03T20:20:04 annoying bot 2010-11-03T20:20:06 lol 2010-11-03T20:20:09 i always lose to your bot 2010-11-03T20:20:12 well not always 2010-11-03T20:20:14 but a lot 2010-11-03T20:20:16 annoying 2010-11-03T20:20:25 And by always you mean about 1/4th of the time :-O 2010-11-03T20:20:26 and its almost a 100 elo points below mine 2010-11-03T20:20:41 cosmickangaroo i guess is yours too 2010-11-03T20:20:44 that one isn't as good 2010-11-03T20:20:50 *** boiserm has joined #aichallenge 2010-11-03T20:20:52 this shampoo one is annoying 2010-11-03T20:21:00 Well it's just me trying different things 2010-11-03T20:21:03 how many versions do you run 2010-11-03T20:21:10 since i seem to run up against yours a lot 2010-11-03T20:21:13 ObligatoryShampoo is a new bot everyday 2010-11-03T20:21:15 4 at once 2010-11-03T20:21:15 anyone around who could tell me what went wrong here? http://ai-contest.com/visualizer.php?game_id=6555583 2010-11-03T20:21:19 ah no wonder 2010-11-03T20:21:25 i get like 4-5 games in a row with your bot 2010-11-03T20:21:42 Yeah 2010-11-03T20:21:43 looks like you timed out? 2010-11-03T20:21:50 or sent an invalid command? 2010-11-03T20:21:57 I timeout occasionally 2010-11-03T20:22:00 my record is full of winning games though 2010-11-03T20:22:03 But I'm pretty sure it's just my internet 2010-11-03T20:22:04 (and losing ones) 2010-11-03T20:22:08 but full games regardless 2010-11-03T20:22:16 *** Appleman1234 has joined #aichallenge 2010-11-03T20:22:19 run the game locally ? 2010-11-03T20:22:24 using that python script 2010-11-03T20:22:27 and see 2010-11-03T20:22:30 what your bot does 2010-11-03T20:22:33 "that python script" == ? 2010-11-03T20:22:42 let me paste it for you hold on 2010-11-03T20:22:45 thanks 2010-11-03T20:23:07 http://pastebin.com/1p3JRhvW 2010-11-03T20:23:13 use that script as follows 2010-11-03T20:23:14 thanks, I'll try it 2010-11-03T20:23:45 ./parse_game_state_url.py 0/1 | 2010-11-03T20:23:55 the 0/1 is to swap player id 2010-11-03T20:24:00 interesting 2010-11-03T20:24:01 if you are playing second on the visualizer page 2010-11-03T20:24:03 thanks 2010-11-03T20:24:04 not my script 2010-11-03T20:24:08 i took it from the forums 2010-11-03T20:24:09 :) 2010-11-03T20:24:29 eh, either way :) 2010-11-03T20:24:35 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-11-03T20:25:07 *** jaspervdj has quit IRC (Quit: NEVER GONNA GIVE YOU UP NEVER GONNA LET YOU DOWN) 2010-11-03T20:25:08 it will output game state to stdout 2010-11-03T20:25:12 and your logs if any to stderr 2010-11-03T20:25:19 you may want to redirect them to a file to analyze 2010-11-03T20:26:17 :-/ not particularly informative in this case... must have been a random timeout 2010-11-03T20:26:19 *** superflit has joined #aichallenge 2010-11-03T20:26:20 no errors. 2010-11-03T20:28:39 possible after all the servers are amazon instances running many games 2010-11-03T20:29:21 *** sigh has quit IRC (Read error: No route to host) 2010-11-03T20:29:48 yeah. :-/ 2010-11-03T20:29:51 ...and it's python. 2010-11-03T20:30:18 inefficiently written python at that, I'm sure 2010-11-03T20:33:37 the replay strings are super easy to parse 2010-11-03T20:33:58 going the other way you just list owner,ships: then fleet infos in sequential order 2010-11-03T20:34:29 *** sigh has joined #aichallenge 2010-11-03T20:34:37 not as easy since you ahve to set up the initial state but it's not too hard 2010-11-03T20:34:46 *** yasith has quit IRC (Ping timeout: 265 seconds) 2010-11-03T20:34:53 i have some C# code around to go from replay string -> values if you want 2010-11-03T20:37:51 I'm too tired to mess with it anymore tonight, but thanks for the offer 2010-11-03T20:38:13 48 games, 36 wins, 12 losses, 4 mysterious crashes/timeouts 2010-11-03T20:38:27 so 1/3 of my losses are a mystery :( 2010-11-03T20:39:21 *** tapwater has joined #aichallenge 2010-11-03T20:39:35 oh well, will check it out tomorrow. Thanks for the suggestions 2010-11-03T20:40:27 doh. Make that 37 losses -_- 2010-11-03T20:40:36 wait... err... 13 losses 2010-11-03T20:40:38 wow, bedtime. 2010-11-03T20:41:00 *** boiserm has quit IRC (Quit: Page closed) 2010-11-03T20:47:27 *** yasith has joined #aichallenge 2010-11-03T20:49:18 *** Cold-Phoenix has quit IRC (Read error: Connection reset by peer) 2010-11-03T20:50:19 *** Cold-Phoenix has joined #aichallenge 2010-11-03T20:57:50 *** amstan has quit IRC (Remote host closed the connection) 2010-11-03T20:58:00 *** amstan has joined #aichallenge 2010-11-03T20:58:00 *** ChanServ sets mode: +o amstan 2010-11-03T21:01:05 *** tapwater_ has joined #aichallenge 2010-11-03T21:02:46 *** bhasker has quit IRC (Ping timeout: 255 seconds) 2010-11-03T21:03:10 *** tapwater has quit IRC (Ping timeout: 250 seconds) 2010-11-03T21:03:10 *** tapwater_ is now known as tapwater 2010-11-03T21:10:24 *** Cold-Phoenix has quit IRC () 2010-11-03T21:10:35 *** cfaftw has quit IRC (Quit: Leaving) 2010-11-03T21:11:27 *** adu has joined #aichallenge 2010-11-03T21:14:52 Top 10 players: bocsimacko(4227), bix0r4ever(4138), jimrogerz(4061), dmj111(4061), george(4055), GreenTea(4037), shangas(4010), Hazard(3989), felixcoto(3956), Accoun(3951) 2010-11-03T21:20:59 gdmt i broke it 2010-11-03T21:24:26 *** davidjliu has joined #aichallenge 2010-11-03T21:27:55 ahahah holy shit i'm a moron 2010-11-03T21:32:05 *** adu has left #aichallenge 2010-11-03T21:35:54 *** Aaaarg has quit IRC (Ping timeout: 265 seconds) 2010-11-03T21:50:54 *** hornairs has quit IRC (Remote host closed the connection) 2010-11-03T21:53:48 one thing that constantly amazes me is how many bugs you can introduce through very simple miscalcs 2010-11-03T22:02:23 demerzel__: not as amazing as when fixing those bugs makes the bot worse 2010-11-03T22:05:21 sigh: my bot can't get much worse atm lol 2010-11-03T22:05:27 it's tracking them down that's my damn problem atm 2010-11-03T22:06:18 if it is reproducible then it is easy... otherwise... good luck 2010-11-03T22:14:53 Top 10 players: bocsimacko(4217), bix0r4ever(4121), dmj111(4052), george(4051), GreenTea(4035), jimrogerz(4010), shangas(4008), Hazard(3977), Raschi(3970), jitanus(3943) 2010-11-03T22:18:13 I used valgrind recently and it tought me a lot about how c++ really works ^^. Still some messages were very obscure. 2010-11-03T22:20:17 I have experimented with an academic problem. Assume all planets sit in one spot and there is no enemy. What is the fastest way to capture them all? 2010-11-03T22:20:59 i think i may have actually found it lol 2010-11-03T22:21:06 i can't use valgrind no windows haha 2010-11-03T22:22:10 The bad thing is that itermediate states have multiple qualities. Available own ships and the turn # for example. If there was only one quality I could solve the problem now with ~4.000.000 attack calculations. 2010-11-03T22:23:11 And still that would be two steps away from the game reality: 1. ignoring travel time is bad, 2. no enemy isn't quite representative 2010-11-03T22:24:23 my bot was about to actually win and crapped out at like turn 120 somehow changing from successfully streaming from 5 planets to trying to launch from 2 haha http://72.44.46.68/canvas?game_id=845431 2010-11-03T22:24:35 why bot why 2010-11-03T22:36:41 any lisp users around? 2010-11-03T22:39:18 janzert: kick demerzel_ i have no idea why it's still up i close it like 5 hours ago 2010-11-03T22:39:33 oh you don't have ops weird haha 2010-11-03T22:40:00 demerzel__: you sure? 2010-11-03T22:40:12 demerzel__: kicking won't mean he'll disconnect 2010-11-03T22:40:17 just means he'll leave this channel 2010-11-03T22:40:22 oh well nm 2010-11-03T22:40:28 maybe it's just sitting at my workstation haha 2010-11-03T22:40:40 ssh in and kill the irc program 2010-11-03T22:41:38 hahaha yea my work uses linux 2010-11-03T22:41:43 they're scared of java 2010-11-03T22:53:35 Frontier: no enemy can be a good enough approximation at the start of the game, as long as the planets are not too close 2010-11-03T22:54:23 *as long as starting planets are not too close 2010-11-03T23:04:43 *** sigh_ has joined #aichallenge 2010-11-03T23:04:43 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-03T23:05:16 *** sigh_ is now known as sigh 2010-11-03T23:07:26 *** bhasker has joined #aichallenge 2010-11-03T23:07:42 sigh: that's right. Maybe I should cut the calculations off after 30 turns into the future or so. 2010-11-03T23:08:47 Frontier: it's worth trying... although I'm skeptical that ignoring travel time will give you any meaningful result 2010-11-03T23:10:42 sigh: ... yeah its not worth trying. + the importance of planets that are the same distance from both players cant be ignored. maybe i find another use for that algorithm somewhere else 2010-11-03T23:11:05 *** Azrathud has joined #aichallenge 2010-11-03T23:11:54 * Frontier steals bocsimacko's source code 2010-11-03T23:13:17 ooo... gimme 2010-11-03T23:13:20 ... please? 2010-11-03T23:14:27 Don't be silly ^^. 2010-11-03T23:14:54 Top 10 players: bocsimacko(4200), bix0r4ever(4103), dmj111(4035), george(4035), jimrogerz(4033), GreenTea(4019), shangas(3997), Raschi(3952), Hazard(3949), jitanus(3940) 2010-11-03T23:15:34 Frontier: btw, what is your name in the name in the main contest? 2010-11-03T23:16:17 hmmm... it seems I've asked you this before 2010-11-03T23:16:28 ok, your name is Frontier in the main contest 2010-11-03T23:17:06 I wish we could search users by username :( 2010-11-03T23:22:50 That would be a long missing feature. I haven't uploaded anything in ages. Im on rank 241 now. 2010-11-03T23:23:37 i'm down to like 3-400 hehe 2010-11-03T23:24:07 when i put it up it was 2 weeks old even cuz of that c# compiling problem 2010-11-03T23:25:04 * janzert idly wonders about adding a simple test for bots to past before they are entered in the contest, basically just that they reply with a valid order after being given an initial map 2010-11-03T23:25:35 mostly just to help catch all the fatal errors in non-compiled languages 2010-11-03T23:25:38 i would just say invalid orders disqualify you 2010-11-03T23:25:50 timing out is an issue that some people can't avoid 2010-11-03T23:26:22 if someone issues an invalid order once in a great while it's not really that big a deal either 2010-11-03T23:28:42 ideally the starter packs would have included code to check for invalid moves 2010-11-03T23:28:57 they don't? 2010-11-03T23:29:05 damn i thought they did 2010-11-03T23:29:10 I mean the bots, not the engine 2010-11-03T23:29:42 *** plediii has quit IRC (Ping timeout: 245 seconds) 2010-11-03T23:30:29 I have code to crash instead of issuing an invalid order... :/ 2010-11-03T23:31:11 as long as it is visible... I just log to an error file locally, ignore silently on the main server 2010-11-03T23:34:45 i have no error handling because i'm in pure C atm haha 2010-11-03T23:34:55 i did just have a try/catch/finally that sent go in C# 2010-11-03T23:44:30 @predict 2010-11-03T23:44:30 I predict that Azrathud will win! 2010-11-03T23:45:09 impressive it managed to spit out 20350 orders in under a second 2010-11-03T23:45:13 all the same unfortunately 2010-11-03T23:56:46 sigh: are you still using checks for invalid moves? 2010-11-03T23:56:56 i've taken them out for the moment, otherwise you don't always see bugs 2010-11-03T23:57:27 antimatroid: I have the checks, but I also log an error 2010-11-03T23:57:40 the only time it is silent is on the main server 2010-11-03T23:57:59 yeah okay, I just let it go without checks atm 2010-11-03T23:58:06 i think i have it right so that it doesn't crash 2010-11-03T23:58:16 anyway, i'm off to do something about lunch