2010-09-13T00:01:21 lol 2010-09-13T00:01:28 PlanetWarrior: in answer to your original question, the answer is IIRC yes 2010-09-13T00:01:47 well, "another quick question" 2010-09-13T00:01:56 git status 2010-09-13T00:01:57 err 2010-09-13T00:02:03 * RobotCaleb backs out slowly 2010-09-13T00:02:06 Thanks for that Mathnerd314. Am I alone in thinking that this is rather unsatisfactory? 2010-09-13T00:02:29 PlanetWarrior: I think it's just fine :-) 2010-09-13T00:05:24 I hope there's something which forbids planets occurring so close together. If you have a bot that has an internal model of the game state which it updates from turn to turn, it should be able to ignore all the planet stuff and the fleets it's already seen and just look at the fleets just launched which to my mind should be equivalent to the list of orders. 2010-09-13T00:06:34 *** preetum has joined #aichallenge 2010-09-13T00:07:13 nope, nothing forbidding it 2010-09-13T00:07:34 I like planets with a distance of 1, sounds fun :) 2010-09-13T00:07:35 as long as two planets have different x,y coords they will be >= 1 turn apart 2010-09-13T00:08:00 true 2010-09-13T00:08:01 we need a close packed map and then develop stealth strategies :D 2010-09-13T00:08:18 ah, ok 2010-09-13T00:08:23 the distance is ceil so they should always show up don't they? 2010-09-13T00:09:27 if they're <= 1 unit distant, you would have: issue order => execute turn => fleet moves 1 unit => caps planet, yes? 2010-09-13T00:10:10 hmm. i think the fleet would exist in flight for one turn 2010-09-13T00:10:26 but maybe not 2010-09-13T00:10:35 nope, if you send a fleet to distance of 1, it never enters Fleet list 2010-09-13T00:11:00 PlanetWarrior: the current map generator seems to avoid distance <=1 planets though 2010-09-13T00:11:05 The fleet would exist in a 'virtual state' while the waveform colapses 2010-09-13T00:11:22 lol 2010-09-13T00:11:40 yes, there is a superposition of fleets if you send multiple ones 2010-09-13T00:11:56 but there aren't any interferance effects 2010-09-13T00:12:00 thankfully... 2010-09-13T00:12:03 @later tell j3camero can we assume that planets are always >1 turn apart? 2010-09-13T00:12:03 Mathnerd314: The operation succeeded. 2010-09-13T00:12:12 Yes, otherwise our algos would have a much harder time 2010-09-13T00:12:34 mmh 2010-09-13T00:12:45 sounds fun to me 2010-09-13T00:12:45 what is contestbot? 2010-09-13T00:12:48 I think 2010-09-13T00:12:49 You may need to account for the fleet diffraction pattern when you send them between the two closely placed planets 2010-09-13T00:12:52 Mathnerd314: stop bothering j3camero directly, there's other people that can answer your questions 2010-09-13T00:13:00 the next AI challenge should be with invisible enemy fleets :D 2010-09-13T00:13:05 preetum: there's a link in the topic about it, it's a bot, that helps people 2010-09-13T00:13:13 amstan: but he's never answered it directly 2010-09-13T00:13:14 preetum: it can also do this: 2010-09-13T00:13:17 contestbot: rankings 2010-09-13T00:13:19 amstan: Top 10 players: dmj111(2835), braddunbar(2815), albert(2792), seigen(2732), Kavish(2707), drewrobb(2695), Obeleh(2679), wtfbbq(2679), sighbot(2676), tinasky(2661) 2010-09-13T00:13:36 Mathnerd314: even i can't get straight answers from him 2010-09-13T00:13:51 Mathnerd314: tbh, i don't think he knows for sure himself 2010-09-13T00:13:53 nice feature 2010-09-13T00:14:07 amstan: surely he'll answer with a straight yes/no 2010-09-13T00:14:21 Mathnerd314: good luck on that, lol 2010-09-13T00:14:42 is the engine down? my bot hasn't played any games in 24 hours. 2010-09-13T00:15:21 *** Theobon has quit IRC (Ping timeout: 252 seconds) 2010-09-13T00:17:16 *** AlliedEnvy has joined #aichallenge 2010-09-13T00:18:16 preetum: yes, looks like it, we're still having performance problems with the server 2010-09-13T00:18:44 ok guys, my newest iteration is doing some runs on the tcp server, come beat him up I need input :) 2010-09-13T00:18:50 alright, no problem. just wanted to make sure I wasn't doing something wrong. 2010-09-13T00:18:58 *** fx4m has joined #aichallenge 2010-09-13T00:20:40 it would be cool if someone could build a web interface for the tcp server. human vs. bot, I mean. 2010-09-13T00:20:46 deepblue: wow, looks awesome :p 2010-09-13T00:20:48 And while youre at it, check out the Qt client - http://github.com/greghaynes/QTcpPlanetWars 2010-09-13T00:20:50 the galcon game higher levels don't show you how many enemies 2010-09-13T00:21:03 thanks :) 2010-09-13T00:23:44 *** montana1 has joined #aichallenge 2010-09-13T00:23:47 http://www.benzedrine.cx/planetwars/canvas?game_id=1284350952|wtfbbq|Cyndre that was a good battle 2010-09-13T00:24:35 do you have a windows version greghaynes? 2010-09-13T00:25:02 *** darkf has joined #aichallenge 2010-09-13T00:25:13 deepblue: is it playing optimal strategy yet? 2010-09-13T00:25:26 i don't think so at all 2010-09-13T00:25:37 ok, have to finish mine then :p 2010-09-13T00:25:48 I mean, I still loose 2010-09-13T00:25:53 soo not optimal :) 2010-09-13T00:25:56 It always is right up till you lose. :P 2010-09-13T00:26:05 lol 2010-09-13T00:27:50 *** preetum has quit IRC (Quit: Page closed) 2010-09-13T00:27:56 deepblue, you reading parameter weights in from a text file? I'm adding that now so I can use some genetic iterations to teak the weights. 2010-09-13T00:28:07 *** DirtyKeyboard has joined #aichallenge 2010-09-13T00:28:31 I don't do things like that :) 2010-09-13T00:28:37 grr beaten by a camel 2010-09-13T00:29:05 shawnshekari: those things can spit man, fierce beasts. 2010-09-13T00:29:43 shawnshekari: your bot just slept about 150 turns against my non-offensive because it had +1 growth.. might wanna look into that 2010-09-13T00:30:25 shawnshekari: nvm, I think it was your old bot 2010-09-13T00:30:26 roger, looking at the log now 2010-09-13T00:31:22 my bot acted like a dick one game. It left the opponent with one node and never attacked again till the turn limit happened 2010-09-13T00:31:41 Mathnerd314: hey, sorry, i never got a chance to extract that code. it might not be until tomorrow evening. 2010-09-13T00:31:54 *** darkf has left #aichallenge ("Leaving") 2010-09-13T00:31:58 Mathnerd314: really, it's not that much code anyway. i wouldn't wait for it unless you just really really want to 2010-09-13T00:32:05 *** Queue29 has quit IRC (Quit: Leaving) 2010-09-13T00:32:07 jmcarthur: ok, no problem. I'm rewriting it all anyway :p 2010-09-13T00:32:28 jmcarthur: Had time to submit your bot, yet? 2010-09-13T00:32:40 bss03: i haven't even had time to write much of it yet 2010-09-13T00:32:51 jmcarthur: So far __init__/grep0 is the best Hakeller on the official server. 2010-09-13T00:33:02 bss03: i have been working too much 2010-09-13T00:33:08 and contributing 2010-09-13T00:33:10 jmcarthur: trying to get a pipeline of functions IO -> [String] -> [Order] -> GameState 2010-09-13T00:33:12 instead of coding my bot 2010-09-13T00:33:16 bss03: is submitting working now? hasn't been working all day for me 2010-09-13T00:33:47 DirtyKeyboard: I haven't tried all day. 2010-09-13T00:34:01 bss03: i fully intend to have the best haskell bot ;) 2010-09-13T00:34:04 DirtyKeyboard: I've been making incremental improvements to my bot. 2010-09-13T00:34:06 DirtyKeyboard: hasn't been any games for hours.. doubt that submitting will be of any help 2010-09-13T00:34:09 i mean, i might not succeed, but i'll try 2010-09-13T00:34:30 jmcarthur: that's impossible, because I'm writing my bot in Haskell :p 2010-09-13T00:34:37 jmcarthur: I'm second at #66. However, I'm pretty sure I can beat __init__ with a few more tweaks. 2010-09-13T00:34:41 I see what's wrong ratatata, somehow a weight got adjusted down to zero... lol 2010-09-13T00:35:02 I submit my bot and don't get an email about failing to compile or succeeding.. and yeah, it just stinks not having ANY version of my bot there 2010-09-13T00:35:22 DirtyKeyboard: I probably won't re-sumbit until I get a much better win/loss ratio out of RageBot. :( 2010-09-13T00:36:06 DirtyKeyboard: I my improvement today moved me from 47/100 -> 57/100 but my next improvement dropped me to 54/100. 2010-09-13T00:36:25 The email has been broken since day one. 2010-09-13T00:36:47 But, your last submission time and whether the cubmission was accepted shows up on your profile page. 2010-09-13T00:36:56 bss03: yeah, nothing there. :| 2010-09-13T00:37:49 DirtyKeyboard: Sometime during day 1, the host got flagged by Google (HA!) has sending too much to be normal, so most likely SPAM. 2010-09-13T00:38:01 bss03: and yeah, ragebot is an annoying thing to beat... I might have to program specifically to counter that strategy 2010-09-13T00:38:48 bss03: I struggle with RageBot too 2010-09-13T00:39:00 especially map 30 2010-09-13T00:39:07 DirtyKeyboard: So far, I dn't have the am-i-winning-or-losing logic that DualBot uses and I don't maintain any state between turns. 2010-09-13T00:39:29 So, there's plenty of room for improvement. >:) 2010-09-13T00:39:42 map26 and map30 are close starters, if you attack > cant gain enough and loose main planet -> you loose .. I just ignore those cases for now :) 2010-09-13T00:39:55 *against RageBot that is 2010-09-13T00:40:24 I use map 1, 20 and 30 alot for testing 2010-09-13T00:40:36 ratatata: I think if my bot was just a little less aggressive with how many ship it is willing to send out the first turn I might have a chance against RB as is. 2010-09-13T00:41:13 I haven't quite figured out map 30 yet - it's easy to hold to a stalemate, but hard to fan out and build growth without sacrificing too much and falling behind. 2010-09-13T00:41:38 As the code stands, there's no limit other than the number of ships on planet for the fleets I send out in a turn, which does really hurt against RB. 2010-09-13T00:42:02 bss03: you leave yourself open for big, quick attacks, right? 2010-09-13T00:42:02 *** joebob has joined #aichallenge 2010-09-13T00:42:26 I hope they use map30 in the final tournament, it's a great way to see good bots 2010-09-13T00:42:40 with map30 you have a little bit of leaway in the number of units that the planet can generate while the attack is in flight 2010-09-13T00:42:47 rwa: Yeah, if we start close, or your bot can simply count how much my home planet is gonna gain by the time you get there, taking my home planet first thing usually kills me. 2010-09-13T00:42:51 you can use this leaway to aquire a neutral 2010-09-13T00:43:01 anyone know if the upload is broken? 2010-09-13T00:43:10 joebob: it's broken 2010-09-13T00:43:13 joebob: Yes, it is. 2010-09-13T00:43:32 Joebob: I think they are working on it. I know I left a message for j3camero. 2010-09-13T00:43:46 site works great, now that upload, games, and leaderboard are turned off. ;) 2010-09-13T00:43:56 joebob: I'm not sure really what the best way to contact the admins about technical issues is. 2010-09-13T00:43:59 rwa: lol 2010-09-13T00:44:06 rwa: Too true. 2010-09-13T00:44:21 no worries.. I need to get to bed.. anyways.. good excuse to get some sleep 2010-09-13T00:44:46 the best way is to file a bug report (after checking to make sure it's not already there, of course) 2010-09-13T00:44:50 I still don't have code to reinforce planets, so if your bot is smart enough to correctly gage the number of ship required to take it over, you'll get my planets. 2010-09-13T00:44:51 @where issues 2010-09-13T00:44:51 jmcarthur: issues = http://code.google.com/p/ai-contest/issues/list 2010-09-13T00:45:19 That's my next improvement; including some code for reinforcing planets. 2010-09-13T00:45:23 actually, that's second best 2010-09-13T00:45:28 the best is to write a patch :D 2010-09-13T00:45:39 bss03: defence is pretty much the key to beating ragebot 2010-09-13T00:46:09 sigh: I was told that last night, but the two improvement I implemented today wouldn't let me think about it until I had them in the code. :( 2010-09-13T00:46:43 yeah, fair enough 2010-09-13T00:47:05 I also have a lot of stuff on my todo list and no time to implement them 2010-09-13T00:47:55 I'm hoping that jaspervdj wil submit the Haskell package and it can become official. 2010-09-13T00:48:00 Java is kind of a bear for this.. I really could see how a more functional language would make things easier.. 2010-09-13T00:48:07 *** craigferguson has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.9/20100825164301]) 2010-09-13T00:48:21 It should be a descent intro to Haskell for those that want to study it. I implemented all the example bots in Haskell for it. 2010-09-13T00:49:32 joebob: There's some O(n^3) stuff I'm "doing" each turn, but since Haskell is lazy only part of that work actually gets done. >:) 2010-09-13T00:50:12 Still, n = number of planets, not n = number of fleets so it would be fast enough to run all the time even in Java/C++ 2010-09-13T00:50:34 O(n^3) for fleets would kill you 2010-09-13T00:51:08 especially against the top bots which set up huge supply chains :P 2010-09-13T00:51:20 sigh: Yeah, I'm betting so. 2010-09-13T00:51:28 sigh: I tend out tons of fleets too. 2010-09-13T00:51:41 planets^3 isn't so bad 2010-09-13T00:51:52 that's effectively constant 2010-09-13T00:51:56 sigh: Plus, if someone *knew* you were doing something like that, they could just send out lots of size=0 fleets. 2010-09-13T00:51:57 yeah, not many planets 2010-09-13T00:52:05 that's not allowed 2010-09-13T00:52:18 (i think) 2010-09-13T00:52:25 sigh: I saw it happen on day one. 2010-09-13T00:52:32 sigh: It might be dis-allowed now. 2010-09-13T00:52:32 ah, ok... my mistake 2010-09-13T00:52:35 i think they are *supposed* to be silently absorbed by the engine 2010-09-13T00:52:38 but are not yet 2010-09-13T00:52:48 i see 2010-09-13T00:52:54 src==dst is illegal 2010-09-13T00:53:00 yup 2010-09-13T00:53:03 and ships>src is illegal of course 2010-09-13T00:53:04 sigh: I explicitly check my generated orders and not emit them if they have size = 0. 2010-09-13T00:53:32 *** WellMux has joined #aichallenge 2010-09-13T00:53:38 yeah, me too... I put a wrapper around issue order to prevent illegal moves and log an error 2010-09-13T00:54:26 I was thinking about treating it like a chess game and scoring the positions on the map and then iterate through a bunch of scenarios and pick the highest scoring. 2010-09-13T00:54:30 *** WellMux has left #aichallenge ("Lähdössä") 2010-09-13T00:54:58 the problem is choosing which scenarios to explore 2010-09-13T00:55:10 *** jmreardon has quit IRC (Quit: jmreardon) 2010-09-13T00:55:10 that's basically game tree search 2010-09-13T00:55:11 you can't just brute force all moves like in chess 2010-09-13T00:55:21 well, you can, but you'll time out :P 2010-09-13T00:55:22 I should pad my orders with [Order s d 0 | s <- IM.keys myPlanets, d <- IM.keys planets]. >:) 2010-09-13T00:55:23 :) 2010-09-13T00:55:41 * jmcarthur scolds bss03 2010-09-13T00:56:00 Just to confuse the guys that are doing fleet calculations, and have the visualizer completely unreadble. 2010-09-13T00:56:23 while you're at it, just do that several times 2010-09-13T00:56:33 the state space is like (25*25)^(highest number of ships on a planet) 2010-09-13T00:56:41 because unless i'm mistaken fleets still aren't merged 2010-09-13T00:56:41 assumed 25 is the max amount of planets 2010-09-13T00:57:02 and that is only for what you can do 2010-09-13T00:57:12 and you have to do that for the enemy too for every round 2010-09-13T00:57:17 one ply! 2010-09-13T00:57:22 a little much for brute force :) 2010-09-13T00:57:53 might be fun to try the game played like speed chess 2010-09-13T00:57:55 yeah, I'm sure I could do a smaller sampling.. 2010-09-13T00:57:56 too late now though 2010-09-13T00:58:02 which is good, this way we won't have a perfect playing bot running around 2010-09-13T00:58:05 but limit based on total time instead of per turn 2010-09-13T00:58:20 i don't think my computer can even calculate that number naively 2010-09-13T00:58:26 even with bignum 2010-09-13T00:58:34 *** mega1 has joined #aichallenge 2010-09-13T00:58:40 25^2^50^2^200 2010-09-13T00:58:41 ^^ 2010-09-13T00:59:02 (assuming 50 max ships on a planet each turn) 2010-09-13T00:59:15 and 200 rounds 2010-09-13T00:59:18 right 2010-09-13T00:59:26 i guess you would only need to go like a few rounds in the future 2010-09-13T00:59:28 but still 2010-09-13T00:59:30 way too much 2010-09-13T00:59:48 even if you reduce that and simplify the ships 2010-09-13T00:59:54 how long does it take for an inactive entry disappear from benzedrine.cx? 2010-09-13T01:00:23 even if you did manage to go a few rounds in, you need a good evaluation function 2010-09-13T01:00:33 it's not immediatly obvious that just plain #ships will cut it 2010-09-13T01:00:45 ^^ 2010-09-13T01:01:13 yeah, the evaluation function is the secret sauce.. 2010-09-13T01:02:22 You might be able to do some rough searching assuming that a player "optimally" attacks a single planet per turn. 2010-09-13T01:03:17 You branching factor is still 625, butwhat searchable. 2010-09-13T01:03:21 dang, (^) is right associative ^_^ 2010-09-13T01:03:36 that's why my computer can't do it 2010-09-13T01:03:43 i did it wrong 2010-09-13T01:03:48 25 ^ 2 ^ 50 un ghci is currently killing my desktop. :( 2010-09-13T01:03:57 bss03: yeah, you need parens ^_^ 2010-09-13T01:04:02 yeah, I tried in python and had to kill it 2010-09-13T01:04:17 mine produces a 55918 digit number 2010-09-13T01:04:31 for this: length . show $ (((25^2)^50)^2)^200 2010-09-13T01:05:38 log(25,10)*(2*50*2*200) = 55917.6 2010-09-13T01:05:43 easier :) 2010-09-13T01:05:45 *** mega1 has quit IRC (Ping timeout: 276 seconds) 2010-09-13T01:05:59 funny it's easier to read like this: foldl1 (^) [25, 2, 50, 2, 200] 2010-09-13T01:06:02 == very big number 2010-09-13T01:06:35 sigh: you cheated. the point was to brute force :P 2010-09-13T01:06:42 heh 2010-09-13T01:06:48 jmcarthur: does cabal work on the server? 2010-09-13T01:06:54 are you guys maintaining a lot of state between moves? 2010-09-13T01:06:59 Mathnerd314: what would you use cabal for? 2010-09-13T01:07:05 joebob: oh yes 2010-09-13T01:07:07 jmcarthur: building 2010-09-13T01:07:18 Mathnerd314: it just uses ghc -O2 --make MyBot.hs -v0 2010-09-13T01:08:00 haha, converted it to Double and it just says Infinity 2010-09-13T01:08:08 duh 2010-09-13T01:08:27 no way it would fit. don't know what i was thinking at all 2010-09-13T01:09:10 jmcarthur: if C++ can use Makefiles, then Haskell should be able to use Cabal files 2010-09-13T01:09:11 Try using Data.Number.CReal.CReal. 2010-09-13T01:09:20 Mathnerd314: C++ is using makefiles? 2010-09-13T01:09:22 That should hold floating point number that are that large. 2010-09-13T01:09:26 Mathnerd314: not to my recollection... 2010-09-13T01:09:32 last time I submitted it ignored my makefile 2010-09-13T01:09:37 *** joebob has quit IRC (Quit: Page closed) 2010-09-13T01:09:39 sigh: oh 2010-09-13T01:09:47 but that was a few days ago 2010-09-13T01:09:50 yeah, it uses its own internal makefile and ignores yours 2010-09-13T01:10:02 Mathnerd314: compilation is not sandboxed. running a makefile would be a major security problem 2010-09-13T01:10:11 Mathnerd314: for that same reason, template haskell is disabled 2010-09-13T01:10:48 would running a Cabalfile be a security problem? 2010-09-13T01:10:52 Woohoo, reminds me on an employer I was at that was very picky about who they gave root. 2010-09-13T01:11:07 rwa: no makefile is used. it's just a python script 2010-09-13T01:11:08 But, they'd give anyone the ability to submit jobs to the compile farm. 2010-09-13T01:11:25 Mathnerd314: yeah. you can have cabal call a makefile or whatever you want 2010-09-13T01:11:27 jmcarthur: ok 2010-09-13T01:11:34 The compile farm would use your Makefile to execute make AS root. 2010-09-13T01:11:38 Mathnerd314: in fact, Setup.hs is an arbitrary program 2010-09-13T01:12:12 Mathnerd314, rwa: http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/backend/compile_anything.py 2010-09-13T01:12:23 I tested by confirming I could write a makefile that ended up dumping the shdaowed passwrd file to my home directory. 2010-09-13T01:12:26 not all of those languages are actually enabled, by the way 2010-09-13T01:12:29 then PROMPTLY deleting it. 2010-09-13T01:12:37 bss03: nice 2010-09-13T01:13:36 *** tapwater has quit IRC (Quit: tapwater) 2010-09-13T01:15:06 urgh, why can't j3camero just sandbox compilation like he does the runtime :-( 2010-09-13T01:15:42 Mathnerd314: honestly the runtime is not well sandboxed yet either. we're going to have to work on it some 2010-09-13T01:15:52 especially if we do this distributed TM idea 2010-09-13T01:16:32 Mathnerd314: why do you need cabal anyway? 2010-09-13T01:17:10 it's just a nicer way to build 2010-09-13T01:17:12 http://www.benzedrine.cx/planetwars/canvas?game_id=1284354371|deepblue|bartwe.4.1 2010-09-13T01:17:28 Mathnerd314: i agree, but it's not like *you* have to invoke ghc manually ;) 2010-09-13T01:17:38 I like what I see there :) 2010-09-13T01:17:58 anybody here go TracerBot10? 2010-09-13T01:18:01 got... 2010-09-13T01:18:07 jmcarthur: also, if we could get it to get packages from hackage... 2010-09-13T01:18:25 Mathnerd314: include them with your bot 2010-09-13T01:18:40 yes, but :-( 2010-09-13T01:18:43 that's what i had to do last year 2010-09-13T01:18:45 i know, it sucks 2010-09-13T01:18:55 i'm still hoping we can clean up the sandbox so we can allow binaries 2010-09-13T01:19:22 i *really* want to use GHC HEAD because i'm using vector and fusion is vastly improved in HEAD 2010-09-13T01:19:42 in fact, with 6.8.2, i might have to abandon vector 2010-09-13T01:19:46 don't remember if i had to do that last time 2010-09-13T01:20:39 can you get HEAD to build on their old, creaky 8.04 system? 2010-09-13T01:21:43 ^ jmcarthur 2010-09-13T01:21:56 *** sigh has quit IRC (Remote host closed the connection) 2010-09-13T01:22:28 I've still successfully avoided MTL -- or anything that's not in the ghc6.8.2 package. 2010-09-13T01:22:59 But, my bot really isn't smart enough to be #66, it's just hyper aggresive in a few different ways. 2010-09-13T01:26:51 *** drewrobb_ has joined #aichallenge 2010-09-13T01:27:18 *** sigh has joined #aichallenge 2010-09-13T01:28:01 Do you think the strateg(ies) for single player work well in multi-player? 2010-09-13T01:28:24 I've been thinking about testing my bot on some 3-player maps. 2010-09-13T01:28:31 it's been decided that there will be no multi-player in the tournament, right? 2010-09-13T01:28:33 Me vs. DualBot vd. RageBot. 2010-09-13T01:28:41 rwa: is that official? 2010-09-13T01:29:08 rwa: I haven't heard anything official, but I doubt multi-player will be part of it. 2010-09-13T01:29:09 specs link in topic 2010-09-13T01:29:10 I think it was j3camero that said so, so I think so 2010-09-13T01:29:44 The spec on the website tends to imply games are always 2 players. 2010-09-13T01:30:09 hmm... regarding channel topic: "No stupid questions." Does this mean that there is no such thing as a stupid question, or that stupid questions are no allowed? 2010-09-13T01:30:17 And was that a stupid question? 2010-09-13T01:30:30 *** drewrobb has quit IRC (Ping timeout: 276 seconds) 2010-09-13T01:30:39 sigh: Yes and yes. :P 2010-09-13T01:30:47 >_< 2010-09-13T01:31:29 It means there are no stupid questions; so, just ask your question and don't ask "Can I ask a question?". 2010-09-13T01:31:53 yes, that makes the most sense :) 2010-09-13T01:32:59 Mathnerd314: no way. i can't even get 6.12 to build on it 2010-09-13T01:33:21 *** hornairs has quit IRC (Quit: hornairs) 2010-09-13T01:33:25 jmcarthur: it isn't easier to build HEAD than 6.12? 2010-09-13T01:33:34 why would that be? 2010-09-13T01:35:01 it gets more portable over time... 2010-09-13T01:35:18 apparently not from 6.8 to 6.12 2010-09-13T01:35:29 the problem is actually running it 2010-09-13T01:35:36 6.12.3 segfaults on it 2010-09-13T01:35:42 is there an issue? 2010-09-13T01:35:44 both prebuilt and custom built 2010-09-13T01:35:49 dunno 2010-09-13T01:35:53 *** caytchen has joined #aichallenge 2010-09-13T01:35:56 i got tired of bothering 2010-09-13T01:36:07 i'd rather just fix the sandbox and be allowed to run binaries 2010-09-13T01:36:52 tired... indeed, I should sleep 2010-09-13T01:37:29 me too 2010-09-13T01:38:43 is there any kind of rough eta on when the official server will start running games again? 2010-09-13T01:40:32 ok 2010-09-13T01:40:37 good night guys 2010-09-13T01:40:48 *** deepblue has quit IRC (Quit: Page closed) 2010-09-13T01:43:15 *** retybok has joined #aichallenge 2010-09-13T01:43:42 *** DirtyKeyboard has quit IRC (Quit: DirtyKeyboard) 2010-09-13T01:44:05 did anything notable happen since 8 hourse ago? ;-) 2010-09-13T01:44:08 *hours 2010-09-13T01:44:29 *** fitzgen has quit IRC (Quit: Lost terminal) 2010-09-13T01:44:36 negative 2010-09-13T01:44:43 jmcarthur: groupBy ((==) `on` f) . sortBy (comparing f) $ array ... is that fast enough, or should be I doing something with IntMap there? 2010-09-13T01:45:44 *** Theobon has joined #aichallenge 2010-09-13T01:46:09 *** caezar has quit IRC (Quit: Page closed) 2010-09-13T01:46:34 jmcartur: like unionsWith mappend $ map (\x -> IM.singleton (f x) [x]) array? 2010-09-13T01:46:47 Er, sorry read "array" as "list". 2010-09-13T01:47:48 bss03: I'm guessing he's asleep, like I will be in ~2 minutes 2010-09-13T01:48:13 Mathnerd314: Your opinion on my question? Also, sleep well. 2010-09-13T01:50:21 Anyone else use extremely simple approaches? Mine is ~60 lines of c++ :) 2010-09-13T01:50:35 bss03: I have no idea as far as performance. But I've heard IntMap is faster than lists 2010-09-13T01:50:47 retybok: including I/O? 2010-09-13T01:51:15 retybok: I'm currently at 161 lines of Haskell. 2010-09-13T01:51:40 mine is already hundreds of lines long just updating data each move :| 2010-09-13T01:51:49 Most of it is the "where" clause on the doTurn function. :P 2010-09-13T01:53:25 i had better go to bed earlier tonight, i skipped it saturday night, small sleep sunday, then was up till 5 again last night having to get up at 7 30 2010-09-13T01:53:53 going to bed, should I leave my bot going on the tcp server? 2010-09-13T01:54:16 depends, if you're not getting debugging data there probably isn't much point ;) 2010-09-13T01:54:30 having said that, i started it this morning before uni so I could watch while I was there 2010-09-13T01:54:55 No, 60 lines of my code, I didn't include the boilerplate from the starter pack. 2010-09-13T01:55:35 I must say I'm quite happy to be 56th on the tcp server with such a simple approach :) 2010-09-13T01:55:50 retybok: Were you have to submit to the official server? 2010-09-13T01:55:50 I do get obliterated by all of the good bots though :-D 2010-09-13T01:56:04 bss03: I haven't yet, the site was too slow yesterday 2010-09-13T01:56:31 shawnshekari: Please do, it provides testing for other bots 2010-09-13T01:56:58 will do 2010-09-13T01:57:16 I modified tcp.c so it will just keep looping :) 2010-09-13T01:57:27 shawnshekari: read as "it helps other people beat your bot" 2010-09-13T01:57:41 that's a good thing :) 2010-09-13T01:58:12 also I am logging and it is tweaking it's parameter weights so it just might get smarter on you. :P 2010-09-13T01:58:52 whatever, I'm asleep 2010-09-13T01:58:56 *** Mathnerd314 has quit IRC (Quit: ChatZilla 0.9.86-rdmsoft [XULRunner 1.9.2.8/20100722155716]) 2010-09-13T01:59:04 gn all 2010-09-13T01:59:07 *** jukkerknott has quit IRC (Quit: Page closed) 2010-09-13T01:59:15 *** shawnshekari has left #aichallenge 2010-09-13T02:02:56 *** retybok has quit IRC (Quit: leaving) 2010-09-13T02:06:38 damn you people, going to bed right as I get home from uni and all 2010-09-13T02:06:50 i might actually have to do some work 2010-09-13T02:07:06 Not true 2010-09-13T02:07:14 I never sleep 2010-09-13T02:07:25 i'm at work and I have to do work :( 2010-09-13T02:07:39 eek 2010-09-13T02:07:55 I just about have a real time visualizer for the tcp contest done 2010-09-13T02:08:04 ah, nice 2010-09-13T02:08:05 rad 2010-09-13T02:12:27 Alright, I have a small shell loop running my bot on the TCP server all night. 2010-09-13T02:12:36 i think i'm going to end up doing at least two different bots 2010-09-13T02:12:42 cause i'm not sure what exactly is going to work best 2010-09-13T02:12:44 Hopefully, I can generate some good data for people. 2010-09-13T02:13:59 ohh, i just had a really good idea though ehehe :) 2010-09-13T02:14:56 leave some ideas for the rest of us! 2010-09-13T02:15:17 it's inevitable that people will overlap on strategies 2010-09-13T02:15:24 *** bss03 is now known as bss03-AFK 2010-09-13T02:15:33 lol 2010-09-13T02:16:22 *** Theobon has quit IRC (Ping timeout: 252 seconds) 2010-09-13T02:22:32 *** Obeleh has joined #aichallenge 2010-09-13T02:22:55 vortex8 your bot is bugged, I beat it - http://www.benzedrine.cx/planetwars/canvas?game_id=1284356246|vortex.8|Cyndre 2010-09-13T02:24:25 Cyndre: you've been taking lesson's from ragebot? 2010-09-13T02:24:28 *lessons 2010-09-13T02:25:31 if myGrowth > enemyGrowth * .5: 2010-09-13T02:25:40 just made the bot think it was winning sooner 2010-09-13T02:26:14 interesting, so it always starts the game in rage mode? 2010-09-13T02:26:26 well playing with it 2010-09-13T02:26:41 and about to rewrite my planet selector - alot of people have said its bad 2010-09-13T02:27:09 yeah, planet selector is one of the trickiest parts... I've seen a lot of bots that do it badly 2010-09-13T02:27:16 mine 2010-09-13T02:27:42 so going to work on a distance, growth, shipCost formula (currently only using distance so this could be the tipping part for my bot) 2010-09-13T02:27:44 *** Obeleh has quit IRC (Read error: Connection reset by peer) 2010-09-13T02:27:45 well, in the game you linked it was good 2010-09-13T02:28:05 Is there a set 'size' for all maps? 2010-09-13T02:28:13 http://www.benzedrine.cx/planetwars/canvas?game_id=1284358198|vortex.8|Cyndre 2010-09-13T02:28:25 *** Obeleh has joined #aichallenge 2010-09-13T02:28:27 was doing good then crapped out 2010-09-13T02:28:56 but there we go picking the hardest planet 2010-09-13T02:29:11 hmmm... taking that center 103 was not good i think 2010-09-13T02:29:14 yeah 2010-09-13T02:29:28 so time to work on some planet taker decider 2010-09-13T02:29:38 good luck 2010-09-13T02:29:41 and Im thinking of a total ships in air limiter 2010-09-13T02:29:45 *** doyoulikeworms has joined #aichallenge 2010-09-13T02:29:47 soowoop 2010-09-13T02:30:05 personally, I can't see the benefit in such a limiter 2010-09-13T02:30:15 over commited 2010-09-13T02:30:33 would be based on my ships/yourShips/airShips 2010-09-13T02:30:33 less planets defended when more ships in the air 2010-09-13T02:30:52 thems the words I needed, ty obeleh 2010-09-13T02:31:08 but so far i havent used a limiter 2010-09-13T02:31:29 im doing fine without 2010-09-13T02:31:42 I jut make sure my planets have enough ships to defend themselves... after that I care how many ships are launched 2010-09-13T02:31:49 was thinking of waiting 1 turn, and seeing if any of the first turn planets the opponent went after would be easy to snag for planetGrowth + 1 ships 2010-09-13T02:32:14 that's a good strategy if the starting planets are close, i think 2010-09-13T02:32:19 not if they are far 2010-09-13T02:32:33 well it would be based on if you could get to the enemyFleets Destination Planet first 2010-09-13T02:32:58 er 1 turn after they land 2010-09-13T02:33:40 oh i see what you mean, if the enemy was silly enoguh to attack neutrals which were closer to you? 2010-09-13T02:34:01 Wow, the canvas visualizer is so much faster and non-laggy now, when did that happen? 2010-09-13T02:34:41 it's been around for while under a different url, but they recently switched it to the official visualiser 2010-09-13T02:35:56 guys anyone want to give me a really simple explanation of how to make multiple planets send fleets in a single turn? 2010-09-13T02:36:35 call IssueOrder multiple times 2010-09-13T02:37:48 so, have a loop around IssueOrder with the code to find source inside the loop? 2010-09-13T02:38:45 you call IssueOrder once for each fleet you want to send, yes 2010-09-13T02:39:34 you can call it as many times as you want (so long as the fleets are valid) 2010-09-13T02:41:40 *** fx4m has quit IRC (Ping timeout: 252 seconds) 2010-09-13T02:41:51 does my_planets or pw.MyPlanets() return how many plannets i have control of? 2010-09-13T02:42:28 sorry for asking so much stuff im not too good at this 2010-09-13T02:42:41 what language are you using? 2010-09-13T02:42:52 c++ 2010-09-13T02:43:17 look in PlanetWars.h, it has a description of the methods you can call 2010-09-13T02:43:40 that will tell you what MyPlanets does and what it returns 2010-09-13T02:43:59 I just made a simple dictionary to track how many ships each planet have so I dont send to many 2010-09-13T02:45:09 and have a command that makes sure everything is valid and safe and removes the ships and issues the order 2010-09-13T02:45:34 Cyndre: PLanet.NumShips() ? 2010-09-13T02:46:01 does it track if I send ships out? (issueOrder, check numships? 2010-09-13T02:46:33 nope, issueorder only outputs a command, doesn't update state 2010-09-13T02:46:47 or make sure the order is even valid 2010-09-13T02:47:05 *** kozlovsky has joined #aichallenge 2010-09-13T02:47:59 Cyndre: no you have to use Planet.RemoveShips 2010-09-13T02:49:04 should actually just build that into issueOrder so you can just check numships 2010-09-13T02:50:08 *** Utkarsh has quit IRC (Ping timeout: 245 seconds) 2010-09-13T02:50:58 cyndre when you issueorder make sure you adjust numships for you source planet 2010-09-13T02:51:09 your 2010-09-13T02:51:24 Ive just been tracking it in a seperate dict, but should just add the remove command to isssueOrder 2010-09-13T02:51:36 yes 2010-09-13T02:52:46 you can update anything in your pw gamestate 2010-09-13T02:52:52 *** EvGenius has joined #aichallenge 2010-09-13T02:53:12 and imo you should do so for as much as you could use 2010-09-13T02:53:12 *** bduc has joined #aichallenge 2010-09-13T02:53:55 dont hesitate to change the original source files 2010-09-13T02:54:11 make them do what you want/expect 2010-09-13T02:54:27 yeah, I've already re-written most of the stuff that the starter pack had 2010-09-13T02:54:52 http://www.benzedrine.cx/planetwars/canvas?game_id=1284360101|Cyndre|caytchen yay 2010-09-13T02:55:08 horrible second planet choice though 2010-09-13T02:55:17 *** Utkarsh has joined #aichallenge 2010-09-13T02:56:10 not horrible... horrible is if that planet was neutral 2010-09-13T02:56:11 yep, my planet wars is almost entirely different 2010-09-13T02:56:19 information about an object is often best stored in that object 2010-09-13T02:56:24 hi guys. I suppose all games, ratings and submissions are currently disabled? I have submitted a bot last evening and now it's morning and still "No game information available at this time" 2010-09-13T02:56:35 *** phreeza has joined #aichallenge 2010-09-13T02:56:59 i havent seen my bot match for a while either 2010-09-13T02:57:23 me neither... I should set myself up on the tcp server 2010-09-13T03:00:11 Use the magical Qt client! 2010-09-13T03:00:23 i was hoping someone would port it to msvc 2010-09-13T03:00:40 if i could i would have 2010-09-13T03:00:48 The Qt client will compile on win and use m$ widgets 2010-09-13T03:00:56 I just dont have a win install to compile on 2010-09-13T03:01:04 yes, cygwin is pretty much a pain in the @$$ 2010-09-13T03:01:18 You dont need cygwin, itll work natively 2010-09-13T03:01:31 Qt is awesome like that :) 2010-09-13T03:01:48 but I need to get QT first somewhere, right? 2010-09-13T03:01:56 To compile it 2010-09-13T03:02:00 its free 2010-09-13T03:02:13 thanks greg ill try it soon 2010-09-13T03:02:18 :) 2010-09-13T03:02:27 hmmm, here's a random way you could debug on the official server if you really wanted, if something goes wrong, send a number of fleets from one planet to signal which error :P 2010-09-13T03:02:46 like say you had to quite your makemove cause you were taking too long 2010-09-13T03:02:50 quit* 2010-09-13T03:03:32 antimatroid: I was doing that - sending zero ships to planets where some rule is kicking in so I know that rule is on 2010-09-13T03:03:40 *** gw has quit IRC (Ping timeout: 252 seconds) 2010-09-13T03:06:19 oh boy, i am now ready (have it sorted mentally) to implement a proper strategy 2010-09-13T03:06:35 hopefully i can do it by the end of the week with all my work, i want to try lots 2010-09-13T03:07:30 *** Obeleh has quit IRC (Quit: Bye) 2010-09-13T03:07:51 *** GordonQuad_ has joined #aichallenge 2010-09-13T03:10:03 what timeout is used at tcp server? 2010-09-13T03:10:30 does it include time to transmit data over TCP? 2010-09-13T03:10:53 EvGenius: Its > 1sec 2010-09-13T03:11:06 Im not sure the exact number 2010-09-13T03:11:52 I mean, I have pings to 213.3.30.106 up to 500 ms 2010-09-13T03:12:13 probably that's the reason why I always get timeout on tcp server 2010-09-13T03:14:32 *** syntaxglitch has quit IRC (Ping timeout: 272 seconds) 2010-09-13T03:14:44 *** Appleman1234 has quit IRC (Read error: Connection reset by peer) 2010-09-13T03:17:54 i would consider maybe changing tcp to give benefit of the doubt and not count a timeout game to either players ranking 2010-09-13T03:18:30 *** Appleman1234 has joined #aichallenge 2010-09-13T03:18:52 *** dstufft has joined #aichallenge 2010-09-13T03:18:54 yay 2010-09-13T03:19:02 *** dstufft_ has quit IRC (Quit: Page closed) 2010-09-13T03:20:20 and i really liked the idea about (pseudo) removing the maximum turns before game is called off, and rather incrementally reduce the band of difference between players number of ships required before a player gets the win if they have a greater difference than the current band, once that gets to a certain point (not zero, as equal matches could presumably cycle on who has the most ships in a particular turn) then the ga 2010-09-13T03:23:24 i'd like to see how the games play differently if you use speed chess style timing instead of 1s per turn 2010-09-13T03:23:44 with unlimited amounts of turns 2010-09-13T03:23:51 yeah, 0.1 secnds or something would be cool 2010-09-13T03:23:55 no 2010-09-13T03:23:56 also, n-player matches would be cool 2010-09-13T03:24:00 speed chess is 2010-09-13T03:24:04 you get 15 minutes TOTAL 2010-09-13T03:24:16 if you take 1 minute per turn, well then you only get 15 turns 2010-09-13T03:24:37 i don't think that's really work very well for this 2010-09-13T03:24:51 hey, if we are talking about interesting variants, how about no time limits... you can fire of commands at anytime 2010-09-13T03:25:14 sigh: you're going to a continuous game 2010-09-13T03:25:20 yup 2010-09-13T03:25:35 would be pretty cool 2010-09-13T03:25:37 antimatroid: sure it would, you'd have to balance between heavy routines and light routines 2010-09-13T03:25:53 i like that idea actually 2010-09-13T03:25:54 aka if you can do a lot of math, but get it done in 15 turns, then take 1 minute a turn 2010-09-13T03:26:06 that should really be how it's done actually 2010-09-13T03:26:30 although would that increase or decrease the speed advantage for us c++ians? 2010-09-13T03:26:37 i think it would increase it yeah? 2010-09-13T03:27:04 probably increase it 2010-09-13T03:27:21 also, did i read right in that we do get 3 seconds for first move again? 2010-09-13T03:27:40 currently with 1second per turn, you get the most "Bang for buck" by trying to toe as closely to the 1 second line as possible 2010-09-13T03:28:00 i couldn't use 0.1 seconds if i tried atm 2010-09-13T03:28:06 *** doyoulikeworms has quit IRC (Quit: Leaving) 2010-09-13T03:28:28 oh god, i'm reminded of 0.02 dollars and 0.02 cents 2010-09-13T03:28:37 *** aerique has joined #aichallenge 2010-09-13T03:28:47 One big advantage of faster bots is that you can run more tests locally 2010-09-13T03:28:50 if you have not seen that on youtube, i would highly recommend it 2010-09-13T03:28:51 faster iterations 2010-09-13T03:29:08 sigh: yep, i'm not looking forward to the bots on tcp taking ages 2010-09-13T03:29:20 antimatroid: I'd hate for the number of turns per match to be changed too far down the track - I suspect some peoples strategies are hardcoded around that number of turns, unless they pass the number in to the bots so it can be part of the strategy 2010-09-13T03:29:49 i wouldn't hardcode anything like that personally 2010-09-13T03:30:08 Zaphus: i doubt it's less hardcoding, and more some stragies favor a long game, some favor a short game 2010-09-13T03:30:23 someone needs to remake the core server part and make it up to like 16 players and maps with like 500 planets 2010-09-13T03:30:39 it already accepts up to 4 2010-09-13T03:30:44 if you wouldn't hardcode it, you'll give some handicap to the opponent) 2010-09-13T03:30:57 and their isn't a limit on planets really 2010-09-13T03:31:15 *** AJC has joined #aichallenge 2010-09-13T03:31:30 other then if you went to 500 planets a lot of bots would time out i bet 2010-09-13T03:31:31 dstufft: there is if they don't want an undesirable amount of bots to fall over in the final tournament 2010-09-13T03:33:23 is it still legal to send fleets to the planet it started at? 2010-09-13T03:33:36 you'd easily have tens of thousands of fleets on such a map 2010-09-13T03:34:06 you also have issues in that numbers have to remain small enough for 32 bit integers 2010-09-13T03:34:07 dstufft: you can't go changing the rules that drastically at the final stages, the rules are basically set now whether they like it or not, there isnt much room for changing them if they want to keep people happy 2010-09-13T03:34:23 antimatroid: game spec says no 2010-09-13T03:34:33 starter packages suggest otherwise 2010-09-13T03:34:39 they'd ruin their own reputation 2010-09-13T03:34:43 *** delt0r___ has quit IRC (Read error: Operation timed out) 2010-09-13T03:35:15 Zaphus: i don't think Baphomet was suggesting to run the contest like that, or atleast that's not how i took it, i think he just wanted to see it 2010-09-13T03:35:20 http://www.ai-contest.com/specification.php says: A bot issues a command with the same source and destination planet and forfeits the game. 2010-09-13T03:35:25 and i admit, id like to see the viz for some huge battles like that 2010-09-13T03:35:31 as an endgame condition 2010-09-13T03:35:32 oh that, cheers 2010-09-13T03:35:33 bugger 2010-09-13T03:35:33 ulope: the framework is giving me some duplicate "ghost" fleets. same size, one turn later. 2010-09-13T03:35:51 oh well, i can handle that, just more mehffort 2010-09-13T03:36:06 *** delt0r___ has joined #aichallenge 2010-09-13T03:36:07 meh, just put a guard in IssueOrder 2010-09-13T03:36:12 siple 2010-09-13T03:36:14 *simple 2010-09-13T03:36:23 but I intend to issue orders there as such 2010-09-13T03:36:34 i will simply have to handle it separately inside that function 2010-09-13T03:37:13 not sure what you mean... that action is equivalent to a no-op, so ignoring it is fine 2010-09-13T03:37:39 i'm not going to spell it out, think about it and i think you'll see why one might want to do that 2010-09-13T03:38:16 *** otherAntimatroid has joined #aichallenge 2010-09-13T03:38:31 ok :) 2010-09-13T03:38:36 :p 2010-09-13T03:39:55 ulope: though it seems only for my own fleets... 2010-09-13T03:40:02 *** sigh has quit IRC () 2010-09-13T03:41:04 *** GordonQuad_ is now known as GordonQuad 2010-09-13T03:43:47 *** dhartmei has joined #aichallenge 2010-09-13T03:44:29 *** dolphin278 has quit IRC (Quit: Page closed) 2010-09-13T03:47:39 *** samulihs_ has joined #aichallenge 2010-09-13T03:48:15 *** samulihs_ is now known as samulihs 2010-09-13T03:50:14 anyone having problems submitting new version of a bot? 2010-09-13T03:50:24 yes, submission is broken 2010-09-13T03:50:59 ok, so it's not just me :) 2010-09-13T03:51:58 someone's probably working on the issue? 2010-09-13T03:52:27 *** GordonQuad has quit IRC (Ping timeout: 252 seconds) 2010-09-13T03:53:31 samulihs: the problems are to do with the load being higher than they anticipated, they're trying to work out how best they can fix this 2010-09-13T03:53:34 may take a little while 2010-09-13T03:53:44 there is an unofficial server you can run on if you want to 2010-09-13T03:53:49 http://www.benzedrine.cx/planetwars/ 2010-09-13T03:54:10 nice! thanks 2010-09-13T03:54:38 samulihs: we've had a lot of issues, we are working on getting through them and renabling things 2010-09-13T03:56:04 yep, I've seen the site's been quite flake here and there. I guess you didn't expect that many people would be interest in the challenge and the infra doesn't take the load 2010-09-13T03:56:30 i hope we get at least a few hundred decent quality bots 2010-09-13T03:56:30 samulihs: don't know what was expected, i just signed on to help last night 2010-09-13T03:57:03 sometimes some of the bots take the system down for one reason or another 2010-09-13T03:57:20 there were some java bots running for hours and shit 2010-09-13T03:57:39 it's always java that's screwing things up :) 2010-09-13T03:57:40 number #10352352 why i dislike java 2010-09-13T03:57:46 Though that's fixed sort of now, harakiri and what not 2010-09-13T03:58:06 procceses get auto killed after using so much cpu time 2010-09-13T03:58:17 kill it with fire? 2010-09-13T03:58:22 yea 2010-09-13T03:58:29 dstufft: for the love of god, timeouts should be a lost game not disqualification 2010-09-13T03:58:32 the kernel tells it to gtfo 2010-09-13T03:58:58 antimatroid: well this has nothing to do with the game engine, at the point this culling happens, the game has been over for a long time 2010-09-13T03:59:08 antimatroid: i don't have any say on those rules though 2010-09-13T03:59:27 i know, just trying to put that opinion across to another person who has slightly more say than i 2010-09-13T03:59:39 also limiting the Address space 2010-09-13T03:59:43 >.> 2010-09-13T03:59:46 *** pierce_ has quit IRC (Ping timeout: 252 seconds) 2010-09-13T03:59:48 hope no bots need 196M of ram 2010-09-13T03:59:54 or else i'll have to boost it 2010-09-13T04:00:00 well 196000kb 2010-09-13T04:00:21 i can't remember how much i was using for tron 2010-09-13T04:00:29 i think i was hitting around 125mb maybe 2010-09-13T04:00:32 otherAntimatroid? 2010-09-13T04:00:42 *** TapewormGalactos has joined #aichallenge 2010-09-13T04:01:08 lots 2010-09-13T04:01:17 that's not very specific 2010-09-13T04:01:30 like up to 500MB at some points i think 2010-09-13T04:01:34 ouch 2010-09-13T04:01:42 maybe we'll have to boost that then ha 2010-09-13T04:01:51 or add formal limits to the spec 2010-09-13T04:01:55 having said that, my current plan isn't going to be using anything near that 2010-09-13T04:03:21 *** cgoddard has joined #aichallenge 2010-09-13T04:03:35 I need an algo for finding a point on a line given two the line endpoints and a percentage distance along that line 2010-09-13T04:03:38 Go! 2010-09-13T04:04:22 I glanced through the sources of the site. why don't you cache stuff like rankings and profile info? 2010-09-13T04:04:42 xnew = xold + cos angle * (length of line * percentage) 2010-09-13T04:05:00 EvGenius: that's going to be happening 2010-09-13T04:05:01 same for y but with sin 2010-09-13T04:05:11 manpower restraints 2010-09-13T04:05:12 TapewormGalactos: xnew and xold are what? 2010-09-13T04:05:20 greghaynes: you can think of it as a weighted average between the two endpoints of the line 2010-09-13T04:06:20 Ah, I get it 2010-09-13T04:06:26 dstufft: OK, I hope that would improve performance 2010-09-13T04:06:34 ie: if the line is (xa,ya) to (xb,yb) then xnew = xa*percentage/100 + xb*(100-percentage)/100 2010-09-13T04:06:37 xold is the x coordinate of one of your two points 2010-09-13T04:06:57 EvGenius: yea it would help, one big issue atm is that we are getting lock contentions on the read/writes 2010-09-13T04:07:09 so i think we'll need to switch some of the tables to innodb 2010-09-13T04:07:12 Cool, and with this ill have the realtime visualization of tcp games complete :) 2010-09-13T04:08:52 *** Obeleh has joined #aichallenge 2010-09-13T04:10:07 greghaynes: x = xA+(xB-xA)*p/100; y = yA+(yB-yA)*p/100; 2010-09-13T04:10:37 ty 2010-09-13T04:10:44 *** otherAntimatroid has quit IRC (Ping timeout: 240 seconds) 2010-09-13T04:11:16 i don't really see how it relate to this contest... 2010-09-13T04:11:27 are you doing some viewer ? 2010-09-13T04:11:30 When displaying the fleet moving 2010-09-13T04:11:31 Yea 2010-09-13T04:13:12 *** otherAntimatroid has joined #aichallenge 2010-09-13T04:15:21 *** j3camero has quit IRC (Ping timeout: 240 seconds) 2010-09-13T04:15:30 guys who knows how to work well with vectors in c++ 2010-09-13T04:15:46 I tend to use Qt' 2010-09-13T04:15:50 but ask 2010-09-13T04:15:53 I may know 2010-09-13T04:16:05 still working in c++ baphomet? 2010-09-13T04:16:14 pretty simple, just trying to make a total of all the numbers in a vector 2010-09-13T04:16:19 *** j3camero has joined #aichallenge 2010-09-13T04:16:34 i have it looping fine just dont know how to make the total add up 2010-09-13T04:16:43 use a for loop or an iterator 2010-09-13T04:16:58 isn't there some reduce function ? 2010-09-13T04:17:01 i just cant get it assigned to a variable 2010-09-13T04:17:37 Ah, total = 0;for(vector::iterator it = v.begin(); it!= v.end(); ++it) total += *it; 2010-09-13T04:18:02 you should declare a variable first and inside the for loop do variable += curValue 2010-09-13T04:18:15 Obeleh thats what im trying to do 2010-09-13T04:18:43 replace v with the name of your vector...that should do it 2010-09-13T04:18:49 should work tho 2010-09-13T04:19:39 that's why PHP is easier that STL in C++: $total = array_sum ($v); :) 2010-09-13T04:20:27 ...int vector_sum(vecotr &v) {total = 0;for(vector::iterator it = v.begin(); it!= v.end(); ++it) total += *it; return total;} 2010-09-13T04:20:33 Now its not 2010-09-13T04:20:45 i just use for(int i=0; i v[i] works 2010-09-13T04:21:04 i fucking love vectors 2010-09-13T04:21:16 eh, I love Qt 2010-09-13T04:21:22 the thing is, you need to write a hundred of such functions to make STL as easy as PHP :) 2010-09-13T04:21:56 Now, lets see your GUI apps in PGP 2010-09-13T04:21:58 er, PHP 2010-09-13T04:22:15 http://pastebin.com/WKpTYcB2 <--- there thats what i have 2010-09-13T04:22:28 making gui's is just rank and unfun 2010-09-13T04:22:42 *** bartwe_ has joined #aichallenge 2010-09-13T04:22:43 heh, only if youre using GTK 2010-09-13T04:23:10 Baphomet: much easier right? 2010-09-13T04:23:13 why GUI? I'm just saying that in PHP it's very easy and quick to manipulate with arrays. they are very flexible and can resemble almost any data structure 2010-09-13T04:23:17 *** espes has joined #aichallenge 2010-09-13T04:23:44 although, for the sake of not going insane, use indentation 2010-09-13T04:23:57 EvGenius: Sure, arrays work fine in php as long as you can figure out what random order the dev who made that function chose for the arguments to be in 2010-09-13T04:24:54 greghaynes: well, yes :) the best case is when you are the author of all functions 2010-09-13T04:25:23 *** vikhyat has joined #aichallenge 2010-09-13T04:25:37 if someone can plz edit that bit in paste bin so i can see how its done, im just trying to work out a total of how many planets i currntly control 2010-09-13T04:25:47 *** mnuriev has joined #aichallenge 2010-09-13T04:25:56 Baphomet: What is the type of my_planets 2010-09-13T04:26:15 Planet ? 2010-09-13T04:26:27 hmm Vector i guess 2010-09-13T04:26:33 ^ 2010-09-13T04:26:58 *** Naktibalda has joined #aichallenge 2010-09-13T04:27:12 I dont know the Planet header, so modify the called function as needed: change line 3 to q += my_planets[i].numShips(); 2010-09-13T04:27:28 oh, wait 2010-09-13T04:27:33 total of number of planets 2010-09-13T04:27:41 am i even doing it right? 2010-09-13T04:27:42 q = my_planets.size() 2010-09-13T04:27:44 I didn't receive any mail within the last 24h. Is the server completely stuck ? 2010-09-13T04:27:50 You dont need a loop. 2010-09-13T04:28:23 so my_planets.size() outputs how many planets you have? no looping needed or anything? 2010-09-13T04:28:25 *** s_denchev has joined #aichallenge 2010-09-13T04:28:32 Yeo 2010-09-13T04:28:33 er, Yep 2010-09-13T04:28:43 @rankings 10 2010-09-13T04:28:44 haaha thanks, ive been trying to do this for like 20 mins 2010-09-13T04:28:44 mnuriev: Top 10 players: dmj111(2835), braddunbar(2815), albert(2792), seigen(2732), Kavish(2707), drewrobb(2695), Obeleh(2679), wtfbbq(2679), sighbot(2676), tinasky(2661) 2010-09-13T04:28:49 Well,how many planets are YOURS 2010-09-13T04:29:10 20 minutes isn't a long time 2010-09-13T04:30:37 *** rafsoaken has joined #aichallenge 2010-09-13T04:30:48 *** mnuriev has left #aichallenge 2010-09-13T04:31:06 Well... 2010-09-13T04:31:10 Is the game stopped ? 2010-09-13T04:31:24 Not on tcp server 2010-09-13T04:31:32 what is tcp server ? 2010-09-13T04:31:42 @whatis tcpserver 2010-09-13T04:31:42 greghaynes: tcpserver = http://www.benzedrine.cx/planetwars/ 2010-09-13T04:31:44 *** pgpaskar_ has quit IRC (Ping timeout: 240 seconds) 2010-09-13T04:32:01 Heres a magical gui client: http://github.com/greghaynes/QTcpPlanetWars 2010-09-13T04:32:02 yes, I cannot load the page. Already tried. 2010-09-13T04:32:10 whaa 2010-09-13T04:32:21 I'm having a little trouble figuring out why this game ends when it does - I don't suppose someone could take a look at it and tell me what I'm missing? http://pastebin.com/n3sK6uLD 2010-09-13T04:32:22 Why not? 2010-09-13T04:32:30 *** pgpaskar_ has joined #aichallenge 2010-09-13T04:36:39 what do you guys use for making guis in c++? 2010-09-13T04:37:10 Qt! 2010-09-13T04:37:12 And only Qt! 2010-09-13T04:38:15 okay, we tried both gtk and wxwidgets, they were both awful 2010-09-13T04:38:26 Yes, you need Qt 2010-09-13T04:38:36 Those two are aweful 2010-09-13T04:39:14 There is a bit of a learning curve with signals / slots if youre not used to that sort of design pattern, it really is a better way though 2010-09-13T04:39:19 greghaynes: Nice but we can't see the game 2010-09-13T04:39:26 Do you accept patches ? 2010-09-13T04:39:32 elijahbal: See branch 2.0 2010-09-13T04:39:34 elijahbal: Yes 2010-09-13T04:40:29 *** rafsoaken has quit IRC (Quit: Page closed) 2010-09-13T04:42:31 greghaynes: so you got the drawing, but you can not play the animation ? 2010-09-13T04:42:45 elijahbal: Im finishing that commit right now 2010-09-13T04:42:58 going to make a copy of the repo 2010-09-13T04:43:10 Go for it 2010-09-13T04:43:38 greghaynes: Why dont you use a pipe with ShowGame.jar ? A lot easier. 2010-09-13T04:43:43 *** hamstah has quit IRC (Ping timeout: 276 seconds) 2010-09-13T04:44:15 elijahbal: eh, all the code for the viewer is about 100 lines, not worth multiprocessing 2010-09-13T04:44:27 how does the main in cpp starter packages know when to finish? 2010-09-13T04:44:35 it's conatined in a while true thingy 2010-09-13T04:45:12 *** sigh has joined #aichallenge 2010-09-13T04:45:24 kill 2010-09-13T04:45:56 wouldn't it have been better for the game engine to just pass a string with a message saying the game is over? 2010-09-13T04:46:22 I probably shouldnt answer that 2010-09-13T04:47:04 do anyway.. :) 2010-09-13T04:51:44 *** cgoddard has quit IRC (Quit: Page closed) 2010-09-13T04:55:22 greghaynes: coding hard , huh ? 2010-09-13T04:55:28 As always 2010-09-13T04:56:03 I was coding my bot in python for so long I got used to being able to float() int's 2010-09-13T04:56:09 Now theres bugs in my placement algo 2010-09-13T04:58:47 *** EvGenius has left #aichallenge 2010-09-13T05:01:45 *** espes has quit IRC (Remote host closed the connection) 2010-09-13T05:04:50 intresting how instable the elo scores are 2010-09-13T05:05:23 *** soswow has joined #aichallenge 2010-09-13T05:07:13 how pairing is done on benzedrine ? 2010-09-13T05:08:20 *** rafsoaken has joined #aichallenge 2010-09-13T05:09:32 Hi to all. Is there will be some way to tell reason of winning/loosing in each game? For example when there is "1 of 1 turns" situation. Does this question looks stupid? =) 2010-09-13T05:10:08 error in bot probably 2010-09-13T05:10:13 soswow: yeah, it's hard to know if it was a server bug or a bot committing suicide. 2010-09-13T05:12:46 AJC: what do you mean? Is it right now hard to tell or it's hard to determine in the game engine? 2010-09-13T05:12:47 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-13T05:12:54 edcba: seems to be a mix of elo and first available 2010-09-13T05:13:19 *** danielvf has joined #aichallenge 2010-09-13T05:13:44 soswow: it's hard to tell watching the replay. 2010-09-13T05:14:01 Naktibalda: YEah, but which one? Illegal turn, time out? Locally works fine. 2010-09-13T05:14:35 soswow: are you using your entire 1s budget? 2010-09-13T05:15:04 AJC: Yeah, I talk exactly about it. There should be some verdict. 2010-09-13T05:15:24 AJC: What do you mean? 2010-09-13T05:15:51 soswow: if you get legal turns offline, then the timeout is most likely... 2010-09-13T05:16:00 soswow: how do you make sure you don't take more than 1s? 2010-09-13T05:16:22 On first turn? it's hard to imagine ... 2010-09-13T05:16:56 soswow: it's your code ;-) 2010-09-13T05:17:07 next up, rolling the tcp code into the binary itself and setup profiling 2010-09-13T05:18:40 AJC: Right! And I can definitely say, that on first turn I don't count factorial of 10^10 2010-09-13T05:19:31 Maybe there something with package. What if I don't add something in my zip? 2010-09-13T05:19:44 soswow: I have a forum post about it, several of the top bots have also succumbed to this, sometimes both bots at the same time - hopefully a way of knowing what happened will be added to the website in teh near future 2010-09-13T05:20:30 Zaphus: Yeah, I saw your post. but there wasn't any answer. =) 2010-09-13T05:21:03 afk 2010-09-13T05:21:56 elijahbal: Ok, fleets are shown now 2010-09-13T05:27:48 *** soswow_mob has joined #aichallenge 2010-09-13T05:32:10 *** soswow_mob has quit IRC (Ping timeout: 252 seconds) 2010-09-13T05:38:59 *** retybok has joined #aichallenge 2010-09-13T05:41:40 hmmm, you can only do a max of 3 players before you no longer have fair starting states when considering the quality of a bot 2010-09-13T05:41:51 meaning you'd nee to do a butload more games to get a reasonable result 2010-09-13T05:41:55 *** EvGenius has joined #aichallenge 2010-09-13T05:42:17 *** Zaphus has quit IRC (Quit: Page closed) 2010-09-13T05:42:30 and you need the map to be symmetric in a 3way split like thing 2010-09-13T05:48:30 guys: http://en.wikipedia.org/wiki/Programmers'_Day is today :D 2010-09-13T05:50:36 Realtime game visualization for the tcp server: http://github.com/greghaynes/QTcpPlanetWars 2010-09-13T05:51:40 *** Ported_ has joined #aichallenge 2010-09-13T05:58:37 *** MatG has joined #aichallenge 2010-09-13T05:59:14 does anyone know what the deal with submissions is at the moment? 2010-09-13T05:59:30 antimatroid: what are you talking about ? 2010-09-13T05:59:34 the whole ai-contest.com site seems to be fried... 2010-09-13T06:00:00 *** infthi has joined #aichallenge 2010-09-13T06:01:10 bartwe_: I want someone to run bots in n player games unofficially at some point in time 2010-09-13T06:01:13 i think it'd be cool 2010-09-13T06:01:39 Ported: right now? 2010-09-13T06:02:46 yep 2010-09-13T06:03:36 is code upload working? i re-submitted code about 12 hours ago, but here are still no changes in my profile 2010-09-13T06:05:20 infthi: i think it is but i haven't tried recently 2010-09-13T06:05:27 http://www.benzedrine.cx/planetwars/ 2010-09-13T06:05:31 that's the best place to test your bot out 2010-09-13T06:05:43 a lot of the top people from last time are testing their early revisions on there 2010-09-13T06:05:56 *** Nuriaion has joined #aichallenge 2010-09-13T06:05:57 aha, so the wrong channel 2010-09-13T06:06:13 well it still worked here 2010-09-13T06:06:25 antimatroid: sounds like fun, i'd need to fix some bugs that don't take enemy vs enemy conditions correctly 2010-09-13T06:06:41 yeah same 2010-09-13T06:06:56 but it'd be cool to see the top bots battle it out if the people were willing to port it over 2010-09-13T06:07:29 shouldn't be too impossible to setup a benzedrine like thing that does that 2010-09-13T06:13:21 Rankings are running again 2010-09-13T06:14:03 And so are official games 2010-09-13T06:15:29 danielvf: i know your guys are busy with the servers but any idea when support for other languages (for which starter packages & instructions have been made) will happen? 2010-09-13T06:16:43 arik181: I'd guess by tomorrow night. It's near the top of the priority list 2010-09-13T06:16:52 *** mistabell has joined #aichallenge 2010-09-13T06:17:02 aerique: which languages ? 2010-09-13T06:17:06 aerique: I'd guess by tomorrow night. It's near the top of the priority list 2010-09-13T06:17:17 Cyndre: congrats ;) 2010-09-13T06:17:23 http://www.benzedrine.cx/planetwars/canvas?game_id=1284364265|dhartmei|caytchen thats some cool game 2010-09-13T06:17:53 edcba: i've seen common lisp, php, ocaml, scheme and possibly more on the forums 2010-09-13T06:18:07 danielvf: alright, thanks 2010-09-13T06:18:19 supporting php... 2010-09-13T06:18:36 it's easy to add support 2010-09-13T06:18:49 i guess you could do almost same client for lisp/ocaml/scheme 2010-09-13T06:19:27 edcba: ruby isn't supported yet i think (not sure) 2010-09-13T06:20:00 I have seen [New language] Ruby topic, but Im not sure do they have a starter package 2010-09-13T06:20:00 *** MatG has quit IRC (Quit: Page closed) 2010-09-13T06:20:36 if you want more ppl to participate i guess php should be first 2010-09-13T06:20:57 you can write in any language for benzedrine just not for the main contest 2010-09-13T06:21:06 php? for ai ? 2010-09-13T06:21:15 it was one of the fun things of the Tron challenge that pretty much any language was supported as long as one made a starter package and it ran on the server 2010-09-13T06:21:16 quick designing 2010-09-13T06:21:33 *** mistabell has quit IRC (Client Quit) 2010-09-13T06:21:35 bartwe_: all winners will be c++ i bet 2010-09-13T06:21:36 is haskell working yet? 2010-09-13T06:21:47 TapewormGalactos: yes 2010-09-13T06:22:10 edcba: why ? 2010-09-13T06:22:14 so you just have to zip your MyBot.hs? 2010-09-13T06:22:19 I want to concentrate on AI, not on learning a "real AI language" 2010-09-13T06:22:21 because you'll need heavy calculation :) 2010-09-13T06:22:27 bartwe_: yes, benzedrine is cool for now 2010-09-13T06:22:30 edcba: not really 2010-09-13T06:22:37 we'll see :p 2010-09-13T06:22:52 1 second per turn is generous 2010-09-13T06:23:40 really ? 2010-09-13T06:23:44 i'm sure if i stay into this i'll try and each every second cpu to make a better future attack predictor 2010-09-13T06:24:10 edcba: what is all winners? top 3, top 10, top 25? 2010-09-13T06:24:11 1 sec is a bit short for the first step 2010-09-13T06:24:27 aerique: top 3 :) 2010-09-13T06:24:45 or top 5 maybe since it seems there is already a lot of ppl 2010-09-13T06:25:33 * edcba looks at current ranking and see top 2 is python powered lol 2010-09-13T06:25:50 but i'm confident ! :) 2010-09-13T06:26:34 edcba: i agree, right now python is to expected because development is faster and these are most likely simple heuristics only, so far 2010-09-13T06:26:59 *** Palmik has joined #aichallenge 2010-09-13T06:27:17 not sure about attack prediction though, i'd guess more time will be spent on search of the decision space 2010-09-13T06:27:32 mono is a slow platform for c# i agree, i might do a source to source when i'm sure i won't be changing anything 2010-09-13T06:27:33 hi 2010-09-13T06:27:37 *** Vio has joined #aichallenge 2010-09-13T06:28:14 mono is slow ?? 2010-09-13T06:28:28 *** espes has joined #aichallenge 2010-09-13T06:28:36 compared to c# on windows? yes 2010-09-13T06:28:42 last time i checked it was faster... 2010-09-13T06:28:47 its slow and has all those subtle api differences 2010-09-13T06:28:56 i'm thinking of maybe building an ai that does extensive search, breaking the time limit, just to see how it plays and maybe derive some heursitics 2010-09-13T06:28:58 but it was mono ppl giving me info :) 2010-09-13T06:29:01 anyone trying that? 2010-09-13T06:29:27 phreeza: i'll do that 2010-09-13T06:30:13 but right now i'm looking how exactly 2010-09-13T06:30:35 alberts c++ engine will probably speed that up quite a bit 2010-09-13T06:30:45 0edcba: the stable mono version you see in contests like this or on debian, etc. is slow as hell 2010-09-13T06:30:50 since you can't really bruteforce every movement with all different amount of ships each turn 2010-09-13T06:31:56 edcba: thats true, i was thinking more like using a heursitic to do the tactics of how exactly to attack/defend planets, and brute forcing which ones to use that heurstic on 2010-09-13T06:32:34 still 2^23 choices per turn, but many can be eliminated as impossible maybe 2010-09-13T06:32:50 2^23 only ?? 2010-09-13T06:33:07 well if you choose to defend or attack a planet or not 2010-09-13T06:33:15 one binary decision per planet 2010-09-13T06:33:30 very simplistic but at least that seems feasible 2010-09-13T06:33:30 the problem is that it isn't binary :) 2010-09-13T06:34:17 no but for the details i was thinking of using a heuristic... like if i attack a planet it sends just the right amount from the closest planets i own, or something like that 2010-09-13T06:34:27 the number of ways to attack 10 planets with 100 units in extremely complex enough 2010-09-13T06:34:27 ok 2010-09-13T06:36:15 Is there organizers people here? 2010-09-13T06:36:27 the ops maybe :) 2010-09-13T06:37:03 US should be sleeping now 2010-09-13T06:38:19 Naktibalda: you must be right ai-contest.com still up ;) 2010-09-13T06:39:00 Hmm... is the new pairing system in action? :) I just jumped in rankings a lot 2010-09-13T06:41:37 *** dmj111 has quit IRC (Quit: ERC Version 5.3 (IRC client for Emacs)) 2010-09-13T06:42:32 hmz, pplaced 2223 with the bartwe.1, which has negative ELO on benzedrine :) 2010-09-13T06:42:37 23 2010-09-13T06:43:21 nice lol 2010-09-13T06:43:22 lucky you. I am placed nowhere and have no games played since yesterday :) 2010-09-13T06:48:46 antimatroid, are you recreating fleets each turn, or are you just manually decreasing the turnReamining counter and when it hits 0, you erase the fleet and just create fleets which have turnsRemaining == totalTripLegth - 1? 2010-09-13T06:50:23 soswow: why did you want organizers ? 2010-09-13T06:51:53 Server is playing 14 games per minute now 2010-09-13T06:52:13 edcba: I just wanted to ask from the first hands about currency situation or ask them to keep us (public) in touch with news, activities, changes, future plans itc. I can't find some constant place, where I can find such info. 2010-09-13T06:52:34 I'm an organizer 2010-09-13T06:52:35 the forum ? 2010-09-13T06:52:51 I think we've got the site stability fixed 2010-09-13T06:52:58 Server super happy this morning 2010-09-13T06:53:00 Yeah, I don't see there much activity from them ... -\ 2010-09-13T06:53:06 Games are playing again. 2010-09-13T06:53:11 danielvf: Nice! 2010-09-13T06:53:19 all seems nice 2010-09-13T06:53:22 We'll be working on the submissions next 2010-09-13T06:53:36 and the bugs in starter packs !? :) 2010-09-13T06:53:37 (some of them don't seem to go through right away) 2010-09-13T06:53:59 After that, it's time to add more languages 2010-09-13T06:54:33 danielvf: Is there will be some kind of status/verdict for each game? Like "time out" or "illegal turn by player #" 2010-09-13T06:54:53 14 games per minute is not a very fast speed 2010-09-13T06:55:23 danielvf: for better understanding of loosing in case of "1 of 1 turns" situation 2010-09-13T06:55:45 danielvf: there are post on forum about it as well 2010-09-13T06:56:07 EvGenius: if you get up to 1 second per turn and up to 200 turns... then 14 games a sec sounds fast 2010-09-13T06:56:16 soswow: Yes, that's an issue. Will probably come after getting the new languages up. 2010-09-13T06:56:25 delt0r___: right 2010-09-13T06:56:35 soswow: did you create an issue? 2010-09-13T06:56:36 14 games a minute 2010-09-13T06:56:41 but still fast 2010-09-13T06:57:40 danielvf: How is java "sand boxed" ? with Systrace? 2010-09-13T06:57:46 Palmik: I have to read in every fleet each turn anyway, so I just add new ones (including my own) to my fleets, also make sure you update your planet information each turn because planets might be distanced 1 apart 2010-09-13T06:58:01 *** retybok has quit IRC (Ping timeout: 265 seconds) 2010-09-13T06:58:17 *** a1k0n_ has joined #aichallenge 2010-09-13T06:58:50 delt0r___: gotta go afk, sorry 2010-09-13T06:58:55 a1k0n_ which one is the real a1k0n ? 2010-09-13T06:59:04 Naktibalda: http://goo.gl/ZnXc 2010-09-13T06:59:21 *** a1k0n_ has quit IRC (Client Quit) 2010-09-13T06:59:34 issues 2010-09-13T06:59:42 bot 2010-09-13T06:59:43 Naktibalda: oh, sorry. 2010-09-13T06:59:43 *** Zaphus has joined #aichallenge 2010-09-13T06:59:48 *** retybok has joined #aichallenge 2010-09-13T06:59:50 where issues 2010-09-13T07:00:13 http://code.google.com/p/ai-contest/issues/list 2010-09-13T07:00:25 contestbot issues 2010-09-13T07:00:26 caytchen: issues = http://code.google.com/p/ai-contest/issues/list 2010-09-13T07:00:28 antimatroid, ok, thanks :) 2010-09-13T07:00:38 Naktibalda: yeah yeah. I didn't make it. 2010-09-13T07:01:23 FYI: message from admins http://ai-contest.com/forum/viewtopic.php?f=18&t=566#p3172 2010-09-13T07:01:27 * danielvf afk 2010-09-13T07:04:00 Hey guys, anyone know where to find the source for Showgame.java ? I can't see it in the trunk at code.google.com 2010-09-13T07:04:32 Zaphus: folder "viz" 2010-09-13T07:04:46 ahh, thanks. 2010-09-13T07:05:18 Zaphus: see "MAkefile" there for more info how ShowGame and PlayGame are made 2010-09-13T07:05:27 Also I wonder whether it would be better passing around pointers to Planets and Fleets or values... by passing pointers, it should have much smaller memory footprint, also you would not have to copy whole objects every time you filter them or something 2010-09-13T07:05:58 *** GordonQuad has joined #aichallenge 2010-09-13T07:06:22 soswow: thanks, although I think I'm in over my head... I just desperately want local playback with the ship count on screen :-( was thinking it might be a trivial mod 2010-09-13T07:06:56 Zaphus: I was thinking of doing this for the java TCP 2010-09-13T07:07:47 It wouldn't be hard to write it better than the applet they have (serious performance issues) 2010-09-13T07:07:58 delt0r___ : cool. If there was some way to run the canvas visuallizer locally, that would do it too - no idea how that works though 2010-09-13T07:09:14 well right now the java is broken on the ranking server... And i am not doing anything else.... 2010-09-13T07:09:36 delt0r___ FTW ! 2010-09-13T07:10:00 Zaphus: did you mean WTF? =) 2010-09-13T07:11:03 I will work on something this afternoon.. But what did you really want Zaphus --real time local playgame with visuals.. or TCP with visuals? 2010-09-13T07:11:13 *** Vio has quit IRC (Disconnected by services) 2010-09-13T07:11:31 updated tcp.c with password support: http:/www.benzedrine.cx/planetwars/tcp.c 2010-09-13T07:11:52 dhartmei: you rock 2010-09-13T07:12:20 DON'T use the same password as on offical web server :) 2010-09-13T07:12:49 dhartmei: IS this TCP things are official or just fans initiative? Still can't get all this ... 2010-09-13T07:13:03 it's backwards compatible, as long as you don't use passwords, nothing changes. once you set a password, you have to supply it every time. 2010-09-13T07:13:35 dhartmei: so if someone sets a pw to your name before, you're pretty much screwed? :) 2010-09-13T07:13:36 soswow: it's completely unofficial, but works quite well :) 2010-09-13T07:13:46 delt0r___ real local, not TCP - so I can do it offline 2010-09-13T07:13:47 *** vikhyat has quit IRC (Ping timeout: 252 seconds) 2010-09-13T07:13:52 ratatata: just add a version number to your name 2010-09-13T07:13:53 ratatata: you just pick a new user name and use a password with that the first time :) 2010-09-13T07:13:57 but yes, that's an attack :) 2010-09-13T07:14:00 ratatata: Right now anyone can just use your username! 2010-09-13T07:14:33 it's not your name, it's bot's name 2010-09-13T07:14:35 Zaphus: What is the problem with ShowGame ? 2010-09-13T07:14:45 dhartmei: can you or anyone update the java version that someone posted ? I cant use the c version :-( 2010-09-13T07:14:49 sure its not perfect --but its not bad 2010-09-13T07:15:07 dhartmei: so, all those people who are in rating on benzedrine.cx - are not connected with those who on ai-contest? 2010-09-13T07:15:09 delt0r___ : I want the ship counts when playing back, like the canvas viewer on th webpage has 2010-09-13T07:15:45 Zaphus: But you get the ship counts in showGame? 2010-09-13T07:15:48 Zaphus: One thing, that I change after first opening of ShowGame - I changed speed of playing and get it out into command line parameter 2010-09-13T07:15:53 It's to slow by default 2010-09-13T07:16:14 *** Obeleh has quit IRC (Quit: Bye) 2010-09-13T07:16:48 soswow: I click furiously on >| 2010-09-13T07:16:59 I also need a new mouse /: 2010-09-13T07:17:01 delt0r___ no you dont, at least not in my version (v1.2, that a1k0n posted) 2010-09-13T07:17:21 *** Obeleh has joined #aichallenge 2010-09-13T07:17:22 Zaphus: I do, i get everything you can know 2010-09-13T07:17:23 delt0r___: YEah, I did this for some time and decided to apply my 4-years of java experience =) 2010-09-13T07:17:30 Perhaps just update? 2010-09-13T07:17:48 delt0r___ sorry, I mean the *totals* not the individual fleet numbers 2010-09-13T07:17:56 Oh 2010-09-13T07:17:57 where is the latest version ? 2010-09-13T07:18:09 *** ratatata has quit IRC (Ping timeout: 252 seconds) 2010-09-13T07:18:31 I guess from the start page.. They have updated once IIRC because early on it was kinda Borked 2010-09-13T07:18:39 Btw, I did update of the official code yesterday and there was compile errors. It's not very nice to commit code, that not compiles =\ 2010-09-13T07:18:48 But totals is *not* shown in mine 2010-09-13T07:19:01 *** EvGenius has left #aichallenge 2010-09-13T07:19:17 I saw totals only on www.benzedrine.cx player 2010-09-13T07:19:26 dhartmei: how does the decay of low number of games bots work ? 2010-09-13T07:19:29 *** Obeleh has quit IRC (Read error: Connection reset by peer) 2010-09-13T07:19:45 http://www.benzedrine.cx/planetwars/getplayer?player=McLeopold 2010-09-13T07:19:47 i purge all data older than 8 hours every hour 2010-09-13T07:19:53 the data file grows >200MB otherwise 2010-09-13T07:19:59 for example doesn't really look like a top 3 ish bot 2010-09-13T07:20:00 *** Obeleh has joined #aichallenge 2010-09-13T07:20:13 what is the timeout on the tcp server? I keep getting timeouts :( 2010-09-13T07:20:25 3 seconds 2010-09-13T07:20:59 dhartmei: also do you consider it spamming the board if i keep the weaker older iterations running ? 2010-09-13T07:21:13 thanks dhartmei 2010-09-13T07:21:30 *** jmreardon has joined #aichallenge 2010-09-13T07:21:32 no, do so if you like 2010-09-13T07:21:37 bartwe_: please keep them running 2010-09-13T07:21:42 it's great to have diversity 2010-09-13T07:21:53 the server can handle many more concurrent connections 2010-09-13T07:22:04 yeah but i got 4 slots in the top10 2010-09-13T07:22:04 if the list gets too long or crowded, i'll solve that :) 2010-09-13T07:22:27 bartwe_: that's because other contestants suck :-D 2010-09-13T07:22:39 (myself included!) 2010-09-13T07:23:05 ok i'll keep the major versions up if they are diverse enugh 2010-09-13T07:23:42 bartwe_: yeah, I think that's the idea 2010-09-13T07:23:49 how do you run bots in a loop? simple shell script or something more complicated? 2010-09-13T07:24:07 Naktibalda: while :; do ... :) 2010-09-13T07:24:33 while true; do tco ./bot ; done 2010-09-13T07:24:49 from inside a makefile 2010-09-13T07:25:08 only problem i have with that approach is that i have no clean way to kill it without losing a battle 2010-09-13T07:25:42 is there the equivalent of if kbhit() ? 2010-09-13T07:25:43 I have a sleep in my loop, crude, but works 2010-09-13T07:25:46 improvement suggestions welcome 2010-09-13T07:26:57 *** vikhyat has joined #aichallenge 2010-09-13T07:27:05 http://www.benzedrine.cx/planetwars/loop.sh 2010-09-13T07:27:20 makes nice logs in a subdir with opponent name and result 2010-09-13T07:27:49 so if you spot an intersting game in the visualizer, you can read the debug log from you bot about it 2010-09-13T07:31:42 *** mistabell has joined #aichallenge 2010-09-13T07:31:53 this game counted as a loss http://www.ai-contest.com/visualizer.php?game_id=4451393 seems more like a draw 2010-09-13T07:32:23 also, noticed a bunch of my previous games disappeared from the "'s latest games" page 2010-09-13T07:34:29 *** mistabell has quit IRC (Client Quit) 2010-09-13T07:35:26 *** mistabell has joined #aichallenge 2010-09-13T07:36:14 http://www.ai-contest.com/profile.php?user_id=6286 several games with Wins after no moves made 2010-09-13T07:36:16 does anyone have the link to albert's c++ game engine? 2010-09-13T07:37:01 bartwe: How many bots do you have? 2010-09-13T07:37:17 sigh: your name isn't alex is it? 2010-09-13T07:37:24 nope 2010-09-13T07:37:34 why? 2010-09-13T07:37:43 nevermind 2010-09-13T07:37:52 ok 2010-09-13T07:38:05 would expect someone to have a username like that who would place like that 2010-09-13T07:38:10 that i know* 2010-09-13T07:39:47 found it: http://github.com/albertz/planet_wars-cpp/blob/master/game.cpp 2010-09-13T07:41:05 *** ratatata has joined #aichallenge 2010-09-13T07:42:18 *** mistabell has quit IRC (Quit: Page closed) 2010-09-13T07:43:27 *** Zaphus has quit IRC (Ping timeout: 252 seconds) 2010-09-13T07:44:08 someone here with a java bot tell me how much memory their bot is using? 2010-09-13T07:46:51 how do you check specifically for mem usage of the bot? 2010-09-13T07:47:12 *** Obeleh has quit IRC (Read error: Connection reset by peer) 2010-09-13T07:47:26 ratatata: what OS are you on 2010-09-13T07:47:32 ubu 2010-09-13T07:48:10 just use top or htop while your bot is running, it's not perfect but it's close enough 2010-09-13T07:49:38 *** RainCT has joined #aichallenge 2010-09-13T07:53:02 dstufft: ~8-12M 2010-09-13T07:53:13 dstufft: but I'm not doing anything Mem heavy 2010-09-13T07:55:31 *** Vi0 has joined #aichallenge 2010-09-13T07:57:03 *** Obeleh has joined #aichallenge 2010-09-13T07:59:35 *** kozlovsky has quit IRC (Ping timeout: 252 seconds) 2010-09-13T08:02:16 i try to use the tcp server with a java bot but i get an error: "execv: java ScalaBot/target/scala_2.8.0/ScalaBot-1.0.min.jar: No such file or directory" 2010-09-13T08:02:25 ./tcp.c.out 213.3.30.106 9999 nuriaion "java ScalaBot/target/scala_2.8.0/ScalaBot-1.0.min.jar" 2010-09-13T08:02:45 remove the quotes 2010-09-13T08:03:04 i suck so bad at this 2010-09-13T08:03:12 execv: java: No such file or directory 2010-09-13T08:03:16 im starting to work out how to program stuff now 2010-09-13T08:03:23 when i removed the quotes 2010-09-13T08:03:24 but everything i do gets me killed 2010-09-13T08:03:28 use full path to the java binary 2010-09-13T08:03:29 Nuriaion: make a wrapper script, it takes a single parameter. 2010-09-13T08:03:36 ah 2010-09-13T08:03:40 ok 2010-09-13T08:03:48 Nuriaion: and a relative path, ./wrapper.sh 2010-09-13T08:03:57 Nuriaion: there is a Java version in the forums (I think under the TCP server thread) which I have been using, fyi 2010-09-13T08:05:27 *** soswow has quit IRC (Ping timeout: 252 seconds) 2010-09-13T08:05:45 Nuriaion: you need to create a script that will call your command 2010-09-13T08:05:56 Thanks everybody 2010-09-13T08:05:57 it's quite stupid but... 2010-09-13T08:05:58 it runs now 2010-09-13T08:06:07 edcba: fixed in the new version of tcp.c :) 2010-09-13T08:06:15 fast 2010-09-13T08:06:38 had same problem with mono mybot.exe :p 2010-09-13T08:06:44 ;) 2010-09-13T08:09:07 *** infthi has quit IRC (Ping timeout: 252 seconds) 2010-09-13T08:09:46 dhartmei: what's an expiration time of games in TCP server? I think that games expire too fast 2010-09-13T08:10:24 *** kuwabara has joined #aichallenge 2010-09-13T08:10:27 hi, there 2010-09-13T08:10:38 the history? 6 hours 2010-09-13T08:10:57 there is still no message on the forums about the java clients all loosing their games on first turn. Any more news here ? 2010-09-13T08:10:58 is there a 1s timeout in tcpserver too ? 2010-09-13T08:11:06 edcba: 3s 2010-09-13T08:11:21 kuwabara: is your bot one of the affected? 2010-09-13T08:11:25 kuwabara: did you use latest PlayGame and ShowGame jars ? 2010-09-13T08:11:41 are the latest available in a repository? 2010-09-13T08:11:47 so much easier to sync :-) 2010-09-13T08:12:01 is this tcp thing for uber programmers who are too paranoid to let their code leave their pc? if so, im sniffing your packets are stealn ur codez 2010-09-13T08:12:16 Baphomet: of course 2010-09-13T08:12:19 dstufft: yes. 2010-09-13T08:12:21 :p 2010-09-13T08:12:33 kuwabara: how long has this been happening? 2010-09-13T08:12:43 dstufft, edcba: look at the top java bots, they are all loosing 100% of their games. 2010-09-13T08:12:53 Baphomet: it's also a way for the server not having to trust any provided code 2010-09-13T08:13:03 and you can program in any languages 2010-09-13T08:13:04 it started this morning, when the games restarted. 2010-09-13T08:13:24 kuwabara: can you tell me how much memory and cputime your bot uses 2010-09-13T08:13:24 that just means java is shit ;) 2010-09-13T08:13:40 on a local run 2010-09-13T08:14:06 dstufft: sorry, no. I'm not at home. 2010-09-13T08:14:15 kuwabara: ok 2010-09-13T08:15:09 well cputime shouldn't matter if its first turn 2010-09-13T08:15:19 the best ranked java bots: Molehill, luismi, OkirFun, Phate666 were all in the top 10 yesterday, and they have lost all of their games on sept 13. 2010-09-13T08:15:51 Wow, my bot played 8 games in single hour, I guess I!m lucky :) 2010-09-13T08:15:56 we added some limits to have the kernel kill run away bot procceses, but cputime is like 10 minutes before the kernel kills it 2010-09-13T08:15:57 is it sad that nearly all the bots i make struggle to beat the default one? 2010-09-13T08:16:05 and ram is like 196000kb 2010-09-13T08:16:50 *** danielvf has quit IRC (Ping timeout: 245 seconds) 2010-09-13T08:16:51 but neither of those should be an issue on the first turn 2010-09-13T08:17:08 *** GordonQuad has quit IRC (Read error: Operation timed out) 2010-09-13T08:17:36 how is RAM computed ? 2010-09-13T08:17:55 computed ? 2010-09-13T08:18:14 a java app doing nothing uses about 26Mb on my host, but VSZ is at 789Mb 2010-09-13T08:18:48 *** EvGenius has joined #aichallenge 2010-09-13T08:19:01 so, if VSZ is checked against 196Mb, there is no way a java bot can enter the aichallenge anymore 2010-09-13T08:19:25 *** danielvf has joined #aichallenge 2010-09-13T08:21:06 look at my profile, the change in rules is quite visible: http://www.ai-contest.com/profile_games.php?user_id=6391 2010-09-13T08:21:13 kuwabara: Address Space, and jesus a javaapp is really got a vsz of 789? 2010-09-13T08:21:44 dstufft: yes, because of all the .so it's linked against. address space is VSZ indeed. 2010-09-13T08:21:53 i'll modify the memory limits then 2010-09-13T08:21:55 kuwabara: what changed? 2010-09-13T08:22:13 actually i'll probably just remove them, atleast for now 2010-09-13T08:22:54 AJC: address space is limited to 196Mb, but java needs a 700+Mb address space 2010-09-13T08:22:56 So I guess that is why I jumped in rankings :/ 2010-09-13T08:23:08 kuwabara: not counting any battles currently in progess, memory limits are now gone 2010-09-13T08:23:15 ok 2010-09-13T08:23:20 dstufft: thanks 2010-09-13T08:23:25 if java bots start winning again we'll know that was the issue 2010-09-13T08:23:48 sorry about that, I don't deal with java at all and didn't realize it needed 700+mb address space 2010-09-13T08:23:49 kuwabara: address space not being the same as actual memory usage? I don't understand Java... that's weird. 2010-09-13T08:24:08 This was a close one :D http://www.ai-contest.com/visualizer.php?game_id=4452689 2010-09-13T08:24:21 ooh, top spot, hope deepblue doesn't wake up 2010-09-13T08:24:41 AJC: when you map a shared library, it enters your address space, but it doesn't "spend" more memory if the library was already mapped by another process 2010-09-13T08:24:54 kuwa: untrue 2010-09-13T08:25:02 kuwa: relocations happen, alot 2010-09-13T08:25:07 also on linux 2010-09-13T08:25:11 bartwe_: are you dmj111? 2010-09-13T08:25:39 no 2010-09-13T08:25:56 *** Maxicube has joined #aichallenge 2010-09-13T08:26:03 ah, ok... you meant top of tcp server? 2010-09-13T08:26:10 Hey all 2010-09-13T08:26:11 yep, sorry about that 2010-09-13T08:26:12 *** ermau has joined #aichallenge 2010-09-13T08:26:14 *** amstan has quit IRC (Ping timeout: 255 seconds) 2010-09-13T08:26:19 *** ermau is now known as ermau|work 2010-09-13T08:26:23 should be able to top the main contest with that too 2010-09-13T08:26:35 bartwe_: so what's your solution? :-) 2010-09-13T08:26:36 :) 2010-09-13T08:26:37 Sup Maxicube 2010-09-13T08:27:02 Nm, friend reeled me in to make a bot :) 2010-09-13T08:27:11 AJC: a good one 2010-09-13T08:27:13 Your friend would be me 2010-09-13T08:27:23 dstufft: It worked: http://www.ai-contest.com/profile.php?user_id=6391 2010-09-13T08:27:25 is it normal that my bot haven't played a single game since yesterday? 2010-09-13T08:27:27 Oh lol.. 2010-09-13T08:27:33 Fail at me. 2010-09-13T08:27:34 nah, its mostly simple heuristics, while trying to avoid magic numbers 2010-09-13T08:27:34 :p 2010-09-13T08:27:42 bartwe_: no searching? 2010-09-13T08:27:47 Hey fish 2010-09-13T08:27:49 Lol 2010-09-13T08:27:53 AJC: what is searching ? 2010-09-13T08:27:54 *** Zaphus has joined #aichallenge 2010-09-13T08:28:14 bartwe_: going through lots of options to pick the best. 2010-09-13T08:28:15 i prepare graphs, collect options, sort and filter 2010-09-13T08:28:36 but its definitly not trying to be exhaustive or even very predictive 2010-09-13T08:28:43 Baphomet: My bot I made at school rapes the default bots 2010-09-13T08:28:44 ive actually started to work out how some of the code works today, just that my strats still suck and i loose anyway 2010-09-13T08:28:46 bartwe_: what goes into your graphs? 2010-09-13T08:29:00 no way to use more time or give an answer before a deadline 2010-09-13T08:29:03 nice, send me the exe so i can play agenst it? 2010-09-13T08:29:05 things needed in the long term i think 2010-09-13T08:29:20 i totally wont decompile it or anything... 2010-09-13T08:29:29 *** Utkarsh has quit IRC (Ping timeout: 272 seconds) 2010-09-13T08:29:36 i actually wont lol, cbf doing that 2010-09-13T08:29:42 Baphomet: Would love to. Cept I'm not on the internets. 2010-09-13T08:29:45 acj: distances, population projections, pathfinding 2010-09-13T08:29:51 kuwabara: does the java bots have 700mb+ rss? 2010-09-13T08:29:52 *** Zeiris has quit IRC (Ping timeout: 265 seconds) 2010-09-13T08:29:59 Maxicube: i know how you feel, im not on the net either 2010-09-13T08:30:08 *** Baphomet has quit IRC () 2010-09-13T08:30:11 how are you on irc then ? 2010-09-13T08:30:15 Baphomet: IPhone ftw 2010-09-13T08:30:17 bartwe_: you need pathfinding? not just a direct flight? 2010-09-13T08:30:24 bartwe_: ^^ 2010-09-13T08:30:39 dstufft: no, RSS is much smaller, probably less than 30Mb. 2010-09-13T08:30:41 AJC: geh 2010-09-13T08:30:45 *** Baphomet has joined #aichallenge 2010-09-13T08:30:52 actually i checked and i was. 2010-09-13T08:30:53 AJC: direct flight takes a long time 2010-09-13T08:31:07 cant be too sure these days 2010-09-13T08:31:08 indirect flight takes a longer time 2010-09-13T08:31:08 kuwabara: ok i'll likely switch the limits from as to rss then 2010-09-13T08:31:09 bartwe_: I know, you become less flexible. 2010-09-13T08:31:18 Lolk. I'm on my iPhone. I hate tethering. 2010-09-13T08:31:29 they are mostly there because we were getting stuck bots every once in awhile 2010-09-13T08:31:34 bartwe_: I bet that's why you're top of the TCP ladder. 2010-09-13T08:31:38 Whats a reasonable time to wait for a new upload to show up in 'submissions' before trying again ? the webpage says 10 minutes, is that OK 2010-09-13T08:31:41 that were chewing up ram and cpu while not playing any games 2010-09-13T08:31:52 AJC: that and some of the stronger bots are offline ;P 2010-09-13T08:31:59 Mainly coz I'm on Linux. 2010-09-13T08:32:13 bartwe_: do the stronger bots also use some kind of max flight time minimization? 2010-09-13T08:32:26 ajc: you would have to ask them 2010-09-13T08:32:33 Sif Linux is cool 2010-09-13T08:32:40 But BDS is better! 2010-09-13T08:32:41 bartwe_: thanks 2010-09-13T08:32:44 BSD* 2010-09-13T08:32:59 contestbot time 2010-09-13T08:32:59 Zaphus: 08:32 AM, September 13, 2010 2010-09-13T08:33:11 It is yes :). I use Linux mint on my laptop. It loves my iPhone. *cough* 2010-09-13T08:33:13 Everyone knows one day desktop-bsd will be the most popular os 2010-09-13T08:33:17 *** Utkarsh has joined #aichallenge 2010-09-13T08:33:39 hmz, haven't used bsd since befire 2k 2010-09-13T08:33:49 dstufft: should I reupload my bot so that its losses are removed ? It won all of its games except for that short period were memory was limited. 2010-09-13T08:33:51 Coz it's free with it's bsdness. 2010-09-13T08:34:08 kuwabara: you're free to 2010-09-13T08:34:51 dstufft: I mean: are all games taken into account for the rankings, on only the last ones (in which case, the losses will disapear with time) 2010-09-13T08:35:36 all games 2010-09-13T08:35:36 kuwabara: I don't know the ranking algo that well, but i think it will take the losses into account until you upload them 2010-09-13T08:35:47 kuwabara: all previous game records disappear upon re-upload afair.. but submission has been absent the past 2 hours 2010-09-13T08:35:50 Phew. That takes some effort to plug that in. 2010-09-13T08:35:53 because your rating is changed verytime you win /lose 2010-09-13T08:35:55 dstufft: ok, thanks. 2010-09-13T08:35:58 *afair=afaik 2010-09-13T08:36:09 and the current position is based offf your rating 2010-09-13T08:37:12 Wonder which language will win? 2010-09-13T08:37:21 ratatata: is anyone looking at why submissions are failing ? or are all the admins asleep (literally) ? 2010-09-13T08:37:53 Probbly the most used one i recon 2010-09-13T08:38:14 actually, looks like python is doing pretty well 2010-09-13T08:38:20 zaphus: no idea 2010-09-13T08:38:26 i'm not asleep, im just tired 2010-09-13T08:38:36 i didn't sleep last night 2010-09-13T08:38:36 lol 2010-09-13T08:39:00 dstufft: are submissions really failing ? my last two attempts have not shown up (over the last 25 minutes) 2010-09-13T08:39:09 that amd i reallllly hate php 2010-09-13T08:39:21 Zaphus: let me see if i can figure out 2010-09-13T08:39:56 *** sinan has joined #aichallenge 2010-09-13T08:41:29 first i have to go see how submission.php works lol 2010-09-13T08:41:55 *** Maxicube_ has joined #aichallenge 2010-09-13T08:44:01 *** sigh_ has joined #aichallenge 2010-09-13T08:44:33 Zaphus: you worry about 25 minutes? I have submitted my bot yesterday 2010-09-13T08:45:20 *** Maxicube has quit IRC (Ping timeout: 240 seconds) 2010-09-13T08:45:24 *** Maxicube_ is now known as Maxicube 2010-09-13T08:45:44 Zaphus: do you know your userid by any chance 2010-09-13T08:45:56 or what's your username on the site 2010-09-13T08:46:03 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-09-13T08:46:42 *** sigh has joined #aichallenge 2010-09-13T08:47:19 dstufft: sorry, was AFK - hang on 2010-09-13T08:47:21 2010-09-13 07:29:49 2010-09-13T08:47:29 that was when the last submission was recorded from you 2010-09-13T08:47:49 ok, but that doesnt appear on the webstite 2010-09-13T08:48:05 looking 2010-09-13T08:48:16 it's status is "ready to be unzipped and compiled" 2010-09-13T08:48:17 nothing on the 13th on the submissions section: http://www.ai-contest.com/profile.php?user_id=3894 2010-09-13T08:48:22 so 2010-09-13T08:48:24 lemme see 2010-09-13T08:48:33 ah, ok, may be in a queue ? 2010-09-13T08:48:36 dstufft: can you check mine please? 4008 evgenius 2010-09-13T08:49:49 same for me, latest submission doesn't appear, but he desn't need to investigate all cases. Once the problem is identified for Zaphus, he can fix it for everybody. 2010-09-13T08:50:04 unless I'm special :-) 2010-09-13T08:50:51 *** sigh_ has quit IRC (Ping timeout: 272 seconds) 2010-09-13T08:51:12 who runs the benzedrine server ? 2010-09-13T08:51:25 oh shit 2010-09-13T08:51:29 let me guess. benzedrine? 2010-09-13T08:51:30 lol 2010-09-13T08:51:31 cron's default is emacs 2010-09-13T08:51:34 how do i exit this 2010-09-13T08:51:40 * dstufft is a vim user 2010-09-13T08:51:49 Ctrl-x Ctrl-c 2010-09-13T08:52:03 http://s.lurkmore.ru/images/4/43/Cartoon284.png 2010-09-13T08:52:21 *** tobiassjosten has joined #aichallenge 2010-09-13T08:53:36 I can't get the TCP-connection working. It's compiled and I can run it but I'm not sure how to use it with my Python bot. Any pointers? 2010-09-13T08:53:45 ok 2010-09-13T08:53:49 found the problem 2010-09-13T08:53:52 time to fix it 2010-09-13T08:53:59 I've made it executable and am trying with: ./tcp 213.3.30.106 9999 username -p password ./MyBot.py 2010-09-13T08:54:23 But that wont work. ": No such file or directory" and "command terminated with exit status 127". 2010-09-13T08:54:35 tobiassjosten: you don't need to type your pass 2010-09-13T08:55:51 EvGenius: Ah, thanks. Though I'm still unable to run my bot. :( 2010-09-13T08:56:36 and I think instead of ./MyBot.py you should type the full command line with the interpreter 2010-09-13T08:56:42 tobiassjosten: does MyBot.py have "#! /usr/bin/python" in the first line? 2010-09-13T08:57:16 *** mega1 has joined #aichallenge 2010-09-13T08:57:58 RainCT: Yeah. /usr/bin/env though. 2010-09-13T08:58:45 tobiassjosten: Dunno then. I ended up hardcoding the interpreter & file name in tcp.c 2010-09-13T08:59:26 RainCT: Ah. Care to share that? :) 2010-09-13T09:00:10 ok 2010-09-13T09:00:14 sort of fixed it 2010-09-13T09:00:27 it's processing the backlog now 2010-09-13T09:00:32 theres like 900 things to compile 2010-09-13T09:00:33 lol 2010-09-13T09:00:50 900 for 700 users ? 2010-09-13T09:00:54 lots of people multiple times 2010-09-13T09:01:05 maybe you could remove the duplicates ? 2010-09-13T09:01:08 yeah just for trying 2010-09-13T09:01:14 the webpage says resubmit in 10 minutes if it doesnt show - thats bad thing to suggest to people! 2010-09-13T09:01:22 dstufft: finally! thank you 2010-09-13T09:01:25 Zaphus: hehe, good point. 2010-09-13T09:02:29 but hey, if it's 1 minute per compile, thats only... 15 hours :-) hopefully it's only a few seconds each compile 2010-09-13T09:02:41 *** ermau|work has quit IRC (Ping timeout: 255 seconds) 2010-09-13T09:02:44 yea it does a bunch per minute 2010-09-13T09:02:50 *** ermau has joined #aichallenge 2010-09-13T09:03:00 tobiassjosten: I don't have it here, and it looks like the tcp.c file on the web has been changed. But it was something like execv(argv[0], args) where args is an array with {"python", "MyBot.py", NULL} (in my case also "--log", "output.log" before the NULL) 2010-09-13T09:03:31 *** ermau is now known as ermau|work 2010-09-13T09:03:48 tobiassjosten: to run on the TCP server, make a wrapper script calling "python path-to/MyBoy.py"... then call it as ./tcp ./ScriptName 2010-09-13T09:05:10 RainCT, ratatata: Sounds like two good suggestions, I'll try them. Thanks! 2010-09-13T09:05:15 hmm, nice http://github.com/albertz/planet_wars-cpp :) 2010-09-13T09:05:23 MyBoy.py, I guess you really cherish your bot :) 2010-09-13T09:05:34 from here http://ai-contest.com/forum/viewtopic.php?f=18&t=550&sid=17966158bb3f2e80ad3a9c01a91117b5 2010-09-13T09:05:50 *** otherAntimatroid has left #aichallenge 2010-09-13T09:05:54 EvGenius: absolutely.. this weekend, we're going fishing 2010-09-13T09:07:17 *** mightybyte has joined #aichallenge 2010-09-13T09:08:03 *** delt0r has joined #aichallenge 2010-09-13T09:08:26 850 items in the backlog 2010-09-13T09:08:32 *** sinan has quit IRC (Ping timeout: 265 seconds) 2010-09-13T09:08:46 only 50 done since then? ouch 2010-09-13T09:09:02 50 in 8 minutes... so 900 in 2.5 hours... 2010-09-13T09:09:02 ratatata: nah, that's pretty good going. 2010-09-13T09:10:17 dstufft: will it pick up submissions from yesterday? 2010-09-13T09:10:23 Ported_: yea 2010-09-13T09:10:24 AJC: with no pileup from hours before, I'd agree 2010-09-13T09:10:27 it's going through them now 2010-09-13T09:10:46 dstufft: i take it it is doing the most recent submissions first? 2010-09-13T09:10:56 it does them in order of submission 2010-09-13T09:10:59 ah 2010-09-13T09:11:00 well 2010-09-13T09:11:02 ok thanks 2010-09-13T09:11:02 i assume it does 2010-09-13T09:11:07 makes the most sense that way 2010-09-13T09:11:36 *** vikhyat has quit IRC (Read error: Connection reset by peer) 2010-09-13T09:12:03 ratatata: it's done more then 50, people are still submiting bots as well 2010-09-13T09:14:21 g'night all, I need to sleep while my bot does all the work :-) 2010-09-13T09:14:29 *** amstan has joined #aichallenge 2010-09-13T09:14:33 *** ChanServ sets mode: +o amstan 2010-09-13T09:14:36 yo 2010-09-13T09:14:36 *** GordonQuad has joined #aichallenge 2010-09-13T09:14:50 contestbot: seen j3camero 2010-09-13T09:14:50 amstan: j3camero was last seen in #aichallenge 16 hours, 27 minutes, and 36 seconds ago: jmcarthur: when the site is up... 2010-09-13T09:14:51 *** Zaphus has quit IRC (Quit: Page closed) 2010-09-13T09:14:55 :( 2010-09-13T09:15:08 amstan: don't suppose i can get a fancy red name on the forums :p 2010-09-13T09:15:18 dstufft: right.. you're an admiun 2010-09-13T09:15:40 link me your profile 2010-09-13T09:16:11 *** Utkarsh has quit IRC (Ping timeout: 255 seconds) 2010-09-13T09:16:24 http://www.ai-contest.com/forum/memberlist.php?mode=viewprofile&u=827 2010-09-13T09:16:49 dstufft: done 2010-09-13T09:17:25 amstan: thanks ! 2010-09-13T09:17:30 770 in the backlog now lol 2010-09-13T09:17:46 dstufft: yeah, we have to organize topics in the techsupport section 2010-09-13T09:17:56 *** sfsgagi has joined #aichallenge 2010-09-13T09:19:27 awww the c++ game engine doesnt work :( 2010-09-13T09:19:32 I thought games were being played again? 2010-09-13T09:20:58 ermau|work: they should be 2010-09-13T09:23:23 *** amstan has quit IRC (Ping timeout: 255 seconds) 2010-09-13T09:24:11 Java is broked last i checked too 2010-09-13T09:25:01 *** pacotabaco has quit IRC (Ping timeout: 252 seconds) 2010-09-13T09:25:03 delt0r: that should be fixed 2010-09-13T09:25:15 cool 2010-09-13T09:25:19 I will check again 2010-09-13T09:25:26 thanks dstufft 2010-09-13T09:26:30 delt0r: i implemented memory limits to prevent bots from running rampant which happened every once in awhile, and was one of the issues bringing the server down 2010-09-13T09:26:44 dstufft, Good idea 2010-09-13T09:26:46 except i don't use java, and didn't realize java had an AS of 700+mb 2010-09-13T09:27:02 What is the new limit 2010-09-13T09:27:06 so limiting AS to 196M was wrecking java bots 2010-09-13T09:27:12 I am under 64meg at the moment 2010-09-13T09:27:26 atm i just removed it (there is still a 10 minute cpu time limit though) 2010-09-13T09:27:27 delt0r: of RSS certainly, but not of AS 2010-09-13T09:27:39 i'll probably switch it to 196MB of RSS 2010-09-13T09:27:48 *** amstan has joined #aichallenge 2010-09-13T09:27:48 *** ChanServ sets mode: +o amstan 2010-09-13T09:28:10 *** vikhyat has joined #aichallenge 2010-09-13T09:28:16 Sorry --what do you mean by AS RSS 2010-09-13T09:28:31 AS is address space, the entire amount of possible ram for a process basically 2010-09-13T09:28:35 including any shared libraries 2010-09-13T09:28:54 cool, yesterday was a new record for the forums 2010-09-13T09:29:04 Java apparently links in 700+mb of shared libraries, that is shared among processes basically 2010-09-13T09:29:07 *** Maxicube has quit IRC (Quit: Colloquy for iPhone - http://colloquy.mobi) 2010-09-13T09:29:17 RSS is how much ram you're particular process alone is using 2010-09-13T09:29:20 your* 2010-09-13T09:29:39 *** GordonQuad has quit IRC (Ping timeout: 252 seconds) 2010-09-13T09:29:54 amstan, Right. Yes java is well known for using everything even when you need nothing 2010-09-13T09:30:03 pros and cons i guess 2010-09-13T09:30:10 hmm? 2010-09-13T09:30:18 oh.. yeah 2010-09-13T09:30:30 except you can't share such memory in a sandbox 2010-09-13T09:30:38 ahh 2010-09-13T09:30:53 So what about using java's own sandbox? 2010-09-13T09:30:55 that and the compile daemon was throwing an exception every time it tried to process a submission (got stuck on one particular submission over and over) 2010-09-13T09:31:08 fixed that, so we are compiling again 2010-09-13T09:31:15 but going through the backlog of submissions 2010-09-13T09:32:04 Is it really realistic to get through all that backlog? -- i mean the comp doesn't run much longer 2010-09-13T09:32:21 the comp runs for 3 months more 2010-09-13T09:32:30 and the backlog was 900 some a bit ago 2010-09-13T09:32:33 it's steadily decreasing 2010-09-13T09:32:38 oh... Perhaps i got all the dates wrong 2010-09-13T09:32:50 * delt0r feels sheepish 2010-09-13T09:32:59 200 items in the last 30 minutes 2010-09-13T09:33:02 about 2010-09-13T09:33:15 so shall i add my java sand box one ;) 2010-09-13T09:33:51 delt0r: we can't have a sandbox for each of our languages 2010-09-13T09:33:53 that would be insane 2010-09-13T09:34:10 but its just a command line argument 2010-09-13T09:34:25 then everything that jvm runs will be "sandboxed" 2010-09-13T09:34:27 then we have to unsandbox it from our normal sandbox.. 2010-09-13T09:34:38 good point 2010-09-13T09:35:21 not to mention that it'll ruin the decentralization thingy 2010-09-13T09:38:49 amstan, So are you using systrace for the sandboxing? if so is there still a java performance issue with that? 2010-09-13T09:40:28 yay! my submissions from yesterday are showing up on the site. Thanks! 2010-09-13T09:41:34 450 int he backlog 2010-09-13T09:41:37 afk a bit 2010-09-13T09:41:48 does anyone know what kind of hardware everything runs on? is this a problem at the moment? 2010-09-13T09:42:07 i mean the infrastructure for running matches, compiling code etc. 2010-09-13T09:42:39 Ported_, I think from how busy it is --its going to be a problem for a little while--the diversity of support langs makes it hard 2010-09-13T09:43:44 delt0r: is it a matter of throwing more hardware at the problem though? 2010-09-13T09:44:29 Manpower too i guess... its run by a university group.. its not really their "day" job... but we got time... 2010-09-13T09:44:29 *** retybok has quit IRC (Ping timeout: 264 seconds) 2010-09-13T09:44:41 *** vikhyat has quit IRC (Ping timeout: 272 seconds) 2010-09-13T09:44:43 I think things will stabilize.... 2010-09-13T09:45:09 *** mietek is now known as mbak 2010-09-13T09:45:11 *** mbak is now known as mietek 2010-09-13T09:46:08 Hm 2010-09-13T09:46:20 *** Appleman1234 has quit IRC (Quit: Leaving) 2010-09-13T09:46:31 *** ippeb has joined #aichallenge 2010-09-13T09:47:46 *** ippeb has left #aichallenge 2010-09-13T09:49:48 Ported: at this point hardware is not the problem 2010-09-13T09:50:01 Ported: Just needed to have the code optimized more 2010-09-13T09:50:18 I think we've got all the huge fires out 2010-09-13T09:52:16 *** zaph has joined #aichallenge 2010-09-13T09:52:58 danielvf, are you talking about optimizations to PlayGame.jar? Or optimizations to the portal? If the former is true, could the cpp port of PlayGame make it even faster? 2010-09-13T09:53:44 *** kateb has joined #aichallenge 2010-09-13T09:54:15 I was talking about the server 2010-09-13T09:54:22 Palmik: it's not playgame.. it's the general health of the server 2010-09-13T09:54:33 Palmik: one moment it's fine, the next it starts swapping badly 2010-09-13T09:54:39 Oh, I see 2010-09-13T09:54:40 It's doing great right now. 2010-09-13T09:54:47 amstan: We've fixed the swapping issue 2010-09-13T09:55:51 *** zaph has quit IRC (Client Quit) 2010-09-13T09:56:58 *** vikhyat has joined #aichallenge 2010-09-13T09:57:15 danielvf: yayt 2010-09-13T09:58:13 sounds like everything is under control then! 2010-09-13T10:00:30 vortex.9: you rule :) 2010-09-13T10:00:47 kicking my ass with an opertunistic neutral stealing strategy 2010-09-13T10:00:58 but two can play that game 2010-09-13T10:01:16 *** cfaftw has joined #aichallenge 2010-09-13T10:02:05 bartwe_: do you get the replays of your own games from the TCP server? 2010-09-13T10:02:18 http://www.benzedrine.cx/planetwars/canvas?game_id=1284385661|bartwe.5.1|vortex.9 2010-09-13T10:03:28 hmz i guess i could capture neutrals from my most remote planets, to see if the enemy is going to steal, but i'd still lose the units 2010-09-13T10:03:30 hmpz 2010-09-13T10:03:40 *** s_denchev has quit IRC (Quit: Page closed) 2010-09-13T10:03:51 it would allow for a smackdown on the recapture 2010-09-13T10:07:38 or a budget of units to spend on neutrals 2010-09-13T10:08:11 *** GordonQuad has joined #aichallenge 2010-09-13T10:09:01 *** kateb has quit IRC (Ping timeout: 252 seconds) 2010-09-13T10:09:24 *** rogue780_ has joined #aichallenge 2010-09-13T10:09:37 Howdy 2010-09-13T10:09:57 hey rogue780_ 2010-09-13T10:10:08 * amstan feels like making a bot, and using it in the tcp tourny 2010-09-13T10:10:43 amstan: try beating these two http://www.benzedrine.cx/planetwars/canvas?game_id=1284385976|vortex.9|bartwe.5.1 2010-09-13T10:11:08 bartwe_: which one of them? 2010-09-13T10:11:09 oh.. 2010-09-13T10:11:10 them 2 2010-09-13T10:11:12 ok 2010-09-13T10:11:20 i'm gonna start with neural networks 2010-09-13T10:11:26 really ? 2010-09-13T10:11:36 bartwe: are you developing your bot full time? 2010-09-13T10:11:48 16h/day ? 2010-09-13T10:11:58 only 16h? 2010-09-13T10:11:59 hehe 2010-09-13T10:12:06 i have never seen a neural net do any succefull gameplaying in a constest like this 2010-09-13T10:12:30 nak: nah maybe 12 h yesterday, nothing so far, just watching, thinking of strategies 2010-09-13T10:12:31 bartwe_: i think it could work as a function approximator for a heuristic or opponent model. 2010-09-13T10:13:01 ajc: i don't see how you can reduce the game state to usefull scalar inputs 2010-09-13T10:13:29 bartwe_: not the game, the value of planets or individual attacks. 2010-09-13T10:13:38 critical things like distances or growthrate 2010-09-13T10:13:38 the likelyhood of a planet being attacked based on its properties 2010-09-13T10:13:50 ah that would be fun 2010-09-13T10:14:09 ulope: which fork of yours is better? 2010-09-13T10:14:10 i was thinking of a bayesian approach to generate possible future attacks 2010-09-13T10:14:14 good news 2010-09-13T10:14:17 the backlog is empty 2010-09-13T10:14:28 great job, dstufft 2010-09-13T10:14:31 from now one your shit should get compiled relatively quickly 2010-09-13T10:15:18 well assuming it doesn't get stuck again 2010-09-13T10:15:22 dstufft: should bots that were in the backlog now be playing? 2010-09-13T10:15:31 I didnt actually fix the cause of the error, I just unjammed it 2010-09-13T10:15:36 Ported_: yea 2010-09-13T10:15:37 amstan: Well I use my fork obviously but I pull in stuff from other peoples if I like them 2010-09-13T10:16:16 ulope: cool, i'll use it as a starting point 2010-09-13T10:16:23 your features do sound good 2010-09-13T10:16:30 *** sinan has joined #aichallenge 2010-09-13T10:16:31 dstufft: I had some games listed on my profile yesterday but they dissappeared and now there is nothing: profile id 6019 2010-09-13T10:17:11 Ported_: did you upload a new submission? 2010-09-13T10:17:38 no, just the ones that were in the backlog, will not be able to upload till I get home tonight... 2010-09-13T10:17:40 amstan: Cool, have fun :) And if you find a bug please put it on the github issuetracker 2010-09-13T10:17:52 *** danielvf has quit IRC (Ping timeout: 252 seconds) 2010-09-13T10:17:52 *** iPanda has joined #aichallenge 2010-09-13T10:17:56 how do you guys test your bots locally? 2010-09-13T10:18:00 hello 2010-09-13T10:18:06 is the java one worth anything? 2010-09-13T10:18:09 Ported_: yea a new version being compiled wipes out your od games 2010-09-13T10:18:16 amstan: i have 2 python scripts for playing games 2010-09-13T10:18:25 dstufft: with gfx? 2010-09-13T10:18:29 dstufft: but should the new one have played a game yet? 2010-09-13T10:18:29 when should my bot start playng? 2010-09-13T10:18:40 amstan: it uses ShowGame.jar to output 2010-09-13T10:19:00 *** danielvf has joined #aichallenge 2010-09-13T10:19:02 iPanda: tourny engine is offline for the moment 2010-09-13T10:19:03 i think... 2010-09-13T10:19:10 http://gist.github.com/574932 2010-09-13T10:19:12 dstufft: ok i got a game now! 2010-09-13T10:19:13 no 2010-09-13T10:19:15 bot's should be playing 2010-09-13T10:19:20 everything should be working 2010-09-13T10:19:33 amstan: that's run.py, it's just used to fight 1 bot vs another 2010-09-13T10:19:49 amstan, you mean against example bots? 2010-09-13T10:19:57 dstufft: Why not use the visualize_locally.py from the canvas visualizer? 2010-09-13T10:20:10 ulope: because when i wrote that i didn't know about it 2010-09-13T10:20:12 ok... so.. what? 2010-09-13T10:20:19 dstufft: Ah ok :) 2010-09-13T10:20:20 g2g 2010-09-13T10:20:23 brb 2010-09-13T10:20:24 rankings 10 2010-09-13T10:20:39 @rankings 10 2010-09-13T10:20:40 .rankings 10 2010-09-13T10:20:40 dstufft: Top 10 players: deepblue(2787), Kavish(2745), cagedmantis(2724), sighbot(2708), dmj111(2703), a1k0n(2702), seigen(2655), wtfbbq(2647), braddunbar(2626), kirkeby(2614) 2010-09-13T10:20:54 please do note that as far as rankings go 2010-09-13T10:21:05 we just proccessed 900 some submissions 2010-09-13T10:21:12 in the last hour or two 2010-09-13T10:21:13 or 3 2010-09-13T10:21:23 so a lot of bots are at 0 or only a few games atm 2010-09-13T10:21:52 i tryed to uppload about 30 times my bot, when server has been off :-[ 2010-09-13T10:22:03 It seems whenever it try to iterate through either my fleets or the enemy's fleets, it causes my not to timeout. I'm using Java. 2010-09-13T10:22:04 ulope: run.py is really just a wrapper around that long ass java command 2010-09-13T10:22:21 *** chevap has quit IRC (Quit: Page closed) 2010-09-13T10:22:21 now they all in my profile O_O 2010-09-13T10:22:28 mostly wrote it because eff trying to remember that command 2010-09-13T10:22:47 iPanda: yea submissions were being added to the compile queue, but the process that compiled them was stuck 2010-09-13T10:23:05 should probably list on the use page when something is submitted but not yet compiled too 2010-09-13T10:23:07 I just made a bat file. 2010-09-13T10:23:09 dstufft: Yeah i have a shell script for that 2010-09-13T10:23:19 yea i thought about that 2010-09-13T10:23:24 but run.py is cross platform 2010-09-13T10:23:29 what does it mean "Warning: mysql_fetch_object(): supplied argument is not a valid MySQL result resource in /var/www/vhosts/ai-contest.com/www/profile_submissions.php on line 16" in my profile->View more 2010-09-13T10:23:30 true 2010-09-13T10:23:35 so i can work on either windows or linux 2010-09-13T10:23:40 http://ai-contest.com/profile_submissions.php?user_id=7010&page=1 2010-09-13T10:23:44 iPanda: it means the site is broken apparently 2010-09-13T10:24:06 oh it's having issues looking up your name 2010-09-13T10:24:23 the joys of php 2010-09-13T10:24:27 wtb a rewrite in Django 2010-09-13T10:24:35 Looking it up 2010-09-13T10:24:43 Hey, if anyone needs the client for windows for the tcp server, I have a compiled one all zipped up at WWW.errgaming.com 2010-09-13T10:25:10 *** amstan has quit IRC (Quit: Konversation terminated!) 2010-09-13T10:25:22 *** amstan has joined #aichallenge 2010-09-13T10:25:22 *** ChanServ sets mode: +o amstan 2010-09-13T10:25:28 cool 2010-09-13T10:26:05 iPanda: Fixed 2010-09-13T10:26:25 ok.. so.. what's the best way to run the bots? 2010-09-13T10:26:26 thanx 2010-09-13T10:26:33 i have nginx setup btw, on port 8080 2010-09-13T10:26:40 but apparently theres a hw fw or something 2010-09-13T10:26:40 dstufft: I saw. 2010-09-13T10:26:52 so gotta port foward via ssh atm 2010-09-13T10:26:55 dstufft: it's serving the site? 2010-09-13T10:27:00 danielvf: yup 2010-09-13T10:27:25 And the other sites on the server? 2010-09-13T10:27:35 btw happy 256th day in year (programmers day =) 2010-09-13T10:27:36 don't have them setup in the conf atm 2010-09-13T10:27:42 easy to add though 2010-09-13T10:27:44 iPanda: lol 2010-09-13T10:28:01 dstufft: Would you do that - I'd want it to be a drop in replacement 2010-09-13T10:28:19 I like nginex, just gotta test it well before we use it. 2010-09-13T10:29:04 danielvf: yea, well now that the proof of concept works, i want to get go ahead to get it running via www-data instead of my personal user, but still bound to non standard port 2010-09-13T10:29:10 i was going to do that then 2010-09-13T10:29:15 Okay 2010-09-13T10:29:21 + i didn't have access to my vps yesterday 2010-09-13T10:29:28 which has my nginx configs on it 2010-09-13T10:29:39 and i didn't feel like looking up all the directives 2010-09-13T10:29:58 dstufft: Heh. I understand 2010-09-13T10:30:02 a nice ui touch for the site would be to generate a queue of scheduled games, and display it. Then people could see that they're at least in line to play, and see how the games are progressing. 2010-09-13T10:30:05 amstan: most people seem to use some sort of wrapper around the PlayGame.jar call 2010-09-13T10:30:22 rwa: Agreed 2010-09-13T10:30:40 rwa: atm that would require a rework of the tournament_manager 2010-09-13T10:30:43 i agree though 2010-09-13T10:31:35 theres a lot of rube goldberg style things going on atm 2010-09-13T10:31:49 amstan: Want to pin this one: http://ai-contest.com/forums/viewtopic.php?f=18&t=567 2010-09-13T10:32:23 You could probably unpin this one Round 2... Fight! 2010-09-13T10:33:57 "Unless you are a horrible programmer it does not mean your player actually timed out, it means that the player did something illegal." 2010-09-13T10:33:59 wrong! 2010-09-13T10:36:32 D: CD D:\ms\csharp_starter_package\csharp_starter_package\ java.exe -jar tools\PlayGame.jar maps\map33.txt 1000 1000 log.txt "iPandaBot.exe" "java -jar example_bots/ProspectorBot.jar" | java.exe -jar tools/ShowGame.jar 2010-09-13T10:37:14 *** sigh has quit IRC (Remote host closed the connection) 2010-09-13T10:37:21 btw, is log.txt hardcoded? 2010-09-13T10:37:32 changing that parameter to /dev/null still creates a log.txt file 2010-09-13T10:37:41 i think it's hardcoded for ShowGame.har 2010-09-13T10:37:46 ah 2010-09-13T10:37:47 *** mark_jn has joined #aichallenge 2010-09-13T10:37:56 .har 2010-09-13T10:38:00 har har 2010-09-13T10:38:02 the pirate programming language 2010-09-13T10:38:13 coffee pirates. 2010-09-13T10:38:15 imo we need a LOLCODE starter kit 2010-09-13T10:38:25 i want to write my AI bot completly in LOLCODE 2010-09-13T10:38:35 not that i can compete 2010-09-13T10:38:36 D: 2010-09-13T10:38:38 Do it in whitespace 2010-09-13T10:38:43 I HAS A fleet 2010-09-13T10:39:39 sounds like a project for tonight 2010-09-13T10:39:44 *** Nuriaion has quit IRC (Quit: Nuriaion) 2010-09-13T10:39:56 *** rogue780_ has quit IRC (Ping timeout: 276 seconds) 2010-09-13T10:40:51 all the best bots are written in LOLCODE 2010-09-13T10:40:57 hope this helps your choice of languages 2010-09-13T10:40:57 *** mega1_ has joined #aichallenge 2010-09-13T10:42:19 *** mega1 has quit IRC (Ping timeout: 272 seconds) 2010-09-13T10:42:50 How do you create an account on benzedrine.cx/planetwars? It's requesting a password but I can't find any information about how to get that. 2010-09-13T10:43:52 eh? password? 2010-09-13T10:43:53 how exactly does it require password? 2010-09-13T10:44:19 tobiassjosten: you don't need a password or even create an account 2010-09-13T10:44:25 "connected to 213.3.30.106:9999, waiting for game" followed by "Password required" 2010-09-13T10:44:35 O_o 2010-09-13T10:44:46 *** deepblue has joined #aichallenge 2010-09-13T10:44:47 And when I tried adding -p password, I got "Password rejected" 2010-09-13T10:45:02 top10 2010-09-13T10:45:09 tobiassjosten: you do use http://www.benzedrine.cx/planetwars/tcp.c right? 2010-09-13T10:45:13 hm 2010-09-13T10:45:15 oh no 2010-09-13T10:45:18 caytchen: Yes. 2010-09-13T10:45:20 I think you have another program named tcp in your system 2010-09-13T10:45:23 oh lol 2010-09-13T10:45:24 the compile daemon got stuck again 2010-09-13T10:45:29 an option was added for a password.. 2010-09-13T10:45:39 man i hate silent updates -_- 2010-09-13T10:45:51 tobiassjosten: try ./tcp 2010-09-13T10:45:54 Nevermind, I switched to another username (other than "username") and it seems to be working. 2010-09-13T10:46:10 yeah probably someone that used that specific name with a password 2010-09-13T10:46:39 Gotcha. 2010-09-13T10:47:16 *** rogue780_ has joined #aichallenge 2010-09-13T10:47:42 cool, got my first game in a couple days 2010-09-13T10:47:59 @rankings 2010-09-13T10:48:06 deepblue: Top 10 players: deepblue(2766), Kavish(2748), dmj111(2699), a1k0n(2686), sighbot(2685), shlewis(2664), bocharov(2645), wtfbbq(2638), seigen(2635), braddunbar(2618) 2010-09-13T10:48:15 teheee 2010-09-13T10:48:39 off to a good start deepblue 2010-09-13T10:48:52 we'll see 2010-09-13T10:49:01 now off to work :) 2010-09-13T10:49:04 later guys 2010-09-13T10:49:11 later 2010-09-13T10:49:11 *** deepblue has quit IRC (Client Quit) 2010-09-13T10:49:18 i should upload .5 to the main contest, then 'pheer' 2010-09-13T10:49:46 lol 2010-09-13T10:50:11 about passwords on benzedrine.cx, dhartmei added pw support to tcp.c; it's optional 2010-09-13T10:50:54 good to know 2010-09-13T10:51:15 bartwe_: what exactly do you use to route fleets through your planets? just some A* with a special crafted cost function? 2010-09-13T10:51:28 My game wasn't saved. Are the lists cached or something, or do I need to do something to have my games saved for playback? 2010-09-13T10:52:09 tobiassjosten: give it a few minutes to show up 2010-09-13T10:52:13 dstufft, you should write your bot in brainf@!k 2010-09-13T10:52:35 I want to see a postscript bot 2010-09-13T10:52:52 *** Utkarsh has joined #aichallenge 2010-09-13T10:53:01 tobiassjosten: which server 2010-09-13T10:53:29 dstufft: 213.3.30.106, port 9999. 2010-09-13T10:53:35 Didn't even know there were others. :) 2010-09-13T10:53:45 i meant the main contest one or the tcp server 2010-09-13T10:53:56 i think the tcp server clears out games every 8 hours or something 2010-09-13T10:54:07 or clears out games > 8 hours ago once an hour 2010-09-13T10:55:04 ulope: do you use windows or something? 2010-09-13T10:55:14 any reason you don't have a #!? 2010-09-13T10:55:58 ugh 2010-09-13T10:56:01 compile Daemon 2010-09-13T10:56:05 why do keep getting stuck 2010-09-13T10:57:53 *** shawnshekari has joined #aichallenge 2010-09-13T10:58:28 dstufft: 6 hours 2010-09-13T10:58:37 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-09-13T11:00:53 dstufft: did something change about the ranking ? Now that my new entry has been entered into the contest, I got 2 games played, but no ranking. 2010-09-13T11:02:01 (it would make sense to have a minimum emtry count, but it wasn't the case before, so I wonder if it's another bug) 2010-09-13T11:02:13 Woah I won against a bartwe bot.. too bad it's an old version 2010-09-13T11:03:42 *** mewo2 has joined #aichallenge 2010-09-13T11:04:13 there fixed compile_daemon correctly this time 2010-09-13T11:04:27 kuwabara: not that i'm aware of, but the rankings are not updating immediatly 2010-09-13T11:04:37 *** cobracom has joined #aichallenge 2010-09-13T11:04:55 they are updated every 7 minutes though 2010-09-13T11:05:10 kuwabara: can you link your profile? 2010-09-13T11:05:25 amstan: No Mac OS. Why? 2010-09-13T11:05:57 ulope: because i want to do ./game.py and it doesn't work.. 2010-09-13T11:06:43 dstufft: it solved itself. It was the 7 minute thing (the 2 games were only separated by 2 minutes). 2010-09-13T11:06:45 amstan: You don't diretly start game.py. You instead start your bot. have a look at one of the example bots 2010-09-13T11:07:40 morning all 2010-09-13T11:09:11 evening, Cyndre 2010-09-13T11:10:11 *** EvGenius has left #aichallenge 2010-09-13T11:11:18 *** tobiassjosten has quit IRC (Read error: Operation timed out) 2010-09-13T11:12:50 mornin Cyndre 2010-09-13T11:13:16 well hopefully now that i modified compile_anything.py, the compile daemon will no longer get stuck 2010-09-13T11:13:56 dstufft: the whole web server is down now 2010-09-13T11:14:23 *** rogue780_ has quit IRC (Ping timeout: 276 seconds) 2010-09-13T11:14:25 back again 2010-09-13T11:14:54 (sorry about this) 2010-09-13T11:15:50 You LOSE against rb_simplebot .. well that's encouraging 2010-09-13T11:15:57 lol 2010-09-13T11:16:21 actually, tutorial bot was stronger than my latest version 2010-09-13T11:16:35 * ermau|work had no problem defeating all of the sample bots 2010-09-13T11:17:03 I already know this bot has most of it's calculations wrong so I'm not really surprised 2010-09-13T11:19:18 ermau|work: Yeah I have that too, I find it surprising how "good" a bot with borked calculations can still play 2010-09-13T11:19:42 ulope: ok.. so... which one is your main file? MyBot.py 2010-09-13T11:19:55 ulope: Well I'm losing most games on the tcp server 2010-09-13T11:20:05 ulope: But was winning basically every game on the real server 2010-09-13T11:20:48 ulope: Most of the issues surround defending planets that will change owners before reinforcements arrive 2010-09-13T11:21:03 amstan: Yes. Do you see Game() at the end of the file? Thats what actually starts the game 2010-09-13T11:21:31 ermau|work: Or not taking the flight time into account correctly 2010-09-13T11:21:56 How can you screw that up? Isn't it just GrowthRate * (Distance + 1) ? 2010-09-13T11:22:12 ? 2010-09-13T11:22:26 1+distance*growthrate 2010-09-13T11:22:26 how do you decide which planet is worth taking? 2010-09-13T11:22:37 edcba: Same thing 2010-09-13T11:22:41 distance to all my other planets + _growth rate_ 2010-09-13T11:22:45 turn*growthrate > numships 2010-09-13T11:22:51 oh and ships of course but thats not too important 2010-09-13T11:23:09 ermau|work: No thats not what i meant. But when sending a fleet from multiple sources each has to have its own adjusted number of ships, but then some don't have enough to actually usefully contribute to the flett.. 2010-09-13T11:23:11 ulope: what file? 2010-09-13T11:23:15 you guys dont rank the individual planets and make sure you taking the next big one? 2010-09-13T11:23:24 ulope: Ah, yeah, I do adjust for that 2010-09-13T11:23:26 bss03-AFK: haha, i have exactly the same line count that you reported last night 2010-09-13T11:23:31 amstan: Look at stupidbot.py for example 2010-09-13T11:23:35 next best one 2010-09-13T11:23:51 ulope: so... if i understand this correctly, the root folder has example bots, the planetwars folder is a module to include that includes the api for the bot 2010-09-13T11:24:01 bss03-AFK: i think the code you pasted should be fast enough, but the only way to be sure is to benchmark 2010-09-13T11:24:02 amstan: Right. 2010-09-13T11:24:07 ulope: and you have no tournament running code 2010-09-13T11:24:08 *** aerique has quit IRC (Quit: ...) 2010-09-13T11:24:13 ulope: for testing 2010-09-13T11:24:49 No, i use the modified PlayGame.jar from a1k0n 2010-09-13T11:25:24 Cyndre: My algorithm for choosing planets is the biggest flaw in my current bot, it's not thought out and basically nothing 2010-09-13T11:25:31 Maybe I should spell that out in the Readme... 2010-09-13T11:25:33 Cyndre: But.. rewrite in progress so :P 2010-09-13T11:26:38 *** vikhyat has quit IRC (Ping timeout: 265 seconds) 2010-09-13T11:26:48 Cyndre: Well determining a global planet rank isn't that simple imho. The distance to the possible attack source planets is an important factor in my opinion 2010-09-13T11:28:36 i guess i just like the cancer model.. progress in your neighborhood ;p 2010-09-13T11:29:11 caytchen: It seems that is the strategy of most of the successful bots at the moment 2010-09-13T11:31:35 *** jmreardon has quit IRC (Ping timeout: 252 seconds) 2010-09-13T11:34:09 hey ulope :) 2010-09-13T11:34:17 hi 2010-09-13T11:35:38 ermau|work, mine currently pcisk the closest planet 2010-09-13T11:36:05 My weighs defenses against growth rate 2010-09-13T11:36:08 *** delt0r_ has joined #aichallenge 2010-09-13T11:36:35 ermau|work: mine doesn't even have an algorithm for choosing planets (yet) :D 2010-09-13T11:37:49 *** delt0r___ has quit IRC (Ping timeout: 252 seconds) 2010-09-13T11:38:00 Demonstration of why choosing targets is important: http://www.benzedrine.cx/planetwars/canvas?game_id=1284390961|rb_simplebot|ermau-1.6 2010-09-13T11:38:14 *** Obeleh has quit IRC (Read error: Connection reset by peer) 2010-09-13T11:38:59 Instead of fighting for the middle, I should of been retaking larger planets 2010-09-13T11:39:03 ermau|work: man, those bots need to learn to stop hating.. ;) 2010-09-13T11:39:06 expand, expand, expand 2010-09-13T11:39:31 caytchen: My attack and expand algo is the same 2010-09-13T11:39:37 it occurs to me now why attacking with streams might actually be a good thing sometimes 2010-09-13T11:39:38 *** Obeleh has joined #aichallenge 2010-09-13T11:39:42 caytchen: (Dumb, yes) 2010-09-13T11:39:46 jmcarthur: Yeah, I figured it out last night 2010-09-13T11:39:58 *** vikhyat has joined #aichallenge 2010-09-13T11:40:00 it lays pressure on the target sooner, so your opponent can't move ships from there for as long 2010-09-13T11:40:21 but it's a tradeoff. you are also subjected to similar constraints by doing it 2010-09-13T11:40:35 You also can reprioritize without having already spent all your forces 2010-09-13T11:40:43 no, that's not it 2010-09-13T11:40:56 if you weren't sending in streams in the first place, you would have had a lot more forces built up over time 2010-09-13T11:41:02 not doing anything 2010-09-13T11:41:08 so they would be even more available 2010-09-13T11:41:15 these streams we see are often just the result of turn based bots 2010-09-13T11:41:35 bots that basically completely reevaluate at each turn and then end up doing things twice 2010-09-13T11:41:47 *** xha has joined #aichallenge 2010-09-13T11:41:47 what's a point of a stream that is equal to growth rate? 2010-09-13T11:41:58 not a very good reevaluation if it doesn't see the fleet it already sent 2010-09-13T11:42:01 it's not a problem of stream 2010-09-13T11:42:07 it's a problem of speed 2010-09-13T11:42:23 and of limited production 2010-09-13T11:42:48 yeah streams equal to growth rate just means they *really* want to attack a planet but don't yet have the forces to do it 2010-09-13T11:42:59 caytchen: either that or bots that empty the planets 2010-09-13T11:43:11 and if the planet is growing as fast or faster than yours then you will never catch up, of course 2010-09-13T11:43:15 *** rogue780_ has joined #aichallenge 2010-09-13T11:43:46 jmcarthur: if you 'really' want to attack a planet but dont have the forces to take over in the predictable future, that is just wasted ships 2010-09-13T11:44:02 better spent on a expansion than harass 2010-09-13T11:44:03 Not really, you're negating ships 2010-09-13T11:44:04 ratatata: i agree 2010-09-13T11:44:14 ermau|work: yeah, but that wasn't the goal 2010-09-13T11:44:35 expansion will usually trump mere pressure 2010-09-13T11:44:42 pressure is for endgame when all planets are taken 2010-09-13T11:44:42 does the planet distance = number of turns for a fleet to get from planet a to planet b? 2010-09-13T11:44:46 *late game 2010-09-13T11:44:55 rogue780_: yup 2010-09-13T11:44:57 rogue780: planet distance rounded up 2010-09-13T11:45:19 rogue780: it's in the game spec (which is now linked from the front page) 2010-09-13T11:45:29 jmcarthur, rounded up? distance() returns and int 2010-09-13T11:45:32 ceil(euclidean_dist(a,b)) 2010-09-13T11:45:36 uses ceil() 2010-09-13T11:45:37 s/and/an 2010-09-13T11:45:40 *** cfaftw has quit IRC (Quit: Leaving) 2010-09-13T11:45:49 rogue780: if there is a distance function in your starter pack it's probably already rounding for you 2010-09-13T11:46:05 perfect 2010-09-13T11:47:26 caytchen: You wanted expansion? http://www.benzedrine.cx/planetwars/canvas?game_id=1284392183|ermau-1.6|rb_simplebot 2010-09-13T11:48:11 is that a full clone of the website? 2010-09-13T11:48:14 thats what i'm talking about heh 2010-09-13T11:48:29 *** hellman has joined #aichallenge 2010-09-13T11:48:50 ermau|work: some bad picks though 2010-09-13T11:48:52 pgpaskar_: benzedrine.cx/planetwars? 2010-09-13T11:48:56 yes 2010-09-13T11:49:01 id go to it but in a lecture atm :P 2010-09-13T11:49:01 if anybody needs a precompile tcp client for windows, I've made one available at www.errgaming.com 2010-09-13T11:49:02 caytchen: It's a bad bot :P 2010-09-13T11:49:03 pgpaskar_: it's a different thing entirely 2010-09-13T11:49:07 oh ok. 2010-09-13T11:49:08 *** iPanda has quit IRC (Ping timeout: 252 seconds) 2010-09-13T11:49:12 pgpaskar_: check it out though. great for testing 2010-09-13T11:49:13 you wasted ~45 ships at a cluster very close to him 2010-09-13T11:49:15 caytchen: The "picking" is awful. 2010-09-13T11:49:35 caytchen: Yeah, I know, it's basically "can I take this right now" 2010-09-13T11:49:38 and thats it 2010-09-13T11:49:39 *** Strepto has joined #aichallenge 2010-09-13T11:49:58 *** Kastra has joined #aichallenge 2010-09-13T11:49:59 heh thats what mine does too.. it doesn't even track fleets because i'm too damn lazy 2010-09-13T11:50:23 thats also why it easily produces 20+ fleets in a few turns lol 2010-09-13T11:50:41 too bad the tcp server prunes old games or i could show you some of the more crazy games 2010-09-13T11:50:43 I've noticed a lot of the bots constantly send out small fleets of 2-4 ships 2010-09-13T11:50:51 Is there a theory behind that? 2010-09-13T11:50:56 caytchen: I do track fleets, just not well enough yet 2010-09-13T11:51:06 caytchen: Working on a new one from scratch 2010-09-13T11:51:31 same.. but i might need a new task approach, right now i'm just pulsing sub modules. 2010-09-13T11:51:37 maybe they send their growthrated ships 2010-09-13T11:52:06 *** vikhyat has quit IRC (Ping timeout: 276 seconds) 2010-09-13T11:52:32 i mean they send some amount and leave 10 ships, next turn they have 13, and they use the same algorithm to calculate a count of ships to send, and get again 3.. 2010-09-13T11:52:59 seems like some of them are wasted on the same planet when they could be colonizing others 2010-09-13T11:53:11 then again i play like a human so I'm not sure how that strategy fits in to bot world :} 2010-09-13T11:53:36 i think sometimes blocking one planet by another is useful 2010-09-13T11:53:42 but dunno how to apply that :) 2010-09-13T11:54:03 *** Obeleh has quit IRC (Quit: Bye) 2010-09-13T11:54:33 *** Obeleh has joined #aichallenge 2010-09-13T11:54:39 it could be good to do while expanding 2010-09-13T11:54:53 it's a common tactic in starcraft games, for sure 2010-09-13T11:55:03 do you play a galcon game? 2010-09-13T11:55:16 *** phreeza has joined #aichallenge 2010-09-13T11:55:27 keep your opponent at bay so they can't attack you after attacking a neutral (while your forces are weak) 2010-09-13T11:55:29 http://www.benzedrine.cx/planetwars/canvas?game_id=1284378905|TracerBot10|wtfbbq this one just keeps sending its growth 2010-09-13T11:55:41 when i play starcraft i attack and expand at roughly the same time 2010-09-13T11:55:44 *** GordonQuad has quit IRC (Read error: Operation timed out) 2010-09-13T11:55:46 but im not sure if that applies here :] 2010-09-13T11:55:49 pgpaskar_: exactly 2010-09-13T11:55:53 i think it does apply 2010-09-13T11:56:08 * Naktibalda should play more 2010-09-13T11:56:11 if you have the ships to do it 2010-09-13T11:59:37 *** vikhyat has joined #aichallenge 2010-09-13T12:02:23 bawww my bot keeps crashing and i dont know why 2010-09-13T12:02:35 this is annoying 2010-09-13T12:02:50 yeah, debugging is unreal 2010-09-13T12:03:12 the emulator doesn't show any error messages :( 2010-09-13T12:03:21 mmm would be so helpfull if it did 2010-09-13T12:03:28 especially for an uber noob programmer like me 2010-09-13T12:03:38 i didnt even know what vectors were till today 2010-09-13T12:04:09 I just use php error logging :) 2010-09-13T12:04:25 just write to a file 2010-09-13T12:05:23 you could also try to capture the game states up to the point that you crash and arrange to send them manually to your bot one by one, separate from the engine, so you can see first hand what's happening 2010-09-13T12:05:27 in the worst case 2010-09-13T12:05:41 *** vikhyat has quit IRC (Ping timeout: 252 seconds) 2010-09-13T12:06:13 i do so :( it's too long 2010-09-13T12:06:50 i dont even know how to debug i just see when something went wrong and guess what change did it 2010-09-13T12:06:59 :( 2010-09-13T12:07:12 You LOSE against deepblue.. this should be entertainingly sad 2010-09-13T12:07:18 heh 2010-09-13T12:07:37 someone debug my code plz :3 hehehe 2010-09-13T12:08:06 Baphomet: it's better to ask in a channel of your programming language 2010-09-13T12:08:20 *** shawnshekari has quit IRC (Quit: Page closed) 2010-09-13T12:08:57 im using c++, but i dont think the guys in #c++ what have a clue whats going on here tho and how it all works 2010-09-13T12:09:21 *** TapewormGalactos has quit IRC (Ping timeout: 245 seconds) 2010-09-13T12:09:21 its something to do with the game thats making it crash, my code compiles ok 2010-09-13T12:09:26 there must be general debugging and logging techniques 2010-09-13T12:09:45 *** Obeleh has quit IRC (Quit: Bye) 2010-09-13T12:09:45 it's called runtime error 2010-09-13T12:10:57 oh, just run your bot stand alone in console and paste a map 2010-09-13T12:11:12 I like the dramatic effect around turn70 :P -> http://ai-contest.com/visualizer.php?game_id=4455045 2010-09-13T12:12:17 *** susu has joined #aichallenge 2010-09-13T12:12:30 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-09-13T12:12:32 paste a map? 2010-09-13T12:12:52 *** susu is now known as susu-magic 2010-09-13T12:12:56 yes, open map file, copy text, and paste it to your bot 2010-09-13T12:13:52 i cant paste inside a console window o.0 ? 2010-09-13T12:13:59 ratatata: i like how there is no turn 10 oO 2010-09-13T12:13:59 yes, you can 2010-09-13T12:14:09 > ./mybot 2010-09-13T12:14:19 im in windows.... 2010-09-13T12:14:22 paste map, type "go" press enter 2010-09-13T12:14:44 run it in command prompt 2010-09-13T12:15:11 there is Paste in context menu of title bar 2010-09-13T12:15:34 ratatata: also, you didn't really waste 65 ships on that center planet when you could have gotten the one next to it with a growthrate of 5 for 70, right? ;) 2010-09-13T12:16:26 *** evgenius has joined #aichallenge 2010-09-13T12:16:52 *** Obeleh has joined #aichallenge 2010-09-13T12:17:33 hey 2010-09-13T12:17:55 hi 2010-09-13T12:18:05 Baphomet: yeah, just compiling doesn't mean your code is correct ;) 2010-09-13T12:18:09 caytchen: actually, there are two turn 11s, you can reach the invisible one using keyboard arrows & my attacks are barely thought out.. just something in the area of "just get the closest whatnot you can take over and move along" 2010-09-13T12:18:21 Baphomet: and not all runtime errors are logic errors. they could still be fundamental language-related errors 2010-09-13T12:18:29 *** vikhyat has joined #aichallenge 2010-09-13T12:18:43 crashes, in particular, are almost always language-related, and relevant in channels like #c++ 2010-09-13T12:18:44 umm yea its not doing anything im doing 'pw.exe ' then typnig and enterting 'go'? 2010-09-13T12:18:48 I had to throw some attack in to test out some utilities.. over time, it turned into this abomination :) 2010-09-13T12:18:50 man i dropped 300 spots in the rankings. and all matches i lost (in a row) have no game data. wtf? 2010-09-13T12:19:08 well, i guess not even almost always 2010-09-13T12:19:11 ratatata: do you just switch modes of operation based on your supremacy? 2010-09-13T12:19:20 often, would be more correct, especially the hands of a newbie 2010-09-13T12:19:31 ulope: http://github.com/amstan/planetwars-python-kit 2010-09-13T12:19:38 *in the 2010-09-13T12:20:05 i lost like 10 games for no apparent reason. 2010-09-13T12:20:06 ulope: note how i have a makefile and it uses the canvas visualizer 2010-09-13T12:20:14 Hello! Which books/articles about theory related to the challenge anyone recommend me to read? 2010-09-13T12:20:35 susu-magic: they're all on the main site, you could also check the forums, strategy section 2010-09-13T12:20:41 maybe some more specific that general "AI programming in game development" 2010-09-13T12:20:54 amstan: Cool. I didn't know one could add links to other repos in github 2010-09-13T12:21:09 ulope: man git submodule 2010-09-13T12:21:27 im pretty sure the bots dont do anything on their own 2010-09-13T12:21:52 caytchen: if you're referring to my 'dramatic moment', its just a moment of overpower where takeover is almost guaranteed.. but then again, my attacks have calculation errors and are not really predictive.. 2010-09-13T12:21:52 amstan: Ahh. Sure... I've never seen one on github tough 2010-09-13T12:22:10 Baphomet: pw.exe then paste a map 2010-09-13T12:22:11 amstan: "Advanced Strategy Guide" is empty, and Strategy section on forums is almost empty 2010-09-13T12:22:12 ulope: it's a git feature, github's just detecting that it's from them and linking it 2010-09-13T12:22:31 caytchen: lets just say, I'm using 30ms max in Java ^^ 2010-09-13T12:22:39 amstan: Yeah I know. I was just saying that i have never seen it used on github before 2010-09-13T12:23:00 susu-magic: strategies for this game are probably very specific to this game. at least, i haven't seen much academic research on games like this 2010-09-13T12:23:03 susu-magic: classic game theory might not have obvious use cases here 2010-09-13T12:23:09 susu-magic: that's kind of the appeal to this game :D 2010-09-13T12:23:48 susu-magic: just do what everyone else does: build heuristics. 2010-09-13T12:24:08 susu-magic: some things could apply to "small" parts of your ai, but the "big" tactics are all up to you and probably won't be derived directly from existing research 2010-09-13T12:24:18 okay 2010-09-13T12:24:27 i have to admit that my first attempt is going to be... kind of classical 2010-09-13T12:24:28 thx 2010-09-13T12:24:37 but with a non-traditional application 2010-09-13T12:24:47 of the theory 2010-09-13T12:25:03 so it's still very experimental and may not (probably won't) work well 2010-09-13T12:25:22 but i figure we have a couple months this time instead of just one, so i might as well experiment more 2010-09-13T12:25:47 amstan: I'm going to pull this into my repo when I get home from work 2010-09-13T12:25:48 heuristics rule. at least until I get to the part where actually good strategy is needed 2010-09-13T12:26:04 heuristics *should* be good strategy 2010-09-13T12:26:04 ok well i have it running on its own (not by that method you said, it doesnt work) and yea it just seems to randomly stop after 30 turns :\ 2010-09-13T12:26:09 well heuristics are pretty much good strategy 2010-09-13T12:26:30 they can easily beat most humans atleat in this game 2010-09-13T12:26:46 and that's unusual for heuristics 2010-09-13T12:27:00 *** hornairs has joined #aichallenge 2010-09-13T12:27:03 i think the main reason is just raw processing power 2010-09-13T12:27:21 *** amstan has quit IRC (Read error: Operation timed out) 2010-09-13T12:27:26 well, more specifically, this game is very amenable to number crunching 2010-09-13T12:27:30 which humans are bad at 2010-09-13T12:27:35 Baphomet: bow you can start debugging your app, use something like ptrace 2010-09-13T12:27:51 can anyone tell me how the contest compiler deals with rubygems? 2010-09-13T12:27:54 anyone tried some kind of genetic algorithms or some self-learning? 2010-09-13T12:28:05 susu-magic ive been doing a bit of GA 2010-09-13T12:28:08 hornairs: have they added ruby support allready? 2010-09-13T12:28:09 so far it hasnt really panned out 2010-09-13T12:28:12 *** malumalu has joined #aichallenge 2010-09-13T12:28:25 Naktibalda i assumed it was in there cause it was last year but perhaps it isnt even supported 2010-09-13T12:28:28 hornairs: it's probably just agnostic to it (which means if gems isn't installed you just can't use those libs unless you include them in your zip, probably) 2010-09-13T12:28:41 susu-magic: someone on forums tried GA 2010-09-13T12:28:45 take a look at starter package page 2010-09-13T12:28:53 hornairs: you're doing GA with ruby? that might not be the greatest language for that application 2010-09-13T12:29:02 jmcarthur yes i know its really shitty 2010-09-13T12:29:06 since GA is very performance-critical 2010-09-13T12:29:14 but its the only language which i can code in well and quickly 2010-09-13T12:29:19 hornairs: to be honest, i've used ruby for GA before too ^_^ 2010-09-13T12:29:33 but i'm not a ruby-ite anymore 2010-09-13T12:29:42 Yay, I got my bot's backed working :) 2010-09-13T12:29:44 ruby spoils you you know, whenever i go anywhere else i miss some of it, its very developer cuddly 2010-09-13T12:29:49 *** CyberjujuM has joined #aichallenge 2010-09-13T12:29:55 i don't feel that way anymore 2010-09-13T12:30:14 jmcarthur really? what did you move on to? 2010-09-13T12:30:33 maybe not only GA, but combine it with heuristics? make fine algoiythm, parametrize it and do GA on it's parametrs? 2010-09-13T12:30:33 my university wasn't on the list when I signed up. Whom should I contact to add it? 2010-09-13T12:30:41 susu-magic thats exactly what im doing 2010-09-13T12:30:49 hornairs: ha, i bet everybody else in here can answer that question by now 2010-09-13T12:30:53 hornairs: haskell 2010-09-13T12:30:59 sfsgagi: there is a thread in forum 2010-09-13T12:31:04 ok, i see ) 2010-09-13T12:31:10 thanks 2010-09-13T12:31:18 jmcarthur ah yes the language of wizards 2010-09-13T12:31:25 brb 2010-09-13T12:31:28 is that what the ruby community says? 2010-09-13T12:31:30 *** hornairs has quit IRC (Client Quit) 2010-09-13T12:32:37 *** susu-magic has quit IRC (Quit: Page closed) 2010-09-13T12:33:07 i think i take the cake for worst programmer 2010-09-13T12:33:15 i cant even unserstand what i did in my own code 2010-09-13T12:33:45 Baphomet: trust me, if you understand that your code doesn't even work, you're already better than more professionals ;) 2010-09-13T12:34:05 *than most 2010-09-13T12:34:11 hahaha 2010-09-13T12:34:20 *** hornairs has joined #aichallenge 2010-09-13T12:34:40 Baphomet: are you progressing though? 2010-09-13T12:34:41 Baphomet: this makes me hate the world http://thedailywtf.com/ 2010-09-13T12:34:57 im needing to install another program that hilights syntax just so i can see how all my nested curley bracket loops lign up lol 2010-09-13T12:35:07 Baphomet: (and if you don't catch the stupidity in some of the snippets there, don't worry too much. you'll learn) 2010-09-13T12:35:21 jmcarthur no just what i say 2010-09-13T12:35:25 thanks for the tips with ruby 2010-09-13T12:35:38 eh, for what they are worth 2010-09-13T12:36:08 *** mega1 has joined #aichallenge 2010-09-13T12:36:19 hmm... I can't login into forums with rekonq, I'm instantly logged out :) 2010-09-13T12:36:19 *** GordonQuad has joined #aichallenge 2010-09-13T12:36:47 Baphomet: are you on windows? i find codeblocks to be wonderful 2010-09-13T12:37:02 *** mega1_ has quit IRC (Ping timeout: 252 seconds) 2010-09-13T12:37:03 and i can just set up a project file to compule with a few key strokes 2010-09-13T12:37:32 antimatroid: thats the program i JUST installed like 2 mins ago to hylight my syntax 2010-09-13T12:37:37 @where spec 2010-09-13T12:37:37 jmcarthur: spec = http://code.google.com/p/ai-contest/wiki/GameSpecification 2010-09-13T12:37:39 old 2010-09-13T12:37:42 @forget spec 2010-09-13T12:37:42 jmcarthur: The operation succeeded. 2010-09-13T12:37:46 @forget specification 2010-09-13T12:37:46 jmcarthur: The operation succeeded. 2010-09-13T12:37:47 was previously using devc++ and couldt see what all my stuff was 2010-09-13T12:37:55 @learn spec as http://ai-contest.com/specification.php 2010-09-13T12:37:55 jmcarthur: The operation succeeded. 2010-09-13T12:38:00 @learn specification as http://ai-contest.com/specification.php 2010-09-13T12:38:00 jmcarthur: The operation succeeded. 2010-09-13T12:39:31 personally i think beginners are better off with higher level languages than c++. c++, IMO, is for experienced OO programmers that need low level code for speed 2010-09-13T12:40:03 my first language is c++ and i love it, everything else seems crappy and weak 2010-09-13T12:40:12 and in fact should only be learned after having used a higher level OO language *and* C first 2010-09-13T12:40:28 c++ was also one of my earlier languages, but i regret that 2010-09-13T12:40:29 jmcarthur, In my experience --even for experienced coders... speed is often over rated. 2010-09-13T12:40:34 delt0r: i agree 2010-09-13T12:40:37 Baphomet: isn't it the only language you know? 2010-09-13T12:40:58 delt0r: for for brute force algorithms or for (some) high end games, it's hard to beat C or C++ 2010-09-13T12:41:01 *but for 2010-09-13T12:41:49 jmcarthur, I would not necessarily agree... but if you need a asm block of sse instructions... then C can rock 2010-09-13T12:41:56 and C++ 2010-09-13T12:42:05 i'm not a c++ fan at all 2010-09-13T12:42:11 Naktibalda: i would say i 'know' it too well, but i have some experance with it, more so that anything else 2010-09-13T12:42:18 But my work code is in java... and its 5x faster than the C code it replaces 2010-09-13T12:42:22 C is also good for non-speed reasons 2010-09-13T12:42:29 (linux kernel programming, for example) 2010-09-13T12:42:33 not that java all that high level 2010-09-13T12:42:42 jmcarthur, True 2010-09-13T12:42:44 most C code is very poorly written 2010-09-13T12:42:50 *** dhartmei has quit IRC (Quit: brb) 2010-09-13T12:42:56 Most Code is very poorly writen 2010-09-13T12:43:01 written 2010-09-13T12:43:05 i'm talking exclusively about very well-written C code here 2010-09-13T12:43:08 *** dhartmei has joined #aichallenge 2010-09-13T12:43:08 by people like me 2010-09-13T12:43:11 lol 2010-09-13T12:43:14 heh 2010-09-13T12:43:47 average java is going to be safer and faster than average c and average c++ 2010-09-13T12:43:50 I am very please this comp is so language diverse. 2010-09-13T12:44:00 but highly tuned java well rarely be on par with highly tuned c or c++ 2010-09-13T12:44:25 with the java 6 and java 7 --it can be 2010-09-13T12:44:32 but not for all thing 2010-09-13T12:44:34 things 2010-09-13T12:44:42 and tuning java is also less obvious than tuning c/c++ 2010-09-13T12:44:47 Its still lacks good SSE optimizations 2010-09-13T12:45:05 what email should I get the signup confirmation from? 2010-09-13T12:45:12 jmcarthur, Thats just knowing your tool. A java programmer would say the same about C 2010-09-13T12:45:18 a similar argument applies to my language of choice 2010-09-13T12:45:27 exactly 2010-09-13T12:45:46 alright, i have a new bot up, not sure if it's going to be better or worse 2010-09-13T12:45:52 seemed to bet some of the better people just before 2010-09-13T12:45:54 I would claim all things being == java has about the same performance as C/C++ 2010-09-13T12:45:57 BUT 2010-09-13T12:46:17 I do heavy numeric cals for days at a time 2010-09-13T12:46:18 still can't bloody beat a1k0n, beat vortex 2010-09-13T12:46:20 *** sjbr has joined #aichallenge 2010-09-13T12:46:28 I don't do GUI's :) 2010-09-13T12:46:56 chick magnet profession 2010-09-13T12:47:18 bad haskell code will be safer but far slower than bad c. average haskell code will be safer and faster than average c. great haskell code will be a little safer and about the same speed as great c. highly tuned haskell will be about the same safety and speed as highly tuned c. 2010-09-13T12:47:22 oh wait, i only beat vortex cause he screwed up 2010-09-13T12:47:29 hmmm, prboably not better, oh well 2010-09-13T12:47:45 actualy, great was wrong 2010-09-13T12:47:51 great should be a bit slower than c 2010-09-13T12:48:02 *** AJC has quit IRC (Ping timeout: 252 seconds) 2010-09-13T12:48:07 I quite liked haskell... 2010-09-13T12:48:25 but i like Scheme more 2010-09-13T12:48:29 The profile can't be edited? I want to change my country which currently is Niger 2010-09-13T12:48:41 unfortunally the compilers and tools just arn't there yet for these langs 2010-09-13T12:48:57 delt0r: yeah, the JVM is awesome for backend stuff. not so for interactive apps, IMO 2010-09-13T12:49:10 i think ghc is great 2010-09-13T12:49:10 Thats what i am told 2010-09-13T12:49:36 oh man i just keep moving my code around and my bot just does even more strange things every time 2010-09-13T12:49:41 delt0r: what do you mean? GHC is quite there. 2010-09-13T12:49:50 good old ghc... I complied it back when 64bit was new for desktops... took a while 2010-09-13T12:49:56 it has a great optimizer, great parallelism primitives, a fast runtime, a built in time and heap profiler, all kinds of niceties 2010-09-13T12:50:00 yet i havnt been able to fix this random crasing bug yet 2010-09-13T12:50:24 that's a long time ago... 2010-09-13T12:50:24 mietek, Error messages and that sort of thing. even for ghc when i last used it 2010-09-13T12:50:25 Compiling it does take a while, that's why there are binaries provided. 2010-09-13T12:50:39 i haven't had any problem with ghc error messages... 2010-09-13T12:51:04 I did --but that was a while ago... I still do with different scheme implementations... 2010-09-13T12:51:09 mietek: compiling it is faster than it used to be at least. or maybe i just have a faster machine :) 2010-09-13T12:51:18 you just can't assume that where it barfed is where the problem is 2010-09-13T12:51:35 I can see how they could be difficult to understand for people new to the language. They've been improved, though -- for example, if you attempt to use a language extension without specifying it, you'll get a helpful suggestion, etc. 2010-09-13T12:51:38 http://pastebin.com/wUU8LxMN maybe someone can quickly see where i went wrong and why it keep crashing after <40 rounds no matter what i do. that is if you dont want to rip your eyes out after seeing my horrible code.... 2010-09-13T12:51:39 eps with macros in scheme 2010-09-13T12:51:42 delt0r: okay, i'm with you there, but that's only for the syntax errors 2010-09-13T12:51:48 the type errors are very clear, IMO 2010-09-13T12:52:11 and once you have learned why the syntax errors are reported the way they are, it's easy to correct anyway 2010-09-13T12:52:17 but my biggest problem is that I need others to be able to take over my code. 2010-09-13T12:52:27 Baphomet: have you considered giving Python a try? 2010-09-13T12:52:32 yeah, that's why i haven't been able to use haskell much in the workplace yet 2010-09-13T12:52:33 I work with mathematicians 2010-09-13T12:52:35 Baphomet: probably division by zero 2010-09-13T12:52:41 Now, Erlang error messages... 2010-09-13T12:52:46 *** RainCT has quit IRC (Remote host closed the connection) 2010-09-13T12:53:03 And I had to explain that java is not really new anymore... and No fortran is not cool 2010-09-13T12:53:05 Naktibalda: hahah really? where did i do that? 2010-09-13T12:53:19 Baphomet: does it always crash on exactly the same turn? 2010-09-13T12:53:20 no idea 2010-09-13T12:53:35 trying to figure out how to connect my bot coded in java to benzedrine.cx, but never coded in c so not sure how to go about it. Any pointers? 2010-09-13T12:53:51 yea, and i can move the code round a little and it changes where it crashes at 2010-09-13T12:53:59 *** cherwin has joined #aichallenge 2010-09-13T12:53:59 Strepto, use the java TCP? 2010-09-13T12:54:24 and if I say that I have no experience of that either :) 2010-09-13T12:54:43 *** evgenius has left #aichallenge 2010-09-13T12:54:46 But if you are on a real linuix distro--just gcc the source and run it.. its just a command line app... you don't need to know C 2010-09-13T12:54:53 Baphomet: it's impossible to find a runtime but in such big piece of code 2010-09-13T12:54:55 Baphomet: does it always crash on the same turn regardless of map or opponent? 2010-09-13T12:55:05 Naktibalda: no it's not :P 2010-09-13T12:55:31 Baphomet: why so many nearly identical else clauses? 2010-09-13T12:55:35 it looks like starter package with strange fleet limit 2010-09-13T12:55:53 because i suck at programming and dont know how to do it any other way :\ 2010-09-13T12:56:10 eh, we won't get into that. let's just find that bug 2010-09-13T12:56:16 *** amstan has joined #aichallenge 2010-09-13T12:56:16 *** ChanServ sets mode: +o amstan 2010-09-13T12:56:29 *** ulope has left #aichallenge 2010-09-13T12:56:30 yea it crashes same time regardless of enemy 2010-09-13T12:56:40 *** ulope has joined #aichallenge 2010-09-13T12:56:58 *** cobracom has quit IRC (Ping timeout: 252 seconds) 2010-09-13T12:57:17 and regardless of map? 2010-09-13T12:57:21 in the same turn? 2010-09-13T12:57:22 *** jmreardon has joined #aichallenge 2010-09-13T12:57:33 the main armount of repeated code is the IssueOrder area, where i have diferent variations dpending on how many planets i have and how many source ships 2010-09-13T12:57:36 Baphomet: also, define "crash" 2010-09-13T12:57:53 crash = suddenly stop in game and i loose 2010-09-13T12:58:00 failure to complete turn 2010-09-13T12:58:08 Baphomet: no error message though? 2010-09-13T12:58:11 oh is it a timeout? 2010-09-13T12:58:24 *** awwaiid has quit IRC (Quit: Page closed) 2010-09-13T12:58:48 sent invalid command maybe... 2010-09-13T12:58:52 * jmcarthur downloads the c++ starter so he can diff with it 2010-09-13T12:59:13 i could just be better off scraping this one and starting it all again 2010-09-13T12:59:21 im probbly waiting you time here guys 2010-09-13T12:59:36 nah, i like helping beginners 2010-09-13T13:00:30 i am in train... 2010-09-13T13:01:47 *** jmreardon has quit IRC (Ping timeout: 252 seconds) 2010-09-13T13:01:56 Baphomet: have you tried opening a log file and printing little tracers into it to show where in the code you are? 2010-09-13T13:02:10 Baphomet: that's a common debugging trick 2010-09-13T13:02:17 *** Obeleh has quit IRC (Read error: Connection reset by peer) 2010-09-13T13:02:20 Baphomet: when it crashes you know which tracer it executed last 2010-09-13T13:03:05 Baphomet: i'm suspecting that you are issuing an invalid order though, personally 2010-09-13T13:03:57 yah, im stripping down all the code right now to see if i can find whats breaking it 2010-09-13T13:04:59 *** AJC has joined #aichallenge 2010-09-13T13:06:21 *** vikhyat has quit IRC (Ping timeout: 265 seconds) 2010-09-13T13:06:48 im thinking it may have something to do with the for loop on issueorder loooping once for every plannet i have? 2010-09-13T13:06:51 *** computerwiz_222 has joined #aichallenge 2010-09-13T13:07:09 but if i dont have that then how to i make all my planets send fleets on 1 turn? 2010-09-13T13:08:48 I can't imagine that the loop is the problem in itself, unless you're exceeding max turn time 2010-09-13T13:08:58 Baphomet: are you making sure that you actually own the planet you are sending ships from? 2010-09-13T13:09:18 the profile can't be edited in http://ai-contest.com 2010-09-13T13:09:22 oh, yeah, i think you are 2010-09-13T13:09:35 well i deleted the loop bit and it doesnt crash now. but theres like 1/0th of the code from before 2010-09-13T13:09:43 Strepto: if he was exceeding time limit it would say that though 2010-09-13T13:09:51 aah 2010-09-13T13:09:57 Baphomet: 1/0 is division by zero :) 2010-09-13T13:10:10 1/10th* lol 2010-09-13T13:10:31 but i usually spend my spare time dividing by 0 anyway just for enjoyment 2010-09-13T13:11:04 yay, bot's first match, a loss, but progress nonetheless 2010-09-13T13:13:49 jmcarthur, thanks for the Game Spec. you wrote :) 2010-09-13T13:14:01 Palmik: you're welcome! i hope it helps a lot of people 2010-09-13T13:14:08 writing it helped me 2010-09-13T13:14:41 Baphomet: line 99 and line 133 both issue the same order i think 2010-09-13T13:16:37 yea, but on- woah i have no idea how i even manage these things. 2010-09-13T13:16:42 Baphomet: which means if the order uses more than half of the ships at the source planet then the second time you issue it you are issuing an illegal order 2010-09-13T13:16:53 yea i see now 2010-09-13T13:17:04 Baphomet: your indentation made it especially hard to follow 2010-09-13T13:17:11 *** jmreardon has joined #aichallenge 2010-09-13T13:17:17 Baphomet: and you should extract some of that duplicated code to a separate function 2010-09-13T13:17:26 Baphomet: it would vastly clean things up and make it easier to read 2010-09-13T13:17:42 Baphomet: the most important thing in programming isn't writing code quickly. it's reading code quickly 2010-09-13T13:17:59 s/reading/understanding/ 2010-09-13T13:18:22 yea im having a cleanup right now 2010-09-13T13:18:34 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-09-13T13:20:15 *** ulope has left #aichallenge 2010-09-13T13:23:14 *** jmreardon has left #aichallenge 2010-09-13T13:23:59 *** jmreardon has joined #aichallenge 2010-09-13T13:25:57 okay, we definitely need the tournament manager to do the +/- 10 rank thing 2010-09-13T13:26:14 what's that? 2010-09-13T13:26:32 or even just form a long-tailed distribution with a shape parameter determined by the confidence interval output by bayeselo 2010-09-13T13:26:40 Palmik: currently it's pitting everybody against everybody else, randomly 2010-09-13T13:27:05 which means people in the top 10 are fighting, on average, people ranked halfway down the list, which adds very little information when they win 2010-09-13T13:27:10 *** phreeza_mobile has joined #aichallenge 2010-09-13T13:27:39 in fact information gain might be a good way to find matches 2010-09-13T13:27:44 yeah, well, +/- 10 would not be ideal imo, what if there is no such person? Why not just that the 10 above you, 10 under you and pick random from them? 2010-09-13T13:27:55 that's what i mean 2010-09-13T13:28:03 not +/- 10 elo points, +/- 10 ranks 2010-09-13T13:28:14 Ah, sorry, misread it 2010-09-13T13:28:33 is that still not in yet? 2010-09-13T13:28:35 geez 2010-09-13T13:28:40 i think it should be overlapping confidence intervals, not ranks 2010-09-13T13:28:52 *** zerd has quit IRC (Read error: Connection reset by peer) 2010-09-13T13:28:56 *** zerd has joined #aichallenge 2010-09-13T13:28:56 *not +/- n ranks 2010-09-13T13:29:30 *** sinan has quit IRC (Ping timeout: 264 seconds) 2010-09-13T13:29:31 maybe next time i write patches i should be on the team, dq myself, etc. 2010-09-13T13:29:33 well, yes. but i think there is value in getting the occasional wildly random match 2010-09-13T13:29:43 we could just throw those in 2010-09-13T13:29:51 +/- 10 ranks could mean local minima anyway 2010-09-13T13:29:57 so we'd still want the occasional random match 2010-09-13T13:30:01 danielvf could probably get that going 2010-09-13T13:30:15 *** ajhager has quit IRC (Ping timeout: 276 seconds) 2010-09-13T13:30:18 i think the long-tailed distribution idea is best 2010-09-13T13:30:52 hmm, that could work 2010-09-13T13:31:19 but i still favor choosing which bots need to play more games based on confidence interval rather than on games played 2010-09-13T13:31:20 cauchy, shape=(max confidence - min confidence / 4 or so) 2010-09-13T13:31:34 why not x% of random match, x% of elho match, and x% of within 10 positions? 2010-09-13T13:31:42 yeah, global information gain. 2010-09-13T13:31:46 Cyndre: we'd only need one of the latter two 2010-09-13T13:31:58 true 2010-09-13T13:32:05 as long as it's constrained not to give 10 games in a row to the same dude 2010-09-13T13:32:10 Cyndre: i think the long tail + prioritize bots with wide confidence intervals would be a good balance 2010-09-13T13:32:19 *** mega1 has quit IRC (Ping timeout: 265 seconds) 2010-09-13T13:32:30 hey, what did you think of my app jmcarthur btw? 2010-09-13T13:32:38 so.. anyone here have a modified tournament engine that has a passthrough for stderr? 2010-09-13T13:32:49 amstan: http://a1k0n.net/temp/PlayGame.jar 2010-09-13T13:32:49 Cyndre: oh, i can't say i really understood it from the slides -_- 2010-09-13T13:32:50 er, 2010-09-13T13:32:53 PlayGame-1.2.jar 2010-09-13T13:33:06 does that not work for that dude? 2010-09-13T13:33:24 amstan: local or server? 2010-09-13T13:33:30 jmcarthur: domain experience helps alot - basicly a replacement for autocad when designing infrastructure overtop of a city 2010-09-13T13:33:33 jmcarthur: local 2010-09-13T13:33:39 will this work? http://gist.github.com/574932 2010-09-13T13:33:51 ah, then a1k0n's version should do that, shouldn't it? 2010-09-13T13:34:03 *** hornairs has quit IRC (Quit: hornairs) 2010-09-13T13:34:08 where do i get the jars from? 2010-09-13T13:34:12 !*(# 2010-09-13T13:34:19 amstan: the url i just scribbled at you 2010-09-13T13:34:25 amstan: http://a1k0n.net/temp/PlayGame-1.2.jar 2010-09-13T13:34:32 oh i see, you actually mean a tourny, not just an engine 2010-09-13T13:34:33 also ShowGmae-1.2.jar 2010-09-13T13:34:44 a1k0n: oh.. sorry, i missed it 2010-09-13T13:35:23 that python is much cleaner than most of the server-side code 2010-09-13T13:36:11 *** ulope has joined #aichallenge 2010-09-13T13:36:24 *** Naktibalda has joined #aichallenge 2010-09-13T13:36:33 *** sjbr has left #aichallenge 2010-09-13T13:38:28 *** kozlovsky has joined #aichallenge 2010-09-13T13:43:49 hmz 2010-09-13T13:43:56 http://ai-contest.com/visualizer.php?game_id=4455908 2010-09-13T13:44:00 and this is a lose ? 2010-09-13T13:44:10 so, ehmz, timeout ? oom ? 2010-09-13T13:44:22 runs very strongly on benzedrine 2010-09-13T13:44:27 bartwe: are you computing minimum spanning tree or is that just an emergent effect of sending to the closest planet? 2010-09-13T13:44:43 bartwe: that was a loss for you? wtf? 2010-09-13T13:44:56 i guess there was a timeout triggered or something 2010-09-13T13:45:00 *** retybok has joined #aichallenge 2010-09-13T13:45:02 mono sucks performance wise 2010-09-13T13:45:15 not sure what to call the algorithm 2010-09-13T13:45:22 bartwe: the entire server situation is frankly a disaster 2010-09-13T13:45:34 Frankly it's easier to simply program in c++, that way you don't have to worry too much about performance 2010-09-13T13:45:38 i pick a central planet, and then mark the shortest edge that can be marked from it and mark it 2010-09-13T13:45:52 Call it "bartwe's fleet army of death" 2010-09-13T13:45:53 i really hope we manage to do a distributed tournament so we can loosen the, IMO, crippling constraints on the bots 2010-09-13T13:45:55 and then keep doing that greedily? 2010-09-13T13:45:59 (i mean for this challenge, not in general of course) 2010-09-13T13:46:03 retybok: except that c++ is not a language i know well 2010-09-13T13:46:14 retybok: i disagree 2010-09-13T13:46:21 a1k0n, bartwe your bots are pretty sweet 2010-09-13T13:46:25 any reasonable language makes it easy enough to bind to C anyway 2010-09-13T13:46:41 bartwe: you probably reinvented prim's algorithm or kruskal's algorithm (they are very simple) 2010-09-13T13:46:42 and there exist languages that are both fast *and* have high levels of abstraction 2010-09-13T13:46:48 if you really need that speed 2010-09-13T13:46:50 there is some additional smartness in determining the numbers to send 2010-09-13T13:47:01 a1k0n: one min, checking it on wikipedia 2010-09-13T13:47:02 (and also have really good foreign function interfaces) 2010-09-13T13:47:20 You can write fortran in any language 2010-09-13T13:47:50 a1k0n: yeah prim sounds about right 2010-09-13T13:47:54 i was considering doing the same thing so was surprised to see it in your bot right after i thought of it 2010-09-13T13:47:55 greghaynes: well my point is that you can write fast enough code in any (reasonable) language, not that you can write fortran *style* in any language 2010-09-13T13:48:11 a1k0n: mind reading 2010-09-13T13:48:16 Yea, mostly true 2010-09-13T13:48:29 a1k0n: actually it came about due to not wanting to have the fleets in flight for long times 2010-09-13T13:48:39 right, less uncertainty in outcome 2010-09-13T13:48:39 I mean, if you need a highly recursive algo you may kill yourself trying to get it to work quickly in pythin 2010-09-13T13:48:46 greghaynes: of course, i don't think that argument applies to, e.g., writing GA in ruby ;) 2010-09-13T13:49:02 i'll put tcpserver support into the binary and start profiling 2010-09-13T13:49:16 bartwe: into the binary? 2010-09-13T13:49:18 and the ships can be reporposed more quickly 2010-09-13T13:49:19 Use the tcpserver visualiser! 2010-09-13T13:49:30 Its magical and in realtime 2010-09-13T13:49:38 2010-09-13T13:49:44 greghaynes: :) 2010-09-13T13:49:50 Yay, that Prim's algorithm was invented by Czech :) 2010-09-13T13:49:58 greghaynes: in all honesty, i don't need realytime, i need unwind 2010-09-13T13:50:04 or found I should say 2010-09-13T13:50:11 i also thought about minimum spanning trees, but i didn't think it would work out so well 2010-09-13T13:50:14 i usually let it run a few games and then do a post mortem on the web interface.. 2010-09-13T13:50:16 i was apparently wrong 2010-09-13T13:50:24 greghaynes, I could never get that to compile for windows 2010-09-13T13:50:30 well, the argument could be made that if you're sending ships to your own planets you might as well take the shortest path 2010-09-13T13:50:34 bartwe: It applies for the rest of us who like watching your algo stomp us in real time 2010-09-13T13:50:56 greghaynes: glad to be of service 2010-09-13T13:50:56 rogue780: Ah, thanks for trying 2010-09-13T13:51:05 the shortest path in this game is almost always going to be directly to the target though, not through an intermediate 2010-09-13T13:51:13 greghaynes: if you want to see stomping 2010-09-13T13:51:17 that's why the results of this trick seem counterintuitive to me 2010-09-13T13:51:23 greghaynes: vortex.9 is killer 2010-09-13T13:51:49 jmcarthur: its about responding to changes 2010-09-13T13:52:03 i kind of suspect that the real benefit this has is that it protects your "front lines" very well 2010-09-13T13:52:05 jmcarthur: if my predictor was worth its salt i could use directer paths 2010-09-13T13:52:09 wow, vortex.9 goes right for the throat 2010-09-13T13:52:11 *** amstan has quit IRC (Remote host closed the connection) 2010-09-13T13:52:16 *** amstan has joined #aichallenge 2010-09-13T13:52:16 *** ChanServ sets mode: +o amstan 2010-09-13T13:52:23 greghaynes, if you know of a way to compile it without qmake, let me know and I'll try that 2010-09-13T13:52:34 jmcarthur: true, routing units through hot zones causes some enemies to mis predict the available force at such a point 2010-09-13T13:52:46 right 2010-09-13T13:52:55 rogue780: I think I can hunt down a win box in my lab today so ill try some things out 2010-09-13T13:53:07 *** Utkarsh has quit IRC (Ping timeout: 252 seconds) 2010-09-13T13:53:25 http://www.benzedrine.cx/planetwars/canvas?game_id=1284397457|bartwe.6|vortex.9 owned ;) 2010-09-13T13:53:34 yeah, was going to post 2010-09-13T13:53:38 kill in 42 moves 2010-09-13T13:53:46 no chance at all 2010-09-13T13:53:54 well i guess the map had a play there 2010-09-13T13:53:59 vortex seems to be an ok bot, with one killer trick 2010-09-13T13:54:09 it wait till you take out a neutral and pounces 2010-09-13T13:54:40 i actually had that in too, but on the very first move on maps with close starter planets 2010-09-13T13:54:42 bartwe.6 is too aggressive, spending all its units on planets 2010-09-13T13:54:49 my wife is currently behind me taking her laptop completely apart 2010-09-13T13:54:52 i love my wife 2010-09-13T13:55:14 The python starter package doesn't work correctly on *nix. You have to convert it with dos2unix. 2010-09-13T13:55:24 i guess i can stomp vortex by not moving first on neutrals 2010-09-13T13:55:30 cherwin: seriously? python scripts aren't agnostic to that? 2010-09-13T13:56:18 jmcarthur: Apparently not. 2010-09-13T13:56:19 if theres one thing i hate with python its that they rely on tabs for syntax 2010-09-13T13:56:24 so this shouldn't come as a surprise heh 2010-09-13T13:56:36 i don't mind indentation-significant syntax 2010-09-13T13:56:40 i just hate the way python does it 2010-09-13T13:56:49 caytchen: It doesn't have *anything* to do with the tabs. 2010-09-13T13:56:53 I think python can handle it fine, it's the #! that has a problem when the shell tries to use it 2010-09-13T13:57:09 what do you guys think i should do to win http://ai-contest.com/visualizer.php?game_id=4455834 ? 2010-09-13T13:57:11 Janzert: Yes, you're right. 2010-09-13T13:57:18 i'm guessing a timeout 2010-09-13T13:57:39 still i want a top10 position and not at the end of the list :) 2010-09-13T13:57:53 bartwe: you should push the admins/organizers to give the bots some more resources 2010-09-13T13:57:55 bartwe: you didn't ? 2010-09-13T13:58:03 bartwe: so did you stay completely clear of LINQ? 2010-09-13T13:58:04 *** Utkarsh has joined #aichallenge 2010-09-13T13:58:09 or have they upgraded? 2010-09-13T13:58:14 caytchen: yeah, avoided it like the plague 2010-09-13T13:58:34 Somebody should put the dos2unix tip in the FAQ or something. 2010-09-13T13:58:35 caytchen: normally i love linq, but in this case i want zero allocation if possible 2010-09-13T13:58:41 *** Vi0 has quit IRC (Ping timeout: 240 seconds) 2010-09-13T13:58:58 *** mega1 has joined #aichallenge 2010-09-13T13:59:04 about LINQ, what's the general concencus on using it? 2010-09-13T13:59:23 know what it costs 2010-09-13T13:59:30 and what it delivers 2010-09-13T13:59:38 its short, easy, powerfull 2010-09-13T13:59:57 also it does indirections, vcalls and allocations 2010-09-13T14:00:15 which often doesn't matter if the algo below it is good 2010-09-13T14:02:28 seems alot of bots use the sample algorithm :( 2010-09-13T14:02:38 * ermau|work wants LINQ 2010-09-13T14:02:39 Badly. 2010-09-13T14:03:13 bartwe, do you only travel with intermediate planets when the time is as long as if it would be direct flight or do you have some maximum margin allowed perhaps? 2010-09-13T14:03:23 it's just a list comprehension, basically 2010-09-13T14:03:54 Palmik: i only travel through intermediaries if the travel is optional 2010-09-13T14:03:55 bartwe: i love the visulisation of your code, how it builds those support chains etc 2010-09-13T14:03:59 yeah, its a perfect match for this competetion... you want functional programming, and you want C#. 2010-09-13T14:04:03 or a specific case of the general idea of monad 2010-09-13T14:04:15 caytchen: python has list comps too, to be fair 2010-09-13T14:05:19 still, i got loads of arrays and forloops 2010-09-13T14:05:39 i'm using haskell vector library for that stuff :D 2010-09-13T14:05:42 really fast 2010-09-13T14:05:56 i just hope we aren't stuck with ghc 6.8.2 in the long run, or it's all for naught 2010-09-13T14:06:09 jmcarthur: geh .... mono ..... 2010-09-13T14:06:18 bartwe: we share this boat 2010-09-13T14:06:27 nah, mono is far worse 2010-09-13T14:06:30 jmcarthur: better not rock it :) 2010-09-13T14:06:31 lol 2010-09-13T14:06:32 especially this old version of it 2010-09-13T14:06:39 ghc 6.8.2 is a cripple 2010-09-13T14:06:47 Yeah, it's ancient 2010-09-13T14:06:52 sounds like the versions debian would ship 2010-09-13T14:07:04 debian actually has more up to date ghc 2010-09-13T14:07:07 far more up to date 2010-09-13T14:07:15 really, supprising 2010-09-13T14:07:24 so ubuntu is the new debian? 2010-09-13T14:07:52 this is an old version of ubuntu to be fair 2010-09-13T14:07:55 This version of Mono only supports the .NET 2 set.. that was released Jan. 2006 2010-09-13T14:08:00 the next version of ubuntu i think will have ghc 6.12.3 2010-09-13T14:08:01 It's a little out of date 2010-09-13T14:08:23 or maybe that's 6.12.1 2010-09-13T14:08:26 don't remember 2010-09-13T14:08:48 http://packages.debian.org/sid/ghc6 2010-09-13T14:08:50 6.12.1 2010-09-13T14:08:52 ermau|work: 3.5 is fun and all, but contains nothing you need 2010-09-13T14:09:05 ermau|work: if you want horror, try java, without generics 2010-09-13T14:09:13 not in this contest ofcourse 2010-09-13T14:09:33 Kavish's bot looks interesting 2010-09-13T14:09:34 Pfft 2010-09-13T14:09:39 Java doesn't have generics, it just thinks it does 2010-09-13T14:09:42 okay, lenny still has only 6.8.2 2010-09-13T14:09:47 java is a different type of "twisted sick". they cripple themselves because its their philosophy. 2010-09-13T14:09:48 so nevermind 2010-09-13T14:09:50 What causes the game to timeout? My bot is timing out randomly now and I can't figure out why 2010-09-13T14:10:05 rogue780: you fail to "go" in 1s 2010-09-13T14:10:16 on official, i think its 5s on tcp 2010-09-13T14:10:25 3sec on tcp i heard 2010-09-13T14:10:35 i'm just doing local testing right now...so 1s? 2010-09-13T14:10:39 http://ai-contest.com/specification.php#bot-io 2010-09-13T14:10:50 3 seconds on first turn, 1 second on subsequent turns 2010-09-13T14:10:52 i guess i could place a gateway on the frontend and force completion at .98 seconds 2010-09-13T14:11:05 * Janzert just realized there is a backward compatible way to give players the number of turns remaining in a game 2010-09-13T14:11:16 Janzert: please tell 2010-09-13T14:11:31 another communication channel? 2010-09-13T14:11:35 file io? 2010-09-13T14:11:43 bartwe: Well, getting subsets without LINQ is a waste of coding time for me 2010-09-13T14:11:52 command argument, maybe? 2010-09-13T14:11:53 slightly ugly, but use a comment at the beginning of the game state 2010-09-13T14:12:01 oh that could work too 2010-09-13T14:12:04 oh yeah, all existing code ignores comments 2010-09-13T14:12:14 i bet some bots will hate that, but it's technically part of the spec 2010-09-13T14:12:14 bartwe: I already know I won't end up maxing out my time so I'm not really worried about that 2010-09-13T14:12:18 yep, and it's a two line change to engine.py 2010-09-13T14:12:19 (except for those who rewrote the parser, like me) 2010-09-13T14:12:21 diff coming up 2010-09-13T14:12:31 my custom parser handles it properly, thank goodness 2010-09-13T14:12:33 bartwe: Besides, the Mono upgrade will be inevitably faster 2010-09-13T14:12:51 Janzert: sweet! you are my hero now 2010-09-13T14:12:56 actually come to think of it all lines not starting iwth 'g', 'F', or 'P' will be ignored so i'm good 2010-09-13T14:13:33 ermau|work: if you have loops in your loops, better not alloc loads of Enumerators and Enumerables 2010-09-13T14:13:38 this seriously would make my bot actually reliable 2010-09-13T14:13:59 hmm...maybe my file io is causing my timeout. every time I make a log entrey i open, write and close the file 2010-09-13T14:14:03 bartwe: Why? 2010-09-13T14:14:07 currently my bot would crash if there is an extra turn beyond 200, and would probably do very badly if there is a turn limit < 200 2010-09-13T14:14:21 rogue780: you can't write logs on the contest site 2010-09-13T14:14:22 ermau|work: GC is slow 2010-09-13T14:14:30 i should know, i wrote one 2010-09-13T14:14:33 ;p 2010-09-13T14:14:42 poor imperative languages and their slow GCs 2010-09-13T14:14:43 a1k0n, I am aware. I haven't submitted to the contest site. I'm still doing local testing 2010-09-13T14:14:43 here's the engine.py diff, I'll try to make a java diff and then submit it to the issue tracker 2010-09-13T14:14:46 http://pastebin.com/6dpmkjss 2010-09-13T14:14:52 *** kozlovsky has quit IRC (Quit: Page closed) 2010-09-13T14:15:00 oh and it's still timing out? i see. 2010-09-13T14:15:10 bartwe: It's not that many enumer(ator|able)s 2010-09-13T14:15:18 a1k0n, as soon as I stopped logging things, it stopped timing out 2010-09-13T14:15:20 Janzert: come to think of it, we could include other things too, like timeout periods 2010-09-13T14:15:28 bartwe: And besides, I'm not running against the time limit, my bot won't be smart enough for any of that 2010-09-13T14:15:29 ermau|work: geh, profile your ass off :) 2010-09-13T14:15:37 Janzert: such as saying 3 for the first turn and 1 for all others 2010-09-13T14:15:41 yep, any optional information could be done this way 2010-09-13T14:15:45 and if it changes we can just change that 2010-09-13T14:15:55 ermau|work: well my main contest entry is randomly getting kicked out, so i do need to shave these kinds of things off 2010-09-13T14:16:00 we could even say whether to use wall time or cpu time 2010-09-13T14:16:13 bartwe: Well I understand, doesn't mean _I_ shouldn't get to use LINQ ;) 2010-09-13T14:16:26 ermau|work: sure, linq your ass off 2010-09-13T14:16:32 bartwe: Like I said, I'm sure the newer version of Mono will be faster, should help you some 2010-09-13T14:16:38 Janzert: i do think perhaps that we should have a special denotation for these comments to say "hey this is some of this special information and not a normal comment" 2010-09-13T14:16:51 #: or something perhaps 2010-09-13T14:17:06 #! 2010-09-13T14:17:08 :P 2010-09-13T14:17:11 not #! :P 2010-09-13T14:17:57 bartwe: http://code.google.com/p/ai-contest/issues/detail?id=101 2010-09-13T14:17:58 actually, since it's not officially spec, we should probably do key-value 2010-09-13T14:18:06 then as we add more we don't break things that are using it 2010-09-13T14:18:19 hmm, maybe just a common format to make parsing easier? Either the bot is expecting it and knows how to handle the comment or it should just ignore it. 2010-09-13T14:18:19 #! key value 2010-09-13T14:18:30 #$ key: value 2010-09-13T14:18:56 or # $ key: value 2010-09-13T14:19:01 i think it should just have a space between key and value since there is probably already code for that 2010-09-13T14:19:11 I seem to recall the python parser needed a space after the # 2010-09-13T14:19:15 ah 2010-09-13T14:19:37 ermau|work: i'm not waiting for magic upgrades 2010-09-13T14:19:44 # : key value 2010-09-13T14:19:45 *** zoned has joined #aichallenge 2010-09-13T14:19:50 bartwe: Oh I'm not waiting either 2010-09-13T14:19:55 bartwe: Doesn't mean it's not worth a push 2010-09-13T14:20:10 i starred it when it was new in the tracker ;p 2010-09-13T14:20:12 the only thing is that limits no spaces in key 2010-09-13T14:20:25 i don't think we will need spaces in keys :P 2010-09-13T14:20:35 bartwe: Can they see how many star an issue, to see demand? 2010-09-13T14:20:36 new languages/getting things upgraded will probably be coming 2010-09-13T14:20:46 *** zoned has quit IRC (Client Quit) 2010-09-13T14:20:53 we mostly were concerned with making the server not blow up 2010-09-13T14:20:54 constantly 2010-09-13T14:21:00 I suppose I'm sure we can do without anyway :) 2010-09-13T14:21:31 ermau|work: just gotta keep annoying them with it.. i think we have a month or two left ;P 2010-09-13T14:21:35 *** phreeza_mobile has quit IRC (Quit: Bye) 2010-09-13T14:21:40 honestly i think the format should have just been json in the first place so we could add to it all we want 2010-09-13T14:21:42 dstufft: But since it's blowing up anyway you're going to upgrade ;) ? 2010-09-13T14:21:44 *** amstan has quit IRC (Ping timeout: 252 seconds) 2010-09-13T14:21:52 would have been nice 2010-09-13T14:22:06 dstufft: Fair enough 2010-09-13T14:22:10 or xml or yaml or whatever the hell 2010-09-13T14:22:22 ulope: i don't know why we're stuck on 8.04 2010-09-13T14:22:37 dstufft: because we're afraid of dist-upgrade 2010-09-13T14:22:39 maybe i can convience them to let me do an upgrade i dunno 2010-09-13T14:22:47 lol 2010-09-13T14:22:51 dstufft: every dist-upgrade i've ever performed has been a disaster 2010-09-13T14:23:00 ive never had one not work 2010-09-13T14:23:02 ever 2010-09-13T14:23:07 i've never had one *work* 2010-09-13T14:23:19 *** xha has quit IRC (Quit: Odcházím) 2010-09-13T14:23:22 Maybe this one will half work? 2010-09-13T14:23:23 I think few hours of downtime for backup + proper upgrade would be worth it :) 2010-09-13T14:23:34 i'd love to have it up to date, myself 2010-09-13T14:23:39 Palmik: Since everyone is playing on the TCP server anyway? Yeah, I'd say so 2010-09-13T14:23:57 although i posted to the mailing list a better server for the same price so if they choose to switch that would be a good time ^.^ 2010-09-13T14:23:58 dstufft: if you are *very* confident that it would work, you are free to try to convince the others, and i will keep with a neutral vote 2010-09-13T14:24:03 jmcarthur: on debian: yes, but in ubuntu with do-release-upgrade I've never had major problems besides exotic pacakges failing 2010-09-13T14:24:12 ermau|work, yep, more importantly, it mich dlow donw the competition for a few hours but it micht get faster afterwars :) 2010-09-13T14:24:13 ulope: ubuntu is exactly what i'm talking about 2010-09-13T14:24:22 i'd rather upgrade debian 2010-09-13T14:24:23 *** hornairs has joined #aichallenge 2010-09-13T14:24:33 than ubuntu 2010-09-13T14:24:38 jmcarthur: well then I guess I've been lucky so far 2010-09-13T14:24:38 *might slow down 2010-09-13T14:24:44 ive done them both, only time ive had an issue was a retarded stable + testing + untable mix 2010-09-13T14:24:51 ew 2010-09-13T14:24:54 that was a horrible mess 2010-09-13T14:24:58 dstufft: Oh yes, that can be "fun" 2010-09-13T14:25:17 i'm an arch guy. i don't mind upgrading normally. but for ubuntu i make an exception due to so many problems in the past 2010-09-13T14:25:35 in fact 2010-09-13T14:25:44 my current vps, that im on right now as my shell server 2010-09-13T14:25:47 was a 8.04 server 2010-09-13T14:25:55 that i went through one release at a time to get to 10.04 2010-09-13T14:25:57 *** phreeza has joined #aichallenge 2010-09-13T14:26:06 god 2010-09-13T14:26:09 *** gw has joined #aichallenge 2010-09-13T14:26:11 that must have sucked to wait for 2010-09-13T14:26:33 i did it while watching tv lol 2010-09-13T14:27:05 i went from 9.10 to 10.04 on my htpc not long ago and killed everything. took me several hours to get working again 2010-09-13T14:27:27 next time i feel the desire to upgrade, i'm just wiping it and installing an OS i like 2010-09-13T14:28:10 and let's not get started on my experiences with ubuntu at my last job. my coworkers decided that ubuntu simply hates me 2010-09-13T14:28:25 even from a fresh install 2010-09-13T14:28:37 lol 2010-09-13T14:28:38 maybe dist-upgrade along isn't my problem. it's just ubuntu :P 2010-09-13T14:29:04 i'm talking two years at this job, and every time i set up an ubuntu machine (we did that a lot since we did kernel dev) it was fail 2010-09-13T14:29:08 i can't decide if i like ubuntu or debian more for a server 2010-09-13T14:29:12 which is why i felt justified in using arch there 2010-09-13T14:29:26 i used to strictly use debian 2010-09-13T14:29:44 but software versions made me start using ubuntu 2010-09-13T14:29:51 next time i set up a server i will suck it up and go with a BSD i think 2010-09-13T14:30:06 need to learn it better anyway 2010-09-13T14:30:26 i like me some BSD too 2010-09-13T14:30:28 it's fun =) 2010-09-13T14:30:33 *** amstan has joined #aichallenge 2010-09-13T14:30:33 *** ChanServ sets mode: +o amstan 2010-09-13T14:30:38 i just started a new job doing freebsd kernel dev a week ago 2010-09-13T14:30:42 used to run all our routers on openbsd 2010-09-13T14:30:43 mark_jn: awesome! 2010-09-13T14:30:52 mark_jn: they pay you to do that? 2010-09-13T14:30:53 back when i still worked at an ISP 2010-09-13T14:30:54 jmcarthur: (Free)BSD is fun (esp. since it has ZFS 2010-09-13T14:30:54 i'm glad to be out of linux kernel dev 2010-09-13T14:31:05 hey i just had an idea: we can evaluate tournament match choice algorithms fairly easily. 2010-09-13T14:31:24 we didn't have the cash for fancy cisco's 2010-09-13T14:31:25 yeah, the company has lots of nonstandard hardware 2010-09-13T14:31:33 so we got openbsd boxes 2010-09-13T14:31:33 they need driver fixes/support 2010-09-13T14:31:46 mark_jn: what company? they hiring? i'm on a job hunt 2010-09-13T14:31:47 generate a set of 'players' which follow the Elo win/loss rules (probability of a win is the sigmoidal function applied to teh difference in underlying elo) and evaluate how well the true elo is found in N games 2010-09-13T14:31:52 openbsd is awesome for routers 2010-09-13T14:31:59 kernel dev isn't really *fun* to me, but i know kernel dev at least 2010-09-13T14:32:04 jmcarthur: sandvine 2010-09-13T14:32:19 used FreeBSD for our DNS servers too 2010-09-13T14:32:23 not sure if they're hiring, i'm just a co-op student 2010-09-13T14:32:30 i wanted to use OpenBSD or FreeBSD for our WWW boxes 2010-09-13T14:32:34 who happens to know freebsd somewhat well 2010-09-13T14:32:36 but they made me pick a linux 2010-09-13T14:32:58 last time I was an admin we used freebsd for all the servers 2010-09-13T14:33:01 *** hornairs has quit IRC (Remote host closed the connection) 2010-09-13T14:33:02 rock solid 2010-09-13T14:33:08 yea i like bsd 2010-09-13T14:33:21 id use it more but i spend most of my time on VPS's now adays 2010-09-13T14:33:37 a1k0n: oh, you mean an objective test. that might be a good idea 2010-09-13T14:33:52 a1k0n: i also think the behavior up to convergence is important though, and i don't think we can test that objectively 2010-09-13T14:34:31 when you play the best bots against all the default java bots in a tournament on all maps, what kind of results do they get? 2010-09-13T14:34:47 jmcarthur: the other option is to blow away the server and just redo it, with back ups it shouldn't take *that* long 2010-09-13T14:34:54 AJC: i think they pretty much get 100/100 2010-09-13T14:35:07 dstufft: :\ 2010-09-13T14:35:09 AJC: yes, mine does 100/100 2010-09-13T14:35:23 though actually, the script i found that with is suspect, so i'll try again 2010-09-13T14:35:42 a1k0n: a python script? 2010-09-13T14:35:47 a shell script 2010-09-13T14:35:48 I think I'm using the same one... 2010-09-13T14:35:52 ah, OK. 2010-09-13T14:36:14 thanks 2010-09-13T14:38:28 *** amstan has quit IRC (Read error: Connection reset by peer) 2010-09-13T14:38:34 *** amstan has joined #aichallenge 2010-09-13T14:38:34 *** ChanServ sets mode: +o amstan 2010-09-13T14:39:02 *** Theobon has joined #aichallenge 2010-09-13T14:40:16 *** Vio has joined #aichallenge 2010-09-13T14:43:18 funny how C++ is taking over from Python in the rankings. 2010-09-13T14:44:03 http://ai-contest.com/visualizer.php?game_id=4456468 ? 2010-09-13T14:44:17 timeout is the ultimate weapon in this game :s 2010-09-13T14:44:31 I didnt timeout (not that I know of 2010-09-13T14:44:33 ) 2010-09-13T14:44:52 in the event of a timeout does player 1 auto win? 2010-09-13T14:44:59 ha, sure enough, ragebot beats my bot and the script i was using is broken 2010-09-13T14:45:07 ragebot is hard 2010-09-13T14:45:24 I have over 90% beaten when conquered with ragebot 2.0 2010-09-13T14:45:26 ha, think i found my timeout 2010-09-13T14:45:33 a1k0n: what's the % against rage? 2010-09-13T14:45:51 a stupid needless list copy in the templates PlanetWars.cs 2010-09-13T14:45:54 mega1: will know in a couple minutes 2010-09-13T14:46:33 and how come Ive gotten 1 battle today and the guy that I lost to looks like hes playing ever 30 mins 2010-09-13T14:47:05 bartwe: hurr? you use the template? 2010-09-13T14:47:06 hmm, does the java engine run games for one turn too long? 2010-09-13T14:47:14 Cyndre: because it's completly random 2010-09-13T14:47:24 Janzert: very probably 2010-09-13T14:47:34 caytchen: ripped out most of it by now 2010-09-13T14:47:35 i noticed that too but haven't tracked it down 2010-09-13T14:47:38 dstufft: and Ive played tons of games on tcp server and local testing without 1 timeout, first time uploaded timedout 2010-09-13T14:47:48 bartwe: the only sane choice with that piece of code he 2010-09-13T14:48:04 Cyndre: theres a good chance your computer is far better then the server 2010-09-13T14:48:13 mega1: 87/100 maps 2010-09-13T14:48:16 also, the server probably has a load of 10.0 2010-09-13T14:48:18 dstufft: nothing complex about it at all 2010-09-13T14:48:18 thanks 2010-09-13T14:48:21 ok, I noticed that it initialized numTurns to 0 and only stops when numTurns is > maxGameTurns 2010-09-13T14:48:35 it uses wall time for the turns 2010-09-13T14:48:35 numTurns could just get initialized to 1 to fix it 2010-09-13T14:48:49 Load of 2.9 2010-09-13T14:49:28 if I do 1 print does the server kill my bot? 2010-09-13T14:49:30 danielvf: any bots running amok? 2010-09-13T14:49:36 Nope 2010-09-13T14:49:38 and do we know why that happens yet? 2010-09-13T14:49:46 a1k0n: shouldn't be, I added limits 2010-09-13T14:49:53 ok cool. 2010-09-13T14:49:54 after 10 minutes of cpu time the kernel kills a process 2010-09-13T14:49:59 dstufft: did you actually add an address space limit before? 2010-09-13T14:50:00 for the jail users 2010-09-13T14:50:21 a1k0n: yea i added an AS limit at first 2010-09-13T14:50:21 or a datasize limit? 2010-09-13T14:50:26 dstufft: i told you that would happen 2010-09-13T14:53:45 a1k0n: yea, well i limitd it to above what you suggested, 196M, I've never actually used Java so didn't know that it addressed some 700+ mb per instance 2010-09-13T14:53:59 atm memory limits are gone, and it is just cpu time limit 2010-09-13T14:54:00 dstufft: yeah, it allocates gigs of address space and then only uses some of it 2010-09-13T14:54:11 10 minutes of cpu time to be specific 2010-09-13T14:55:00 figure that's good enough, theortically a game should only last 3 minutes and change wall time max 2010-09-13T14:56:17 can probably re add the memory limit and use rss instead of as, but not going to bother unless issues stop poping up again 2010-09-13T15:01:07 *** RainCT has joined #aichallenge 2010-09-13T15:01:21 ermau|work: slowest bit in my code after a stupid list copy ? enumerators.... 2010-09-13T15:02:07 *** computerwiz_222 has left #aichallenge 2010-09-13T15:03:30 bartwe: Doesn't mean it's a significant slow down 2010-09-13T15:03:38 ermau|work: geh... 2010-09-13T15:03:45 *** hellman has quit IRC (Remote host closed the connection) 2010-09-13T15:06:43 dstufft, a1k0n: allocating lots of AS upfront is what you'll see with sbcl too 2010-09-13T15:06:52 yep 2010-09-13T15:07:17 really i was just trying to get rid of the inferior languages :starts a language war: 2010-09-13T15:08:39 if the upper limit on memory used were defined ... 2010-09-13T15:08:49 and it should be, as there is one 2010-09-13T15:09:02 even if right now it'll just get you OOM-killed 2010-09-13T15:09:13 there isn't an uper limit 2010-09-13T15:09:15 not right now 2010-09-13T15:09:21 *** jmreardon has quit IRC (Ping timeout: 245 seconds) 2010-09-13T15:09:22 yes but there is 2010-09-13T15:09:28 the box only has 2 gigs of RAM 2010-09-13T15:09:35 + paging 2010-09-13T15:09:38 right 2010-09-13T15:09:49 for all practical purpuses the amount of physical memory available to the process is what matters 2010-09-13T15:09:50 so if a bot starts swapping the box and then runs it out of VM, it gets killed 2010-09-13T15:10:10 therefore there's an upper limit. somewhere. 2010-09-13T15:10:23 is it possible to get a game log from www.benzedrine.cx ? 2010-09-13T15:10:59 a1k0n: yea true 2010-09-13T15:11:01 you basically get one from the data that is sent to the canvas thingy.. 2010-09-13T15:11:16 *** jmreardon has joined #aichallenge 2010-09-13T15:11:22 view source on the benzedrine visualizer page 2010-09-13T15:12:04 I see 2010-09-13T15:12:26 on the last turn of a game your bot receives variable called turns_remaining would you excpect it to have a value of one or zero? 2010-09-13T15:12:27 I get too many timeouts, should be a network issue 2010-09-13T15:12:56 Janzert: 1, if the bot is able to perform any actions at all 2010-09-13T15:13:05 you're getting timeouts on tcp? lol 2010-09-13T15:13:29 yes, I am 2010-09-13T15:14:05 a1k0n: I was leaning that way as well, so I'll go with that 2010-09-13T15:16:00 *** Bobng has joined #aichallenge 2010-09-13T15:16:18 *** Theobon has quit IRC (Ping timeout: 252 seconds) 2010-09-13T15:18:08 weird.. I timed out in last 3 games.. started all of a sudden 2010-09-13T15:18:22 *benzedrine.cx 2010-09-13T15:21:44 @tcpserver 2010-09-13T15:21:45 dstufft: tcpserver = http://www.benzedrine.cx/planetwars/ 2010-09-13T15:24:53 *** MorkBork has joined #aichallenge 2010-09-13T15:25:38 are C# submissions compiled with mono or ms csc 2010-09-13T15:26:23 mono 2010-09-13T15:26:31 up to .net 2.0 2010-09-13T15:27:06 *** amstan has quit IRC (Ping timeout: 264 seconds) 2010-09-13T15:27:13 how are they sandboxed 2010-09-13T15:27:30 i.e. any network connectivity? 2010-09-13T15:28:01 wondering if its possible to have a bot keep track of other peoples strategies 2010-09-13T15:28:05 no network, in a VM, limited cpu time and ram 2010-09-13T15:28:14 figured as much 2010-09-13T15:28:23 dhartmei: is there any chance of problem with TCP server? 2010-09-13T15:28:27 i just noticed the no writing to disk, so i wondered if that was an alternative 2010-09-13T15:28:28 *** tapwater has joined #aichallenge 2010-09-13T15:28:55 ratatata: it looks fine, other games play ok 2010-09-13T15:29:10 MorkBork: hardcoding behaviour by your enemies name is forbidden, but if you still intend on using a learning approach, you'll have to train your bot beforehand 2010-09-13T15:29:20 yea not by name 2010-09-13T15:29:27 ratatata: I just played a couple of games fine 2010-09-13T15:29:28 must have been a glitch I guess.. thanks 2010-09-13T15:29:33 im talking about an evolving strategy 2010-09-13T15:29:54 dhartmei: I'm getting timeouts often too 2010-09-13T15:30:19 dwchandler: yes me to.. then I timed out on 3 games out of 4.. hundreds of local games play just fine.. and I test with 300ms limit 2010-09-13T15:32:25 *** amstan has joined #aichallenge 2010-09-13T15:32:25 *** ChanServ sets mode: +o amstan 2010-09-13T15:32:45 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-09-13T15:33:08 *** ajhager has joined #aichallenge 2010-09-13T15:33:14 yo 2010-09-13T15:33:15 *** phreeza has joined #aichallenge 2010-09-13T15:33:55 can you go in the ranks from 47 to 26 without having played a single game or is that wonky? 2010-09-13T15:34:15 rafsoaken, you can 2010-09-13T15:34:24 dhartmei: ok this is weird.. game I actually won resulted with me timing out.. which is not possible.. http://www.benzedrine.cx/planetwars/canvas?game_id=1284404969|vortex.9|ratatata-0.3a 2010-09-13T15:34:50 Palmik: thx, I guess I wont complain then 2010-09-13T15:34:52 ranks are not recalculated after every single game, that and someone before you might have dropped or uploaded new code 2010-09-13T15:35:11 *** jmreardon has quit IRC (Ping timeout: 245 seconds) 2010-09-13T15:35:23 rafsoaken, a wouldn't either :D 2010-09-13T15:35:23 *** Strepto has quit IRC (Ping timeout: 252 seconds) 2010-09-13T15:35:23 sounds reasonable 2010-09-13T15:35:25 *I 2010-09-13T15:36:36 *** amstan_ has joined #aichallenge 2010-09-13T15:36:36 *** ChanServ sets mode: +o amstan_ 2010-09-13T15:36:58 *** amstan has quit IRC (Read error: Connection reset by peer) 2010-09-13T15:37:25 *** num1 has joined #aichallenge 2010-09-13T15:37:41 *** amstan_ is now known as amstan 2010-09-13T15:37:41 *** jmreardon has joined #aichallenge 2010-09-13T15:39:17 *** ntroPi has joined #aichallenge 2010-09-13T15:39:21 hi 2010-09-13T15:39:55 hey 2010-09-13T15:40:18 I just registered (and activated) an account ... but seem to have the password wrong (kinda doubt it). Anyways ... is there a way to reset the PW? 2010-09-13T15:41:05 ntroPi: are you yepugac? 2010-09-13T15:41:10 nope 2010-09-13T15:41:14 sigh.. 2010-09-13T15:41:15 2 ppl now 2010-09-13T15:41:22 there's no password reset feature 2010-09-13T15:41:44 amstan: there isn't? maybe i'll get that worked on 2010-09-13T15:41:52 although im half asleep so prolly not until i wake back up 2010-09-13T15:41:54 what is cameron.jp's handle? 2010-09-13T15:41:57 dstufft: nope.. there isn't 2010-09-13T15:42:00 Bobng: why? 2010-09-13T15:42:20 wanted to discuss something with him, but I guess dstufft will do, he contributed to the file in question 2010-09-13T15:42:25 mind if I pm you dstufft? 2010-09-13T15:42:30 so i guess it will take some time? 2010-09-13T15:42:50 Bobng: you can talk here, as long as it's not something account related 2010-09-13T15:43:02 I'll go for some coding without testing then ^^ 2010-09-13T15:43:05 Oh, ok. Well I have just re-written compile_anything.py 2010-09-13T15:43:09 ntroPi: we have to achieve a handshake by email 2010-09-13T15:43:16 ntroPi: then you send me a md5 of your new password 2010-09-13T15:43:26 great 2010-09-13T15:43:43 its a complete re-write, I got rid of the hundreds of "if" statements for each language. 2010-09-13T15:43:57 if you are a admin in the forum ... my mail there is the same as for the ai account (and the nick is the same too) 2010-09-13T15:44:12 *** deepblue has joined #aichallenge 2010-09-13T15:44:23 Bobng: the only thing ive touched on that file is to replace os.unlink() with an if statement to check for if its a dir and if so use shutil.rmtree 2010-09-13T15:44:39 A language and its compile chain is defined a dictionary like so: http://pastebin.com/6DK11wtA 2010-09-13T15:44:40 hi guys 2010-09-13T15:45:19 The file looks a lot better and easier on the eyes, and the amount of re-used code has been reduced a lot 2010-09-13T15:45:38 *** Theobon has joined #aichallenge 2010-09-13T15:45:44 But it might seem a bit complex at first, I just wanted some reactions I guess 2010-09-13T15:46:02 *** kozlovsky has joined #aichallenge 2010-09-13T15:46:25 ntroPi: check your email 2010-09-13T15:48:50 amstan: thx, replied. 2010-09-13T15:49:27 Hey deepblue 2010-09-13T15:49:42 ntroPi: btw.. your old password started with "22", if you're interested 2010-09-13T15:49:59 passwords arnt encrypted? 2010-09-13T15:50:02 k, should be up 2010-09-13T15:50:22 s/your/the md5 for your 2010-09-13T15:50:39 hmz, think i can get pretty close to being alloc free 2010-09-13T15:51:08 amstan: that replace made my head hurt 2010-09-13T15:51:42 the md5 for your old password? (not feeling good, sorry) 2010-09-13T15:51:49 *** TapewormGalactos has joined #aichallenge 2010-09-13T15:52:11 Cyndre: As in "the md5 for your old password started with..." 2010-09-13T15:52:13 so, I kept making edits and running tcp, and having MAJOR trouble understanding how my bot was (wasn't) working. I was editing a different version the whole time 2010-09-13T15:52:22 * dwchandler <-- ai-challenged 2010-09-13T15:52:42 yea, got it now ermau|work, just messed up somehow 2010-09-13T15:52:59 * Cyndre comforts dwchandler 2010-09-13T15:53:09 I did that last night 2010-09-13T15:53:13 *** td123 has quit IRC (Quit: WeeChat 0.3.3) 2010-09-13T15:53:51 well the other night - I was working on a new bot called MyBotNew.py and nothing I was doing was working - then I realized I was happy enough it became MyBot.py sometime at 4 in the morning.... 2010-09-13T15:54:29 2 more timeouts on tcp.. my score is screwd big time too :) anyone else unusually timing out on TCP? 2010-09-13T15:54:55 Cyndre: heh. at least I'm not alone :) 2010-09-13T15:55:06 ratatata: what part of the world are you coming from? 2010-09-13T15:55:47 *** mistabell has joined #aichallenge 2010-09-13T15:56:16 dwchandler: eastern europe 2010-09-13T15:57:01 * deepblue wants to brag 2010-09-13T15:57:04 @rankings 2010-09-13T15:57:05 deepblue: Top 10 players: deepblue(2727), Kavish(2695), a1k0n(2667), sighbot(2652), zuko3d(2638), wtfbbq(2628), braddunbar(2597), Tracer(2590), Loks1982(2587), dmj111(2586) 2010-09-13T15:57:16 :) 2010-09-13T15:57:32 bleh, all your wins are against randombots 2010-09-13T15:57:34 gz :) I would totally gratz you with emote but I dont know how :) 2010-09-13T15:57:40 yeah 2010-09-13T15:57:46 no real opponent so far 2010-09-13T15:57:50 hope the servers like my resource reduced bot 2010-09-13T15:58:12 hey, montanalow is back 2010-09-13T15:58:33 ratatata: did you see our game? It's the first time I see two bots agreeing on a border :) 2010-09-13T15:58:42 and wtfbbq must have resubmitted which is why i'm in 3rd with a bunch of randombot wins. huh. 2010-09-13T15:59:16 Is the timeout still 1000 ? I see some games I lost where no one made a move 2010-09-13T15:59:33 well if bartwe.7 is bugfree it should also attack the top 10 2010-09-13T15:59:37 retybok: server said that I timed out on that :S 2010-09-13T16:00:18 I actually received time-out response on a game I won.. so .. I don't know :) 2010-09-13T16:00:34 http://ai-contest.com/profile_games.php?user_id=3885 2010-09-13T16:00:39 all losses but the first 2 no one moved 2010-09-13T16:00:46 ratatata: yeah but still, look at the game before the time out 2010-09-13T16:00:49 i like how there's no name 2010-09-13T16:01:08 amstan: yeah, that's a nice touch 2010-09-13T16:01:20 retybok: yeah, just did.. it's awesome! :) 2010-09-13T16:02:23 I swear to god if a bot timesout player 1 auto wins in tourny endine 2010-09-13T16:02:25 er engine 2010-09-13T16:02:42 I had the same problem mistabell 2010-09-13T16:02:44 that's what I'm wondering 2010-09-13T16:02:50 I don't think mine is timing out 2010-09-13T16:02:51 it has to be 2010-09-13T16:03:24 (it beats all the default bots 100/100 without errors or timeouts) I don't see why it would not make a first move 2010-09-13T16:03:49 unless maybe it's right on that 1 second limit 2010-09-13T16:04:27 also.. if no one moves, seems like it would be a draw, unless I'm making an illegal move. But tricky without some sort of logging. 2010-09-13T16:04:43 ratatata: :) for anyone interested: http://www.benzedrine.cx/planetwars/canvas?game_id=1284407096|ratatata-0.3a|retybok 2010-09-13T16:04:47 website should have a reason for ending match 2010-09-13T16:04:57 winner + wiinner code 2010-09-13T16:05:11 retybok: we are of peace.. always 2010-09-13T16:05:21 if 0 player won normally, if 1 other player timed out, if 2 other player illegal move 2010-09-13T16:05:22 very simple 2010-09-13T16:05:28 retybok: lol 2010-09-13T16:07:33 *** ratatata has quit IRC (Quit: Page closed) 2010-09-13T16:07:53 * num1 is figuring out how to adapt Chess AI algorithms to PlanetWars 2010-09-13T16:08:06 *** anoek has quit IRC (Quit: Leaving) 2010-09-13T16:08:20 *** bss03-AFK is now known as bss03 2010-09-13T16:08:22 you know some parts of the future, like when fleets will arive, but you don't know others, it's an interesting problem 2010-09-13T16:08:55 heh, I just got connect: Connection timed out error with tcp 2010-09-13T16:09:06 Crap on a stick. My modification that enables reinforcing planets made my bot worse! 2010-09-13T16:09:22 *** anoek has joined #aichallenge 2010-09-13T16:09:35 that's a prove that it's a network issue 2010-09-13T16:11:18 deepblue, I'm impressed by your game with buddhabrot (http://ai-contest.com/visualizer.php?game_id=4455309#) At one part you're covering your own ass when he tries to take your planet 2010-09-13T16:12:22 *** sinan has joined #aichallenge 2010-09-13T16:12:47 *** ingdas has joined #aichallenge 2010-09-13T16:13:45 *** ratatata has joined #aichallenge 2010-09-13T16:14:07 that's a nice game I agree :) 2010-09-13T16:14:25 quick :D 2010-09-13T16:15:26 i like the trade at the end 2010-09-13T16:17:46 *** Theobon has quit IRC (Ping timeout: 276 seconds) 2010-09-13T16:17:59 site down again.. 2010-09-13T16:18:02 btw, patch is now up to send bots the turns remaining http://code.google.com/p/ai-contest/issues/detail?id=74 2010-09-13T16:18:11 if someone could take a look at it would be great 2010-09-13T16:19:18 why not just send "max turns" in the first turn? 2010-09-13T16:20:02 it's the same 2010-09-13T16:20:15 in the first turn you get max turns 2010-09-13T16:20:22 sending turns remaining accomplishes the same thing and keeps the property that the full global state is sent every turn so bots do not need to incrementally update if they don't want 2010-09-13T16:20:45 Well, whatever :) 2010-09-13T16:21:27 *shrug* it just seemed more in line with the rest of the protocol is all :) 2010-09-13T16:22:16 I like the property that I don't have to worry about messing up keeping track of state :) 2010-09-13T16:22:17 Yeah, that's true :) 2010-09-13T16:23:02 Hmm, I do not see how "T " would be backward incompatible, at least C++ starter's package did just parse those lines starting with "P" or "F" 2010-09-13T16:23:08 how quickly are patches usually reflected on the serveR? 2010-09-13T16:23:08 http://www.benzedrine.cx/planetwars/canvas?game_id=1284408449|deepblue|deepblue-test 2010-09-13T16:23:32 how about go 2010-09-13T16:23:46 most of the example bots will handel that 2010-09-13T16:23:56 I mean 2010-09-13T16:24:02 go whatever 2010-09-13T16:24:06 take a look the python start package. it'll crash on an unexpected line 2010-09-13T16:24:14 Janzert, I see :/ 2010-09-13T16:24:49 "go" is allready a whole line--what extra line? 2010-09-13T16:25:09 I was replying to the T suggestion 2010-09-13T16:25:15 sorry 2010-09-13T16:26:47 I like the idea of # : key value though, it leave room for future extra values 2010-09-13T16:26:52 *** amstan has quit IRC (Ping timeout: 276 seconds) 2010-09-13T16:26:54 yeah, go might have worked fine as well, but it seems more risky since it is a incompatible change. 2010-09-13T16:26:58 s/values/keys/ 2010-09-13T16:27:45 Is this the command you would use for the tcp.c -> ./tcp 213.3.30.106 9999 mistabell "java MyBot.java" ? 2010-09-13T16:27:48 the protocol already specifies comments so any bot that has a problem with it can be said to be out of spec :) 2010-09-13T16:27:53 I can't get it to see my bot 2010-09-13T16:28:01 mistabell: no 2010-09-13T16:28:27 you need to wrap the "java -cp classpath MyBot" in a script 2010-09-13T16:28:41 and compile first 2010-09-13T16:28:43 I can my script AI2000 2010-09-13T16:28:54 ah, ok. thanks delt0r 2010-09-13T16:29:03 yea that too --compile tcp.c 2010-09-13T16:29:21 yeah, it's compiled but I wasn't sure if a wrapper was necessary 2010-09-13T16:29:45 *** amstan has joined #aichallenge 2010-09-13T16:29:45 *** ChanServ sets mode: +o amstan 2010-09-13T16:30:01 * Janzert as refering to the bot actually, since you referred to a .java file 2010-09-13T16:30:09 Hmm, it's a shame fleets are not sorted by turns remaining in the format, have to do it manually :D 2010-09-13T16:30:12 the wrapper keeps the paths etc sane 2010-09-13T16:30:50 *** ermau|work has quit IRC (Quit: ermau|work) 2010-09-13T16:30:56 where if you pass "java MyBot" to the c --when it runs the process the paths are all borked 2010-09-13T16:31:30 Palmik: They are always in the order they where created in 2010-09-13T16:31:35 *** Accoun has joined #aichallenge 2010-09-13T16:31:43 delt0r_, yeah :) 2010-09-13T16:32:02 well for every gameserver i have seen... a gameserver somewhere may not do this 2010-09-13T16:32:15 forum is working? 2010-09-13T16:32:33 *** Theobon has joined #aichallenge 2010-09-13T16:32:48 *** amstan has quit IRC (Client Quit) 2010-09-13T16:32:53 blank page on forum link http://ai-contest.com/forum/ 2010-09-13T16:32:55 Accoun: it is for me 2010-09-13T16:33:03 *** amstan has joined #aichallenge 2010-09-13T16:33:03 *** ChanServ sets mode: +o amstan 2010-09-13T16:33:19 *** Ported has quit IRC (Ping timeout: 252 seconds) 2010-09-13T16:34:08 *** hornairs has joined #aichallenge 2010-09-13T16:35:06 game viewer dont worck 2010-09-13T16:35:09 Ah, hrm. DualBot wins 2/100 at 200 turns. 0/100 at 1000 turns. 2010-09-13T16:35:20 I'll take that as I always actually beat DualBot. 2010-09-13T16:35:34 Still, against RageBot I'm back down to 45/100. :( 2010-09-13T16:36:00 it's Accoun! 2010-09-13T16:36:08 Have anyone tested the rage bot on tcpserver or the official one? How did he manage? 2010-09-13T16:36:11 Accoun: zomg, welcome back 2010-09-13T16:36:56 Accoun: how so? 2010-09-13T16:37:34 bss03: Is there a script or tool somewhere that plays two bots on all the maps and gives the results? 2010-09-13T16:37:46 theobon, it's on the forum 2010-09-13T16:37:53 Theobon, there is... search for "batch" 2010-09-13T16:37:58 Thanks. 2010-09-13T16:38:18 Theobon: I have my own script. 2010-09-13T16:38:20 Theobon: http://gist.github.com/574904 2010-09-13T16:38:57 Palmik: I think Rage and Dual are both always in the official mix. 2010-09-13T16:39:20 Plamik: Something like StandardBot_Dual and StandardBot_Rage are their "usernames". 2010-09-13T16:39:32 bss03, ok.. will look them up 2010-09-13T16:39:54 *** mega1 has quit IRC (Ping timeout: 255 seconds) 2010-09-13T16:39:56 are the tcp.c games 1000 200 ? 2010-09-13T16:40:10 (it seems to take a long time to play after the map is chosen and the opponent is listed.) 2010-09-13T16:40:20 DualBot: http://ai-contest.com/profile.php?user_id=3930 2010-09-13T16:40:44 http://ai-contest.com/profile.php?user_id=3928 here rage 2010-09-13T16:41:47 I think those are the only two StarterPack bots in the official competition. 2010-09-13T16:41:49 Not as goos as I would expect hit to be :) 2010-09-13T16:42:02 *good, him :/ 2010-09-13T16:43:21 I mean, even some better custom bots have problems beating him. 2010-09-13T16:45:31 *** ntroPi has left #aichallenge 2010-09-13T16:46:42 *** malumalu has quit IRC (Quit: Lost terminal) 2010-09-13T16:50:21 *** mistabell has quit IRC (Quit: Page closed) 2010-09-13T16:53:29 *** ratatata has quit IRC (Ping timeout: 252 seconds) 2010-09-13T16:54:23 http://code.google.com/p/ai-contest/issues/detail?id=130 2010-09-13T16:54:29 can anyone review my patch? 2010-09-13T16:55:00 *** Accoun has quit IRC () 2010-09-13T16:57:47 *** ratatata has joined #aichallenge 2010-09-13T16:58:55 *** jaspervdj has quit IRC (Quit: leaving) 2010-09-13T17:00:18 *** td123 has joined #aichallenge 2010-09-13T17:01:41 Bobng: ExternalCompiler.compile has 'out, err = ""' but i presume you want 'out, err = "", ""' 2010-09-13T17:01:52 *** jmreardon has quit IRC (Ping timeout: 245 seconds) 2010-09-13T17:02:06 *** Vio has quit IRC (Read error: Connection reset by peer) 2010-09-13T17:02:51 i also recommend having the nuke_glob loop not all on one line in compile_function, but that's just style preference 2010-09-13T17:03:32 *** Accoun has joined #aichallenge 2010-09-13T17:06:43 *** Kastra has quit IRC (Ping timeout: 276 seconds) 2010-09-13T17:11:07 *** sinan has quit IRC (Ping timeout: 276 seconds) 2010-09-13T17:16:13 *** ermau has joined #aichallenge 2010-09-13T17:16:51 *** s_denchev has joined #aichallenge 2010-09-13T17:17:00 would someone here be willing to help with a bit of java debugging? 2010-09-13T17:17:47 *** ermau has quit IRC (Client Quit) 2010-09-13T17:18:31 rogue780: what is going wrong? 2010-09-13T17:20:59 rafsoaken, I have this code in a method :http://pastebin.ca/1940019 2010-09-13T17:21:14 *** ermau has joined #aichallenge 2010-09-13T17:21:48 it doesn't iterate through that forloop nor does it send any orders 2010-09-13T17:22:57 *** Mathnerd314 has joined #aichallenge 2010-09-13T17:23:28 *** s_denchev has quit IRC (Quit: Page closed) 2010-09-13T17:23:57 *** s_denchev has joined #aichallenge 2010-09-13T17:25:45 rafsoaken, any idea? 2010-09-13T17:26:33 *** amstan has quit IRC (Ping timeout: 252 seconds) 2010-09-13T17:26:34 well if it does not iterate, then it is clear that it wont send any orders 2010-09-13T17:26:51 *** CyberjujuM has quit IRC (Ping timeout: 252 seconds) 2010-09-13T17:27:17 *** mark_jn has quit IRC (Quit: leaving) 2010-09-13T17:28:00 so what do you get if you print pw.NotMyPlanets().size() to stderr? 2010-09-13T17:28:23 memory location of that list's reference :) 2010-09-13T17:29:24 *my bad, didn't see .size() 2010-09-13T17:29:31 *** drewrobb_ has quit IRC (Ping timeout: 264 seconds) 2010-09-13T17:29:40 :) 2010-09-13T17:29:48 *** jmreardon has joined #aichallenge 2010-09-13T17:31:58 *** ippeb24 has joined #aichallenge 2010-09-13T17:32:39 *** sigh has joined #aichallenge 2010-09-13T17:32:40 lol http://www.benzedrine.cx/planetwars/canvas?game_id=1284412490|rafsaoken|homeowner 2010-09-13T17:32:52 *** s_denchev has quit IRC () 2010-09-13T17:33:09 @rafsoaken nice curve though 2010-09-13T17:33:10 ippeb24: Error: "rafsoaken" is not a valid command. 2010-09-13T17:33:21 nice curve though xD 2010-09-13T17:33:41 *** Zeiris has joined #aichallenge 2010-09-13T17:34:17 rafsoaken, the size is consistantly 22 2010-09-13T17:34:30 sounds about right 2010-09-13T17:34:45 rogue780: what calls that method? 2010-09-13T17:35:00 (is it being called?) 2010-09-13T17:35:32 *** Palmik has quit IRC (Remote host closed the connection) 2010-09-13T17:35:40 Zannick, yes. It is being called 2010-09-13T17:35:48 *** ippeb24 has quit IRC (Client Quit) 2010-09-13T17:36:31 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-09-13T17:37:03 I'm tempted to just put my entire bot source code up. I'm getting fed up 2010-09-13T17:37:18 rogue780: put a vanilla for loop in there and check whether it iterates, somthing is fishy 2010-09-13T17:37:35 what is IssueOrder? 2010-09-13T17:37:40 Does it take Planets or ints? 2010-09-13T17:37:58 Zannick: both 2010-09-13T17:38:10 2 methods 2010-09-13T17:38:10 ah, it's overloaded? 2010-09-13T17:38:13 ok 2010-09-13T17:38:19 if not you should do both anyway :) 2010-09-13T17:38:41 Distance takes only id iirc :) 2010-09-13T17:38:55 that's stupid :) 2010-09-13T17:39:01 check that findWeakest and findStrongest are actually returning values? 2010-09-13T17:39:43 Zannick: he said it is not even iterating 2010-09-13T17:40:28 yes, hence the "is it called" question 2010-09-13T17:41:17 this snippet does not show any other reason it wouldn't iterate, so i think about why it isn't issuing orders 2010-09-13T17:43:07 Zannick, those methods are returning a planet object 2010-09-13T17:43:13 i verified it 2010-09-13T17:43:59 i am guessing that your problem lies outside this snippet 2010-09-13T17:44:09 So what are the thoughts on relying on the current fleet order in the input? should i do that? 2010-09-13T17:44:18 I already do for planets 2010-09-13T17:44:57 *** eburnette has joined #aichallenge 2010-09-13T17:46:00 so you should look at what calls that method, make sure that is called, etc. 2010-09-13T17:46:11 make sure that something isn't making it exit the method early 2010-09-13T17:46:36 How long is it supposed to take from when I upload code until when I see game results? 2010-09-13T17:46:49 eburnette, have you tried the tcp client? 2010-09-13T17:47:05 *** deepblue has quit IRC (Quit: Page closed) 2010-09-13T17:47:12 No, just found that. Anybody have a java version or do I need to write one? 2010-09-13T17:47:15 *** TapewormGalactos has quit IRC (Quit: Ex-Chat) 2010-09-13T17:47:24 eburnette, do you run windows or something else? 2010-09-13T17:47:29 windows 2010-09-13T17:47:39 eburnette: someone wrote a java client in the msgboard somewhere 2010-09-13T17:47:47 go to www.errgaming.com and download tcp.zip 2010-09-13T17:47:47 there was a java code in the forums, yes 2010-09-13T17:47:54 it has everything compiled to run on windows 2010-09-13T17:49:48 eburnette, I use it to run my java based bot 2010-09-13T17:49:54 when my bot is working, that is 2010-09-13T17:52:42 Zannick, would you be willing to look at my whole bot? 2010-09-13T17:53:09 eburnette: http://ai-contest.com/forum/viewtopic.php?f=18&t=424#p2616 2010-09-13T17:53:51 sure 2010-09-13T17:53:53 thanks i'll try it 2010-09-13T17:54:05 *** eburnette has quit IRC (Quit: Page closed) 2010-09-13T17:56:11 *** Appleman1234 has joined #aichallenge 2010-09-13T17:56:26 *** Heckel has joined #aichallenge 2010-09-13T17:57:13 *** Tegid has joined #aichallenge 2010-09-13T17:57:23 *** Heckel has quit IRC (Remote host closed the connection) 2010-09-13T17:57:38 Zannick, did you get my pm? 2010-09-13T17:57:43 yes 2010-09-13T17:58:03 *** sigh has quit IRC (Remote host closed the connection) 2010-09-13T18:00:54 *** mightybyte has quit IRC (Quit: leaving) 2010-09-13T18:03:57 *** jmreardon has quit IRC (Ping timeout: 245 seconds) 2010-09-13T18:06:02 *** Bobng has quit IRC (Ping timeout: 265 seconds) 2010-09-13T18:06:57 *** ingdas has quit IRC (Quit: Page closed) 2010-09-13T18:07:21 Fixed what I thought was a bug and my wins against RageBot whent from 45% to 5%. :( 2010-09-13T18:07:48 unofficial go starter kit: http://github.com/ckolbeck/Waterloo-AI-Contest---Go-Starter-Kit cbeck++ 2010-09-13T18:07:55 *** amstan has joined #aichallenge 2010-09-13T18:07:55 *** ChanServ sets mode: +o amstan 2010-09-13T18:08:09 bss03: you seem to be in a lucky spree :P 2010-09-13T18:09:38 I'm sure the bot would be better if I'd bite the bullet and wrote something to actually resolve all the fleets currently in the air. 2010-09-13T18:09:56 bss03: I feel the pain... Happened to me more than once. 2010-09-13T18:10:02 Right now, I'm going the lazy way and just summing them by destination, without regard to clock ticks. 2010-09-13T18:10:37 *** mightybyte has joined #aichallenge 2010-09-13T18:10:39 bss03: I took me two hours to implement the same thing, I don't regret it one second 2010-09-13T18:11:20 bss03: So far that's all I'm doing at it works great when playing against the example bots 2010-09-13T18:11:37 * RainCT goes to implement some "which planet is better" rating now 2010-09-13T18:13:33 bss03: the method is in some python unofficial starter kit 2010-09-13T18:15:34 http://github.com/simu/planetwars-python-kit/blob/master/planetwars/planet2.py 2010-09-13T18:16:21 rafsoaken: I've got some haskell code somewhere in my git tree that does it. 2010-09-13T18:16:44 rafsoaken: I just haven't merged and debugged it. 2010-09-13T18:16:55 I keep battling my bot on the TCP server even though I know it's going to lose because I don't yet have time to improve it 2010-09-13T18:16:58 @seen accoun 2010-09-13T18:16:58 amstan: accoun was last seen in #aichallenge 1 hour, 41 minutes, and 51 seconds ago: game viewer dont worck 2010-09-13T18:17:16 @quotegrabber accoun 2010-09-13T18:17:17 amstan: Error: "quotegrabber" is not a valid command. 2010-09-13T18:17:22 @grab accoun 2010-09-13T18:17:22 amstan: The operation succeeded. 2010-09-13T18:17:28 @qoute accoun 2010-09-13T18:17:29 amstan: Error: "qoute" is not a valid command. 2010-09-13T18:17:33 @quote accoun 2010-09-13T18:17:33 amstan: game viewer dont worck 2010-09-13T18:17:38 heheh 2010-09-13T18:17:38 *** Ported has joined #aichallenge 2010-09-13T18:18:17 javagamer_Cpp: And we thank you for that. 2010-09-13T18:18:25 heh 2010-09-13T18:18:27 does anyone else have trouble compiling the vis directory from http://code.google.com/p/ai-contest 2010-09-13T18:18:34 javagamer_Cpp: More game we play, the better we refine our strategies. 2010-09-13T18:18:35 @seen j3camero 2010-09-13T18:18:35 blkperl: j3camero was last seen in #aichallenge 1 day, 1 hour, 31 minutes, and 21 seconds ago: jmcarthur: when the site is up... 2010-09-13T18:19:01 Ported: you mean the java visualizer? why do you need it compiled? 2010-09-13T18:20:01 @help 2010-09-13T18:20:01 pgpaskar_: (help [] []) -- This command gives a useful description of what does. is only necessary if the command is in more than one plugin. 2010-09-13T18:20:04 bss03, my bot doesn't have what I'd really consider a strategy, still trying to best organize the data so I have optimal access to it 2010-09-13T18:20:09 * pgpaskar_ slaps contestbot 2010-09-13T18:20:48 i want to rewrite a version of the engine so i can run different versions of my bots against eachother without having to start a new jvm for every command 2010-09-13T18:21:02 *** Theobon has quit IRC (Ping timeout: 245 seconds) 2010-09-13T18:21:14 Ported: I am already doing that 2010-09-13T18:21:18 @amstan:i dont need the visualizer just the engine code 2010-09-13T18:21:19 Ported: Error: "amstan:i" is not a valid command. 2010-09-13T18:21:42 delt0r: you have some code somewhere? 2010-09-13T18:21:46 Though i don't know when it will be finished. It will be BSD/apache/gpl take your pick 2010-09-13T18:21:58 Ported: i see, can't help you there, sorry, lol 2010-09-13T18:22:08 Ported: prefer to relase when finsihed 2010-09-13T18:22:14 dam my spelling is bad 2010-09-13T18:22:28 delt0r_: no worries 2010-09-13T18:22:30 sigh.. 2010-09-13T18:22:46 Ported: Keep in mind that you may get the wrong idea with performance on timeouts 2010-09-13T18:22:50 delt0r_: did you start from the version on google code?\ 2010-09-13T18:22:55 no 2010-09-13T18:23:19 Its not a showGa,e 2010-09-13T18:23:24 showGame 2010-09-13T18:23:34 Its a game player and show game in one 2010-09-13T18:23:36 delt0r_: yeah, its more just for testing a bunch of things without having to wait... 2010-09-13T18:23:46 like what i want 2010-09-13T18:24:01 and it has a "play" tcp button 2010-09-13T18:24:29 and you get to watch real time 2010-09-13T18:24:41 wow, my bot won a game, someone should be worried 2010-09-13T18:24:43 or go gui less and do the command line thing 2010-09-13T18:25:13 delt0r_ sounds good 2010-09-13T18:25:31 Well give me a few days... I have a day job /: 2010-09-13T18:25:50 And i stand on my head :/ 2010-09-13T18:27:18 delt0r: me too, very inconvenient at times! 2010-09-13T18:27:48 there's already this: http://github.com/albertz/planet_wars-cpp 2010-09-13T18:28:08 http://www.benzedrine.cx/planetwars/canvas?game_id=1284416218|real-1-1-1|ErrBot_0.0.3 2010-09-13T18:28:26 ErrBot is mine 2010-09-13T18:28:53 Hey, I just won against that guy too, interesting looking strategy, but it doesn't seem all to effective 2010-09-13T18:29:12 Wow, you made quite a comeback 2010-09-13T18:30:23 actually, looks like he never had a higher growth rate, he just seemed so intimidating w/ all those fleets everywhere 2010-09-13T18:30:30 indeed 2010-09-13T18:30:41 i just beat him again too 2010-09-13T18:30:52 http://www.benzedrine.cx/planetwars/getplayer?player=real-1-1-1 2010-09-13T18:30:59 I think everyone is beating him right now. 2010-09-13T18:31:19 I want to play against bartwe again 2010-09-13T18:31:44 i will set copybot running again on the tcp server later tonight 2010-09-13T18:31:54 bartwe us kind of like my nemesis 2010-09-13T18:32:00 *** erh has joined #aichallenge 2010-09-13T18:32:04 *** erh is now known as blister 2010-09-13T18:32:37 *** Mathnerd314 has quit IRC (Disconnected by services) 2010-09-13T18:32:51 How was http://www.benzedrine.cx/planetwars/canvas?game_id=1284416233|real-1-1-1|rogerbraun a draw? 2010-09-13T18:32:56 *** biffabacon has joined #aichallenge 2010-09-13T18:33:27 *** Mathnerd314_ has joined #aichallenge 2010-09-13T18:33:29 *** Mathnerd314_ is now known as Mathnerd314 2010-09-13T18:33:37 *** AJC has left #aichallenge 2010-09-13T18:33:47 Ported: Oh yea, its also threaded so you can have more than one game at a time if you have a "n core" machine. 2010-09-13T18:33:50 they would have had to both timeout on the same turn 2010-09-13T18:33:54 I got one 32 core machine 2010-09-13T18:34:18 *** biffabacon has quit IRC (Client Quit) 2010-09-13T18:34:22 the cluster is over 500 cores :D 2010-09-13T18:34:29 *** GordonQuad has quit IRC (Read error: Operation timed out) 2010-09-13T18:34:32 *** biffabacon has joined #aichallenge 2010-09-13T18:34:34 thats a lot of games an hour 2010-09-13T18:35:05 *** drewrobb has joined #aichallenge 2010-09-13T18:35:34 *** Theobon has joined #aichallenge 2010-09-13T18:35:43 Naktibalda, are you running your php bot right now? 2010-09-13T18:35:44 javagamer_Cpp: it happens a lot 2010-09-13T18:35:54 yes, I am 2010-09-13T18:35:57 ratatata, do you know why? 2010-09-13T18:36:02 gah! I just lost to a PHP bot 2010-09-13T18:36:13 shame on you 2010-09-13T18:36:19 rogue780: ? why is that bad? 2010-09-13T18:36:34 php != general purpose langauge 2010-09-13T18:36:35 javagamer_Cpp: no clue 2010-09-13T18:36:52 oh PHP 2010-09-13T18:36:56 I read python 2010-09-13T18:37:02 my bad 2010-09-13T18:37:22 Naktibalda, I admire your moxy 2010-09-13T18:37:49 moxy? 2010-09-13T18:37:53 Naktibalda: next you should do a java script one 2010-09-13T18:38:01 or even better --Hypercard 2010-09-13T18:38:05 basic? 2010-09-13T18:38:08 http://www.benzedrine.cx/planetwars/canvas?game_id=1284416675|Naktibalda-002|ErrBot_0.0.3 2010-09-13T18:38:10 brainfuck 2010-09-13T18:38:17 malbolge 2010-09-13T18:38:34 zerd: ook is better 2010-09-13T18:40:27 nah.. I'm oldschool :p 2010-09-13T18:40:41 ook 2010-09-13T18:40:47 on off key? 2010-09-13T18:40:54 Ook. Ook? 2010-09-13T18:41:25 rogue780: what's a purpose of PHP? 2010-09-13T18:41:43 http://www.dangermouse.net/esoteric/ook.html 2010-09-13T18:41:52 for rogue780 2010-09-13T18:42:14 Naktibalda, Well, seeing as it stands for either Personal Home Page or PHP Hypertext Preprocessor, I'm going to go with it is either for making home pages, or preprocessing hypertext 2010-09-13T18:42:23 Win with that and i will buy you whisky 2010-09-13T18:42:26 Naktibalda, I hath defeated thee 2010-09-13T18:42:36 my bot timeouted 2010-09-13T18:42:46 but only because of a timeout I see 2010-09-13T18:42:46 must be my wifi 2010-09-13T18:43:10 Naktibalda: you are doing a TCP php bot over wifi! 2010-09-13T18:43:29 lame. timeout after 4 turns 2010-09-13T18:43:30 I'm doing that right now 2010-09-13T18:43:40 tcp bot over wifi 2010-09-13T18:43:52 at a busy university campus 2010-09-13T18:43:56 *** danielvf has left #aichallenge 2010-09-13T18:44:11 how come I get picked to play real-1-1-1 so often :/ 2010-09-13T18:44:32 dwchandler: And you have no problems? Wifi is fine for browsing for me... but you know a game of quake suxs 2010-09-13T18:44:38 I'm playing real-1-1-1 a lot, too. multiple bots at once? 2010-09-13T18:45:16 delt0r_: it's ok. I'm not pushing 1 sec for my turn, so a little lag doesn't matter 2010-09-13T18:45:25 well.. I figured connected usernames are unique.. so.. max 1 2010-09-13T18:45:40 i don't think so 2010-09-13T18:45:55 I'm playing against real right now 2010-09-13T18:46:55 woot 2010-09-13T18:47:01 i just won my first game (i think) 2010-09-13T18:47:20 13/40 of my games are real-1-1-1 2010-09-13T18:47:40 whats real-1-1-1? 2010-09-13T18:48:08 i've played 8/30 against real-1-1-1 2010-09-13T18:48:16 pretty sure it's multiple bost 2010-09-13T18:48:18 bots 2010-09-13T18:48:34 lol, he has bidirectional streams 2010-09-13T18:48:58 just won another against real.. lol 2010-09-13T18:49:32 *** caytchen has quit IRC (Ping timeout: 265 seconds) 2010-09-13T18:49:33 *** drewrobb has quit IRC (Ping timeout: 272 seconds) 2010-09-13T18:50:25 *** jmreardon has joined #aichallenge 2010-09-13T18:50:26 Utkarsh: what do you mean by real ? 2010-09-13T18:50:35 real-1-1-1 the bot 2010-09-13T18:50:57 Oh 2010-09-13T18:51:14 * delt0r_ hits his head on the wall 2010-09-13T18:51:30 1am brain fart 15min early 2010-09-13T18:51:49 heh 2010-09-13T18:52:17 well, i just beat real-l-l-l 2010-09-13T18:52:19 so woot 2010-09-13T18:52:54 real bot sucks, I want to play against other bots 2010-09-13T18:53:18 Naktibalda: same 2010-09-13T18:53:40 Its a shame that the tcp server is not ... more full 2010-09-13T18:53:51 I get matched against real pretty often, but mixed with others 2010-09-13T18:53:59 if 'oppo' number is some sort of rating for past opponent quality.. I'm the highest in rank with negative rating .. all thanks to real-1-1-1 :D 2010-09-13T18:55:10 I wonder what real-1-1-1's owner is trying to achieve. 2010-09-13T18:55:46 meow__, get the lowest ranked bot? 2010-09-13T18:56:04 mewo2: global warming through excess CO2 production ? 2010-09-13T18:56:19 *** Queue29 has joined #aichallenge 2010-09-13T18:56:25 smremde has the same naming convention as real 2010-09-13T18:56:31 *** rogue780__ has joined #aichallenge 2010-09-13T18:57:17 ratatata: which is your bot? 2010-09-13T18:57:43 dwchandler: same name? 2010-09-13T18:58:22 *** rogue780_ has quit IRC (Ping timeout: 240 seconds) 2010-09-13T18:58:40 dwchandler: call me paranoid but I quickly looked through smremde s.. and none has ever played real-x-x-x.. what are the odds? I alone played it 15 times 2010-09-13T18:58:41 ah, since you said neg ranking I was looking too low 2010-09-13T18:58:57 nah, highes rank with negative oppo rating :d 2010-09-13T19:00:31 *** drewrobb has joined #aichallenge 2010-09-13T19:00:50 ratatata: compare game time 2010-09-13T19:01:58 *** sigh has joined #aichallenge 2010-09-13T19:02:35 Naktibalda: what do you mean? 2010-09-13T19:03:08 maybe smremde played a few hours ago 2010-09-13T19:03:36 maybe switched names at that time 2010-09-13T19:03:50 connect: Connection timed out is a little bit better than timeout during game 2010-09-13T19:04:01 but it sucks still :( 2010-09-13T19:04:06 *** syntaxglitch has joined #aichallenge 2010-09-13T19:04:12 just played 5 in a row with real 2010-09-13T19:05:35 *** num1 has quit IRC (Read error: Connection reset by peer) 2010-09-13T19:05:44 If anyone is interested in writing bots in Go, I wrote a starter package before I realized that I'd have no time. The starter is at http://github.com/ckolbeck/Waterloo-AI-Contest---Go-Starter-Kit if anyone cares to pick it up. 2010-09-13T19:05:47 Hah, a few days ago I had been a little discouraged when some of my initial strategy ideas didn't do well. I didn't do much with it until now, and I just discovered that my whole strategy had been bracketed with if ( myFleetsInFlight.size() == 0 ) { ...do strategy... } 2010-09-13T19:06:53 haha. so you were sending pulses instead of streams 2010-09-13T19:06:56 :) 2010-09-13T19:06:57 Yeah 2010-09-13T19:07:27 I had noticed the pulse behavior before, but thought it might have been an emergent property of the fairly simple strategy idea I was using. 2010-09-13T19:07:55 this one was touch and go for a while http://www.benzedrine.cx/planetwars/canvas?game_id=1284417986|infthi|dwc-003 2010-09-13T19:08:36 cbeck: Have you started on your bot? 2010-09-13T19:10:51 *** rogue780__ has quit IRC (Quit: Bye) 2010-09-13T19:18:45 wow, im just about sick and tired of playing real-1-1-1 2010-09-13T19:19:13 *** dmj111 has joined #aichallenge 2010-09-13T19:19:46 improve your bot 2010-09-13T19:21:29 It doesn't seem to have much to do with how good your bot is. 2010-09-13T19:23:31 on tcp is has more to do with who's connected and ready to play 2010-09-13T19:24:11 Right, as soon as two bots are connected and free, they'll pair up and play. 2010-09-13T19:24:16 *** hornairs has quit IRC (Quit: hornairs) 2010-09-13T19:24:30 And at the moment, there seem to be a lot of real-1-1-1s connecting. 2010-09-13T19:24:46 there are a lot 2010-09-13T19:25:02 * ulope is struggling with graphs... 2010-09-13T19:25:14 real-1-1-1 doesn't seem to be very good though. :( 2010-09-13T19:25:40 hah. and yet real-1-1-1 never plays against himself 2010-09-13T19:26:01 I guess the server is smart enough to avoid that? 2010-09-13T19:27:05 It would probably break the Elo system. 2010-09-13T19:27:12 *** DirtyKeyboard has joined #aichallenge 2010-09-13T19:27:26 it does. very much. i was just playing with that. 2010-09-13T19:28:32 I play 15games/h .. real-1-1-1 plays that amount in 1 minute :D 2010-09-13T19:29:38 is ELO supposed to award 'loosing against high rank' more than 'winning against low rank' ? 2010-09-13T19:32:22 wow.. 2010-09-13T19:32:55 played 2 actual games in between all that, and actually made a useful observation 2010-09-13T19:33:35 *** yowaken has joined #aichallenge 2010-09-13T19:34:58 *** sr3d has joined #aichallenge 2010-09-13T19:35:32 Grr. 2010-09-13T19:35:41 I just drew with real because the server send bad information. 2010-09-13T19:35:53 I got a line that was "F 2go". :P 2010-09-13T19:36:41 *** delt0r___ has joined #aichallenge 2010-09-13T19:36:54 I wish I could play on the tcp server without timing out :( 2010-09-13T19:37:27 sad. I can't view any matches on the tcp server att 2010-09-13T19:37:30 I think we have to create petition to lift timeout limit up to 10 seconds (just in case) 2010-09-13T19:37:43 10? D: 2010-09-13T19:37:50 *** sr3d has left #aichallenge 2010-09-13T19:38:18 3 seconds are hit too often 2010-09-13T19:38:21 What? Nah. The tcpserver should have a little leeway, but not that much. 2010-09-13T19:38:21 *** delt0r_ has quit IRC (Ping timeout: 252 seconds) 2010-09-13T19:39:30 to prevent abuse it could be limited to 3 long answers per game 2010-09-13T19:40:16 *** danielvf has joined #aichallenge 2010-09-13T19:40:34 sigh: why can't you? connection issue? 2010-09-13T19:40:56 I think it has to do with the fact that i'm on the other side of the world 2010-09-13T19:40:57 Any dev's up for quick code review? 2010-09-13T19:41:23 I'm loosing half of my games just because I have 1 slow answer in the middle of game 2010-09-13T19:41:29 *** lurker has joined #aichallenge 2010-09-13T19:42:16 I added logging to look at the timing of my bot and none of my moves even take close to 0.1s 2010-09-13T19:42:27 so it has to be my connection 2010-09-13T19:43:12 * Naktibalda is going to ping server of a few minutes 2010-09-13T19:43:15 *** eburnette has joined #aichallenge 2010-09-13T19:43:28 benzedrine needs timeseal. http://www.freechess.org/Help/HelpFiles/timeseal.html 2010-09-13T19:43:55 sigh: I can run 4-5 instances for you right now, my machine is pretty idle.. but has to be hassle free, something I can just start and forget on ubuntu (ergo compiled java/c++) 2010-09-13T19:43:59 Is benzedrine.cx tcp up? 2010-09-13T19:44:07 eburnette: it is 2010-09-13T19:44:14 interval is 40ms - 900ms so far 2010-09-13T19:44:39 ratatata: thanks 2010-09-13T19:44:59 At the moment though we don't have the code split up so the games can run on different machines 2010-09-13T19:45:07 *** inverselimit has joined #aichallenge 2010-09-13T19:45:15 ratatata, thanks for the offer but my bot is not really stable enough to be worth the effort 2010-09-13T19:45:38 maybe if I'm still having trouble down the track, and I have a bot I am proud of :) 2010-09-13T19:45:55 my getplayer page is returning a blank page 2010-09-13T19:46:12 danielvf: still need reviewing? 2010-09-13T19:46:23 I'm terribly sleep deprived at the moment.. can't wait to get the job meeting done tomorrow and work on it with a clear mind 2010-09-13T19:46:29 yeah, got three small website changes 2010-09-13T19:46:40 eburnette: wait for a few minutes 2010-09-13T19:46:43 (that have been running in production all day) 2010-09-13T19:46:50 http://code.google.com/p/ai-contest/source/list 2010-09-13T19:46:59 Well, small small and medium 2010-09-13T19:47:22 *** num1 has joined #aichallenge 2010-09-13T19:47:43 heh. so you're using tha tbranch in production now? 2010-09-13T19:48:14 a1k0n: Just the changed files. Have to know if it works. :P 2010-09-13T19:49:16 danielvf: so why does changing the query from -1 days to -3? 2010-09-13T19:49:29 *** DirtyKeyboard has quit IRC (Remote host closed the connection) 2010-09-13T19:49:31 ...help? 2010-09-13T19:49:49 oh duh 2010-09-13T19:49:49 *** DirtyKeyboard has joined #aichallenge 2010-09-13T19:49:52 looking at wrong rev 2010-09-13T19:49:54 two commits 2010-09-13T19:49:56 trick 2010-09-13T19:49:58 *y 2010-09-13T19:50:06 *** sigh has quit IRC (Remote host closed the connection) 2010-09-13T19:50:11 *** RainCT has quit IRC (Remote host closed the connection) 2010-09-13T19:50:48 ah. they're indexed by timestamp? 2010-09-13T19:51:11 Yeah, two big indexes 2010-09-13T19:51:12 oh that explains why games from three days ago aren't showing up anymore 2010-09-13T19:51:16 so, can anyone access the playbacks right no? 2010-09-13T19:51:28 rogue780: anyone 2010-09-13T19:51:28 I think tcp server just sneezed screwing up a couple probably alive games ^^ 2010-09-13T19:51:44 rogue780: no, people without computers don't have access to your playbacks. 2010-09-13T19:51:46 a1k0n: I can try taking that three day thing out. 2010-09-13T19:51:50 yeah, just timeouted 2010-09-13T19:52:04 danielvf: but doesn't the optimization depend on that? 2010-09-13T19:52:17 or just changing g.player_two to g.loser, etc, helps? 2010-09-13T19:52:24 allow me to rephrase my question, is anyone else in this channel unable to playback battles right now? 2010-09-13T19:52:27 *** sigh has joined #aichallenge 2010-09-13T19:52:32 yeah, because I have two uber indexes 2010-09-13T19:52:33 KEY `winner_3` (`winner`,`timestamp`,`draw`,`game_id`,`loser`), 2010-09-13T19:52:33 KEY `loser_3` (`loser`,`timestamp`,`draw`,`game_id`,`winner`) 2010-09-13T19:52:55 rogue780: works fine for me, what seems to be the issue? 2010-09-13T19:52:55 everything this query needs is in the index 2010-09-13T19:52:57 am I allowed to setup a sandbox like the one at http://www.benzedrine.cx/planetwars without disqualifying myself from the tournament? 2010-09-13T19:53:08 DirtyKeyboard, pages aren't loading 2010-09-13T19:53:08 ah. but why include timestamp? 2010-09-13T19:53:12 which means it doesn't have to hunt all over the disk for the game details 2010-09-13T19:53:16 danielvf: oh, could you pastebin output of EXPLAIN ? 2010-09-13T19:53:17 DirtyKeyboard, can you access http://www.benzedrine.cx/planetwars/getplayer?player=ErrBot_0.0.3? 2010-09-13T19:53:20 that last where? 2010-09-13T19:53:26 ...which you added? 2010-09-13T19:53:32 rogue780: yep 2010-09-13T19:53:33 rogue780: I can't open my page 2010-09-13T19:53:41 can you send more than one fleet per turn? 2010-09-13T19:53:41 rogue780: Yeah, in a sec 2010-09-13T19:53:57 amstan, yes 2010-09-13T19:54:01 a1k0n: Since it would get all the games before combining them, and then limiting 2010-09-13T19:54:07 darn 2010-09-13T19:54:10 *games for that player 2010-09-13T19:54:31 Which might not be too bad 2010-09-13T19:54:34 But anyway 2010-09-13T19:54:36 well, right 2010-09-13T19:54:41 unless it's on the full listing, or whatever 2010-09-13T19:55:05 so i guess it amkes sense to include timestamp in the index 2010-09-13T19:55:12 as like the last part of the key 2010-09-13T19:55:32 or no, just as you have it 2010-09-13T19:55:39 hmm. yeah. 2010-09-13T19:56:45 Naktibalda, what browser are you running? 2010-09-13T19:56:50 *** kozlovsky has quit IRC (Ping timeout: 252 seconds) 2010-09-13T19:56:58 rogue780: it works now 2010-09-13T19:57:14 http://www.benzedrine.cx/planetwars/canvas?game_id=1284418010|blister|ErrBot_0.0.3 DOMINATED 2010-09-13T19:57:51 rogue780: have you a questionmark at the and of address in browser? 2010-09-13T19:58:19 no. i had a question mark at the end of a sentence 2010-09-13T20:00:22 Is there a faster way to try my bot against others? ai-contest takes hours and benzedrine.cx takes about 5 min per game 2010-09-13T20:00:26 Would j3cameron be who needs to make the decision on adding turns remaining output to the bots? 2010-09-13T20:00:46 oops s/j3cameron/j3camero 2010-09-13T20:00:56 eburnette: Only if someone will give you their bot to run locally. 2010-09-13T20:01:43 eburnette: which will not be of any help since test against 1 bot does not mean much unless that is the 'ultimate' bot :) 2010-09-13T20:01:55 me thinks 2010-09-13T20:01:57 Then, you can treat it like one of the example bots. Still match/5mins. isn't exactly slow. 200s > 180s = 3 mins. 2010-09-13T20:02:00 Janzert: wasn't it added allready? 2010-09-13T20:02:47 eburnette: you can play against example bots 2010-09-13T20:02:52 I created a patch for it, but I don't believe it's actually been accepted or used 2010-09-13T20:02:54 When I submit to ai-contest I want to run against all bots 2010-09-13T20:03:07 already beat example bots 2010-09-13T20:03:11 Who's "r6" on the tcp server. You totally owned me twice in a row. :( 2010-09-13T20:03:50 *** lurker has quit IRC (Quit: Page closed) 2010-09-13T20:03:53 owned me in 90 moves 2010-09-13T20:04:31 If CPU power is a problem we could farm out to ec2 2010-09-13T20:05:28 should be expensive 2010-09-13T20:05:46 Janzert: probably, yes. did you submit a patch to the issue tracker though? 2010-09-13T20:05:57 yes, issue 74 2010-09-13T20:06:04 oh, cool 2010-09-13T20:06:11 I'll try and catch him on later then 2010-09-13T20:06:39 Google is sponsoring, they could chip in or we could take up a collection 2010-09-13T20:06:56 Naktibalda: I survived 89 turns and 106 turns, respectively. 2010-09-13T20:06:58 that suggestion has been made so many times before :( 2010-09-13T20:07:04 Naktibalda: But, they weren't even close. 2010-09-13T20:07:10 Sorry just started today 2010-09-13T20:07:16 no problem 2010-09-13T20:07:27 just...google appears to sponsor in not much more than name 2010-09-13T20:07:56 i would very much like to see google provide more 2010-09-13T20:08:02 If I volunteer to help with the issue tracker (i.e. closing bugs, marking dups) does that invalidate me from the contest? 2010-09-13T20:08:03 bss03: O owned that in 85 & 150 moves :) but got owned twice as well 2010-09-13T20:08:57 Janzert: no 2010-09-13T20:09:14 dammit, i just got a timeout 2010-09-13T20:09:29 but...i work for google, and even i don't understand its relationship to this contest fully 2010-09-13T20:09:35 anyway 2010-09-13T20:09:37 * Zannick -> 2010-09-13T20:11:33 Janzert: Only server access takes you out of the contest 2010-09-13T20:11:38 weird loss, 1 move: http://www.benzedrine.cx/planetwars/canvas?game_id=1284421248|eburnette|stephen-smith.50cf 2010-09-13T20:12:01 Janzert: yeah, you could be a committer. i am. 2010-09-13T20:12:10 ok, thanks. I got that impression from the eligibility page, but it isn't explicitly mentioned I thought 2010-09-13T20:12:14 Janzert: j3camero would be happy to have you i'm sure 2010-09-13T20:13:10 I'd be happy to help, if it's wanted. I'll try and ask him about that next time I see him on as well :) 2010-09-13T20:13:14 eburnette: Means you crashed or timed out. 2010-09-13T20:14:15 This one was kind of interesting, I was up for a long time and then lost it http://www.benzedrine.cx/planetwars/canvas?game_id=1284422285|stephen-smith.50cf|eburnette 2010-09-13T20:14:51 wow i now have two followers on twitter 2010-09-13T20:15:02 a1k0n: lol 2010-09-13T20:15:22 Wait till the spammers want to be your followers 2010-09-13T20:15:55 a1k0n: thats great, my only followers on twitter that I dont use are I think either spammers or prostitutes 2010-09-13T20:16:18 rad. yeah i don't use it either. i just signed up to have an account, and posted one thing. 2010-09-13T20:16:30 i think that describes 99.94% of twitter accounts 2010-09-13T20:17:07 * Janzert goes back to watch resumed US Open final 2010-09-13T20:17:53 this is weird...http://www.benzedrine.cx/planetwars/canvas?game_id=1284422438|ErrBot_0.0.3|TracerBot11.1 2010-09-13T20:17:58 http://www.benzedrine.cx/planetwars/canvas?game_id=1284422285|stephen-smith.50cf|eburnette -- I am the comeback KING! 2010-09-13T20:18:57 rogue780: why weird? 2010-09-13T20:19:11 my bot didn't launch a single fleet 2010-09-13T20:19:43 bss03, wow, that's impressive, though admittedly your bot probably shouldn't get to the point that a comeback is necessary 2010-09-13T20:20:23 got defense mechanism? if you dont account for that situation, there is no coming back from that 2010-09-13T20:20:30 *** danielvf_ has joined #aichallenge 2010-09-13T20:22:35 bss03 you're stephen-smith.50cf? 2010-09-13T20:23:32 *** danielvf has quit IRC (Ping timeout: 245 seconds) 2010-09-13T20:23:33 *** danielvf_ is now known as danielvf 2010-09-13T20:23:50 bss03: we are playing right now /grin 2010-09-13T20:24:01 *** dmj111 has quit IRC (Ping timeout: 252 seconds) 2010-09-13T20:24:45 @(*$!^% timeouts 2010-09-13T20:24:46 eburnette: Error: "(*$!^%" is not a valid command. 2010-09-13T20:24:57 stupid timeouts :( 2010-09-13T20:28:26 r6 is pretty good 2010-09-13T20:28:29 amstan: you around? 2010-09-13T20:28:36 danielvf: yo? 2010-09-13T20:28:56 maybee i am retarded or something.. but where is the PlayGame source? 2010-09-13T20:29:04 *** drg has joined #aichallenge 2010-09-13T20:29:41 delt0r___: http://code.google.com/p/ai-contest/source/browse/#svn/trunk/planet_wars/viz 2010-09-13T20:29:59 playgame's main is Engine.java 2010-09-13T20:30:09 thanks... 2010-09-13T20:30:34 *** Ported has quit IRC (Ping timeout: 252 seconds) 2010-09-13T20:30:35 I was even there on that page and just didn't think to try the viz folder 2010-09-13T20:30:46 yeah it's kinda odd that it's there 2010-09-13T20:31:11 - I think the official website just went down.. also there hasn't been any games in the last hour for the top 50ish.. fyi 2010-09-13T20:31:27 ratatata: yeah, I caused the website grief. :( 2010-09-13T20:31:29 *nvm, site back live 2010-09-13T20:31:39 ratatata: will be slow for a bit 2010-09-13T20:34:23 bss03: if you are stephen-smith.50cf - http://www.benzedrine.cx/planetwars/canvas?game_id=1284423490|ratatata-0.3b|stephen-smith.50cf 2010-09-13T20:34:35 quite good initial deploy, work the system :) 2010-09-13T20:35:18 I'm stephen-smith on the tcp-server. 2010-09-13T20:35:33 The stuff after the dot is the first part of the git commit hash. 2010-09-13T20:36:42 I've gotta modify the bot to not send out so many attacking fleets. :( 2010-09-13T20:37:00 I leave my planets too hard to defend later. 2010-09-13T20:37:20 But, if I can stablise, I can usually come back. 2010-09-13T20:44:43 g'night all 2010-09-13T20:46:31 clear AB-algoritm with sample heurisic make terible rezilt? 2010-09-13T20:47:06 or not so bad? 2010-09-13T20:47:09 Accoun: alpha-beta? 2010-09-13T20:47:10 *** ratatata has quit IRC (Quit: Page closed) 2010-09-13T20:47:13 yes 2010-09-13T20:47:16 the action space is so huge i don't know if anyone has attempted it 2010-09-13T20:47:37 Accoun, that's what I'm currently working on 2010-09-13T20:47:44 might work really well with sort of high-level actions 2010-09-13T20:47:51 a1k0n> not alpha-beta? 2010-09-13T20:47:56 a1k0n> u use not alpha-beta? 2010-09-13T20:48:14 Accoun: my bot right now? it just doing arithmetic on fleets incoming 2010-09-13T20:48:17 *** yowaken has left #aichallenge 2010-09-13T20:48:18 it's very simple and dumb 2010-09-13T20:48:21 I'm getting over the problem a1k0n mentioned by pre-pruning out the moves that *really* don't make any sense 2010-09-13T20:49:08 num1: but among the moves that do make sense, are you searching for all values of "send 20 ships to X" through "send 50 ships to X", and what about simultaneous stuff going on with other planets? 2010-09-13T20:49:29 a1k0n> 8n plase on tcp server, not so bad 2010-09-13T20:49:31 a good pruning algorithm is really difficult, for me at least 2010-09-13T20:49:46 Accoun: yeah, it does pretty well for how simple it is 2010-09-13T20:49:55 i was very surprised to see it dominating at first 2010-09-13T20:50:38 a1k0n, i haven't completely decided yet, but If I do end up having a range of possible ships, I'll most likely only search through the endpoints of the range and the midpoint, and maybe not even the midpoint 2010-09-13T20:51:11 num1: not a bad idea. i do have a range of ships now, maybe i'll give it a shot later 2010-09-13T20:51:50 when I get it up and running (probably Sunday) we should compete on tcp :) 2010-09-13T20:51:51 i think it may be better in this challenge to try to model your opponent though 2010-09-13T20:52:15 though last time i discovered it was dangerous to assume your opponent was dumb 2010-09-13T20:52:26 I thought about that, but I don't know nearly enough about that sort of analysis to pull it off quickly 2010-09-13T20:52:52 it comes down to probability distributions 2010-09-13T20:52:58 and every use-case I can think of *should* be automatically corrected for by the ab tree 2010-09-13T20:53:15 a1k0n: You think the optimal strategy in this game is pure or mixed? 2010-09-13T20:53:19 right, as long as you're searching the moves your opponent is actually going to make... 2010-09-13T20:53:28 a-b can do everything in *theory* 2010-09-13T20:53:28 *** bhowmisu has joined #aichallenge 2010-09-13T20:53:30 mightybyte: no idea honestly. 2010-09-13T20:53:43 but i'm leaning towards analytical vs probabilistic 2010-09-13T20:53:49 if I have time (and compute time left too!) I'll probably try to model the enemy and use the model to provide weights for my search algorithm 2010-09-13T20:54:47 a1k0n> jast use fast expand algoritm ignore oponent? 2010-09-13T20:55:29 Accoun: that sort of tactic is vulnerable to rage tactics :P 2010-09-13T20:55:31 Accoun: yes but you need to consider how many ships your opponent can send to each planet 2010-09-13T20:55:38 a1k0n, that's where the I'm pruning out the bad moves part comes in, I'll make it a property (if possible) that everything pruned out is guaranteed to be a worse option than something I've left in, so I'll be fighting the worse-case opponent, and if they mess up I'm just at an advantage 2010-09-13T20:56:22 Accoun, ignore opponent is probably not a good idea, due to some of the game mechanics 2010-09-13T20:57:08 a1k0n> its like AB on 1 move forward 2010-09-13T20:57:38 Accoun: sure, if your heuristic is good. that's basically what i'm working on now 2010-09-13T20:58:01 *** ss_ has joined #aichallenge 2010-09-13T20:58:35 *** drewrobb has quit IRC (Ping timeout: 276 seconds) 2010-09-13T20:59:39 *** amstan has quit IRC (Remote host closed the connection) 2010-09-13T20:59:55 wait...how did copybot time out 2010-09-13T21:00:00 i don't believe you 2010-09-13T21:00:12 :P tpc 2010-09-13T21:00:13 cp 2010-09-13T21:00:15 * tcp 2010-09-13T21:00:46 which bot is copybot? 2010-09-13T21:00:57 it's named "copybot" on the tcp server 2010-09-13T21:01:15 the official site is a little too slow to watch 2010-09-13T21:01:37 *** amstan has joined #aichallenge 2010-09-13T21:01:37 *** ChanServ sets mode: +o amstan 2010-09-13T21:01:44 ah, i see it now, it wasn't in the ranking list 2010-09-13T21:01:50 Zannick: you should submit copybot 2010-09-13T21:02:01 it's up 2010-09-13T21:02:16 i submitted it on the main site as a test 2010-09-13T21:02:21 and i haven't gotten a real bot yet 2010-09-13T21:02:36 It's really quite amazing that ragebot is as good as it is. 2010-09-13T21:03:03 under 'Zannick'? 2010-09-13T21:03:09 'bswolf' 2010-09-13T21:03:18 yeah, it's worth looking at making an improved version of ragebot 2010-09-13T21:03:26 oh haha, wasn't that in the top 10 for a while? 2010-09-13T21:05:09 hahah http://ai-contest.com/visualizer.php?game_id=4459337 2010-09-13T21:05:42 Hah, that's hilarious. 2010-09-13T21:06:22 bswolf is ever so slightly better 2010-09-13T21:06:38 heh 2010-09-13T21:06:47 i'm still not sure how he ended up ahead 2010-09-13T21:06:58 oh, bottom planet 2010-09-13T21:07:21 turn 62-63 decided it 2010-09-13T21:09:03 yeah, there's always one in a win 2010-09-13T21:09:09 most of the time it's the center 2010-09-13T21:09:19 omg, I want a copy of copybot! :D 2010-09-13T21:09:44 also, the currently running copybot is not quite the same as bswolf 2010-09-13T21:10:40 copybot and real-1-1-1 keep timing out on the same turn 2010-09-13T21:11:51 i presume it is the sheer number of fleets 2010-09-13T21:12:44 this is interesting: http://ai-contest.com/rankings.php?page=5 2010-09-13T21:13:02 lots of 404's... 2010-09-13T21:13:16 haha 2010-09-13T21:13:31 Mathnerd314: It's doing ok for me. 2010-09-13T21:13:37 elo score is 1984... ominous 2010-09-13T21:13:57 wooo 3 googlers in the top 100 2010-09-13T21:14:15 (why is everyone "other"?) 2010-09-13T21:14:39 *** td123 has quit IRC (Quit: WeeChat 0.3.3) 2010-09-13T21:14:40 Anyone know what kind of a rating ragebot gets? 2010-09-13T21:15:02 on a red to green scale I would say incredible hulk 2010-09-13T21:16:00 mightybyte: see http://ai-contest.com/profile.php?user_id=3928 2010-09-13T21:16:06 an elo rating 2010-09-13T21:16:23 Ahh, yes. 2010-09-13T21:17:13 *** drg has quit IRC (Quit: Page closed) 2010-09-13T21:17:18 hm, ragebot didn't have the one-fleet rule? 2010-09-13T21:18:12 nope 2010-09-13T21:18:12 * Mathnerd314 checks starter pack 2010-09-13T21:18:19 another draw with real-1-1-1 2010-09-13T21:18:20 gee 2010-09-13T21:18:23 *** rafsoaken has quit IRC (Quit: Page closed) 2010-09-13T21:21:01 @rankings 2010-09-13T21:21:03 amstan: Top 10 players: deepblue(2699), sighbot(2679), wtfbbq(2675), Kavish(2668), a1k0n(2648), hellman(2643), zuko3d(2628), luismi(2609), braddunbar(2606), Loks1982(2599) 2010-09-13T21:21:14 o_O 2010-09-13T21:21:24 I need to get back to work writting my planet selection 2010-09-13T21:22:08 my bot always targets planets too far away. I need to remedy this 2010-09-13T21:22:16 *** gw has quit IRC (Ping timeout: 252 seconds) 2010-09-13T21:22:25 for each notmyplanets: for each my planets: and turns to conquer, ship cost and growth rate gained 2010-09-13T21:22:49 sound good? 2010-09-13T21:23:15 If it was, we might tell you it's not and steal it 2010-09-13T21:23:16 ;) 2010-09-13T21:23:32 so...there's definitely an upper limit on the number of fleets 2010-09-13T21:23:49 the ratio between the 3 is what counts 2010-09-13T21:23:53 at a certain point, the input parser is taking a second on it 2010-09-13T21:25:54 who runs the stephen smith bot again? 2010-09-13T21:25:56 what do you mean, Zannick? 2010-09-13T21:25:59 Zannick, really? Have you found an upper limit in the engine code? 2010-09-13T21:26:21 rogue780: bss03 2010-09-13T21:26:29 IIRC 2010-09-13T21:26:50 not in the engine source 2010-09-13T21:27:11 sigh, yes. I believe you are quite right 2010-09-13T21:27:12 but 4 out of 4 games, copybot vs real-1-1-1, draw by both timeout 2010-09-13T21:27:17 bss03, I just defeated thee 2010-09-13T21:27:25 *** McLeopold has joined #aichallenge 2010-09-13T21:28:16 hmm... how many fleets does it take to time it out? 2010-09-13T21:30:04 *** yowaken has joined #aichallenge 2010-09-13T21:32:21 http://www.benzedrine.cx/planetwars/canvas?game_id=1284426952|ErrBot_0.0.3|stephen-smith.50cf 2010-09-13T21:33:08 i have 468 for the third game, 467 for the second, and 481 for the first 2010-09-13T21:33:19 from the tcp server 2010-09-13T21:33:46 of course, i don't know that it's just the parser 2010-09-13T21:33:57 but that's how many caused us to both time out 2010-09-13T21:34:07 ah i see 2010-09-13T21:34:42 what language? 2010-09-13T21:35:00 ha, and the fourth was 501 2010-09-13T21:35:15 the numbers are pretty close 2010-09-13T21:35:21 yeah 2010-09-13T21:35:21 copybot is mostly the c++ starter bot 2010-09-13T21:35:28 dunno about real-1-1-1 2010-09-13T21:35:40 damn... if parsing is the problem on c++ that's very bad 2010-09-13T21:35:49 rougue780: Yeah, after the changes I made today, my bot is not very good. :/ 2010-09-13T21:35:57 i'm writing entirely from scratch in c 2010-09-13T21:36:12 so with scanf instead of that string parsing thing 2010-09-13T21:36:38 I would profile first 2010-09-13T21:36:54 and without all the random vector stuff 2010-09-13T21:37:02 sigh: I'm also seeing double-timeouts against real-1-1-1 2010-09-13T21:37:12 The server doesn't send back a valid game state. 2010-09-13T21:37:17 jmreardon: likewise 2010-09-13T21:37:23 ah, that might be it 2010-09-13T21:37:24 err, likewise 2010-09-13T21:37:29 sigh: I'll get something like "F 2go" as the last line. 2010-09-13T21:37:32 ah well, there you go then 2010-09-13T21:37:38 i'm filtering those out of my stdout 2010-09-13T21:37:53 real-1-1-1 must be overflowing some buffer on the server. 2010-09-13T21:37:57 bss03: same here: F 2 1 go 2010-09-13T21:38:00 so all i see is "Your opponent is" followed by "You win" or something 2010-09-13T21:38:19 bss03, don't worry. I'm sure you'll fix it and get back to kicking my ass again like you did yesterday 2010-09-13T21:39:13 rogue780: I think I can get some marked improvement by just tweaking 2 lines. 2010-09-13T21:39:36 the starter code might actually be exiting on a bad line 2010-09-13T21:39:45 rouge780: But, I actually want to replace those two lines with something more complex, so the behavior of my bot changes once it is ahead. 2010-09-13T21:41:36 Another case where I come back from rather far behind: http://www.benzedrine.cx/planetwars/canvas?game_id=1284425853|eburnette|stephen-smith.50cf 2010-09-13T21:41:57 eburnette just doesn't keep the pressure on enough. 2010-09-13T21:49:20 bss03, yeah, I saw that game. it was tight 2010-09-13T21:50:31 workin on it 2010-09-13T21:52:01 *** xvinyl has joined #aichallenge 2010-09-13T21:53:05 2:47AM here :( it looks like I had more productive hours tonight than during work day 2010-09-13T21:53:25 Naktibalda, where do you live? 2010-09-13T21:53:31 UK 2010-09-13T21:53:44 *** bobcolem has joined #aichallenge 2010-09-13T21:54:06 ah. Only part of the UK I've been to is Northern Ireland 2010-09-13T21:55:32 *** bobcolem has left #aichallenge 2010-09-13T21:55:58 wtf. I wish I could set it so I don't get stuck playing with real1-1-1 2010-09-13T21:56:04 i hate him 2010-09-13T21:56:52 it looks like retybok has improved 2010-09-13T21:57:08 wait, copybot just beat real-1-1-1 2010-09-13T21:57:11 no timeout 2010-09-13T21:57:21 holy shit 2010-09-13T21:57:27 Mathnerd314: i'm finally working on extracting (very) simple starter from my code 2010-09-13T21:57:38 just after losing to him, but... 2010-09-13T21:57:45 jmcarthur: yay! :-) 2010-09-13T21:57:47 it's actually a result 2010-09-13T21:57:56 Mathnerd314: i would have done it earlier today, but my wife and i ended up having to waste the day tearing her laptop into little pieces and putting them back together again 2010-09-13T21:58:14 a few times 2010-09-13T21:58:41 Mathnerd314: i'm making the starter by inserting it into the specification as literate haskell. should be pretty nice i hope :) 2010-09-13T22:00:26 *** Queue29 has quit IRC (Read error: Connection reset by peer) 2010-09-13T22:00:59 well played, bss03 2010-09-13T22:02:21 *** Naktibalda has quit IRC (Remote host closed the connection) 2010-09-13T22:03:35 rogue780: It was a come-back again. :/ 2010-09-13T22:03:47 hey, a comeback is still a win 2010-09-13T22:04:00 and you didn't come back from that much of a deficit 2010-09-13T22:04:02 rogue780: I've got to save this particular version of the bot. It has an amazing propensity for doing that. 2010-09-13T22:04:30 Maybe switch back to this simple strategy when I'm behind, or smth likt that. 2010-09-13T22:04:38 it was a fun game though http://www.benzedrine.cx/planetwars/canvas?game_id=1284429037|ErrBot_0.0.3|stephen-smith.50cf 2010-09-13T22:05:30 for some reason my defense mode isn't kicking in like it should 2010-09-13T22:08:41 Planet-Wars-Canvas-Visualizer: Daniel Von Fange master * r40974e1 / js/visualizer.js : Minor tweak. Remove uneeded config line. Replace with grabbing the canvas's actual size. - http://bit.ly/aSSunc 2010-09-13T22:08:42 Planet-Wars-Canvas-Visualizer: Daniel Von Fange master * r2c91c0f / js/visualizer.js : Fix for pausing when no fleets are in the air - http://bit.ly/cDmm8L 2010-09-13T22:11:24 Looks remarkably close, but I have the lead the whole time: http://www.benzedrine.cx/planetwars/canvas?game_id=1284428605|stephen-smith.50cf|jmreardon.2.2 2010-09-13T22:12:48 bss03: mostly 'cause we spent the better part of the game going at the one planet 2010-09-13T22:13:16 My bot thought it would be easy to take over. :P 2010-09-13T22:15:58 you bot is stuck up 2010-09-13T22:16:09 How so? 2010-09-13T22:16:31 wasn't so easy, was it? 2010-09-13T22:16:34 * bss03 giggles. 2010-09-13T22:16:36 *** Queue29 has joined #aichallenge 2010-09-13T22:16:55 Nope, it wasn't, but I did take it eventually. 2010-09-13T22:17:20 rouge780: It hasn't hit the website, but I think I just beat you again. 2010-09-13T22:17:41 rogue780: Timing out? Is your connection sucking? 2010-09-13T22:17:54 bss03, did I time out more than once? 2010-09-13T22:18:20 Just once while I was watching, but I only have TCP-server on in the background. 2010-09-13T22:18:38 jmcarthur: ETA on finishing? 2010-09-13T22:18:51 I was holding the scroll bar to read something and then I realized that it paused the process (I love windows) and I think that caused the timeout 2010-09-13T22:19:03 Mathnerd314: umm.... sometime tonight :) 2010-09-13T22:19:09 if it happened more than once though, there may be another problem 2010-09-13T22:19:11 LOL 2010-09-13T22:19:12 i really only just got started 2010-09-13T22:19:27 and this is also making me implement some functionality that i've been neglecting, and refactor a fair bit 2010-09-13T22:19:42 retybok and I are playing right now 2010-09-13T22:19:49 You know, I suspect my bot might do better if I solve all the invalid orders it seems to want to give 2010-09-13T22:19:49 anyone want to take bets on the winner? 2010-09-13T22:20:03 Not really. 2010-09-13T22:20:04 rogue780: what sort of bet? 2010-09-13T22:20:05 (if you want to bribe me to throw the match, I could just hold my scrollbar and time it out) 2010-09-13T22:20:10 Mathnerd314, bacon 2010-09-13T22:20:47 ermau, I found a lot of mine were 0 sized fleets 2010-09-13T22:20:54 Mine aren't -_- 2010-09-13T22:21:02 Which means it thinks it's getting forces out that it's not 2010-09-13T22:21:26 and then I wrote a helper method that transposed the destination id with the source id and caused a lot of invalid moves.... 2010-09-13T22:21:28 For some reason I keep tweaking the old bot even though I'm rewriting it 2010-09-13T22:23:42 bss03, if you wanna see a comeback... http://www.benzedrine.cx/planetwars/canvas?game_id=1284429756|ErrBot_0.0.3|real-1-1-1 2010-09-13T22:24:42 Grr, I hate watching real-1-1-1. 2010-09-13T22:24:47 It boggs my browser. 2010-09-13T22:25:51 Still, good. comeback 2010-09-13T22:26:07 I need to have smarter resource management I think 2010-09-13T22:26:41 well, retybok defeated me 2010-09-13T22:26:59 rogue780: Why does it stop at turn 116 with no winner? Was that the round limit? 2010-09-13T22:28:06 *** sfsgagi has left #aichallenge 2010-09-13T22:28:58 madsy, I'm not sure what you mean...blue doesn't have any planets left and its last fleet just failed to take my topright planet 2010-09-13T22:32:02 It's almost hard to use ships as inefficienttly as real-1-1-1 does there. 2010-09-13T22:32:40 *** preetum has joined #aichallenge 2010-09-13T22:33:04 rogue780: Might be Chrome that decides to die on me again.. 2010-09-13T22:33:18 *** amstan_ has joined #aichallenge 2010-09-13T22:33:18 *** ChanServ sets mode: +o amstan_ 2010-09-13T22:36:56 *** amstan has quit IRC (Ping timeout: 272 seconds) 2010-09-13T22:37:23 *** preetum has left #aichallenge 2010-09-13T22:37:28 mightybyte, ++ 2010-09-13T22:37:53 *** McLeopold has quit IRC (Ping timeout: 252 seconds) 2010-09-13T22:40:57 *** drewrobb has joined #aichallenge 2010-09-13T22:41:52 mightybyte: yeah... that bot is not smart. :P 2010-09-13T22:44:43 *** pacotabaco has joined #aichallenge 2010-09-13T22:45:24 hi, anybody is using "New python kit (unofficial)"? 2010-09-13T22:50:39 Mathnerd314: maybe not tonight 2010-09-13T22:50:43 bedtime is already approaching 2010-09-13T22:50:51 jmcarthur: oh noes! 2010-09-13T22:51:14 Mathnerd314: i have the parsing and IO stuff in it already. i was just going to also implement some basic engine-ish functionality 2010-09-13T22:51:23 jmcarthur: how about you just use my code instead. It probably does what you want. :p 2010-09-13T22:52:01 http://github.com/Mathnerd314/planet-wars-haskell 2010-09-13T22:52:22 although the step function is broken 2010-09-13T22:52:36 Mathnerd314: i actually have much of it written anyway. i just have to factor it out properly 2010-09-13T22:52:45 Mathnerd314: plus i'm trying to make it flow nicely with the spec 2010-09-13T22:55:45 Gah! this is keeping me away from my homework 2010-09-13T22:57:43 Mathnerd314: i don't want it linked around yet, but just for you and anybody listening, here's the work in progress: http://jake.devio.us/starter/PlanetWars.html 2010-09-13T22:58:06 Mathnerd314: i will also be distributing a markdown .lhs version eventually 2010-09-13T22:58:15 jmcarthur: this channel is logged, not certain how secret that is :p 2010-09-13T22:58:18 *** td123 has joined #aichallenge 2010-09-13T22:58:18 pacotabaco: I'm using that library 2010-09-13T22:58:34 i have a new strategy I want to employ, but I must finish this annoying art homework 2010-09-13T22:58:35 Mathnerd314: i don't care that it's a secret. i just don't want it publicized much 2010-09-13T22:58:42 until it's ready 2010-09-13T22:59:21 pacotabaco: have any problems with it? 2010-09-13T22:59:22 I changed by name on the TCP server, now it is based on a git-describe of my tree, which is based on the git tag I just added. 2010-09-13T22:59:35 I've never seen the allure of haskell 2010-09-13T22:59:36 It might make it easier to compare with my bot on the official server. 2010-09-13T22:59:59 Mathnerd314: no lhs file yet, but in the meantime one can just copy the web page 2010-09-13T23:00:07 not that i really expect anybody to yet 2010-09-13T23:00:41 <3 literate haskell 2010-09-13T23:02:46 jmcarthur, is haskell interpreted or compiled mainly? 2010-09-13T23:04:08 jmcarthur: I don't like how your Owner type limits play to 2 players. 2010-09-13T23:04:42 rogue780: compiled 2010-09-13T23:04:46 jmcarthur: I haven't done it, but it would be nice to have the bot easily play on a 3 or 4 player map. 2010-09-13T23:04:57 bss03: the spec very clearly states that it's for the 2-player variant of the game 2010-09-13T23:05:58 really, i don't care that some of our engines can do >2 players. the contest is only for two 2010-09-13T23:06:24 and the spec is only for two, as well 2010-09-13T23:08:00 hmm, it occurs to me that that code requires a few extensions and i'm not including them as pragmas 2010-09-13T23:08:04 i'll have to add that 2010-09-13T23:08:17 i can only think of OverloadedStrings right now 2010-09-13T23:08:29 anyway, i think i'm done for tonight 2010-09-13T23:09:53 *** amstan has joined #aichallenge 2010-09-13T23:09:53 *** ChanServ sets mode: +o amstan 2010-09-13T23:10:09 *** deepblue has joined #aichallenge 2010-09-13T23:10:13 http://www.benzedrine.cx/planetwars/canvas?game_id=1284432111|rb_st_egc|deepblue-test 2010-09-13T23:10:16 woho 2010-09-13T23:10:20 action 2010-09-13T23:11:53 *** amstan_ has quit IRC (Ping timeout: 245 seconds) 2010-09-13T23:12:40 cool game! nice comeback 2010-09-13T23:12:41 deepblue: lol, i like how your bot saves a bunch of forces up and then strikes big 2010-09-13T23:13:05 I liked the planet trade :) 2010-09-13T23:13:07 both wait and then the final showdown including changing all planets owners 2010-09-13T23:13:17 very cool 2010-09-13T23:13:26 all kind of chaos in the middle there 2010-09-13T23:14:47 does you bot explictly decide to abandon planets, or did it just happen that way? 2010-09-13T23:16:27 *** hornairs has joined #aichallenge 2010-09-13T23:16:28 ohh.... red attacked two 79-ship neutrals near you... that gave you the advantage 2010-09-13T23:18:22 yeah 2010-09-13T23:18:28 it used the weakness that opened 2010-09-13T23:18:33 *** yowaken has quit IRC (Quit: Page closed) 2010-09-13T23:18:43 and it abandoned the planets to safe the fleet 2010-09-13T23:19:22 hi all 2010-09-13T23:19:36 jmcarthur: reading your code right now 2010-09-13T23:20:39 jmcarthur: I think someone mentioned a problem about ceiling and floor in ghc 2010-09-13T23:21:30 jmcarthur: on haskell-cafe this week 2010-09-13T23:24:46 deepblue: That was a cool game. 2010-09-13T23:28:08 *** drewrobb has quit IRC (Ping timeout: 245 seconds) 2010-09-13T23:28:23 bartwe.6's behavior is curious 2010-09-13T23:29:13 mine lost a few times against bartwe earlier today gave me great valuable input :) 2010-09-13T23:29:28 I'm a net loser to barwe.6 so far 2010-09-13T23:30:07 ah, I think I understand the reasoning behind what it's doing. very clever. 2010-09-13T23:31:12 amstan: My uni says we can donate a 4core 64bit with 6gb of ram 2010-09-13T23:31:17 amstan: Do you need? 2010-09-13T23:31:49 We have >1gbit connection also : 2010-09-13T23:31:52 greghaynes: sounds nice, can you send me an email? 2010-09-13T23:31:58 What addr? 2010-09-13T23:31:59 greghaynes: jeff's not here 2010-09-13T23:32:05 greghaynes: amstan@ai-contest.com 2010-09-13T23:32:16 greghaynes: actually.. cc to jeff@ai-contest.com too 2010-09-13T23:32:37 Ok 2010-09-13T23:32:49 *** hellman has joined #aichallenge 2010-09-13T23:32:56 greghaynes: thanks 2010-09-13T23:33:01 np 2010-09-13T23:34:26 *** hellman has quit IRC (Remote host closed the connection) 2010-09-13T23:34:52 *** hellman has joined #aichallenge 2010-09-13T23:36:04 deepblue, you are going to be very hard to beat 2010-09-13T23:36:45 I think I lost to Damien just know 2010-09-13T23:36:46 now 2010-09-13T23:36:51 so there are ways to beat me 2010-09-13T23:36:57 but damien loses against others 2010-09-13T23:37:03 so it's all good and fun so far :) 2010-09-13T23:38:25 lol I can't wait to see how this all turns out 2010-09-13T23:38:41 deepblue, is your bot playing the tcp server? I want a shot at you... ;) 2010-09-13T23:39:00 looks like he quit for the day 2010-09-13T23:39:17 back in the tron days there weren't this many games played on there per unit time 2010-09-13T23:39:29 so they wouldn't roll over so fast 2010-09-13T23:39:39 it's running on the tcp right now 2010-09-13T23:40:23 http://www.benzedrine.cx/planetwars/canvas?game_id=1284434318|deepblue|damien 2010-09-13T23:40:24 there you are 2010-09-13T23:40:59 there are some weird things going on that are not supposed to be happening 2010-09-13T23:41:15 * deepblue stupid bot 2010-09-13T23:43:30 Wonder why damien didn't win that. 2010-09-13T23:44:09 Although I must admit, it is fun to watch a lost opponent suffer while you accumulate masses of ships. 2010-09-13T23:47:20 deepblue, i still think you might have been able make a comeback if the turn limit hadn't been hit 2010-09-13T23:48:13 maybe, at least I think I found the bug 2010-09-13T23:48:14 :) 2010-09-13T23:48:22 wait, in the game where damien beat deepblue, did he refrain from wiping deepblue out because he wanted to rack up points? I looked in the ranking system and I thought it only took into consideration win/loss 2010-09-13T23:48:37 if it makes you feel any better, I've lost to damien too... 2010-09-13T23:48:44 Is the bug that you have if username == 'damien': dont attack? 2010-09-13T23:49:28 *** pacotabaco has quit IRC (Quit: Page closed) 2010-09-13T23:49:32 hehe 2010-09-13T23:49:52 it has todo with a if that doesnt what it's supposed to do :) 2010-09-13T23:51:59 hrmm, my bot is too agressive it seems 2010-09-13T23:52:38 Having this visualizer sure makes it easier to tweak 2010-09-13T23:52:47 *** whiteside has joined #aichallenge 2010-09-13T23:52:50 oh yes 2010-09-13T23:52:58 single stepping back and forth 2010-09-13T23:53:14 Ah, well being able to do it without waiting for the page to update 2010-09-13T23:53:21 Since I would just refresh like a madman 2010-09-13T23:53:27 now testing the bug fix locally and if it turns out working it's going on the tcp while I watch some arrested development :) 2010-09-13T23:54:39 *** jmreardon has quit IRC (Quit: jmreardon) 2010-09-13T23:55:52 http://www.benzedrine.cx/planetwars/canvas?game_id=1284435402|damien|deepblue 2010-09-13T23:55:59 here on can see the bug quite well 2010-09-13T23:56:07 accumulating in the upper right corner :/ 2010-09-13T23:56:20 it still manages to win but that's not aceptable behaviour 2010-09-13T23:56:54 *** l4u_ has quit IRC (Ping timeout: 276 seconds) 2010-09-13T23:56:59 eek, http://www.benzedrine.cx/planetwars/canvas?game_id=1284435910|OMG_PONIES|dhartmei 2010-09-13T23:57:21 OMG PWNIES not PONIES -_- 2010-09-13T23:57:32 Thatlle be v2.0 2010-09-13T23:58:09 so many fleets flying in that game 2010-09-13T23:59:01 haha. OMG_PONIES actually returned from death twice there 2010-09-13T23:59:26 Well, when theres pones involved...