2010-09-14T00:01:50 the hopefully bugfixed version is doing some runs on the tcp now 2010-09-14T00:02:14 and I'm gonna watch some tv now and hope for either a clean record or some easy to spot bugs ;) 2010-09-14T00:02:28 *** Queue29_2 has joined #aichallenge 2010-09-14T00:02:42 *** Obeleh has joined #aichallenge 2010-09-14T00:02:56 *** l4u has joined #aichallenge 2010-09-14T00:03:33 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T00:04:04 hey all 2010-09-14T00:07:30 *** ajhager has joined #aichallenge 2010-09-14T00:08:25 *** Obeleh has quit IRC (Ping timeout: 252 seconds) 2010-09-14T00:12:19 *** ermau has quit IRC (Quit: ermau) 2010-09-14T00:22:43 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T00:22:55 *** inverselimit has quit IRC (Quit: Leaving) 2010-09-14T00:23:47 *** ajhager has joined #aichallenge 2010-09-14T00:24:03 *** certhia has joined #aichallenge 2010-09-14T00:24:38 *** preetum has joined #aichallenge 2010-09-14T00:25:38 *** certhia has quit IRC (Client Quit) 2010-09-14T00:28:59 *** preetum has quit IRC (Client Quit) 2010-09-14T00:34:50 *** zklapow has joined #aichallenge 2010-09-14T00:38:20 *** ajhager has quit IRC (Ping timeout: 252 seconds) 2010-09-14T00:39:22 ok, so, in linux AND freebsd (same machine), the Java example bots time out for me in round 1 2010-09-14T00:39:30 *** ajhager has joined #aichallenge 2010-09-14T00:39:42 I'm attempting to play on tcp, but my client seems to fail with "EOFError: EOF when reading a line" after every game 2010-09-14T00:39:44 however in windows, they all work fine, and my bots (c++) work in all three 2010-09-14T00:39:49 anybody know what could cause this? 2010-09-14T00:41:01 *** dmj111 has joined #aichallenge 2010-09-14T00:41:15 num1, the server is down 2010-09-14T00:41:46 whiteside, when you run java MyBot by itself, does it throw an exception? or does it read from stdin like it should? 2010-09-14T00:42:16 Queue29_2: I'm playing the tcp server right now, doesn't look down to me 2010-09-14T00:42:18 Queue29, however I can still connect and play games against opponents and have them show up at http://www.benzedrine.cx/planetwars/getplayer? ? 2010-09-14T00:42:33 last one was for Queue29_2 , sorry 2010-09-14T00:43:21 I seem to be able to play games on the tcp server but the site www.benzedrine.cx is down. 2010-09-14T00:43:27 oh that's weird.. http://www.benzedrine.cx/planetwars/ is down for me 2010-09-14T00:43:36 dunno if my bot is still running 2010-09-14T00:44:49 bunch of bots playing "There are currently 42 connections and 19 games running" 2010-09-14T00:45:22 *** Obeleh has joined #aichallenge 2010-09-14T00:45:38 num1, are you using the new tcp client or the old one? 2010-09-14T00:46:02 I'm using tcp.c 2010-09-14T00:46:18 version 2.2 2010-09-14T00:46:22 according to the source 2010-09-14T00:47:30 *** mightybyte has left #aichallenge 2010-09-14T00:47:30 *** deepblue has quit IRC (Quit: Page closed) 2010-09-14T00:47:59 Queue29_2, I don't specifically have the java MyBot, but the included example bots don't throw exceptions no 2010-09-14T00:48:47 it'll take a few minutes, but I guess I should try compiling the Java one myself 2010-09-14T00:49:50 whiteside, which version of java? 2010-09-14T00:50:25 1.6 something-or-another... just a second 2010-09-14T00:50:28 Queue29_2, what's the newest version? 2010-09-14T00:50:37 of the tcp client 2010-09-14T00:50:44 (build 1.6.0_20-b02) 2010-09-14T00:51:07 num1, not sure, the guy who makes it posted a new version on the forums not too long ago, but it links back to his site, which is down... 2010-09-14T00:51:10 num1, I have 2.2 2010-09-14T00:51:12 i'm running 2.1 though 2010-09-14T00:51:57 looks like website is back up, I'm going to try to play another game 2010-09-14T00:52:38 oops, I forgot I only have the JRE... so I guess that's a nogo on compiling it myself 2010-09-14T00:53:02 whiteside, can you not install a jdk? 2010-09-14T00:54:01 does the client auto-disconnect after a single game or does it keep playing game after another? 2010-09-14T00:54:11 yeeeaaahhh I guess I can't really avoid it, I'll go look for the CD 2010-09-14T00:54:17 num1, auto disconnects 2010-09-14T00:54:56 num1, i just posted this to the forums, to make life a little easier: http://gist.github.com/578352 2010-09-14T00:55:48 Queue29_2, that looks great, thanks! 2010-09-14T00:56:17 whiteside, it's posted on Oracles website if you have windows, and it's available in your pack-man if you're on linux 2010-09-14T00:58:34 now that's just embarassing http://www.benzedrine.cx/planetwars/canvas?game_id=1284439733|bartwe.6|ErrBot_0.0.3 2010-09-14T00:59:44 rogue780: Ouch. 2010-09-14T00:59:53 *** Zeiris has quit IRC (Ping timeout: 255 seconds) 2010-09-14T00:59:55 you guys can connect to the site ?? 2010-09-14T00:59:59 rogue780: Can you beat RageBot on a map like that? 2010-09-14T01:00:21 I'm working on ErrBot0.0.4 right now. it shoud improve things a lot. 2010-09-14T01:00:24 Queue29_2: What site? The official one? If so, yes. 2010-09-14T01:00:31 bss03, not sure. I've only ever tried one map locally 2010-09-14T01:00:38 i mean the benzedrine.cx site 2010-09-14T01:00:53 bartwe.6 does something exceedingly clever. bartwe is one to watch. 2010-09-14T01:01:36 bartwe has a very smart bot 2010-09-14T01:01:52 whose is vortex? 2010-09-14T01:01:58 me 2010-09-14T01:02:31 it is very good without looking clever 2010-09-14T01:02:36 lol 2010-09-14T01:02:39 thanks I think 2010-09-14T01:02:47 stupid comebacks http://www.benzedrine.cx/planetwars/canvas?game_id=1284439851|ErrBot_0.0.3|DaTwinkDaddy-s8-31-ga91f 2010-09-14T01:02:48 i mean it just.. wins, no gimmicks 2010-09-14T01:02:51 Queue29_2: Yeah, I'm connected to the TCP server, haven't check the webpages in a few minutes. 2010-09-14T01:02:53 rwa, you're vortex? 2010-09-14T01:02:58 * rogue780 shakes fist at rwa 2010-09-14T01:03:23 rogue780: I think map 26 or 29 is a really hard one to beat RageBot on. 2010-09-14T01:03:38 http://www.benzedrine.cx/planetwars/canvas?game_id=1284438320|vortex.10|a1k0n-0.2 2010-09-14T01:03:41 heh. my bot hates you. 2010-09-14T01:03:52 your bot is just like "whatever" 2010-09-14T01:04:07 bss03, I'll have to check it out. 2010-09-14T01:04:11 well, g'night everybody 2010-09-14T01:04:39 remember, if you live in Maryland, don't forget to vote tomorrow 2010-09-14T01:04:48 or today really 2010-09-14T01:05:01 wow, it's really smart to bomb me with 445 and then cascade the attack 2010-09-14T01:05:04 a1kon, I'm surprised my bot pulled that out, it was tight for awhile 2010-09-14T01:05:28 the cascaded attack is what did it. 2010-09-14T01:05:30 that's what she said 2010-09-14T01:05:41 my bot has no concept that that's possible 2010-09-14T01:06:48 a1kon: good, I'm intentionally trying to make my bot unpredictable 2010-09-14T01:07:02 *** Snowplan has joined #aichallenge 2010-09-14T01:08:04 dhartmei has a line on me 2010-09-14T01:08:08 have to figure out what's going on there 2010-09-14T01:08:11 Does the comment at the bottom of the Planet Wars source code drop the f bomb 2010-09-14T01:08:26 *** rebelxt has joined #aichallenge 2010-09-14T01:08:41 Snowplan, there's one in the starter package as well iirc 2010-09-14T01:08:53 but first I have to fix a big glaring bug and see where I stand 2010-09-14T01:09:40 num1,Queue29_2, argh it's not included on the base CD, so I'll have to go download it anyway... I have a slow connection so it'll take a good while, but thanks 2010-09-14T01:09:44 Ha was just reading thru 2010-09-14T01:09:52 good stuff 2010-09-14T01:10:51 *** ajhager has quit IRC (Ping timeout: 272 seconds) 2010-09-14T01:12:04 whiteside, it's only ~65 Mb prepackaged on ubuntu, what kind of connection do you have? 2010-09-14T01:12:16 *** Utkarsh has quit IRC (Ping timeout: 240 seconds) 2010-09-14T01:12:55 *** rebelxt has quit IRC (Ping timeout: 252 seconds) 2010-09-14T01:13:29 anyone know if theyre working on the matching algorithm? idd like to play better bots :) 2010-09-14T01:13:59 seriously. i thought jeff was going to test that tonight. perhaps tomorrow. 2010-09-14T01:14:14 *** ajhager has joined #aichallenge 2010-09-14T01:14:31 it'd be nice if a new bot had priority as well, I think 2010-09-14T01:14:41 its starting to annoy :) 2010-09-14T01:14:47 rwa: they will :) 2010-09-14T01:15:04 Queue29_2, ahh, that's not as bad as I thought... 71 mb for a slacker, but yeah not the hundreds I was anticipating 2010-09-14T01:15:07 I think the first bot everyone should play against is the Rank 1 bot. 2010-09-14T01:15:19 First, a good smackdown. Then, they can crawl back up the ranks. 2010-09-14T01:15:40 i end up playing bots ranked 300 spots lower 2010-09-14T01:16:47 don't feel too bad, they're playing bots ranked 300 spots higher :P 2010-09-14T01:16:53 wtf http://www.benzedrine.cx/planetwars/canvas?game_id=1284440391|real-1-1-1|ErrBot_0.0.3 2010-09-14T01:17:11 *** sigh has quit IRC (Remote host closed the connection) 2010-09-14T01:17:12 yea but i want to learn aswel :) 2010-09-14T01:17:26 *** sigh has joined #aichallenge 2010-09-14T01:17:48 i think real-1-1-1's strategy is to overwhelm other bots' cpu usage 2010-09-14T01:17:56 I'm a big fan of swiss. You should be paired with a closely ranked bot that you haven't played against in ceil(lg(entires)) games. 2010-09-14T01:18:21 a1k0n, he gives me a lot of free wins, so it's okay 2010-09-14T01:18:24 heh 2010-09-14T01:18:32 real-1-1-1 might actually beat me like that 2010-09-14T01:18:38 *** Utkarsh has joined #aichallenge 2010-09-14T01:18:48 is 1s wall time or cpu time? 2010-09-14T01:18:58 right now, wall. 2010-09-14T01:19:08 seems like cpu is better, no? 2010-09-14T01:19:23 yes, everyone agrees on that 2010-09-14T01:19:24 yes indeed. this debate has been raging for a week now. there is a patch iirc. 2010-09-14T01:19:34 rogue: Your bot won't attack growth 0 planets? 2010-09-14T01:19:43 im thinking about posting some basic strategies to get the new ones stated on the ladder anyone intrested in writing some pieces? 2010-09-14T01:19:59 bss03, apparently it doesn't think they're worth it...i'll be modifying that 2010-09-14T01:20:03 errbot seems to have a bit of a problem with the endgame...lol 2010-09-14T01:20:30 yeah i had to modify mine to not ignore the 0-growth planet 2010-09-14T01:20:41 (for that reason only) 2010-09-14T01:20:58 is there any strategic use for 0-growth planets? 2010-09-14T01:21:06 I don't ignore anything; I just put off very inefficent captures until later. 2010-09-14T01:21:15 sigh, well, you could amass a huge fleet there without my bot knowing it 2010-09-14T01:21:19 sigh: Staging grounds / waypoints. 2010-09-14T01:21:25 *** tapwater has quit IRC (Quit: tapwater) 2010-09-14T01:21:26 ah, haha 2010-09-14T01:21:29 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T01:21:29 a1k0n you perhaps? 2010-09-14T01:21:43 sigh: rwa might use one to do a bomb/cascade attack on two nearby planets simultaneously. 2010-09-14T01:21:56 most of the maps with a 0-growth center planet have larger planets right next to it so it kinda diminishes the need for staging grounds there 2010-09-14T01:22:21 Obeleh: in writing a strategy guide? maybe after i have a strategy.. 2010-09-14T01:22:37 it's more that i'm wary about telling my bot to pay units to grab one... if my enemy captured it, that's another story 2010-09-14T01:22:44 a1k0n: Agreed. 0-growth planets are very low on my list -- I basically only attack them if I contol all the other planets. 2010-09-14T01:22:59 just some basic things 2010-09-14T01:23:06 same here, only because i add 0.01 to growth when evaluating 2010-09-14T01:23:16 not the killer ones ;) 2010-09-14T01:23:33 a1k0n: is that to avoid div 0? :P 2010-09-14T01:24:03 no it's because i require a >0 value :) 2010-09-14T01:24:43 come to think of it i could have just changed > to >= 2010-09-14T01:24:44 Obeleh: I can probably write down a few things. Heck, the source code to my bot is open, so I don't mind divulging anything I'm doing right. 2010-09-14T01:24:45 I discovered the 0-growth planets when my bot started erroring with div0 errors :P 2010-09-14T01:25:32 sigh: welcome to the party :) 2010-09-14T01:25:37 :) 2010-09-14T01:25:39 Obeleh: You have a preferred format for submissions? 2010-09-14T01:26:07 ill start writing a bit and then show it here for ya to pitch in ok? 2010-09-14T01:26:30 haha, my two year old is learning chinese from ni hao kai lan 2010-09-14T01:26:42 Obeleh: Sure. I'm cool w/ that. 2010-09-14T01:26:44 *** chad has joined #aichallenge 2010-09-14T01:27:23 dangit, both bartwe.6 and deepblue have an edge on me 2010-09-14T01:27:41 dangit, my bot is obsolete and i'm not ready with a replacement 2010-09-14T01:28:35 *** Utkarsh has quit IRC (Ping timeout: 272 seconds) 2010-09-14T01:29:20 lol 2010-09-14T01:29:38 I am glad I am focusing on other things at the moment 2010-09-14T01:30:33 *** Queue29_2 has quit IRC (Quit: Leaving) 2010-09-14T01:33:09 Queue29_2, alright, if you're still around, I unpacked the JDK and compiled the MyBot myself--same thing, java bots all time out while my C++ bots could care less 2010-09-14T01:33:29 *** ajhager has joined #aichallenge 2010-09-14T01:33:35 *** Utkarsh has joined #aichallenge 2010-09-14T01:33:55 *** retybok has quit IRC (Quit: leaving) 2010-09-14T01:34:28 * whiteside makes the "damn" face 2010-09-14T01:35:54 why is it java bots are timing out? 2010-09-14T01:36:19 on the contest server or on tcpserver? 2010-09-14T01:36:34 that's what I'd like to know... oh sorry, I mean on my local machine 2010-09-14T01:36:43 weird! 2010-09-14T01:36:50 even the examples? 2010-09-14T01:37:23 yes, exactly the examples... I just nabbed my distro JDK to try compiling it myself, but alas 2010-09-14T01:37:40 my bots java just wondering if there is a flaw 2010-09-14T01:38:05 I dunno, but if it gives you any peace, I haven't been able to find any clues by searching the forum 2010-09-14T01:38:09 is there still a problem with non-english locales? 2010-09-14T01:38:18 snowplan, use a try catch 2010-09-14T01:38:20 ah yes, very possibly 2010-09-14T01:38:24 or rather, the problems that people were having seemed to get solved, and I can't find anyone else with my problem 2010-09-14T01:38:31 no I know my problem 2010-09-14T01:38:32 and log the error to file 2010-09-14T01:38:34 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T01:38:47 whiteside: do you have a non-english locale possibly? 2010-09-14T01:38:59 I just wanna know now if programming in another language is better 2010-09-14T01:39:03 no, it's definitely en_US 2010-09-14T01:39:05 or prefrence 2010-09-14T01:39:15 ok, I'm out of ideas then :/ 2010-09-14T01:39:15 I'm trying to get better in java 2010-09-14T01:41:49 *** ajhager has joined #aichallenge 2010-09-14T01:42:30 anybody have speed comparisons of java vs python? I was think about switching to python. 2010-09-14T01:42:45 For this, no? 2010-09-14T01:42:56 In general, the language shoot-out has some. 2010-09-14T01:45:19 *** syntaxglitch has quit IRC (Ping timeout: 240 seconds) 2010-09-14T01:46:09 *** chad has quit IRC (Quit: Page closed) 2010-09-14T01:50:14 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T01:50:45 chad: I wouldn't make that move for primarily performance reasons 2010-09-14T01:51:07 *** hornairs has quit IRC (Quit: hornairs) 2010-09-14T01:54:33 *** caytchen has joined #aichallenge 2010-09-14T01:54:43 rwa: chad is already gone 2010-09-14T01:54:52 yep, thanks 2010-09-14T01:56:01 *** ajhager has joined #aichallenge 2010-09-14T01:57:21 ...maaaaybe the java bots start up, take a look at my 256 mb ram, and decide that this ain't their scene, so they leave and go party elsewhere without informing me? It sounds like a stretch to me, but I literally have no other ideas 2010-09-14T01:58:22 whiteside, what does the error look like again? 2010-09-14T01:59:05 if I run playgame.jar with a java bot as either or both players, the java ones time out every time 2010-09-14T02:00:09 *** phreeza has joined #aichallenge 2010-09-14T02:00:14 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T02:00:43 whiteside, try putting everything in a giant try/catch, and catch OutOfMemoryError, and have it send a stack trace to your log file 2010-09-14T02:00:55 they're not throwing exceptions, and if there's any sort of relevant information in there anywhere that would provide more clues, I'm unsure about how to examine it... I don't have much experience with java 2010-09-14T02:01:08 ah, I'll see if I can do that 2010-09-14T02:01:15 stdout and stderror are redirected, so you won't see exceptions or anything 2010-09-14T02:01:52 oh, even with the 1.2 playgame? 2010-09-14T02:02:22 i think so.. it's how the bots communicate with the game manager 2010-09-14T02:02:28 *** biffabacon has quit IRC (Quit: Page closed) 2010-09-14T02:02:31 stderr isn't redirected 2010-09-14T02:02:36 but the PlayGame.jar swallows it 2010-09-14T02:03:33 *** GordonQuad has joined #aichallenge 2010-09-14T02:04:02 whiteside, if you haven't realized this yet, you can't use System.out.print() anywhere in your program either 2010-09-14T02:04:03 maybe try adding a maximum heap size limit for java, something like "java -Xmx:100m MyBot" 2010-09-14T02:04:45 I guess see if it will even start up with that limit first though :) 2010-09-14T02:05:12 oh, actually I have no idea, I'm using c++ on my personal bot... and for that guy, I just write a log file (but if and only if I detect a file named "safetolog" in the same directory) 2010-09-14T02:06:12 *** DirtyKeyboard has quit IRC (Quit: DirtyKeyboard) 2010-09-14T02:06:43 or maybe you should just buy more ram :p 2010-09-14T02:07:40 I know earlier in the day all the java bots were crashing on the contest server because a memory limit had been imposed 2010-09-14T02:08:15 and I think that limit may have been 700mb so you may not have any luck at all with 256 2010-09-14T02:09:23 but also I think that limit was imposed outside of the JVM without the JVM's knowledge so working with it may be feasible 2010-09-14T02:10:09 Janzert: no the limit was 196M of Address Space 2010-09-14T02:10:34 Me not ever working with JAva didn't realize that Java addresses 700+mb of shared libs 2010-09-14T02:10:35 ahh 2010-09-14T02:10:58 there are currently no memory limits (besides the fact the box currently only has 2gb of ram in it) 2010-09-14T02:13:24 does PlayGame-1.2.jar incorporate the latest battle resolution specs? 2010-09-14T02:13:54 I think someone wrote a C++ PlayGame that doesn't swallow stderr 2010-09-14T02:14:04 and gives you realtime viz 2010-09-14T02:14:43 sweet, took two matches against deepblue 2010-09-14T02:16:48 rwa: yes 2010-09-14T02:17:22 a1k0n: great, thanks. 2010-09-14T02:17:25 *** hellman has quit IRC (Remote host closed the connection) 2010-09-14T02:18:10 yeah i gotta get the c++ simulator, this java crap is ridiculous 2010-09-14T02:18:27 lol 2010-09-14T02:18:45 *** smf68 has quit IRC (Read error: Operation timed out) 2010-09-14T02:22:13 *** Queue29 has quit IRC (Quit: Leaving) 2010-09-14T02:23:19 *** jukkerknott has joined #aichallenge 2010-09-14T02:23:43 I think I'm just going to change a few lines and compile those java rapscallions as prim and proper c++ bots (no offense, java guys) 2010-09-14T02:24:23 *** jeyk has joined #aichallenge 2010-09-14T02:24:44 *** jeyk is now known as jey 2010-09-14T02:27:28 *** genericbob has joined #aichallenge 2010-09-14T02:28:27 moin 2010-09-14T02:28:58 *** ratatata has joined #aichallenge 2010-09-14T02:33:33 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-09-14T02:35:30 *** caytchen has quit IRC (Ping timeout: 265 seconds) 2010-09-14T02:36:06 *** jukkerknott has quit IRC (Quit: Page closed) 2010-09-14T02:37:54 Qt game similator ftw! 2010-09-14T02:38:17 For those looking for it: http://github.com/greghaynes/QTcpPlanetWars 2010-09-14T02:38:28 a1k0n: Do you work at yahoo hq? 2010-09-14T02:40:47 *** jey has quit IRC (Remote host closed the connection) 2010-09-14T02:41:02 *** Any_Key has joined #aichallenge 2010-09-14T02:42:06 huh. someone still hasn't added the Production method to the PlanetWars class in the top level source files of the java starter package (but it's included in the PlanetWars class in the example_bots directory) 2010-09-14T02:45:26 whiteside: probably because it's just some simple iteration 2010-09-14T02:47:38 granted it's trivial, but the two (presumably) same files in the package are inconsistent 2010-09-14T02:48:34 greghaynes: in sunnyvale, yeah 2010-09-14T02:48:55 if someone copied the bot files to the next directory up to use as a basis for their own bots, they wouldn't compile out-of-the-box unless they overwrote the planetwars class with the one from the bot directory 2010-09-14T02:49:07 Nice, I have a friend who works there - seanbruno 2010-09-14T02:49:11 Hes newish 2010-09-14T02:50:03 ah. he is upstairs from me it seems 2010-09-14T02:50:14 wow he reports directly to filo? 2010-09-14T02:50:24 * greghaynes shrugs 2010-09-14T02:50:29 Went to school with him 2010-09-14T02:50:42 *** mossplix has joined #aichallenge 2010-09-14T02:50:45 Hes a smart guy though 2010-09-14T02:51:10 a1k0n: you're at the office? 2010-09-14T02:51:11 works on freebsd 2010-09-14T02:51:16 Yep 2010-09-14T02:51:19 rwa: well, not right now 2010-09-14T02:51:43 ok, thought you were checking your directory or something 2010-09-14T02:51:58 does yahoo encourage discourage you from competing this year? 2010-09-14T02:52:01 i am checking the employee directory yes 2010-09-14T02:52:13 (or) 2010-09-14T02:52:24 or neither 2010-09-14T02:52:26 yahoo as a corporate entity has no opinion. my boss thinks it's great, is considering entering, and my coworker is competing 2010-09-14T02:53:04 *** jey has joined #aichallenge 2010-09-14T02:53:19 *** jukkerknott has joined #aichallenge 2010-09-14T02:55:48 that's cool 2010-09-14T02:58:15 greghaynes: even though it says "bot running", I'm timing out at start, so I'm assuming I provide wrong input.. what should I toss to 'locate your bot' ? 2010-09-14T02:58:55 hrmm 2010-09-14T02:59:16 If it says bot runnig then it shouldnt be an issue with passing the correct bot 2010-09-14T02:59:29 Unless you passed something executable that wasnt your bot 2010-09-14T02:59:30 it there a comman-line way of supplying the bot? 2010-09-14T02:59:47 No 2010-09-14T03:00:00 Before you make the first move it times out? 2010-09-14T03:00:07 greghaynes: correct 2010-09-14T03:00:14 What language? 2010-09-14T03:00:17 btw: I use java 2010-09-14T03:00:33 it's always java 2010-09-14T03:00:43 * greghaynes downloads starter pack to test 2010-09-14T03:01:57 *** Theobon has quit IRC (Ping timeout: 276 seconds) 2010-09-14T03:02:12 oo, it doesnt like java for some reason 2010-09-14T03:02:35 Give me a few to figure this out 2010-09-14T03:05:27 *** GordonQuad has quit IRC (Quit: Ухожу я от вас) 2010-09-14T03:12:51 What commands do you give to compile the java sstarter package, and how do you exec it? 2010-09-14T03:13:18 I did javac MyBot.java, it made the .class's, I cant ./MyBot.class or java MyBot.class though 2010-09-14T03:15:01 greghaynes: "java MyBot" is the bot execution 2010-09-14T03:15:21 greghaynes: How do you enter a password? Is that capability not written yet? 2010-09-14T03:15:52 aha 2010-09-14T03:16:12 genericbob: Its not written, itll take me all of two seconds to add though 2010-09-14T03:17:48 *** sigh_ has joined #aichallenge 2010-09-14T03:17:49 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-09-14T03:18:00 *** Theobon has joined #aichallenge 2010-09-14T03:18:07 greghaynes: any idea if this will fly on OS X? 2010-09-14T03:18:17 rwa: It should, its qt only 2010-09-14T03:18:22 cool 2010-09-14T03:19:01 anyone else finding it a lot harder than they expected to get a non naive bot working very well? 2010-09-14T03:19:19 antimatroid: Non-native? 2010-09-14T03:19:21 *** Utkarsh has quit IRC (Ping timeout: 264 seconds) 2010-09-14T03:19:30 naive* 2010-09-14T03:19:43 dunno, I think my bot is kinda naive 2010-09-14T03:19:46 What do you mean by non-naive? 2010-09-14T03:19:47 *** genericbob has quit IRC (Ping timeout: 252 seconds) 2010-09-14T03:19:48 define non-naive? 2010-09-14T03:20:16 well i've been trying to start actually implementing stuff, but it all seems to suck compared to even just adding moves onto a priority queue with a very simple payoff function 2010-09-14T03:21:55 antimatroid: My bot has been getting worse as I add "better" stuff too. 2010-09-14T03:22:04 I guess it depends what kind of "stuff" you're going after 2010-09-14T03:22:13 it's not hard I think but without extensive planning beforehand, the behaviour is not always the expected.. I just remembered why my bot was sending kamikaze attacks.. stupidest thing 2010-09-14T03:22:24 antimatroid: But, I think a lot of it is how flawed my evaluation function is in reality. 2010-09-14T03:23:01 yea i know it's my strategy being flawed 2010-09-14T03:23:13 my point was that it's a lot harder than i initially thought to get it playing quite well 2010-09-14T03:23:22 i am slowly working out bugs 2010-09-14T03:23:33 antimatroid: I think if my evaluation function (for a planet) was better, the techniques I am using would show better. 2010-09-14T03:23:34 *** Utkarsh has joined #aichallenge 2010-09-14T03:23:39 small bugs can kill you in this game 2010-09-14T03:23:55 off-by-one is the difference between winning and losing if you fail to capture a planet 2010-09-14T03:23:58 I had a bug which made my bot do better, once 2010-09-14T03:24:04 of course, that is rare 2010-09-14T03:24:12 antimatroid: Yeah. It's not harder than I initially thought, but it is getting significantly harder as time goes by. 2010-09-14T03:24:30 antimatroid: I feel other bots are getting better much more quickly than my bot. 2010-09-14T03:24:35 I already implemented a couple "improvements" that made my bot worse 2010-09-14T03:24:36 well, if you are completely stuck for ideas watch some of the better bots like bartwe or deepblue and see what they do 2010-09-14T03:25:09 i'm all good with ideas atm :) 2010-09-14T03:25:29 aw, he left, I just added passwords for him 2010-09-14T03:25:31 well then, the only thing to it is implement them and see 2010-09-14T03:27:22 http://pastebin.com/Z9Ggi4PE bss03 and a1kon or anyone who wants to help. mailme to sjuulj at hotmail 2010-09-14T03:28:31 *** Lithosphere has joined #aichallenge 2010-09-14T03:29:18 *** Palmik has joined #aichallenge 2010-09-14T03:31:06 *** soswow has joined #aichallenge 2010-09-14T03:31:11 i hope its good and that people want to write a piece for it 2010-09-14T03:31:15 hi to all. 2010-09-14T03:31:21 hi 2010-09-14T03:32:46 Is there official twitter account of ai-contest? 2010-09-14T03:33:03 Where we can get latest news, updates, important info itc ... 2010-09-14T03:33:38 Obeleh: Hrm, that seems fine. I'm not great at finding problems though. 2010-09-14T03:34:02 Obeleh: I'm a bit more goal-oriented. Is there anything in particular you'd like me to try and exapnd on? 2010-09-14T03:34:14 *** sigh_ has quit IRC (Remote host closed the connection) 2010-09-14T03:34:20 you know i love it when you talk to a dedi hosting company and they offer to throw in 5tb of extra b/w a month for free 2010-09-14T03:35:37 sweet, hopefully it's for running a game server... 2010-09-14T03:36:43 *** delt0r_ has joined #aichallenge 2010-09-14T03:36:47 ratatata: I think I fixed it 2010-09-14T03:36:49 ratatata: update 2010-09-14T03:37:19 ? 2010-09-14T03:37:20 delt0r___: hi 2010-09-14T03:37:28 rwa looking at similar priced but better servers for the contest server itself 2010-09-14T03:37:40 as a replacement? 2010-09-14T03:37:43 yea 2010-09-14T03:37:47 nice 2010-09-14T03:37:53 Currently 2gb of Ram/p4 2010-09-14T03:38:11 *** delt0r___ has quit IRC (Ping timeout: 265 seconds) 2010-09-14T03:38:11 I made an offer to the admins for a 4 core 64bit machine with 6gb of ram on a >1 gbit connection... 2010-09-14T03:38:11 just something usefull bss03 2010-09-14T03:38:14 So well see 2010-09-14T03:38:38 we need about 10 of those and we'll be good 2010-09-14T03:39:05 Really? Theres a 16core I can probably get but I didnt think it was needed... 2010-09-14T03:39:34 greghaynes: confirmed, nice and quick fix, thanks 2010-09-14T03:39:46 ratatata: It works!? 2010-09-14T03:39:58 greghaynes: positive 2010-09-14T03:40:03 nice 2010-09-14T03:40:18 greghaynes: you can't run a node if you wan't to compete 2010-09-14T03:40:21 I had to do a if path.endsWith(".clas") ..domagic 2010-09-14T03:40:23 want* 2010-09-14T03:40:38 dstufft: There are other people who will admin them for me 2010-09-14T03:40:52 er, for the competition 2010-09-14T03:41:57 greghaynes: ah. Well that's up to jeff then I guess, or amstan 2010-09-14T03:42:00 alright, I need timeout code 2010-09-14T03:42:09 anybody done something with timeout checking in C++ that worked well? 2010-09-14T03:42:33 i mean more machines arn't strictly required, but if we go distributed, then the more machines/power we got the more games we can run concurrently 2010-09-14T03:42:50 rwa: isn't there something to get the current milliseconds since january1st 1970 as usual? 2010-09-14T03:43:04 I was sort of hoping this could keep from needing to go distributed 2010-09-14T03:43:32 greghaynes: honestly, pesonally i don't think we need to go distributed at all 2010-09-14T03:43:37 but they are talking about it 2010-09-14T03:43:39 ratatata: yea, thats a very expensive call though 2010-09-14T03:43:56 dstufft: who is they? 2010-09-14T03:44:25 contestants? 2010-09-14T03:44:29 soswow: people working on the code/admins, are talking about the possiblity of doing it, not that we are going to though 2010-09-14T03:44:37 One of the admins seems to really be pushing for it 2010-09-14T03:45:01 IMO it seems likely to cause a lot of issues 2010-09-14T03:45:58 I don't tihnk it's needed really. I think it was thrown out as an idea when the server kept getting steamrolled 2010-09-14T03:46:02 If theres no other option though, its understandable 2010-09-14T03:46:06 dstufft: why don't you think you need it? And why does it could issues? It's seems atomic action (game between 2 players). Only concurency problem is DB writing score, but it's simple one, isn' it? 2010-09-14T03:46:27 dstufft: I think so also 2010-09-14T03:46:32 To both points 2010-09-14T03:47:12 I really think distributed is essential. The sample sizes needed to distinguish closely matched bots are going to be very large. The current rate of games is like two orders of magnitude off from where it should be. 2010-09-14T03:47:17 *** ss_ has quit IRC (Ping timeout: 252 seconds) 2010-09-14T03:47:33 soswow: i'm not arguing that it wouldn't allow more games to be played concorently, but I think that a) it add's extra complexity to an already rube goldber esque system, and b) we are unlikely to get many nodes since hosting a node disqualifies you from competeting 2010-09-14T03:47:56 we are currently running 14 games per minute, on a shitbox p4 2010-09-14T03:48:15 dstufft: isn't it to small number? 2010-09-14T03:48:20 Ideally it seems you would want a high game rate and a very level game field, the distributed option makes the level game field a difficult thing to assure 2010-09-14T03:48:22 14 games a minute is horrible 2010-09-14T03:48:23 greghaynes: currentMillis is expensive? between 2 calls of it, I do a whole bunch of stuff and return difference as low as 1-2ms.. is that expensive? (considering that I spent most of that 1-2ms for other calculations) 2010-09-14T03:48:24 *** liehann has joined #aichallenge 2010-09-14T03:48:39 dstufft: And it's only a beginning, people will come I think 2010-09-14T03:49:10 hi 2010-09-14T03:49:26 ratatata: Its all relative, that seems like its a higher level language call so they probably abstract it, in C API land it requires a context switch which is expensive 2010-09-14T03:49:39 dstufft: Better say how many games average contestant have per minute. 1 in half hour, in hour? 2010-09-14T03:49:45 it's not so grate for us 2010-09-14T03:49:49 there's also the trust issue, Some people expressed concern over their bots being passed to other systems not controlled by us 2010-09-14T03:49:49 * great 2010-09-14T03:50:14 dstufft: Em. Who talk about system not controlled by you? 2010-09-14T03:50:22 greghaynes: I'll keep that in mind, thanks 2010-09-14T03:50:24 dstufft: Make it controlled by you! =) It's simple 2010-09-14T03:50:52 *** vikhyat has joined #aichallenge 2010-09-14T03:50:57 soswow: currently games are chosen completly randomly, on the todo list is modifying the scheduler so that games are scheduled based on narrowing the confidence gap of the bayeselo ranking 2010-09-14T03:51:23 ratatata: The server is a linux box, so if you can put a thread to sleep for so many milisecs (using select or some other call) and have it signal to the other thread when it wakes up, it should be farly efficient due to preemption 2010-09-14T03:51:36 dstufft: yeah, I know that. But Slow game rate is the problem. 2010-09-14T03:51:54 greghaynes: if I recall you're not allowed to spawn threads, but i may be wrong on that rule 2010-09-14T03:52:00 ah 2010-09-14T03:52:04 That would be an issue then 2010-09-14T03:52:10 dstufft: If there is 1000 players and there is only 14 games per minute, what is time for one game for each player? 2010-09-14T03:53:02 14 games a minute is 840 games an hour, each game has 2 players so 1680 players involved each hour 2010-09-14T03:54:11 so 1.68 games an hour on average assuming you schedule it so that it goes round robin (which none of the proposed systems are going to do) 2010-09-14T03:54:16 per player 2010-09-14T03:54:23 dstufft: does it mean ~ each player have one game every half an hour? 2010-09-14T03:54:36 soswow: assuming a round robin schedular 2010-09-14T03:54:45 *** jaspervdj has joined #aichallenge 2010-09-14T03:54:54 we don't have a round robin schedular though 2010-09-14T03:55:01 the current game schedular is completly random 2010-09-14T03:55:16 so people can go hours without a game, and then can get 5 games in a row 2010-09-14T03:55:18 Yea, so that would be the average 2010-09-14T03:55:35 And you get a gaussian (sp?) dist around that 2010-09-14T03:55:42 that is being changed, but the average will still remain the same 2010-09-14T03:55:51 I sucked at prob/stats 2010-09-14T03:55:52 dstufft: Just confirm my idea that me as a player have only 1 game per half an hour? (or something like this) 2010-09-14T03:56:06 soswow: it's not that simple 2010-09-14T03:56:13 soswow: basically yes 2010-09-14T03:56:18 *** liehann has quit IRC (Quit: Page closed) 2010-09-14T03:56:24 dstufft: But it's very very slow! =( 2010-09-14T03:56:56 They arent near that right now...I thing its due to running the web and mysql servers on the same box 2010-09-14T03:57:03 When I upload my new engine, I need to wait half a day 12H to play 24 games ... 2010-09-14T03:57:08 With some inefficient queries 2010-09-14T03:57:32 greghaynes: actually i havn't checked into it, but supposidly we are currently playing about 14 games a miinute right now 2010-09-14T03:57:43 Maybe it's ok for such competitions, btw! I am not so familiar with such things ... 2010-09-14T03:57:44 Definitely not 2010-09-14T03:57:50 Ive played <20 all day 2010-09-14T03:57:53 But it's just feels slow ==) 2010-09-14T03:58:09 what worse than feeling slow is that it will not distinguish good bots from bad before the cows come home 2010-09-14T03:58:10 i'll have to check into it myself then 2010-09-14T03:58:10 Mt last game was hours ago 2010-09-14T03:58:15 er, My 2010-09-14T03:58:38 Better pairing would virtually speed things up as well 2010-09-14T03:59:05 that is being worked on too 2010-09-14T03:59:09 You you are constantly matched against people 400 below you, climbing up is very difficult :) 2010-09-14T03:59:12 Separating the web and bot challenge boxes would help a lot 2010-09-14T03:59:26 at the moment pairing is completly random 2010-09-14T03:59:28 Palmik: Yeah, they working on better paring 2010-09-14T03:59:31 dstufft, yeah, I heard that :) 2010-09-14T04:00:05 *** ajhager has joined #aichallenge 2010-09-14T04:00:40 It's just that it might increase the effectivness more then ano or two adition boxes would 2010-09-14T04:00:48 imo 2010-09-14T04:01:24 dstufft: Ok. My final thought: If it were 1 game per minute for each player - that would be just awesome! =) 2010-09-14T04:01:25 *one or two additional 2010-09-14T04:01:44 Yes, you would need a very large number of boxes to account for the rank accuracy gained by a better scheduler 2010-09-14T04:01:53 Youre correct 2010-09-14T04:02:29 Especially when youre at the extremes of the ranking 2010-09-14T04:03:58 Also, what is the rating you start with? 2010-09-14T04:05:23 *** _jsimpson <_jsimpson!62b9eca7@gateway/web/freenode/ip.98.185.236.167> has joined #aichallenge 2010-09-14T04:06:45 select count(game_id) from games where timestamp >= DATE('2010-09-14'); 2010-09-14T04:06:51 1800 games 2010-09-14T04:07:01 1807 now 2010-09-14T04:07:25 dstufft: are you counting games per minute? 2010-09-14T04:07:40 so it's 4am 2010-09-14T04:07:41 @15 games per min and 500 players it would take almost 6 days for everyone to play everyone... 2010-09-14T04:07:42 delt0r_: Error: "15" is not a valid command. 2010-09-14T04:07:56 oh there is a bot 2010-09-14T04:07:59 Theres 1k players 2010-09-14T04:08:10 1800 / 4 / 60 = 7.5 2010-09-14T04:08:14 so 2010-09-14T04:08:19 greghaynes: and it will goes bigger 2010-09-14T04:08:22 oh -then thats x4 = 24 days 2010-09-14T04:08:23 in the last 4 hours we've averaged 7.5 games a minute 2010-09-14T04:08:31 Its also random assignment, so its in terms of zindex ;) 2010-09-14T04:08:56 dstufft: US? 4am? didn't sleep yet? 2010-09-14T04:08:58 i don't profess to understand bayeselo, so i don't know what it needs 2010-09-14T04:08:58 well i wasn't pointing it was a good idea --i am pointing out that random/all pairings won't work 2010-09-14T04:09:11 time 2010-09-14T04:09:15 delt0r_: yeah, and just once. Good bots will be like 55/45, and will need many many samples to know which is better. 2010-09-14T04:09:18 Its used in chess a lot 2010-09-14T04:09:23 delt0r_: and distribution =) 2010-09-14T04:09:32 Yes, and I was trying to show its actually worse than what you say due to the distribution 2010-09-14T04:10:10 also remember that the current rankings are just there to see how well you're doing, atleast that's my understanding 2010-09-14T04:10:29 i think when we "end" it, we are stopping new submissions, and thenr unning the bots against each other 2010-09-14T04:10:38 for idk how long 2010-09-14T04:10:47 bayeselo also doesn't really stabilize well. It only takes one outlier to give you a big jump 2010-09-14T04:11:01 Hopefully well get to see how this 4core box works, and if it doesnt ill be more than happy to obtain the 16core 2010-09-14T04:11:31 mmm its not like a 16 core box? 2010-09-14T04:11:47 huh? 2010-09-14T04:11:55 also soswow yea its 4am here, but i went to sleep around 5pm 2010-09-14T04:12:04 just woke up around 1am 2010-09-14T04:12:12 dstufft: Yeah, But before stopping submitions players it would be nice for each player have real understanding of wellness of his bot. 2010-09-14T04:12:19 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T04:12:20 the games --for 7.5 games per min--- i was assuming its already on a 4-16 core box 2010-09-14T04:12:36 no 2010-09-14T04:12:39 The persons box it is, is actually the secratary of x.org ;) 2010-09-14T04:12:42 it's on a p4 shitbox 2010-09-14T04:12:44 i would have assumed 16 core... since we have a few 16 core boxes just sitting aroudn 2010-09-14T04:12:58 oh 2010-09-14T04:13:16 I was talking to quadranet earlier, and I'm going to suggest moving to a new server 2010-09-14T04:13:17 *** m4rw3r has joined #aichallenge 2010-09-14T04:13:29 off of a p4 and onto a C2d or a i3 2010-09-14T04:13:30 well then... some of the issues do make sense now... i was wondering 2010-09-14T04:13:38 for the same price as the current box 2010-09-14T04:13:39 roughly 2010-09-14T04:14:02 *** ajhager has joined #aichallenge 2010-09-14T04:14:10 dstufft: I emailed j3c and amstan about the x4, so we should have an aswer tomorrow 2010-09-14T04:14:10 the game runner communicates with the tournament_manager via ssh 2010-09-14T04:14:27 so splitting that out to it's own box isn't terribly hard 2010-09-14T04:16:16 I havn't looked really close at the bot running code, it might not be running paralell bot fights i don't know 2010-09-14T04:16:17 *** _jsimpson <_jsimpson!62b9eca7@gateway/web/freenode/ip.98.185.236.167> has quit IRC (Quit: Page closed) 2010-09-14T04:16:35 hah, it better be 2010-09-14T04:17:09 dstufft: It would be easy to do... just have the code running the bots run twice... only need to "sync" the results 2010-09-14T04:17:17 *** Nuriaion has joined #aichallenge 2010-09-14T04:17:47 @repo 2010-09-14T04:17:48 dstufft: repo = http://code.google.com/p/ai-contest/ 2010-09-14T04:17:55 if a match is duplicated (improbable) its not really the end of the world 2010-09-14T04:18:19 yea 2010-09-14T04:18:32 atm we are pegging the cpu of hte server 2010-09-14T04:18:37 pretty constantly 2010-09-14T04:18:44 well 2010-09-14T04:18:53 loads in the 2-3's 2010-09-14T04:19:11 if find that load number very uninformative 2010-09-14T04:19:21 Between a significant # of php requests per sec (with a sig. number of mysql queries each) and running the bots, im not surprised 2010-09-14T04:19:27 our cluster head node can be at .5 and its slow to respond 2010-09-14T04:19:39 haha 2010-09-14T04:19:41 other times it can be at 15 and its really snappy 2010-09-14T04:19:52 i'm hoping that if they let me switch to quadranet, and let me set itup 2010-09-14T04:19:55 Thats no good, load > .5 2010-09-14T04:19:59 then i can get rid of apache 2010-09-14T04:20:04 and get nginx on board 2010-09-14T04:20:06 with php-fpm 2010-09-14T04:20:11 and varnish 2010-09-14T04:20:13 in front of it 2010-09-14T04:20:23 or memcache 2010-09-14T04:20:39 cache the web requests somewhere 2010-09-14T04:20:49 should ease up the www portion of the box 2010-09-14T04:20:51 Just wondering why the hardware is ...... so thin? is this more people than you were expecting? 2010-09-14T04:20:53 That helps some but in my experience its about a 1:100 cpu time usage between server and php itself 2010-09-14T04:21:12 delt0r_: i just started helping like 2-3 days ago 2010-09-14T04:21:23 so I don't know 2010-09-14T04:21:28 I wouldn't have gotten a p4 2010-09-14T04:21:31 ic 2010-09-14T04:24:27 *** Naktibalda has joined #aichallenge 2010-09-14T04:25:16 *** Zeiris has joined #aichallenge 2010-09-14T04:25:30 I did as about twitter earlier. didn't get answer. Why don't make twitter account for contest? Where you can latest news, updates, important info itc. Like HTC did =) : "Current power is 3.7 TeV", "We are reaching 10 TeV, hold you pens!" =) But games per minute in your case =) 2010-09-14T04:26:20 * as => ask 2010-09-14T04:26:28 soswow: i dunno 2010-09-14T04:26:35 im just a code monkey and computer janitor 2010-09-14T04:26:51 *** rwa has quit IRC (Quit: rwa) 2010-09-14T04:27:22 dstufft: ok =) but if you involved, you can pass it further =) 2010-09-14T04:27:33 Or I'll better write about it on forum 2010-09-14T04:27:43 post it on the forum please 2010-09-14T04:35:48 *** wartalker has joined #aichallenge 2010-09-14T04:37:52 ive posted a strategy guide on the forums. any comments are welcome 2010-09-14T04:39:00 *** ajhager has quit IRC (Ping timeout: 276 seconds) 2010-09-14T04:39:50 *** ajhager has joined #aichallenge 2010-09-14T04:39:59 *** bss03 is now known as bss03-AFK 2010-09-14T04:40:15 *** td123 has quit IRC (Ping timeout: 240 seconds) 2010-09-14T04:41:05 ...is there no ExpandBot? 2010-09-14T04:41:47 whiteside: you can compile it to .class yourself and use it 2010-09-14T04:42:28 *** kozlovsky has joined #aichallenge 2010-09-14T04:42:29 hey bapho :) still c++? 2010-09-14T04:42:55 ya ive been working on mine, learning curve has been unbelievable the last 24h 2010-09-14T04:43:09 er, what I mean to say is, the ExpandBot.java source is identical to BullyBot.java ... so it won't even compile, much less be different from BullyBot if it did 2010-09-14T04:43:38 whiteside: well in that case they obviously never finished making one :p 2010-09-14T04:43:50 bapho. check the forums. ive posted a strategy guide :) 2010-09-14T04:44:02 i'm just waiting for my latest entry to play a game 2010-09-14T04:44:20 *** Vi0 has joined #aichallenge 2010-09-14T04:44:26 cool, ill take a read 2010-09-14T04:45:51 *** sigh has joined #aichallenge 2010-09-14T04:48:41 *** ratatata has quit IRC (Quit: Page closed) 2010-09-14T04:48:42 Does anyone have specialised knowledge in what an intelligent game scheduling algorithm would entail? 2010-09-14T04:48:59 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T04:49:10 Rather, what algorithm would be suitable for this problem. 2010-09-14T04:49:12 greghaynes: that sounds like a pretty general term 2010-09-14T04:49:40 basically: a smart bot 2010-09-14T04:49:49 *** ajhager has joined #aichallenge 2010-09-14T04:50:00 Well, at the moment the two opponents are chosen at random, I know how to fix this I just do not know the algorithm to correctly chose the two opponents 2010-09-14T04:50:27 *** rafsaoken has joined #aichallenge 2010-09-14T04:50:37 *** sinan has joined #aichallenge 2010-09-14T04:50:55 Obviously, anything better than random (which shouldnt be difficult) is better than our current situation 2010-09-14T04:53:45 I was considering creating a gaussian distribution centered over player one proportianal to the total number of opponents and determining an opponent out of that 2010-09-14T04:53:51 greghaynes, a1k0n proposed taking X people above you and X under you and random from them. As for picking the first player, you can go for the one who did not play recently. That should be more effective then the current system. 2010-09-14T04:53:54 Im not sure this is the best tactic though 2010-09-14T04:54:28 *** Utkarsh has quit IRC (Ping timeout: 276 seconds) 2010-09-14T04:54:37 Sounds similar 2010-09-14T04:54:45 Like, the longer you bot did not play, the more likely it is to be picked... 2010-09-14T04:54:59 That only goes for the first player 2010-09-14T04:55:14 The second should be random from the given ranking margin 2010-09-14T04:55:15 I do know that bayeselo has a confidence number 2010-09-14T04:55:25 aha 2010-09-14T04:55:25 *** Yoshi-TS4 has joined #aichallenge 2010-09-14T04:55:40 and someone was talking about picking out fights that would narrow the confidence gap 2010-09-14T04:55:59 currently your ranking is the mean of the range that the bayeselo thinks your ELO would be 2010-09-14T04:56:07 Palmik: I was proposing to make the first player a choice out of sequentually following user id's 2010-09-14T04:56:19 Then the opponent would be decided inteligentlyu 2010-09-14T04:56:21 the more games you play, the narrower that gaps become 2010-09-14T04:56:24 dstufft, what is the confidence based on, out of interest :) 2010-09-14T04:56:35 Palmik: that's part of bayeselo i think 2010-09-14T04:56:45 bayeselo is a system that tries to guess what your ELO would be 2010-09-14T04:56:48 dstufft: Yes, that seems correct, I just dont know of an algo that does this 2010-09-14T04:56:59 I would imagine there is someone around here who has studied this... 2010-09-14T04:57:04 *** staii has joined #aichallenge 2010-09-14T04:57:13 i'm trying to find the issue where they were talking about it 2010-09-14T04:57:36 *** sinan has quit IRC (Ping timeout: 265 seconds) 2010-09-14T05:00:10 *** ajhager has quit IRC (Ping timeout: 252 seconds) 2010-09-14T05:00:34 *** mega1 has joined #aichallenge 2010-09-14T05:01:50 *** ajhager has joined #aichallenge 2010-09-14T05:05:36 *** retybok has joined #aichallenge 2010-09-14T05:06:09 Is there now a way to get compilation errors when entering the official challenge? 2010-09-14T05:06:54 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T05:07:10 *** jey has quit IRC (Remote host closed the connection) 2010-09-14T05:07:54 What is the meaning of the "+" and the "-" columns on http://www.benzedrine.cx/planetwars/getratings ? 2010-09-14T05:09:30 i don't know i give up trying to find the discussion 2010-09-14T05:10:06 heh 2010-09-14T05:10:44 *** caytchen has joined #aichallenge 2010-09-14T05:12:14 *** bss03-AFK has quit IRC (Remote host closed the connection) 2010-09-14T05:14:24 *** dmj111 has quit IRC (Ping timeout: 245 seconds) 2010-09-14T05:15:39 *** ajhager has joined #aichallenge 2010-09-14T05:16:24 *** bss03 has joined #aichallenge 2010-09-14T05:21:18 *** ajhager has quit IRC (Ping timeout: 265 seconds) 2010-09-14T05:23:35 *** ajhager has joined #aichallenge 2010-09-14T05:25:56 *** eburnette has quit IRC (Ping timeout: 252 seconds) 2010-09-14T05:26:51 *** kozlovsky has quit IRC (Quit: Page closed) 2010-09-14T05:27:21 *** kozlovsky has joined #aichallenge 2010-09-14T05:30:21 *** otherAntimatroid has joined #aichallenge 2010-09-14T05:34:28 *** zerd_ has joined #aichallenge 2010-09-14T05:34:50 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T05:35:05 *** Utkarsh has joined #aichallenge 2010-09-14T05:35:35 *** ajhager has joined #aichallenge 2010-09-14T05:36:01 *** zerd has quit IRC (Ping timeout: 276 seconds) 2010-09-14T05:36:10 *** ratatata has joined #aichallenge 2010-09-14T05:36:31 fml i think that my bot can still get killed by the default one 2010-09-14T05:36:41 im so shit at this 2010-09-14T05:38:38 i wish i could work out how to use pw.Distance() as that would help heaps 2010-09-14T05:39:12 what's wrong with it? 2010-09-14T05:39:29 which bot is default one? 2010-09-14T05:39:30 everything, because i faul 2010-09-14T05:39:46 fail, i cant even write fail withought failing :( 2010-09-14T05:39:50 lol http://ai-contest.com/visualizer.php?game_id=4464921 2010-09-14T05:40:10 rofl 2010-09-14T05:40:17 atleast mine doesnt to that 2010-09-14T05:42:19 pw.Distance(source.PlanetID,destinmation.PlanetID) 2010-09-14T05:42:49 .PlanerID() 2010-09-14T05:43:03 yea i know how to do that just not include the output into my other calculations 2010-09-14T05:43:25 danielvf: 2 stupids bots it seems 2010-09-14T05:43:33 int distance = pw..... 2010-09-14T05:44:00 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T05:44:04 if (distance < 15) bla 2010-09-14T05:44:49 ok ill see what i can do with that 2010-09-14T05:44:52 thanks 2010-09-14T05:44:53 *** ratatata has quit IRC (Quit: Page closed) 2010-09-14T05:45:05 edcba: Yup 2010-09-14T05:45:14 just curious how old are you? 2010-09-14T05:45:26 me? 2010-09-14T05:45:34 yea 2010-09-14T05:45:36 18 2010-09-14T05:45:45 i just dont have too much programming experance 2010-09-14T05:45:47 *** ajhager has joined #aichallenge 2010-09-14T05:46:04 try some tutorial first man 2010-09-14T05:46:49 some basic c++ or preferable an easier language 2010-09-14T05:47:04 *** acieroid` has joined #aichallenge 2010-09-14T05:47:20 *** caytchen has quit IRC (Ping timeout: 240 seconds) 2010-09-14T05:47:50 i know basic c++ already. and i know it much more than anything else and i like working in c++ 2010-09-14T05:50:04 Stick in there, im one of the few where I work who actually likes c++ 2010-09-14T05:50:10 *** hellman has joined #aichallenge 2010-09-14T05:52:07 haha the db seems to have stuffed up and make 2 copies of a game but one of them says i played myself :p 2010-09-14T05:52:18 and that i won :p 2010-09-14T05:52:20 *** m4rw3r has quit IRC (Ping timeout: 252 seconds) 2010-09-14T05:52:27 URL? 2010-09-14T05:52:58 c++ is not bad. But its not what i would call a good choice for a first language. 2010-09-14T05:53:17 I started with Z80 Assembly ! 2010-09-14T05:53:21 then lisp 2010-09-14T05:53:41 lisp was hard 2010-09-14T05:53:44 i didnt actually play myself, its just a copy of another game. http://ai-contest.com/visualizer.php?game_id=4464895 but if you look here http://ai-contest.com/profile.php?user_id=6239 the bottom 2 are both the same game 2010-09-14T05:54:15 also thats the fml game where i lose to what im pretty sure is the default bot :( 2010-09-14T05:54:33 bah, I learned to program using C++.. it makes you really learn memory managemen 2010-09-14T05:54:38 i 2010-09-14T05:55:02 why is this user playiong itself so much for the ranking ?http://www.ai-contest.com/profile.php?user_id=5837 2010-09-14T05:55:05 greghaynes: as does C or assmbley... I thinking more of OO/class and templates 2010-09-14T05:55:13 bartwe <3 your bot its sexy to watch 2010-09-14T05:55:28 No one just learning C++ writes their own templates ;) 2010-09-14T05:55:30 Its a lot to start with since a lot of code in c++ will all of that 2010-09-14T05:55:34 Although templates ftw 2010-09-14T05:55:37 Baohiomet: thanks :) 2010-09-14T05:55:54 greghaynes: But when you look at code examples..... 2010-09-14T05:56:02 Baphomet: now if only if it won all rounds .... :) 2010-09-14T05:56:09 its there and kinda forced on new coders 2010-09-14T05:56:29 Yea...if youre using C++ as a learning step you should really be using Qt ;) 2010-09-14T05:56:59 meh, i'd advise a new programmer to learn something basic, python or java 2010-09-14T05:57:02 and Boost + STL >) 2010-09-14T05:57:09 !@#$! boots 2010-09-14T05:57:14 er. boost 2010-09-14T05:57:23 bartwe: me too--but you can start with c/c++ 2010-09-14T05:57:23 hello 2010-09-14T05:57:30 have you a game engine in python ? 2010-09-14T05:57:34 STL is crap and boost is just overtemplated magic 2010-09-14T05:57:40 organisers about ? 2010-09-14T05:57:44 its hard though---but you get the idea of a reference better than if you start with python 2010-09-14T05:57:45 hello greghaynes 2010-09-14T05:57:52 Hey 2010-09-14T05:57:53 templates rock, the compilation model suck 2010-09-14T05:57:58 did your tcp client been upgraded Q 2010-09-14T05:58:03 ? 2010-09-14T05:58:13 greghaynes: My point exatly ... hence evil eyes >) 2010-09-14T05:58:15 elijahbal: Yep, it does real time visualizatin now 2010-09-14T05:58:28 :) 2010-09-14T05:58:40 ok 2010-09-14T05:58:44 Will try it. 2010-09-14T05:58:45 Baphomet: does this look bugged to you ? http://www.ai-contest.com/profile.php?user_id=5837 2010-09-14T05:58:57 For the speed gain boost template magic gives you its not worth the added API complexity IMO 2010-09-14T05:59:32 bartwe: yea fo sure theres no why he should be playing the same guy so much 2010-09-14T05:59:47 theres like 1k players and its suposed to be random? 2010-09-14T05:59:48 For some systems I can definitely see the advantage though 2010-09-14T05:59:48 Baphomet: himself... the ids match 2010-09-14T06:00:08 wait wat lol 2010-09-14T06:00:08 Baphomet: which has zero information value for the elo score 2010-09-14T06:00:33 Welcome to the wonderful world of probability ;) 2010-09-14T06:00:41 Baphomet: so i'm looking for an admin to kick 2010-09-14T06:00:42 now thats strange 2010-09-14T06:01:02 i don't think kicking an admin for a bug is strange 2010-09-14T06:02:34 * delt0r_ kicks bartwe_ for a bug in his code 2010-09-14T06:02:48 bartwe_: I know you have one ;) 2010-09-14T06:02:53 at least one 2010-09-14T06:02:55 auch, right in the right alligned allocation buffers 2010-09-14T06:03:01 *** acieroid` has quit IRC (Quit: "Be free.") 2010-09-14T06:03:18 delt0r_: ha, an optimist 2010-09-14T06:03:23 greghaynes: It doesn't work 2010-09-14T06:03:26 only one.... 2010-09-14T06:03:31 *** ajhager has quit IRC (Ping timeout: 276 seconds) 2010-09-14T06:03:33 I dont have rt visualization. 2010-09-14T06:03:42 yea ---thats what i tell my boss all the time... 2010-09-14T06:03:46 it was just one bug 2010-09-14T06:03:48 elijahbal: What did you downloat? 2010-09-14T06:03:52 er, download 2010-09-14T06:03:57 I pull the master branch 2010-09-14T06:04:01 on github 2010-09-14T06:04:16 You should get it then.. 2010-09-14T06:04:20 *** ajhager has joined #aichallenge 2010-09-14T06:05:03 The visualization window keeps stuck on the game start. 2010-09-14T06:05:14 Nothing is moving. 2010-09-14T06:05:20 Does your bot time out? 2010-09-14T06:05:26 No log 2010-09-14T06:05:35 Run it from a terminal 2010-09-14T06:06:06 yes my bot timed out 2010-09-14T06:06:13 but why ? 2010-09-14T06:06:19 What language are you using? 2010-09-14T06:06:22 It works pretty well with the java game engine 2010-09-14T06:06:24 python 2010-09-14T06:06:27 Im using python 2010-09-14T06:06:30 hrmm 2010-09-14T06:06:41 I use python, it works find 2010-09-14T06:06:50 Do you have modules in a folder. 2010-09-14T06:06:52 chmod +x your MyBot.py 2010-09-14T06:06:55 MyBot.py 2010-09-14T06:07:03 PlanetWars/*.py are the dependencies. 2010-09-14T06:07:10 yes, my bot has already +x 2010-09-14T06:07:32 Will try to copy PlanetWars in the PythonPATH 2010-09-14T06:08:16 hrmm, it should still work.. 2010-09-14T06:08:59 Well, I donot get the point 2010-09-14T06:09:13 What system are you on? 2010-09-14T06:10:17 *** acieroid` has joined #aichallenge 2010-09-14T06:10:44 antimatroid, hi. Are you passing planets around as a pointers to save some memory or do you just pass them as objects? :) 2010-09-14T06:11:28 elijahbal: Is your #! at the top of your MyBot.py file correct? 2010-09-14T06:13:07 yes I had #!/usr/bin/env python and now I'm trying with #!/usr/bin/python 2010-09-14T06:13:34 #!/usr/bin/env python is ok.. 2010-09-14T06:13:35 Does that work? 2010-09-14T06:13:47 Palmik: we dont really pass planets around, we just pass the whole planetwars object, but we pass that around as a pointer 2010-09-14T06:17:31 *** Ported_ has quit IRC (Quit: Page closed) 2010-09-14T06:17:40 Palmik: I don't do any passing 2010-09-14T06:17:45 *** Ported has joined #aichallenge 2010-09-14T06:17:57 greghaynes: I put a debug line in my bot 2010-09-14T06:18:04 *** acieroid` has quit IRC (Quit: "Be free.") 2010-09-14T06:18:14 Apparently the only line I get from your program is "initializing" 2010-09-14T06:18:35 I never fetch the list of planet, although it is displayed on the console by your program. 2010-09-14T06:18:54 That's why my program is waiting the list of planet but never get it 2010-09-14T06:19:33 animatroids, ok... well, that could work by accessing the member variable directly or have some getter returning const reference, still you would ne to copy them when filtering etc. 2010-09-14T06:20:10 *** acieroid` has joined #aichallenge 2010-09-14T06:20:17 greghaynes: my bot takes input from stdin 2010-09-14T06:21:53 *** Bobng has joined #aichallenge 2010-09-14T06:22:04 But never read the list of planet. 2010-09-14T06:22:08 How could it be ? 2010-09-14T06:23:32 *** acieroid has quit IRC (Quit: "Be free.") 2010-09-14T06:23:51 *** acieroid` is now known as acieroid 2010-09-14T06:24:00 check your parser 2010-09-14T06:24:14 python's starter package reads and outputs good. You just put bot's code int DoTurn function 2010-09-14T06:25:01 http://pastebin.ca/1940425 <- here is my bot 2010-09-14T06:25:49 edcba: my parser is good since I win or lose against java starter package's bots 2010-09-14T06:25:58 is the starter package client do import PlanetWares.doturn too ? 2010-09-14T06:26:05 Well my bot is running correctly. 2010-09-14T06:26:12 edcba: not, that is my work 2010-09-14T06:26:23 maybe it is checked by organizers 2010-09-14T06:27:02 ie file contains def DoTurn() 2010-09-14T06:27:17 http://pastebin.ca/1940429 <- here is my IO /parser code 2010-09-14T06:27:44 I do not see how it could be wrong, but tell me if something is bad in it. 2010-09-14T06:28:10 Can anyone code-review this patch: http://code.google.com/p/ai-contest/issues/detail?id=130 2010-09-14T06:28:35 import PlanetWars.doturn wtf? is it defined in PlanetWars.py ? 2010-09-14T06:29:11 http://pastebin.ca/1940430 2010-09-14T06:29:25 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T06:29:35 elijahbal: why are you importing "planet" and "fleet"? 2010-09-14T06:29:35 http://pastebin.ca/1940431 2010-09-14T06:29:44 because I rewrited the code 2010-09-14T06:29:51 splitting the files up so one class is in each file is a horrible way to do things 2010-09-14T06:30:37 java-way? :/ 2010-09-14T06:30:41 Bobng: you have not to like it. 2010-09-14T06:31:05 IMO a 40 lines file is better than a 100 lines file. 2010-09-14T06:31:15 modular and quickest to work on it. 2010-09-14T06:31:19 but that's not the point. 2010-09-14T06:31:31 Hardly, it contains classes you need not touch 2010-09-14T06:31:42 why ? 2010-09-14T06:31:46 Where ? 2010-09-14T06:32:24 elijahbal, try http://pastebin.ca/1940433 as input for you bot 2010-09-14T06:32:25 *** ajhager has joined #aichallenge 2010-09-14T06:32:28 elijahbal: try without return False maybe in IO 2010-09-14T06:32:30 http://pastebin.ca/1940429 2010-09-14T06:33:08 are people still having issues getting emails? 2010-09-14T06:33:10 you pretty much just removed some key functions from the PlanetWars class and made them into module functions. 2010-09-14T06:33:19 seems pointless. 2010-09-14T06:36:43 bash: ./MyBot.py : /usr/bin/python^M 2010-09-14T06:36:48 Bobng: seems, 2010-09-14T06:36:58 *** ajhager has quit IRC (Ping timeout: 255 seconds) 2010-09-14T06:37:19 Some people completely rewrite the start package. Is it pointless ? 2010-09-14T06:37:45 I just put functions I don't take care about in another file 2010-09-14T06:38:39 nothing wrong, I was just saying what came to my head when I read it 2010-09-14T06:39:00 you did ask if you could be told if something was bad in it, and in my mind thats what was bad, but its a personal choice I guess. 2010-09-14T06:39:15 *** Utkarsh has quit IRC () 2010-09-14T06:40:20 I have a problem with the Sheebang 2010-09-14T06:40:27 ./MyBot.py doesn't work 2010-09-14T06:40:54 *** td123 has joined #aichallenge 2010-09-14T06:41:11 you have windows line endings 2010-09-14T06:41:14 dos2unix 2010-09-14T06:41:37 also hardcoding /usr/bin/python is bad, use /usr/bin/env python 2010-09-14T06:42:04 windows line endings never were a issue whenever I uploaded code to a *nix box 2010-09-14T06:42:26 they affect #! line 2010-09-14T06:42:30 try "python MyBot.py" ... 2010-09-14T06:42:34 ahh that makes sense 2010-09-14T06:44:34 *** staii has quit IRC (Ping timeout: 264 seconds) 2010-09-14T06:52:34 *** EvGenius has joined #aichallenge 2010-09-14T06:53:12 problem solved 2010-09-14T06:53:16 thank you 2010-09-14T06:54:34 *** madsy_ has joined #aichallenge 2010-09-14T06:54:38 *** madsy has quit IRC (Remote host closed the connection) 2010-09-14T06:56:20 no any good ideas 2010-09-14T06:57:45 *** vikhyat has quit IRC (Ping timeout: 245 seconds) 2010-09-14T07:00:29 Accoun: it's no fun if everyone uses the same strategies 2010-09-14T07:05:37 *** kjeldahl has joined #aichallenge 2010-09-14T07:09:59 *** vikhyat has joined #aichallenge 2010-09-14T07:12:20 antimatroid: the maps are "reflect" about the center, so identical stratigies should give a draw every time :) 2010-09-14T07:12:34 I know that not what you ment... 2010-09-14T07:12:48 but there is a string incentive to come up with something inventive 2010-09-14T07:14:16 *** Pitel has joined #aichallenge 2010-09-14T07:15:51 Hi, I submitted python starter package entery, which should be working, by all my games results in loss after just one turn. 2010-09-14T07:16:19 delt0r_: re: Identical strategies giving draws. It doesn't seem to be true. I put DualBot against DualBot, and it's not a draw. It's probably related to "randomness" in how a language sorts identical planets, if distance does not enter into calculations. 2010-09-14T07:17:03 Pitel: probably you submited a file with windows line endings 2010-09-14T07:17:29 kjeldahl: starting position matters too 2010-09-14T07:17:51 if both bots attack the same planet, the first or the second may have a benefit 2010-09-14T07:18:29 True, if a planet is put in the square middle of the map... 2010-09-14T07:18:41 Naktibalda: ok, it seems so 2010-09-14T07:19:06 simple algorithms are predictable, they allways attack the same target 2010-09-14T07:19:33 *** jmreardon has joined #aichallenge 2010-09-14T07:19:34 Or if identical planets are put at identical distances, and both players attack the same planet... 2010-09-14T07:19:51 I think at least. :-) 2010-09-14T07:20:05 Eh, read that as "I believe...". 2010-09-14T07:20:28 it's not so easy add evaluation of distance to standard fomula 2010-09-14T07:20:37 I discovered that yesterday :) 2010-09-14T07:21:49 *** EvGenius has left #aichallenge 2010-09-14T07:23:11 well, today at 2am :) 2010-09-14T07:24:30 *** mossplix has quit IRC (Quit: Ex-Chat) 2010-09-14T07:27:17 if both bots attack the same planet at identical distance, neither can capture it, so a bot with higher growth_rate in initial planet wins, if growth rate is equal, it should result in draw 2010-09-14T07:27:21 kjeldahl: True--but if there is "randomness" in a bot, then its not really an identical stratagy. Also over a set of games it would show as a draw 2010-09-14T07:28:01 Naktibalda: Yes it does. My bot did that agaist prospector or something... 2010-09-14T07:28:10 *** mega1 has quit IRC (Ping timeout: 245 seconds) 2010-09-14T07:28:25 delt0r_: Agree. But it's easy to forget the implicit randomness in sorting algorithms when planets rank equal... 2010-09-14T07:28:32 true 2010-09-14T07:28:45 stable sort? quick sort? etc 2010-09-14T07:30:08 also kjeldahl i think if the behavior is implicit with fleet order, you get player order dependence too 2010-09-14T07:30:19 delt0r: Regarding your comments over a set of _identical_ games, I would expect that the kind of "sorting randomness" I hinted at would make the one instance of similar bots to win every time, i.e. not a draw. 2010-09-14T07:30:52 could do 2010-09-14T07:31:17 Which basically boils down to sorting randomness not being random at all. :-) 2010-09-14T07:31:28 but its not what i would expect if there is not "extra state" or player order dependance 2010-09-14T07:31:35 Agree. 2010-09-14T07:34:49 *** s_denchev has joined #aichallenge 2010-09-14T07:35:34 WARNING: player 1 timed out. 2010-09-14T07:35:34 WARNING: player 2 timed out. 2010-09-14T07:35:34 Turn 1 2010-09-14T07:35:34 Draw! 2010-09-14T07:36:36 impressive 2010-09-14T07:36:48 it's so easy to get a draw 2010-09-14T07:36:53 :) 2010-09-14T07:37:19 java -jar tools/PlayGame.jar maps/map7.txt 1001 1001 _log.txt "java -jar example_bots/BullyBot.jar" "java -jar example_bots/DualBot.jar" | java -jar tools/ShowGame.jar 2010-09-14T07:37:32 dont work 2010-09-14T07:37:37 *** retybok has left #aichallenge 2010-09-14T07:37:55 hm ok for me 2010-09-14T07:37:58 dualbot won 2010-09-14T07:43:23 mb bad java version 2010-09-14T07:44:31 java version "1.6.0_20" 2010-09-14T07:44:33 Java(TM) SE Runtime Environment (build 1.6.0_20-b02) 2010-09-14T07:47:28 java version "1.6.0_01" 2010-09-14T07:47:28 Java(TM) SE Runtime Environment (build 1.6.0_01-b06) 2010-09-14T07:47:28 Java HotSpot(TM) Client VM (build 1.6.0_01-b06, mixed mode) 2010-09-14T07:47:52 *** caytchen has joined #aichallenge 2010-09-14T07:50:05 *** marla_singer has joined #aichallenge 2010-09-14T07:50:41 *** marla_singer has left #aichallenge 2010-09-14T07:53:48 zaphos: you should joing the tcpserver 2010-09-14T07:53:53 it check russian windows and deny execution i think 2010-09-14T07:54:09 java version "1.6.0_21" 2010-09-14T07:54:09 Java(TM) SE Runtime Environment (build 1.6.0_21-b07) 2010-09-14T07:54:09 Java HotSpot(TM) Client VM (build 17.0-b17, mixed mode, sharing) 2010-09-14T07:56:51 Accoun: I'm having the same problem... timeouts 2010-09-14T07:57:19 *** PurplePanda has joined #aichallenge 2010-09-14T07:57:30 java version "1.6.0_20" 2010-09-14T07:57:30 Java(TM) SE Runtime Environment (build 1.6.0_20-b02) 2010-09-14T07:57:30 Java HotSpot(TM) 64-Bit Server VM (build 16.3-b01, mixed mode) 2010-09-14T07:57:49 do you run your bot or example bots? 2010-09-14T07:57:55 example ones 2010-09-14T07:58:11 ubuntu 10.04 64bit, btw 2010-09-14T07:58:31 an both example bots time out 2010-09-14T07:59:07 works for me in kubuntu 10.04, but I never tried running 2 example bots 2010-09-14T08:00:41 Pitel, you've tried increasing the timeouts? May be needed if you're on a slow computer. 2010-09-14T08:01:31 It's a pain in the ass that the java gamehost filters out stderr. Makes it a lot harder to catch errors of various kinds. 2010-09-14T08:01:36 *** retybok has joined #aichallenge 2010-09-14T08:01:39 kjeldahl: how? and I gues my core2duo E8400 isn;t slow computer ;) 2010-09-14T08:01:46 What's going on on the tcp server? 2010-09-14T08:02:08 The numbers 3000 and 1000 in the command line to run bots refer to 3 and 1 second timeouts. But your computer should be fast enough. 2010-09-14T08:03:00 pitel@mana3:~/Plocha/ai-contest$ java -jar tools/PlayGame.jar maps/map43.txt 3000 3000 log.txt "java -jar example_bots/RageBot.jar" "java -jar example_bots/ProspectorBot.jar" | java -jar tools/ShowGame.jar 2010-09-14T08:03:00 WARNING: player 1 timed out. 2010-09-14T08:03:00 WARNING: player 2 timed out. 2010-09-14T08:03:00 Turn 1 2010-09-14T08:03:00 Draw! 2010-09-14T08:03:03 what is cameron.jp's handle here? 2010-09-14T08:03:39 tcp server is down 2010-09-14T08:04:04 *** s_denchev has quit IRC (Quit: Page closed) 2010-09-14T08:06:19 Pitel, looks fine. Try 6000 just to be safe. I'm running on Ubuntu 10.04 as well with the same java version, and it works fine here. 2010-09-14T08:06:35 *** Utkarsh has joined #aichallenge 2010-09-14T08:07:21 kjeldahl: 10 seconds, and still nothing. but it produced notable 10 second spike on one CPU core... probably because polling? 2010-09-14T08:09:18 Pitel: Yeah, sounds like some poor IO model. I have no idea really, other than I got lots of timeouts as well when trying to port a bot to Clojure, and since the gamehost owns stderr without showing it, there were lots of easy to fix errors that I struggled to catch. 2010-09-14T08:10:27 *** mightybyte has joined #aichallenge 2010-09-14T08:11:09 If there are any enterprising java programmers around, it should be an easy fix to output stderr from running bots in the game host... ;-) 2010-09-14T08:14:50 kjeldahl: new java playgame version from a1k0n got that 2010-09-14T08:15:12 pinned in the forums, under tech issues 2010-09-14T08:15:24 rafsaoken: Excellent, thanks. I'll go fetch it now. 2010-09-14T08:15:37 retybok: works here 2010-09-14T08:16:55 *** staii has joined #aichallenge 2010-09-14T08:17:25 *** ermau has joined #aichallenge 2010-09-14T08:18:36 *** ermau has joined #aichallenge 2010-09-14T08:19:38 kjeldahl: I am also writing my own... it also has it 2010-09-14T08:19:46 and it won't restart the jvm 2010-09-14T08:19:52 *** Zeiris has quit IRC (Ping timeout: 276 seconds) 2010-09-14T08:20:11 *** s_denchev has joined #aichallenge 2010-09-14T08:22:49 lol on my profile i have written 2010-09-14T08:22:53 About Me: 2010-09-14T08:22:54 Pineapple Is Tasty. 2010-09-14T08:23:10 amd i just beat someoen called burnt.pineapple 2010-09-14T08:25:19 Re stderr, thanks, running the versions from a1k0n now. Works great. It also seems faster. 2010-09-14T08:27:29 *** ermau is now known as ermau|work 2010-09-14T08:28:24 kjeldahl> whow to fix it? 2010-09-14T08:28:37 zerd_: yeah, it's back on 2010-09-14T08:30:10 Accoun: The game host (PlayGame) used to suck up all stderr output, which is where error messages from bots gone wild goes. The versions from a1k0n fixes that, so stderr is now echoed through the game host. I.e. lets you see what is going on. 2010-09-14T08:32:07 *** Vi0 has quit IRC () 2010-09-14T08:33:00 anty pessian bag 2010-09-14T08:33:05 anti pessian bag 2010-09-14T08:33:08 anti ressian bag 2010-09-14T08:33:13 *** Mathnerd314 has quit IRC (Quit: ChatZilla 0.9.86-rdmsoft [XULRunner 1.9.2.8/20100722155716]) 2010-09-14T08:33:16 anti russian bag 2010-09-14T08:33:26 http://ai-contest.com/forum/viewtopic.php?f=18&t=498 solve 2010-09-14T08:33:28 *** otherAntimatroid has left #aichallenge 2010-09-14T08:33:32 wtf? 2010-09-14T08:33:46 jast read link 2010-09-14T08:33:57 um 2010-09-14T08:34:03 anti russian bag? 2010-09-14T08:34:31 satrter pacage dont worck in russian windows 2010-09-14T08:34:47 lol what is wrong with you its a locale issue 2010-09-14T08:35:00 the java starter packages don't work in any non-EN locale at the moment which is being attented to quite urgently 2010-09-14T08:35:21 2009 yaer starter packages worck well 2010-09-14T08:36:00 Completely unrelated to eachother 2010-09-14T08:36:07 forum also dont worck if brouser declare rusian lagiage 2010-09-14T08:36:29 Then that must be a phpBB issue which perhaps i can help with 2010-09-14T08:36:42 *** amstan_ has joined #aichallenge 2010-09-14T08:36:42 *** ChanServ sets mode: +o amstan_ 2010-09-14T08:37:16 hello 2010-09-14T08:37:28 Mornin' amstan_ 2010-09-14T08:37:40 stupid bus being late 2010-09-14T08:37:44 amstan: you're discriminating against the russian!!!111 2010-09-14T08:37:46 and stupid class for not having outlets 2010-09-14T08:37:51 caytchen: huh? 2010-09-14T08:38:09 lol amstan_ 2010-09-14T08:38:37 the bus i take stops at laurier and also a highschool 2010-09-14T08:38:40 its a nightmare 2010-09-14T08:39:00 pgpaskar_: it gets full, fast 2010-09-14T08:39:17 i'm thinking i should get car technologies 2010-09-14T08:39:22 pgpaskar_: any ideas where to park? 2010-09-14T08:39:33 nope i do not drive to school 2010-09-14T08:40:03 driving would cut my travel time by 10 though. since i can go the other direction through R&T and be there in 2 minutes 2010-09-14T08:40:07 vs 30m bus lol 2010-09-14T08:40:41 and besides the fact that i'm late and my battery is dying, there's no prof to be seen in this class 2010-09-14T08:41:19 why is your battery dying? 2010-09-14T08:41:24 also ugh wifi congestion 2010-09-14T08:41:44 today is such a crappy day 2010-09-14T08:41:50 nah 2010-09-14T08:41:54 *** odinsbane has joined #aichallenge 2010-09-14T08:41:56 this day 3 months from now will be crappy 2010-09-14T08:41:58 i woke up at 7:20.. bus at 7:45 2010-09-14T08:42:03 today is a good day, sons of anarchy is on tonight 2010-09-14T08:42:07 pgpaskar_: it can only get crappier 2010-09-14T08:42:16 yay! a prof 2010-09-14T08:42:31 pgpaskar_: what class are you in?> 2010-09-14T08:42:46 *** dabino has joined #aichallenge 2010-09-14T08:43:44 cs245 2010-09-14T08:44:21 people are still trickling in 2010-09-14T08:44:26 now realizing that seets are QUITE limited 2010-09-14T08:44:28 ftw being on time 2010-09-14T08:44:52 *** eburnette has joined #aichallenge 2010-09-14T08:45:19 c++ pacage more and more object oriented 2010-09-14T08:45:26 ))) 2010-09-14T08:45:51 pgpaskar_: it's fine, there's still 5% more spaces 2010-09-14T08:46:09 woot, OO, in cs246 2010-09-14T08:46:13 Accoun: are you spying on me? lol 2010-09-14T08:46:38 *** dabino has quit IRC (Client Quit) 2010-09-14T08:47:46 pgpaskar_: ? 2010-09-14T08:47:51 Err wow 2010-09-14T08:47:53 amstan_: ? 2010-09-14T08:48:04 Are you in this class? 2010-09-14T08:48:15 pgpaskar_: cs246: object oriented programming, and Accoun just guesses it 2010-09-14T08:48:21 pgpaskar_: my prof wants us to use c++, lol 2010-09-14T08:48:25 Oh 2010-09-14T08:48:30 Good luck w/ dat 2010-09-14T08:48:49 ew! wtf? 2010-09-14T08:48:56 you'll program in c++ using emacs 2010-09-14T08:49:01 noO!! 2010-09-14T08:49:10 someone has a nokia ringtone 2010-09-14T08:49:11 how awesome 2010-09-14T08:52:36 will be perl package? 2010-09-14T08:52:45 will be perl starter package? 2010-09-14T08:53:16 Accoun: there is an unoffical perl starter package, I don't know if the server supports perl yet though 2010-09-14T08:54:54 Accoun: you can make one 2010-09-14T08:55:10 Accoun: you put it on the forums when finished 2010-09-14T08:55:34 its not make me famous 2010-09-14T08:55:39 ) 2010-09-14T08:56:30 contestbot, newlanguage 2010-09-14T08:56:31 amstan_: newlanguage = post in here: http://ai-contest.com/forum/viewtopic.php?f=18&t=414 2010-09-14T08:56:37 Accoun: that ^ 2010-09-14T08:57:02 amstan_> lern russian 2010-09-14T08:57:08 ) 2010-09-14T08:57:46 Accoun: no thanks, romanian and english is enough for me for now 2010-09-14T08:57:52 maybe later 2010-09-14T08:57:52 all book about atomic nuclear bomp prodiction writeln in russian )))) 2010-09-14T08:58:13 lol 2010-09-14T08:58:24 I'm fairly certain they are written in English as well 2010-09-14T08:58:24 contestbot: grab Accoun 2010-09-14T08:58:24 amstan_: The operation succeeded. 2010-09-14T08:58:27 lol 2010-09-14T08:58:44 grab? 2010-09-14T08:58:57 dstufft: contestbot: quote Accoun 2010-09-14T08:59:05 oh 2010-09-14T08:59:19 in English, not in canadian ))) 2010-09-14T08:59:39 Canadian isn't a language 2010-09-14T09:00:06 realy? canadian not have any langiage ))) 2010-09-14T09:00:14 Accoun: canadians speak english 2010-09-14T09:00:51 or french 2010-09-14T09:01:13 Canadians can be proud of ))) 2010-09-14T09:01:40 Hold on I need to balance you 2010-09-14T09:01:43 (((((((((((((((((((((((((((((((((((( 2010-09-14T09:01:46 Alright 2010-09-14T09:01:47 carry on 2010-09-14T09:02:01 lol 2010-09-14T09:02:27 it looks like lisp 2010-09-14T09:02:32 Anyone have any idea how much bandwidth the tcpserver uses (on average) for one game? 2010-09-14T09:02:32 *** Queue29 has joined #aichallenge 2010-09-14T09:02:45 ok, now I'm sure my startup entry has Unix line endings, but my bot still always lose after 1 turn :( 2010-09-14T09:02:50 I only ask because I pay for bandwidth per gb and I'm running my bot fairly frequently 2010-09-14T09:02:54 blister: ping it from your computer 2010-09-14T09:03:06 if you're on windows 2010-09-14T09:03:07 ping 213.3.30.106 2010-09-14T09:03:11 from command prompt 2010-09-14T09:03:17 mine is like 500ms 2010-09-14T09:03:20 antimatroid: I don't need response times, I mean actual traffic requirements for a game 2010-09-14T09:03:23 not latency 2010-09-14T09:03:24 oh 2010-09-14T09:03:27 sorry :P 2010-09-14T09:03:35 like, from start to finish, how much traffic is used. 2010-09-14T09:03:48 blister> ban em if on final have hude overtime 2010-09-14T09:03:55 i know there are some linux apps to monitor that stuff, but I'm not really a sysadmin and don't know all the cool tricks 2010-09-14T09:03:56 --- 213.3.30.106 ping statistics --- 2010-09-14T09:03:56 18 packets transmitted, 17 received, 5% packet loss, time 17002ms 2010-09-14T09:03:56 rtt min/avg/max/mdev = 66.735/296.078/725.099/228.341 ms 2010-09-14T09:03:58 it depends on a number of fleets mainly :) 2010-09-14T09:04:04 blister: why dont you monitor it from your end, and multiply it by 2? 2010-09-14T09:04:16 judging from the log files ~100-150k per game 2010-09-14T09:04:21 Pitel link me your profile 2010-09-14T09:04:24 rough guestimate 2010-09-14T09:04:26 Baphomet: that's kinda what I'm asking. I don't know any apps to monitor bandwidth (linux). Any ideas? 2010-09-14T09:04:33 Guest77645: thanks. 2010-09-14T09:04:33 dstufft: http://ai-contest.com/profile.php?user_id=7372 2010-09-14T09:04:35 blister: get wireshark, and run a game 2010-09-14T09:04:36 thats close enough 2010-09-14T09:04:39 write trafic optimized client/server 2010-09-14T09:05:08 i like accoun, he gives useful advice all the time, lol 2010-09-14T09:05:18 lmfao 2010-09-14T09:05:23 *** askhader has joined #aichallenge 2010-09-14T09:05:56 if you redirect your tcp client output to a file you can get an approx count for your games 2010-09-14T09:06:33 That's a good idea. 2010-09-14T09:06:39 I'll give that a shot... 2010-09-14T09:07:06 hey can anyone show me how to use pw.Distance() to fine the closest planets for dest? 2010-09-14T09:07:14 Pitel: http://pastebin.com/iwbr2GuQ 2010-09-14T09:07:19 Baphomet: what language? 2010-09-14T09:07:20 It's not a big deal, I pay like, $0.15/gb, but I see from log.txt that it's sending a pretty huge gamestate stream and was worried that tcp.c might be a fairly heavy bandwidth hog 2010-09-14T09:07:22 c++ 2010-09-14T09:07:25 awww 2010-09-14T09:07:43 blister: How would you mitigate this, assuming you're correct? 2010-09-14T09:07:44 i have pw.Distance working, just not sure how to integrate that with the dest finding calculatins 2010-09-14T09:07:52 dstufft: oh, I thought I should just submit mybot.py :] thanks 2010-09-14T09:08:03 askhader: stop playing on my server with metered bandwidth and go back to playing on my home account... ;0 2010-09-14T09:08:04 the code in python that i have for my strongest planet: max(self.universe.my_planets,key=lambda planet:planet.ship_count) 2010-09-14T09:08:06 heh 2010-09-14T09:08:10 or how i can use to it find the closest/furthest 2010-09-14T09:08:55 measure distance to all planets and find shortest 2010-09-14T09:09:09 wifi is soooo laggy 2010-09-14T09:09:23 Naktibalda but im not sure how to do that in code :( 2010-09-14T09:09:31 ha, i finally got a win against a1k0n 2010-09-14T09:09:35 that only took me a week 2010-09-14T09:09:46 Baphomet: foreach(other_planets) { if(distance < min_distance) { min_distance = distance; min_planet = planet; } } 2010-09-14T09:09:50 my thing is still really buggy though D: 2010-09-14T09:09:56 how can i pass all the diferent souce and dest planets through pw.Distance() and extract the lowest val? 2010-09-14T09:09:59 *** kozlovsky has quit IRC (Ping timeout: 252 seconds) 2010-09-14T09:10:10 Baphomet, with loops? 2010-09-14T09:10:11 same code as to find strongest/weakest planets 2010-09-14T09:10:12 askhader: is it just me or is he using freaking win telnet for ssh? 2010-09-14T09:10:28 * askhader wasn't looking 2010-09-14T09:10:31 *** ajhager has joined #aichallenge 2010-09-14T09:10:33 hyperterminal.. that's the one 2010-09-14T09:10:38 *** panchi has joined #aichallenge 2010-09-14T09:10:55 is there actually an already made min_distance? 2010-09-14T09:11:06 also im doing this in c++ and i dont know other languages... 2010-09-14T09:11:24 no, but use a priority queue and it's pretty easy to generate on the first move 2010-09-14T09:11:33 (i'm lazy) 2010-09-14T09:11:42 compare the distance for all the planets 2010-09-14T09:11:47 askhader: in batchfiles... 2010-09-14T09:11:52 =P 2010-09-14T09:11:54 antimatroid: pretty sure that's a term only used in windows 2010-09-14T09:12:25 Baphomet: if you know C++, you could pretty much jump right in to Python or Java 2010-09-14T09:12:34 they're "easy" in comparison 2010-09-14T09:12:43 *** Yoshi-TS4 has quit IRC (Quit: Minus One IRC) 2010-09-14T09:12:49 Baphomet: i agree, it's a very cool language to learn 2010-09-14T09:12:50 *** rafsaoken has quit IRC (Quit: Page closed) 2010-09-14T09:13:11 java looks pretty similar and python looks confusing but c++ is anways nice and organised and so powerfull 2010-09-14T09:13:24 Python is the Master Language 2010-09-14T09:13:27 Baphomet: its true, but they're all fairly similar 2010-09-14T09:13:27 You have that backwards 2010-09-14T09:13:27 >.> 2010-09-14T09:13:30 "and organized" lol 2010-09-14T09:13:30 Python is the supreme language 2010-09-14T09:13:34 roflcopter 2010-09-14T09:13:34 :) 2010-09-14T09:13:37 C++ is a mess 2010-09-14T09:13:48 i implemented dualbot in python in 3 lines of code 2010-09-14T09:13:55 well from my experance ;P 2010-09-14T09:13:55 * blister says that as someone who knows C++. 2010-09-14T09:13:55 dest=min(self.universe.not_my_planets,key=lambda planet:planet.ship_count) 2010-09-14T09:13:57 source=max(self.universe.my_planets,key=lambda planet:planet.ship_count) 2010-09-14T09:13:58 source.send_fleet(dest,source.ship_count/2) 2010-09-14T09:14:10 send_fleet hehe 2010-09-14T09:14:11 Tabs? 2010-09-14T09:14:20 hmm? 2010-09-14T09:14:26 blister> write trafic optimized client/server - send over net only moves, not all map 2010-09-14T09:14:27 Baphomet: If the only tool you have is a hammer, you tend to see every problem as a nail 2010-09-14T09:14:47 amstan_: nvm, my fail locale settings 2010-09-14T09:14:49 Accoun: make one.. everything's open source this time.. 2010-09-14T09:14:58 amstan: you should rewrite python starter package in this style 2010-09-14T09:15:22 tabs? 2010-09-14T09:15:26 Naktibalda: http://github.com/amstan/planetwars-python-kit 2010-09-14T09:15:34 Accoun: rgr. that might work. Don't know if it's worth my time though. Seems easier just to play on my home DSL connection or just pay the few dollars extra that it'll cost to play from my server 2010-09-14T09:16:03 I need to get an ssd for my laptop :) 2010-09-14T09:16:23 i need more joule technologies 2010-09-14T09:16:35 14% left 2010-09-14T09:16:43 owned 2010-09-14T09:16:47 wish the server could do games quicker 2010-09-14T09:16:47 4:44 remaining 2010-09-14T09:17:02 amstan: it looks like your stupid bots don't attack enemy 2010-09-14T09:17:15 Naktibalda: it's just a starter package, don't mind me in the rankings 2010-09-14T09:17:19 *** drewrobb has joined #aichallenge 2010-09-14T09:17:20 it was just a test 2010-09-14T09:17:57 Tracerbot has just uploaded a seemingly unbeatable bot. No one has beaten him in a real game yet: http://www.benzedrine.cx/planetwars/getplayer?player=TracerBot11.1 2010-09-14T09:18:03 it seems more agressive than usual 2010-09-14T09:18:28 i see 3 losses 2010-09-14T09:18:37 edcba: there are not real losses 2010-09-14T09:18:42 more like server glitches 2010-09-14T09:18:54 or timeouts.. 2010-09-14T09:19:06 actually one of them is a real defeat 2010-09-14T09:19:14 against bartwe.6 2010-09-14T09:24:08 *** ajhager has quit IRC (Ping timeout: 240 seconds) 2010-09-14T09:27:42 huzzah, bot seems semi decent for mostly still being heuristic based, might need to take the rest of the week off and do some uni work finally 2010-09-14T09:29:08 c++ have scanf function? 2010-09-14T09:29:46 Accoun, in maybe, but you should be using 2010-09-14T09:29:55 @contestbot help 2010-09-14T09:29:56 eburnette: Error: "contestbot" is not a valid command. 2010-09-14T09:30:10 @help grab 2010-09-14T09:30:10 pgpaskar_: (grab [] ) -- Grabs a quote from by for the quotegrabs table. is only necessary if the message isn't sent in the channel itself. 2010-09-14T09:30:31 @rankings 2010-09-14T09:30:32 eburnette: Top 10 players: deepblue(2779), sighbot(2714), a1k0n(2697), Jerome(2697), Dazibao(2671), Rustyoldman(2651), wtfbbq(2649), Kavish(2630), zuko3d(2628), hellman(2617) 2010-09-14T09:30:49 o_O how am I second with my crappy bot 2010-09-14T09:31:03 o_O i am 10th again 2010-09-14T09:31:17 clearly beating all those people ranked 700 helped 2010-09-14T09:31:21 *** amstan_ has quit IRC (Quit: Konversation terminated!) 2010-09-14T09:31:34 *** amstan_ has joined #aichallenge 2010-09-14T09:31:34 *** ChanServ sets mode: +o amstan_ 2010-09-14T09:31:39 @game eburnette 2010-09-14T09:31:39 eburnette: Error: 'eburnette' is not a valid integer. 2010-09-14T09:31:44 @game 12345 2010-09-14T09:31:45 eburnette: An error has occurred and has been logged. Please contact this bot's administrator for more information. 2010-09-14T09:32:11 *** hexapode has joined #aichallenge 2010-09-14T09:32:12 eburnette: only integers 2010-09-14T09:32:14 @ping 2010-09-14T09:32:14 pong 2010-09-14T09:32:21 eburnette: and valid integers 2010-09-14T09:32:31 *** hexapode has quit IRC (Client Quit) 2010-09-14T09:32:56 sscanf scanf full functional an don have memory leaks 2010-09-14T09:32:57 ranking is about correct for bot separated by 100 ranks, but it's about random on any 20 consecutive place 2010-09-14T09:33:47 Accoun: why do you need scanf? there's already a starter package for c++ 2010-09-14T09:33:53 haha 2010-09-14T09:33:57 just made a suicidal bot 2010-09-14T09:33:58 the 10 first bots almost never fought each other, so their relative scores is only about how many 150+ opponents they fought. 2010-09-14T09:34:00 never attacks 2010-09-14T09:34:00 :D 2010-09-14T09:34:14 blister: not that hard, lol 2010-09-14T09:34:19 heh 2010-09-14T09:34:25 amstan_: yeah, but i did it on accident... :D 2010-09-14T09:34:40 we should have a side contest on the *worst* bot 2010-09-14T09:34:48 if i suddently quit, it means my battery died 2010-09-14T09:34:49 that's against the rules :( 2010-09-14T09:34:50 blister: it could be worse: you could send N-1 fleet on each neutral planet, and loose first faster 2010-09-14T09:35:10 kuwabara that's what I was thinking 2010-09-14T09:35:11 *** zaph has joined #aichallenge 2010-09-14T09:35:31 sigh: what is? 2010-09-14T09:35:38 intentionally losing 2010-09-14T09:35:44 is it? 2010-09-14T09:35:47 oh.. yeah 2010-09-14T09:35:47 at least that's what the site says 2010-09-14T09:35:58 to prevent boosting other bots 2010-09-14T09:36:01 yeah 2010-09-14T09:36:06 and to save cpu cycles 2010-09-14T09:36:10 haha 2010-09-14T09:36:20 the site has very strict language regarding it 2010-09-14T09:36:24 the funniest part is that its doing pretty well against good bots 2010-09-14T09:36:37 they're having a hard time beating me 2010-09-14T09:36:43 up to 150 turns right now 2010-09-14T09:36:46 doing absolutely nothing 2010-09-14T09:36:47 hehe 2010-09-14T09:36:57 blister: you still get beaten though 2010-09-14T09:37:26 wow 2010-09-14T09:37:34 have an elo of 350 with a bot that never attacks 2010-09-14T09:37:35 hahaha 2010-09-14T09:37:38 :D 2010-09-14T09:37:39 thats my highest score ever 2010-09-14T09:38:02 it would do quite well against ragebot too 2010-09-14T09:38:05 http://www.benzedrine.cx/planetwars/getplayer?player=cobalt-3.0 2010-09-14T09:39:22 http://pastebin.ca/1940527 hello 2010-09-14T09:39:28 I have a real problem with it. 2010-09-14T09:39:54 I don t understand How can I send ships from neutral planets, but the planet 2 is mine on the map 18. 2010-09-14T09:40:43 *** ajhager has joined #aichallenge 2010-09-14T09:41:13 *** amstan_ has quit IRC (Remote host closed the connection) 2010-09-14T09:41:29 *** zaph has quit IRC (Quit: Bye) 2010-09-14T09:41:48 omg 2010-09-14T09:41:55 my suicide bot just won against a good player 2010-09-14T09:41:57 hahahaha 2010-09-14T09:42:06 went 100+ turns and then he timed out... :D 2010-09-14T09:42:08 help ? 2010-09-14T09:42:17 new contest strategy: never attack! 2010-09-14T09:42:17 woot 2010-09-14T09:42:34 blister, give a link please :) 2010-09-14T09:42:50 as soon as it is available, ill paste it 2010-09-14T09:43:04 how long is the delay between game completion and being listed on the server? 2010-09-14T09:43:08 2-3 minutes? 2010-09-14T09:43:24 yea 2010-09-14T09:43:40 k 2010-09-14T09:44:00 *** vikhyat has quit IRC (Read error: Connection reset by peer) 2010-09-14T09:46:44 http://www.benzedrine.cx/planetwars/canvas?game_id=1284471296|cobalt-3.1|hn1 2010-09-14T09:46:53 haha 2010-09-14T09:47:13 *** Tegid has quit IRC (Quit: Page closed) 2010-09-14T09:47:33 his bot apparently doesn't know what to do when there are no neutral planets to attack 2010-09-14T09:47:56 new bot algorithm: if (opponent == 'hn1'): return 2010-09-14T09:47:57 haha 2010-09-14T09:54:13 wow, apparently a lot of bots refuse to attack a suicidal bot 2010-09-14T09:54:24 since none of their planets have more ships than my non-attacking planet 2010-09-14T09:54:26 they ignore it 2010-09-14T09:54:30 boring 2010-09-14T09:56:03 *** iPanda has joined #aichallenge 2010-09-14T09:56:32 could anybody says, why no games played until 4 hours? 2010-09-14T09:56:48 is it bug at tournament script? 2010-09-14T09:58:09 http://ai-contest.com/profile.php?user_id=7010 2010-09-14T09:58:32 What, iPanda ? 2010-09-14T09:58:50 why no games played until 4 hours? 2010-09-14T09:59:16 since 5 Am 2010-09-14T09:59:39 Somebody is working on it. 2010-09-14T09:59:53 Changing the code I think. 2010-09-14T09:59:53 but other bots plaing 2010-09-14T10:00:01 *** vikhyat has joined #aichallenge 2010-09-14T10:00:15 *** xha has joined #aichallenge 2010-09-14T10:00:32 Yeah, they were playing up until not too long ago. 2010-09-14T10:00:42 iPanda: I will look into it. 2010-09-14T10:02:05 *** drewrobb has quit IRC (Ping timeout: 272 seconds) 2010-09-14T10:03:19 wheee... beat bartwe.6: http://www.benzedrine.cx/planetwars/canvas?game_id=1284471933|bartwe.6|sigh-timeout 2010-09-14T10:04:48 sigh: you're second on main site :P 2010-09-14T10:04:50 or were before 2010-09-14T10:05:03 although i think unofficial is obviously more competitive 2010-09-14T10:05:16 Any admin's around for a code review? 2010-09-14T10:05:43 antimatroid: yeah, but main site is a fluke... all the games re against people ranked like 500 2010-09-14T10:05:58 what is non-official? 2010-09-14T10:06:03 whats different? 2010-09-14T10:06:35 http://www.benzedrine.cx/planetwars/ 2010-09-14T10:06:42 iPanda: you get more games, and against better bots 2010-09-14T10:06:58 *** s_denchev has quit IRC (Quit: Page closed) 2010-09-14T10:07:00 sigh: ;) 2010-09-14T10:07:18 sigh: just started 7.1 hope it will do better 2010-09-14T10:07:25 I thought this was an interesting match 2010-09-14T10:07:25 http://www.benzedrine.cx/planetwars/canvas?game_id=1284469310|hexist|ErrBot_0.0.3 2010-09-14T10:07:30 bartwe_: don't worry, you beat me next game anyone 2010-09-14T10:07:37 bartweI wish I could make a bot as well as you 2010-09-14T10:07:52 *anyway 2010-09-14T10:08:05 rogue: you should be able to 2010-09-14T10:08:16 there are loads of crap ones before the .7 2010-09-14T10:08:30 and alot more then 6 too ;) 2010-09-14T10:08:33 *** rwa has joined #aichallenge 2010-09-14T10:10:40 *** ath888 has joined #aichallenge 2010-09-14T10:10:47 sigh: oh it's on :P 2010-09-14T10:10:58 *** tapwater has joined #aichallenge 2010-09-14T10:11:15 haha... 2010-09-14T10:11:23 I bet my bot will timout on you, it's good at tht 2010-09-14T10:11:28 so is mine 2010-09-14T10:11:43 and i am still randomly messing around with numbers to see what sort of works best aha 2010-09-14T10:12:00 but i shall be fighting you for top spot in australia :P 2010-09-14T10:12:00 :) 2010-09-14T10:12:12 antimatroid: magic number == trouble 2010-09-14T10:12:30 are, nice to see another aussie in here :D 2010-09-14T10:12:42 bartwe_: i'm just trying to get a feel for what seems to work better 2010-09-14T10:12:51 Zaphus is australian too 2010-09-14T10:12:52 damn you! 2010-09-14T10:13:00 I win! 2010-09-14T10:13:00 oh, again 2010-09-14T10:13:04 ugh zaphus beat me on the official 2010-09-14T10:13:19 lol 2010-09-14T10:13:56 antimatroid: this game looks closer 2010-09-14T10:14:14 *** drewrobb has joined #aichallenge 2010-09-14T10:15:22 damnit 2010-09-14T10:15:24 :P 2010-09-14T10:16:38 http://ai-contest.com/visualizer.php?game_id=4467509 look graphs =) 2010-09-14T10:16:59 *** Cyndre has quit IRC (Quit: Leaving) 2010-09-14T10:18:35 danielvf: im around 2010-09-14T10:18:47 dstufft: Great. 2010-09-14T10:19:08 sigh: pantsed me both times 2010-09-14T10:19:33 maybe next time ;) 2010-09-14T10:22:10 *** Cyndre has joined #aichallenge 2010-09-14T10:23:36 hmmm, i wonder if the main site is even running games atm? 2010-09-14T10:24:16 antimatroid: it should be unless someone paused it or it got stuck 2010-09-14T10:24:17 main site is always iffy 2010-09-14T10:24:44 yes 2010-09-14T10:24:46 its running games 2010-09-14T10:25:09 *** xha has quit IRC (Quit: Odchzm) 2010-09-14T10:25:29 7 games in 10 hours... ugh 2010-09-14T10:25:51 *** HakanD has joined #aichallenge 2010-09-14T10:26:47 hey 2010-09-14T10:26:53 hey 2010-09-14T10:27:02 we are averaging 8.6 games a minute today 2010-09-14T10:27:19 and picking players of the games it's still 100% random 2010-09-14T10:27:24 http://www.benzedrine.cx/planetwars/canvas?game_id=1284473770|cheese_bot|antimatroid1.3 2010-09-14T10:27:28 i came back from behind, yeah 2010-09-14T10:28:37 nice work 2010-09-14T10:29:19 alright, time for me to sleep... goodnight 2010-09-14T10:29:25 *** sigh has quit IRC (Quit: Leaving...) 2010-09-14T10:31:17 hmz, 7.1 is a dud so far 2010-09-14T10:31:20 *** Appleman1234 has quit IRC (Quit: Leaving) 2010-09-14T10:31:24 deepblue: great tactics update 2010-09-14T10:31:53 bartwe_: it's weird how easy it is to break it 2010-09-14T10:32:07 anti: break what ? 2010-09-14T10:32:17 make the bot just completely retarded 2010-09-14T10:32:58 definitly 2010-09-14T10:33:15 yeah, I just fixed a terrible bug and my bot got significantly worse... ugh 2010-09-14T10:33:42 it was much harder to break my tron bot 2010-09-14T10:34:17 making the score function more technically correct did wonders 2010-09-14T10:35:13 fuzzy things like 'aggresiveness' could work, but i would need hundreds of benzedrine runs before i can come close to evaluating it 2010-09-14T10:35:19 so i just leave it running for for ever 2010-09-14T10:35:46 a1k0n: you better have a new bot ready soon, you're starting to get trampled by people :P 2010-09-14T10:36:43 Is anybody using the java tcp client from http://ai-contest.com/forum/viewtopic.php?f=18&t=424&hilit=tcp+java#p2616 on windows? 2010-09-14T10:37:16 *** Ported has quit IRC (Quit: Page closed) 2010-09-14T10:38:12 eburnette: yes i am 2010-09-14T10:38:43 although someone emailed me the .class file so i didn't actually have to get a proper java compiler 2010-09-14T10:39:04 i could do you the same favour if you weren't too worried about pming me your email 2010-09-14T10:39:20 it only has a small virus 2010-09-14T10:39:29 just hangs for me, bot never responds to the go command 2010-09-14T10:39:33 nothing too deadly 2010-09-14T10:39:44 eburn: .Flush() 2010-09-14T10:39:59 java TCP "213.3.30.106" "9999" "antimatroid1.3" ".\projectfiles.exe" "10000" 2010-09-14T10:40:09 i'm using a command like that, requires some random useless parameter at the end 2010-09-14T10:40:40 why not use the tcp.c ? 2010-09-14T10:40:47 works on windows with cygwin too 2010-09-14T10:41:00 i couldn't get cygwin to pick up gcc 2010-09-14T10:41:07 this works, so i'm happy :) 2010-09-14T10:41:10 I prefer to use Java if I could get this to work 2010-09-14T10:41:30 cygwin is like an elaborate tower of stacked dominos 2010-09-14T10:41:35 *** amstan_ has joined #aichallenge 2010-09-14T10:41:35 *** ChanServ sets mode: +o amstan_ 2010-09-14T10:41:42 it's awesome utnil one part falls and takes the rest out with it 2010-09-14T10:41:43 >.> 2010-09-14T10:41:59 yo 2010-09-14T10:42:22 meh, use alot of cygwin 2010-09-14T10:42:42 well that was fun, waited 4.5 hours to play "noobwaffle", a bot that doesn't run 2010-09-14T10:42:43 hi amstan: i've been seeing players that continually combat their own in the rankings 2010-09-14T10:42:59 yea i had that today 2010-09-14T10:43:14 I just use virtualization 2010-09-14T10:43:39 What version of gcc is being used to compile the c++ bots? 2010-09-14T10:44:18 latest one? 2010-09-14T10:44:21 rwa: the official server will always be like that, i really really hope they run a top 50 round robin this time 2010-09-14T10:44:34 *** madsy_ has quit IRC (Quit: leaving) 2010-09-14T10:44:51 bartwe: huh? 2010-09-14T10:44:57 i wish the server would play people more quickly 2010-09-14T10:45:05 so i can find out if my code is any good 2010-09-14T10:45:32 Baphomet: use the tcp server 2010-09-14T10:45:33 http://www.benzedrine.cx/planetwars/ 2010-09-14T10:45:51 i struggle to code very basic bots lets alone that 2010-09-14T10:45:57 is it easy to setup? 2010-09-14T10:46:10 So using the tcp.c, what servers are available? And is the username/passwd the ones from the contest? 2010-09-14T10:46:27 sigh what was that site that had the precompiled cygwin tcp version? 2010-09-14T10:46:30 yep you download the tcp.c file he provides, then run your unmodified bot and the tcp program launches your bot, connets to the server, adn does all of that 2010-09-14T10:46:41 antimatroid: we are working on fixing that 2010-09-14T10:46:50 amstan: maybe it is fixed now, but saw alot of players that had Win lines with as enemy their own name 2010-09-14T10:46:58 cool sounds good, i will give this a try 2010-09-14T10:46:59 odinsbane: no, the username is just for stats collecting, you don't need to create an account anywhere.. the password is optional as well 2010-09-14T10:47:08 dstufft: fixing a round robin? 2010-09-14T10:47:17 don't use the same password as in main site 2010-09-14T10:47:18 anoek: is the ip www.benzedrine.cx' s ip? 2010-09-14T10:47:24 are you guys going to sweep out all starter package bots? 2010-09-14T10:47:35 and how the hell will you do that will all the revisions that have happened to them? 2010-09-14T10:47:40 odinsbane: it's just there in case you are worried that someone is going to pose as you and mess up your unofficial stats 2010-09-14T10:47:50 oh shit so do i need to run cygwin to do the tcp.c ver? 2010-09-14T10:47:57 Baphomet: probably 2010-09-14T10:47:59 Baphomet: now 2010-09-14T10:48:15 now? 2010-09-14T10:48:17 tcp/tcp 213.3.30.106 9999 ./MyBot 2010-09-14T10:48:17 Somebody gave me a link last night but it was on my other computer. Is there a log of this chat? 2010-09-14T10:48:29 Baphomet: you just need to be able to compile it. You could probably do it with ming. 2010-09-14T10:48:42 there is a java file that you just need in the same folder as your bot .exe 2010-09-14T10:48:51 then just run a command though normal command prompt 2010-09-14T10:49:08 so anyone have any idea what version of gcc we are using? 2010-09-14T10:49:46 it doesnt compile, dont have all the includes 2010-09-14T10:49:51 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T10:50:01 pm me your email address 2010-09-14T10:50:11 i'll email you the .class file so you don't have to compile 2010-09-14T10:50:15 i had it emailed to me like that 2010-09-14T10:50:19 anoek: I can't access that webserver for the tcp server, is this limited to a specific range of ips? 2010-09-14T10:50:24 anoek: 2010-09-14T10:50:26 Using built-in specs. 2010-09-14T10:50:27 Target: i486-linux-gnu 2010-09-14T10:50:29 Configured with: ../src/configure -v --enable-languages=c,c++,fortran,objc,obj-c++,treelang --prefix=/usr --enable-shared --with-system-zlib --libexecdir=/usr/lib --without-included-gettext --enable-threads=posix --enable-nls --with-gxx-include-dir=/usr/include/c++/4.2 --program-suffix=-4.2 --enable-clocale=gnu --enable-libstdcxx-debug --enable-objc-gc --enable-mpfr --enable-targets=all --enable-checking=release --build=i486-linux-gnu -- 2010-09-14T10:50:30 host=i486-linux-gnu --target=i486-linux-gnu 2010-09-14T10:50:32 Thread model: posix 2010-09-14T10:50:33 gcc version 4.2.4 (Ubuntu 4.2.4-1ubuntu4) 2010-09-14T10:50:35 ok.. 2010-09-14T10:50:36 that's too much, sorry 2010-09-14T10:50:43 Perfect, thanks amstan_ :) 2010-09-14T10:50:53 *** amstan has quit IRC (Remote host closed the connection) 2010-09-14T10:50:54 antimatroid: it's not a round robin 2010-09-14T10:50:56 *** amstan_ is now known as amstan 2010-09-14T10:51:05 it currently just selects 2 people completely at random 2010-09-14T10:51:17 why not run one at the end to finalise the top 50? 2010-09-14T10:52:11 I don't know what will be done at the end, but currently it's 100% random, and folks are looking into changing it to a much smarter and fairer algo for selecting games 2010-09-14T10:53:22 *** vikhyat has quit IRC (Ping timeout: 265 seconds) 2010-09-14T10:54:31 dstuff: gausing distribution 2010-09-14T10:54:37 relative to the current ranking 2010-09-14T10:55:39 bartwe_: yea, i failed prob stats so i figure i'll let that to someone else :p 2010-09-14T10:57:25 *** panchi has quit IRC (Ping timeout: 252 seconds) 2010-09-14T10:58:55 bartwe_, dstufft: the current algorithm we are looking at picks a bot with a wide confidence interval and then pairs it with a bot near it in rank, but with a long tail so we still get the occasional wild game 2010-09-14T10:59:10 at least, i think that's what we are agreeing to try 2010-09-14T11:02:23 So I used java with a bash script. Should I get the results afterwards, or do I have to go online to see what happened? 2010-09-14T11:03:58 *** vikhyat has joined #aichallenge 2010-09-14T11:04:33 Oh, it tells me if I win, but not if I lose. 2010-09-14T11:06:38 jmcarthur: just checking, you explicitly check so that a bot does not fight itself 2010-09-14T11:07:20 bartwe_: no idea 2010-09-14T11:07:32 jmcarthur: a binary search strategy for new bots would be a quick way to find a good starting point 2010-09-14T11:07:37 bartwe_: that is certainly something to watch for in the future scheduler. i have no idea about the current one 2010-09-14T11:08:07 bartwe_: no, the confidence interval is going to be better than binary search because we also are uncertain about the ranks of the bots we are "inserting" the new bot into 2010-09-14T11:08:59 it would be like trying to use binary search to insert an element into an unsorted array 2010-09-14T11:10:37 my preference would be to favor matches where the bots' confidence intervals have the greatest overlap, but we have to admit some more variety in the matches, so the close-rank-with-long-tail version will work 2010-09-14T11:13:20 *** ajhager has joined #aichallenge 2010-09-14T11:15:09 *** mega1 has joined #aichallenge 2010-09-14T11:16:03 *** Vi0 has joined #aichallenge 2010-09-14T11:16:34 will it matter wheter you win in 200 rounds or in 50? 2010-09-14T11:17:45 *** PurplePanda has left #aichallenge 2010-09-14T11:17:48 i have the perl starter pacadage 2010-09-14T11:17:55 is it possible to win against strong bot in 50 rounds? 2010-09-14T11:18:47 i think it still happens but thats not my point 2010-09-14T11:19:13 Naktibalda: given the rigth map i think it could 2010-09-14T11:19:17 it doesn't matter 2010-09-14T11:19:22 dependent on what the players did on the first move 2010-09-14T11:19:31 it should enable the better bots to advance faster because they were stronger 2010-09-14T11:19:34 *** phreeza has joined #aichallenge 2010-09-14T11:19:53 advance in the ladder i mean 2010-09-14T11:20:00 If you guys spent on your bots at least half the energy you spend discussing here, you would be in the top ten :-D 2010-09-14T11:20:15 clanbase does something similar like that 2010-09-14T11:20:16 Obeleh: then people will take over the galaxy, and just waste time till the end of the game, before putting the other person out of their misery 2010-09-14T11:20:20 best part about tcp.c ver is that you can open as many as you want at once :D 2010-09-14T11:20:26 i have 6 games running right now 2010-09-14T11:20:26 im in the bus :p 2010-09-14T11:21:02 Obeleh: a win is a win 2010-09-14T11:21:10 dstufft: why would you want to win with fewer rounds? 2010-09-14T11:21:11 *** Vi0 has quit IRC (Ping timeout: 276 seconds) 2010-09-14T11:21:15 Baphomet: Shakey boa? 2010-09-14T11:21:25 retybok: i believe i just beat you :P 2010-09-14T11:21:36 boa? 2010-09-14T11:21:40 Obeleh: the faster games are overwith the more games the server can play 2010-09-14T11:21:41 ah, no, I see Baphomet 2010-09-14T11:21:42 Obeleh: seeing as how the goal is simply to win within 200 rounds and not to win as fast as possible, changing the score criteria based on number of turns to win would be deceptive 2010-09-14T11:21:59 dstufft: ^^ 2010-09-14T11:22:09 Baphomet: glad you didn't give up the other day now? :P 2010-09-14T11:22:32 sry i made a mistake 2010-09-14T11:22:32 i beat vortex11 2010-09-14T11:22:36 hehehe yea im still loosing most of the time but atleast i can find out if my bot sucks really quickly :p 2010-09-14T11:22:52 nice antimatroid ;) 2010-09-14T11:23:23 i meant why would you want to win with like 200 rounds 50 is better 2010-09-14T11:23:46 Obeleh: but would you go out of your way to change your strategy so that that happens at the risk of having a bot that overall performs worse? 2010-09-14T11:24:05 antimatroid: you're part of the lucky few who still beat me :-D 2010-09-14T11:24:12 Baphomet: has your bot any intelligence? it looks like it sends 6 ship fleets to all sides 2010-09-14T11:24:18 i think it broadens the game yes 2010-09-14T11:24:42 i still have lots to add to mine, so i'm glad it's at least doing alright off of data and a few well chosen heuristics 2010-09-14T11:24:44 I try :( 2010-09-14T11:24:57 I havnt worked out how to factor distance into the dest calculation yet 2010-09-14T11:24:59 still seem to get caned by a couple of people 2010-09-14T11:25:15 every aspect ads another dimension of skill 2010-09-14T11:25:16 antimatroid: same for me :) 2010-09-14T11:25:19 Baphomet: i pretty much rewrote planetwars for that 2010-09-14T11:26:04 i also havnt worked out how to make my planets send fleets to diferent locations in a single turn 2010-09-14T11:26:17 so when they attack all my planets shoot for the same dest :( 2010-09-14T11:26:19 *** staii has quit IRC (Ping timeout: 252 seconds) 2010-09-14T11:26:26 *** vikhyat has quit IRC (Quit: Leaving) 2010-09-14T11:27:37 just mark planet as protected and check this property before sending fleet from other planet 2010-09-14T11:27:55 Obeleh: if you factor in how many MORE ships a player wins by, then you provide incentive for a bot to "win" by getting the opponent down to one planet, and sending in ships to cull the herd, but not enough to take it over 2010-09-14T11:28:10 and then just sit and do nothing until the turns run out where you finally destroy them 2010-09-14T11:28:13 Baphomet: you are allowed to make as many IssueFleet calls as you want in a turn 2010-09-14T11:28:32 he exploits this right totally :) 2010-09-14T11:28:39 experimental support for fighting bots near in ranking is being tested. 2010-09-14T11:28:45 dstufft: i wouldnt do that no 2010-09-14T11:28:47 *** ademar_ has joined #aichallenge 2010-09-14T11:28:51 is it still legal to send multiple fleets with the same source and destination in one move? 2010-09-14T11:28:54 Obeleh you might not, but people would 2010-09-14T11:29:06 It first picks one of 5,10,20,40,100,200,1000 2010-09-14T11:29:07 antimatroid: I would think so 2010-09-14T11:29:09 i've gone under the assumption that's going to remain, but not spaming, just not to combine 2010-09-14T11:29:22 and then picks a random bot that close to you in rankings 2010-09-14T11:29:30 just give more points to someone who wins in fewer rounds sounds fair to me 2010-09-14T11:29:42 so...errbot didn't do very well overnight 2010-09-14T11:29:47 antimatroid2.0 is kicking my butt 2010-09-14T11:30:26 antimatroid: no its not actually ive done it many times 2010-09-14T11:30:43 in fact my current code does it 2010-09-14T11:31:12 bartwe still beats me everytime 2010-09-14T11:31:13 antimatroid, be careful though, I do believe there is an upper-limit to the number of fleets you can send at once... 2010-09-14T11:31:27 rogue780: since when? 2010-09-14T11:31:51 since rogue780 lost to PHP bot :) 2010-09-14T11:31:56 since around ~450 fleets causes most games to time out 2010-09-14T11:32:07 Naktibalda, that was ONCE 2010-09-14T11:32:09 ;) 2010-09-14T11:32:34 ah crud, is there a limit posted somewhere? i'm going to have to incorporate that in 2010-09-14T11:32:37 *** amstan has quit IRC (Remote host closed the connection) 2010-09-14T11:33:13 antimatroid, i don't think it's an explicit limit, I think it's the result of a software deficiency 2010-09-14T11:33:45 *** amstan has joined #aichallenge 2010-09-14T11:33:45 *** ChanServ sets mode: +o amstan 2010-09-14T11:34:24 i wish i knew how to make my bot attack the closest planets on first turn 2010-09-14T11:34:36 would help so much lol 2010-09-14T11:35:08 baphomet: create lists of planet indexes based on shortest distances 2010-09-14T11:35:32 remember, pencil and paper are your friend 2010-09-14T11:35:48 Baphomet, pw.Distance(planet1, planet2) 2010-09-14T11:36:17 rogue780: i know how to do that, i just dont know how i use that when i find the dest 2010-09-14T11:36:20 around 460-500 fleets caused the engine on benzedrine to give a malformed state 2010-09-14T11:36:47 i need to find a way to loop through all the planets and find the smalest ones 2010-09-14T11:37:08 Baphomet, what language are you using> 2010-09-14T11:37:11 *** delt0r___ has joined #aichallenge 2010-09-14T11:37:13 c++ 2010-09-14T11:37:14 antimatroid: just beat you :-P 2010-09-14T11:37:22 :p 2010-09-14T11:37:31 that was a long game 2010-09-14T11:38:19 Naktibalda: example http://www.benzedrine.cx/planetwars/canvas?game_id=1284477417|a1k0n-0.2|antimatroid2.0 2010-09-14T11:38:39 wow--the provided java playgame was slower than i thought... 2010-09-14T11:38:50 Baphomet, here's how I do it in Java. It might help you. 2010-09-14T11:38:51 http://pastebin.ca/1940650 2010-09-14T11:38:58 *** delt0r_ has quit IRC (Ping timeout: 252 seconds) 2010-09-14T11:39:01 *** Nuriaion has quit IRC (Ping timeout: 245 seconds) 2010-09-14T11:39:05 thanks it probbly will 2010-09-14T11:39:25 yeah, can't really say it was an overwhelming victory :-) 2010-09-14T11:39:44 a win is a win 2010-09-14T11:40:53 *** amstan has quit IRC (Ping timeout: 272 seconds) 2010-09-14T11:41:13 cya all 2010-09-14T11:42:51 Posted code review, dstufft 2010-09-14T11:44:06 antimatroid lost the match when he left his homeplanet 2010-09-14T11:44:54 danielvf: link? 2010-09-14T11:44:56 fuck google code 2010-09-14T11:45:04 yeah, it sucks 2010-09-14T11:45:25 btw, to find code reviews, go to http://code.google.com/p/ai-contest/source/list 2010-09-14T11:45:27 Obeleh: this map is horrible for expanding bots 2010-09-14T11:45:27 *** HakanD has quit IRC (Ping timeout: 252 seconds) 2010-09-14T11:45:34 (source > changes) 2010-09-14T11:46:05 It shows that the context of your planets should really influence your choices 2010-09-14T11:46:43 or simply put... its expected statw 2010-09-14T11:46:47 state 2010-09-14T11:47:30 aww, it keeps giving me games against bartwe who owns me 2010-09-14T11:47:53 retybok: I think it's a distance to enemy planet that matters 2010-09-14T11:48:14 http://ai-contest.com/forum/viewtopic.php?f=17&t=593 2010-09-14T11:48:58 Naktibalda: the amount of ennemies on those planets is also important 2010-09-14T11:49:01 *** McLeopold has joined #aichallenge 2010-09-14T11:49:21 i was hoping for some feedback on my post but nothing came :( 2010-09-14T11:49:24 it's the same on start 2010-09-14T11:49:35 danielvf: is the change to not choose a random first player for a match being delayed for later? 2010-09-14T11:49:45 *** AlliedEnvy has quit IRC (Ping timeout: 272 seconds) 2010-09-14T11:50:29 Janzert: In the last half hour, the server started running experimental code that usually picks near bots. 2010-09-14T11:50:49 danielelvf: i'd take any game :P 2010-09-14T11:51:08 I meant, I was reading through the code review and saw that the first player is still random choice 2010-09-14T11:51:44 *** Kastra has joined #aichallenge 2010-09-14T11:52:10 danielvf: you mean ai-cotest.com is now choosing closely ranked bots? 2010-09-14T11:52:34 Obeleh: Yes. 2010-09-14T11:52:42 koooool 2010-09-14T11:52:44 also I'm wondering why the magic array of numbers for pickable_spread instead using one of the distributions from the random module? 2010-09-14T11:52:59 Janzert: I am open to improvement 2010-09-14T11:53:10 i was really waiting for that :) 2010-09-14T11:53:16 Now if it would only send compilation errors by email... 2010-09-14T11:53:17 You want to pastebin me a reccomendation? 2010-09-14T11:53:17 danielvf: instead of Random should we instead try and pick someone who hasn't played a game? 2010-09-14T11:53:26 or hasn't played it recently 2010-09-14T11:53:44 I've been in the challenge for two days now, have a good ranking on the tcp server but I still haven't managed to make a single submission :-/ 2010-09-14T11:53:55 retybok link me your profile 2010-09-14T11:54:19 gotta take off for work soon, but I can try to get something real quick. Is there a specific distribution or goal for pickable_spread now? 2010-09-14T11:54:27 damn you retybok, you just beat me again 2010-09-14T11:54:32 http://ai-contest.com/profile.php?user_id=7425 2010-09-14T11:54:45 *** rwa has quit IRC (Quit: rwa) 2010-09-14T11:54:48 I just got destroyed by last year's winner :-/ 2010-09-14T11:55:03 *** amstan has joined #aichallenge 2010-09-14T11:55:03 *** ChanServ sets mode: +o amstan 2010-09-14T11:55:13 Janzert: nothing more than usually play a close bot, occasionally play a bot farther away. 2010-09-14T11:55:14 sorry :( 2010-09-14T11:55:21 ok 2010-09-14T11:55:36 :) 2010-09-14T11:55:48 *** amstan has quit IRC (Client Quit) 2010-09-14T11:55:51 anyway, it's bed time 2010-09-14T11:55:51 :) 2010-09-14T11:55:52 night 2010-09-14T11:55:57 night 2010-09-14T11:55:58 night 2010-09-14T11:56:01 *** amstan has joined #aichallenge 2010-09-14T11:56:01 *** ChanServ sets mode: +o amstan 2010-09-14T11:56:11 danielvf: does it favor bots that haven't played in a while or bots that have wide confidence intervals? 2010-09-14T11:56:38 jmcarthur: Are the confidence intervals in the db? 2010-09-14T11:56:55 danielvf: it'd be awesome if you could get the Elo score of the opponent in the profile page games list (i have a patch but it runs to slow on the db) 2010-09-14T11:56:59 danielvf: i think maybe they should be, but they aren't right now unless something has changes 2010-09-14T11:57:00 danielvf: they aren't. 2010-09-14T11:57:06 *changed 2010-09-14T11:57:12 jmcarthur: At the moment it just picks a random starting bot 2010-09-14T11:57:26 jmcarthur: Will look and see how hard that is to change 2010-09-14T11:57:29 i agree with a1k0n 2010-09-14T11:57:36 oh, i believe it should at least pick a bot based on how long it's been since the last game it played 2010-09-14T11:57:44 a1k0n: it's using the rank from the scoring rankings. 2010-09-14T11:57:52 a1k0n: rank, not score 2010-09-14T11:58:05 better would be confidence interval based i think, but that would be more work, and i don't think it matters *too* much 2010-09-14T11:58:11 no, i mean the score of all the people you've played, not your own score 2010-09-14T11:58:12 a1k0n: but should have the same effect, right? 2010-09-14T11:58:16 man, i just got pounded, perhaps i should stay up and make it better aha 2010-09-14T11:58:18 not strictly better either 2010-09-14T11:58:29 danielvf: http://code.google.com/p/ai-contest/issues/detail?id=87 2010-09-14T11:58:37 a1k0n: Oh - yeah 2010-09-14T11:58:42 *** Vi0 has joined #aichallenge 2010-09-14T11:58:56 http://code.google.com/p/ai-contest/source/branch?spec=issue90&branch=%2Fbranches%2Fgamelist_elo 2010-09-14T11:58:59 antimatroid: stop watching it and get some sleep :-D 2010-09-14T11:59:24 Was almost gonna do that. Gonna have to test to see if there is a fast way, or if we need to store more stuff in the DB. 2010-09-14T11:59:31 http://www.benzedrine.cx/planetwars/canvas?game_id=1284478921|a1k0n-0.2|Baphomet-ai-v5 did i do good? 2010-09-14T11:59:56 haha you keeled me! 2010-09-14T12:00:04 im not too good at this programming 2010-09-14T12:00:07 *** iPanda has quit IRC (Ping timeout: 252 seconds) 2010-09-14T12:00:24 i can usually only beat the default bot 2010-09-14T12:00:54 Baphomet: good job! 2010-09-14T12:01:29 I can't beat example bots yet 2010-09-14T12:01:37 I'm going to be headed out soon. Might code the "fair" scheduling next. 2010-09-14T12:02:40 *** AlliedEnvy has joined #aichallenge 2010-09-14T12:02:47 danielvf: does a normal distribution with a standard devation of 10 ranks sound good for choosing the opponent? 2010-09-14T12:02:55 *** ademar_ has quit IRC (Remote host closed the connection) 2010-09-14T12:03:21 *** staii has joined #aichallenge 2010-09-14T12:03:33 Janzert: That's not going to play people far away often enough 2010-09-14T12:03:36 yea i cant beat bullybot or ragebot or dualbot 2010-09-14T12:04:00 You'd probably want to choose opponents who's high and low confidence barely overlap. 2010-09-14T12:04:32 1/3 of the time it will be farther, but do you want the standard deviation larger or are you wanting a flatter distribution? 2010-09-14T12:05:01 maybe the standard deviation should be a percentage of the total number of submissions? 2010-09-14T12:05:21 5-10%? 2010-09-14T12:05:43 challenge guys: why do you give the compile log on success, but not on failure? It would seem more useful to do the contrary :) 2010-09-14T12:06:28 it should send on both 2010-09-14T12:06:46 *** ermau|work has quit IRC (Ping timeout: 252 seconds) 2010-09-14T12:07:01 retybok did you resubmit? 2010-09-14T12:07:11 I think i just fixed you 2010-09-14T12:07:15 unless you resubmited 2010-09-14T12:07:31 er wait 2010-09-14T12:07:33 no i didnt 2010-09-14T12:07:37 i found your error tho 2010-09-14T12:07:51 Baphomet: good job beating a1k0n! 2010-09-14T12:08:42 fluke really considering i cant beat any of the starter bots :p thanks anyway 2010-09-14T12:08:47 retybok: see pm 2010-09-14T12:10:22 dstufft: http://pastebin.org/867879 2010-09-14T12:11:01 oops, slight fix http://pastebin.org/867887 2010-09-14T12:11:13 this uses a standard deviation of 10% 2010-09-14T12:11:20 Janzert: im the worst at prob stats 2010-09-14T12:11:27 the only real change from your code is the offset = line 2010-09-14T12:11:30 but random.gauss is faster then random.normalvariate 2010-09-14T12:11:34 you mean daniel's code 2010-09-14T12:11:48 *** criisis has quit IRC (Ping timeout: 240 seconds) 2010-09-14T12:12:15 oops, yes sorry I meant danielvf. Is speed a concern here? 2010-09-14T12:12:16 *** kirkeby has quit IRC (Ping timeout: 252 seconds) 2010-09-14T12:12:40 either way I think a real distribution is probably better, gauss would be fine 2010-09-14T12:12:47 trueskill has a probabilistic matchmaking formula 2010-09-14T12:12:53 dstufft: I teach statistic 2010-09-14T12:12:55 http://research.microsoft.com/en-us/projects/trueskill/details.aspx#matchmaking 2010-09-14T12:13:02 Janzert: it's not an immensve concern, but (unless my understanding is wrong), normal and gauss are the same thing 2010-09-14T12:13:17 yea 2010-09-14T12:13:32 so no reason not to use gauss since it's faster 2010-09-14T12:13:50 when you guys talk about staggered attacks, what do you mean? 2010-09-14T12:13:56 dstufft: I just resubmitted using the script I had made for the last challenge and it worked 2010-09-14T12:13:57 oops, yes you are correct 2010-09-14T12:14:00 sorry for the nois 2010-09-14T12:14:08 *** Obeleh has quit IRC (Quit: Bye) 2010-09-14T12:14:09 well generating normals or gauss random variates should be very fast 2010-09-14T12:14:43 a normal distribution *is* a Gaussian distribution... Just to be clear 2010-09-14T12:14:44 dstufft: thanks for the help 2010-09-14T12:15:36 so we have http://pastebin.org/867946 :) 2010-09-14T12:16:00 the tails on a gaussian distribution are too light; i'd use cauchy 2010-09-14T12:16:06 *** HakanD has joined #aichallenge 2010-09-14T12:16:06 *** Obeleh has joined #aichallenge 2010-09-14T12:16:14 *** ermau has joined #aichallenge 2010-09-14T12:16:41 a1k0n: Or a Laplace 2010-09-14T12:17:09 there should be a nonzero chance of getting opponents from far ranks, just to be sure we have a global optimum of ranking 2010-09-14T12:17:56 hmm. laplace is an interesting idea 2010-09-14T12:18:01 a1k0n: that would be best done with a mixture model 2010-09-14T12:18:03 i read that was lapdance 2010-09-14T12:18:06 as 2010-09-14T12:18:26 say a 5% chance of picking anyone at random 2010-09-14T12:18:27 delt0r___: in other words, just use a normal distribution X% of the time and uniform 100-X% 2010-09-14T12:18:33 yea 2010-09-14T12:18:42 a1k0n, rankings update right? So as you move up and down your range of potential opponents changes, and you have th epossibility of global optimum 2010-09-14T12:18:48 i suspect that we might have a lot of A beats B beats C beats A situations in this contest. i don't think we are likely to have very clear optimum. i guess we are likely at least in the top ranks 2010-09-14T12:18:53 *** sjbr has joined #aichallenge 2010-09-14T12:18:58 I don't see how allowing battles between far ranks helps 2010-09-14T12:19:03 http://docs.python.org/release/2.5/lib/module-random.html shows the builtin distributions we have available with the version of python on the server any of those good? :) 2010-09-14T12:19:25 num1: Its stops clusters from forming... 2010-09-14T12:19:35 or would a1k0n's method be better? 2010-09-14T12:19:35 well it kind of establishes a baseline such that the difference in ranks between the top player and the bottom player doesn't require inferring from a chain of 500 different games 2010-09-14T12:19:37 num1: local maxima are possible (likely, i suspect) 2010-09-14T12:20:05 delt0r, a1k0n jmcarthur you're right, I stand down :) 2010-09-14T12:20:06 difference in elo score, sorry, not ranks 2010-09-14T12:20:48 how long has the typical game been taking lately? 2010-09-14T12:21:07 wall clock wise or turns wise? 2010-09-14T12:21:14 wall clock 2010-09-14T12:24:49 i tested the mixture model yesterday actually on a fake game generator 2010-09-14T12:24:57 it does work, even if the random games are 1% of the time 2010-09-14T12:25:11 it takes a lot of games to differentiate the top 10 though 2010-09-14T12:25:28 how many? 2010-09-14T12:25:45 *** ermau is now known as ermau|work 2010-09-14T12:25:52 and what is your criteria for differentiation? 2010-09-14T12:26:02 minimal overlap of confidence intervals? 2010-09-14T12:26:08 correct ordering 2010-09-14T12:26:24 ah wow, so that it takes a lot is really saying something 2010-09-14T12:26:29 *** retybok has quit IRC (Ping timeout: 272 seconds) 2010-09-14T12:26:32 i generate games according to the elo model (probability that a beats b is 1/(1+10^(a-b)/400) 2010-09-14T12:26:50 good idea 2010-09-14T12:27:08 so it's really testing the model itself without saying anything about real games 2010-09-14T12:27:11 so how about this way of choosing an opponent http://pastebin.org/868056 2010-09-14T12:27:37 *** soswow_ has joined #aichallenge 2010-09-14T12:27:45 uniform distribution 5%, normal distribution 95% with stddev of 10% of rankings 2010-09-14T12:27:47 *** evgenius has joined #aichallenge 2010-09-14T12:28:27 after 5000 games, the top player is ranked 5th, 2nd is 1st, 4th is 2nd, 3rd is 3rd 2010-09-14T12:28:33 :o 2010-09-14T12:29:05 but i'm not doing it iteratively with confidence intervals or anything 2010-09-14T12:29:26 *** amstan has quit IRC (Ping timeout: 276 seconds) 2010-09-14T12:29:28 it's actually just uniform +/-20 ranks, or uniform across the board 2010-09-14T12:29:35 this is with 100 players btw 2010-09-14T12:29:52 do you think it would be better or worse with more players 2010-09-14T12:29:56 we have 1020 on the server right now 2010-09-14T12:30:00 clearly worse 2010-09-14T12:30:10 well, 1020 that have played games 2010-09-14T12:30:47 also, how is the spread of elo scores of your virtual top 5 or so players? 2010-09-14T12:30:51 are they close are distant? 2010-09-14T12:31:00 *close or distant 2010-09-14T12:31:20 Janzert: that looks good; it'd be cooler if we could use the confidence interval output by bayeselo as ~3*stddev 2010-09-14T12:31:44 yeah, hopefully that will be available in the future 2010-09-14T12:31:46 yeah i think we need to put the confidence in the database for that 2010-09-14T12:31:47 jmcarthur: i'm going to pastebin some stuff, one sec 2010-09-14T12:31:51 a1k0n: alright 2010-09-14T12:32:29 perl game generator; http://pastebin.org/868114 2010-09-14T12:32:43 oops, there was a bug in that last pastebin. this should work better http://pastebin.org/868110 2010-09-14T12:33:06 output: http://pastebin.org/868116 2010-09-14T12:33:48 er, http://pastebin.org/868123 has the actual elo it was using 2010-09-14T12:34:19 danielvf: if you're around here is the modified while loop for choosing opponents using a1k0n's method http://pastebin.org/868110 2010-09-14T12:34:38 what is it doing now? uniform +/- 10 ranks or something? 2010-09-14T12:35:23 uniform using a magic array to choose the distance 2010-09-14T12:35:43 see http://code.google.com/p/ai-contest/issues/detail?id=151 2010-09-14T12:35:45 a1k0n: it picks a range from an array randomly, then picks a bot within that range randomly 2010-09-14T12:35:51 iiuc 2010-09-14T12:36:06 *** kozlovsky has joined #aichallenge 2010-09-14T12:36:39 oh. huh. 2010-09-14T12:36:57 kindof approximates a laplace distribution or something. 2010-09-14T12:37:06 a steppy one. 2010-09-14T12:37:19 time for work, later all 2010-09-14T12:37:35 something like that. i didn't pay attention to the actual distribution. it just had that tail i was looking for 2010-09-14T12:37:53 (i didn't write that code, in case my language implies otherwise) 2010-09-14T12:38:24 wtf? a geek squad repair unit just showed up here. how embarassing. 2010-09-14T12:38:31 lol 2010-09-14T12:40:10 it's actually kinda clever. i wonder how well it works. 2010-09-14T12:40:26 probably is reasonable 2010-09-14T12:40:38 *** td123 has quit IRC (Quit: WeeChat 0.3.3) 2010-09-14T12:40:49 *** HakanD has quit IRC (Quit: Colloquy for iPhone - http://colloquy.mobi) 2010-09-14T12:40:54 you're going to get a shitton of games against the same people, but that's probably good as it establishes correct probabilities of winning 2010-09-14T12:41:15 i'd be most interested in the games i keep losing anyway 2010-09-14T12:42:28 if eventually about half of my games are losses then that's probably a sign that the scheduler is working well 2010-09-14T12:42:42 the sooner it gets to that point the better 2010-09-14T12:43:00 maybe that could be a good criteria to check for as we try to nail this down 2010-09-14T12:43:09 it's not as important as getting an accurate rank though 2010-09-14T12:43:36 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T12:44:49 alright....I'm starting over. 2010-09-14T12:47:13 a1k0n: what do you think of this idea? http://ai-contest.com/forums/viewtopic.php?f=19&t=568#p3370 2010-09-14T12:47:26 i bet a lot of people further down do want a chance to play some top bots every once in a while 2010-09-14T12:47:36 *** td123 has joined #aichallenge 2010-09-14T12:47:36 that obviously can't be the only thing we look for 2010-09-14T12:48:09 but if we could perhaps weight the choice based on distance from top in addition to distance from bot, maybe that could lead to interesting matchups? 2010-09-14T12:48:17 for the lower ranks at least 2010-09-14T12:49:08 why not just random pairings ? 2010-09-14T12:49:10 it might be bad though because it means lower ranked bots might be more starved for games 2010-09-14T12:49:28 edcba: that's what we did up to today and it has led to uninteresting, uninformative games 2010-09-14T12:49:57 edcba: we need games to cause the score to narrow in more quickly, and we need the games to be more interesting matchups 2010-09-14T12:50:23 ok then swiss 2010-09-14T12:50:59 another idea if you don't mind: when you submit a new version of you bot, why not give you a starting rating the same as it was before submission? it's too painful to get through ranks as games are played too rare 2010-09-14T12:51:18 *** Bobng has quit IRC (Quit: Leaving) 2010-09-14T12:51:42 edcba: that's kind of the direction we're moving in already 2010-09-14T12:51:43 *** Bobng has joined #aichallenge 2010-09-14T12:51:55 evgenius: the system is unable to work that way, unfortunately 2010-09-14T12:52:00 are we talking about tcp or officials btw ? 2010-09-14T12:52:09 evgenius: the ratings are updated in batches based on games played. there is no seed value 2010-09-14T12:52:16 edcba: official 2010-09-14T12:52:58 *** deepblue has joined #aichallenge 2010-09-14T12:53:01 hm. but batches use your current rating in the calculations? 2010-09-14T12:53:06 evgenius: at the most it would buy you one game close to your original rank, but it won't be informative enough to keep you there 2010-09-14T12:53:09 morning 2010-09-14T12:53:21 evgenius: no, the batch jobs completely recalculate based solely on games played 2010-09-14T12:53:25 official should use swiss seems no brainer :) 2010-09-14T12:53:46 edcba: it's not that simple 2010-09-14T12:54:01 why ? 2010-09-14T12:54:07 edcba: the scoring system is a bayesian estimate of your eventual elo score. our goal is to narrow in on that more quickly 2010-09-14T12:54:21 edcba: it's not just a "add points when you win" kind of deal 2010-09-14T12:54:47 *** Pitel has quit IRC (Quit: KTHXBYE) 2010-09-14T12:55:40 i'm chess player and we use elo and swiss pairing in tournaments 2010-09-14T12:55:57 edcba: but this isn't straight up elo 2010-09-14T12:56:02 edcba: this is a bayesian estimate of it 2010-09-14T12:56:08 I just played a match on the tcp server. How do I find it? my username is Obeleh 2010-09-14T12:56:20 Obeleh: go to your profile 2010-09-14T12:56:23 oh tcp 2010-09-14T12:56:30 edcba: also there are no real rounds, since new submissions are constantly coming in and sometimes replacing old ones 2010-09-14T12:56:32 Obeleh: sorry, just look at the rankings and find your name 2010-09-14T12:56:41 evgenius: Using the previous rating as an estimate for a new version's rating seems reasonable but can totally fail, as happened to me years ago when I accidentally introduced a bug into my chess program that effectively changed the eval to "return 0;" 2010-09-14T12:56:42 its not there (yet?) 2010-09-14T12:56:46 Obeleh: give it time 2010-09-14T12:56:51 Obeleh: it has to run the rank updater 2010-09-14T12:56:59 ah ok 2010-09-14T12:57:00 thanks 2010-09-14T12:57:32 Does the tcp server mind if you run your bot twice? 2010-09-14T12:57:43 Obeleh: run it as many times as you can 2010-09-14T12:57:53 Thanks again :) 2010-09-14T12:57:54 *** JCS^ has joined #aichallenge 2010-09-14T12:57:55 Obeleh: your rank will become more accurate as you do it 2010-09-14T12:58:08 Obeleh: as long as you keep using the same name 2010-09-14T12:58:24 yes i understand 2010-09-14T12:58:25 Obeleh: http://www.benzedrine.cx/planetwars/getplayer?player=Obeleh 2010-09-14T12:59:19 Tank you :) 2010-09-14T13:00:03 Janzert: wait for the rounds to be played before replacing bots 2010-09-14T13:00:57 why using bayesian estimation ? 2010-09-14T13:01:04 *** ulope has left #aichallenge 2010-09-14T13:01:59 *** ajhager has joined #aichallenge 2010-09-14T13:02:37 edcba: we can't play enough games to arrive at an accurate number otherwise 2010-09-14T13:02:44 edcba: I imagine mostly because bayeselo is freely available, pretty easy to integrate and well regarded 2010-09-14T13:03:05 edcba: it arrives at fairly close numbers more quickly 2010-09-14T13:03:11 important for testing your bot 2010-09-14T13:04:53 swiss pairings could still be done with bayeselo running later, but it doesn't really work if everyone's submitting bots at different times 2010-09-14T13:05:18 jmcarthur: re: that forum post. unsure. might make sense to give the top players more games just so they have to defend their title more 2010-09-14T13:05:27 yeah... 2010-09-14T13:05:29 i am also unsure 2010-09-14T13:05:35 i would want to test it first 2010-09-14T13:05:54 to see what effects it might have 2010-09-14T13:06:17 a1k0n: i think the weights suggested are far too heavy though 2010-09-14T13:06:19 a mixture model sounds pretty ideal here: we could have some element of uniform game picks, some element of nearby rank picks, and some element of p(rank) ~ 1/rank or so 2010-09-14T13:06:31 yeah 2010-09-14T13:07:13 i would still like to toy with rankings where confidence interval is available as a variable too, but that can wait 2010-09-14T13:07:37 i want to see what trueskill's matchmaking formula would end up being like when applied to bayeselo 2010-09-14T13:08:34 it's calculates the probability that a given pair of players should play against each other based on how "fun" the game is likely to be, where "fun" means "likely to draw". not sure how well that applies here 2010-09-14T13:09:25 is it alright that i run the tcp one on a loop so im always playing? 2010-09-14T13:09:31 Baphomet: yes 2010-09-14T13:09:35 im pretty sure a few other guys are going it anyway 2010-09-14T13:09:36 Baphomet: although i might make a suggestion: 2010-09-14T13:09:43 maybe a wait? 2010-09-14T13:09:51 Baphomet: put a delay in your loop so you have a chance to stop it without forfeiting a game 2010-09-14T13:10:11 good idea 2010-09-14T13:10:26 Baphomet: also try to pick a delay that others are "unlikely" to pick so that you might get some more variety in your matchups 2010-09-14T13:10:32 since a lot of people use delays too 2010-09-14T13:11:14 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-09-14T13:14:53 * mightybyte suggests a prime number 2010-09-14T13:15:49 *** hornairs has joined #aichallenge 2010-09-14T13:16:23 mightybyte: everybody do that! :P 2010-09-14T13:16:27 *does 2010-09-14T13:16:39 you have to be sneakier 2010-09-14T13:17:04 (not really) 2010-09-14T13:17:09 :| 2010-09-14T13:17:15 a mersenne prime divided by 2^n 2010-09-14T13:17:53 lol i actually have a program that calculates mersenne primes that i wrote in c++ 2010-09-14T13:18:13 was one of my first programs ever too 2010-09-14T13:18:15 map pred $ iterate (*2) 1 2010-09-14T13:18:20 done 2010-09-14T13:18:24 your code is obsolete 2010-09-14T13:18:52 what is pred there? 2010-09-14T13:19:06 actually that's just mersennes, not mersenne primes, my bad 2010-09-14T13:19:11 i was misreading what Baphomet said 2010-09-14T13:19:27 a1k0n: pred just decrements by one 2010-09-14T13:19:36 oh, predecrement or something? 2010-09-14T13:19:40 indeed, condering mine is 40 lines long total lol. but i was like 15 when i made it 2010-09-14T13:19:50 predecessor 2010-09-14T13:19:56 ah 2010-09-14T13:20:05 Baphomet: if yours did mersenne primes then it does more than mine 2010-09-14T13:20:13 isn't that a little general for numbers? 2010-09-14T13:20:27 who said (2*) was just for numbers? 2010-09-14T13:20:36 who says the whole idea can't be for more general things? 2010-09-14T13:20:45 it's the haskell way to generalize 2010-09-14T13:20:47 the 1 on the end there 2010-09-14T13:20:54 1 is not just integers 2010-09-14T13:21:01 ok ok, i give in 2010-09-14T13:21:13 haskell has overloaded numeric literals 2010-09-14T13:21:38 Here is my code: http://pastebin.com/Tgen9UVJ 2010-09-14T13:21:52 they are read as Integers and then fromInteger is implicitly applied 2010-09-14T13:22:07 and something similar for decimal numbers 2010-09-14T13:22:32 The main problem was, even when using an unsigned long int it still wasnt big enough considering the mersenne primes are like 10+ digits long after the 7th one 2010-09-14T13:22:51 Baphomet: yeah you need to use a bignum library or something to do it for larger numbers 2010-09-14T13:22:54 or roll your own 2010-09-14T13:23:37 hey can you guys think of any reason why PlayGame-1.2.jar might work fine when called from bash but not when called programatically (like from a system/exec call)? 2010-09-14T13:23:45 im getting really weird behaviour 2010-09-14T13:23:48 *** retybok has joined #aichallenge 2010-09-14T13:23:58 hornairs: inheriting file descriptors maybe? i dunno 2010-09-14T13:24:57 hornairs: Its often "paths" and pwd settings 2010-09-14T13:25:17 I got some strange behavour when i was testing with my own player 2010-09-14T13:25:21 its so weird, if i run the command my tournament program runs, pitching my bot against an example bot, my bot times out. if i run the command in my shell it works fine, and if i run the command pitching two example bots against each other it times out and doesn't do anything, process never exits but nothing comes out on its stdout/err 2010-09-14T13:25:39 that last example is run programatically that is 2010-09-14T13:25:53 It was due to the fact that I expected the proccess to keep the jvm alive 2010-09-14T13:25:56 its doesn't 2010-09-14T13:26:22 a1k0n: as a basic example of a sensible generalization of the arithmetic operations, they could operate on functions that take some parameter and return a number. for example (+5) * (+1) could be applied to 2 and you would get 21 as the result 2010-09-14T13:26:29 as the jvm shuts down it also closes the std in and err and out ect. The bot sees and eof and exits 2010-09-14T13:26:43 a1k0n: although that is actually still what we think of as arithmetic. there are other kinds of mappings that could make sense too 2010-09-14T13:26:48 yes yes 2010-09-14T13:27:12 delt0r___ ok thanks for the tips 2010-09-14T13:27:15 *** damag has joined #aichallenge 2010-09-14T13:27:16 i was just saying (-1) might be clearer. 2010-09-14T13:27:22 but perhaps it's not idiomatic. 2010-09-14T13:27:25 that is "negative one" in haskell 2010-09-14T13:27:31 you'd have to say subtract 1 2010-09-14T13:27:36 hornairs: But if works from a shell script its proably the envorment.. ie PATH etc 2010-09-14T13:27:37 that's what i was afraid of 2010-09-14T13:27:38 *** ulope has joined #aichallenge 2010-09-14T13:27:39 (kind of lame IMO) 2010-09-14T13:27:49 guys how i actually put a wait into a batch script? 2010-09-14T13:28:09 *** retybok has quit IRC (Read error: Operation timed out) 2010-09-14T13:28:12 and (\x x-1) is just silly 2010-09-14T13:28:41 *** Naktibalda has joined #aichallenge 2010-09-14T13:28:48 oh i see people use ping now to wait 2010-09-14T13:28:59 there isn't a 'sleep'? 2010-09-14T13:29:19 there is a sleep in most shells 2010-09-14T13:29:21 *** evgenius has left #aichallenge 2010-09-14T13:29:41 man sleep 2010-09-14T13:29:46 oh wrong window 2010-09-14T13:29:54 sleep is part of GNU coreutils 2010-09-14T13:30:00 yep 2010-09-14T13:30:16 Should be on most Linux systems I'd imagine. 2010-09-14T13:30:26 i like how everyone assumes im on linux. 2010-09-14T13:30:33 win xp ftw :D 2010-09-14T13:30:37 yes, i meant on windows 2010-09-14T13:30:43 i think cmd suupports sleep, doesn't it? 2010-09-14T13:30:44 well have fun Baphomet 2010-09-14T13:30:49 better you than me 2010-09-14T13:31:08 I booted to windows once... it was a saturday 2010-09-14T13:31:26 *** soswow_ has quit IRC (Ping timeout: 252 seconds) 2010-09-14T13:31:27 is the tournament manager running at all? 2010-09-14T13:31:41 so ehmz, what tactics do you guys use to avoid losing to much to neutrals ? 2010-09-14T13:31:43 must be 2010-09-14T13:32:05 bartwe: if you leave them alone they leave you alone. :) 2010-09-14T13:32:38 never lost to a neutral ? :D 2010-09-14T13:32:41 a1k0n: well, you want growth rate, also you want to not lose units to neutral when you can lose them against the enemy 2010-09-14T13:33:05 right. so you can try to wait for your opponent to send ships to neutrals and then follow up on the next turn 2010-09-14T13:33:08 not attacking neutral sounds a bit.... wastefull 2010-09-14T13:33:08 haha, this is funny :) http://www.benzedrine.cx/planetwars/canvas?game_id=1284484499|sambot|deepblue 2010-09-14T13:33:18 it's kinda gettin nowhere.. 2010-09-14T13:33:50 wow, deepblue doesn't go in for the kill 2010-09-14T13:34:02 lal 2010-09-14T13:34:05 it prefers to drain it seems 2010-09-14T13:34:08 just bouncing huge fleets around 2010-09-14T13:34:08 works just as well 2010-09-14T13:34:27 it's randomly flitting 4000 ships around, holy crap 2010-09-14T13:34:45 it sent 15 at the end 2010-09-14T13:34:52 Defense is the best offense? =/ 2010-09-14T13:35:03 I'd hate to be in that mission briefing room... 2010-09-14T13:35:22 ok now we know deepblue is crazy 2010-09-14T13:36:35 well, my bot is at least as crazy... 2010-09-14T13:36:51 probably, almost, 2010-09-14T13:37:04 mine is mentally challenged, but not mentally ill 2010-09-14T13:37:07 *** hellman has quit IRC (Remote host closed the connection) 2010-09-14T13:39:48 Who's declared the winner when a game is stopped, most ships or most planets? 2010-09-14T13:39:54 ships 2010-09-14T13:39:56 including fleets 2010-09-14T13:40:16 which makes sense 2010-09-14T13:40:25 So then, you don't need to ever go for the kill. 2010-09-14T13:40:47 if you knew how long the game would take, and your sure of the superior growth rate 2010-09-14T13:41:30 *** bhowmisu has quit IRC (Quit: Leaving) 2010-09-14T13:41:58 it's true. i guess it might have been too risky to do it in that situation. 2010-09-14T13:41:58 you don't need to know the length of the game if you know your growth rate is superior 2010-09-14T13:42:10 bartwe: Yea -- if your own more that half the growth rate... and have the same or more ships. Then turtle... 2010-09-14T13:42:13 Unless your comming from behind. 2010-09-14T13:42:56 but a concentrated attack could still remove planets from you...turtleing would be that easy 2010-09-14T13:43:01 thats the problem, different tactics for different opponents 2010-09-14T13:43:18 wouldn't be that easy 2010-09-14T13:43:52 I suppose it depends on the map if you can turtle or not. 2010-09-14T13:44:11 Is there a way for a bot to surrender if it knows it can't win? Might save some cpu time. I tried issuing a 0,0,0 order but benzedrine's server didn't kill me 2010-09-14T13:44:17 if a concentrated attack removes a planet from you then it's still easy to counter after an epic turtle like that 2010-09-14T13:44:41 eburnette: i don't think i would do that. there's always a chance that the opponent screws up 2010-09-14T13:44:56 like, crashes or something 2010-09-14T13:45:00 eburnette: maybe just do something extremely desperate, like launch all ships at one planet of theirs 2010-09-14T13:45:31 i would just try to hold out for as long as possible if at all possible 2010-09-14T13:45:36 more time for opponent to make a mistake 2010-09-14T13:45:49 yeah this is no place for altruism, heh 2010-09-14T13:45:54 I don't count opponent crash as a true win 2010-09-14T13:46:01 eburnette: the server does 2010-09-14T13:46:31 I know, but if I'm losing at move 30 I'd rather give up than watch 170 more moves inch by 2010-09-14T13:46:42 you don't have to watch it :P 2010-09-14T13:46:47 you have no fighting spirit 2010-09-14T13:47:03 I have a debugging spirit 2010-09-14T13:47:16 eburnette: if you ^C the game it will count as a loss 2010-09-14T13:47:18 just watch the first 30 and then let it be while you debug the first 30 2010-09-14T13:47:27 yeah and it's interactive just do that 2010-09-14T13:47:29 *if it's 2010-09-14T13:47:45 so yeah, you can just disconnect to forfeit 2010-09-14T13:47:56 which finally answers your original question 2010-09-14T13:48:05 eburnette: yeah you could always just divide by zero if you want to do it programmatically :) 2010-09-14T13:48:11 or exit 2010-09-14T13:48:11 (or just exit()) 2010-09-14T13:48:22 dividing my zero is more epic 2010-09-14T13:48:25 *by 2010-09-14T13:48:31 mine too though 2010-09-14T13:48:43 I'll try exiting. If it were C++ I'd *((char *)null)=0 :) 2010-09-14T13:49:18 eburnette: you should make sure to send a polite "gg" first ;) 2010-09-14T13:49:32 hah! there could be a 'gg' command instead of 'go' 2010-09-14T13:49:46 a1k0n: we don't even have to modify the engine 2010-09-14T13:50:00 i think that would just get ignored 2010-09-14T13:50:15 really? bad syntax doesn't kill the bot? 2010-09-14T13:50:29 kill in the losing sense 2010-09-14T13:50:55 i don't remember 2010-09-14T13:51:45 my bot happens to ignore things it doesn't understand, but i would think that a tournament engine would be more strict 2010-09-14T13:51:59 whatever 2010-09-14T13:52:01 *** hornairs has quit IRC (Quit: hornairs) 2010-09-14T13:52:03 i should actually code this thing... 2010-09-14T13:55:20 *** hornairs has joined #aichallenge 2010-09-14T13:58:33 hmm, my bot won against bot with 270 higher elo score and lost 30 elo points 2010-09-14T13:59:07 Baphomet: for the record, given a function isPrime, here's your program in haskell: filter isPrime . map pred . iterate (*2) $ 1 2010-09-14T13:59:18 isPrime p = p > 1 && (all (\n -> p `mod` n /= 0 ) $ takeWhile (\n -> n*n <= p) [2..]) 2010-09-14T13:59:24 hmm, it looks like ranking update is delayed 2010-09-14T13:59:32 could be faster though 2010-09-14T13:59:50 Naktibalda: it is 2010-09-14T13:59:56 Naktibalda: yeah ranks are calculated in batches 2010-09-14T13:59:59 Naktibalda: it's a script that runs every 7 minutes 2010-09-14T14:00:05 ah, ok 2010-09-14T14:00:21 usually it changes every game 2010-09-14T14:00:41 coincidence i'm sure 2010-09-14T14:00:54 mmmh 2010-09-14T14:01:03 just had a fight against the second placed bot 2010-09-14T14:01:04 yeah, I had a few fast games now, and it was the same 2010-09-14T14:01:10 I expected something more intense: 2010-09-14T14:01:11 http://ai-contest.com/visualizer.php?game_id=4469323 2010-09-14T14:02:26 *** braddunbar has joined #aichallenge 2010-09-14T14:04:40 alk0n doing an exit counts as a time out. I propose a "surrender" command that bot will send instead of "go". 2010-09-14T14:05:02 you mean "gg", not "surrender" ;) 2010-09-14T14:05:06 :D 2010-09-14T14:05:09 ^ 2010-09-14T14:05:18 ok sure :) 2010-09-14T14:05:24 i'll not implement giving up 2010-09-14T14:05:29 me either 2010-09-14T14:05:32 never give up! 2010-09-14T14:05:41 i may just for debugging to make things quicker 2010-09-14T14:06:12 that makes sense 2010-09-14T14:06:19 just exit the whole program and timeout 2010-09-14T14:06:23 *** retybok has joined #aichallenge 2010-09-14T14:06:58 yeah that works but i'd rather the other guy know I surrendered and he got a true win. not a big deal I guess. 2010-09-14T14:07:22 *** Baphomet has quit IRC (Ping timeout: 240 seconds) 2010-09-14T14:08:15 *** jaspervdj has quit IRC (Quit: leaving) 2010-09-14T14:08:18 i'll not implement let you down 2010-09-14T14:08:38 i'll not implement run around and desert you 2010-09-14T14:08:41 *** jaspervdj has joined #aichallenge 2010-09-14T14:08:56 lol 2010-09-14T14:09:02 oh bow 2010-09-14T14:09:05 boy * 2010-09-14T14:09:06 I don't think I'm coming back from this one: End of turn 50, my ships=113, enemy ships=909 2010-09-14T14:09:21 back for about 10min 2010-09-14T14:09:38 What we need to rankings calculations, is something that doesn't have to batch process 2010-09-14T14:09:43 but can update 'online' 2010-09-14T14:09:48 conceivable if you have 16 more growth than he has and are completely isolated 2010-09-14T14:09:52 don't give up eburnette! 2010-09-14T14:09:59 danielvf: trueskill! 2010-09-14T14:10:08 add a new battle, instantly the relevant adjustments are made 2010-09-14T14:10:11 jmcarthur: ah 2010-09-14T14:10:43 a little complex to implement maybe, but we don't have to worry about >2 players and stuff 2010-09-14T14:11:24 but trueskill has the benefits of bayeselo, and also comes with all the little ideas i've been proposing for our bayeselo setup like reporting with lower confidence bound and even has a matchmaking formula 2010-09-14T14:11:34 i don't know of any libraries for it though :( 2010-09-14T14:11:38 just papers and articles 2010-09-14T14:11:47 Hmm 2010-09-14T14:12:47 This seems like the best thing yet. 2010-09-14T14:12:52 there appears to be a ruby library for it, at the top of my google search for trueskill library 2010-09-14T14:12:55 bots are noticeably tougher today compared to last night 2010-09-14T14:13:04 jmcarthur: I saw a java implementation of it somewhere 2010-09-14T14:13:09 let me look... 2010-09-14T14:13:10 danielvf: we'd definitely need to test it hard before using it 2010-09-14T14:13:42 jmcarthur: Yeah, I won'nt just drop this one in on the server. 2010-09-14T14:14:06 danielvf: i don't know how quickly bots would rise to the top, etc. i also don't know if the mathmaking algorithm would make sense for us since it's optimized for "fun" (likelihood of draw) rather than for nailing down a rank quickly 2010-09-14T14:14:17 danielvf: either way, it is designed to nail it down fairly quickly at least 2010-09-14T14:14:28 *matchmaking 2010-09-14T14:14:48 i don't know if there is a tradeoff, is what i mean 2010-09-14T14:15:05 *** Baphomet has joined #aichallenge 2010-09-14T14:15:06 jmcarthur: C#: github.com/moserware/skills Java: github.com/nsp/jskills 2010-09-14T14:15:11 phreeza: nice 2010-09-14T14:15:26 Yeah, I don't see a con to it, compared to the existing system, other than that we would have to implement it. 2010-09-14T14:15:50 jmcarthur: they state in the paper that there's a F# implementation out there 2010-09-14T14:15:58 xvinyl: yeah 2010-09-14T14:16:10 xvinyl: i just don't know if we need to integrate it into existing code easily or not 2010-09-14T14:16:20 jmcarthur: may be easily convertible to Haskell and integrated easily 2010-09-14T14:16:27 jmcarthur: with game engine code 2010-09-14T14:16:36 xvinyl: you mean for a new game engine? 2010-09-14T14:16:47 our existing engine on the server is in python 2010-09-14T14:16:54 huh did i miss something? game enine, haskell? 2010-09-14T14:16:56 jmcarthur: hmmm i thought you were writing a game engine in Haskell 2010-09-14T14:17:07 xvinyl: no, just a simple starter 2010-09-14T14:17:16 yeah, a tournament manager. 2010-09-14T14:17:19 xvinyl: mightybyte was considering it at one time i think 2010-09-14T14:17:22 jmcarthur: ok sorry 2010-09-14T14:17:31 danielvf: hmm? 2010-09-14T14:17:39 jmcarthur: should not be that hard to do I guess 2010-09-14T14:17:57 jmcarthur: hmm at what? 2010-09-14T14:18:00 *** soswow has quit IRC (Ping timeout: 252 seconds) 2010-09-14T14:18:02 jmcarthur: as the bots already have to track and parse game states 2010-09-14T14:18:07 danielvf: i don't know what exactly you were responding to 2010-09-14T14:18:20 *** evgenius has joined #aichallenge 2010-09-14T14:18:21 ": hmmm i thought you were writing a game engine in Haskell" 2010-09-14T14:18:23 trueskill can also do team fights, just in case anyone would ever want to do that :) 2010-09-14T14:18:24 xvinyl: yeah. main difference is in managing two players instead of just one 2010-09-14T14:18:29 phreeza: yeah :) 2010-09-14T14:18:34 actually that would be exciting :) 2010-09-14T14:18:41 danielvf: who's writing a tournament manager in haskell? 2010-09-14T14:18:42 2v2 2010-09-14T14:18:43 ? 2010-09-14T14:18:51 lol, my bad 2010-09-14T14:19:10 jmcarthur: ok i think i confused everyone here... sorry guys 2010-09-14T14:19:11 we are modifying one in python - 2010-09-14T14:19:12 phreeza: would be cool, but certainly *after* this contest 2010-09-14T14:19:19 jmcarthur: sorry for confusion 2010-09-14T14:19:20 danielvf: heh, okay 2010-09-14T14:19:21 jmcarthur: yes 2010-09-14T14:19:51 jmcarthur: or as a general idea for the next contest, something that requires cooperation with other AIs 2010-09-14T14:20:03 phreeza: i thought of a few things like that 2010-09-14T14:20:21 jmcarthur: Diplomacy, for example 2010-09-14T14:20:25 the board game 2010-09-14T14:20:27 phreeza: i think it would be neat if the AIs have some communication channel but we don't specify the protocol so the contestants have to come up with one together ad hoc 2010-09-14T14:20:46 communications channel? how would that work 2010-09-14T14:20:49 via IRC, forums, mailing list, something 2010-09-14T14:20:53 they send age of empires taunts? 2010-09-14T14:20:58 hehe yea 2010-09-14T14:21:00 pgpaskar_: i dunno just a broadcast thing to teammates 2010-09-14T14:21:11 ascii strings or something 2010-09-14T14:21:19 pgpaskar_: that would be so great 2010-09-14T14:21:27 30 -- where is my mother? 2010-09-14T14:21:30 15 -- alahaha 2010-09-14T14:21:34 heh 2010-09-14T14:21:37 16 - im in ur base... killin ur dudes.. 2010-09-14T14:21:41 hehehe 2010-09-14T14:23:42 i should update bash 2010-09-14T14:23:43 danielvf: btw, if you are pursuing trueskill, i told somebody the other day that trueskill adapts to changing skill levels over time. i was wrong. that's a different scoring system. so if somebody mentions that when you talk about it you can correct them for me :) 2010-09-14T14:23:47 3.2.48 on osx.... 2010-09-14T14:23:51 i don't remember who i told that to 2010-09-14T14:24:10 jmcarthur: It really doesn't adapt to changes over time? 2010-09-14T14:24:22 danielvf: not that i'm aware of 2010-09-14T14:24:29 danielvf: after reading over how it works i don't see how it could 2010-09-14T14:24:39 jmcarthur: I'll be on the lookout for that then 2010-09-14T14:24:42 danielvf: it just models uncertainty very well with an incremental algorithm 2010-09-14T14:24:58 danielvf: and has other nice things that we need like leaderboard generation and matchmaking :) 2010-09-14T14:25:21 At least in the summary they were talking about "surprises" affecting rankings a lot. 2010-09-14T14:25:32 yeah, but elo does too. that's nothing all that unique 2010-09-14T14:25:43 they just include uncertainty in with the surprises 2010-09-14T14:25:49 Yeah, this needs to both generate leader boards and generate matches. 2010-09-14T14:26:00 it's still will always *narrow* the confidence interval. i don't think it ever widens it 2010-09-14T14:26:07 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-14T14:26:07 widening would be necessary to adapt to changing skill 2010-09-14T14:27:20 danielvf: another nice thing is that the matchmaking will pit new bots against established average bots rather than against other new bots 2010-09-14T14:27:22 jmcarthur: will be on the lookout 2010-09-14T14:27:47 *** hornairs has quit IRC (Ping timeout: 245 seconds) 2010-09-14T14:28:04 that way you don't end up with new bad bots playing against new awesome bots 2010-09-14T14:28:08 *** hornairs has joined #aichallenge 2010-09-14T14:28:30 jmcarthur: Ah 2010-09-14T14:32:53 i guess in a sense it does something like binary tree insertion for new bots when establishing mean skill, but reports the lower confidence bound so they still appear to be rising in the ranks rather than jumping around 2010-09-14T14:33:20 *** retybok has quit IRC (Ping timeout: 252 seconds) 2010-09-14T14:34:35 jmcarthur: yeah 2010-09-14T14:35:38 how do you guys feel about a 1 second grace period at the start of the game for library loading and initialization? 2010-09-14T14:36:07 hornairs: there may be something like that already in 2010-09-14T14:36:26 it could be before the map is sent so you cant get an edge by precomputing 2010-09-14T14:36:44 but maybe each bot should say ready before the first state is sent or something? 2010-09-14T14:37:07 *** retybok has joined #aichallenge 2010-09-14T14:37:27 might make it easier to diagnose complilation errors too 2010-09-14T14:37:33 Yeah 2010-09-14T14:37:51 hornairs: the first turn is 3 seconds long 2010-09-14T14:37:57 hornairs: after that is 1 second 2010-09-14T14:38:06 It seems like players can die, even if they still have a fleet. 2010-09-14T14:38:09 is that respected in showgame? 2010-09-14T14:38:11 playgame* 2010-09-14T14:38:18 hornairs: just the tournament i think 2010-09-14T14:38:21 afk 2010-09-14T14:38:22 *** HakanD has joined #aichallenge 2010-09-14T14:38:28 startup is 2 seconds, the first turn is then 1 second 2010-09-14T14:38:32 Zannick: oh? 2010-09-14T14:38:35 i thought it was 3 2010-09-14T14:38:37 * jmcarthur peeks 2010-09-14T14:38:40 according to the tourney engine 2010-09-14T14:38:44 it's like 3. 2010-09-14T14:38:48 i dont think playgame respects taht 2010-09-14T14:38:50 "like" 3? 2010-09-14T14:38:53 it is written as 3? 2010-09-14T14:39:09 *** canihazbacon has joined #aichallenge 2010-09-14T14:39:21 engine.py doesn't have a special case for the first turn 2010-09-14T14:39:23 whatever the case im getting my ass kicked by the first timeout 2010-09-14T14:39:34 but it waits 2 seconds for the startup 2010-09-14T14:39:37 the first turn is longer? 2010-09-14T14:39:40 yes 2010-09-14T14:39:40 s/for the/after/ 2010-09-14T14:40:00 ah okay 2010-09-14T14:40:14 i think i put that in the spec wrong then 2010-09-14T14:40:24 so not 2 seconds more for the first round 2010-09-14T14:40:32 it's two seconds for your bot to load up right? 2010-09-14T14:40:36 yes 2010-09-14T14:40:45 such as the jvm 2010-09-14T14:40:49 Zannick: the last paragraph of Bot I/O is wrong then, isn't it? http://ai-contest.com/specification.php#bot-io 2010-09-14T14:41:07 jmcarthur: yeah, slightly 2010-09-14T14:41:26 I hope you all know my grades are going to plummet because of this contest 2010-09-14T14:41:35 it's certainly something that may lead some to program their bots in a way that will make them crash 2010-09-14T14:41:43 http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/backend/engine.py#287 2010-09-14T14:42:00 2 seconds to spin up 2010-09-14T14:42:05 then each turn is 1 second 2010-09-14T14:42:19 i wonder if it worked that same way in the last contest 2010-09-14T14:42:29 that would explain why my bot timed out so much 2010-09-14T14:42:44 you'd have to ask someone who was here for that 2010-09-14T14:42:45 i was just flatly told that the first turn was 3 seconds, at the time 2010-09-14T14:43:42 well, thats what i understand from the docs 2010-09-14T14:44:16 HakanD: if you mean for this contest, i wrote those docs with an incorrect understanding of it ;) 2010-09-14T14:44:22 i'm making a patch now 2010-09-14T14:44:41 yep i mean this contest (: 2010-09-14T14:45:00 HakanD: well, fyi, the first turn is still just 1 second. sorry about that 2010-09-14T14:45:32 np, glad to figured that out sooner 2010-09-14T14:45:51 I ll play with it a bit to get it right 2010-09-14T14:47:28 hmz, thinking of going with a total reimplementation 2010-09-14T14:47:42 this sucks 2010-09-14T14:47:52 happens to me all the time 2010-09-14T14:47:55 it's the haskell way! 2010-09-14T14:47:59 rewrite rewrite rewrite 2010-09-14T14:48:07 (not officially) 2010-09-14T14:48:48 testing is turning out to be difficult 2010-09-14T14:49:37 bad strategy 2010-09-14T14:49:57 first strategy, grab all neutrals, then group and pickoff the easy planets 2010-09-14T14:50:20 however that became problematic when enemies started to wait for the other to grab neutrals first 2010-09-14T14:50:22 *** staii has quit IRC (Ping timeout: 240 seconds) 2010-09-14T14:50:54 *** Naktibalda has quit IRC (Ping timeout: 272 seconds) 2010-09-14T14:51:54 first strategy: win 2010-09-14T14:52:00 if that doesn't work, second strategy: don't lose 2010-09-14T14:52:07 har 2010-09-14T14:52:20 deepblue tries that it seems 2010-09-14T14:52:25 ;) 2010-09-14T14:52:30 simply going for growthrate + population 2010-09-14T14:52:37 If that doesn't work, third strategy: bitch in forums 2010-09-14T14:52:40 copybot has taught me very interesting things about the center planet 2010-09-14T14:55:47 an updated version of my errbot is out http://www.benzedrine.cx/planetwars/canvas?game_id=1284489442|DaTwinkDaddy-s8-33-g1808|ErrBot_0.0.4 2010-09-14T14:55:51 be careful what you learn from copybot 2010-09-14T14:56:03 *** sjbr has left #aichallenge 2010-09-14T14:56:11 *** alexshendi has joined #aichallenge 2010-09-14T14:56:39 copybot merely recites facts that you just told it 2010-09-14T14:56:49 you guys make beautiful patterns :D 2010-09-14T14:59:24 *** ajhager has joined #aichallenge 2010-09-14T15:00:11 rogue780: some more meat for my bot to chew on :-P 2010-09-14T15:00:46 jmcarthur: so did you finish your haskell starter package ? or do you use __init__'s one ? 2010-09-14T15:00:48 retybok, let not your heart be troubled. the current version is only a partial implementation of my new strategy 2010-09-14T15:01:33 this contest is a lot of fun :) 2010-09-14T15:01:41 even moreso with this channel :) 2010-09-14T15:01:49 yes :) 2010-09-14T15:02:04 I'm having more fun with this than tron 2010-09-14T15:06:42 xvinyl: still polishing it while i work on my bot 2010-09-14T15:07:23 xvinyl: it's more of a mini-starter. i'm not working about adding any convenience functionality. i'm just implementing some details from the spec and that's it 2010-09-14T15:07:30 *convenience functions 2010-09-14T15:07:45 jmcarthur: yep 2010-09-14T15:08:06 Hello folks a quick question. Can I use a shell script to start my bot? 2010-09-14T15:08:20 On the server? or localy? 2010-09-14T15:08:58 On the server, of course. Locally is no problem. 2010-09-14T15:09:44 alexshendi: no 2010-09-14T15:09:47 Hm.. maybe MyBot.sh ? ;) 2010-09-14T15:09:53 for the tcpserver, yes 2010-09-14T15:10:02 anyone have a fixed bashscript tot test against examplebots with a1k0n new playgame and showgame 2010-09-14T15:10:13 the server compiles and runs a command to start your bot based on the language 2010-09-14T15:10:44 alexshendi: why would you want to do that? 2010-09-14T15:10:56 ughs, bugs bugs bugs 2010-09-14T15:11:02 also it runs in the sandbox to prevent hacks 2010-09-14T15:11:53 jmcarthur: "Before starting to determine the new skill beliefs of all participating players for a new game outcome, the TrueSkill ranking system assumes that the skill of each player may have changed slightly between the current and the last game played by each player." 2010-09-14T15:12:16 danielvf: ah, yeah, it does do that 2010-09-14T15:12:16 retybok: To add command line arguments. 2010-09-14T15:12:25 danielvf: but it does that uniformly between all matches 2010-09-14T15:12:33 alexshendi: hard code them 2010-09-14T15:12:48 danielvf: we could remove that and simply add some uncertaintly artificially upon resubmission though 2010-09-14T15:12:52 Is ANSI C acceptable? 2010-09-14T15:12:53 would have the same effect 2010-09-14T15:12:57 (for the submission only, of course!) 2010-09-14T15:13:05 jmcarthur: exactly 2010-09-14T15:13:14 danielvf: that would be nice. we should do it :) 2010-09-14T15:13:41 It looks like the uber complicated parts of trueskill are the multi players on a team 2010-09-14T15:13:44 We don't need that 2010-09-14T15:13:47 yeah, which we don't care about 2010-09-14T15:14:12 I can't read their paper - too much greek. 2010-09-14T15:14:15 heh 2010-09-14T15:14:32 the web page is simple enough though i think 2010-09-14T15:14:36 Yeah 2010-09-14T15:15:00 danielvf: do you actually want to keep track of the player's rank and not bot skill ? 2010-09-14T15:15:09 danielvf: i mean 2010-09-14T15:15:11 *** montana1 has left #aichallenge 2010-09-14T15:15:12 xvinyl: I think that's the right way to go. 2010-09-14T15:15:24 Though I am open to being convenced otherwide 2010-09-14T15:15:27 *wise 2010-09-14T15:15:39 no, you want to track both 2010-09-14T15:15:47 jmcarthur: Why? 2010-09-14T15:15:49 danielvf: well, one 'plus' for the TrueSkill is, as jmcarthur said 2010-09-14T15:16:07 we want to report the rank as the lower confidence bound, but we can't do that unless we track all variables 2010-09-14T15:16:17 danielvf: uncertainty over skill always decreases 2010-09-14T15:16:22 in the db we actually want to store the mean and uncertainty 2010-09-14T15:16:24 You only have to track sigma and whatever the other one is 2010-09-14T15:16:32 danielvf: mu :) 2010-09-14T15:16:48 xvinyl: what we were talking about a second was to increase uncertainty when somebody uploads a new version of their bot 2010-09-14T15:16:49 xvinyl: no, unceranty goes up on submission, and possibly a tiny bit before each game 2010-09-14T15:16:57 xvinyl: rather than starting over from scratch 2010-09-14T15:17:00 Except for the message passing, the TrueSkill algo isn't too bad. 2010-09-14T15:17:09 xvinyl: Thanks 2010-09-14T15:17:16 danielvf: not before each game. we don't need that 2010-09-14T15:17:19 I've implemented it in python before. 2010-09-14T15:17:24 *** evgenius has left #aichallenge 2010-09-14T15:17:31 McLeopold: On github? 2010-09-14T15:17:31 danielvf: bots are not like humans in that their skills don't change unless you upload a new one ;) 2010-09-14T15:17:37 danielvf: my point is : a bot's skill never changes 2010-09-14T15:17:44 jmcarthur: exactly 2010-09-14T15:17:48 xvinyl: we should still do it upon resubmission though 2010-09-14T15:17:53 instead of starting over 2010-09-14T15:18:06 jmcarthur: i don't really see why 2010-09-14T15:18:21 xvinyl: so that they can go ahead and play similarly ranked players without having that rampup period 2010-09-14T15:18:32 jmcarthur: exactly 2010-09-14T15:18:35 xvinyl: but while still acknowledging that their skills may have changed 2010-09-14T15:18:44 jmcarthur: yep :) what i mean is 2010-09-14T15:19:04 jmcarthur: i think this algorithm finds the good skill level pretty fast 2010-09-14T15:19:14 jmcarthur: *right skill level 2010-09-14T15:19:34 Hmm. Does planet growth get added before or after fights? 2010-09-14T15:19:34 xvinyl: yeah, but it will always tighten the confidence interval unless we artifically loosen it 2010-09-14T15:19:50 before 2010-09-14T15:19:52 ath888: before 2010-09-14T15:19:55 Thanks. :) 2010-09-14T15:20:06 jmcarthur: again, if we are ranking bots and not players, confidence intervals should always tighten 2010-09-14T15:20:15 xvinyl: except for resubmission 2010-09-14T15:20:21 in which case they should loosen 2010-09-14T15:20:36 danielvf: no, not on github, i'd have to dig it up 2010-09-14T15:20:49 xvinyl: so our choices are to (a) reset all values for a bot upon resubmission, (b) increase uncertainty upon resubmission, or (c) do nothing upon resubmission 2010-09-14T15:20:58 Ah. I may ask you later for it. 2010-09-14T15:21:06 xvinyl: we are not considering adding uncertainty between games for the same version of the bot 2010-09-14T15:21:13 woop, inverted condition in cost function 2010-09-14T15:21:17 that won't help.... 2010-09-14T15:21:43 increase uncertainty, that is how trueskill did it. 2010-09-14T15:21:49 jmcarthur: i see it this way : each time a player uploads a new bot, it's just like if it was a new player 2010-09-14T15:21:56 McLeopold: well, trueskill proper does it between every game, which is wrong for us 2010-09-14T15:22:09 xvinyl: but it's not. it's highly likely to just be a changed version of the old one 2010-09-14T15:22:13 xvinyl: which is just like a skill change 2010-09-14T15:22:43 TrueSkill was based on time. You should probably use a constant. Assume a new bot will move 100% closer to high and 50% closer to bottom. 2010-09-14T15:23:03 McLeopold: why do it at all though? 2010-09-14T15:23:10 just curious, what about a sandbox period, 2-3 matches maybe, after resubmission? 2010-09-14T15:23:10 McLeopold: bots don't change skill over time unless you resubmit 2010-09-14T15:23:11 and don't change the mean, I guess 2010-09-14T15:23:22 retybok: no, you never change the mean artificially 2010-09-14T15:23:24 *** Yoshi-TS4 has joined #aichallenge 2010-09-14T15:23:29 *** braddunbar has quit IRC (Ping timeout: 240 seconds) 2010-09-14T15:23:40 If you have a tight deviation, then a new bot will move up the ranks slower if it never gets to play good bots. 2010-09-14T15:23:53 McLeopold: point 2010-09-14T15:24:10 If a bad bot gets improved, one game against a top player will tell you much more than serveral along the way. 2010-09-14T15:24:12 jmcarthur: ok so you assume each player's successive bots will be around the same level with slight alterations in level 2010-09-14T15:24:13 McLeopold: but once it's up there then it should not have a problem staying there even after resubmission 2010-09-14T15:24:17 jmcarthur: makes sense 2010-09-14T15:24:17 xvinyl: yeah 2010-09-14T15:24:21 jmcarthur: i guess :) 2010-09-14T15:24:22 *** ath888 has quit IRC (Ping timeout: 252 seconds) 2010-09-14T15:24:38 Yes, only after resubmittion. 2010-09-14T15:24:43 McLeopold: i'm just saying it's okay if the bot moves up slowly overall as long as it doesn't drop drastically between resubmissions 2010-09-14T15:24:45 okay 2010-09-14T15:24:50 McLeopold: we are in agreement then 2010-09-14T15:24:54 ;) 2010-09-14T15:25:21 xvinyl: we can pretty drastically increase the uncertainty without a huge penalty 2010-09-14T15:25:26 what is cameron.jp's handle in this channel? 2010-09-14T15:25:31 Bobng: j3camero 2010-09-14T15:25:59 *** ajhager has quit IRC (Ping timeout: 265 seconds) 2010-09-14T15:26:00 Oh, 35 hours idle -_- 2010-09-14T15:26:17 jmcarthur: one risk, though, is that in the case your new bot is actually worse than your previous one, you won't see it immediately 2010-09-14T15:26:25 A zero growth planet has been the bain of my bots existence. 2010-09-14T15:26:31 xvinyl: it would just mean that rank upon resubmission is low, but the bot will still get to play some higher ranked bots as we narrow back in on its score 2010-09-14T15:26:37 rock on. http://atom.research.microsoft.com/trueskillranking/rankcalculator.aspx 2010-09-14T15:26:53 xvinyl: no, you'll see it because the mean is more "mobile" when the uncertainty is high 2010-09-14T15:27:41 danielvf: yeah it's nice stuff 2010-09-14T15:27:44 *** RainCT has joined #aichallenge 2010-09-14T15:28:31 jmcarthur: okok you won :) it should be good enough this way 2010-09-14T15:28:48 This is exactly what I needed 2010-09-14T15:28:55 jmcarthur: in the end, the real problem is that the server can only play a ridiculous amount of games in one day 2010-09-14T15:29:14 xvinyl: yeah, which is why we need to be really careful about picking a great scoring system 2010-09-14T15:29:28 20000 games a day on this server 2010-09-14T15:29:39 Probably 40000 on the next 2010-09-14T15:29:40 My TrueSkill code looks like baby poo. And it's tied to a database. 2010-09-14T15:29:44 danielvf: it even shows match quality prediction :) 2010-09-14T15:30:42 *** espes has quit IRC (Ping timeout: 255 seconds) 2010-09-14T15:30:48 danielvf: well, i was talking about the official server 2010-09-14T15:30:53 :) 2010-09-14T15:30:54 Is there a faster way to do match quality prediction for us? 2010-09-14T15:31:06 danielvf: faster in what sense? 2010-09-14T15:31:08 danielvf: I'm talking about the official server 2010-09-14T15:31:12 *** sinan has joined #aichallenge 2010-09-14T15:31:34 danielvf: ok sorry 2010-09-14T15:31:36 danielvf: the formula just takes two players info and spits out a number between 0 and 1 2010-09-14T15:31:40 Right 2010-09-14T15:31:45 danielvf: where 1 is "awesome match" and 0 is "terrible match" 2010-09-14T15:32:10 which would mean to find the best next match would be num players squared ish 2010-09-14T15:32:13 danielvf: i guess it's pretty much a scaled difference of ELO ranks 2010-09-14T15:32:19 it's not that slow. i'm suspecting you mean "can we change it to take multiple players as input and spit out some matches" 2010-09-14T15:32:24 or something similar 2010-09-14T15:32:28 jmcarthur: Yeah 2010-09-14T15:32:42 danielvf: here's how i think we should do it... 2010-09-14T15:32:51 Are you talking about the trueskill match prediction? 2010-09-14T15:32:58 McLeopold: yup 2010-09-14T15:33:04 danielvf: we should make a pool of players that haven't played in a while, then pick the best match out of that pool 2010-09-14T15:33:08 I don't think you want close matches at all. 2010-09-14T15:33:17 I might be wrong... 2010-09-14T15:33:31 McLeopold: isn't the match prediction also tuned for picking matches that tighten the confidence interval quickly? 2010-09-14T15:34:06 McLeopold: We do want to try to get close matches 2010-09-14T15:34:07 danielvf: so the algorithm would still be n^2, but for a smaller number of players 2010-09-14T15:34:12 No, i think it's for picking the matches that involve very closely ranked bots. 2010-09-14T15:34:19 McLeopold: not really 2010-09-14T15:34:41 McLeopold: rank is a different idea from mean in this case 2010-09-14T15:34:59 Which would in turn tighten the confidence interval if it was right. 2010-09-14T15:35:00 McLeopold: a new bot has rank 0, but its mean is actually at the halfway point, so it will be matched with an average bot 2010-09-14T15:35:11 McLeopold: right. we want that 2010-09-14T15:35:20 Oh good, you calculating rank as mean - u 2010-09-14T15:35:34 mean - 3*stddev, iirc 2010-09-14T15:35:49 Basically if deep blue beats a starter bot, you gain no information about deep blue, and just that starter bot is not a top level bot 2010-09-14T15:35:51 or is it variance? 2010-09-14T15:35:53 don't remember 2010-09-14T15:36:09 danielvf: right, and that matchup would also probably not be picked 2010-09-14T15:36:23 jmcarthur: std dev in this case 2010-09-14T15:36:34 we can try these scenarios with that link that danielvf pasted earlier 2010-09-14T15:36:40 since it shows quality of match 2010-09-14T15:36:59 jmcarthur: some other measures use variance actually, but i don't think it would be of interest here 2010-09-14T15:37:02 I suppose if 2 bots have tight deviation, you won't gain much from matching them up. The don't have a chance to move up. 2010-09-14T15:37:05 *** espes has joined #aichallenge 2010-09-14T15:37:43 McLeopold: of course, if they are that tight, they don't need to move anyway 2010-09-14T15:38:04 McLeopold: and they will still get matched with newer bots, and unexpected results could move them quite a bit 2010-09-14T15:38:09 Unless it turns out there is some sort of paper, sissors, rock issue with bot stragegies. 2010-09-14T15:38:13 right 2010-09-14T15:38:18 which we are counting on not happening 2010-09-14T15:38:19 Oh yea, new bot.. 2010-09-14T15:38:23 (it's likely to happen though) 2010-09-14T15:38:24 McLeopold: that's the whole problem of ranking 2010-09-14T15:38:40 we have to just assume there are no rock paper scissors situations 2010-09-14T15:39:05 *** braddunbar has joined #aichallenge 2010-09-14T15:39:12 *** rwa has joined #aichallenge 2010-09-14T15:39:24 jmcarthur: trueskill takes care of this for us 2010-09-14T15:39:37 yeah 2010-09-14T15:39:45 trueskill is really quite awesome 2010-09-14T15:40:07 xvinyl: wait, do you mean rock paper scissors situations? 2010-09-14T15:40:10 jmcarthur: there will be rock/paper/scissors situations beacause it's inevitable 2010-09-14T15:40:15 *** hornairs has quit IRC (Quit: hornairs) 2010-09-14T15:40:32 xvinyl: yeah, which is exactly why we can't worry about it unless we have multidimensional rankings 2010-09-14T15:40:39 which is just ridiculous 2010-09-14T15:40:42 jmcarthur: yep ; the whole problem of ranking is that you have this kind of situation 2010-09-14T15:40:45 but awesome 2010-09-14T15:40:49 McLeopold: heh yeah 2010-09-14T15:41:08 jmcarthur: no because you still wouldn't have a total order 2010-09-14T15:41:26 xvinyl: that's the point of what i just said 2010-09-14T15:41:53 If you saved every match, you could recalculate rankings for only matches involving 1 person. 2010-09-14T15:42:11 McLeopold: what do you mean? 2010-09-14T15:43:00 I was trying to think of a way to easily have multi-dim rankings. 2010-09-14T15:43:07 Well I've got to work for a few. 2010-09-14T15:43:07 More of a "personal" ranking. 2010-09-14T15:43:35 McLeopold: what do you call a multidimensional ranking exactly ? 2010-09-14T15:43:39 I would vote for a tournament with looser bracket best of 5 :) 2010-09-14T15:43:41 Or at least to detect r/p/s 2010-09-14T15:44:18 Rankings for different bot types 2010-09-14T15:46:23 what different kind of bots? 2010-09-14T15:47:09 *** rwa has quit IRC (Quit: rwa) 2010-09-14T15:47:09 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-14T15:47:32 *** danielvf has joined #aichallenge 2010-09-14T15:48:00 Idea: 2010-09-14T15:48:04 I don't know yet. 2010-09-14T15:48:09 *** alexshendi has left #aichallenge ("Verlassend") 2010-09-14T15:48:18 Bots that stream... Bots that don't go for the kill 2010-09-14T15:48:21 We don't want to use match quality to choose matches 2010-09-14T15:48:30 Bots that fly 4000 ships around for fun... 2010-09-14T15:48:51 We want to start with a bot with high sigma 2010-09-14T15:49:05 And then pick, based on that. 2010-09-14T15:49:10 *** amstan has joined #aichallenge 2010-09-14T15:49:10 *** ChanServ sets mode: +o amstan 2010-09-14T15:49:13 What would the next pick be? 2010-09-14T15:49:20 yellow 2010-09-14T15:49:48 :P 2010-09-14T15:49:48 lol 2010-09-14T15:49:51 The losers will never get matches 2010-09-14T15:49:54 danielvf: trueskill would match such a bot with a player that has a similar mean but a tight sigma 2010-09-14T15:50:04 McLeopold: why not? 2010-09-14T15:50:47 jmcarthur: Ah. 2010-09-14T15:50:47 If you always pick high sigma, you'll never match them to bad bots, except for an unknown bad bot. 2010-09-14T15:51:04 High sigma for the first bot 2010-09-14T15:51:10 i think that's a bad idea 2010-09-14T15:51:19 i think it should just always be between two bots that haven't played in a while 2010-09-14T15:51:20 then look for match 2010-09-14T15:51:46 Once all the bots have sigma pushed down, then it will start to play around 2010-09-14T15:51:46 The first bot should be one that hasn't played, then find a good match. 2010-09-14T15:52:03 danielvf: the first bot should be the bot that has played the least recently, IMO 2010-09-14T15:52:10 I agree 2010-09-14T15:52:15 danielvf: and the bot to match with should be matched by trueskill formula 2010-09-14T15:52:28 we actually don't have to do a pool for that i guess 2010-09-14T15:52:34 jmcarthur: won't that result in the bot playing the same other bots over and over? 2010-09-14T15:52:48 jmcarthur: With no change in ranking? 2010-09-14T15:52:49 You could check for that. 2010-09-14T15:52:49 danielvf: only when it's the seed bot 2010-09-14T15:53:01 danielvf: if another bot is the seed then it's likely to get more variety 2010-09-14T15:53:08 No rematches for 10 rounds? 2010-09-14T15:53:09 since the seed bot will tend to have more uncertainty anyway 2010-09-14T15:53:16 McLeopold: i disagree with such a restriction 2010-09-14T15:53:26 *** hornairs has joined #aichallenge 2010-09-14T15:54:03 Hmm. Bot #1 and #2 should maybe be played across every map? 2010-09-14T15:54:21 danielvf: the best matchup for bot A might be bot B, but that doesn't mean the best matchup for bot B is bot A 2010-09-14T15:54:29 No, it does. 2010-09-14T15:54:33 right? 2010-09-14T15:54:37 don't think so 2010-09-14T15:55:00 danielvf: if bot A has high uncertainty, why would bot B want to play bot A? 2010-09-14T15:55:16 A would certainly want to play B 2010-09-14T15:55:19 jmcarthur: You are right 2010-09-14T15:55:31 danielvf: and that gives me an idea about picking a seed 2010-09-14T15:55:35 jmcarthur: I forgot they have different pools of options 2010-09-14T15:55:38 danielvf: we should pick the bot that was least recently a seed 2010-09-14T15:55:46 danielvf: not the bot that least recently played 2010-09-14T15:56:09 for bot in submissions: go 2010-09-14T15:56:22 jmcarthur: and play a certain number of times? 2010-09-14T15:56:28 danielvf: nah 2010-09-14T15:56:37 jmcarthur: Just once? 2010-09-14T15:57:03 I think you should do %50 new bots and %50 last seed bots. 2010-09-14T15:57:16 McLeopold: new bots are always the least recently seeded ;) 2010-09-14T15:57:23 doh 2010-09-14T15:57:28 danielvf: just once 2010-09-14T15:57:31 danielvf: then pick a new seed 2010-09-14T15:57:37 danielvf: most fair to new submissions, i think 2010-09-14T15:57:45 danielvf: plus, we don't want to build up a huge queue 2010-09-14T15:58:00 That would make a new bot take a while to get settle, while settled bots get the same amount of time 2010-09-14T15:58:06 *settled 2010-09-14T15:58:14 danielvf: hmm, point 2010-09-14T15:58:34 we should at least bias it for least recently seeded, but maybe give some priority to new bots too 2010-09-14T15:58:36 Bot with lowest confidence? 2010-09-14T15:58:54 McLeopold: we want all bots to get a chance to seed though 2010-09-14T15:58:58 McLeopold: That's what I was originally thinking 2010-09-14T15:59:02 that won't work 2010-09-14T15:59:05 That would include new bots. 2010-09-14T15:59:10 bots that have been around for a while would never get to seed 2010-09-14T15:59:19 But they would still be played against 2010-09-14T15:59:28 yeah, but uncertain bots 2010-09-14T15:59:32 they would never get to progress 2010-09-14T15:59:36 &by uncertain bots 2010-09-14T15:59:38 **by 2010-09-14T15:59:41 okay 50% low conf, 50% last seed 2010-09-14T16:00:04 new bots are low conf and would get a few matches right away 2010-09-14T16:00:04 McLeopold: Something like that seems reasonable. 2010-09-14T16:00:05 i think confidence shouldn't play into it 2010-09-14T16:00:43 i think bots should be guaranteed some maximum amount of games the server run before choosing that bot for another game 2010-09-14T16:00:45 * danielvf really has to work - darn for being freelance 2010-09-14T16:01:17 i also don't think it's so bad for bots to take a while to rise in the ranks 2010-09-14T16:01:24 because they won't drop so fast upon resubmission anyway 2010-09-14T16:01:34 since we are just increasing uncertainty rather than starting over 2010-09-14T16:01:56 unless you new bot divides by zero... 2010-09-14T16:02:01 so i'm sticking to my "seed with the least recently seeded" suggestion 2010-09-14T16:02:20 McLeopold: we shouldn't increase uncertainty enough that that hurts too badly 2010-09-14T16:02:31 just enough to have some mobility 2010-09-14T16:02:46 You mean, you don't want new bots to get scheduled a few more times? 2010-09-14T16:02:55 oh, wait, now I see. 2010-09-14T16:03:02 it won't matter, is the point 2010-09-14T16:03:09 because they aren't starting over 2010-09-14T16:03:10 New bots are different from, improved bots. 2010-09-14T16:03:13 right 2010-09-14T16:03:28 absolutely new bots can take some time to rise in the ranks. i think that is perfectly fine 2010-09-14T16:03:42 probably more exciting anyway 2010-09-14T16:03:53 No way, more frustrating for the bot owner. 2010-09-14T16:03:58 * jmcarthur shrugs 2010-09-14T16:04:03 I agree that's frustrating 2010-09-14T16:04:03 you can rise pretty quickly per game anyway 2010-09-14T16:04:12 if your bot is any good at all 2010-09-14T16:04:21 and the higher your uncertainty the fast you rise 2010-09-14T16:04:24 *faster 2010-09-14T16:04:26 Especially for people that uploaded the crappy default bots and let it play a bunch of games =) 2010-09-14T16:04:32 damag: heh 2010-09-14T16:04:52 *** Zeiris has joined #aichallenge 2010-09-14T16:04:57 okay, we should probably increase uncertainty pretty drastically for resubmissions, but keep means where they were 2010-09-14T16:05:00 Maybe allow a user to clear their matches? 2010-09-14T16:05:06 When they choose to 2010-09-14T16:05:07 You should have a list of the first 5 moves the default bot takes on each map, then schedule those less. 2010-09-14T16:05:14 damag: no need with uncertainty increase 2010-09-14T16:05:22 McLeopold: lol 2010-09-14T16:05:25 Ah ok 2010-09-14T16:05:36 there is some merit to wanting new bots to have more matches 2010-09-14T16:05:40 maybe need to think about it some more 2010-09-14T16:05:59 Uncertainty could go all the way up with the same mean and you'd probably be okay. 2010-09-14T16:06:16 But that means a new submission on an average bot gets to play the best bot. 2010-09-14T16:06:25 *** Accoun has quit IRC () 2010-09-14T16:06:25 maybe "newness" could just be a factor multiplied by the time since it was last seeded 2010-09-14T16:06:42 McLeopold: no it doesn't 2010-09-14T16:06:49 McLeopold: it would play an average bot 2010-09-14T16:07:47 McLeopold: if the range is 0-50 and the bot's mean is 25 and uncertainty is huge, it will be matched with a bot that has a mean 25 and small uncertainty 2010-09-14T16:07:55 ok 2010-09-14T16:08:02 ignore the range i quoted. unimportant 2010-09-14T16:08:31 so a improved bad bot gets a chance to play an average bot on resubmission. 2010-09-14T16:08:34 *it will probably be matched wit 2010-09-14T16:08:36 h 2010-09-14T16:08:41 it's still probabilistic, we need to remember 2010-09-14T16:08:52 McLeopold: an improved bad bot will first play a bad bot 2010-09-14T16:09:08 McLeopold: but if it wins it has the mobility to move up after that since the uncertainty is big 2010-09-14T16:10:05 i suspect it would probably be best to just always reset uncertainty upon resubmission rather than just making it bigger 2010-09-14T16:10:11 that way it has maximal mobility 2010-09-14T16:10:18 bot has range(2-4) mean(3), after resubmission has range(0-50) still mean of (3), so would play a 3 2010-09-14T16:10:26 *** deepblue has quit IRC (Quit: Page closed) 2010-09-14T16:10:37 yeah that's the case if we reset uncertainty completley 2010-09-14T16:10:53 in fact no matter what we do it will prefer to play a 3 2010-09-14T16:11:00 (not always. again, this is probability) 2010-09-14T16:11:02 Then you can use uncertainty to help schedule resubmitted bots sooner. 2010-09-14T16:11:33 but that could starve bots that are around for a while. what if those developers are waiting for new matches for analysis? 2010-09-14T16:11:40 we need to be fair to all bots 2010-09-14T16:11:45 next ai challenge: find smartest ranking algo ;-) 2010-09-14T16:11:59 ranking/matching 2010-09-14T16:12:05 That's why I like the 50/50 split of uncertainty/last seeded 2010-09-14T16:12:39 (what about a "request matches" button) 2010-09-14T16:12:55 Yea, everyone wants to play deep blue. 2010-09-14T16:13:09 You should have a deepblue only server. 2010-09-14T16:13:10 dwchandler: actually that's a very real problem 2010-09-14T16:13:15 i think we should pick the bot with max(k*time_since_last_seed*some_factor_relating_to_newness_of_submission) 2010-09-14T16:13:17 Isn't deep blue playing on tcp 2010-09-14T16:13:37 actually the k is probably unnecessary 2010-09-14T16:13:43 yeah, won't affect the max 2010-09-14T16:14:22 and maybe not *completely* reset uncertainty on resubmission. (maybe we should just run some simulations) 2010-09-14T16:14:32 I honestly think its more a issue of having the hardware to simply play enough game per day. I mean I ran well over 1000 games in the last hour 2010-09-14T16:15:26 Some 16 core monster should be able to churn through games pretty quickly. 2010-09-14T16:15:57 here's a stupid idea: priority to winning bots to challenge up to higher ranked losing bots. 2010-09-14T16:16:23 so winning bots get picked always, and losing bots get matched 2010-09-14T16:16:35 dwchandler: that would starve losing bots though 2010-09-14T16:16:49 i think we should try to give all bots as even rate of matches as possible 2010-09-14T16:16:50 no, they get matched against 2010-09-14T16:16:55 people *will* complain otherwise 2010-09-14T16:17:10 also, if they only play top ranked bots, they will never get to tighten their confidence intervals 2010-09-14T16:17:17 nor will the top 2010-09-14T16:17:28 *** McLeopold has quit IRC (Quit: Leaving) 2010-09-14T16:17:41 it would just be a series of games with the same outcomes over and over 2010-09-14T16:18:46 for anywhere except extreme top/bottom bots would bubble to their level 2010-09-14T16:19:09 for top/bottom you could have some random matching 2010-09-14T16:20:26 i don't see the rationale behind this 2010-09-14T16:20:46 what's the status of decentralization? 2010-09-14T16:20:55 (just wondering, seems like an interesting idea) 2010-09-14T16:20:57 no idea 2010-09-14T16:23:06 jmcarthur: rationale is it's just a ladder, which is easy as pie to schedule fairly, and where bots play other bots close to their level 2010-09-14T16:25:05 *** retybok has quit IRC (Ping timeout: 240 seconds) 2010-09-14T16:25:07 *** Naktibalda has joined #aichallenge 2010-09-14T16:25:08 mobility should be high enough 2010-09-14T16:25:31 *** MorkBork has quit IRC (Read error: Operation timed out) 2010-09-14T16:26:00 *** MorkBork has joined #aichallenge 2010-09-14T16:26:08 dwchandler: but we are talking about actually arriving at *accurate* rankings from a limited number of games. it's a more complex problem than it appears, and the naive solution is very suboptimal 2010-09-14T16:26:50 jmcarthur: yeah, I guess the number of entries really throws a wrench into things 2010-09-14T16:27:36 *** Accoun has joined #aichallenge 2010-09-14T16:27:40 dwchandler: fortunately there's been a ton of research in this area already (e.g. online gaming, where trueskill comes from) 2010-09-14T16:28:34 * dwchandler nods 2010-09-14T16:29:16 *** HakanD has quit IRC (Ping timeout: 252 seconds) 2010-09-14T16:30:13 http://ai-contest.com/forum/viewtopic.php?f=17&t=593 2010-09-14T16:30:37 Basic Strategy guide @ http://ai-contest.com/forum/viewtopic.php?f=17&t=593 2010-09-14T16:30:45 *** Obeleh has quit IRC (Quit: Page closed) 2010-09-14T16:32:56 *** amstan has quit IRC (Ping timeout: 240 seconds) 2010-09-14T16:33:00 *** retybok has joined #aichallenge 2010-09-14T16:36:25 oh wow, I'm a moron. I was having such a hard time getting the java bots to work this morning so I translated them to C++... and I just figured out that java bots timed out when my (more complicated) C bots didn't because I set the round limit to 400 2010-09-14T16:36:40 I need anywhere from 500 to the default 1000 to accommodate them, d'oh 2010-09-14T16:37:32 whiteside: seems like a nice exercise though :P 2010-09-14T16:37:59 *** ermau|work has quit IRC (Quit: ermau|work) 2010-09-14T16:39:06 *** danielvf_ has joined #aichallenge 2010-09-14T16:39:06 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-14T16:39:06 *** danielvf_ is now known as danielvf 2010-09-14T16:40:16 *** Mathnerd314 has joined #aichallenge 2010-09-14T16:41:04 isn't Math.ceil supposed to always round up? 2010-09-14T16:41:44 did it rounded down something? 2010-09-14T16:41:57 it appears to be 2010-09-14T16:42:04 rogue780: yes. ceil(0.8) => 1.0, ceil(3.0) => 3.0, ceil(-2.5) => 2.0 2010-09-14T16:42:33 last example is incorrect :) 2010-09-14T16:42:36 rogue780: sorry, last part ceil(-2.5) => -2.0 2010-09-14T16:42:43 Naktibalda: lol, was fixing it :P 2010-09-14T16:42:51 I have Integer numAttackPlanets = (int)Math.floor(pw.MyPlanets().size() / 2);, and it is giving me a value of 0 when pw.MyPlanets().size() = 1 2010-09-14T16:43:13 you're calculating int / int 2010-09-14T16:43:15 floor != ceil 2010-09-14T16:43:22 err ceil rather 2010-09-14T16:43:22 rogue780: 1/2 is int division, you'll get zero 2010-09-14T16:43:23 te only possible result is an int.. 2010-09-14T16:43:27 I copied from the wrong method 2010-09-14T16:43:39 try pw.MyPlanets().size / (double)2 2010-09-14T16:43:43 sinan, ahah. that's it. I've run into it before 2010-09-14T16:45:36 ok. now my bot is crashing after four turns 2010-09-14T16:48:17 rogue780: add more logging to know what is crashing it 2010-09-14T16:48:36 sinan, I'm removing elements from an ArrayList while I'm iterating through it 2010-09-14T16:48:55 * rogue780 is looking up how to clone an ArrayList 2010-09-14T16:49:40 j3camero ping 2010-09-14T16:50:43 sinan, ok...this isn't working List _targetPlanets = (ArrayList)targetPlanets.toArray().clone(); how should I be going about this? 2010-09-14T16:51:28 rogue780: i think you're better of asking those questions in a langauge-specific channel :) I use Python 2010-09-14T16:51:46 ah. mybad. I thought you were using Java for a minute 2010-09-14T16:51:58 *** tobiassjosten has joined #aichallenge 2010-09-14T16:52:15 rogue780: no worries ;) the int/int issue is common in many languages, but i am not really that good at Java :) 2010-09-14T16:52:29 Has anyone gotten logging working using that GitHub hosted Python starter pack? 2010-09-14T16:54:37 yes 2010-09-14T16:54:58 I think if you download the latest PlayGame code it'll spit out stderr to the console too 2010-09-14T16:55:09 Guest77645: How do you use it? I'm getting: Exception in thread "Thread-2" java.lang.ArrayIndexOutOfBoundsException: 1 2010-09-14T16:55:13 that works for my logging needs now 2010-09-14T16:55:30 *** braddunbar has quit IRC (Ping timeout: 276 seconds) 2010-09-14T16:55:33 anoek: I'm raising exceptions for logging purposes now. It's very crude. :P 2010-09-14T16:56:03 *** Mathnerd314 has quit IRC (Quit: ChatZilla 0.9.86-rdmsoft [XULRunner 1.9.2.8/20100722155716]) 2010-09-14T16:56:07 Not crude if they are errors :) 2010-09-14T16:56:13 tobiassjosten: I'm the author of the alternate python kit. That message isn't from the python code 2010-09-14T16:56:19 I log stuff like decisions my bot is making and why 2010-09-14T16:56:41 ulope: Lots of thanks for sharing that code! 2010-09-14T16:57:22 tobiassjosten: np. I assume you are running PlayGame when you get that error? 2010-09-14T16:57:24 anoek: That's what I need to log as well. Raising an exception to see what data it's basing a single decision on isn't working otu for me. 2010-09-14T16:57:30 ulope: Exactly. 2010-09-14T16:57:52 can you paste the command? 2010-09-14T16:58:11 ulope: I might very well be using this all wrong. The way I'm running my bot is: java -jar ai-contest-read-only/planet_wars/starter_packages/tools/PlayGame.jar ai-contest-read-only/planet_wars/starter_packages/maps/map57.txt 1000 1000 log.txt "./plague" "java -jar example_bots/RandomBot.jar" | java -jar tools/ShowGame.jar 2010-09-14T16:58:25 *** retybok has quit IRC (Quit: leaving) 2010-09-14T16:58:35 ulope: The 'planet' directory is a clone of your repo. 2010-09-14T16:58:54 did anyone by change make a version of PlayGame with relaxed timeout requirements? 2010-09-14T16:59:37 You specify the timeout on the PlayGame command-line. 2010-09-14T16:59:39 tobiassjosten: seems normal. is "plague" your bot? 2010-09-14T17:00:05 hornairs: The first "1000" means timeouts are 1000ms. The second "1000" means games are 1000 turns long. 2010-09-14T17:00:45 ulope: Sorry, that's a recent command I was testing. Wrapping the python in a bash script. My command is: java -jar ai-contest-read-only/planet_wars/starter_packages/tools/PlayGame.jar ai-contest-read-only/planet_wars/starter_packages/maps/map57.txt 1000 1000 log.txt "python planetwars/plaguebot.py" "java -jar example_bots/RandomBot.jar" | java -jar ai-contest-read-only/planet_wars/starter_packages/tools/Sho 2010-09-14T17:00:45 wGame.jar 2010-09-14T17:00:53 (Sorry for spam people) 2010-09-14T17:01:14 ulope: And the problem arises when I try "python --log plague.log planetwars/plague.py" 2010-09-14T17:01:51 tobiassjosten: Ah... the --log goes after the file name. Else it's an option for the python interpreter 2010-09-14T17:02:26 bss03 no freakin way thanks a bunch 2010-09-14T17:02:35 Ohhh! There we go! 2010-09-14T17:02:40 :) 2010-09-14T17:02:40 *** ajhager has joined #aichallenge 2010-09-14T17:02:41 ulope: Thanks a lot! 2010-09-14T17:03:53 *** Kastra has quit IRC (Ping timeout: 240 seconds) 2010-09-14T17:04:39 This will help a lot. My bot is behaving in the weirdest way. :P 2010-09-14T17:05:17 *** sinan has quit IRC (Ping timeout: 245 seconds) 2010-09-14T17:06:38 tobiassjosten: Logging always helps :) 2010-09-14T17:06:56 ulope: Oh, and another newbie question if you will. How can I (should I) use your code without putting my bot within the repo? 2010-09-14T17:08:46 Say I want to use your repo as a submodule, for instance. 2010-09-14T17:09:43 tobiassjosten: Either copy the planetwars dir into your pythons site-packages (not recommended) or (if you're on *nix / mac os) export PYTHONPATH=$PYTHONPATH:/path/to/the/dir/below/planetwars (still a bit ugly) or use virtualenv and put it into the virtualenv's site-packages (best option imho) 2010-09-14T17:10:15 virtualenv ftw 2010-09-14T17:10:17 Googling for virtualenv then. :) Thanks! 2010-09-14T17:10:38 *** Mathnerd314 has joined #aichallenge 2010-09-14T17:11:12 tobiassjosten: also be sure to google for virtualenvwrapper. makes dealing with virtualenvs a ton easier 2010-09-14T17:11:52 * ulope should probably add a setup.py... 2010-09-14T17:12:02 *** Tidan2 has joined #aichallenge 2010-09-14T17:12:07 *** eburnette has quit IRC (Quit: Page closed) 2010-09-14T17:12:16 *** Tidan2 has quit IRC (Client Quit) 2010-09-14T17:13:25 dhartmei, hi, the tcp.c needed by the server code is the same as the one you use to play the games? 2010-09-14T17:13:37 *** Vi0 has quit IRC (Ping timeout: 245 seconds) 2010-09-14T17:15:27 *** alexshendi has joined #aichallenge 2010-09-14T17:15:27 *** rafsoaken has joined #aichallenge 2010-09-14T17:16:53 how do you mean? the server contains other tcp code as well. but i play my bot against the server using the same tcp.c client, yes. 2010-09-14T17:17:22 dhartmei, I mean if I want to compile the server myself... it requires tcp.c but does not contain it 2010-09-14T17:17:42 how does the server require tcp.c? 2010-09-14T17:18:36 you tell me :) this is the server makefile http://codepad.org/p0rgGuBm 2010-09-14T17:18:38 you mean the Makefile, building tcp.c as well? yes, that's just because i use a single Makefile to build both 2010-09-14T17:18:46 ahh, I see 2010-09-14T17:18:47 *** sinan has joined #aichallenge 2010-09-14T17:18:49 it's not linked into the server binary :) 2010-09-14T17:19:13 so make server and make expire should be enought :) 2010-09-14T17:19:22 yes 2010-09-14T17:19:49 thanks a lot :) 2010-09-14T17:19:56 np :) 2010-09-14T17:23:48 *** jaspervdj has quit IRC (Quit: leaving) 2010-09-14T17:26:18 *** jmreardon has quit IRC (Quit: jmreardon) 2010-09-14T17:28:00 *** jared has joined #aichallenge 2010-09-14T17:30:35 *** sigh has joined #aichallenge 2010-09-14T17:31:18 *** hornairs has quit IRC (Quit: hornairs) 2010-09-14T17:37:07 *** dmj111 has joined #aichallenge 2010-09-14T17:41:57 *** HakanD has joined #aichallenge 2010-09-14T17:43:21 *** rafsoaken has quit IRC (Quit: Page closed) 2010-09-14T17:49:53 alright. I'm off to class. new version of my bot is up. I doubt it'll win anything though 2010-09-14T17:50:46 rogue780: good luck :) you never know! 2010-09-14T17:54:07 battles will be here 2010-09-14T17:54:08 http://www.benzedrine.cx/planetwars/getplayer?player=ErrBot_0.0.5 2010-09-14T17:56:32 *** mega1 has quit IRC (Ping timeout: 245 seconds) 2010-09-14T17:56:37 *** sigh has quit IRC (Remote host closed the connection) 2010-09-14T17:58:33 *** Any_Key has quit IRC (Ping timeout: 272 seconds) 2010-09-14T17:58:49 *** ermau has joined #aichallenge 2010-09-14T18:00:54 *** Any_Key has joined #aichallenge 2010-09-14T18:02:37 *** uair00 has joined #aichallenge 2010-09-14T18:03:16 *** Ported has joined #aichallenge 2010-09-14T18:03:37 *** Bobng has quit IRC (Ping timeout: 276 seconds) 2010-09-14T18:06:14 *** alexshendi has left #aichallenge ("Verlassend") 2010-09-14T18:07:55 hi, I'm unable to visualizer on linux 2010-09-14T18:08:09 ERROR: failed to start client: "java Exception in thread "Thread-1" java.lang.ArrayIndexOutOfBoundsException: 1 at ViewerPanel.run(ViewerPanel.java:85) at java.lang.Thread.run(Thread.java:619) 2010-09-14T18:08:33 how do you run it? 2010-09-14T18:08:49 java -jar tools/PlayGame.jar maps/map7.txt 1000 1000 log.txt "java -jar example_bots/RandomBot.jar" "java -jar example_bots/DualBot.jar" | java -jar tools/ShowGame.jar 2010-09-14T18:09:14 taken from http://ai-contest.com/using_the_tools.php 2010-09-14T18:09:43 http://github.com/DanielVF/Planet-Wars-Canvas-Visualizer 2010-09-14T18:10:06 the canvas visualizer is better anyway 2010-09-14T18:10:33 thanks, I'll give it a try 2010-09-14T18:11:32 *** mightybyte has quit IRC (Quit: leaving) 2010-09-14T18:11:36 *** tapwater has quit IRC (Ping timeout: 265 seconds) 2010-09-14T18:11:48 *** Zeiris has quit IRC (Ping timeout: 255 seconds) 2010-09-14T18:14:30 *** preetum has joined #aichallenge 2010-09-14T18:21:08 well i have empirical evidence of a rock-paper-scissorsing 2010-09-14T18:21:26 a1k0n-0.3 can beat a1k0n-0.2 about 65% of the time, but it is objectively worse 2010-09-14T18:22:15 *** smhanov_ has joined #aichallenge 2010-09-14T18:24:43 dhartmei: why do I get mv: cannot stat `MyBot.log': No such file or directory errors when using loop.sh? 2010-09-14T18:24:45 the evolutionary steps of a single bot are very likely to have that situation since the developer is more likely to overfit their bot to beat their previous incarnations of it 2010-09-14T18:25:01 *** smhanov has quit IRC (Ping timeout: 276 seconds) 2010-09-14T18:25:12 I have problems with the canvas visualizer 2010-09-14T18:25:24 java -jar tools/PlayGame.jar maps/map7.txt 1000 1000 log.txt 'java -jar example_bots/RandomBot.jar' 'java -jar example_bots/DualBot.jar' | python visualizer/visualize_localy.py 2010-09-14T18:25:36 seems to run smoothly 2010-09-14T18:26:15 but I get a white page in firefox that says "Loading" and the play button on the bottom doesn't do anything 2010-09-14T18:28:11 *** td123 has quit IRC (Quit: WeeChat 0.3.3) 2010-09-14T18:28:51 *** awwaiid has joined #aichallenge 2010-09-14T18:34:02 *** Mathnerd314 has quit IRC (Ping timeout: 240 seconds) 2010-09-14T18:34:16 *** preetum has quit IRC (Quit: Page closed) 2010-09-14T18:36:27 *** pdbpdb80 has joined #aichallenge 2010-09-14T18:36:31 I got the original visualizer to work 2010-09-14T18:36:41 I made an stupid mistake 2010-09-14T18:37:23 forgot that my shell doesn't treat double quotes specially 2010-09-14T18:37:30 sorry for the noise 2010-09-14T18:37:42 np 2010-09-14T18:37:57 *** phreeza has quit IRC (Quit: kthxbai) 2010-09-14T18:38:33 hi everybody 2010-09-14T18:38:45 *** Mathnerd314 has joined #aichallenge 2010-09-14T18:39:05 *** uair00 has quit IRC (Quit: Page closed) 2010-09-14T18:40:03 @later tell jaspervdj Couple of pull requests sent via GitHub. When is this starter kit going to be official? 2010-09-14T18:40:03 bss03: The operation succeeded. 2010-09-14T18:41:02 Is it possible to have class A have a member variable of an object of class B while class B has a member variable of an object of class A? 2010-09-14T18:41:22 totaly 2010-09-14T18:41:35 javagamer_Cpp: If there's a pointer in there, sure. 2010-09-14T18:41:41 How can you do that if one class has to be defined before the first? 2010-09-14T18:41:45 javagamer_Cpp: (C++, right?) 2010-09-14T18:41:52 Yeah, hence the name ;) 2010-09-14T18:41:53 if its cpp, expect trouble. 2010-09-14T18:42:02 but forward declarations are your friend 2010-09-14T18:42:18 *** mightybyte has joined #aichallenge 2010-09-14T18:42:20 caytchen, ooh, sounds useful, looking up now 2010-09-14T18:42:23 javagamer_Cpp: class B; class A { ... }; 2010-09-14T18:42:26 javagamer_Cpp: Otherwise, you'll have issues no matter how you order them. One will be an incomplete (or undefined) type when you need it to be complete. 2010-09-14T18:42:57 * jmcarthur shoots c++ in the face 2010-09-14T18:43:19 javagamer_Cpp: If you use a pointer somewhere, the type can still be incomplete and things be okay. 2010-09-14T18:43:20 * Naktibalda registered in a free shell server for the first time since 2001 or 2002 :) 2010-09-14T18:43:40 *** Palmik has quit IRC (Remote host closed the connection) 2010-09-14T18:44:11 bss03, but it's not possible to have two objects w/ member variables that can store each other? (I don't need this, pointers will work for what I'm doing, but I'm curious) 2010-09-14T18:44:15 javagamer_Cpp: You can use a forward-defintion like in jmcarthur's example "class B;" to define and incomplete type. 2010-09-14T18:44:38 *** Appleman1234 has joined #aichallenge 2010-09-14T18:44:41 javagamer_Cpp: c++ aside, that can not work, not even on a logical level ;o 2010-09-14T18:44:50 javagamer_Cpp: No, because then class A and class B would be of infinite size. 2010-09-14T18:44:57 javagamer_Cpp: is here any chance that you are talking about inheritance? 2010-09-14T18:45:19 javagamer_Cpp: sizeof(A) = sizeof(B) + /* sizes of A's other members */ + /* padding */ 2010-09-14T18:45:25 javagamer_Cpp: sizeof(B) = sizeof(A) + /* sizes of A's other members */ + /* padding */ 2010-09-14T18:45:36 bss03, I see 2010-09-14T18:47:06 It work in java because java uses references -- which are like pointers -- everywhere, so you don't need to know the size of your members to calculate the size of your class. 2010-09-14T18:48:35 Various high-level languages use references or boxing or other techniques to allow "class A { B stuff; }; class B { A stuff; };" to "work". 2010-09-14T18:50:53 *** td123 has joined #aichallenge 2010-09-14T18:53:37 *** syntaxglitch has joined #aichallenge 2010-09-14T18:54:20 http://ai-contest.com/forum/viewtopic.php?f=18&t=621 perl starter packages 2010-09-14T18:57:17 *** awwaiid has quit IRC (Ping timeout: 272 seconds) 2010-09-14T18:57:37 *** sigh has joined #aichallenge 2010-09-14T19:04:00 *** tapwater has joined #aichallenge 2010-09-14T19:07:19 *** Snowplan has left #aichallenge 2010-09-14T19:11:51 *** awwaiid has joined #aichallenge 2010-09-14T19:14:46 *** tobiassjosten has quit IRC (Quit: Leaving) 2010-09-14T19:17:42 *** pdbpdb80 has quit IRC (Quit: Page closed) 2010-09-14T19:25:44 awesome: http://ai-contest.com/visualizer.php?game_id=4469485 2010-09-14T19:27:32 wtf? 2010-09-14T19:27:45 stacks of 1-unit fleets 2010-09-14T19:28:05 haha 2010-09-14T19:28:10 man the tension is killing me 2010-09-14T19:28:15 trying to weed out crappy bots? 2010-09-14T19:28:15 lol 2010-09-14T19:28:19 ahhh, was going to test another game and see if it was a visualizer bug 2010-09-14T19:28:25 that stack of 1 unit fleets are awesome 2010-09-14T19:28:33 you look at it and are like 2010-09-14T19:28:34 wtf?? 2010-09-14T19:28:42 good argument for combining fleets on the server 2010-09-14T19:28:44 though we should probably combine them transparently 2010-09-14T19:28:49 yeah 2010-09-14T19:29:04 in the server would be best, but at least in teh visualiser, if nothing else 2010-09-14T19:29:04 Hm but whats the point? Confusing opponents who cant sum() ? 2010-09-14T19:29:05 i think this is done on tcp 2010-09-14T19:29:22 compressing visualizations better? 2010-09-14T19:29:25 *** kozlovsky has quit IRC (Quit: Page closed) 2010-09-14T19:29:26 i've been doing the same for some time until i realized it was pointless heh 2010-09-14T19:29:36 heh 2010-09-14T19:29:38 was initially just to stump some example users 2010-09-14T19:29:38 how about any bot that iterates over the fleets badly 2010-09-14T19:30:04 Are there any rules against shipping .so's and dlopen'ing them? 2010-09-14T19:30:06 i don't even represent fleets as objects anymore 2010-09-14T19:30:26 a1k0n: me neither 2010-09-14T19:30:34 ? 2010-09-14T19:30:47 what you do? sum up Incoming/Allied in the planet? 2010-09-14T19:30:48 they're just future state changes of a planet 2010-09-14T19:30:55 *** McLeopold has joined #aichallenge 2010-09-14T19:30:56 I store fleets on the destination planet, indexed by day of arrival 2010-09-14T19:30:59 caytchen: yeah pretty much 2010-09-14T19:31:08 nice idea 2010-09-14T19:31:11 not that any of this code is being used yet 2010-09-14T19:31:23 yes, split all flets to 1 ship flits is anywey good idea 2010-09-14T19:31:36 that's what I'm coding up right now. not in any active bot yet 2010-09-14T19:31:39 if server dont combinre it 2010-09-14T19:32:12 regardless, the visualiser should combine them 2010-09-14T19:32:28 I hate showing my code (afraid of peer review) http://pastebin.com/0teqDEvV 2010-09-14T19:33:00 I believe the tcp server does combine them. 2010-09-14T19:33:11 I'm not sure about the contest server. 2010-09-14T19:33:22 there's a patch to do so for the contest server 2010-09-14T19:33:24 well, we have evidence right there it doesn't 2010-09-14T19:33:36 a1k0n> make perl support 2010-09-14T19:34:41 2 diferent perl starter pachages avalable 2010-09-14T19:34:46 Accoun: i'll bug j3camero about it 2010-09-14T19:34:53 i assume the contest thingy just needs to recognize .pls 2010-09-14T19:35:27 any canadian do perl support 2010-09-14T19:35:33 ))) 2010-09-14T19:36:07 Cyndre: the way I is store fleets is effectivly (in psuedo-code): planets[destination].incoming_fleets[player][remaining] += num_ships 2010-09-14T19:37:44 *** delt0r_ has joined #aichallenge 2010-09-14T19:37:53 *** coventry has joined #aichallenge 2010-09-14T19:38:53 *** delt0r___ has quit IRC (Ping timeout: 240 seconds) 2010-09-14T19:38:55 What's the URL for pulling the game data off of benzedrine? (There's a game I can view in the visualizer which I would like to include in my notes.) 2010-09-14T19:39:45 you can extract it from the page source 2010-09-14T19:40:32 sigh: remaining is turns till it lands right? 2010-09-14T19:40:44 Cyndre: yup 2010-09-14T19:41:17 that way I can just iterate over the incoming_fleets array to determine the state of the planet at any time 2010-09-14T19:41:20 *** lawlit has joined #aichallenge 2010-09-14T19:41:32 Thanks, Guest77645. (It's taking me a while to verify that, because suddenly, I can't connect to benzedrine.) 2010-09-14T19:41:55 sigh: yeah, that's very smart 2010-09-14T19:42:10 great idea - mind if I steal it now? 2010-09-14T19:42:14 sigh: Do you track your total ship count separately? 2010-09-14T19:42:17 go ahead :) 2010-09-14T19:42:27 awsome, now at least you wont be mad when I borrow it.... :) 2010-09-14T19:42:39 ulope: total ship count for the planet, or the game? 2010-09-14T19:42:45 the game 2010-09-14T19:43:19 a1k0n: i have been wondering if this game will end up with mixed strategies. If i can get one strategy working, maybe I will consider mixing it. 2010-09-14T19:43:33 *** ath888 has joined #aichallenge 2010-09-14T19:43:39 atm, I have a function which calculates it by iterating over all planets, I don't store it anywhere 2010-09-14T19:43:51 mostly because I don't use the total ship count yet 2010-09-14T19:44:06 I just calculate it for logging 2010-09-14T19:44:19 sigh: But the timing of the ships is critical, isn't it? Just adding them together throws that information away. 2010-09-14T19:44:47 *** l4u has quit IRC (Ping timeout: 272 seconds) 2010-09-14T19:45:16 coventry, that's why he indexes by "remaining" 2010-09-14T19:45:19 coventry: the "remaining" index is the timing information 2010-09-14T19:45:40 Ah, gotcha. 2010-09-14T19:46:00 *** RainCT has quit IRC (Remote host closed the connection) 2010-09-14T19:46:10 Are other people having trouble connecting to benzedrine? DFEOJM and I are. 2010-09-14T19:46:42 tried the web site just now and no connection 2010-09-14T19:46:57 Website is having hiccups. 2010-09-14T19:47:01 *** Yoshi-TS4 has quit IRC (Quit: Minus One IRC) 2010-09-14T19:47:33 *** genericbob has joined #aichallenge 2010-09-14T19:47:47 *** Naktibalda has quit IRC (Quit: No Ping reply in 180 seconds.) 2010-09-14T19:47:54 no connection to the actual server, too 2010-09-14T19:48:11 *** Naktibalda has joined #aichallenge 2010-09-14T19:48:13 i guess someone tried those 1-ship fleets.. ;p 2010-09-14T19:49:24 *** amstan has joined #aichallenge 2010-09-14T19:49:24 *** ChanServ sets mode: +o amstan 2010-09-14T19:49:26 is benzedrine server dead? 2010-09-14T19:49:27 zero ship fleets are pretty funny 2010-09-14T19:49:43 Naktibalda: it seems so 2010-09-14T19:50:17 Naktibalda: are we the only ones using PHP? 2010-09-14T19:50:35 no idea 2010-09-14T19:50:59 I've also got no connection to benze. 2010-09-14T19:51:03 I'm using Haskell. 2010-09-14T19:51:15 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-14T19:51:16 PHP is definitely the wrong choice for this problem. 2010-09-14T19:51:17 I'd hate to port my bot over to something else at the moment. 2010-09-14T19:51:28 *** danielvf has joined #aichallenge 2010-09-14T19:51:32 lol.. 2010-09-14T19:51:43 Are you having problems with the 1sec timeout with PHP? 2010-09-14T19:51:54 not at all. 2010-09-14T19:51:59 no, I have problem with dead server 2010-09-14T19:52:09 *** madsy has joined #aichallenge 2010-09-14T19:52:25 Finally, my LP-solver works :-) 2010-09-14T19:52:25 well, only when I have my debuging output on and I'm running xdebug in order to profile the app with kcachegrind. 2010-09-14T19:52:49 How are you using LP? 2010-09-14T19:53:15 ath888: my bot plays 235 turn game in 4 seconds accorting time 2010-09-14T19:53:59 1000 turns in 11.3s 2010-09-14T19:54:32 Alright, either im writing very bad code, or you write beautiful PHP. :) Im guessing the first. 2010-09-14T19:54:59 have you a php bot? 2010-09-14T19:55:16 *** anoek has left #aichallenge ("Leaving") 2010-09-14T19:55:23 *** anoek has joined #aichallenge 2010-09-14T19:55:40 it's possible that you are a very good AI programmer, and I'm not, so my codebase is much smaller :) 2010-09-14T19:56:13 Mine is a bit slower, 1000 turns in 20.2s 2010-09-14T19:57:10 But I'm doing a lot of possibly unnecessary computations at the moment. 2010-09-14T19:58:17 greghaynes: look at that, there are two people in here using php 2010-09-14T19:58:35 Meanwhile, deepblue, who is actually placing on the board, says that his *C++ code* is using the full 1s time limit... Maybe you're considering the wrong metric. 2010-09-14T19:58:35 what's wrong with greghaynes? 2010-09-14T19:58:55 Naktibalda: he just find it unbelievable 2010-09-14T19:59:06 can you guys reach the tcp server? 2010-09-14T19:59:08 Naktibalda: i don't. i disagree with your decision to use it, but i don't find it unbelievable 2010-09-14T19:59:35 *** McLeopold has quit IRC (Quit: Leaving) 2010-09-14T19:59:55 If I hit 1 second limit, I consider rewriting it in another language 2010-09-14T20:00:22 does the tcp server work for you at the moment? 2010-09-14T20:00:28 jmcarthur: I think all programming languages are just tools. I considered writing the bot in prolog at one time ;) 2010-09-14T20:00:43 canihazbacon: nope 2010-09-14T20:01:02 damn it, i just wanted to test my new python client for it :-( 2010-09-14T20:01:04 genericbob: prolog would at least make sense from an academic standpoint ;) 2010-09-14T20:01:18 I had to use prolog for my AI subject at uni 2010-09-14T20:01:22 jmcarthur: oh my... 2010-09-14T20:01:43 for a real challenge, how about using brainfuck? ;-) 2010-09-14T20:01:53 lolcode! 2010-09-14T20:02:01 or shakespeare! 2010-09-14T20:02:02 "You can write brainfuck in any language." 2010-09-14T20:02:13 :) 2010-09-14T20:02:23 * greghaynes facepalms 2010-09-14T20:02:26 prolog is not like brainfuck 2010-09-14T20:02:50 we could start the ogaicsc (obfuscated google ai contest submission contest) 2010-09-14T20:02:59 prolog is not obfuscated 2010-09-14T20:03:01 hahaha 2010-09-14T20:03:32 jmacarthur: it's just that nobody understands it in the first place? kind of like perl? ;-) 2010-09-14T20:03:37 If I have to port my bot, I would prolly pick Python. But if I were to start over, Haskell would prolly be my choice, I didn't see a starter pack for it when I started. 2010-09-14T20:03:38 I imagine some paranoid people are already putting their code through an obfuscator prior to submission. 2010-09-14T20:03:40 prolog is funky... I just find it had to work out whats slow and fast. 2010-09-14T20:03:41 *** drg has joined #aichallenge 2010-09-14T20:03:45 canihazbacon: and yet people are using perl for this and nobody is amazed 2010-09-14T20:03:56 delt0r_: i'll agree there 2010-09-14T20:03:57 Im amazed 2010-09-14T20:04:00 nobody noticed 2010-09-14T20:04:07 I dont know whats more amazing, perl or php ai's 2010-09-14T20:04:17 delt0r_: it's fast when you want to search in the order that it wants to search, slow otherwise 2010-09-14T20:04:23 at least in my experience 2010-09-14T20:04:23 greghaynes: php, definitely 2010-09-14T20:04:42 coventry: Not related to my AI, but who knows. Maybe I'll find a problem I can apply it to. 2010-09-14T20:04:53 greghaynes: at least perl is *somewhat* more principled... 2010-09-14T20:05:02 madsy: Oh, OK. Sounds like a happy development in any case. 2010-09-14T20:05:19 I write Perl, and its quite readable. Perl can be obscure, but its really more about the programmer than the language. 2010-09-14T20:05:31 are there any tcp-servers apart from the benzedrine one? 2010-09-14T20:05:32 ath888: i agree. i still hate perl, but i agree 2010-09-14T20:05:40 coventry: Yeah, I'm kind of sucky at math, so I am satisfied with this working with an arbitrary number of dimensions/unknowns 2010-09-14T20:05:44 ath888: I think you protest too much. No one mentioned readability until you did. :-) 2010-09-14T20:06:16 coventry: canihazbacon has mentioned "nobody understands it" 2010-09-14T20:06:23 *** rogue780_ has joined #aichallenge 2010-09-14T20:06:26 but with a wink 2010-09-14T20:06:28 Anyone here familiar with how the python engine deals with multi-player matches? 2010-09-14T20:06:30 coventry: Well, readability and obscurity does have some connection. :) 2010-09-14T20:06:38 Oh, OK. 2010-09-14T20:06:52 bss03: i assume you mean >2 players? 2010-09-14T20:06:55 Specifically, when a player issues an invalid move, to his ships on planet revert to neutral, or are they removed? 2010-09-14T20:07:01 Gah! Errbot0.0.5 is an epic fail 2010-09-14T20:07:03 jmcarthur: Yeah. 2010-09-14T20:07:04 no idea 2010-09-14T20:07:17 bss03: You lose 2010-09-14T20:07:17 neutral would make the most sense to me 2010-09-14T20:07:20 instantly 2010-09-14T20:07:29 delt0r_: yeah, but what about the remaining 2-3 players? 2010-09-14T20:07:36 delt0r_: this is the case that there are >2 players 2010-09-14T20:08:00 bss03: why? 2010-09-14T20:08:04 *shrug* I guess it doesn't matter too much. 2010-09-14T20:08:08 well iirc they just sit there --as a netrual that earns growth 2010-09-14T20:08:21 earns growth? nice... 2010-09-14T20:08:23 Just writing on engine parts in Haskell, and I want to be able to run 3-4 player games. 2010-09-14T20:08:26 Thats from the code... 2010-09-14T20:08:26 iirc, the game engine will asplode. 2010-09-14T20:08:42 delt0r_: so what if they had a majority of planets and nobody can overtake them? who's gonna win that? 2010-09-14T20:08:59 My Old Bot vs. Dual vs. Rage vs. My New Bot, for example. 2010-09-14T20:09:08 jmcarthur: I have not idea... I think they should go over to netrual... 2010-09-14T20:09:14 earn nothing 2010-09-14T20:09:29 but I don't recall seeing code to do that 2010-09-14T20:09:36 mine doesn't 2010-09-14T20:09:49 buts its untested on more than 2 players 2010-09-14T20:09:53 I'll just implement it as 'error "Figure out what happens here."' in Haskell. 2010-09-14T20:10:03 you guys are talking about the improved python engine from the forums? 2010-09-14T20:10:29 *** eburnette has joined #aichallenge 2010-09-14T20:12:59 news? 2010-09-14T20:13:03 delt0r_: It seems to set all their planets' owner to 0 2010-09-14T20:13:38 bss03: when you timeout or issue an invalid order, you are "kicked" from the game, which deletes all your fleets and makes your planets neutral 2010-09-14T20:14:05 I concur with Zannick 2010-09-14T20:14:32 Zannick: Do your ships on planet stay and the planet is a populated Neutral, or are the ships deleted and the planet is a 0-pop Neutral? 2010-09-14T20:14:43 ships stay, only ownership changes 2010-09-14T20:15:42 Oddly enough the fleets become 0 ship neutral fleets with 1 turn remaining. 2010-09-14T20:16:30 yeah, i dislike that, but it works 2010-09-14T20:16:48 Is there an alternate tcp site than benzedrine? 2010-09-14T20:22:38 *** AJC has joined #aichallenge 2010-09-14T20:23:01 eburnette: i would like to know that, too 2010-09-14T20:23:44 Since I'm waiting for benzedrine.cx I'm going to pit my bot against the example bots again. 2010-09-14T20:23:59 will the tourney be structured with >2player games? :S 2010-09-14T20:24:03 *** Any_Key has quit IRC (Ping timeout: 272 seconds) 2010-09-14T20:24:05 Anybody have hard bots to download and play locally? 2010-09-14T20:24:43 num1: doubtful 2010-09-14T20:24:44 num1: no 2010-09-14T20:25:13 eburnette: use the tcp server 2010-09-14T20:25:17 eburnette: you can do it semi locally 2010-09-14T20:25:30 *** Snellface has joined #aichallenge 2010-09-14T20:25:33 it's down 2010-09-14T20:25:42 dualbot? 2010-09-14T20:26:22 eburnette: I would propose mine, but you beat them already. :( 2010-09-14T20:26:30 i should write some assembly for my bot 2010-09-14T20:26:35 or all of it 2010-09-14T20:26:45 how much fun would that be? :P 2010-09-14T20:26:46 what maps are hardest to beat with the example bots? dualbot, perspector, rage 2010-09-14T20:27:07 I've heard 26 and 30 are hard. 2010-09-14T20:27:17 Dual and Rage are the only good ones I think. 2010-09-14T20:27:24 I concur with 30 being hard. 2010-09-14T20:27:33 amstan: Did my email find you? 2010-09-14T20:27:40 I've also had issues with 27 and 34 2010-09-14T20:27:41 Prospector is pretty easy to beat because he only sends out one fleet at a time. 2010-09-14T20:28:06 eburnette: try running a tournament http://gist.github.com/574904 2010-09-14T20:28:43 *** ath888 has quit IRC (Quit: Page closed) 2010-09-14T20:29:12 any map where the starting planets are close together is hard 2010-09-14T20:29:18 http://ai-contest.com/forum/viewtopic.php?p=3533#p3533 2010-09-14T20:29:23 i haz an idea 2010-09-14T20:30:40 bss03: one of my bots had a very hard time against RandomBot because of the one fleet at one time business. My bot was too defensive and wouldn't do hardly anything because RandomBot had a planet close to mine that had sent small fleets, thus ensuring that if I ordered an attack, RandomBot could attack and take over my planet. 2010-09-14T20:34:31 *** Caffeinated has joined #aichallenge 2010-09-14T20:34:32 does anyone have additional maps? 2010-09-14T20:34:44 Mornings all 2010-09-14T20:34:50 *** AJC has left #aichallenge 2010-09-14T20:35:00 genericbob: you can just use the python map generator 2010-09-14T20:35:15 Quick q, since I'm slow and can't see anything on the site - is the zip meant to be the entire game, or just the bot file? 2010-09-14T20:35:20 ahh good point, I feel silly for not thinking of that. 2010-09-14T20:35:38 Caffeinated: everything what you need 2010-09-14T20:35:43 Caffeinated: just the files required for the bot. 2010-09-14T20:36:14 and an engine and visualizer you can run on your own machine 2010-09-14T20:36:26 Awesome =) thanks 2010-09-14T20:37:26 *** Naith has joined #aichallenge 2010-09-14T20:38:35 Caffeinated: if you can't build or run your bot without it, it needs to be in the zip 2010-09-14T20:39:04 that includes PlanetWars.ext, if you are using a starter 2010-09-14T20:39:47 *** Any_Key has joined #aichallenge 2010-09-14T20:39:51 *** camedee has joined #aichallenge 2010-09-14T20:40:16 Has anyone made an interface that allows a human to play against a bot? 2010-09-14T20:40:44 ntiko 2010-09-14T20:41:12 * jmcarthur realizes that looks like a nick 2010-09-14T20:41:16 Not That I Know Of 2010-09-14T20:41:25 Yeah, I got it. 2010-09-14T20:41:36 Any new developments on the haskell support front? 2010-09-14T20:41:36 the game of galcon itself? 2010-09-14T20:41:43 :P 2010-09-14T20:41:51 Zannick: But that's different rules 2010-09-14T20:41:55 mightybyte: still the same situation on the server, if that's what you mean 2010-09-14T20:42:00 Yeah 2010-09-14T20:42:12 And no support for binaries? 2010-09-14T20:42:15 not yet 2010-09-14T20:42:17 mightybyte: i don't think you would want to play a bot 2010-09-14T20:42:20 we need to make the sandbox better first 2010-09-14T20:42:24 Zannick: I do. 2010-09-14T20:42:24 these games are hundreds of turns long 2010-09-14T20:42:32 mightybyte: when i was asking about supporting binaries before, i didn't realize how bad the sandbox was 2010-09-14T20:42:46 jmcarthur: Ahh 2010-09-14T20:42:53 honestly, i *still* think we might as well support binaries just because we are already supporing arbitrary code 2010-09-14T20:42:55 maybe if there were a better human-readable interface 2010-09-14T20:43:02 but it is scarier than i at first thought, either way 2010-09-14T20:43:20 i would not want to support binaries 2010-09-14T20:43:40 the only difference is we can't look at the source retroactively 2010-09-14T20:43:47 If code is compiled in the forest and nobody is around to look at it, is it indistinguishable from a binary? 2010-09-14T20:43:49 no, there's also linked libraries 2010-09-14T20:44:06 i don't see the significance of that 2010-09-14T20:44:10 oh, but those can be included 2010-09-14T20:44:14 yeah 2010-09-14T20:44:24 static binaries and all that 2010-09-14T20:44:28 anyway g2g 2010-09-14T20:44:36 oh 2010-09-14T20:44:39 Ok, later. 2010-09-14T20:44:45 people would submit windows binaries and they wouldn't work 2010-09-14T20:45:25 by compiling it on the server, it's more likely that it'll run on the server 2010-09-14T20:45:30 i dunno 2010-09-14T20:45:34 Yeah 2010-09-14T20:45:38 also the security issues are D: 2010-09-14T20:45:50 but they always are 2010-09-14T20:45:56 The security issues are the same as with code. 2010-09-14T20:46:21 ...with the exception of having the possibility to run static code analysis tools. 2010-09-14T20:47:06 what about using something like freebsd jails? 2010-09-14T20:47:35 the sandbox is one of the (many) issues we want to work on 2010-09-14T20:47:38 Im still suggesting a client that connects to the main server that your bot runs in 2010-09-14T20:47:48 Cyndre: that's unfair 2010-09-14T20:47:50 Heh, you could go with Trusted Solaris. 2010-09-14T20:47:53 dstufft: how so? 2010-09-14T20:48:29 Cyndre: I, with my big dual quad core server, will have an advantage over many people who don't have as much processing power at their fingertips. 2010-09-14T20:48:49 Cyndre: A) people with better computers will get an advantage (some one with a shitty computer get's less cpu time per 1s), B) People with high latency get penalized (they have to stop their turn before someone else does) 2010-09-14T20:49:15 C) If you live in a portion of the world your bandwidth is metered, it's asking you to pay to host your bot to compete 2010-09-14T20:49:17 mightybyte: the client tracks your bot and kills it if it uses over x cpu time or x ram (create hard limits instead of 1 second of cpu time soft changing limits) 2010-09-14T20:49:30 did someone ping me? 2010-09-14T20:49:37 and what's stopping someone from just using a different client 2010-09-14T20:49:46 it's not like our protocol is super hard 2010-09-14T20:49:54 dstufft: nothing, but when it comes to finals your bot wont win and you will dq and the next person below you will replace you 2010-09-14T20:50:05 20:21 < greghaynes> amstan: Did my email find you? 2010-09-14T20:50:05 oh, the same idea again 2010-09-14T20:50:11 dstufft: you would only be confusing yourself 2010-09-14T20:50:14 amstan: ^ 2010-09-14T20:50:20 dstufft: well cheating yourself 2010-09-14T20:50:27 greghaynes: it did, but as usual, jeff's inactive 2010-09-14T20:50:37 amstan: i talked to him earlier 2010-09-14T20:50:50 briefly 2010-09-14T20:50:53 *** braddunbar has joined #aichallenge 2010-09-14T20:50:58 dstufft: well.. here's the thing.. 2010-09-14T20:51:06 dstufft: this could of been hosted at uwaterloo just fine 2010-09-14T20:51:18 dstufft: you could send your code to someone you trust to run your bot for you, or rent a dedicated server and play way more games 2010-09-14T20:51:21 it's just because of some dissagreements that jeff moved(his decision alone) 2010-09-14T20:51:32 er. 2010-09-14T20:51:37 dstufft: and the disagreements mostly happened because of crappy fragile code 2010-09-14T20:52:43 amstan: i try to stay out of politics my life has less drama that way. I was just saying i talked to him earlier, and he seemed receptive to greghaynes offer 2010-09-14T20:52:50 but didn't say yes/no 2010-09-14T20:53:24 i just don't like his miscommunication 2010-09-14T20:53:53 and his mislabeling of things, it's neither a csclub event not google 2010-09-14T20:54:15 how did it end up being related to google? 2010-09-14T20:54:29 Zannick: jeff worked at google 2010-09-14T20:54:51 and he ran one while at google? 2010-09-14T20:54:56 and then kept the name? 2010-09-14T20:55:14 no 2010-09-14T20:55:28 he just asked them if he can use the name for mutual advertisment 2010-09-14T20:55:41 ah 2010-09-14T20:55:52 that's how I found it, it has googles name 2010-09-14T20:56:10 so apart from the name google has no part in it? 2010-09-14T20:56:10 because with all of the server issues, many people have suggested google lend machines 2010-09-14T20:56:23 and that they aren't and will not 2010-09-14T20:56:47 too bad app engine doesn't allow persistent connections, could run a bot from there... 2010-09-14T20:57:04 It's much more impressive to win a "Google AI challenge", than a "Random AI challenge" good job on getting the name. 2010-09-14T20:57:15 yes, i agree 2010-09-14T20:57:33 danielvf: you definitely got a point there... 2010-09-14T20:57:43 Random AI Challenge == a few dozen people 2010-09-14T20:57:46 ideally i would like jeff to make peace with the club so we can move the contest back there 2010-09-14T20:58:03 i just dislike how unfortunate the whole thing has turned out, and being powerless to help even through i work for the "sponsor" 2010-09-14T20:58:04 it can be just fine if it's waterloo ai challenge 2010-09-14T20:58:26 Zannick: so like... he worked for one term as an intern there 2010-09-14T20:58:35 anyway 2010-09-14T20:58:38 that's not the issue 2010-09-14T20:58:52 wait a minute... If we patched the tcp server to allow connections over xmpp, we could run bots on Google App Engine! 2010-09-14T20:58:52 no, it's a separate complaint of mine :) 2010-09-14T20:58:59 it would be very nice if we could just go back to uwaterloo, it's just much better there 2010-09-14T20:59:06 anyway, i must head out 2010-09-14T20:59:06 Zannick: get me a job imo 2010-09-14T20:59:08 Pfft.. So after all this I won't get a job working with google's AI machine to take over different websites based on popularity? =P 2010-09-14T20:59:09 we have a server network, no bandwidth concerns 2010-09-14T20:59:42 anyway.. 2010-09-14T20:59:49 i have to talk to jeff about all this 2010-09-14T20:59:50 Are you talking about Jeff Cameron? He owns the ai-contest.com domain right? 2010-09-14T20:59:56 maybe i said things that i shouldn't of said 2010-09-14T21:00:19 *** hornairs has joined #aichallenge 2010-09-14T21:00:34 How was the tron contest handled, all at waterloo? 2010-09-14T21:01:13 so this isn't hosted at uwaterloo ? 2010-09-14T21:01:31 *** kozlovsky has joined #aichallenge 2010-09-14T21:01:45 eburnette: yes 2010-09-14T21:02:00 eburnette: it's still there: http://csclub.uwaterloo.ca/contest/ 2010-09-14T21:02:33 I checked for a while waiting for it to reopen 2010-09-14T21:02:49 *** deepblue_ has joined #aichallenge 2010-09-14T21:02:55 Can somebody rehost the tcp server at uwaterloo now? 2010-09-14T21:02:58 tbh b/w concerns arn't that big of a deal at this point, not if things are done intelligently., the biggest "issue" is that the more raw compute power that's had, the more bots can be run 2010-09-14T21:03:05 eburnette: i could 2010-09-14T21:03:11 eburnette: in an hour or so 2010-09-14T21:03:19 *** lawlit has quit IRC (Ping timeout: 272 seconds) 2010-09-14T21:03:40 dstufft: agreed 2010-09-14T21:03:42 that would be great 2010-09-14T21:04:12 More compute power just means more bots fighting - we aren't overloading the server anymore 2010-09-14T21:04:23 dstufft: I think Cyndre has a point there for the pre-contest activity. 2010-09-14T21:05:15 what point? I missed it apparently 2010-09-14T21:05:36 mightybyte: and it would allow you to run multiple bots without cheating others out of games (I assume thats why you can only have one) 2010-09-14T21:05:38 Change to self-hosted bots via an interface program 2010-09-14T21:05:38 or do you mean the tcp server? 2010-09-14T21:05:42 Yeah 2010-09-14T21:05:45 *** kozlovsky has quit IRC (Ping timeout: 252 seconds) 2010-09-14T21:05:48 i don't see a point to it tbh 2010-09-14T21:05:48 ...but not for the final contest. 2010-09-14T21:06:03 you can cheat it, and if you want to use that, there is already a tcp server set up 2010-09-14T21:06:12 what's the point in duplicating something that works 2010-09-14T21:06:25 *** Any_Key has quit IRC (Ping timeout: 276 seconds) 2010-09-14T21:06:28 Yeah, but since it's not official, there aren't as many people playing there. 2010-09-14T21:06:29 giving the ability to the other 400 that want to have more games 2010-09-14T21:06:50 and if its a client it could be a live visualizer 2010-09-14T21:06:52 the way it's setup nwo, you can use tcp server to do testing of various bot types, and then upload to the official server to get a "tested on the final gaming platform" 2010-09-14T21:06:58 Cyndre: Well, technically they all have the ability to play on the current tcp server. 2010-09-14T21:07:16 mightybyte: if they come to irc, or read the forums 2010-09-14T21:07:34 and scheduling is being working on to get people to have more games 2010-09-14T21:07:35 mihgtybyte: and use linux, and know how to compile it, and install gcc 2010-09-14T21:07:37 Yeah. 2010-09-14T21:07:53 see my point? 2010-09-14T21:08:03 I see both points. 2010-09-14T21:08:04 if the contest supported it we could have a very nice unofficial testing client 2010-09-14T21:08:08 *** Yoshi-TS4 has joined #aichallenge 2010-09-14T21:08:41 There probably also is a rather large number of "drop-outs" that uploaded the startet package and lost interest... 2010-09-14T21:08:46 and a few have already put significant work into this 2010-09-14T21:09:00 which a visualizer and realtime results would have kept as well 2010-09-14T21:09:13 Cyndre: And those who have put significant work into it will figure out ways to do adequate amounts of testing. 2010-09-14T21:10:01 mightybyte: no, I meant into a visual client for tcp server 2010-09-14T21:10:50 we really should run a hash against the starter packs 2010-09-14T21:11:01 and exclude them from competing 2010-09-14T21:11:02 imo 2010-09-14T21:11:05 dstufft: lol, I was just thinking about something similar. 2010-09-14T21:11:48 Or maybe comparing their games to the starter pack bots and labelling them accordingly. 2010-09-14T21:12:14 that is signicantly harder 2010-09-14T21:12:37 my bot is going to get smarter soon - my two biggest pit falls is planet selection which I am almost done fixing, and sending to many reinforcements to a planet which shouldnt be hard to fix 2010-09-14T21:12:38 hueristics on how a bot plays i wouldn't want to write it lol 2010-09-14T21:12:59 hashing (or a diff I suppose, but meh id just hash) 2010-09-14T21:13:23 *** jmreardon has joined #aichallenge 2010-09-14T21:13:29 good night everybody 2010-09-14T21:13:31 *** caytchen has quit IRC (Ping timeout: 265 seconds) 2010-09-14T21:14:00 canihazbacon: night 2010-09-14T21:14:04 wow i left off the rest of my thought 2010-09-14T21:14:10 *** canihazbacon has quit IRC (Quit: Leaving.) 2010-09-14T21:14:17 hashing is quick and easy, and get's rid of everyone who did nothing but upload a starter kit 2010-09-14T21:14:27 if you've changed the bot at all, well then imo your a contestest 2010-09-14T21:14:37 *** Queue29 has quit IRC (Read error: Connection reset by peer) 2010-09-14T21:14:42 Won't catch adding wtitespace... 2010-09-14T21:14:48 whitespace* 2010-09-14T21:15:04 that's ok, really 2010-09-14T21:15:07 at least they opened the zip, resaved and rezipped 2010-09-14T21:15:09 I think the idea is just to stop people spamming the default bots 2010-09-14T21:15:10 *** lawlit has joined #aichallenge 2010-09-14T21:15:11 yea but how many people are going to unzip, add some whitespace, and then rezip 2010-09-14T21:15:23 * Cyndre cheers dstufft on 2010-09-14T21:15:28 the quick start guide encourages people to upload the default bot 2010-09-14T21:15:31 If they're lazy enough to upload the defaults..... 2010-09-14T21:15:39 *** Olathe has joined #aichallenge 2010-09-14T21:15:43 so im sure there is atleast some people whose accounts are nothing but the default bots 2010-09-14T21:16:00 I mean we could add "official" accouts for each of the example bots 2010-09-14T21:16:04 Caffeinated: Uploading the default is an explict step in the tutorial 2010-09-14T21:16:10 and exlude them from the hash checks 2010-09-14T21:16:21 so not necessarily lazy 2010-09-14T21:16:22 if people want to compete against RageBot etc in the tournament 2010-09-14T21:16:24 dstufft: can you find accounts with single submition? 2010-09-14T21:16:32 Naktibalda: we can yes 2010-09-14T21:16:38 update the tutorial to have the user make a small change in the logic before they upload 2010-09-14T21:16:53 many of them should be starter packs 2010-09-14T21:17:15 other_donkey: that's already in the tutorial 2010-09-14T21:17:18 Naktibalda: probably, but some of them arnt so doing it like that would suck 2010-09-14T21:17:20 probably should be worth to check manually 2010-09-14T21:17:35 could easily add it in to compile_anything.py 2010-09-14T21:17:47 to update it's status to Default Bot 2010-09-14T21:17:47 make them fix an error in the starter pack 2010-09-14T21:18:04 i think the purpose of that step is to give you a sense of accomplishment 2010-09-14T21:18:06 right away 2010-09-14T21:18:11 *** coventry has left #aichallenge ("ERC Version 5.3 (IRC client for Emacs)") 2010-09-14T21:18:20 you feel like it's easy because you didnt have to do anything, so are more likely to crack it open 2010-09-14T21:18:28 and play around 2010-09-14T21:18:37 also it allows people to understand the upload process without having to worry about compilation/code issues 2010-09-14T21:19:03 you could even play an "unoffical" game 2010-09-14T21:19:08 ie a game that doesn't factor into the rankings 2010-09-14T21:19:11 once per default bot 2010-09-14T21:19:37 or just pretend 2010-09-14T21:19:46 and link them to a preplayed game with the names changed lol 2010-09-14T21:20:04 *** dhaval has joined #aichallenge 2010-09-14T21:20:56 I also think that games lost/won due to crashing or timing out shouldn't count lol 2010-09-14T21:21:23 Naktibalda: I wonder if php-cli is installed on the sandbox.... If it already is maybe we could create a MyBot.cpp where all it does is exec php MyBot.php.... 2010-09-14T21:21:52 maybe a strike system, you crash/time out X times (maybe x times in y amount of time), and your submission get's disqualified 2010-09-14T21:22:02 *** braddunbar has quit IRC (Ping timeout: 265 seconds) 2010-09-14T21:22:24 prevents broken bots from wasting time and what not 2010-09-14T21:23:28 *** bss03 is now known as bss03-AFK 2010-09-14T21:25:19 hrm. benzedrine still down, eh? :( 2010-09-14T21:26:14 *** Any_Key has joined #aichallenge 2010-09-14T21:28:07 alrighty. I'm back 2010-09-14T21:28:35 *** Mathnerd314 has quit IRC (Quit: ChatZilla 0.9.86-rdmsoft [XULRunner 1.9.2.8/20100722155716]) 2010-09-14T21:29:42 *** Appleman1234 has quit IRC (Ping timeout: 240 seconds) 2010-09-14T21:30:31 unrelated, Sons of Anarchy is on in an hour and a half! 2010-09-14T21:30:54 http://65.255.180.155/docmon/ss1.png then click the play button and then http://65.255.180.155/docmon/ss2.png 2010-09-14T21:31:52 wth is that 2010-09-14T21:32:07 why is there an edit box with code in it 2010-09-14T21:32:13 sweet huh? 2010-09-14T21:32:32 i don't understand 2010-09-14T21:32:32 just for easier editing and testing 2010-09-14T21:32:34 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-14T21:32:41 yea im completly confused 2010-09-14T21:32:47 what are you trying to tell us 2010-09-14T21:32:55 *** deepblue_ has quit IRC (Quit: Page closed) 2010-09-14T21:33:03 *** danielvf has joined #aichallenge 2010-09-14T21:33:09 i don't want the browser to be my ide 2010-09-14T21:33:16 if that's what you are getting at 2010-09-14T21:33:22 *** Naith has quit IRC (Quit: Naith) 2010-09-14T21:33:53 we have plenty of good ways to test locally 2010-09-14T21:33:59 yea 2010-09-14T21:34:06 that's a terrible idea tbh no offense lol 2010-09-14T21:34:23 allows people to actually get their feet wet very easily 2010-09-14T21:34:45 reading up a see that you complained that it isn't supported on windows and stuff like that, but really, these tools were made by people not directly involved, and others could port it and tweak it and whatever else all they want if they'd just get off their asses and do it 2010-09-14T21:34:48 for you guys yes - but your not the target 2010-09-14T21:34:49 *i see 2010-09-14T21:35:26 if I had the knowledge/ability I would jmcarthur 2010-09-14T21:35:47 or you could just run linux. it's easy. you don't even have to kill your windows install 2010-09-14T21:35:56 Im on the tcp server and such 2010-09-14T21:36:17 http://wubi-installer.org/ 2010-09-14T21:36:20 Im not the target for this stuff either 2010-09-14T21:36:34 i guess i just don't see what your complaint is really about 2010-09-14T21:36:48 none really, just more accessability 2010-09-14T21:36:55 Cyndre: somebody made a visual client for the tcp server already 2010-09-14T21:37:03 greghaynes did it, iirc 2010-09-14T21:37:07 Cyndre: Easy to build a javascript browser engine 2010-09-14T21:37:07 yea, greghaynes did, but is it lined on the site 2010-09-14T21:37:15 Cyndre: Tie into the visualizer 2010-09-14T21:37:27 Cyndre: its notthe worst idea ever, but we have limited amount of manhours 2010-09-14T21:37:29 er listed on the site 2010-09-14T21:37:50 Cyndre: it's unofficial. we can't have everything be officially supported. that would be way too much work 2010-09-14T21:37:53 Wow, I changed the weight for planet->num_ships for the planets' ratings to 0, and now my bot wins all the time (against example bots ATM). discarding the info of the number of ships defending is something I wouldn't have thought to do before.... 2010-09-14T21:37:53 *** melkor has joined #aichallenge 2010-09-14T21:38:07 dstufft: nah, I dont want time wasted on new stuff like that, I was just bored :) - but the winows client for tcp server with easy bot loading and running would be awsome included on site 2010-09-14T21:38:28 *** mightybyte has quit IRC (Ping timeout: 265 seconds) 2010-09-14T21:38:42 how easy is it really? i don't find writing code in a text box and submitting it just for testing to be convenient at all 2010-09-14T21:38:55 Cyndre: you can make an issue about providing better info to the TCP server 2010-09-14T21:39:34 *** mightybyte has joined #aichallenge 2010-09-14T21:41:09 jmcarthur: actually it allows you to tweak your bot from anywhere - change a setting quickly from work (nothing more then a neat feature, but it is alot simpler then download file, unzip, edit, zip, upload) 2010-09-14T21:41:32 Cyndre: ssh is a wonderful tool 2010-09-14T21:41:40 Cyndre: as is distributed version control 2010-09-14T21:41:56 genericbob: I guess than means you attack close planets? 2010-09-14T21:42:02 jmcarthur: not every one intrested in participating is that organized/advanced 2010-09-14T21:42:02 *** shawnshekari has joined #aichallenge 2010-09-14T21:42:12 Cyndre: and a nice makefile with an uploader makes your job really easy 2010-09-14T21:42:28 Cyndre: then they need to learn. this is a contest for developers, not grade schoolers 2010-09-14T21:42:37 i don't mean to sounds like an ass, i promise 2010-09-14T21:42:39 *sound 2010-09-14T21:43:05 *** Appleman1234 has joined #aichallenge 2010-09-14T21:43:10 jmcarthur: its a test of ai programming not intimate comp knowledge 2010-09-14T21:43:14 the whole point of this contest is to learn, and there are tons of people in the forums and irc channel that are a wealth of information 2010-09-14T21:43:25 programming has a lot of obstacles aside from just code 2010-09-14T21:43:29 danielvf: The planet ratings are now based on growth_rate and distance 2010-09-14T21:43:35 ai programming is no different 2010-09-14T21:43:50 genericbob: Yeah, that probably would stomp a default bot. 2010-09-14T21:43:58 *** rwa has joined #aichallenge 2010-09-14T21:44:00 Cyndre: eh, i'll just stop. i'm looking like an ass 2010-09-14T21:44:08 jmcarthur: Lol. :P 2010-09-14T21:44:13 lmao 2010-09-14T21:44:19 nah, just expressing your view 2010-09-14T21:44:46 and in the 3rd screen shot was a live link to this irc channel from the site so they can ask you noob programming questions 2010-09-14T21:45:27 Cyndre: i'm with you that noob friendliness is awesome. i just don't think it's something the organizers should worry about spending time on. you should feel free to write code and contribute patches if you would like to see it, of course 2010-09-14T21:45:39 is the tcp match server down? 2010-09-14T21:45:41 Cyndre: and if it's high quality enough i'm sure it would be merged in 2010-09-14T21:45:48 shawnshekari: yup :( 2010-09-14T21:45:55 that ones a little tougher with manipulating the users bot files when logged in 2010-09-14T21:45:55 ok, thanks 2010-09-14T21:46:01 not where I would like to dive into the code at 2010-09-14T21:46:11 What's it called when you want the third sigma of of a Gaussian curve from a specific mean? 2010-09-14T21:46:30 *** Any_Key has quit IRC (Ping timeout: 252 seconds) 2010-09-14T21:46:35 I think my bot just pissed itself 2010-09-14T21:46:41 The third moment? 2010-09-14T21:46:41 *** shawnshekari has quit IRC (Client Quit) 2010-09-14T21:46:47 Cyndre: i think tryhaskell.org is an awesome example of a web site for noobs. you get a live console, interactive tutorial, and are even prompted to join #haskell via a web interface after you have gone through the first couple steps 2010-09-14T21:47:00 are you serious? 2010-09-14T21:47:02 yeah 2010-09-14T21:47:14 do the people in #haskell appreciate that? 2010-09-14T21:47:21 oh yeah. #haskell loves noobs 2010-09-14T21:47:34 #haskell is overwhelmingly helpful at times 2010-09-14T21:47:41 #haskell appreciates all the help it can get, have you tried it lately. 2010-09-14T21:48:58 I think thats great :) 2010-09-14T21:49:28 probably one of the wierdest things they have seen all day 2010-09-14T21:49:32 *** num1_ has joined #aichallenge 2010-09-14T21:49:33 heh, what you just said in #haskell followed by leaving suddenly :P 2010-09-14T21:49:57 nah, we get the occasional troll that says even weirder things 2010-09-14T21:51:10 I hate trolls - I never understand them, people confuse me with them sometimes, and I am oblvious and normally last to catch on to being trolled 2010-09-14T21:51:29 we have a fun strategy for handling trolls in #haskell 2010-09-14T21:51:43 ? 2010-09-14T21:52:45 we wave our banhammer stick and say they need to be nice if they want to stay and, usually, because they want to stick around long enough to keep trolling as all trolls do, they quiet down a little and start learning haskell just so they can earn their right to stay. sometimes we have to talk them to death, but it works amazingly well 2010-09-14T21:52:48 *** num1 has quit IRC (Ping timeout: 245 seconds) 2010-09-14T21:52:49 So is there a haskell starter pack? I haven't seen one. 2010-09-14T21:52:57 genericbob: there are a few around 2010-09-14T21:53:26 education solves everything 2010-09-14T21:53:37 we do have to kickban sometimes, but it's a lower number proportional to trolls than is usual for an irc channel 2010-09-14T21:53:47 Has anybody run into a limit on the number of fleets? Like 100 or 150? 2010-09-14T21:53:55 jmcarthur: I meant for the AI Challenge. 2010-09-14T21:54:01 genericbob: yeah, me too 2010-09-14T21:54:01 eburnette: what do you mean? 2010-09-14T21:54:02 *** num1_ has quit IRC (Read error: Operation timed out) 2010-09-14T21:54:14 jmcarthur: what other ones are there besides http://github.com/jaspervdj/planet-wars-haskell ? 2010-09-14T21:54:14 I really need to get to work on this 2010-09-14T21:54:18 genericbob: jaspervdj has one on github. that's the oldest and most popular, afaik 2010-09-14T21:54:26 OK, Being overly impatient like I am, any suggestions as to why my upload isn't appearing even on my profile page to say its been submitted? Waited the 10 minutes it says, but no email or anything.. 2010-09-14T21:54:26 tapwater: i'm working on one, but it's incomplete 2010-09-14T21:54:52 ah, that's right 2010-09-14T21:54:56 genericbob: i think bss03-AFK has one... i could be wrong 2010-09-14T21:55:08 *** melkor has quit IRC (Quit: out) 2010-09-14T21:55:16 *** dhaval has quit IRC (Ping timeout: 252 seconds) 2010-09-14T21:55:18 genericbob: and i think mathnerd has also been working on one 2010-09-14T21:55:26 not in here atm it appears 2010-09-14T21:55:39 ahh cool, thanks. 2010-09-14T21:56:10 genericbob: ah mathnerd's is forked from jaspervdj's http://github.com/Mathnerd314/planet-wars-haskell 2010-09-14T21:56:30 genericbob: there are a few forks it seems... http://github.com/jaspervdj/planet-wars-haskell/network 2010-09-14T21:57:11 ah, here's another one: http://github.com/ajhager/hs-planet-wars 2010-09-14T21:57:42 ajhager: Yeah, I'm working on sample bots now. 2010-09-14T21:57:46 oops 2010-09-14T21:58:13 I mean I'm trying to figure out why my bot stopped and I was wondering if lots of fleets could be the culprit. Local run. 2010-09-14T21:58:16 But it should be usable now. 2010-09-14T21:58:29 I enjoy reading different haskell implementations of the same problem 2010-09-14T21:58:57 you can almost say haskell is too expressive haha 2010-09-14T21:59:40 i'm writing mine into the spec as literate haskell, which means i'm reimplementing a lot of the engine too 2010-09-14T21:59:55 kind of ugly at the moment 2010-09-14T22:00:05 eburnette: are you sure you're not sending more ships than you have? I was accidentally doing that. 2010-09-14T22:00:57 possible I guess, I'll check for that too 2010-09-14T22:01:06 If you run PlayGame.jar without piping it into ShowGame.jar, it should give you a reason your bot stopped in the log file you specify. 2010-09-14T22:01:30 *** rwa has quit IRC (Quit: rwa) 2010-09-14T22:01:30 @amstan it's been an hour, where's the server? :) 2010-09-14T22:01:31 eburnette: Error: "amstan" is not a valid command. 2010-09-14T22:02:11 is the tcp server down? 2010-09-14T22:02:53 will I be disqualified for hosting a sandbox? 2010-09-14T22:03:23 tcp server pings, but can't connect a bot or get the web site 2010-09-14T22:03:31 rogue780: yes, though currently we arnt asking for anyone to donate one 2010-09-14T22:04:14 dstufft, it wouldn't be an official one. I'm wanting to setup a tcp server since the current one seems to be bogged down a lot and has issues at times (like now) 2010-09-14T22:04:29 *** Apophis has quit IRC (Read error: Connection reset by peer) 2010-09-14T22:04:34 oh you can setup anything unoffical you want 2010-09-14T22:04:53 *** Apophis has joined #aichallenge 2010-09-14T22:04:57 rogue780: TCP server be fine 2010-09-14T22:05:11 perfect. I'm purchasing a linode node right now...how much ram should I need. would 768mb be enough to run about 20games at a time? 2010-09-14T22:05:21 *** Apophis is now known as Guest7327 2010-09-14T22:05:25 rogue780: umm.. for what? 2010-09-14T22:05:30 rogue780: linode? 2010-09-14T22:05:38 I don't know what the requirements are, but i can't imagine they are much, since the tcp server runs the bots locally 2010-09-14T22:05:42 and not on the end server 2010-09-14T22:05:56 for the tcp server you don't need something very strong 2010-09-14T22:06:06 though.. what's a linode? 2010-09-14T22:06:11 its a company 2010-09-14T22:06:12 amstan, im going to host a tcp server 2010-09-14T22:06:14 linode.com 2010-09-14T22:06:16 they offer vps slices 2010-09-14T22:06:23 purchasing and server with 768 mb of ram does not sound like a good idea 2010-09-14T22:06:23 they provide vps...yeah waht dstufft said 2010-09-14T22:06:23 decently cheap, i used to use them 2010-09-14T22:06:27 its a VPS 2010-09-14T22:06:29 aha 2010-09-14T22:06:44 you pay like (idk what the 768 ones cost, 40 bucks or so?) 2010-09-14T22:06:47 rogue780: i usually prefer to run my own servers, have them at home 2010-09-14T22:06:48 a month 2010-09-14T22:06:51 rogue780: neads hardly any ram 2010-09-14T22:06:59 example: hypertriangle 2010-09-14T22:07:00 $19.95/mo for 512mb ram, $29.99/mo for 768 2010-09-14T22:07:07 rogue780: you could probably get away with the cheapest linode 2010-09-14T22:07:11 ok. i'll go with the cheap one then 2010-09-14T22:07:14 rogue780: seems like a lot, what nets does it have? 2010-09-14T22:07:22 all tcp server really does is act like a go between 2010-09-14T22:07:24 *** num1_ has joined #aichallenge 2010-09-14T22:07:46 they have like 200 or 300gb a month last i lookd 2010-09-14T22:07:55 and i think i pegged my connection at 30mbps 2010-09-14T22:07:58 the cheap one is 200gb 2010-09-14T22:08:01 last time i was using them 2010-09-14T22:08:08 for transfer' 2010-09-14T22:08:20 I wonder if dreamhost vps would be better 2010-09-14T22:08:34 amstan: your home server is not in a DC 2010-09-14T22:08:41 and is going to be less stable because of it 2010-09-14T22:09:01 my home server is less stable because I have comcast 2010-09-14T22:09:20 also a lot of residental net specifically disallow hosting servers 2010-09-14T22:09:22 dstufft: dc? 2010-09-14T22:09:26 and you can have your account terminated 2010-09-14T22:09:28 amstan, datacenter 2010-09-14T22:09:28 dstufft: not mine 2010-09-14T22:09:28 data center 2010-09-14T22:09:41 data centers have multihomed networks 2010-09-14T22:09:51 it seems less of a headake 2010-09-14T22:10:08 running this tcp server can't take more power than running an urbanterror server did 2010-09-14T22:10:18 I could run this on my home server to if you guys want a server up and running 2010-09-14T22:10:19 heck, lemme host it here and see how well that works 2010-09-14T22:10:20 amstan: vps's are very little headache, you don't have to deal with the hardware issues at all 2010-09-14T22:10:31 same with dedicated 2010-09-14T22:10:40 the hardware isn't your problem 2010-09-14T22:10:42 dstufft: but you deal with tons of software issues, sometimes you can't even touch the insall 2010-09-14T22:10:52 is the server hosted on github? 2010-09-14T22:10:53 contestbot, rankings 2010-09-14T22:10:54 num1_: Top 10 players: deepblue(2895), Tracer(2882), pdbpdb80(2867), _Shurf(2850), damien(2849), Loks1982(2845), youwillfail(2803), a1k0n(2790), liehann(2757), sighbot(2756) 2010-09-14T22:11:02 i've never seen an unmanaged vps where you didn't have complete control 2010-09-14T22:11:08 linode you get complete control 2010-09-14T22:11:28 you can even install whole new OS's, (as long as it's linux and can use their kernel) 2010-09-14T22:11:33 RapidKVM you can install any OS 2010-09-14T22:12:05 managed is a completly different story 2010-09-14T22:12:18 they run the server for you, so of course they don't want you in there screwing around 2010-09-14T22:12:18 anyone know where I can get the tcpserver from? 2010-09-14T22:13:19 dstfft: thats how they keep it running 2010-09-14T22:14:02 i don't buy managed servers or vps's though 2010-09-14T22:14:06 since i can just do it myself 2010-09-14T22:16:37 *** Utkarsh has quit IRC () 2010-09-14T22:16:48 *** McLeopold has joined #aichallenge 2010-09-14T22:21:29 McLeopold, I know you have enjoyed destroying my bot as if it were a rose frozen with liquid nitrogen. Well let me tell you, my bot has just acquired thorns. 2010-09-14T22:21:59 a call out with a simile 2010-09-14T22:23:00 *** Naktibalda has quit IRC (Remote host closed the connection) 2010-09-14T22:23:13 *** Any_Key has joined #aichallenge 2010-09-14T22:28:59 Can someone explain the difference between wall clock time and CPU time? 2010-09-14T22:29:08 rogue780: Have you been working on that line all day? 2010-09-14T22:29:32 *** DirtyKeyboard has joined #aichallenge 2010-09-14T22:29:38 McLeopold: 1 second of wall clock doesn't necessarily = 1 second of cpu time 2010-09-14T22:29:58 if your app is only using 50% of the CPU it will take 2 wall clock seconds to hit 1 second of CPU time 2010-09-14T22:30:02 rogue780: I'm glad I have a fan. 2010-09-14T22:30:35 It sounds like someone is going to watch a clock and count to mississippi 1 for each match. 2010-09-14T22:30:42 lol 2010-09-14T22:30:53 *** rwa has joined #aichallenge 2010-09-14T22:31:09 no, they'll just check the system clock for when you start and make sure you give an order before 1 second has passed 2010-09-14T22:31:12 oh... how would my bot calculate that? 2010-09-14T22:31:18 same way 2010-09-14T22:31:42 McLeopold: wall clock is real time 2010-09-14T22:31:46 RobotCaleb: it also works the other way, if your bot is multi threaded (and there is more then one processor) it might take up 2sec of CPU time for every 1sec of wall time. 2010-09-14T22:31:48 like, 1 second in the real world 2010-09-14T22:31:56 cache the current time when you get the order 2010-09-14T22:32:05 there are issues either way 2010-09-14T22:32:17 genericbob: I thought I read no threading? 2010-09-14T22:32:26 if you use wall time, then if the server is under a lot of load, you might get less cpu time per second of wall time 2010-09-14T22:32:53 if you use cpu time, people can do while: sleep(999) and it punishes less effecient languages 2010-09-14T22:32:56 is TCP down? D: 2010-09-14T22:32:58 They basically chose wall time because they said they can't accurately measure cpu time for the process. 2010-09-14T22:33:11 DirtyKeyboard: The site hasn't been pulling up for me 2010-09-14T22:34:07 genericbob: Yeah. "Use of multiple processes or threads is prohibited." 2010-09-14T22:34:25 tcp has been down for me 2010-09-14T22:36:17 DirtyKeyboard, yes. 2010-09-14T22:36:20 *** genericbob has quit IRC (Ping timeout: 252 seconds) 2010-09-14T22:36:37 If someone could point me in the direction to the tcp server code and such, I'd be happy to host another 2010-09-14T22:36:51 It's on the tcp server. :) 2010-09-14T22:37:08 I have a copy from the 10th 2010-09-14T22:37:43 does dhartmei know? he's probably sleeping at this time of night 2010-09-14T22:38:08 rogue780: http://res0l.net/server.tar.gz 2010-09-14T22:43:17 *** Queue29 has joined #aichallenge 2010-09-14T22:48:06 oh www.benzedrine.cx/planetwars ... where art thou? 2010-09-14T22:50:42 Is there any word on the max number of planets we will see in a map? 2010-09-14T22:51:00 967,345,723 2010-09-14T22:51:07 lol 2010-09-14T22:51:08 heh 2010-09-14T22:51:11 there goes 1 second 2010-09-14T22:51:39 That would wreak havoc with my approach. 2010-09-14T22:51:59 I think RandomBot would win that contest. 2010-09-14T22:52:41 amstan it's been 2 hours, where's the server? :) 2010-09-14T22:52:46 That would be interesting. Having too many planets to calculate well for all of them. You'd have to narrow your approach 2010-09-14T22:52:46 RobotCaleb, thanks. I'm installing a new copy of ubuntu on one of my extra boxes right now. It's running and AMD Athlon XP 2400+ and has a gig of ram 2010-09-14T22:53:36 rogue780: what is your bot name on the tcp server? 2010-09-14T22:53:44 ErrBot 2010-09-14T22:54:03 the latest version was a dismal failure 2010-09-14T22:54:06 Oh 2010-09-14T22:54:14 I have been beating you... 2010-09-14T22:54:19 Indeed you have 2010-09-14T22:54:39 the version I'm working on now should ::crosses fingers:: remedy that situation 2010-09-14T22:54:41 I wish I could see a match now... 2010-09-14T22:55:06 I should have my tcp server up in under an arn 2010-09-14T22:55:15 one whole arn? 2010-09-14T22:55:26 Hey, a farscape fan! 2010-09-14T22:55:57 crap. That was unintentional 2010-09-14T22:56:18 but yes. I love farscape 2010-09-14T22:56:35 definately in the top 5 sci-fi shows of all time, if not number one 2010-09-14T22:56:49 meh, babylon 5 ftw 2010-09-14T22:56:54 Doctor Who 2010-09-14T22:57:25 farscape was awesome 2010-09-14T22:57:28 not sure dr. who qualifies :) but I like it 2010-09-14T22:57:40 kind of cheesy to look back on, but i watched it religiously at one time 2010-09-14T22:58:02 new series is fantastic imho 2010-09-14T22:58:18 "Bad Wolf" 2010-09-14T22:58:25 lemme see. in my top five, Farscape, Dr. Who, Stargate (SG-1 and Atlantis), Andromeda, and (braces for backlash) Star Trek DS-9 2010-09-14T22:58:28 *** lawlit has quit IRC (Ping timeout: 255 seconds) 2010-09-14T22:58:43 actually 2010-09-14T22:58:44 DS9? Really? 2010-09-14T22:59:20 No BSG? Sure it got a little weird toward the end, but the rest was pretty awesome 2010-09-14T22:59:21 i'm more of a TNG trekkie, myself, but i never *really* got into it 2010-09-14T22:59:24 Actually, replace Andromeda with the new Battle Star Galactica 2010-09-14T22:59:26 i loved BSG 2010-09-14T22:59:35 okay, good 2010-09-14T22:59:47 BSG=frack 2010-09-14T22:59:57 I would put firefly in there, but being only one season makes it more of a mini-series to me than a tv show 2010-09-14T23:00:14 * jmcarthur doesn't know whether eburnette is saying good things or bad things 2010-09-14T23:00:15 "Yesterday's Enterprise" 2010-09-14T23:00:33 ^ all time best ST episode 2010-09-14T23:00:35 *** dhaval has joined #aichallenge 2010-09-14T23:00:50 but I'm definately sticking to my guns about DS9. I friggin loved that show 2010-09-14T23:01:08 i would put firefly right that top of my five scifi shows 2010-09-14T23:01:14 *right at the top 2010-09-14T23:01:20 jmcarthur, +1 2010-09-14T23:01:30 The stuff in the wormhole with the aliens was cool, but Bajoran politics was kind of a snooze 2010-09-14T23:02:57 *** PabloM has joined #aichallenge 2010-09-14T23:03:04 hi 2010-09-14T23:03:13 is the tcp server working? 2010-09-14T23:03:14 hello 2010-09-14T23:03:19 yep, works great 2010-09-14T23:03:33 could you tell me the IP? 2010-09-14T23:03:51 213.3.30.106 2010-09-14T23:03:53 tcpserver is down 2010-09-14T23:04:05 rogue780, you ruined it... 2010-09-14T23:04:15 oh. 2010-09-14T23:04:16 * rogue780 blushes 2010-09-14T23:04:35 PabloM, I'm working on setting up a new one right now 2010-09-14T23:04:39 *smile 2010-09-14T23:04:49 @smile 2010-09-14T23:04:50 eburnette: Error: "smile" is not a valid command. 2010-09-14T23:04:50 rogue780: cool 2010-09-14T23:04:56 @@ 2010-09-14T23:04:57 RobotCaleb: Error: "@" is not a valid command. 2010-09-14T23:05:00 @yay 2010-09-14T23:05:00 Double rainbow all the way! 2010-09-14T23:05:03 @@@@@@@ 2010-09-14T23:05:04 rogue780: Error: "@@@@@@" is not a valid command. 2010-09-14T23:05:05 one thing seems clear about this game 2010-09-14T23:05:06 showoff 2010-09-14T23:05:09 @help 2010-09-14T23:05:09 Queue29: (help [] []) -- This command gives a useful description of what does. is only necessary if the command is in more than one plugin. 2010-09-14T23:05:13 a good bot will slaughter the best human player 2010-09-14T23:05:18 contestbot: sigh is awesome 2010-09-14T23:05:18 i know... 2010-09-14T23:05:27 @get in the kitchen and bake me a pie 2010-09-14T23:05:27 rogue780: Error: The command "get" is available in the Dunno, Herald, Quote, QuoteGrabs, Success, and Topic plugins. Please specify the plugin whose command you wish to call by using its name as a command before "get". 2010-09-14T23:05:32 @ranking 2010-09-14T23:05:33 eburnette: Top 10 players: Tracer(2968), _Shurf(2899), Loks1982(2882), deepblue(2880), damien(2879), pdbpdb80(2877), youwillfail(2863), xtile1(2827), a1k0n(2790), liehann(2762) 2010-09-14T23:05:35 @"get in the kitchen and bake me a pie" 2010-09-14T23:05:36 rogue780: Error: "get in the kitchen and bake me a pie" is not a valid command. 2010-09-14T23:05:44 ah, contestbot, just like my wife 2010-09-14T23:05:45 ah, I've finally been kicked out of the top 10 2010-09-14T23:05:46 @ 2010-09-14T23:05:47 about time 2010-09-14T23:06:30 nice to see some new blood up there 2010-09-14T23:06:36 Anyone finding problems with uploads? My stuff compiles fine locally, just refuses to do ANYTHING on the site >__> 2010-09-14T23:06:48 good to see the top bots are finally playing each other a bit 2010-09-14T23:06:49 Okay, when I make the server I get a "getplayer.cpp:151: error: 'atol' was not declared in this scope" 2010-09-14T23:07:05 I'm on gentoo, any thoughts? 2010-09-14T23:07:05 Caffeinated, yea, it's been that way for me since the contest started 2010-09-14T23:07:13 did the new game selection stuff go in? I noticed that I started playing strong bots recently. 2010-09-14T23:07:13 McLeopold, i'm sorry 2010-09-14T23:07:54 Queue29: Any way around it? Or just keep trying? 2010-09-14T23:08:11 McLeopold: it's not in stdlib.h? 2010-09-14T23:08:12 Caffeinated, i've just been playing the TCP server only, but now that's down 2010-09-14T23:08:19 *** rogue780_ has quit IRC (Ping timeout: 240 seconds) 2010-09-14T23:08:27 so no, i don't really know what's wrong or how to fix it 2010-09-14T23:08:45 Eh, least it isn't just me then :) 2010-09-14T23:09:04 What do you all think is the hardest of the example bots? 2010-09-14T23:09:12 for a while i thought maybe they weren't letting starter packs play, but now i don't think that's true 2010-09-14T23:09:13 rage is said to be the toughest 2010-09-14T23:09:16 Rage.. 2010-09-14T23:09:17 RageBot 2010-09-14T23:09:17 *** hellman has joined #aichallenge 2010-09-14T23:09:20 *** dhaval has quit IRC (Ping timeout: 252 seconds) 2010-09-14T23:09:23 they're all pretty easy.. 2010-09-14T23:09:43 *** rxdh54xd5h has joined #aichallenge 2010-09-14T23:09:43 I finally found a mod which destroys RageBot... unfortunately it loses badly to the original bot 2010-09-14T23:09:58 ragebot rulez 2010-09-14T23:10:38 *** Any_Key has quit IRC (Ping timeout: 272 seconds) 2010-09-14T23:11:00 ragebot on map 30 is a puzzle to solve 2010-09-14T23:11:30 my bot does nothing on that map against ragebot 2010-09-14T23:11:33 ends up in a draw 2010-09-14T23:11:47 playing it to a draw is easy - beating it is not obvious 2010-09-14T23:11:54 yeah 2010-09-14T23:12:00 but it can be shellacked in like 20 some moves 2010-09-14T23:12:06 *** Yoshi-TS4 has quit IRC (Ping timeout: 240 seconds) 2010-09-14T23:12:19 *** rxdh54xd5h has quit IRC (Client Quit) 2010-09-14T23:12:22 dangit. I cut my inder fingernail too short and now I can't adequately expel this booger from my nose. 2010-09-14T23:12:30 ... 2010-09-14T23:12:41 lol 2010-09-14T23:12:48 use your nostril and wind power 2010-09-14T23:13:22 the trick is to get it open just enough so that it's not blocked but it's also not freely blowing around the booger. you need enough drag on the booger to accelerate it. 2010-09-14T23:13:24 What if it's a calm night? 2010-09-14T23:13:33 rwa, it just flapps there. as soon as I insert an instrument to assist in the extraction, it just lays back against the innerwall and mocks me. 2010-09-14T23:14:11 I just have to say, I love this channel 2010-09-14T23:14:28 * rogue780 thanks God for samurai sword-style letter openers. problem solved. 2010-09-14T23:14:34 ewwwwwww 2010-09-14T23:14:42 wow, the things I see when I switch windows 2010-09-14T23:15:00 use a scraper 2010-09-14T23:15:04 Ladies and gentlemen, the Aristocrats! 2010-09-14T23:15:38 while we're on the subject of boogers...anyone else from Maryland out there? 2010-09-14T23:15:49 rogue780: formerly 2010-09-14T23:16:23 rwa, congrats on getting out. The air force dumped me here and now that I'm out, I'm stuck until I finish my degree 2010-09-14T23:16:41 which part. I was in the Towson area. 2010-09-14T23:16:41 *** td123 has quit IRC (Read error: Operation timed out) 2010-09-14T23:16:51 I'm in...Dundalk... 2010-09-14T23:16:57 is anyone else running a tcp server currently? 2010-09-14T23:17:11 other_donkey, in about 30 - 60minutes I'll have one up 2010-09-14T23:17:12 rogu780: lol... sorry 2010-09-14T23:17:44 rogue780: Meade? 2010-09-14T23:17:45 rogue780: awesome! greghaynes might have one going soon too, dunno for sure 2010-09-14T23:17:46 hmm. I hope I didn't have anything important on that hard drive before I formatted it... 2010-09-14T23:18:02 Doesn't matter now. I wouldn't worry about it. :) 2010-09-14T23:18:03 rogue780: do people still play duck pins around there? 2010-09-14T23:18:30 RobotCaleb, I was stationed there, and my new job will put me there 2010-09-14T23:18:36 rwa, I'm not sure what that is? 2010-09-14T23:18:53 maybe that finally died - it's this weird form of bowling that was big in that area 20 years ago 2010-09-14T23:19:10 rwa, ah. I don't bowl...so i dunno. 2010-09-14T23:19:10 Contractor? I did six years in AZ. They wanted to send me to Alabama, which I wasn't too keen on, so I jumped ship for game development. 2010-09-14T23:19:37 you getting the accent yet? 2010-09-14T23:19:45 the weird 'o's? 2010-09-14T23:19:54 RobotCaleb, yeah. I'm getting a job with a contractor here. My degree I'm working on is CS with the game programming option 2010-09-14T23:20:06 although, I'm still going to Anne Arundel Community College. 2010-09-14T23:20:26 I'm taking Calc 1, Java programming 1 and Art 100 right now. 2010-09-14T23:20:33 That's fine. I skipped the school part. I'm still trying to decide if that's good or bad 2010-09-14T23:20:57 I have two other AA degrees, but not much transfers from a degree in Farsi or a degree in Communications Technology 2010-09-14T23:21:45 Did you learn Farsi at DLI? 2010-09-14T23:21:47 *** PabloM has quit IRC (Quit: y) 2010-09-14T23:22:43 RobotCaleb, yup 2010-09-14T23:22:52 47 weeks of fun in the sun 2010-09-14T23:22:54 The source was missing stdlib.h. I added it. How in the world was this compiled before? 2010-09-14T23:22:56 Nice. I would like to have gone. How was it? Pretty intense? 2010-09-14T23:23:32 Add it still has a tron comment. 2010-09-14T23:23:45 Any c coders? 2010-09-14T23:23:55 the school was just about the hardest thing I've done. 6 hours a day learning a foreign language from teachers who think they're God's gift to humanity. the area was nice and I met my wife there, so it's not all bad. 2010-09-14T23:24:02 McLeopold, I wish I were 2010-09-14T23:24:15 I'm an amateur C++ coder 2010-09-14T23:24:33 You might need a newer version of the code to get another server running. 2010-09-14T23:24:33 I didn't ever deploy :(. The only sand I saw was AZ and TX sand. I did jump out of a couple planes and flew an F-16. I guess it evens out. 2010-09-14T23:24:52 Or, I'm just completely lost. 2010-09-14T23:24:54 I always wanted to go to DLI, though. I fancy the idea of my job being to learn a new language 2010-09-14T23:25:12 night all, please post in forum when you get the servers up 2010-09-14T23:25:16 McLeopold: Does it work now? 2010-09-14T23:25:22 *** eburnette has quit IRC (Quit: Page closed) 2010-09-14T23:25:25 RobotCaleb, I never deployed even though I volunteered for Afghanistan 4 times. I never used my langauge and ended up spending the last 2 years doing PHP dev 2010-09-14T23:25:42 Wow. That's not surprising, but disappointing. 2010-09-14T23:25:52 DLI can't be cheap and they just wasted it 2010-09-14T23:25:56 Um..does what work? The server or the source code? 2010-09-14T23:26:06 the server. You fixed the compile, right? 2010-09-14T23:26:27 build a turtle fence! build a turtle fence! we need a rattlesnake fence! you shall build a turtle fence! 2010-09-14T23:26:59 I'm trying to setup my own. I got past 1 compile error, but now I have more. 2010-09-14T23:27:08 warning: format '%s' expects type 'char *', but argument 2 has type 'struct FILE *' 2010-09-14T23:28:03 RobotCaleb, were you army or air force? 2010-09-14T23:28:18 AF 2010-09-14T23:29:31 *** adi_ has joined #aichallenge 2010-09-14T23:29:53 Hey Guys, I'm trying to run PlayGame.jar and ShowGame.jar but it gives me an "incorrect version error" 2010-09-14T23:29:57 well butter my biscuits, I need to plug the network cable in before I can get on the network... facepalm 2010-09-14T23:30:07 adi_, get a newer version 2010-09-14T23:30:20 *** rogue780 is now known as unhelpful_advice 2010-09-14T23:30:31 I'm on Mac OSX 10.5 2010-09-14T23:30:31 rogue780: 1. Is it plugged in? 2010-09-14T23:30:35 and I just ran an update 2010-09-14T23:30:36 adi_, get a newer version 2010-09-14T23:30:38 and I now have java version "1.5.0_24" Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0_24-b02-357-9M3165) 2010-09-14T23:30:39 I think it's your java verson 2010-09-14T23:30:40 *** unhelpful_advice is now known as rogue780 2010-09-14T23:30:47 it didn't run for me on java 1.5 2010-09-14T23:30:57 open java preferences, drag Java SE 6 to top of both lists 2010-09-14T23:30:58 lemme see...i'm gonna turn it off and on again... 2010-09-14T23:31:47 Thanks drg, trying it out 2010-09-14T23:32:21 RobotCaleb, sweet 2010-09-14T23:34:29 alrighty...just finishing up ubuntu install, then I'll get to work on compiling the server 2010-09-14T23:35:20 drg, that works great! :) 2010-09-14T23:35:22 Thanks again 2010-09-14T23:35:27 No problem! 2010-09-14T23:36:24 *** rogue780|server has joined #aichallenge 2010-09-14T23:36:55 http://res0l.net/server.tar.gz 2010-09-14T23:37:44 rogue780: got maps? 2010-09-14T23:40:26 all I have is in that tarball right there 2010-09-14T23:40:46 and apparently ubuntu doesn't come with g++ pre installed 2010-09-14T23:41:41 *** patm has joined #aichallenge 2010-09-14T23:42:25 *** jmreardon has quit IRC (Quit: jmreardon) 2010-09-14T23:42:59 *** patm has quit IRC (Client Quit) 2010-09-14T23:43:03 *** genericbob has joined #aichallenge 2010-09-14T23:43:19 I'm assuming at this point I can use maps from the starter kit? 2010-09-14T23:43:49 I think that's a fair assumption 2010-09-14T23:44:07 I just created some 3-player and 4-player maps 2010-09-14T23:44:21 sounds busy 2010-09-14T23:44:24 *** jmreardon has joined #aichallenge 2010-09-14T23:45:10 is this thing supposed to serve the games and the web site as well? 2010-09-14T23:45:17 I'm so confused as to how this all works right now 2010-09-14T23:45:31 I haven't a clue. I never even opened it 2010-09-14T23:45:32 :) 2010-09-14T23:46:26 crap. I'm not sure if I'll be able to do this :\ 2010-09-14T23:46:37 did someone say greghaynes was working on it too? 2010-09-14T23:46:56 I'm also going to create some mirrored maps. (ie. if there is a planet at x,y there will be another planet at x,-y instead of -x,-y) 2010-09-14T23:47:49 genericbob: I was thinking of writing something that would take a 2-player map and turn it into an n-player map. 2010-09-14T23:47:59 danielvf, this server code is yours, right? 2010-09-14T23:48:48 rogue780: What's the issue? 2010-09-14T23:48:56 genericbob: Just map all objects with phase x in (0, pi] to (0, pi * 2 / n] and repeat them. 2010-09-14T23:49:10 genericbob: Then, I would instantly have 100 n-player maps. :) 2010-09-14T23:49:16 Dammit, Jim, I'm a PHP developer, not a C server compiler guy! 2010-09-14T23:49:40 bss03-AFK: That's what I do, I modified the map_generator.py to do that 2010-09-14T23:49:55 bss03-AFK, I believe your handle is a misnomer 2010-09-14T23:50:07 *** bss03-AFK is now known as bss03 2010-09-14T23:50:08 rogue780|server: greghaynes *might* be working on getting his server on 2010-09-14T23:50:19 rogue780: Oops, I always forget to take that off. 2010-09-14T23:50:30 rogue780: Looks like there's no database, just raw files it writes to and reads from 2010-09-14T23:50:49 I think I've got a tcp server up at ./tcp 140.99.51.114 9999 ... 2010-09-14T23:51:23 no web site, just the tcp server 2010-09-14T23:51:37 Is benxe not coming back? :( 2010-09-14T23:51:38 It would be cool to have a tcp server with 3 or 4 players possible. 2010-09-14T23:51:55 Is there an engine that does n players? 2010-09-14T23:51:58 *** deepblue has joined #aichallenge 2010-09-14T23:52:14 The java engine and python engine both work 2010-09-14T23:52:14 :( the tcp is still down 2010-09-14T23:52:21 *** jmreardon has quit IRC (Quit: jmreardon) 2010-09-14T23:52:24 * deepblue cries 2010-09-14T23:52:26 RobotCaleb: I believe the python one hadles it fine. 2010-09-14T23:52:31 cool 2010-09-14T23:52:39 deepblue: I just put one up at 140.99.51.114 2010-09-14T23:52:46 oh 2010-09-14T23:52:46 RobotCaleb: I'm also writing one in Haskell as part of my bot. 2010-09-14T23:52:48 dwchandler, dwchandler sweet. that's awesome 2010-09-14T23:52:48 anyone want to try and see if it's working? 2010-09-14T23:52:51 awesomesauce 2010-09-14T23:52:57 dwchandler, it works for me 2010-09-14T23:53:05 are you going to try and get the website part working too? 2010-09-14T23:53:07 cool! thanks, rogue780 2010-09-14T23:53:19 port 9999? 2010-09-14T23:53:21 I lost against your bot though...although it's my really crappy bot 2010-09-14T23:53:25 yes, 9999 2010-09-14T23:53:26 *** adi_ has quit IRC (Quit: Page closed) 2010-09-14T23:53:28 ITS OVER 9000!!!! 2010-09-14T23:53:32 ok will be there in a sex 2010-09-14T23:53:34 sex 2010-09-14T23:53:35 sec 2010-09-14T23:53:37 -_- 2010-09-14T23:53:44 hey now, let's not get too friendly 2010-09-14T23:53:49 they don't call you DEEPblue, for nothing 2010-09-14T23:54:13 Opponent disconnected 2010-09-14T23:54:14 You WIN against ErrBot_0.0.5 2010-09-14T23:54:19 hehe 2010-09-14T23:54:22 yeah, I just hit ctrl+c 2010-09-14T23:54:43 I will see if I can reverse engineer a web site without too much trouble 2010-09-14T23:54:54 ErrBot_0.0.5 is a failbot. 0.0.5b should be a marked improvement 2010-09-14T23:54:54 but it's getting late for me 2010-09-14T23:54:56 dwchandler: What's wrong with the site provided? 2010-09-14T23:54:57 dwchandler: how did you compile? does your getplayer.cpp have a reference to stdlib.h? 2010-09-14T23:55:03 I'm connected to the new tcp server. I've got a shell loop connecting 5 secs. after each game. 2010-09-14T23:55:06 dwchandler, I believe most of the front-end for it is on github 2010-09-14T23:55:21 So, If you need data, you can play my crappy bot, while I work on making it awesome again. 2010-09-14T23:55:51 McLeopold: there's a fair chance daniel wrote this on openbsd, which is what I compiled on 2010-09-14T23:56:03 dwchandler: There's site code in htdocs folder in the archive, right? 2010-09-14T23:56:10 so I just typed 'make' 2010-09-14T23:56:39 crap, I'm on gentoo and it doesn't compile 2010-09-14T23:56:55 " I'm on gentoo" <------ problem found 2010-09-14T23:57:16 rogue780: Ppft. 2010-09-14T23:57:19 har har, gentoo is suppose to compile EVERYTHING 2010-09-14T23:57:19 ;) 2010-09-14T23:57:48 ah, yes, htdocs looks like it has all the goodies :) 2010-09-14T23:57:54 I abandoned Gentoo years ago, as I was spending too much time maintaining it and not enough time using it. 2010-09-14T23:58:06 McLeopold, of course it is. To install it, it has to compile itself, after you write and assembler by manually flipping bits on the hard drive with a neodinium magnet 2010-09-14T23:58:16 But, It's still not always the clusterfsck that everyone claims it is. 2010-09-14T23:58:30 when I type make it complains about not having tcp.c 2010-09-14T23:58:33 i used gentoo for a long time 2010-09-14T23:58:34 dwchandler: works and is rather fast 2010-09-14T23:58:44 then I was forced to use macosx 2010-09-14T23:58:56 can someone link me to tcp.c?