2010-09-12T00:00:00 Thanks 2010-09-12T00:00:54 amstan: suhosin isn't right for everything, it disables or modifies some of the more dangerous parts of PHP 2010-09-12T00:01:10 and if your app uses one of them, it won't work obviously 2010-09-12T00:01:12 a1k0n: I'm a student at Ghent university, where they coined the name reservoir computing around 2005, doing my dmasters dissertation on it 2010-09-12T00:01:18 Snowplan: if you want to test your bot right now, http://www.benzedrine.cx/planetwars/ is highly recommended 2010-09-12T00:01:28 JensT1: ahha! well, this should be a good test of it then! 2010-09-12T00:01:47 JensT1: planning to write a paper about applying it to this problem? 2010-09-12T00:01:49 *** shawnshekari has joined #aichallenge 2010-09-12T00:02:11 a1k0n: I'm just starting on it this month, I've already got a different problem :) 2010-09-12T00:02:20 hello all 2010-09-12T00:02:31 JensT1: well a blog post about it or something would be awesome later on 2010-09-12T00:02:33 Thanks, I appreciate it I need to test. 2010-09-12T00:02:35 @whereis tcpserver 2010-09-12T00:02:35 greghaynes: tcpserver = http://www.benzedrine.cx/planetwars/ 2010-09-12T00:02:39 don't know if I'll have enought time to really do it 2010-09-12T00:02:54 I'm leaving for barcelona in 45mins 2010-09-12T00:03:04 oh heh! conference? 2010-09-12T00:03:05 and will only be able to do anything ai related in 2weeks 2010-09-12T00:03:10 holidays with my GF 2010-09-12T00:03:18 :p 2010-09-12T00:03:23 that's more important :) 2010-09-12T00:03:26 even better 2010-09-12T00:03:29 Where are your priorities? 2010-09-12T00:03:33 convince your GF to participate too 2010-09-12T00:03:38 rofl 2010-09-12T00:03:52 hah. well ECML PKDD is in barcelona in two weeks 2010-09-12T00:03:57 coworker of mine is going 2010-09-12T00:03:57 she's a forensic archaeologists master student 2010-09-12T00:04:11 that won't work I guess ;) 2010-09-12T00:04:27 JensT1, you have a few months that is plenty of time 2010-09-12T00:04:44 JensT1, enjoy barcelona 2010-09-12T00:04:49 thx :) 2010-09-12T00:05:10 I use vacations to lay in the sun and plan bots 2010-09-12T00:05:19 * Appleman1234 goes back to coverletter writing and sorting clothes whilst thinking about this contest lol 2010-09-12T00:06:13 Appleman1234: i know the feeling 2010-09-12T00:06:16 I just thing i'll be hard to find something to predict 2010-09-12T00:06:49 oponnents moves, yeah, but then you still have to figure out a countermove 2010-09-12T00:07:01 amstan, once again I would like to thank you for an awesome contest so far 2010-09-12T00:07:03 countermoves are easy 2010-09-12T00:07:03 will be hard to do it all in 1sec 2010-09-12T00:07:07 predicting is hard 2010-09-12T00:07:16 Appleman1234: thank j3camero too 2010-09-12T00:07:39 I wonder if the AI challenge can be represented as an LP problem. 2010-09-12T00:07:49 friend of mine just made a bot that beats other bots in 80 moves 2010-09-12T00:08:05 get him on the tcp server 2010-09-12T00:08:05 j3camero, thank you very much for being awesome and making an awesome contest, regarding development, I may help out, but not sure if I want to participate or help out yet 2010-09-12T00:08:07 hard to get prediction going there 2010-09-12T00:08:11 my bot need cannon fooder 2010-09-12T00:08:29 *** l4u_ has joined #aichallenge 2010-09-12T00:08:39 *** acieroid` has joined #aichallenge 2010-09-12T00:08:47 Appleman1234: you can do both, as long as you're not hosting a node or have a shell account on the main server 2010-09-12T00:08:50 madsy, yes it can , although like all models lp isn't a smooth isomorphic mapping to the domain 2010-09-12T00:08:52 http://ai-contest.com/profile.php?user_id=5953 2010-09-12T00:09:00 okay, now that i've actually begun the IO for my starter, i think i made a mistake in the spec 2010-09-12T00:09:04 about player numbers 2010-09-12T00:09:15 amstan, well which ones I do, whether one or the other or both, I will see 2010-09-12T00:09:24 oh, no, nevermind 2010-09-12T00:09:30 my confusion was actually planet ids 2010-09-12T00:09:42 i was assuming planet 1 always belongs to player 1 2010-09-12T00:09:47 amstan, at the moment I am still happy / content about graduating from my undergraduate recently :) 2010-09-12T00:09:55 i did not write that in the spec though, so it's okay 2010-09-12T00:09:58 i spend an hour debugging yesterday until I noticed that the maps are mirrored -_- 2010-09-12T00:10:13 yeah, how about that 2010-09-12T00:10:14 thought I had a bug 2010-09-12T00:10:21 duplicating planets 2010-09-12T00:10:25 anyone using the fact that they are mirrored? 2010-09-12T00:10:33 *** JensT1 has quit IRC (Quit: Leaving.) 2010-09-12T00:10:35 probably shouldn't depend on that 2010-09-12T00:10:44 JensT1, not yet, but I could use it later 2010-09-12T00:10:58 Did tcpserver just reset? 2010-09-12T00:11:15 nvm 2010-09-12T00:12:02 McLeopold is really milking his one win against deepblue 2010-09-12T00:12:16 yeah 2010-09-12T00:12:19 yeah 2010-09-12T00:12:22 stupis elo :D 2010-09-12T00:12:42 what is the Elo initial assumption? 2010-09-12T00:13:03 as in the prior? 2010-09-12T00:13:15 yeah 2010-09-12T00:13:15 probably 0 with a gigantic stddev 2010-09-12T00:13:51 if all the information you have is one win against the #1 person then you have to assume they might be better 2010-09-12T00:13:53 i'm telling you guys... adjusted elo 2010-09-12T00:14:02 lower confidence bound 2010-09-12T00:14:04 it's delicious 2010-09-12T00:14:26 yeah except on the main site where you don't get any games played and your rank is totally inaccurate unless you let it sit there for a week 2010-09-12T00:14:54 jmcarthur: we will give it a shot soon, now that we have so many people in the contest. We shall see if it messes up the rankings. If not, I would probably give it a try. 2010-09-12T00:15:19 j3camero: http://tinyurl.com/distributedai 2010-09-12T00:15:33 tinyurl fail 2010-09-12T00:15:34 http://code.google.com/p/ai-contest/issues/detail?id=133 2010-09-12T00:15:40 j3camero: ai.dstufft.com 2010-09-12T00:15:43 nginx 2010-09-12T00:15:47 jmcarthur: s/distributed/decentralized 2010-09-12T00:15:55 is this ever going to work? http://ai-contest.com/organization_profile.php?org_id=80 2010-09-12T00:16:27 a1k0n: make a patch? 2010-09-12T00:16:41 it looks like data is missing from the db or something 2010-09-12T00:16:49 dstufft: how are you generating the leaderboard? Is it a static copy, or do you have access to the main db? 2010-09-12T00:17:02 a1k0n: nope, it's all there, check the sample db 2010-09-12T00:17:11 j3camero: sampledb 2010-09-12T00:17:18 *** vikhyat has joined #aichallenge 2010-09-12T00:17:30 j3camero: hey is it ok that i committed a patch from someone else to trunk after reviewing and testing it? 2010-09-12T00:17:35 j3camero: has everything except passwords, and 99.9% of the games 2010-09-12T00:17:41 i figured branching, reviewing, and then merging was a bit redundant 2010-09-12T00:17:44 a1k0n: yeah that's fine. 2010-09-12T00:17:47 j3camero: that copy is just a checkout of www and the sampledb 2010-09-12T00:17:55 *** Theobon has quit IRC (Ping timeout: 252 seconds) 2010-09-12T00:18:08 nginx + php + php-fpm + apc + suhosin 2010-09-12T00:18:14 amstan: okay cool. 2010-09-12T00:18:22 j3camero: check the wiki, add to it 2010-09-12T00:18:25 apc is an opcode cache, prevents you from having to read the php files from disk every pageload 2010-09-12T00:18:55 dstufft: this seems pretty slick. 2010-09-12T00:19:21 j3camero: i can set it up if you want, on port 81 or whatever to test, or whatever you all want to do 2010-09-12T00:19:22 dstufft: how much smaller would you say the memory footprint is? 2010-09-12T00:19:38 http://blog.webfaction.com/a-little-holiday-present 2010-09-12T00:19:40 graphs 2010-09-12T00:19:45 second one in particular 2010-09-12T00:20:07 nginx doesn't really increase in footprint 2010-09-12T00:20:11 it stays constant 2010-09-12T00:20:18 well, but the php runners will to some extent 2010-09-12T00:20:21 yea 2010-09-12T00:20:22 they will 2010-09-12T00:20:27 php is it's own process with this 2010-09-12T00:20:38 but caching is your friend 2010-09-12T00:20:51 but with apache and mod_php (assuming?) every apache process has a copy of the php interpreter 2010-09-12T00:21:13 though it should be in shared memory so that's not as much of an issue, except for the stuff that's copied on write 2010-09-12T00:21:30 but yes, apache uses a shitton of ram regardless 2010-09-12T00:21:38 i also have static files being served directly from nginx etc too 2010-09-12T00:21:41 dstufft: oh.. how does phpbb handle that setup? 2010-09-12T00:21:58 amstan: i don't have a phpbb install setup atm, but ive used it before with this 2010-09-12T00:22:00 Where to change my password on ai-contest.com? 2010-09-12T00:22:10 I've run phpbb before behind nginx with no problem 2010-09-12T00:22:14 RobotCaleb: is it that important? 2010-09-12T00:22:15 drewrobb is giving me some losses 2010-09-12T00:22:50 yay 2010-09-12T00:23:00 you were up something like 10-2 against me 2010-09-12T00:23:04 it is only fair 2010-09-12T00:23:04 if i recall php-fpm also logs pages that load slow 2010-09-12T00:23:07 amstan: Have youall figured out a plan for getting the contest going again? 2010-09-12T00:23:12 much like mysql slow query log 2010-09-12T00:23:14 hehe 2010-09-12T00:23:20 good games to analyse :) 2010-09-12T00:23:26 greghaynes: yep, nxgiwhatever and decentralization 2010-09-12T00:23:34 nginx :p 2010-09-12T00:23:37 engine - x 2010-09-12T00:23:49 greghaynes: http://code.google.com/p/ai-contest/wiki/Decentralization 2010-09-12T00:23:57 dstufft: i wonder how those results would change for dynamic pages? A lot of our problems are coming from serving people's profile pages, where even caching won't help much. 2010-09-12T00:24:27 j3camero: an opcode cache isn't a cache like that 2010-09-12T00:24:37 j3camero: database schema changing may also help 2010-09-12T00:24:37 dstufft: only one way to really find out I guess. Perhaps we could set it up on port 81 on the prod webserver? 2010-09-12T00:24:54 dstufft: write me a tutorial, i can set it up 2010-09-12T00:24:59 when you interpret php, it get's compiled down to opcodes 2010-09-12T00:25:12 an opcode cache caches those, sort of like .pyc files 2010-09-12T00:25:18 dstufft: yep 2010-09-12T00:26:02 dstufft: if I get you a shell account and an ordinary user account, would you be able to run a copy of the site off the live database as yourself without needing root? (for now) 2010-09-12T00:26:25 sorry, I just mean an ordinary non-root shell account. 2010-09-12T00:26:44 I guess you would have to use port 8080 or something. 2010-09-12T00:27:23 yea i can do that, a slong as someone is around to aptitude install anything i need 2010-09-12T00:27:32 I am going to turn on the TM and see what happens. amstan: be prepared! 2010-09-12T00:27:40 dstufft: yeah I can help out. 2010-09-12T00:27:44 j3camero: for what? there's not much i can do 2010-09-12T00:27:45 lol 2010-09-12T00:27:53 using the new code? 2010-09-12T00:27:53 amstan: just letting you know 2010-09-12T00:28:13 *** l4u has quit IRC (*.net *.split) 2010-09-12T00:28:13 *** Queue29 has quit IRC (*.net *.split) 2010-09-12T00:28:13 *** acieroid has quit IRC (*.net *.split) 2010-09-12T00:28:17 j3camero: you want a pubkey for me? 2010-09-12T00:28:30 or do you allow passworded logins 2010-09-12T00:29:11 dude i have an easy idea: when picking people at random, weight the probability they are picked by the width of the last confidence interval generated by bayeselo (or 1000 if none) 2010-09-12T00:29:12 a1k0n: ah yes sorry. I got sidetracked from the changes I was helping you with when the server started acting up. Sorry about that. Let me know what changes have been made, and what changes you need help making, and then we will restart the TM after making the changes. 2010-09-12T00:29:30 a1k0n: that is a slick idea. Damn. 2010-09-12T00:29:32 j3camero: no worries, i don't need to do anything to the TM 2010-09-12T00:29:44 j3camero: just fixing up the java code before new starter packs go out 2010-09-12T00:29:45 *** shawnshekari has quit IRC (Quit: Page closed) 2010-09-12T00:30:17 a1k0n: okay I will turn on the TM and we will see what happens. Then I will go back to testing the starter pack fix you did earlier. Unless somebody else already did that and they've already been merged with trunk? 2010-09-12T00:30:47 j3camero: i have tested and merged all java viz code with trunk 2010-09-12T00:30:53 no outstanding issues now 2010-09-12T00:30:53 Okay the rankings should start updating again within ten minutes. 2010-09-12T00:31:06 the python engine code i haven't touched 2010-09-12T00:31:14 a1k0n: awesome. What about the fixes to the Java engine that I said I would review earlier today? 2010-09-12T00:31:26 j3camero: i merged them; someone had independently reviewed 2010-09-12T00:31:48 a1k0n: excellent. Sorry again about that. I still actually have the review open and a terminal open where I am testing them. :-) 2010-09-12T00:32:03 hah, no problem. i figured you had more important things to worry about 2010-09-12T00:32:20 plus i had to go take my kid to the park so i figured i'd ask for forgiveness instead of permission 2010-09-12T00:32:22 a1k0n: what's next to fix? Do you think I should focus on putting up more starter packs, or is the python engine more important? 2010-09-12T00:32:34 a1k0n: you made the right call. 2010-09-12T00:32:50 j3camero: python engine has gotta get into spec 2010-09-12T00:33:08 a1k0n: okay cool. Is that messy half-merged branch still around? 2010-09-12T00:33:09 but people are hanging out there with starter packs as well, and the TM isn't running that many games yet, so.. i guess it's your call 2010-09-12T00:33:25 j3camero: of java stuff? no, it's all merged in, i can nuke the branch 2010-09-12T00:33:30 ohh wait 2010-09-12T00:33:36 yeah, the python code is still half-merged. 2010-09-12T00:33:42 j3camero: check pm's when you get a chance 2010-09-12T00:33:47 a1k0n: no I mean the branch that had java engine fixes and python engine fixes from one or two days ago. 2010-09-12T00:33:51 j3camero: issue64_69_72_73 python code, yeah 2010-09-12T00:34:00 that one needs reviewing. 2010-09-12T00:34:28 Dumb question, Im trying to connect my java bot through the tcp server. saying execv: ./MyBot: failed. Is there a command line? 2010-09-12T00:35:05 @rankings 2010-09-12T00:35:06 amstan: Top 10 players: dmj111(2829), albert(2810), seigen(2776), Kavish(2746), strank(2711), drewrobb(2702), danielvf(2699), sighbot(2681), cheese_bot(2665), vortex(2649) 2010-09-12T00:35:34 Snowplan: well, do you ahve a ./MyBot? make a script with #!/bin/sh and java -jar yourbot.jar, chmod 755 it, and run that with the tcp thing 2010-09-12T00:36:04 thanks 2010-09-12T00:37:00 *** sigh has joined #aichallenge 2010-09-12T00:38:20 http://www.benzedrine.cx/planetwars/ 2010-09-12T00:38:37 sorry Snowplan, there is on the first page in forums 2010-09-12T00:39:06 also there's a sample commandline in the comments of tcp.c 2010-09-12T00:39:06 http://ai-contest.com/forum/viewtopic.php?f=18&t=424 2nd last post 2010-09-12T00:39:15 j3camero: we're doing pretty well 2010-09-12T00:42:35 I just played a match! 2010-09-12T00:43:10 greghaynes: hence the pretty well, lol 2010-09-12T00:44:50 amstan: any idea what the memory impact of the TM is? 2010-09-12T00:45:12 Mem: 2026 1946 79 0 39 1691 2010-09-12T00:45:15 *** sigh_ has joined #aichallenge 2010-09-12T00:45:38 *** caytchen has quit IRC (Ping timeout: 240 seconds) 2010-09-12T00:45:43 was 2026 1766 259 0 9 1568 2010-09-12T00:46:08 so.. that's idk.. 100 mb more.. 2010-09-12T00:46:33 I see the free memory is a lot lower than it was before. It's almost like the TM is using more RAM, and Linux did not decrease the cache size to compensate. Does that sound right? 2010-09-12T00:46:59 j3camero: ideally you always have 0 memory available, and cache everything 2010-09-12T00:47:02 linux tries to use as much memory as is available for cache 2010-09-12T00:47:19 otherwise ram goes to waste.. 2010-09-12T00:47:30 it starts dropping cache as you get near the swappiness 2010-09-12T00:47:34 *** sigh has quit IRC (Ping timeout: 255 seconds) 2010-09-12T00:47:36 jesus what shell is this 2010-09-12T00:47:38 how do i get free to actually show used memory, without cache? 2010-09-12T00:47:40 no tab complete ~ 2010-09-12T00:47:41 dstufft: sh i think 2010-09-12T00:47:59 dstufft: huh? 2010-09-12T00:48:03 i think the visualizer has a bug 2010-09-12T00:48:04 http://ai-contest.com/visualizer.php?game_id=4441702# 2010-09-12T00:48:14 it stops in the middle when there's no fleet flying around 2010-09-12T00:48:22 dstufft: oh.. main contest thingy? chsh /bin/bash 2010-09-12T00:48:22 deepblue: it's a known bug 2010-09-12T00:48:30 dstufft: then relogin 2010-09-12T00:48:34 ok sorry, didn't say anyhing :) 2010-09-12T00:48:38 yea i was just looking at what shells you guys had installed 2010-09-12T00:48:47 deepblue: don't apologize. we want people reporting bugs! 2010-09-12T00:48:48 dstufft: you can use chsh to change it. Sorry :-) 2010-09-12T00:48:48 zsh ! 2010-09-12T00:48:56 deepblue: i was just letting you know 2010-09-12T00:49:03 woo, finally beat hsbot10.01 2010-09-12T00:49:13 ok, just don't want to bug anyone :) 2010-09-12T00:50:23 @greg: now bear dmj.4, a1k0n, dhartmei, drewrobb and me 2010-09-12T00:50:24 deepblue: Error: "greg:" is not a valid command. 2010-09-12T00:50:31 beat 2010-09-12T00:50:47 talk to my hand bot! 2010-09-12T00:50:56 I beat dhartmei, but I have no idea how 2010-09-12T00:51:00 *** zaphus has joined #aichallenge 2010-09-12T00:51:18 a1k0n: did anyone update the starter packages yet, since you merged the fixes to the Java engine? 2010-09-12T00:51:34 dwchandler: happens to me all the time ;) 2010-09-12T00:52:07 j3camero: have you seen this? http://code.google.com/p/ai-contest/wiki/Decentralization 2010-09-12T00:52:25 Let me give it a read. I have it open. 2010-09-12T00:52:44 j3camero: how did your calculations go? 2010-09-12T00:55:10 amstan: just about to start them. 2010-09-12T00:55:34 j3camero: see: Comeback: bandwidth for the server is 4TB/month, bandwidth for hypertriangle is 200GB/month, the nodes will have that issue much sooner. 2010-09-12T00:55:59 amstan: true. But I don't want to fuck hypertriangle either. 2010-09-12T00:56:13 j3camero: what's the average bot size? 2010-09-12T00:56:15 10mb? 2010-09-12T00:56:17 Do you know if we can use corn-syrup? 2010-09-12T00:56:19 *** ihodes has quit IRC (Quit: ihodes) 2010-09-12T00:56:28 amstan: I am about to find out. 2010-09-12T00:57:01 deepblue: Im working on it 2010-09-12T00:57:14 I did beat djm.4 though 2010-09-12T00:57:28 j3camero: http://www.wolframalpha.com/input/?i=10MB+/+(1sec+*+300)+*+1+month 2010-09-12T00:57:49 j3camero: 80GB/month/sandbox(1 bot at a time) 2010-09-12T00:58:17 just watched that replay :) 2010-09-12T00:59:01 http://www.benzedrine.cx/planetwars/canvas?game_id=1284264352|wtfbbq|greghaynes 2010-09-12T00:59:06 *** gw is now known as whiteside 2010-09-12T00:59:06 you need to be more aggressive 2010-09-12T00:59:11 :P 2010-09-12T00:59:25 *** whiteside is now known as gw 2010-09-12T00:59:39 In the output of the du command, are those sizes in KB, bytes, or what? 2010-09-12T00:59:40 hah, I thought I could just scare him into suicide 2010-09-12T00:59:57 tehe 2010-09-12T01:00:01 j3camero: du -h 2010-09-12T01:00:05 j3camero: human readable 2010-09-12T01:00:18 that's not going to be useful anyway 2010-09-12T01:00:18 j3camero: -h also works for ls 2010-09-12T01:00:24 you have to look at the compressed sizes 2010-09-12T01:00:34 jmcarthur: we have them compressed too i think 2010-09-12T01:00:48 chunked up the way they would be sending to sandbox nodes? 2010-09-12T01:01:05 presumably it wouldn't just be sending them individually 2010-09-12T01:01:14 i guess not a big difference there 2010-09-12T01:01:18 n/m me 2010-09-12T01:01:19 As is, I believe the engine sends them to the compute node uncompressed. Unless scp silently compresses things. 2010-09-12T01:01:24 Are the 'nodes' low or high bandwith / low or high cpu? 2010-09-12T01:01:38 scp could silently compress in theory if you configure it to 2010-09-12T01:01:49 scp does by default IIRc 2010-09-12T01:02:06 so then you still have to do the test for compressed 2010-09-12T01:02:24 j3camero: sent you some apt-get's 2010-09-12T01:03:20 j3camero: 0.4MB 2010-09-12T01:03:24 http://www.wolframalpha.com/input/?i=1.3GB/2788 2010-09-12T01:03:36 1.3G . 2010-09-12T01:03:37 root@ltfs1110:/home/contest/ai-contest/planet_wars/submissions# ls|wc -l 2010-09-12T01:03:38 2788 2010-09-12T01:03:40 uncompressed 2010-09-12T01:03:42 amstan: Did you see my queston? 2010-09-12T01:03:58 I ask because I could offer boxes of one or the other, just not both 2010-09-12T01:03:59 greghaynes: low bandwidth, high cpu 2010-09-12T01:04:16 I have a home box that would probably work well for that 2010-09-12T01:04:32 greghaynes: you'll need 3.5GB/month/sandbox(1 bot at a time) 2010-09-12T01:04:47 and disqualification from the contest... 2010-09-12T01:04:50 greghaynes: if you host a node you cannot compete 2010-09-12T01:04:54 ah 2010-09-12T01:05:01 McLeopold extends his lead! 2010-09-12T01:05:07 Thats a tough decision 2010-09-12T01:05:09 sad that volunteers get stuck in such a situation 2010-09-12T01:05:21 jmcarthur: i think we'll have enough volunteers 2010-09-12T01:05:31 probably, but it still saddens me 2010-09-12T01:05:35 deepblue: You just beat me 2010-09-12T01:05:35 we only need like 5 servers 2010-09-12T01:05:40 woho 2010-09-12T01:05:43 the volunteers are the ones who deserve the chance to compete the most 2010-09-12T01:05:43 amstan: 3.5GB isn't bad - only about 110 mb per day and i assume that's spread out reasonably 2010-09-12T01:05:49 i want to beat leopold that would make me happy 2010-09-12T01:06:06 computerwiz_222: yep, it's spread out 0.4MB/match 2010-09-12T01:06:13 wow, that's little 2010-09-12T01:06:15 computerwiz_222: each match is 300 seconds long 2010-09-12T01:06:18 amstan: I think we could be safe if we assume 1 MB per bot per game. 2010-09-12T01:06:20 amstan: i have a c2d with 4gb of ram sitting under my desk atm 2010-09-12T01:06:26 for my fileserver 2010-09-12T01:06:37 1MB/bot > 8.7GB 2010-09-12T01:06:42 it's running solaris though 2010-09-12T01:06:48 slowlaris, eh? 2010-09-12T01:06:49 dstufft: we don't need ram, we need cpu 2010-09-12T01:06:50 well opensolaris 2010-09-12T01:06:53 c2d 2010-09-12T01:06:55 core 2 duo 2010-09-12T01:07:05 *** Zeiris has quit IRC (Excess Flood) 2010-09-12T01:07:08 *** rwa has joined #aichallenge 2010-09-12T01:07:12 greghaynes: yea, ZFS is ++ 2010-09-12T01:07:13 dstufft: we can probably use it, you're forfeited, right? 2010-09-12T01:07:21 *** Zeiris has joined #aichallenge 2010-09-12T01:07:26 dstufft: Thats on fbsd now also 2010-09-12T01:07:27 i have a login to the server so im pretty sure yes 2010-09-12T01:07:39 greghaynes: in a limited way, so i hear 2010-09-12T01:07:53 Supposedly its going to be in kernel linux very soon 2010-09-12T01:07:55 i've been wanting to try dfly 2010-09-12T01:07:57 greghaynes: last time i seriously looked at it there was still a bunch of kernel panics all the time and shit 2010-09-12T01:08:02 greghaynes: i haven't heard that 2010-09-12T01:08:03 eek 2010-09-12T01:08:12 hammer looks cool to me 2010-09-12T01:08:14 greghaynes: id be surprised, ZFS's lic isn't compatable with the GPL 2010-09-12T01:08:23 unless it's been relic'd 2010-09-12T01:08:40 http://linux.slashdot.org/story/10/08/27/2259253/Native-ZFS-Is-Coming-To-Linux-Next-Month?from=rss 2010-09-12T01:08:47 j3camero: umm is this an error: To be clear, this does not include forum moderators, committers, contributors, employees of the sponsor, or people who are lending server time to the contest. 2010-09-12T01:08:56 there's also zfs for fuse 2010-09-12T01:09:01 slooow 2010-09-12T01:09:03 Yea, this is in kernel thoough 2010-09-12T01:09:04 Yep 2010-09-12T01:09:17 oh it's not mainline 2010-09-12T01:09:48 Yea, im guessing due to license 2010-09-12T01:10:00 yea they said in that article they are realeasing it under CDDL 2010-09-12T01:10:07 and are not going to use any GPL'd symbols 2010-09-12T01:11:01 i hated linux kernel development at my last job. the GPL got in the way so freaking much 2010-09-12T01:11:21 i tend to get into arguments about the GPL 2010-09-12T01:11:36 heh, me and amstan have gotten into those 2010-09-12T01:11:42 :) 2010-09-12T01:11:43 we actually used a lot of freebsd code due to the license. we even contributed a lot of code to the project. we couldn't really do the same for linux because we couldn't even use it that much 2010-09-12T01:11:44 dstufft: are you danielvf? i cannot find his account on the server 2010-09-12T01:11:50 greghaynes: no we didn't, darkf did 2010-09-12T01:11:55 *** mistabell has joined #aichallenge 2010-09-12T01:12:00 greghaynes: i don't really care 2010-09-12T01:12:05 in my experiences the GPL hinders contributions more than encourages them 2010-09-12T01:12:10 I thought I remembered you giving me crap for gpl... 2010-09-12T01:12:15 amstan: no i am dstufft 2010-09-12T01:12:18 greghaynes: i don't remember 2010-09-12T01:12:18 who me? 2010-09-12T01:12:19 jmcarthur: eh, it all depends 2010-09-12T01:12:25 oh amstan 2010-09-12T01:12:46 eh, i too don't really want to argue about it. i understand the user vs. developer favoritism and all that 2010-09-12T01:13:02 and positive and negative rights, and this and that. nobody agrees what freedome is 2010-09-12T01:13:04 *freedom 2010-09-12T01:13:09 :) 2010-09-12T01:13:23 no need to get a squabble over licenses lol 2010-09-12T01:13:38 speaking of which... what license is the contest code under? 2010-09-12T01:13:40 jmcarthur: one thing i didn't like about gpl this only time, pyqt was licensed gpl only, not lgpl, so i can't make a commercial app using it 2010-09-12T01:13:43 i never looked 2010-09-12T01:13:44 I tend to use all sorts of licences for all sorts of projects since I tend to find what works best varies a lot 2010-09-12T01:14:22 amstan: Theres a pyqt alternative coming out with a more permissive license 2010-09-12T01:14:26 I forget the name... 2010-09-12T01:14:35 Hey guys... Java noob question that should be easy for someone to answer. How do you make a list of lists? Not right -> http://pastebin.com/tctVSUbe 2010-09-12T01:14:36 pyside, sucks for now 2010-09-12T01:15:13 mistabell: try #java if there's no response here 2010-09-12T01:16:09 IIRC just use arraylist and its pretty straightforward, the container type is ArrayList< ArrayList< MyClass > > 2010-09-12T01:16:21 Oh, duh. Thanks greg 2010-09-12T01:16:32 np 2010-09-12T01:17:32 amstan: updated http://code.google.com/p/ai-contest/wiki/Decentralization 2010-09-12T01:17:42 I put in there a bandwidth estimate for decentralization. 2010-09-12T01:18:59 j3camero: http://code.google.com/p/ai-contest/wiki/ContestEligibility 2010-09-12T01:19:03 j3camero: i added a list of admins 2010-09-12T01:19:27 j3camero: in the exclusion for admins... there's a point which says "or people who are lending server time to the contest." 2010-09-12T01:19:40 j3camero: this is exactly the node stuff.. it seems to be a mistake in the wiki then 2010-09-12T01:19:57 j3camero: people hosting nodes should be classified as admins and disqualified 2010-09-12T01:21:41 j3camero: is there a license on all this source code? 2010-09-12T01:22:45 jmcarthur: apache. 2010-09-12T01:22:56 jmcarthur: closest thing to the WTFPL I could find. 2010-09-12T01:23:04 *** ByteCoin has quit IRC (Quit: Page closed) 2010-09-12T01:23:10 WTFPL is an awesome lic 2010-09-12T01:23:22 WTFPL and BSD are my go to's lol 2010-09-12T01:23:34 amstan: okay go ahead and edit the wiki to exclude people lending server time. They should be excluded, since they could in principle gain unfair advantage. 2010-09-12T01:24:02 yay, my first game http://ai-contest.com/visualizer.php?game_id=4441862 2010-09-12T01:24:04 zlib is close to wtfpl 2010-09-12T01:24:19 *** Utkarsh has quit IRC (*.net *.split) 2010-09-12T01:24:19 *** JCS^ has quit IRC (*.net *.split) 2010-09-12T01:24:33 j3camero: cool thanks 2010-09-12T01:24:56 wtfpl is close to wtfpl 2010-09-12T01:25:21 jmcarthur: on google code you have to select a license from a drop down box. There is no wtfpl. 2010-09-12T01:25:27 ah 2010-09-12T01:25:34 heh 2010-09-12T01:25:52 *** mistabell has quit IRC (Quit: Page closed) 2010-09-12T01:28:44 I am taking down the TM to test the fixed python engine created by a1k0n. 2010-09-12T01:29:53 *** rwa has quit IRC (Quit: rwa) 2010-09-12T01:30:54 j3camero: no, no starter packages have been updated that i know of 2010-09-12T01:31:24 (sorry, had to assist my kid with a thomas the train set) 2010-09-12T01:31:39 it didn't occur to me that i could have repackaged those up 2010-09-12T01:32:36 j3camero: what about game data bandwidth, you know.. playing the game 2010-09-12T01:32:47 *** arik181 has joined #aichallenge 2010-09-12T01:35:19 *** rwa has joined #aichallenge 2010-09-12T01:36:20 *** tebele has joined #aichallenge 2010-09-12T01:36:35 a1k0n, are u thee? 2010-09-12T01:36:37 there 2010-09-12T01:36:45 or sb could help me please? 2010-09-12T01:37:39 come on 2010-09-12T01:37:44 ? 2010-09-12T01:37:50 tebele: don't ask to ask please 2010-09-12T01:37:56 tebele: just ask, it's a very busy channel 2010-09-12T01:38:11 I'm using the C# pack 2010-09-12T01:38:27 * greghaynes waves to arik181 2010-09-12T01:38:38 and it has a lot of bugs: some fleets dissapear, and there're zeros (0) flying 2010-09-12T01:39:12 there's a new C# starter pack, but it doesn't work 2010-09-12T01:39:26 tebele: where are you getting these issues? main server? 2010-09-12T01:39:33 no, locally 2010-09-12T01:39:41 amstan: do a free -m now. Why is there so little cache space all of a sudden? 2010-09-12T01:39:45 tebele: i think it's an issue with the visualizer 2010-09-12T01:39:51 yes, it is 2010-09-12T01:40:03 I've talked with a1k0n before 2010-09-12T01:40:08 j3camero: idk, someone cleared the cached stuff, somehow.. lol 2010-09-12T01:40:28 i wouldn't be so concerned with that cache, really 2010-09-12T01:40:30 s/one/thing 2010-09-12T01:40:35 ok guys 2010-09-12T01:40:40 i'm heading to bed 2010-09-12T01:40:44 j3camero: run htop 2010-09-12T01:40:46 and told me dowload the newest PlayGame.jar from the newest C# pack, but it doesn't work 2010-09-12T01:41:00 j3camero: for the mem field, the green stuff is the important one 2010-09-12T01:41:12 my bot will run a little longer 2010-09-12T01:41:12 so have fun with it ;) 2010-09-12T01:41:12 *** deepblue has quit IRC (Quit: Page closed) 2010-09-12T01:41:13 140mb/2026 2010-09-12T01:41:49 zaphus: replied to the wiki comment 2010-09-12T01:41:54 jmcarthur: ^^ 2010-09-12T01:41:56 *** Utkarsh has joined #aichallenge 2010-09-12T01:42:35 amstan, any solution? 2010-09-12T01:42:55 tebele: ask a1k0n? 2010-09-12T01:43:06 ¬ he is not here now 2010-09-12T01:43:31 amstan: is that free space? 2010-09-12T01:43:38 jmcarthur: no, used 2010-09-12T01:43:55 looks pretty great to me, unless it just means something is horribly wrong 2010-09-12T01:44:05 a1k0n: when I look at the diff for engine.py, the only change appears to be taking out the time.sleep(1) statement in the main game loop, and reducing the startup time for players from 3 to 2 seconds. Is that right? Is there no change to the battle resolution stuff? 2010-09-12T01:44:32 j3camero: read the backlog pls 2010-09-12T01:44:36 j3camero: instructions for stuff 2010-09-12T01:44:41 oh it changes the startup time? 2010-09-12T01:44:47 i might need to update the spec if so 2010-09-12T01:45:01 tebele: look.. i don't know this myself, try to figure it out, look through bugs, even try fixing this yourself 2010-09-12T01:45:12 tebele: if you can build a bot, you can do this too 2010-09-12T01:46:02 a1k0n: nvm, I found the full diff. Google Code was not showing me all the changes for some reason. 2010-09-12T01:49:51 looks like good throughput until around 7:45 pm then something happened today? 2010-09-12T01:50:02 *** tebele has quit IRC (Quit: Page closed) 2010-09-12T01:50:17 rwa: tournament manager was off for a while 2010-09-12T01:50:23 j3camero: yeah you have to tell it to diff between r220 and head or something 2010-09-12T01:50:23 rwa: httpd problems 2010-09-12T01:51:28 *** Baphomet has joined #aichallenge 2010-09-12T01:52:30 j3camero: how are you getting 12 games per minute? What's the average game size? something's fishy there 2010-09-12T01:53:29 oh.. i see, turn duration 2010-09-12T01:53:39 bots might be faster than 1second/turn 2010-09-12T01:53:46 danielvf: if you're around, you probably need to Math.ceil() instead of parseInt on the xmax and ymax of the dirty regions.. it's leaving a little junk on the right/bottom borders of moving objects 2010-09-12T01:54:47 a1k0n: i think we fixed that 2010-09-12T01:55:26 * Mathnerd314 finishes up his proof of the triangle inequality for the contest 2010-09-12T01:56:23 programming is so hard and confsing :( 2010-09-12T01:56:28 my bot fails so hard 2010-09-12T01:57:45 Yea -393 Elo 2010-09-12T01:58:13 just me or did the site just stop loading? 2010-09-12T01:58:20 site 2010-09-12T01:58:38 dammit was trying to update my bot and the site crashed 2010-09-12T01:59:50 obviously it's your fault 2010-09-12T02:00:04 lol 2010-09-12T02:00:10 blkperl: you crashed the site, wtf? 2010-09-12T02:00:19 blkperl: man.. it took forever to get it up 2010-09-12T02:00:19 He did? 2010-09-12T02:00:26 jk 2010-09-12T02:00:28 :) 2010-09-12T02:00:45 *** Mathnerd314 has quit IRC (Quit: ChatZilla 0.9.86-rdmsoft [XULRunner 1.9.2.8/20100722155716]) 2010-09-12T02:00:47 lol 2010-09-12T02:00:48 Aw, I would have loved to give him crap for that 2010-09-12T02:00:53 we need to sticky a thread about the starter packs not working and the updated jars and blah 2010-09-12T02:01:11 a1k0n: can you make the thread? i'll give you mods 2010-09-12T02:01:20 *** pierce_ has quit IRC (Ping timeout: 252 seconds) 2010-09-12T02:01:26 nvm, that's too complicated.. 2010-09-12T02:01:31 i'll sticky it myself 2010-09-12T02:01:35 just make the thread 2010-09-12T02:02:26 lets see if it works this time :) 2010-09-12T02:02:44 success! 2010-09-12T02:03:40 *** castle_bravo has joined #aichallenge 2010-09-12T02:05:13 contestbot: topic help 2010-09-12T02:05:13 amstan: (topic []) -- Returns the topic for . is only necessary if the message isn't sent in the channel itself. 2010-09-12T02:05:19 contestbot: help topic 2010-09-12T02:05:19 amstan: (topic []) -- Returns the topic for . is only necessary if the message isn't sent in the channel itself. 2010-09-12T02:05:23 ugh 2010-09-12T02:06:01 hey so anyone want to help me program stuff lol :p 2010-09-12T02:06:32 contestbot: topic replace 4 No stupid questions. Don't ask to ask. 2010-09-12T02:06:32 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || http://contestbot.hypertriangle.com/ || Developers wanted, contact j3camero. || No stupid questions. Don't ask to ask. || Sorry about the slowness. || Latest Idea, decentralization: http://tinyurl.com/decentralizedai" 2010-09-12T02:06:59 Baphomet: so.. what/ 2010-09-12T02:07:01 thing i mainly want to do is choose dest by planet distance from my largest plannets 2010-09-12T02:07:29 but well, i have lery limited programming experance and no idea how to do these things... :\ 2010-09-12T02:07:43 Baphomet: what language? 2010-09-12T02:07:46 c++ 2010-09-12T02:08:03 *** dstufft has quit IRC (Ping timeout: 240 seconds) 2010-09-12T02:08:05 Baphomet: ok.. hold on 2010-09-12T02:08:10 @where repo 2010-09-12T02:08:10 amstan: repo = http://code.google.com/p/ai-contest/ 2010-09-12T02:08:30 Baphomet: you know how to get maximum from an array? 2010-09-12T02:08:33 thing is, i swear i saw something about it written somewhere yesterday but i cant seem to find it again 2010-09-12T02:09:23 Baphomet: you know how to get maximum from an array? 2010-09-12T02:09:24 not really.. i've only done really basic work with arrays 2010-09-12T02:09:50 hrm, I wonder when ill get to play another match 2010-09-12T02:09:51 thats how not very experanced i am. 2010-09-12T02:10:34 *** dstufft_ has joined #aichallenge 2010-09-12T02:10:36 amstan: http://ai-contest.com/forum/viewtopic.php?f=18&t=534 2010-09-12T02:11:33 Baphomet: http://www.kirupa.com/developer/actionscript/array_max_min.htm use Planet::NumShips with that 2010-09-12T02:11:48 Baphomet: that's how you get largest planet 2010-09-12T02:12:08 that was fast 2010-09-12T02:12:15 * blkperl wonders the same thing as greghaynes 2010-09-12T02:12:16 a1k0n: yep.. i just did it 2010-09-12T02:12:40 a1k0n: it's a link at the end of the topic: Quick-mod tools: Sticky topic GO 2010-09-12T02:13:58 greghayes: dunno, games seem to be much slower than earlier today 2010-09-12T02:14:09 amstan: did you install ruby to the server yet? :) 2010-09-12T02:14:28 blkperl: starter packages are in the works, i don't know more than than myself 2010-09-12T02:14:47 i know about the starter pacakages im just wondering if ruby is actually on the server 2010-09-12T02:15:21 amstan@ltfs1110:~$ ruby 2010-09-12T02:15:23 The program 'ruby' is currently not installed. You can install it by typing: 2010-09-12T02:15:26 ugh 2010-09-12T02:15:31 my vps's router went out 2010-09-12T02:15:34 blkperl: ruby support is coming soon. It was supposed to happen today, but we had rather a web server malfunction. 2010-09-12T02:15:35 ...apt-get install ruby? 2010-09-12T02:15:46 j3cemero i sent you a pm again, on this nick though 2010-09-12T02:15:58 blkperl: even after installation, it still needs to be added to the compiling script 2010-09-12T02:16:01 j3camero: thanks, make sure to announce it :) 2010-09-12T02:16:08 when its ready :) 2010-09-12T02:16:08 a1k0n: this python engine is workign decently. I am running the TM manually and watching it now to test it. I fixed a few little array-out-of-bounds things, but overall it looks good. 2010-09-12T02:16:24 will do 2010-09-12T02:16:26 so whats actually wrong with the site? is it just overloaded or something? 2010-09-12T02:16:48 *** dhartmei has joined #aichallenge 2010-09-12T02:16:55 oh steadfast decided to upgrade their core router 2010-09-12T02:16:56 j3camero: ok, awesome. 2010-09-12T02:16:56 Yeah way too much traffic. Plus I tried to save money by getting a shitty server and now I am regretting it. 2010-09-12T02:17:02 dhartmei: yo 2010-09-12T02:17:04 so glad they told me about it 2010-09-12T02:17:09 hi 2010-09-12T02:17:11 j3camero: it's all your fault! 2010-09-12T02:17:11 instead of just killing my connection 2010-09-12T02:18:22 dstufft_: hahah, right after you got your site set up? 2010-09-12T02:18:45 a1k0n: well it's been setup for a few now, but yea basically 2010-09-12T02:18:56 my irc shell is on that vps too 2010-09-12T02:19:08 i feel so cutoff >.> 2010-09-12T02:19:53 demand a refund 2010-09-12T02:19:56 hey just a question about running bots locally 2010-09-12T02:20:01 in the command line 2010-09-12T02:20:08 when you have '1000 1000' 2010-09-12T02:20:19 what exactly are those 2 values? 2010-09-12T02:20:27 2nd one is max turns 2010-09-12T02:20:41 *** Utkarsh has quit IRC (Read error: Connection reset by peer) 2010-09-12T02:20:44 first is time per turn 2010-09-12T02:21:02 maximum time per turn, that is 2010-09-12T02:21:09 ah ok, thanks 2010-09-12T02:21:13 1000 = 1 second 2010-09-12T02:22:20 ok any reasons why when i run it at max 200 rounds the program runs heaps slower? 2010-09-12T02:23:13 oh i see 2010-09-12T02:23:15 Because you touch yourself at night. 2010-09-12T02:23:23 they had this elaborate plan to not have downtime 2010-09-12T02:23:37 So did we, dstufft_ 2010-09-12T02:23:38 but they fucked it up and: BGP Convergence 2010-09-12T02:23:53 and now i can't reach my vps because fios is dumb 2010-09-12T02:24:00 movie! 2010-09-12T02:24:01 @yay 2010-09-12T02:24:01 Double rainbow all the way! 2010-09-12T02:24:54 WHAT DOES IT MEEEEAN?!!? 2010-09-12T02:24:57 * a1k0n sobs 2010-09-12T02:26:04 I wonder when ill play another match.. 2010-09-12T02:26:09 me too. 2010-09-12T02:26:21 i gave up 1st or 2nd place to change a number in my code. :( 2010-09-12T02:26:31 *** Baphomet has quit IRC (Ping timeout: 240 seconds) 2010-09-12T02:26:35 thing is.. a lot of submissions were.... submitted in the downtime 2010-09-12T02:26:39 the TM has to catch up 2010-09-12T02:27:09 was that double rainbow guy on heavy drugs or what? 2010-09-12T02:27:10 is the new TM running now? 2010-09-12T02:27:11 awesome 2010-09-12T02:27:21 a1k0n: the TM is back up and running with the fixed python engine. 2010-09-12T02:27:26 rwa: he's working for microsoft now 2010-09-12T02:27:32 rwa: i saw a subsequent interview where someone asked him that. "not at the time" was his answer 2010-09-12T02:27:39 can this be an engine failure? what are the odds of them both crashing at 1st turn? http://ai-contest.com/visualizer.php?game_id=4434927 2010-09-12T02:27:43 the microsoft commercial is fucking terrible 2010-09-12T02:28:08 none of the original... je ne sais quois 2010-09-12T02:28:17 ratatata: uhhhh that was my fault. I fixed the issue causing that. Nevermind you. 2010-09-12T02:28:18 a1k0n: wow. I was kinda hoping he was tripping on something, otherwise I think he may not be "well". 2010-09-12T02:28:31 the dude is just nuts 2010-09-12T02:28:42 *** Utkarsh has joined #aichallenge 2010-09-12T02:28:45 also, he sounds exactly like 'the dude' (aka jeff bridges) 2010-09-12T02:29:10 j3camero: /cry.. I was so happy to have found a draw in high ranks :P 2010-09-12T02:29:15 he was literally knocked backward by the power of the double rainbow, and stuff 2010-09-12T02:29:31 *** Baphomet has joined #aichallenge 2010-09-12T02:29:36 wow 2010-09-12T02:29:49 i was just testing to see how laggy the program gets with round numbers 2010-09-12T02:29:59 and it lagged my comp so much mirc broke 2010-09-12T02:30:06 aaand the rankings page is slow again 2010-09-12T02:30:29 round numbers? 2010-09-12T02:30:34 @rankings 2010-09-12T02:30:45 amstan: Error: There was a problem accessing the interface to ai-contest.com 2010-09-12T02:30:49 awwww 2010-09-12T02:31:20 j3camero: site is getting slow again, my bot can't access stuff 2010-09-12T02:31:43 j3camero: did you remember to nice -n 19 it? 2010-09-12T02:32:10 2 java bots are hogging the cpu 2010-09-12T02:32:50 j3camero: i'm an idiot and forgot to tell you to apt-get upgrade >.> 2010-09-12T02:33:15 amstan: the rankings seem fine to me. They are responsive. 2010-09-12T02:33:21 j3camero: actually... 2010-09-12T02:33:25 they waver 2010-09-12T02:33:38 j3camero: the 2 java bots with 100% cpu, they have a cpu time of 1 hour 2010-09-12T02:33:39 each 2010-09-12T02:33:46 j3camero: they must be infinite looping 2010-09-12T02:33:50 o_O 2010-09-12T02:33:59 sweet 2010-09-12T02:34:05 lol 2010-09-12T02:34:05 i reniced them, but.. 1 hour is not normal 2010-09-12T02:34:26 contestbot: rankings 2010-09-12T02:34:35 amstan: Error: There was a problem accessing the interface to ai-contest.com 2010-09-12T02:34:37 site's stuck currently 2010-09-12T02:34:39 :) 2010-09-12T02:34:40 aww 2010-09-12T02:34:50 SHould be good now? 2010-09-12T02:34:57 j3camero: shall i kill them? 2010-09-12T02:35:01 no no 2010-09-12T02:35:03 its cool 2010-09-12T02:35:17 j3camero: it's the same 2 processes though 2010-09-12T02:35:39 The site just went down for a few seconds because of the apt-get upgrade that dstufft_ told me to do. It updated apache, so it stopped the apache server. That is so retarded. 2010-09-12T02:35:49 contestbot: rankings 2010-09-12T02:35:50 amstan: Top 10 players: dmj111(2824), albert(2796), seigen(2755), Kavish(2745), drewrobb(2690), strank(2680), danielvf(2679), sighbot(2661), cheese_bot(2656), vortex(2642) 2010-09-12T02:35:59 ok... 2010-09-12T02:36:00 amstan: are you sure those java processes aren't just two bots playing a game? 2010-09-12T02:36:12 one hour of cpu time though.. 2010-09-12T02:36:14 don't think so 2010-09-12T02:36:19 that's over 3600 game turns 2010-09-12T02:36:19 *** Palmik has joined #aichallenge 2010-09-12T02:36:52 amstan: oh in that case, kill them individually. 2010-09-12T02:36:59 amstan: thanks for noticing that. 2010-09-12T02:37:13 amstan: are they running as jailuser*? 2010-09-12T02:37:23 j3camero: user 10022 2010-09-12T02:37:29 there we go! 2010-09-12T02:37:31 who is that? 2010-09-12T02:37:34 cpu is back down 2010-09-12T02:37:39 j3camero: probably a jail 2010-09-12T02:37:46 lol I hope those processes weren't important... 2010-09-12T02:37:47 contestbot: rankings 2010-09-12T02:37:48 amstan: Top 10 players: dmj111(2824), albert(2796), seigen(2755), Kavish(2745), drewrobb(2690), strank(2680), danielvf(2679), sighbot(2661), cheese_bot(2656), vortex(2642) 2010-09-12T02:37:54 j3camero: much better 2010-09-12T02:38:01 lol well atleast apache is updated and ruby is installed :p 2010-09-12T02:38:02 Nice and snappy. 2010-09-12T02:38:02 RobotCaleb: site faster? 2010-09-12T02:38:14 ya its working good right now 2010-09-12T02:38:20 so um, do you have a cpu time limit on processes in the sandbox or jail or whatever? 2010-09-12T02:38:22 stupid java.. 2010-09-12T02:38:36 amstan: Yep 2010-09-12T02:38:41 (ulimit -t) 2010-09-12T02:38:44 a1k0n: no cpu time limit, no. We don't even relaly know how to measure the bots' cpu time. 2010-09-12T02:38:46 j3camero: yes.. so what happened there? why did they get that far? 2010-09-12T02:39:07 j3camero: ulimit -t ought to be a safe way to do it 2010-09-12T02:39:20 amstan: I dunno. Perhaps the engine or TM died and did not kill the two java bots, so they just kept spinning? 2010-09-12T02:39:43 why is the round limit on the server so low :( my bot only really gets good late game 2010-09-12T02:39:51 j3camero: can you renice the tournament engine? 2010-09-12T02:39:59 j3camero: it will make the site go much faster? 2010-09-12T02:40:00 amstan: that won't do anything. 2010-09-12T02:40:24 amstan: it's not the TM process itself that's using all the CPU. It's probably the actual bots. 2010-09-12T02:40:36 j3camero: nice is inhereted 2010-09-12T02:40:44 *** sinan has joined #aichallenge 2010-09-12T02:40:48 amstan: plus the TM restarts every few minutes with a new process. 2010-09-12T02:40:57 j3camero: how? cron? 2010-09-12T02:41:02 amstan: yeah 2010-09-12T02:41:02 you can start a process with a particular nice level 2010-09-12T02:41:02 j3camero: put the nice in there then 2010-09-12T02:41:22 amstan: I dunno. Why do that? CPU isn't much concern is it? 2010-09-12T02:41:38 j3camero: nice 19 command 2010-09-12T02:41:40 j3camero: it is... 2010-09-12T02:41:49 not a lot, but it is 2010-09-12T02:41:52 amstan: to be honest i think the main issue is the database 2010-09-12T02:41:57 amstan: if we nice the bots, I am afraid that some bot will get screwed and time out just because someone happened to load a heavy page on the website as it was finishing its turn. 2010-09-12T02:42:03 amstan: the TM i think uses a lot of it when it's starting up 2010-09-12T02:42:11 amstan: slowing down the TM might actually make it work 2010-09-12T02:42:14 jmcarthur: that's why bots should of used cpu time 2010-09-12T02:42:23 *make it worse 2010-09-12T02:42:39 j3camero: that's why bots should of used cpu time, wasn't this fixed already? 2010-09-12T02:42:47 The TM doesn't appear to be slow as it starts up. I was just running it manually and it seemed fine. 2010-09-12T02:42:54 there are 2 games that are marked as losses for me, when i try to view them, there is only 1 turn (and nothing happens). what could it be? 2010-09-12T02:43:05 sinan: was it java? 2010-09-12T02:43:10 i think the main issue is lock contention on the DB, or atleast one of the issues 2010-09-12T02:43:14 i'm just saying that if the TM still has a slow query or something then that will just take longer to finish at a lower priority 2010-09-12T02:43:18 sinan: that was a little problem that is fixed now. Don't worry about it. 2010-09-12T02:43:19 amstan: no, i'm using python. 2010-09-12T02:43:27 if anything you would want the TM to completely *faster* 2010-09-12T02:43:28 j3camero: ah ok, thanks :) 2010-09-12T02:43:31 *complete 2010-09-12T02:43:42 sinan: reupload your code if you want to get rid of the lost game 2010-09-12T02:43:51 j3camero: does it mean my AI didn't play in them, or actually lost but i can't view the game? 2010-09-12T02:44:07 amstan: cool :) i have a new version coming up soon so i'll do it then 2010-09-12T02:44:09 jmcarthur: idk.. site seems a lot more important 2010-09-12T02:44:17 amstan: no that's what i'm saying 2010-09-12T02:44:24 sinan: it means either you or your opponent arbitrarily lose the game through no fault of your own. But don't worry, it will even out. 2010-09-12T02:44:35 amstan: if the TM has a slow query and is locking the db that could explain the occasional slowdown 2010-09-12T02:44:37 j3camero: if that's the case, great ;) thanks 2010-09-12T02:44:44 amstan: so speeding up the TM will release those locks faster 2010-09-12T02:45:00 amstan: let's not go nice-ing things. We tried that kind of stuff in the first GAC, and it always had disastrous unintended consequences. 2010-09-12T02:45:11 jmcarthur: does renicing the TM actually renice the queries too? i didn't think it was the case 2010-09-12T02:45:23 amstan: it means the TM can process them more quickly on its end 2010-09-12T02:45:37 most queries are incremental, after all 2010-09-12T02:45:40 no it doesn't change the queries 2010-09-12T02:45:48 the queries are done in the DB processes 2010-09-12T02:45:54 it changes nothing in the DB 2010-09-12T02:46:06 SELECTS will be in paralell with other selects 2010-09-12T02:46:20 but UPDATES are 1 at a time, and the DB is either SELECTing or INSERT/UPDATING 2010-09-12T02:46:36 so a long running SELECT can get a long queue for updates 2010-09-12T02:46:37 We seem to be golden on memory and apache processes. I wonder how things got so bad earlier in the day. Do you think we might have had a huge traffic spike all at once or something? Like maybe a famous person put a link to the leaderboard in a twitter post or some shit? 2010-09-12T02:46:43 somebody said earlier that the TM locks a lot of stuff in the db 2010-09-12T02:46:43 j3camero: so.. how is the time done now? cpu time or wall time? 2010-09-12T02:47:03 still wall time i'm sure 2010-09-12T02:47:06 wall time 2010-09-12T02:47:09 j3camero: before it was bad becasue we had that limit set too low, it wasn't the ram 2010-09-12T02:47:15 j3camero: can we fix it? change to cpu time? 2010-09-12T02:47:26 There has been a conversation about cpu time. We just need to think it through very carefully. It's not as simple as it seems at first. 2010-09-12T02:47:41 indeed it's not :( 2010-09-12T02:48:04 if i submit a while(1); bot, will it get killed after it times out? 2010-09-12T02:48:05 j3camero: i'm just saying.. wall time is the most unfair thing ever, especially since you have http and mysql on the same box 2010-09-12T02:48:10 jmcarthur: on a singlecore cpu.. 2010-09-12T02:48:16 j3camero: ^ 2010-09-12T02:48:23 i agree. wall time *killed* my bot in the last contest 2010-09-12T02:48:30 It's appealing to "just switch to cpu time to make things fairer" but there are some bad things that can happen. What if somebody submit this bot? while(1) {sleep(999);} 2010-09-12T02:48:38 jmcarthur: sorry for highlighting you too, too many ppl here, my autocomplete limit is set too low 2010-09-12T02:48:41 j3camero: that won't hurt anything 2010-09-12T02:48:44 and mine. many of my games had errors that were only made if it could only search for like 100ms 2010-09-12T02:48:46 j3camero: it's not *doing* anything 2010-09-12T02:49:01 j3camero: use both then, 10 seconds wall, 1 second cpu 2010-09-12T02:49:09 j3camero: plus, the ideal solution would do 1 cpu .... yeah that 2010-09-12T02:49:29 10 wall results in DQ, 1 cpu results in losing the match 2010-09-12T02:49:52 j3camero: but i agree, sleep(999) does not actually use up the cpu, it just waits forever 2010-09-12T02:49:53 well, the server load could be at 200 and it takes 10 wall clock seconds to run 1 cpu second easily 2010-09-12T02:50:05 amstan: that's one possible solution, but then somebody could submit a bot that waits for 10 seconds every turn, just to slow down out games/minute rate, which then fucks the leaderboard. 2010-09-12T02:50:36 j3camero: you're being a bit paranoid 2010-09-12T02:50:40 i really really hate dependencies 2010-09-12T02:50:42 if somebody wants to fuck things up they can anyway 2010-09-12T02:50:46 j3camero: once you have cpu time setup, you can run like 20 games concurrently, since cpu time will be accurate in giving everyone good cpu usage 2010-09-12T02:50:47 The reason for the time limit is to guarantee a minimum throughput on the games, not to be fair to the cotnestants. 2010-09-12T02:50:49 and we can attack that on a case by case basis 2010-09-12T02:51:01 the point is that the wall time is *unfair* to contestants 2010-09-12T02:51:09 rather than neutral 2010-09-12T02:51:21 i'm with jmcarthur 2010-09-12T02:51:36 right now it is unfair, someone loads a page, a bot gets less time 2010-09-12T02:52:16 The interim rankings don't really matter though. They're just to get an idea how you're doing. 2010-09-12T02:52:33 did you switch to cpu time for the final tournament last time? 2010-09-12T02:52:37 j3camero: it'll be the same problem in the final tournament 2010-09-12T02:52:37 During the final tournament, we turn off the leaderboard, traffic drops for a few days, and we can run a fair tournament. 2010-09-12T02:52:39 *** gilaniali has quit IRC (Quit: gilaniali) 2010-09-12T02:52:44 it's still not fair 2010-09-12T02:52:55 j3camero: mysql, same problem 2010-09-12T02:53:08 even with only two bots on the machine at once and that's it it's not fair 2010-09-12T02:53:10 other bots as well 2010-09-12T02:53:22 j3camero: apt-get install libevent and see if that get's the newer version of libevent (when you got a sec) 2010-09-12T02:53:45 dstufft_: i can do it 2010-09-12T02:53:48 dstufft_: but why install? 2010-09-12T02:54:39 amstan: if i recall the version of libevent in the mainline repo is named libevent1 or something along that lines 2010-09-12T02:54:40 personally, i think game throughput is less important than game fairness, even when just testing. hard to test when you don't even know if your bot gets 0.95 seconds or 0.05 seconds 2010-09-12T02:55:19 jmcarthur: are you getting all this? 2010-09-12T02:55:20 I'm trying to get a newer version from a ppa, but it looks like they named it libevent 2010-09-12T02:55:24 j3camero: are you getting all this? 2010-09-12T02:56:23 anyway, it's time for me to head to bed 2010-09-12T02:56:30 noo, I dropped during the night :D 2010-09-12T02:56:30 gnite guys 2010-09-12T02:56:31 dstufft_: k.. so like.. good enough: http://packages.ubuntu.com/search?keywords=libevent&searchon=names&suite=hardy§ion=all 2010-09-12T02:56:35 good night 2010-09-12T02:56:56 yeah, i'm off too 2010-09-12T02:57:16 *** Appleman1234 has quit IRC (Ping timeout: 276 seconds) 2010-09-12T02:58:38 *** Snowplan has left #aichallenge 2010-09-12T02:58:40 amstan: we can take another look at CPU time if someone can figure out how to enforce a CPU time limit accurately, I guess. 2010-09-12T02:58:59 amstan: it's not clear to me that anyone has figured it out yet. 2010-09-12T02:59:12 amstan: anyways it's time for bed for me. 2010-09-12T02:59:46 is the new tournament manager running? 2010-09-12T02:59:53 yes 2010-09-12T02:59:55 cool 2010-09-12T03:00:01 * amstan going away 2010-09-12T03:00:03 *** castle_bravo has quit IRC (Quit: leaving) 2010-09-12T03:00:18 *** Appleman1234 has joined #aichallenge 2010-09-12T03:00:23 does new include a new pairing algorithm other than random yet? 2010-09-12T03:00:33 I think i found the reason why I sometimes send invalid orders even if they should be valid... it's probably because in one turn, I am sending more then once from the same planet and the data structure is not updating... happy to figure it out, unhappy that I have to rewrite quite a lot of things :P 2010-09-12T03:01:18 What is the new turnament manager? 2010-09-12T03:01:59 Palmik: you should check total amount of ships sent from single planet. 2010-09-12T03:02:21 j3camero: You could parse /proc//stat for each of the bots regularly, it has a field for how much user-mode time a process has had. 2010-09-12T03:02:22 i thought it was the one in the patch which matched people up preferentially against people with similar elo ratings 2010-09-12T03:02:37 kirkeby: yeah that's what I was thinking. 2010-09-12T03:03:26 dolphin278, I of course, do that... the thing is, that after you issue order, your planets are not updated (the numShips is not lowered), so when I send again from that planet, I got invalid data 2010-09-12T03:03:37 a1k0n Janzert: I haven't had a chance to switch to a decent pairing algorithm yet. Was supposed to do this today, but the server meltdown took up all the time. Hopefully this will happen tomorrow. For now, the rankings just converge really slowly. 2010-09-12T03:03:50 it should be easy to fix that though 2010-09-12T03:04:07 I am going to bed now. See you all tomorrow! 2010-09-12T03:04:14 Thanks everyone for their help! 2010-09-12T03:04:26 a1k0n, that will be sweet :) 2010-09-12T03:04:39 j3camero: fair enough. night! 2010-09-12T03:04:45 Definitely appreciated the help of this whole channel this afternoon when the server was being overloaded. You guys are like super-google. 2010-09-12T03:05:07 good night 2010-09-12T03:05:10 a1k0n: thanks for all these sweet patches you've been contributing. You are the bomb. 2010-09-12T03:05:23 np, many of them came from other people actually 2010-09-12T03:05:52 my patches are mainly out of self-interest to get the damn engine to work, and fst 2010-09-12T03:05:55 fast 2010-09-12T03:09:02 ahha. at least the new tournament engine in production does battle resolution correctly 2010-09-12T03:09:57 later j3camero, sent you one last pm unless you already left :p 2010-09-12T03:10:09 dstufft_: stop pming them 2010-09-12T03:10:13 dstufft_: use the bot.. 2010-09-12T03:10:53 *** vikhyat has quit IRC (Ping timeout: 240 seconds) 2010-09-12T03:11:19 oh amstan your here still :p wanna http://pastebin.com/gMXUpRyM real quick for me? If not tht's cool, gonna run and grab some food though, later :p 2010-09-12T03:11:44 dstufft_: dude.. i still need to know why 2010-09-12T03:11:52 and no.. i already started my movie 2010-09-12T03:12:15 amstan: oh i thought i said why 2010-09-12T03:12:23 DualBot on Map26 beats me -- starting planets are annoying close, and the planets I want to take are hella far away. :( 2010-09-12T03:12:30 php-fpm requires version 1.4.11 of libevent 2010-09-12T03:12:44 ubuntu 8.04 has 1.3 2010-09-12T03:14:46 bss03-AFK how do you fair on map 30 ? that one sucks :-) 2010-09-12T03:14:57 *** bss03-AFK is now known as bss03 2010-09-12T03:15:06 I didn't view the game, but I won it. 2010-09-12T03:15:32 how come there isn't and oregon state org? 2010-09-12T03:15:37 On turn 187, I guess. 2010-09-12T03:16:37 On that one, Dualbot is less interested in my home planet and sends to many fleets "prospecting". 2010-09-12T03:18:07 *** jmreardon has quit IRC (Quit: jmreardon) 2010-09-12T03:20:02 http://ai-contest.com/forum/viewtopic.php?f=17&t=486&sid=fc68b3c3a1a2772b1b36e2d8d053a463 nice, so it's happening :) 2010-09-12T03:22:20 *** computerwiz_222 has quit IRC (Quit: computerwiz_222) 2010-09-12T03:22:30 Palmik: I'm thinking of doing some kind of genetic/etc version as well, just to see what happens - then if it gets better than my handcrafted one I'll submit it instead :-) 2010-09-12T03:23:01 *** vikhyat has joined #aichallenge 2010-09-12T03:23:30 Hope you or someone else releases their code after the contest... I was always attracted by this tuff :D 2010-09-12T03:23:59 Palmik: I think there is some kind of Geneva convention about having my code released, offensive to the senses 2010-09-12T03:24:26 naah :D 2010-09-12T03:24:55 Palmik: I'm not doing GP, but my bot is LGPL'd, so anyone can view it. 2010-09-12T03:24:57 *** hornairs has quit IRC (Remote host closed the connection) 2010-09-12T03:25:14 I'm still on the first page of results, but there's tons of stuff my bot is stupid about. 2010-09-12T03:26:55 hey so just wondering how i compile .java to .jav on windows? i have latest java jdk but running javac *.java doesnt do anything? 2010-09-12T03:27:19 javac only produces the .class files. 2010-09-12T03:27:38 Found jaspervdj's bot on google under "planet wars bot" http://github.com/jaspervdj/planet-wars-haskell I know you posted the git link yesterday (my yesterday), but I forgot to bookmark it, would mind sharing it again? :) 2010-09-12T03:27:42 ok then thats fine too, but it still doesnt work 2010-09-12T03:27:47 Baphomet: But, that should be enough to test with. 2010-09-12T03:27:53 im just trying to follow the guide on the site 2010-09-12T03:28:02 The jaspervdj's bot is just a starter package though 2010-09-12T03:29:04 http://github.com/stephen-smith/ai-contest-2010 -- I'm a fork of the starter kit so you can also find me linked from http://github.com/jaspervdj/planet-wars-haskell/network/members 2010-09-12T03:30:10 aah :) thanks a lot, I'm sure many people will find it useful as well :) 2010-09-12T03:30:10 jaspervdj has pulled a few things from me that I licensed under the WTF license; stuff I wanted to contribute back to the starter kit. 2010-09-12T03:31:04 But, the majority of my stuff is LGPL, so share alike, plz. 2010-09-12T03:31:15 so um about compiling .java to .class how does one do this? 2010-09-12T03:31:24 javac 2010-09-12T03:31:44 just says its not a recognosed program 2010-09-12T03:31:52 and i just installed the latest ver 2010-09-12T03:32:02 bss03, I'm working in c++ but I already released my AbstractBot for basic stuff 2010-09-12T03:32:07 Apparently not correctly 2010-09-12T03:32:32 :( just did default install 2010-09-12T03:32:39 Palmik: Sweet. Got a link? I'll share with my roommate that is working in C++. 2010-09-12T03:32:46 http://www.benzedrine.cx/planetwars/canvas?game_id=1284274900 this map is crazy, almost all planets are 5 growth :) 2010-09-12T03:33:12 *** RobinB has joined #aichallenge 2010-09-12T03:33:17 morning! 2010-09-12T03:33:52 bss03, sure, here http://gitorious.org/planet-wars-abstract-bot It's a bit outdated comapred to my recent version, but I got some problems with the recent one and I'm not sure it they were coused by me playing around with the PlanetWars core :) 2010-09-12T03:34:43 Palmik: Almost all were growth 5 and 'dwc-001' decided to go for the one that wasnt, and was most expensive :-) 2010-09-12T03:35:01 *** delt0r_ has joined #aichallenge 2010-09-12T03:35:04 *** FlagCapper has joined #aichallenge 2010-09-12T03:35:21 *compared, if, caused... ah, my typos :) 2010-09-12T03:35:27 zaphus, yeah, well :)) 2010-09-12T03:35:42 one has to start somewhere :) 2010-09-12T03:35:48 lol 2010-09-12T03:36:21 *** delt0r___ has quit IRC (Ping timeout: 252 seconds) 2010-09-12T03:37:40 damn, how long it takes between battles :((( 2010-09-12T03:37:46 I was going after it as well, since it has 3 growth rate, by capturing it, I will virtually gain 6 growth gain :) 2010-09-12T03:39:11 I wonder if ill have played another match when I wake up tomorro 2010-09-12T03:40:08 dolphin278: my last two intervals were about 80 minutes and about 17 minutes 2010-09-12T03:41:12 rwa: after last submission i have one fight more that 1.5 hours ago 2010-09-12T03:41:32 hmm - just reconnected to the tcp server and it says I have a -999999 Elo, even though I can see my score up on the leaderboards (same name, I checked) 2010-09-12T03:41:50 after my first sumbmission I waited for more then 5 hours for 1st fight :) do not worry, you will fight 2010-09-12T03:42:15 zaphus, i resets imo after certain treshold is breached 2010-09-12T03:42:29 oh, time limit sort of thing - ok 2010-09-12T03:42:42 yeah, probably 2010-09-12T03:42:51 first day they was pretty often. hope guys will find a way to parallelize computations for several boxes... 2010-09-12T03:43:02 *** Theobon has joined #aichallenge 2010-09-12T03:44:17 hey i just tried that abstract bot but it doesnt seem to do anything? 2010-09-12T03:44:52 it all compiled ok but when used it just sits there and doesnt send any ships? 2010-09-12T03:45:12 right now the odds of you being picked for a match are over 300-1 (2 players out of 600 for each match) - so it could obviously be a while before you play :-) the scheduler will be changing soon to help with that 2010-09-12T03:45:34 there's a match every five seconds or so tho 2010-09-12T03:45:42 *** bss03 is now known as bss03-AFK 2010-09-12T03:46:03 a1k0n so with luck you'd get a game every 25 minutes (5 seconds * 300 games) 2010-09-12T03:46:27 i think it's completly random 2010-09-12T03:46:31 so 2010-09-12T03:46:37 highly unlikely to get 1 every 25min 2010-09-12T03:46:47 it probably is right now... as is the pairing 2010-09-12T03:46:48 *** sinan has quit IRC (Ping timeout: 252 seconds) 2010-09-12T03:47:00 if it's completely random then that is the rate you'd get them, except it's a poisson process 2010-09-12T03:47:38 a1k0n: that's only the rate you'd get them if it's a random permutation not a combination 2010-09-12T03:47:39 we can fit a gamma distribution to the observed set of games to find the rate of games played :) 2010-09-12T03:47:53 dstufft_: but it's a combination of 2 out of 500, so it's close enough 2010-09-12T03:48:06 600+ now 2010-09-12T03:48:18 643 2010-09-12T03:48:47 there is a bot who hasn't played a game in the last ~20 hours 2010-09-12T03:49:20 something flaky on tcp server, my bot just got told it has -999999 elo a few seconds after completing a match, and that deepblue has -999999 - guess it couldnt retrieve the numbers 2010-09-12T03:53:14 mine shows 481 2010-09-12T03:53:18 does anyone have c++ code to set closest planet as dest that they wana share with me? :p 2010-09-12T03:53:32 *** tapwater has quit IRC (Quit: tapwater) 2010-09-12T03:53:41 zaphus: fixed now, calculating elo takes > 10 s now, i didn't write to a temp file 2010-09-12T03:54:03 if i even had the slightest clue how to calculate distances i would do it myself 2010-09-12T03:54:06 ahh 2010-09-12T03:54:17 Baphomet: pw.Distance()? 2010-09-12T03:54:32 dhartmei: ah, ok - my bot also never finished its game against deepblue - it was just sitting there waiting for the next turn, I checked the website and it seems I lost the match quite a while ago :-) 2010-09-12T03:54:43 yeah, and then just sort them... or if you want just the closest, sorting is a bit overkill 2010-09-12T03:55:33 if you wanted to do it yourself, dist = sqrt(dx^2 +dy^2) 2010-09-12T03:56:20 oh ok 2010-09-12T03:58:42 baphomet: and if all you want is to sort for order (closest vs furthest), not actual distance, then drop the sqrt and save some time (just dont use it as actual distance) 2010-09-12T03:58:43 how do i actually use the pw.Distance() function tho? 2010-09-12T03:58:55 sorry im kinda noob at this :\ 2010-09-12T03:59:23 pass 2 planet ids 2010-09-12T03:59:29 as arguments 2010-09-12T03:59:33 *** vikhyat has quit IRC (Ping timeout: 252 seconds) 2010-09-12T04:01:39 *** antimatroid has joined #aichallenge 2010-09-12T04:02:53 *** antimatroid1 has quit IRC (Ping timeout: 240 seconds) 2010-09-12T04:06:09 *** AJC has joined #aichallenge 2010-09-12T04:09:18 *** Theobon has quit IRC (Ping timeout: 252 seconds) 2010-09-12T04:09:32 Would it be possible to add a bar to the graph of the visualiser, which indicates where in the replay you're at? 2010-09-12T04:11:26 *** vikhyat has joined #aichallenge 2010-09-12T04:13:05 did the webserver just die ? 2010-09-12T04:13:05 *** McLeopold has quit IRC (Read error: Connection reset by peer) 2010-09-12T04:13:20 yah 2010-09-12T04:13:49 crap, was just preparing my forum post on a possible engine bug :-) 2010-09-12T04:14:28 blarg everyone tries to help me and i still dunno how to do anything 2010-09-12T04:15:36 if there is pw.NotMyPlanets() and pw.MyPlanets() is there just a pw.planets()? 2010-09-12T04:16:03 they all exist as well as pw.EnemyPlanets() 2010-09-12T04:17:39 Can I send ships so that the planet is left with 0 ships or is that illegal? 2010-09-12T04:17:50 that's fine 2010-09-12T04:17:56 What is the algo for determining the growth rate on a planet? growth rate * num_ships? 2010-09-12T04:18:13 no, constant, not dependant on number of ships 2010-09-12T04:18:30 hrmm 2010-09-12T04:20:22 *** sinan has joined #aichallenge 2010-09-12T04:22:52 *** Utkarsh has quit IRC (Ping timeout: 276 seconds) 2010-09-12T04:26:01 *** rwa has quit IRC (Quit: rwa) 2010-09-12T04:27:12 *** mtcbot has joined #aichallenge 2010-09-12T04:27:20 Baphomet: Read the PlanetWars.java file if you can. 2010-09-12T04:27:48 Baphomet: However, I bet the function is named "Planets". So, "pw.Planets()" is how you would call it. 2010-09-12T04:27:54 *** bss03-AFK is now known as bss03 2010-09-12T04:27:57 *** Utkarsh has joined #aichallenge 2010-09-12T04:28:01 Baphomet: true 2010-09-12T04:28:11 ya i had a look and worked some stuff out 2010-09-12T04:28:25 posted my "I think theres a bug in the engine" forum post: http://www.ai-contest.com/forum/viewtopic.php?f=18&t=537 I'll probably be proved wrong :-) 2010-09-12T04:30:19 zaphus: I asked about that couple hours ago.. specifically about the 1st match example.. I believe it was j3camero saying that he fixed the cause of that 2010-09-12T04:30:46 so yes, there was a bug and hotfixed me thinks 2010-09-12T04:30:54 ratatata: ah, I wasnt here for that - was teaching my boy to ride without training wheels :-) 2010-09-12T04:31:20 how old? 2010-09-12T04:31:25 yay, solved the invalid orders problem... 2010-09-12T04:31:44 ratatata: 5yo 2010-09-12T04:32:04 here is the solution if anyone alse runs into the same problem http://codepad.org/X6rwNZ40 2010-09-12T04:33:52 *** vikhyat has quit IRC (Quit: Leaving) 2010-09-12T04:35:26 Palmik: that is one long conditional :P 2010-09-12T04:35:35 *** jaspervdj has joined #aichallenge 2010-09-12T04:35:47 *** jaspervdj has quit IRC (Client Quit) 2010-09-12T04:36:38 ratatata, yeah, well :P I'm thinking about remaking the whole order system 2010-09-12T04:36:52 just have to evaluate the gains :) 2010-09-12T04:38:46 *** jaspervdj has joined #aichallenge 2010-09-12T04:39:23 Javache: for your game visualizer... I'm liking the graph at the bottom 2010-09-12T04:39:44 *** ratatata has quit IRC (Ping timeout: 252 seconds) 2010-09-12T04:39:53 Javache: It would be cool if there would be some cursor showing the current position on the graph 2010-09-12T04:40:02 (i.e. the time) 2010-09-12T04:41:46 oh, loving that graph 2010-09-12T04:42:44 http://ai-contest.com/visualizer.php?game_id=4442686 -- I am the comeback king. :P 2010-09-12T04:44:48 they need to plot ship growth as well 2010-09-12T04:45:18 They do. 2010-09-12T04:45:33 Production = d(Ships) / dt 2010-09-12T04:45:53 So, it's the slope of the graph. 2010-09-12T04:47:38 it seems ExpandBot in the example bots is just a copy of BullyBot, am I right? 2010-09-12T04:51:28 *** mtcbot has quit IRC (Ping timeout: 252 seconds) 2010-09-12T04:54:55 hey guys i see this on one of the guides on the site but it doesnt seem to work in c++ 'if (pw.Production(1) >= pw.Production(2)) {' basically a conparison between the 2 bots of production 2010-09-12T04:55:26 how does one use that/another function to identify/compare the players production? 2010-09-12T04:58:11 i guess it sums the production of the posessed planets 2010-09-12T04:58:36 ya, and you would think (pw.Production(1) is player 1 etc, but it doesnt work for some reason 2010-09-12T05:00:05 website down again? 2010-09-12T05:00:33 *** Obeleh has joined #aichallenge 2010-09-12T05:03:04 contestbot, visualiser 2010-09-12T05:03:05 Palmik: I do not know about 'visualiser', but I do know about these similar topics: 'visualizer' 2010-09-12T05:03:06 blkperl: nope, it's up 2010-09-12T05:03:09 contestbot, canvas 2010-09-12T05:03:10 Palmik: Error: "canvas" is not a valid command. 2010-09-12T05:03:16 :'( 2010-09-12T05:03:19 contestbot: where visualizer 2010-09-12T05:03:19 amstan: visualizer = (#1) http://github.com/DanielVF/Planet-Wars-Canvas-Visualizer, or (#2) http://csclub.uwaterloo.ca/~amstan/aichallenge/canvas-visualizer/ 2010-09-12T05:03:22 amstan: just lag i guess ;) 2010-09-12T05:03:31 amstan, thanks :) 2010-09-12T05:03:59 amstan: although the compile script hasn't email me yet... 2010-09-12T05:04:00 I think he clicked a button and then messaged that once it was fixed ;) 2010-09-12T05:04:25 greghaynes: oh man.. guys in #python.. rudest thing ever 2010-09-12T05:04:37 heh, I think ive mentioned this before? 2010-09-12T05:04:42 greghaynes: you help one person, and 10 other people tell you in how many ways it's wrong 2010-09-12T05:04:48 they're always better somehow.. 2010-09-12T05:05:04 also.. fucksleep 2010-09-12T05:05:09 All they can say is "Are you using twisted?" "Why arent you using twisted?" 2010-09-12T05:05:31 so um does anyone know how to fix that small code error i said previouly? 2010-09-12T05:05:33 k.. what should i do the rest of the night? 2010-09-12T05:05:54 *** mtcbot has joined #aichallenge 2010-09-12T05:05:54 amstan: Work on the bot or xsbs ;) 2010-09-12T05:06:24 what bot? 2010-09-12T05:06:29 Your bot 2010-09-12T05:06:33 greghaynes: also.. you notice i didn't touch xsbs in a long time 2010-09-12T05:06:36 i don't have a bot 2010-09-12T05:06:41 unless you mean contestbot 2010-09-12T05:06:42 contestbot: hi 2010-09-12T05:06:42 hi amstan 2010-09-12T05:06:49 Post it on benzedrine 2010-09-12T05:06:58 so like.. 2010-09-12T05:07:02 i can just get... 2010-09-12T05:07:04 contestbot: rankings 1 2010-09-12T05:07:11 contestbot: rankings 1 2010-09-12T05:07:13 gar!! 2010-09-12T05:07:16 amstan: Error: There was a problem accessing the interface to ai-contest.com 2010-09-12T05:07:26 amstan: Error: There was a problem accessing the interface to ai-contest.com 2010-09-12T05:07:28 when the rankings appera.. 2010-09-12T05:07:49 that would require working servers ;) 2010-09-12T05:07:59 so.. i can get dmj111's entry 2010-09-12T05:08:03 in a second, lol 2010-09-12T05:08:11 If I wasnt doing well in the contest I would offer a node 2010-09-12T05:08:26 greghaynes: i can fix that for you, lol 2010-09-12T05:08:32 heh 2010-09-12T05:08:41 *** m4rw3r has joined #aichallenge 2010-09-12T05:08:54 a few sql queries.. lol 2010-09-12T05:09:03 *** sigh_ has quit IRC (Remote host closed the connection) 2010-09-12T05:09:15 oh man 2010-09-12T05:10:44 the server seems pretty fine 2010-09-12T05:10:48 cpu usage is a big problem 2010-09-12T05:11:07 mono MyBot.exe 2010-09-12T05:11:09 heheh 2010-09-12T05:11:17 I have a 4cor cpu sittine here... 2010-09-12T05:11:21 er, core 2010-09-12T05:11:26 i do too 2010-09-12T05:11:32 greghaynes: work on code to distribute it 2010-09-12T05:11:44 contestbot, issues 2010-09-12T05:11:45 amstan: issues = http://code.google.com/p/ai-contest/issues/list 2010-09-12T05:11:58 You should find a way to plug into boinc or @home 2010-09-12T05:12:14 you need kernel modules for the sandbox 2010-09-12T05:12:27 sandbox? 2010-09-12T05:12:29 Why? 2010-09-12T05:12:29 and then ppl will try to install it on windows.. 2010-09-12T05:12:32 and all that fun 2010-09-12T05:12:44 greghaynes: ... so evil bots don't escape? 2010-09-12T05:12:51 and rm -Rf / 2010-09-12T05:12:58 hrmm 2010-09-12T05:13:08 I would run in a vm 2010-09-12T05:13:18 does anyone know how to make pw.Production(1) etc work in c++? 2010-09-12T05:13:40 i want to use it as a comparison between the 2 players productin but it doesnt work 2010-09-12T05:14:10 greghaynes: perhaps i should make another channel 2010-09-12T05:14:19 *** ratatata has joined #aichallenge 2010-09-12T05:15:14 Go for it 2010-09-12T05:16:35 *** Palmik_ has joined #aichallenge 2010-09-12T05:17:15 the function is listed on the site in one of the guides but in java, but it doesnt seem to work 2010-09-12T05:18:33 anyone? page im looking at is here http://ai-contest.com/simple_strategy_guide.php 2010-09-12T05:18:53 greghaynes: k.. wtf? http://ai-contest.com/forum/viewtopic.php?f=19&t=528#p2992 2010-09-12T05:18:58 *** vikhyat has joined #aichallenge 2010-09-12T05:19:00 look at the last line 2010-09-12T05:19:00 *** Palmik has quit IRC (Ping timeout: 265 seconds) 2010-09-12T05:19:06 greghaynes: do i look like i asked something? 2010-09-12T05:19:37 huh? 2010-09-12T05:19:40 I dont get it 2010-09-12T05:19:51 *amstan's mailbox is already the proud owner* 2010-09-12T05:19:58 Yea... 2010-09-12T05:20:00 only thing i said: I think you can attach them in here just fine. 2010-09-12T05:20:01 I dont get that 2010-09-12T05:20:04 ah 2010-09-12T05:20:07 he sent me a mail.. 2010-09-12T05:20:11 with his code 2010-09-12T05:20:25 Maybe hes not so good at englush 2010-09-12T05:22:50 he probably doesn't use lunix either 2010-09-12T05:22:59 ;) 2010-09-12T05:23:08 the answer is to use github ;) 2010-09-12T05:23:18 Im not so goot at typing, I accept that 2010-09-12T05:23:35 blkperl: sure, as long as you dont want a wiki 2010-09-12T05:23:36 blkperl: nah.. just use email 2010-09-12T05:23:42 greghaynes: theres a wiki 2010-09-12T05:23:45 gollum :) 2010-09-12T05:23:55 blkperl: even better.. give me your address, i'll burn you a cd and you can load it on your pc 2010-09-12T05:24:02 you need a 2x cdrom drive 2010-09-12T05:24:07 greghaynes: http://github.com/github/gollum 2010-09-12T05:24:14 Yea, the wiki sucks so hard its just sad 2010-09-12T05:24:22 greghaynes: which wiki? 2010-09-12T05:24:24 gollum is the nw one 2010-09-12T05:24:28 github 2010-09-12T05:24:29 new* 2010-09-12T05:24:45 *** mtcbot has quit IRC (Quit: Page closed) 2010-09-12T05:24:48 amstan: We switched off that wiki purely because it sucked, if you dont remember 2010-09-12T05:24:52 greghaynes: have you checked out hthe new wiki? 2010-09-12T05:25:00 No, whats new? 2010-09-12T05:25:11 greghaynes: what were the main issues with it? 2010-09-12T05:25:14 Baphomet: I just entered and couldn't see what function you were talking about but I assume pw.Production() ? that is not implemented in PlanetWars.java, correct 2010-09-12T05:25:26 greghaynes: i don't remember, i didn't actually actively develop xsbs, just played around with irc 2010-09-12T05:25:42 greghaynes: mathematical equations, syntax highlighting, api documentation 2010-09-12T05:25:52 amstan: It was basicly a weekend project, no revisions, bare markup 2010-09-12T05:26:06 aww 2010-09-12T05:26:23 you know what i hate about github? 2010-09-12T05:26:34 this: http://github.com/amstan/hyperserv/commits/master 2010-09-12T05:26:47 it marked every commit that i have, with an invalid-email-address 2010-09-12T05:27:00 heh 2010-09-12T05:27:01 there's no way to reverse it 2010-09-12T05:27:10 it's one of their magic accounts 2010-09-12T05:27:12 Thats git, not githupb 2010-09-12T05:27:13 ratatata: yes thats the function i am talking about, i was assuming it would be in java but i was wanting it for c++ 2010-09-12T05:27:22 is there something wrong with your local git config? 2010-09-12T05:27:27 greghaynes: no.. git, those commits actually have Alexandru Stan as the name 2010-09-12T05:27:36 *** shawnshekari has joined #aichallenge 2010-09-12T05:27:36 and alex@alex-desktop.local as email 2010-09-12T05:27:45 so like.. you don't detect it.. 2010-09-12T05:27:47 fine 2010-09-12T05:27:48 leave it as is 2010-09-12T05:27:51 don't just hide it 2010-09-12T05:27:56 that might be the problem, I think you might need to have the github account email 2010-09-12T05:28:18 m4rw3r: i cannot go changing all this history for 100+ commits that i ever contributed to github 2010-09-12T05:28:21 even if not that function, i was after a way to make a comparison between the 2 players total production size 2010-09-12T05:28:25 *** Kastra has joined #aichallenge 2010-09-12T05:28:45 contestbot: topic swap 3 4 2010-09-12T05:28:46 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || http://contestbot.hypertriangle.com/ || No stupid questions. Don't ask to ask. || Developers wanted, contact j3camero. || Sorry about the slowness. || Latest Idea, decentralization: http://tinyurl.com/decentralizedai" 2010-09-12T05:28:46 yeah, that'd be quite annoying yes 2010-09-12T05:28:54 amstan: My git install alerts me if I havent set my email before I commit 2010-09-12T05:29:14 greghaynes: it does me too, but like.. i don't care, i don't want my email in there 2010-09-12T05:29:24 Baphomet: you can loop over planets, adding growthRates according to players and compare it 2010-09-12T05:29:30 *** nibalizer has joined #aichallenge 2010-09-12T05:29:41 * greghaynes waves to nibalizer 2010-09-12T05:29:53 * amstan waves to Kastra 2010-09-12T05:29:55 amstan: Then it doesnt matter 2010-09-12T05:30:01 ah well that sounds a bit complicaed for me im not too good at all this 2010-09-12T05:30:22 * Kastra waves back to amstan 2010-09-12T05:31:27 bit if i was to try would be something along the lines of looping pw.MyPlanets to get a total? 2010-09-12T05:32:14 actually would probbly be my_planets[i] 2010-09-12T05:32:18 to loop 2010-09-12T05:35:09 dam the website is lagging again 2010-09-12T05:35:15 * nibalizer waves at greghaynes 2010-09-12T05:35:30 blkperl: back up 2010-09-12T05:35:37 blkperl: was doing mysqls again 2010-09-12T05:36:25 mysqls-- then :) 2010-09-12T05:37:30 jaspervdj: Around? 2010-09-12T05:37:50 * bss03 wonders how to use the bot to send messages. 2010-09-12T05:37:56 @help 2010-09-12T05:37:56 bss03: (help [] []) -- This command gives a useful description of what does. is only necessary if the command is in more than one plugin. 2010-09-12T05:38:13 bss03: Yep, I'm here 2010-09-12T05:38:30 bss03: @later tell who message 2010-09-12T05:38:55 jaspervdj: I'm about to send a couple of pull requests. Both merge cleanly with your master on GitHub. 2010-09-12T05:39:14 cool, fire :-) 2010-09-12T05:39:23 jaspervdj: And, I think after merging those, our package will be in as good shape as the Python package. 2010-09-12T05:39:34 \o/ 2010-09-12T05:39:40 jaspervdj: So, I was wondering if there's a way to make it "official" some how. 2010-09-12T05:40:04 just zip it up with the maps and the java stuff and ping the organizers will do :-) 2010-09-12T05:40:33 Does anyone know how frequently we should see our bot matched up? 2010-09-12T05:41:54 shawnshekari: Last I heard, they were still doing random pairings. 2010-09-12T05:42:20 shawnshekari: So, it could be fast or slow, depending on your luck with random number generators. 2010-09-12T05:42:30 bss03: I'm afk, go ahead and send me the pull requests, I'll merge them later this afternoon 2010-09-12T05:42:48 jaspervdj: One already your way; the other coming soon. 2010-09-12T05:43:24 ok, thanks. 2010-09-12T05:43:52 @later tell jaspervdj I'll let you decide when the stuff is ready to be submitted to the organizers. Let me know if there's something else that needs work. 2010-09-12T05:43:52 bss03: The operation succeeded. 2010-09-12T05:43:57 *cries* i cant code for shit 2010-09-12T05:44:30 Baphomet: perhaps you should start with something easier 2010-09-12T05:44:50 Baphomet: Blah, it just takes practice, like many things. 2010-09-12T05:45:18 i have some basic c++ skils but nothing seems to be working for me 2010-09-12T05:45:33 Baphomet: Your first language is the hardest; I find it's now relatively easy for me to pick up another. 2010-09-12T05:46:04 Baphomet: Compiler errors, run-time crashes, or just trouble putting your strategy in code. 2010-09-12T05:46:30 Baphomet: I can help with the first, and *maybe* the second, the last is tricky. 2010-09-12T05:46:52 *** Palmik_ has quit IRC (Ping timeout: 240 seconds) 2010-09-12T05:47:21 *** Palmik has joined #aichallenge 2010-09-12T05:47:38 compile errors dut to inability to unserstand how all the code goes togeather. i can pretty much understand what i want to do and how i want to do it the first reason :p 2010-09-12T05:48:21 just confused what all the functions do, and how to work with them 2010-09-12T05:49:39 Ah, documentation failure. I haven't read the C++ starter code, yet. But what I saw looked like most of the functions were self-descriptive. 2010-09-12T05:50:13 *** sigh has joined #aichallenge 2010-09-12T05:50:20 It does use the STL, and some of the more interesting features of the STL can be a be hairy, especially if you haven't coded C++ in a while. 2010-09-12T05:51:05 *** shawnshekari has quit IRC (Quit: Page closed) 2010-09-12T05:51:47 yah if there was a document saying what the default inbuilt functions did it would be helpfull 2010-09-12T05:54:24 right now im just trying to implement 'if (pw.Production(1) >= pw.Production(2))' but it has to be done manually because that function doesnt actually exist :\ 2010-09-12T05:54:46 Baphomet: Yeah, I noticed that was missing. 2010-09-12T05:55:20 Baphomet: Model it after NumShips(). 2010-09-12T05:56:19 Baphomet: You don't have to iterate across the fleets in Production(), but you will basically do the same walk of the planets. 2010-09-12T05:56:31 Baphomet: You'll just add up a different field. 2010-09-12T05:57:29 HAHAHA. I didn't realize the example_bots where in Java in the C++ starter package. 2010-09-12T05:57:46 For the Haskell starter package, I ported them to Haskell. 2010-09-12T06:01:07 *** sigh has quit IRC (Ping timeout: 252 seconds) 2010-09-12T06:01:38 *** elijahbal has joined #aichallenge 2010-09-12T06:02:37 *** ath88 has joined #aichallenge 2010-09-12T06:02:52 Hello 2010-09-12T06:03:15 When running my two java bots, The program says me 'Player 1 timeout, Player 2 timeout' 2010-09-12T06:04:38 elijahbal: does it work with the default bots? 2010-09-12T06:04:59 no 2010-09-12T06:05:41 elijahbal: try this: http://ai-contest.com/forum/viewtopic.php?f=18&t=534 2010-09-12T06:05:44 I think you are not running them corrently. :( 2010-09-12T06:06:00 make sure that you're using "java BotFile" for your compiled Bot and "java -jar BotName.jar" for example ones 2010-09-12T06:09:10 *** mistabell has joined #aichallenge 2010-09-12T06:09:22 *** sigh has joined #aichallenge 2010-09-12T06:09:23 please take a look at http://lvi.toile-libre.org/data/planetwar_bug_screenshot.png 2010-09-12T06:09:38 where's the bug? 2010-09-12T06:09:48 I cannot see the fight. 2010-09-12T06:09:59 I click on play, but it doesn't work 2010-09-12T06:10:08 Nothing is moving on the screen. 2010-09-12T06:10:33 It keeps stuck on the first round, but Do not play more. 2010-09-12T06:10:49 ah that is strange. I've only played with the Java one. Python guys? 2010-09-12T06:11:06 Oh my... my new bot gets it's ass handed by my older one :'( 2010-09-12T06:11:27 mistabell: It's not a matter of python. 2010-09-12T06:11:34 I use the defaults java bots. 2010-09-12T06:11:46 Maybe it's my java openjdk 2010-09-12T06:12:04 elijahbal: If a bot times bot, PlayGame.jar removes it from the game. 2010-09-12T06:12:11 rewrite mine 2010-09-12T06:12:22 it takes forever to climb the ranks now that people are actually playing 2010-09-12T06:12:23 elijahbal: The player is operating correctly, there's no more turns to display. 2010-09-12T06:12:30 *I need to rewrite mine 2010-09-12T06:12:49 Why I do not see any fight ? 2010-09-12T06:13:11 How could I test quickly my bot ? 2010-09-12T06:13:13 elijahbal: It looks like you are calling PlayGame.jar correctly though. Those java bots must be crashing or otherwise failing somehow. 2010-09-12T06:13:46 Well, it doesn't take one minute to say me « time out » 2010-09-12T06:14:00 it's strange also. 2010-09-12T06:14:10 because the time out is one minute afaik 2010-09-12T06:14:13 elijahbal: What happens if you do ({ cat maps/map1.txt; echo go; } | java -jar example_bots/BullyBot.jar)? 2010-09-12T06:14:16 timeout is 1 second 2010-09-12T06:14:17 *** GordonQuad has joined #aichallenge 2010-09-12T06:14:37 The first "1000" on the command line means a timeout of 1000ms. = 1 sec. 2010-09-12T06:14:42 david@dkremer-laptop:~/python_starter_package$ ({ cat maps/map1.txt; echo go; } | java -jar example_bots/BullyBot.jar) 2010-09-12T06:14:44 1 15 50 2010-09-12T06:14:45 go 2010-09-12T06:14:47 david@dkremer-laptop:~/python_starter_package$ 2010-09-12T06:15:11 it seems to be working ? 2010-09-12T06:15:28 Well, that bot starts up correctly and runs, so that doesn't seem to be the problem. 2010-09-12T06:15:31 wtfbbq has a strong technique. I need to send more ships at a time 2010-09-12T06:17:03 http://ai-contest.com/visualizer.php?game_id=4443929 interesting game 2010-09-12T06:17:31 max_turn_time max_num_turns 2010-09-12T06:17:35 1000 1000 2010-09-12T06:17:49 elijahbal: How about (java -jar tools/PlayGame.jar maps/map1.txt 1000 200 PlayGame.log "java -jar example_bots/BullyBot.jar" "java -jar example_bots/BullyBot.jar") -- don't paste the output, just describe it. 2010-09-12T06:17:56 The output should be really long. 2010-09-12T06:18:34 I see again this two warning P1 timeout P2 timeout 2010-09-12T06:18:48 actually the same output than above. 2010-09-12T06:19:02 and a list of numbers after the DRAW! message 2010-09-12T06:19:05 profile_games.php doesn't list the username at the top. Probably already reported 2010-09-12T06:19:07 elijahbal: Your PlayGame.jar is broken, methinks. 2010-09-12T06:19:20 dhartmei: you underpredict the force needed to take a planet by one it seems, 2010-09-12T06:19:38 but only if it has growth 2010-09-12T06:20:07 md5sum ? 2010-09-12T06:20:11 Er, profile_games.php doesn't show it once you click More games. 2010-09-12T06:20:29 mistabell: I see some ridicilous moves by wtfbbq there 2010-09-12T06:22:10 http://ai-contest.com/visualizer.php?game_id=4438909 -- Another is example of failing to take a planet. (On purpose?) 2010-09-12T06:22:51 lol, that is interesting 2010-09-12T06:23:38 so, first it sends an amount to get it to 0, then sends an amount based on the growthrate? or something. 2010-09-12T06:24:16 bartwe: which game? 2010-09-12T06:24:47 rofl.. that bot is messed up :) 2010-09-12T06:24:52 My bot only has 2 losses, and that is one of them. 2010-09-12T06:25:26 My attempt to take his small planet fails miserably, and puts me way behind. :( 2010-09-12T06:28:02 *** sigh has quit IRC (Remote host closed the connection) 2010-09-12T06:31:05 What I have to do with a PlayGame.jar broken ? 2010-09-12T06:31:15 *** jaspervdj has quit IRC (Quit: leaving) 2010-09-12T06:33:22 *** m4rw3r has quit IRC (Quit: Page closed) 2010-09-12T06:33:47 *** blueiris has quit IRC (Ping timeout: 252 seconds) 2010-09-12T06:33:51 http://ai-contest.com/forum/viewtopic.php?f=18&t=534 2010-09-12T06:33:53 Found it. 2010-09-12T06:33:59 It works with the new one. 2010-09-12T06:36:24 *** zaph has joined #aichallenge 2010-09-12T06:36:53 dhartmei: i saw it in several games, you keep sending fleets to keep an enemy at 0, but never take the planet 2010-09-12T06:37:02 dhartmei: just look for long chains of 4's or something 2010-09-12T06:37:16 not that i want you to be stronger ofcourse :) 2010-09-12T06:37:42 *** zaph has quit IRC (Client Quit) 2010-09-12T06:39:10 I hope __init__ figures some way to break into the top 20. Haskell is being enbarassed. :( 2010-09-12T06:39:28 one of these days, I need to learn haskell 2010-09-12T06:39:36 Hey, let me recap just for see if I'm ok with the internal working of the context. 2010-09-12T06:39:42 All the game is text-file based ? 2010-09-12T06:40:02 With me as like the 2nd best ranked Haskeller, that's kinda sad. I just picked up Haskell like... this(?) January. 2010-09-12T06:40:24 And you can rewrite your Galaxy class ? 2010-09-12T06:41:05 sounds about right. haha. 2010-09-12T06:41:12 elijahbal: You can sumbit anything that the official server will compile. You can start from scratch if you want. I.e. Yes, you can modify any calss. 2010-09-12T06:41:33 yes, but eg what the lib available for python ? 2010-09-12T06:41:36 Is it just std lib ? 2010-09-12T06:41:54 kudos to the guys organizing this contest. everything has been super smooth for me 2010-09-12T06:42:05 Will the server advertise you if you give a bad file ? 2010-09-12T06:42:22 elijahbal: It's not strictly file-based though. The stuff coming from the server to your program is likely via UNIX pipe, but your program just needs to read stdin and write stdout. 2010-09-12T06:42:23 yeah, it'll bounce back 2010-09-12T06:42:28 my first 2 submissions got sent back 2010-09-12T06:42:35 *** s_denchev has joined #aichallenge 2010-09-12T06:42:54 elijahbal: You might not get an email at this point, but your profile will show that your last submission failed to compile or somesuch. 2010-09-12T06:43:30 elijahbal: If you don't get the email, I don't think you can see anything other than a pass/fail. 2010-09-12T06:44:19 elijahbal: As far as python libs, there are probably a few available. IIRC, the official game engine is written in Python. 2010-09-12T06:44:40 elijahbal: For Haskell, we just get bog-standard GHC 6.8 from Ubuntu 8.04. :( 2010-09-12T06:44:59 Hmm... rating does not reset by uploading new version? 2010-09-12T06:45:18 rating does reset 2010-09-12T06:45:28 takes a few minutes 2010-09-12T06:45:37 http://ai-contest.com/profile.php?user_id=3891 this is weird 2010-09-12T06:45:48 elijahbal: It still hasn't made it into an Ubuntu LTS, but there is an "official" set for libraries to go with GHC--called the Haskell Platform--which would make my life easier, but we don't get it on the official server. :( 2010-09-12T06:45:57 it seems that from last upload he had played only 2 games, one of which he lost nad is ranked 27th 2010-09-12T06:46:07 *and 2010-09-12T06:46:21 Palmik: it's part of the scoring, it's weird like that 2010-09-12T06:46:25 That ranking is probably only based on the first game. 2010-09-12T06:46:40 *** Utkarsh has quit IRC (Ping timeout: 252 seconds) 2010-09-12T06:46:42 but still, he only beat someone ranked 127 2010-09-12T06:46:43 The ranking update as a separate process from the games being played. 2010-09-12T06:47:12 also, can't your score go up if you lose to someone with a high score? 2010-09-12T06:47:18 and go down if you beat someone with a low score? 2010-09-12T06:47:26 The current ranking system is very... generous if you have a W/L record of 100%. 2010-09-12T06:47:49 Yes, your ranking can change if you don't even play a game. 2010-09-12T06:48:44 Cause it's a ranking. If someone's BayesianElo goes above yours, your rank increases since they jump from below you to above you. 2010-09-12T06:49:08 What will happen if you got a timeout ? 2010-09-12T06:49:29 Do you lose immediatly ? 2010-09-12T06:49:39 yes 2010-09-12T06:49:41 elijahbal: The official server is like PlayGame.jar there. It removes you from the game. 2010-09-12T06:49:53 But can you play another match ? 2010-09-12T06:50:08 elijahbal: Unless the opponent times out at the same time you'll get a loss. 2010-09-12T06:50:19 k 2010-09-12T06:50:36 elijahbal: I believe you will still get scheduled for future games. 2010-09-12T06:50:52 k 2010-09-12T06:51:31 elijahbal: There are certain conditions when a bot might stop getting scheduled at all, but that's only if the engine thinks you violated (or tried to violate) the sandbox. 2010-09-12T06:51:33 Cannot write a bot in C, huh ? 2010-09-12T06:51:48 In C++ a bot should be very fast. 2010-09-12T06:51:49 elijahbal: I bet you can. 2010-09-12T06:51:55 What do you think ? 2010-09-12T06:52:00 *** Utkarsh has joined #aichallenge 2010-09-12T06:52:14 Is really a so big problem so you need a lot of CPU to win ? 2010-09-12T06:52:14 elijahbal: But, there's no starter package for C, ATM. 2010-09-12T06:52:29 bss03: nope, but there is one for C++ 2010-09-12T06:52:39 that's not bad if you want speed. 2010-09-12T06:52:42 elijahbal: I'm not sure; my bot plays exteremely fast and is written in Haskell. 2010-09-12T06:52:53 Haskell is a compiled language ? 2010-09-12T06:53:16 Yes-ish. 2010-09-12T06:53:21 woot, my first win against hsBot10.01 ha, i guess that means i beat a haskell ot ? 2010-09-12T06:53:44 There's no "eval" or something like that, so you can definitely compile it without the binary containing a Haskell runtime. 2010-09-12T06:54:23 And it is commonly compiled. 2010-09-12T06:54:37 *** phreeza_ has quit IRC (Quit: Computer has gone to sleep) 2010-09-12T06:54:49 However, GHC also ships with an interactive evaluator that uses most of the same internals. 2010-09-12T06:55:02 omg I just got an strategy idea 2010-09-12T06:55:28 Haskell does not do incredible at the langauge shoot-out when it come to speed, but ti's not horrible either. 2010-09-12T06:55:57 It's my understanding that compiled haskell runs only slightly slower than C 2010-09-12T06:56:29 I think the main advantage to Haskell is the static typing, which make for fewer run-time errors. But, I've also used it's non-strict nature quite a lot in this compeition. 2010-09-12T06:57:09 http://shootout.alioth.debian.org/ -- High-quality language comparisions. 2010-09-12T06:58:31 mistabell, you should not get rating fro losing... 2010-09-12T06:58:53 That's a nice site 2010-09-12T06:59:05 http://shootout.alioth.debian.org/u64q/benchmark.php?test=all&lang=ghc&lang2=gcc -- Haskell vs. C 2010-09-12T06:59:45 If you don't agree that a language is "that slow" they do accept improvements/rewites for the programs being tested. 2010-09-12T07:00:10 So, if your favorite language looks too bad, get to work proving the shootout wrong! 2010-09-12T07:00:30 Heh 2010-09-12T07:00:46 weird that the haskell source code is often longer that the C one 2010-09-12T07:01:13 Palmik: When I read one of the Haskell programs it wasn't very idiomatic Haskell. 2010-09-12T07:01:56 Plamik: So instead of doing things the "Haskell way" they just found a Haskell function (sometimes from the FFI) that behaved exactly like the C function and used it. 2010-09-12T07:02:12 But, that could have changed. 2010-09-12T07:02:39 Plus, sometimes that really is the fastest way to write stuff. However, like many high-level languages,... 2010-09-12T07:02:59 dang, my bot lost again. I'm out of the 20 2010-09-12T07:03:12 experienced Haskellers recommend getting the program working first, and then optimizing as indicated by profiling. 2010-09-12T07:03:48 *** canihazbacon has joined #aichallenge 2010-09-12T07:04:11 what's the benefit to sending a stream instead of 1 large fleet? 2010-09-12T07:04:37 bss03 actually for one of the benchmarks they actually say the newest ghc is faster 2010-09-12T07:05:36 6.12 is supposed to be faster than 6.8 2010-09-12T07:05:57 I hope that the final turnament will not be ranked based on rating but rather some "everyone against everyone" or something 2010-09-12T07:06:06 Note for future reference: with GHC HEAD as of 6/13/10, 2010-09-12T07:06:06 -- works something like 3x as fast as GHC 6.12.1. Reevaluate 2010-09-12T07:06:06 -- in future! -LW 2010-09-12T07:06:10 http://shootout.alioth.debian.org/u64q/program.php?test=regexdna&lang=ghc&id=2 2010-09-12T07:06:10 Also, 7.0 is supposed to be *really* fast, rewritten around LLVM, IIRC. 2010-09-12T07:06:16 mistabel, may be this is more flexible ? 2010-09-12T07:06:50 Galcon is a terrible game with a terrible UI and terrible balance :( 2010-09-12T07:06:51 ah, that's true. I hadn't thought about that 2010-09-12T07:06:56 I just blew at least 30 minutes on it. 2010-09-12T07:07:22 Palmik: No it'll be based on rating, but there may be some tweaks in scheduling, ratings calculation, and rate of play. 2010-09-12T07:07:26 Zeiris actually the developer said the high level ai cheats 2010-09-12T07:07:31 by getting more ships 2010-09-12T07:07:44 galcon is the galcon ai in the ranking 2010-09-12T07:07:47 The maps seem balanced in its favor, too. 2010-09-12T07:08:05 Zeiris, just random the ap until it's in your favor 2010-09-12T07:08:11 Heh. 2010-09-12T07:08:31 bss03 is there a plan for 7.0 2010-09-12T07:08:37 But the AI is cheating in the highest levels... it took me like 30 tries to beat the last one 2010-09-12T07:08:47 I looked around the wiki and the llvm stuff there is from months ago 2010-09-12T07:08:59 bss03, hmm 2010-09-12T07:09:04 iFire: Yeah, what was going to be 6.14 is now 7.0, I think. 2010-09-12T07:09:32 bss03 any idea what's the best too for programically generating wrappings from c++ to haskell? 2010-09-12T07:09:40 iFire: They was to releace in time of the ICFP this year, I think. The GHC devs say that a lot though. 2010-09-12T07:09:40 tool* 2010-09-12T07:10:16 iFire: Nope. I've not yet even really ventured into the Haskell FFI, except for reading about it. 2010-09-12T07:10:20 icfp conference or contest ? 2010-09-12T07:10:31 Conference, I think. 2010-09-12T07:10:37 bss03 if you ever though of making a haskell iphone program Apple removed all unreasonable programming language restrictions 2010-09-12T07:11:04 seems like it would be a ton of work to get a haskell iphone app running 2010-09-12T07:11:08 The GHC devs usually miss the ICFP by a month or two though, as far as releases. 2010-09-12T07:11:10 llvm again 2010-09-12T07:11:27 I think there's a objective c haskell binding 2010-09-12T07:11:39 i'm a bit sad that i can't write a bot in pluk for this contest 2010-09-12T07:11:41 but all of the API stuff 2010-09-12T07:11:45 I'm not sure whether it's map layout or cheaty AI, but the bots seem to capture way more planets than actual galcon game does. 2010-09-12T07:12:02 I think they just start with more ships 2010-09-12T07:12:05 iFire: I heard. But, no, I don't want to write to the iPhone. I have an HTC Hero, and will probably continue to use Android in the future. 2010-09-12T07:12:05 compiling on their server i can unerstand, but still, it forces weird versions of compilers and such 2010-09-12T07:12:14 as in low levels they start with more ships 2010-09-12T07:12:25 *lower amout of ships 2010-09-12T07:12:40 by the way Gacon's AI is http://ai-contest.com/profile.php?user_id=3907 2010-09-12T07:12:54 Has anyone seen games where A got ahead of B in income/units, but then B regained the upper hand? 2010-09-12T07:12:58 Phil (developer of galcon) ported it 2010-09-12T07:13:01 *** ratatata has quit IRC (Ping timeout: 252 seconds) 2010-09-12T07:13:05 bartwe, they just should update, it would be enought imo 2010-09-12T07:13:14 bartwe: jmcarthur was working on trying to get the organizers to accept binary submitions. Maybe pluk (or whatever) will be available by the end. 2010-09-12T07:13:27 Zeiris: I have one... 2010-09-12T07:13:32 bss03: that would be nice 2010-09-12T07:14:01 Does the Erlang runtime + my code count as a binary? :D 2010-09-12T07:14:03 dhartmei: could the elo of players on benzedrine, that have only a few games or are old be degraded ? 2010-09-12T07:14:13 dhartmei: several single game wonders in the top10 2010-09-12T07:14:21 where's the benzedrine source again? 2010-09-12T07:14:21 I think I fall behind on this one: http://ai-contest.com/visualizer.php?game_id=4442686 2010-09-12T07:14:46 bartwe: tcp benzedrine right? 2010-09-12T07:15:01 Zeiris: I think it would be whatever you can pack under an ELF header. :P 2010-09-12T07:15:07 @tcp 2010-09-12T07:15:08 bss03: tcp = tcpserver = http://www.benzedrine.cx/planetwars/ 2010-09-12T07:15:18 haha, nice. it's got it's own command 2010-09-12T07:16:19 bss03, nice, by the end of the game my FPS starts to drop :D 2010-09-12T07:16:40 Zeiris: My bot tends to send a lot of fleets. 2010-09-12T07:17:03 Zeiris: That not necessarily bad, but I need to optimize how it sends fleets. 2010-09-12T07:17:08 hmz, intresting 2010-09-12T07:17:52 http://www.benzedrine.cx/planetwars/canvas?game_id=1284289370|bartwe|ObligatoryShampoo it looks like the winner wins because it waits for its oponent to softenup the neutrals and then take over 2010-09-12T07:18:04 Right now, it decides what planet to attack and how many ships to send and then just divides that proportionally across all my planets. :P 2010-09-12T07:19:02 Anyone care to take a look at my unordered_map problem? http://codepad.org/PMg2WSKf I'm just trying to access a value from it, yet it seems I can't do the simplest thing. I can assign it values, but once I try and access one it refuses to work. 2010-09-12T07:19:35 *** AJC has left #aichallenge 2010-09-12T07:20:23 *** ratatata has joined #aichallenge 2010-09-12T07:21:04 javagamer_Cpp: blah blah blah ... discards qualifiers means you are trying pass a "const" where you need a non-"const", usually. 2010-09-12T07:21:18 javagamer_Cpp: That's what it means here, in fact. 2010-09-12T07:21:27 hmm, is there some script you use to run your bots automatically until you stop the script... that might be handy :) 2010-09-12T07:21:48 bss03, Oh! Wow, I thought that error was near useless to me, thanks! 2010-09-12T07:22:20 javagamer_Cpp: I rarely find a compiler error I can't make some sense out of these days. 2010-09-12T07:22:22 javagamer_Cpp, what bss03 or that in const member function you are using non-const one or are changing mamber 2010-09-12T07:22:43 *member 2010-09-12T07:22:45 javagamer_Cpp: For more than a year I ported C/C++/Perl/Shell/etc. code from AIX to SLES 10. 2010-09-12T07:23:05 I'm just trying to read from the unordered_map though, does that change it somehow? 2010-09-12T07:23:25 bss03, I can see why you know a lot more about compiler errors then 2010-09-12T07:23:39 I think the problem is that ShowCloest() is const, so you can't modify member variable in it (they are "const"-ified). 2010-09-12T07:24:45 You can probably drop the "const" from ShowClosest(), but that might cascade your error. You could make the member you are touching "mutable". 2010-09-12T07:25:01 Am I modifying distances though? I'm just trying to print a value from it 2010-09-12T07:25:25 is GetPoint const? 2010-09-12T07:25:44 Oh, that could be it, checking 2010-09-12T07:25:56 Yeah, it is 2010-09-12T07:26:20 *** rogue780 has joined #aichallenge 2010-09-12T07:26:25 Looks like the operator[] you are using isn't "const" 2010-09-12T07:26:59 yes... it can not be treated as a const, because you can not know ehther it's used as lvalue or rvalue 2010-09-12T07:27:04 so that's it 2010-09-12T07:27:22 *whether 2010-09-12T07:27:47 that is why Qt containers have at() which is const 2010-09-12T07:27:48 *** gefunkt has quit IRC (Remote host closed the connection) 2010-09-12T07:27:49 Oft-times, there will be "T& operator[](stuff);" and a "const T& operator[](stuff) const;" so you can use it both ways. 2010-09-12T07:27:53 *** gefunkt has joined #aichallenge 2010-09-12T07:28:11 *** rogue780__ has quit IRC (Ping timeout: 276 seconds) 2010-09-12T07:28:14 Aha, so I just need to use another function to get the value? 2010-09-12T07:28:15 But, other times, they will give the "const"-ified one a different name. 2010-09-12T07:28:43 Like "const T& at(stuff) const;" as Plamik said. 2010-09-12T07:29:24 It might be easier to diagnose if we had the corresponding source, or at least some context. 2010-09-12T07:30:07 Man, how do I know all this stuff and still can't get a fsckin' job as a developer. 2010-09-12T07:30:35 I've gotta move away from Arkansas. :( I love my state! (Particularly, my city!) :( 2010-09-12T07:31:16 bss03, you can't get a job as a developer!? 2010-09-12T07:33:05 Nope. :( They say nothing is available. 2010-09-12T07:33:14 http://ai-contest.com/profile.php?user_id=3915 1 game 2010-09-12T07:33:16 ranked 17 2010-09-12T07:33:20 delay though. 2010-09-12T07:34:36 mistabell, this is even more interesting http://ai-contest.com/profile.php?user_id=3891 :) 2010-09-12T07:34:44 with how much rating do you start? 2010-09-12T07:35:00 javagamer_Cpp: I don't want to move. But, if you have a job I can do from home, I'm cheap. (Living expenses aren't that high here.) 2010-09-12T07:35:10 *** gefunkt___ has joined #aichallenge 2010-09-12T07:35:47 Is it possible that by winning a match I can drop more than 50 spots on the ranking board? 2010-09-12T07:35:47 javagamer_Cpp: I'll even do some travel, I just don't want to be away from home all the time. 2010-09-12T07:36:13 Obeleh: overall, after you play a few games the rating seems fair. it's just weird for the first few 2010-09-12T07:36:15 Obeleh: Unfortunately, yes. For two reasons: 2010-09-12T07:36:26 Obeleh: it can if someone you lost against also loses some games 2010-09-12T07:36:40 k 2010-09-12T07:36:58 you can submit a new bot and climb up to a higher spot than you were before in the same day if your code's better than the last one 2010-09-12T07:37:00 Obeleh: More matches means tighter Bayesian bounds on your Elo, and the current rating system is a git generous. 2010-09-12T07:37:51 Obeleh: Second, if others increase their rating, your RANKing might fall because they passed you -- which might happen even if you don't play a game. 2010-09-12T07:38:23 *** gefunkt has quit IRC (Ping timeout: 264 seconds) 2010-09-12T07:38:24 *** gefunkt___ is now known as gefunkt 2010-09-12T07:38:44 oh, nice, I won my first game over tcp 2010-09-12T07:38:52 Thanks bss03 and Palmik, got it working w/ the find method 2010-09-12T07:39:06 Another question. the previous bot I've uploaded ended up playing against only 4 better bots. The rest like 12 were all little competition. Shouldnt bots match more against other bots that are more equally matched? 2010-09-12T07:39:17 javagamer_Cpp: Sweet! 2010-09-12T07:39:42 Obeleh: That's definitely in the works. The current game scheduler is extremely simple. 2010-09-12T07:39:46 Obeleh, it's random now 2010-09-12T07:39:49 Obeleh: Random pairings. :( 2010-09-12T07:40:02 Ok thanks for the answers guys :) 2010-09-12T07:40:12 ah, I was thinking it paired you with those who hadn't had many games 2010-09-12T07:40:21 but I guess all the new signups and the random function end up doing that 2010-09-12T07:40:34 I dont know if any of the organizers are reading but... They're doing a nice job !!! :) 2010-09-12T07:41:17 Really I like building newer better versions and see it compete :D 2010-09-12T07:41:27 is there any chance of the offical round length being increased? 2010-09-12T07:41:33 200 seems pretty low to me 2010-09-12T07:41:49 my bot looses many tims due to the short round length 2010-09-12T07:41:51 Baphomet: My bot rarely goes to 200. 2010-09-12T07:42:07 Baphomet: It either wins or loses by about 150. 2010-09-12T07:42:08 I havent had a match yet where the match wasn't decided in 200 rounds 2010-09-12T07:42:09 well it always depends on the style of the bot 2010-09-12T07:42:23 a bot that loses in 200 tends to lose in 400 turns against better bots 2010-09-12T07:42:28 all the top ones use that rapid spreading method 2010-09-12T07:42:33 And when it goes to 200 you can see one or the other side being the stronger one 2010-09-12T07:42:51 Baphomet: I have seen matches between two simple-ish bots that I actually thought were decided the wrong way since it was called at 200 turns. 2010-09-12T07:43:11 *** otherAntimatroid has left #aichallenge 2010-09-12T07:43:12 200 seems low until you really start optimizing 2010-09-12T07:43:21 I agree 2010-09-12T07:43:34 lol, my opponent sent a streem of 1 ship fleets, and it couldn't capture a planet with 0 ships 2010-09-12T07:43:35 currently the rules just force everyone to use a similar method or loose 2010-09-12T07:43:42 Baphomet: But, incresing the game length means throwing more CPU at each game... which might be a problem right now. 2010-09-12T07:44:31 Baphomet: But, asking about it on the Forums shouldn't hurt. 2010-09-12T07:44:44 4:40am, time to stop playing with bots. lol. Have a great night /day you guys. Awesome competition. 2010-09-12T07:44:47 I don't think any of the organizers / admins are around right now. 2010-09-12T07:44:57 mistabell: 6:40 Here. 2010-09-12T07:45:16 *** mistabell has quit IRC (Quit: Page closed) 2010-09-12T07:45:17 mistabell: Have fun on Pascifc time! 2010-09-12T07:46:10 Baphomet: I've come back from being behind and still finsihed in under 200 turns. 2010-09-12T07:46:28 well maybe i just suck at programming 2010-09-12T07:46:42 You also need strategy 2010-09-12T07:46:49 So blame that ;) 2010-09-12T07:47:03 Baphomet: I doubt it. But, "wait" is rarely a good strategy. 2010-09-12T07:47:04 You could be a perfect programmer still :P 2010-09-12T07:47:06 i have strategy, i just dont know how to program 2010-09-12T07:47:33 i have multiple tactics i want to implement i just dont know how 2010-09-12T07:47:38 if you need some help just ask 2010-09-12T07:47:38 Baphomet: Also, keep in mind that you can all IssueOrder mutltiple times in DoTurn, if needed. 2010-09-12T07:47:57 ah intresting, i didnt know that 2010-09-12T07:47:57 Baphomet: If you watch one of my games, you'll see me sending crap everywhere! 2010-09-12T07:48:09 i always wondered how people did that 2010-09-12T07:48:50 but beware if you send more then once from the same planet, you will have to modify the issueOrder method to update the ships count on the planet 2010-09-12T07:48:52 http://ai-contest.com/visualizer.php?game_id=4442686 -- is a come-from-behind win, and I have a LOT of fleets in the air at one point. 2010-09-12T07:49:18 Palmik: I wouldn't modify IssueOrder to do that. 2010-09-12T07:49:28 I would, because it's rational :) 2010-09-12T07:49:39 Palmik: I would update the bookkeeping, just not in IssueOrder. 2010-09-12T07:49:45 It has no place in your "logic" code 2010-09-12T07:50:05 *** sigh has joined #aichallenge 2010-09-12T07:50:08 oh god, why do i have uni work, i want to keep programming :( 2010-09-12T07:50:18 Nah, IssueOrder is an input-output method. It shouldn't be responsible for updating state. 2010-09-12T07:50:44 so should not you logic :) 2010-09-12T07:50:49 *your 2010-09-12T07:50:52 *** zaphus has quit IRC (Quit: Page closed) 2010-09-12T07:51:08 antimatroid, the same reason I have jobs to apply for and various other distractions, on the bright side you have written more code than me 2010-09-12T07:51:37 Palmik: In some ways I agree with you, but I still wouldn't put it in IssueOrder. 2010-09-12T07:51:50 if you are working with the starter package, you have to update planets which are in planets_ vector which is private, so you have to do it in some PlanetWars method unless you are upto major rewrite :) 2010-09-12T07:51:59 *** smhanov has joined #aichallenge 2010-09-12T07:52:00 Palmik: I might toss it and IssueOrder in a separate function. 2010-09-12T07:52:42 Palmik: Try writing in Haskell. Varables cannot be modified. 2010-09-12T07:54:18 If you wanna enhace readability, yeah, put it in different function :) If you wanna concentrate on more important things, leave it and refactor later if needed... this is not competition of clean code after all :) 2010-09-12T07:54:34 And yes, Haskell is different, dunno what it has to do with this :P 2010-09-12T07:54:49 *** td has joined #aichallenge 2010-09-12T07:55:08 Palmik: Your issue = I can't modify private data; My "issue" = I can't modify ANYTHING. 2010-09-12T07:55:28 That is not really an issue in Haskell but rather a feature :P 2010-09-12T07:56:07 Palmik: Well, it certainly helps when reasoning about code. 2010-09-12T07:56:10 I've been meaning to learn Haskell, does it really make you a better programmer like some claim? 2010-09-12T07:56:40 Appleman1234: it's not very good yet though, I got pantsed by any of the better bots and I don't think they've done much yet either 2010-09-12T07:56:47 Palmik: Plus, I wouldn't be writing in Haskell if such things were really an issue. 2010-09-12T07:57:03 antimatroid, I haven't even finished my initial changes lol 2010-09-12T07:57:10 Palmik: I went for the major rewrite option, everything auto updates now as I parese the game state 2010-09-12T07:57:24 javagamer_Cpp: No, I don't think it makes you a better programmer, not really. 2010-09-12T07:57:24 ah doh 2010-09-12T07:57:30 i need dykstras algorithm 2010-09-12T07:57:32 ofcourse 2010-09-12T07:57:32 *** delt0r___ has joined #aichallenge 2010-09-12T07:57:35 doh doh doh 2010-09-12T07:57:47 although one had to be careful there wrt the fact that moves made from the oppenent with trip length 0 are never seen by you as fleets 2010-09-12T07:58:30 antimatroid, so if you issueOrder to send 30 ships form planet A which has 50 ships, will it autoupdate as well? :) 2010-09-12T07:58:49 I mean, if you check the size immediately after the order 2010-09-12T07:59:05 javagamer_Cpp: It does change the way you think about certain tasks. 2010-09-12T07:59:08 *** delt0r_ has quit IRC (Ping timeout: 255 seconds) 2010-09-12T08:00:01 javagamer_Cpp: Have you ever written any Scheme / Lisp / Ocaml / ML ? Anything with "first-class functions" ? 2010-09-12T08:00:31 javagamer_Cpp: Have you even experimented with the C++x0 "lambdas"? 2010-09-12T08:00:52 which are fugly :D 2010-09-12T08:00:53 I've written some very simple Haskell and Scheme 2010-09-12T08:01:12 I get some of the core ideas, but I know very little of either language 2010-09-12T08:01:13 Palmik: it does 2010-09-12T08:01:40 allthough the bot I ran some games with before was literally just using some of my information to add possible moves to a priority queue and pick the best while the move is still fesible 2010-09-12T08:01:43 nothing intelligent at all 2010-09-12T08:01:59 antimatroid, nice nice :) Well, I will see... for now, I can't say that the code in default PlanetWars.cc is my bottleneck 2010-09-12T08:02:15 it's not, but I expect it to be 2010-09-12T08:02:28 It's hard to describe exactly what I'm talking about... but occationally, I'll have an "epiphany" in Haskell. ... 2010-09-12T08:03:09 Suddenly, my brain will start using (or at least "reaching for") some of the Haskell tools automatically. ... 2010-09-12T08:04:34 That's usually when I realize (a) "Oh, I finally get that!" and (b) "Wow, this would be a lot longer in C++ / Java." 2010-09-12T08:04:49 but that doesn't really change how you look at other languages? 2010-09-12T08:04:59 Still, no langauge can save a bad programmer from themselves. 2010-09-12T08:05:25 Oh, it definitely changes how you look at new lanaguages, and it will sometimes give you insight into old languages. 2010-09-12T08:06:37 But, partially because some of the tools I am getting used to using aren't readily available, I don't think my C++ would be much different than before I learned Haskell. 2010-09-12T08:08:11 Then again, my C++ is getting somewhat rusty. My last job a PHP, and before that was "plain" C. I've probably forgetten some of the neat templating trick I learned. 2010-09-12T08:08:27 I've definitely forgetten entire sections of the STL. 2010-09-12T08:10:24 *** sigh has quit IRC (Remote host closed the connection) 2010-09-12T08:12:16 i'm still learning the basics of c++ and have no real knowledge of other languages :| 2010-09-12T08:12:39 In a match like http://ai-contest.com/visualizer.php?game_id=4445504 does that mean the enemy timed out? 2010-09-12T08:12:54 i'm not sure, i had a win and a lose of those 2010-09-12T08:13:25 It's my second orso 2010-09-12T08:13:35 Obeleh: what time was the game? 2010-09-12T08:13:46 Sep 12 06:57:55 AM 2010-09-12T08:13:48 my two happened at 12:22 and 12:23 2010-09-12T08:14:15 http://ai-contest.com/profile.php?user_id=6026 2010-09-12T08:15:04 *** vikhyat has quit IRC (Ping timeout: 245 seconds) 2010-09-12T08:15:21 *** GordonQuad has quit IRC (Ping timeout: 252 seconds) 2010-09-12T08:15:22 It probably means the server had a hiccup and didn't give one of the bots enough CPU time. :P 2010-09-12T08:15:43 i still wonder if that's what happened to me last tournament :( 2010-09-12T08:20:02 Alright, I seriously need to sleep a little bit. 7 AM to too late to be hacking on A.I. 2010-09-12T08:20:07 Have fun everyone. 2010-09-12T08:20:12 *** bss03 is now known as bss03-AFK 2010-09-12T08:20:43 bss03: I didn't sleep last night either :| 2010-09-12T08:20:53 hahaha 2010-09-12T08:21:02 i was too busy fixing up planetwars.h 2010-09-12T08:21:08 it's now 10 15pm sunday night aha 2010-09-12T08:21:10 A dedicated Botter :D 2010-09-12T08:21:54 *** mightybyte has joined #aichallenge 2010-09-12T08:22:02 still fairly uncompetitive to get in the top 50 it seems 2010-09-12T08:22:08 i hope that changes 2010-09-12T08:22:15 and soon 2010-09-12T08:22:34 is the site not loading for others? 2010-09-12T08:22:44 btw does anyone know if there is a cap on memory you can use? 2010-09-12T08:22:59 antimatroid, yep... you are not alone 2010-09-12T08:23:58 Obeleh: people literally raped it for game trees last time 2010-09-12T08:24:03 i think you are pretty safe 2010-09-12T08:24:36 k 2010-09-12T08:24:38 gotta go 2010-09-12T08:24:39 Laters 2010-09-12T08:24:56 *** Obeleh has quit IRC (Quit: Page closed) 2010-09-12T08:25:42 Hmm... much first bot seems to be much dumber comapred to my second yet he is doing better :( 2010-09-12T08:26:28 They both suck though :D 2010-09-12T08:26:43 does everyone else completely fail against ragebot on map 27? 2010-09-12T08:27:03 will try that 2010-09-12T08:27:10 *** sinan has quit IRC (Ping timeout: 245 seconds) 2010-09-12T08:27:56 *** vikhyat has joined #aichallenge 2010-09-12T08:31:25 antimatroid, I do and quite quickly :P 2010-09-12T08:32:36 how many turns? 2010-09-12T08:32:40 i can last like 88 2010-09-12T08:32:55 57 :'( 2010-09-12T08:33:34 What is special about that map, I did not look at the visualiser 2010-09-12T08:33:39 i think it's a clear example of a map with second mover advantage 2010-09-12T08:33:47 *visualization 2010-09-12T08:33:49 or probably wait to fight over the middle 2010-09-12T08:33:59 my stupid bot wins in 99 turns 2010-09-12T08:34:01 both bots start near the middle 2010-09-12T08:34:08 I see, that one 2010-09-12T08:34:12 and the ideal growth to ships to take planets are near the edge 2010-09-12T08:34:31 *** braddunbar is now known as Guest75371 2010-09-12T08:34:43 ragebot tries to capture my starting planet, but it fails 2010-09-12T08:34:59 is your bot waiting on the first move? 2010-09-12T08:35:04 no 2010-09-12T08:35:08 mine is trying to claim planets towards the outer edges 2010-09-12T08:35:13 *** braddunbar has joined #aichallenge 2010-09-12T08:35:41 morning! 2010-09-12T08:36:33 deepblue is doing great on benzedrine 2010-09-12T08:36:50 almost 300 rating lead 2010-09-12T08:38:02 my computer becomes slow when there are a lot of fleets on screen :) 2010-09-12T08:40:05 *** vikhyat has quit IRC (Ping timeout: 255 seconds) 2010-09-12T08:50:14 *** albert has joined #aichallenge 2010-09-12T08:52:15 *** vikhyat has joined #aichallenge 2010-09-12T09:07:29 *** Strepto has joined #aichallenge 2010-09-12T09:11:08 my new version lost agains RageBot in 66 moves and agaist DuelBot in 668 moves in map27 :) 2010-09-12T09:14:09 hello 2010-09-12T09:14:28 I got no ranking currently 2010-09-12T09:14:33 What the problem ? 2010-09-12T09:14:38 *** RainCT has joined #aichallenge 2010-09-12T09:14:58 antimatroid: I don't think map 27 is necessarily a 2nd mover advantage. 2010-09-12T09:16:05 In a for loop i++ and ++i are equivalent, right? The difference is only when the statement contains other stuff like array[i++] 2010-09-12T09:16:31 *** acieroid` is now known as acieroid 2010-09-12T09:16:46 mightybite: neither, i made a comment just after 2010-09-12T09:17:12 Ahh, yes. 2010-09-12T09:17:22 javagamer_Cpp: yep, basically ++i increments i before using it, i++ uses i then increments it 2010-09-12T09:17:31 if you do it in it's own line, it doesn't really matter 2010-09-12T09:17:50 i'm doing a gross pde's assignment 2010-09-12T09:17:52 oh intresting i never knew that ++i thing 2010-09-12T09:17:56 i hate applied maths, it's so... messy 2010-09-12T09:18:09 but, i dont see how it actually makes a difernc 2010-09-12T09:18:10 e 2010-09-12T09:18:32 *** ermau has joined #aichallenge 2010-09-12T09:18:35 Baphemot: it's extremely useful ;) 2010-09-12T09:18:54 antimatroid: yea, i dropped that course and went to function theory instead 2010-09-12T09:19:16 albert: i took one 6 units this semester, i'm regretting that now with ai 2010-09-12T09:19:35 does anyone know how to use the pw.Distance() function? 2010-09-12T09:19:41 in c++ 2010-09-12T09:19:49 yeah 2010-09-12T09:19:57 put in the two planet index numbers as parameters 2010-09-12T09:20:13 although i suggest you index distances between planets on the first move as they're static 2010-09-12T09:20:37 i wish i knew how to do these things 2010-09-12T09:20:51 just have a global 2 dimensional vector ;) 2010-09-12T09:21:09 i dont see how everyone understands all the functions so well 2010-09-12T09:21:21 its not even documented anywhere what everything is 2010-09-12T09:21:36 google "type c++" 2010-09-12T09:21:56 has anyone already reimplemented the game engine to c++? i asked that already a few days ago but now i am actually starting this. but maybe someone else already did it now 2010-09-12T09:21:58 once you do a programming project you are actually interested in, you start wanting to do things you can't, so you go learn how to do those things 2010-09-12T09:22:03 at least that's how it's working fo rme 2010-09-12T09:22:21 albert: i pretty much completely redid planetwars last night 2010-09-12T09:22:30 but i'm keeping that new copy private 2010-09-12T09:22:46 i dont even know how to use the functions to get planet locations tho 2010-09-12T09:23:00 because nothing is documented 2010-09-12T09:23:20 Baphemot: look in planetwars.h and it'll give you the function definitions 2010-09-12T09:23:35 i would also suggest getting rid of all privates and changing classes to structs 2010-09-12T09:23:41 but i'm a communist when it comes to programming 2010-09-12T09:23:52 antimatroid: a why did you redid it? the engine itself? still in java/python or some other lang? 2010-09-12T09:24:04 oh, not the actual engine sorry 2010-09-12T09:24:13 just couldn't stand the way planet wars was written 2010-09-12T09:24:17 *** Zaphus has joined #aichallenge 2010-09-12T09:24:32 especially having a new planet wars object created on every damn turn as well as creating planet and fleet vectors 2010-09-12T09:24:40 it did not sit well with how I wanted fleets to be done 2010-09-12T09:24:46 yea 2010-09-12T09:25:43 i could probbly work things out if i just knew what function returned planet locations :\ 2010-09-12T09:26:05 i am reimplementing the game engine now in c++ (and will publish it when i'm finished). one standalone version which is just compatible to the old and also some functions to be able to use it in my bot as some easy forward simulations 2010-09-12T09:26:06 do all the planets have preset id's or something? 2010-09-12T09:26:10 Planets have functions to return X() and Y() from memory 2010-09-12T09:26:20 but getter functions are the grosest thing ever invented 2010-09-12T09:26:29 antimatroid: full ack 2010-09-12T09:26:32 :\ why? 2010-09-12T09:26:38 Careful before you start a religious war 2010-09-12T09:26:51 ermau: you say that like I wouldn't love it :D 2010-09-12T09:27:05 Baphomot: I like to be able to just access what I want however I Want 2010-09-12T09:27:12 Publicly exposed state is the grossest thing ever ;) 2010-09-12T09:27:18 although I often end up with ridiculously complicated code as a result 2010-09-12T09:28:13 ok well i see planet can get info about planets, but how does one know how many planets there are on the current map? 2010-09-12T09:28:17 hehe, yea, i guess it maybe really is a religious war. but in languages like c++ where you could reimplement some property wrapper member, there is really no point in having stuff public in the first place instead of having such dummy trivial wrapper getters/setters functions 2010-09-12T09:28:34 albert: why would you reimplement the game engine in c++, for speed? (i have a vague desire to do so in python as well, but i'm not sure why) 2010-09-12T09:28:40 Smart compilers will inline those getters/setters anyway 2010-09-12T09:29:14 big question... do I dare upload new code that works differently, does better than my old code against what I have to test with, but on *some* maps it does worse... overall mathematically it seems better, but I risk losing my hard earned 15th place, I'll bet it will take a while to get that back even if I resubmitted the same bot :-) 2010-09-12T09:29:18 braddunbar: my bot is in c++ and i want to do some explorations on the game tree 2010-09-12T09:29:53 Zaphus: don't worry about your place before the final tournament 2010-09-12T09:30:00 albert: ah, ok 2010-09-12T09:30:02 *** sigh has joined #aichallenge 2010-09-12T09:30:04 albert: good luck with that 2010-09-12T09:30:05 Oh wow, we have until the end of November? *phew* 2010-09-12T09:30:34 antimatroid: oh, I'm not worried about it in that sense, it's purely egotistacal at this point :-) 2010-09-12T09:30:59 yes, but the fact that my bot is sitting in 40 atm says a lot about how poor they are 2010-09-12T09:31:26 There are a _lot_ of starter bots running sadly 2010-09-12T09:31:31 it is very poorly adding possible fleets to a priority queue with a very bad payoff function 2010-09-12T09:32:34 i look forward to when people actually start implementing stuff 2010-09-12T09:33:16 most people have spent a lot less time looking at the problem than those in the IRC :-) apparently some of them even have lives ! 2010-09-12T09:33:42 what is that? 2010-09-12T09:33:43 Or they followed the tutorial, submitted the starter bot, realized they don't have a clue how to do this and moved on 2010-09-12T09:34:14 Current Rank: N/A. No ranking available 2010-09-12T09:34:15 they will remove just starter package bots eventually 2010-09-12T09:34:21 you wont have to deal with them in the final tournament 2010-09-12T09:34:31 i may just give up, i cant really program anything 2010-09-12T09:34:43 Baphomet: if you play around for a bit you will get going 2010-09-12T09:34:47 antimatroid: Yeah, but they're taking up rounds _now_ 2010-09-12T09:34:49 then it really is a lot of fun 2010-09-12T09:34:56 antimatroid: Guess that's why the TCP server is best 2010-09-12T09:35:01 antimatroid: how will they know which ones are starter bots ? with all the different variants of the way the tutorial could be typed in (including the fact that the C++ tutorial doesnt even work) 2010-09-12T09:35:06 ive been trying for 2 days and i still struggle to defeat the default bot 2010-09-12T09:35:15 ermau: tcp is best because you get lots of games against better bots 2010-09-12T09:35:17 thats how bad i am 2010-09-12T09:35:22 antimatroid: Yeah, no kidding 2010-09-12T09:35:41 antimatroid: In the competition, I kept winning and winning.. tcp server I kept losing and losing 2010-09-12T09:35:44 Zaphus: no idea, but the seemed to do a decent job of it last time 2010-09-12T09:35:56 antimatroid: Though my bot has a couple big bugs, haven't gotten around to rewriting and taking a new approach 2010-09-12T09:36:05 ermau: a lot of the top competitors from last time have already got their first play arounds up on there 2010-09-12T09:36:10 i wouldn't take that too much to heart 2010-09-12T09:36:14 I guess any bot submitted only once, right at the start, is a good clue - although alberts has done well :-) 2010-09-12T09:36:44 antimatroid: Oh I'm not 2010-09-12T09:36:50 what's all this talk about a tcp server? 2010-09-12T09:37:00 http://www.benzedrine.cx/planetwars/ 2010-09-12T09:37:03 antimatroid: I'm just pointing out how worthless the current entries in the real site are right now 2010-09-12T09:37:20 yeah, it's going to be like that most of the contest unfortunately 2010-09-12T09:37:34 you will get best games on tcp, but it's still cool to have the final tournament 2010-09-12T09:37:38 I think some of these guys just have scripts running their bot on the tcp constantly 2010-09-12T09:37:43 i wish they would run a final round robin on top x bots 2010-09-12T09:37:48 they do 2010-09-12T09:38:14 :\ im almost rank 500 2010-09-12T09:38:25 I told myself if I get a couple features done on my main project that I can work on the AI bot :P 2010-09-12T09:38:26 ermau: a lot of people will generate stats off those games to help improve their bots too 2010-09-12T09:38:30 i swear my bot is getting worse 2010-09-12T09:38:33 once I submit the new bot, the old playbacks become very hard to find, right ? 2010-09-12T09:38:34 antimatroid: Right, tis what I figured 2010-09-12T09:38:45 yep, although can be done 2010-09-12T09:38:47 antimatroid: I'm not expecting anything, personally, it's just fun as hell. 2010-09-12T09:38:58 Why I'm not ranked ? 2010-09-12T09:39:04 i feel like i need to make up for last time, otherwise i wont be satisfied 2010-09-12T09:39:04 my bot is too bad ? 2010-09-12T09:39:16 elijahbal: when did you submit your bot? 2010-09-12T09:39:18 elijahbal: takes a while for the server to allocate you a game 2010-09-12T09:39:23 antimatroid: I wasn't around the last time, found out via an email from GalCon 2010-09-12T09:39:46 i'm a little worried the bots might end up too good in this 2010-09-12T09:39:48 does the server notify you of matches, etc... ? 2010-09-12T09:39:50 antimatroid: Last AI competition I looked at was for StarCraft, and that's a little over my head 2010-09-12T09:39:52 and it's a mass of draws 2010-09-12T09:39:53 but we'll see 2010-09-12T09:40:03 because i did not get any mails... 2010-09-12T09:40:16 i would consider infinite mario ai comp if it wasn't like all set up with java 2010-09-12T09:40:21 even though this is too, ergh 2010-09-12T09:40:22 Your ranking is deleted when you are beginning with a new bot ? 2010-09-12T09:40:27 yep 2010-09-12T09:40:59 website down again, or incredibly slow ? 2010-09-12T09:41:18 damn, the page appeared right when I pressed enter :-) 2010-09-12T09:41:55 *** xkpe has joined #aichallenge 2010-09-12T09:42:15 *** mightybyte has left #aichallenge 2010-09-12T09:42:15 *** bobng has joined #aichallenge 2010-09-12T09:42:37 rankings are offline? 2010-09-12T09:43:12 yes 2010-09-12T09:44:05 i finally got an email to activate my accout but when i press the link i get "Failed to activate the account. (102)" 2010-09-12T09:44:22 its because you dont exist 2010-09-12T09:44:51 *** Zeiris has quit IRC (Ping timeout: 265 seconds) 2010-09-12T09:45:04 damm it i know it :( 2010-09-12T09:45:20 does it mean i have to register again? 2010-09-12T09:50:52 *** phreeza has joined #aichallenge 2010-09-12T09:51:08 Ok, new code uploaded... 72 hours since the last one was up. will be interesting to see how it goes 2010-09-12T09:53:24 wow so many bots now... 2010-09-12T09:53:46 a lot of them are just the starter pack 2010-09-12T09:54:05 good for me cuz they are the only ones i can beat >.< 2010-09-12T09:54:13 *** anoek has joined #aichallenge 2010-09-12T09:58:09 Wish there was a "detailed" leaderboard like the one on benzedrine, with wins/losses number of games, etc 2010-09-12T09:58:57 i did ask for it, i think they were going to put it into profiles at least 2010-09-12T09:59:02 but seems to have been forgotten 2010-09-12T09:59:26 well, we wish for a stable fast webserver first I guess 2010-09-12T09:59:50 ^ 2010-09-12T10:00:14 *** td has quit IRC (Quit: Page closed) 2010-09-12T10:00:46 lol 2010-09-12T10:00:51 priorities 2010-09-12T10:04:38 site down again? 2010-09-12T10:05:07 http://www.ai-contest.com/forum/viewtopic.php?f=18&t=539 2010-09-12T10:05:11 i think that's actually a pretty good idea 2010-09-12T10:06:15 site comes up, forums go down. 2010-09-12T10:07:13 lol 2010-09-12T10:07:32 and it's officialyl monday 2010-09-12T10:07:43 gah, i don't foresee much sleep in the next few months 2010-09-12T10:08:37 antimatroid, sleep is important, I say as I type up coverletters instead of sleeping 2010-09-12T10:08:59 sleep is a pandemic that needs curing 2010-09-12T10:09:01 ah crap, the weekend is over... where'd that go? 2010-09-12T10:09:03 *** jmreardon has joined #aichallenge 2010-09-12T10:09:08 don't get me wrong, I love it, but I wish I didn't need so damn much of it 2010-09-12T10:10:04 I was averaging 2 games per hour over the last 8 hours... resubmitted, lets see if I have those sort of figures in 8 hours time or not... sleep calls - g'night all ! 2010-09-12T10:10:16 night 2010-09-12T10:10:25 antimatroid, I am lucky like that, I have less sleep waves (delta waves) in brain in comparison to normal people 2010-09-12T10:10:56 *** AndroUser has joined #aichallenge 2010-09-12T10:11:05 my bot sucks and i get heaps of games and loose em all 2010-09-12T10:11:10 so I am insomniac / nightowl by design 2010-09-12T10:11:23 i mostly win when its a default bot or their bot crashes 2010-09-12T10:11:26 *** Doctor has joined #aichallenge 2010-09-12T10:11:34 baphoment: what is your bot doing? 2010-09-12T10:11:46 nothing really i cant really program too well 2010-09-12T10:11:49 i could offer you some suggestions, but i'm not willing to really share what I intend to implement for my final bot 2010-09-12T10:12:07 i know how i would make it good, i just dont know how to implement anything 2010-09-12T10:12:10 does anyone have a script for windows to run a bot against all the available bots on all the maps...? 2010-09-12T10:12:11 try considering what the number of fleets will be at another players planet in the time it takes your planet to send fleets there 2010-09-12T10:12:16 and send the amount required to take it etc. 2010-09-12T10:12:34 do you know how vectors work? 2010-09-12T10:12:38 How long does a program have each turn before it times out? 2010-09-12T10:12:38 Doctor, a few people do, they aren't me 2010-09-12T10:12:39 they are the most wonderful thing 2010-09-12T10:12:44 Is there a more detailed problem description anywhere? I have a few questions, like, what is the travel speed for the fleets, are the distances rounded or truncated? 2010-09-12T10:12:44 1 second 2010-09-12T10:12:48 nop. i only have minimal c++ experance 2010-09-12T10:12:59 xkpe: there is one in the works, not sure of the link 2010-09-12T10:13:02 Baphomet, then use another language 2010-09-12T10:13:05 jmcarthur: you about? 2010-09-12T10:13:13 i know more c++ than anything else. 2010-09-12T10:13:21 i just dont know much about programming 2010-09-12T10:13:38 well still python would be easier baphomet 2010-09-12T10:13:41 xkpe: http://code.google.com/p/ai-contest/wiki/GameSpecification 2010-09-12T10:13:44 maybe 2010-09-12T10:13:47 Baphomet, can you write what you want your bot to do in english ? 2010-09-12T10:13:50 i will try it 2010-09-12T10:14:06 i guess so 2010-09-12T10:14:08 Baphomet, if you can write your strategy in english or your native language 2010-09-12T10:14:11 like psudocode? 2010-09-12T10:14:12 if you are already somewhat comfortable with c++ i don't think changing to python is going to help 2010-09-12T10:14:24 Baphomet, like english 2010-09-12T10:14:24 C# maybe 2010-09-12T10:14:34 *** Zaphus has quit IRC (Quit: Page closed) 2010-09-12T10:14:42 i dont think c# is really any easier than c++ 2010-09-12T10:14:44 Baphomet, write it in forms you understand, then convert it to pseudocode 2010-09-12T10:14:58 then worry about the language implementation 2010-09-12T10:15:21 i've never properly sat down and used anything but c++ 2010-09-12T10:15:32 * Appleman1234 knows lots of languages 2010-09-12T10:15:36 sit* 2010-09-12T10:15:59 well ive heard before that python is easy so i may give that a look quickly 2010-09-12T10:16:03 C,C++,Python, Scheme / Racket, Javascript, C#, Java, Perl,etc 2010-09-12T10:16:07 I "know" C++, I'm just not comfortable with it and I don't like it. 2010-09-12T10:16:29 every time i use c++ i learn something new 2010-09-12T10:16:30 Baphomet, what parts of the c++ are you having issues with ? 2010-09-12T10:16:44 C# is my typical weapon of choice 2010-09-12T10:16:49 knowing what all the functions do 2010-09-12T10:16:57 and how to work with the functions 2010-09-12T10:17:09 Baphomet, just read the PlanetWars.h 2010-09-12T10:17:17 and correct syntax/usage/implementation 2010-09-12T10:17:20 your DoTurn function in planetwars.cc takes in a PlanetWars object 2010-09-12T10:17:28 that contains a vector of fleets and a vector of planets 2010-09-12T10:17:33 is that new? 2010-09-12T10:17:47 vikhyat, ty 2010-09-12T10:17:50 is a vector similar to an array? 2010-09-12T10:18:02 yes, except it's size is dynamic 2010-09-12T10:18:12 without having to worry about all the dynamic memory crap 2010-09-12T10:18:26 TBH, if you don't know what a vector is, I say you really don't know C++. 2010-09-12T10:18:33 webservr is being all flaky again 2010-09-12T10:18:43 ermau: he said he knows c++ more than anything else, not that he knows c++ 2010-09-12T10:18:51 i know, i never said i was any good at c++, i have basic skils at best 2010-09-12T10:19:15 just played round making a few console based programs etc 2010-09-12T10:19:21 nothing really complicated 2010-09-12T10:19:34 Sorry, missed the distinction 2010-09-12T10:19:36 baphomet, i would suggest playing around with console apps again to work out how vectors work 2010-09-12T10:19:38 Baphomet, your bot won't be really complicated either 2010-09-12T10:20:24 Baphomet, also see http://cplusplus.com/ and http://cprogramming.com 2010-09-12T10:20:33 the second site has a c++ tutorial 2010-09-12T10:20:42 antimatroid: yea i have done some small console programs. Appleman1234: yea i know i just dont know how to work with this stuff too well 2010-09-12T10:21:02 k i will take a look at those 2010-09-12T10:21:04 Baphomet, practice makes perfect, the compiler will tell you if you are wrong 2010-09-12T10:21:15 hehe....yea...it does.... 2010-09-12T10:21:39 whenever I don't know how to use something in c++, google "vector c++" and it'll show up one of the c++ reference page telling you the different functions it has and variabls etc. 2010-09-12T10:21:52 errors the compiler tells you are easy to fix for the most part 2010-09-12T10:21:54 Appleman1234: not if it's semantically wrong :P 2010-09-12T10:22:18 antimatroid, yes and no, depending on your setup 2010-09-12T10:22:33 but no need to debate here 2010-09-12T10:22:55 it was not intended to form a debate, just a pokey comment :) 2010-09-12T10:22:59 and the googling the error strategy also works for the most part 2010-09-12T10:23:17 do the hokey pokey and you turn about 2010-09-12T10:23:26 and that's what it is all about 2010-09-12T10:24:04 *** sigh has quit IRC (Remote host closed the connection) 2010-09-12T10:24:21 I could write some tutorials / strategy ideas on the site or the wiki, but not sure if I will ever finish my bot if I start that ... 2010-09-12T10:25:35 Baphomet: take a read of this page http://code.google.com/p/ai-contest/wiki/GameSpecification 2010-09-12T10:25:37 wee im happy... ive made it to 7th spot 2010-09-12T10:26:13 you're wtfbbq? 2010-09-12T10:26:15 fuuuuu 2010-09-12T10:26:20 obeleh 2010-09-12T10:26:30 oh, that's alright 2010-09-12T10:26:35 wtfbbq kept beating me in tcp before 2010-09-12T10:26:51 im on my phone, thats why my name is androuser 2010-09-12T10:27:25 antimatroid: thankyou that looks helpfull 2010-09-12T10:30:02 lol 2010-09-12T10:31:06 *** gilaniali has joined #aichallenge 2010-09-12T10:32:22 I want to reimplement a lot of the core java starter pack. Like the std reading in, the fleet object etc. 2010-09-12T10:32:52 does the server just compile everything in the zip and run MyBot.class? 2010-09-12T10:32:54 deslt0r___: what's stopping you? 2010-09-12T10:32:55 *** sinan has joined #aichallenge 2010-09-12T10:32:56 yep 2010-09-12T10:32:59 If I have something like an else statement with only a comment in it, will the compiler just remove it entirely? 2010-09-12T10:33:01 sweet 2010-09-12T10:33:13 you need to have the mybot.java or whatever though 2010-09-12T10:33:31 the server screamed at me for uploading mine with bot.cc last night 2010-09-12T10:33:34 antimatroid: All as java src... not classes 2010-09-12T10:34:01 I assume to scan for FileOutputStream and File().delete() :) 2010-09-12T10:34:10 delt0r: sorry I don't really know anything about java 2010-09-12T10:34:21 well I will try it and see 2010-09-12T10:34:28 but provided you meet certain standards, you can rewrite everything 2010-09-12T10:35:07 *** edcba has joined #aichallenge 2010-09-12T10:35:18 The provided java... is well very C --and quite slow compared to what it could be 2010-09-12T10:36:13 I felt the same with the cpp one and redid a lot of that 2010-09-12T10:36:25 its just an example 2010-09-12T10:36:35 Yea 2010-09-12T10:37:07 what's the easiest way to print out stuff while running your bot? 2010-09-12T10:37:13 And its one way to document the interface... But i think a description of the input output would be better 2010-09-12T10:37:13 output files 2010-09-12T10:37:45 http://code.google.com/p/ai-contest/wiki/GameSpecification 2010-09-12T10:37:50 try that delt0r 2010-09-12T10:37:50 *** Strepto has quit IRC (Quit: Page closed) 2010-09-12T10:38:10 *** AndroUser has quit IRC (Quit: Bye) 2010-09-12T10:38:12 *** Strepto has joined #aichallenge 2010-09-12T10:38:28 thanks for the tip 2010-09-12T10:38:29 Now why isn't that on the getting started page :/ 2010-09-12T10:38:47 it's still in the works i think 2010-09-12T10:38:51 antimatroid: Thanks --thats exatlly what i wanted 2010-09-12T10:40:20 *** vikhyat has quit IRC (Ping timeout: 265 seconds) 2010-09-12T10:40:53 antimatroid: yes? 2010-09-12T10:44:05 were you just asking so i could link to the spec? 2010-09-12T10:44:59 yeah, nevermind :) 2010-09-12T10:45:08 *** cobracom has joined #aichallenge 2010-09-12T10:45:11 although i just got a lot of appreciation that you should have got 2010-09-12T10:45:15 so consider this me passing that on 2010-09-12T10:45:24 i haven't received a ton of feedback on it. i'm going to assume that means it's good and write up a patch for the main site 2010-09-12T10:45:28 okay cool :D 2010-09-12T10:46:00 read up like 10 lines and you'll see someone questioning why it's not linked off the main page 2010-09-12T10:48:08 *** Shachar has joined #aichallenge 2010-09-12T10:51:59 *** vikhyat has joined #aichallenge 2010-09-12T10:52:34 *** caytchen has joined #aichallenge 2010-09-12T10:52:59 hi ! 2010-09-12T10:53:08 i have some questions : 2010-09-12T10:53:12 shoot 2010-09-12T10:53:18 bbl 2010-09-12T10:53:24 is a turn is time limited ? 2010-09-12T10:53:30 yes, one second 2010-09-12T10:53:39 ok 2010-09-12T10:53:51 if you exceed that you get disqualified, so be careful 2010-09-12T10:53:58 are maps randomly generated during contest ? 2010-09-12T10:54:13 no, but they aren't necessarily all going to be the same set you started with 2010-09-12T10:54:43 besides, you want bots playing on the same maps 2010-09-12T10:54:54 that means some are in the set some aren't ? 2010-09-12T10:54:58 i would personally like a fresh set of maps used on the final tournament, but i wouldn't count on it 2010-09-12T10:55:20 no, that means you shouldn't count on it being either case 2010-09-12T10:55:27 ok lol 2010-09-12T10:55:32 but it could be a combination of the two etc. etc. 2010-09-12T10:57:36 wow, i won a game again a1kon-0.2 2010-09-12T10:57:49 http://www.benzedrine.cx/planetwars/canvas?game_id=1284302135|bartwe|a1k0n-0.2 2010-09-12T10:58:02 bartwe: have you been improving your bot? 2010-09-12T10:58:09 definitly 2010-09-12T10:59:00 looks like you locked him to that specific planet heh. 2010-09-12T10:59:15 *** Appleman1234 has quit IRC (Ping timeout: 245 seconds) 2010-09-12T10:59:21 bartwe: that's a visually interesting strategy you have 2010-09-12T10:59:37 thanks 2010-09-12T11:00:16 damn this is fun 2010-09-12T11:00:17 trying to make a tree like structure that feeds into ever greater planets with more and more clout 2010-09-12T11:00:46 kind of looks like a minimal spanning tree? 2010-09-12T11:00:57 its not, but should be 2010-09-12T11:00:57 *** Strepto has quit IRC (Quit: Page closed) 2010-09-12T11:01:10 well, it looks bidirectional, also 2010-09-12T11:01:40 i found that feeding through the intermediate planets allows me more flexibility to redirect and protect other nodes 2010-09-12T11:02:07 bartwe: i'd be careful about times in which you need to fight perfectly 2010-09-12T11:02:08 i think i can guess why that's working so well 2010-09-12T11:02:22 antimatroid: what do you mean ? 2010-09-12T11:02:31 whistles 2010-09-12T11:02:45 *** Strepto has joined #aichallenge 2010-09-12T11:02:50 i suspect it's confusing opponents that only look at what planets can immediately send to points of conflict and are not really paying attention to planets that will be able to send out forces on the next couple turns 2010-09-12T11:02:57 *** Appleman1234 has joined #aichallenge 2010-09-12T11:03:34 and so they are attacking with insufficient forces 2010-09-12T11:03:39 one of the future extensions i'm planning is delaying attacks 2010-09-12T11:03:55 alot of attacks use just enough power 2010-09-12T11:03:58 like my own 2010-09-12T11:04:08 which allows me to send just enough defense mostly 2010-09-12T11:04:37 having all planets do a synchronized attack on a single point would be fun :) 2010-09-12T11:04:58 :D i found some of my old c++ programs i write a while back 2010-09-12T11:05:06 maybe the last but one turn ? 2010-09-12T11:05:07 sadly i dont think it will help me much :\ 2010-09-12T11:05:08 Is there a maximum number of planets on any given map? 2010-09-12T11:05:26 anoek: is there a minimum ? 2010-09-12T11:05:35 bartwe: 2? 2010-09-12T11:05:47 2 is kinda of boring 2010-09-12T11:05:51 lol 2010-09-12T11:05:51 i'd say so 2010-09-12T11:06:12 3 won't be less boring i guess 2010-09-12T11:06:25 i think 3 could be a good degenerate map for testing 2010-09-12T11:06:34 i was going to say, if you want a fun afternoon with pen and paper, sit down with the 3+ planet case 2010-09-12T11:06:35 4 maybe better 2010-09-12T11:06:50 above 4 they start to blur in my head 2010-09-12T11:07:02 the number of variables is insane 2010-09-12T11:07:06 yeah even 3 can be a mind job if you look at the game theory behind it 2010-09-12T11:07:21 this is a great game for an AI contest 2010-09-12T11:07:38 i'm still unsure how well people will get to approximating perfect play 2010-09-12T11:07:41 so no official maximum yet? 2010-09-12T11:07:43 i hope it doesn't get too bad 2010-09-12T11:07:48 anoek: nope 2010-09-12T11:07:50 okie 2010-09-12T11:07:52 is the website down again ? 2010-09-12T11:08:02 probbly 2010-09-12T11:08:04 The DB is getting destroyed 2010-09-12T11:08:06 antimatroid: "too bad" == too close to perfect or too far from perfect? 2010-09-12T11:08:11 Gave me a too many connections error a second ago 2010-09-12T11:08:16 yeah me too 2010-09-12T11:08:23 we really need to get those bots on other machines now 2010-09-12T11:08:32 actually 2010-09-12T11:08:48 the TM doesn't really use much db anymore, does it? 2010-09-12T11:08:58 jmcarthur: i hope the top bots don't end up just drawing all the time 2010-09-12T11:09:03 antimatroid: ah 2010-09-12T11:09:15 but so far it looks like the games diverge quite well to a clear winner 2010-09-12T11:09:18 is anyone else seeing the site is down right now? 2010-09-12T11:09:19 i am leaning toward the top bots actually having more variety in this one than in the tron one 2010-09-12T11:09:26 j3camero: db connections 2010-09-12T11:09:27 but depends how well people penetrate near perfect play (or even perfect) 2010-09-12T11:09:32 antimatroid: if you have a near-perfect player, please put it up against others at benzedrine 2010-09-12T11:09:34 j3camero: yep 2010-09-12T11:09:37 j3camero: find a way to lower them or close them more quickly 2010-09-12T11:09:56 bartwe: i don't, although i am not ruling out trying that as an option 2010-09-12T11:10:02 although that is not the aim is it not :P 2010-09-12T11:10:11 antimatroid: perfect play seems very hard to approach considering the variables involved, though 2010-09-12T11:10:22 hence the near perfect 2010-09-12T11:10:29 heance the "approach" 2010-09-12T11:10:33 *hence, even 2010-09-12T11:10:41 *** Appleman1234 has quit IRC (Ping timeout: 252 seconds) 2010-09-12T11:11:05 i think another obstacle to 'perfect play' is that it is dependent on the opponent 2010-09-12T11:11:17 bartwe: we are defining perfect play slightly differently here 2010-09-12T11:11:26 j3camero: a lot of bots are just starter packages getting games, is there a way to stop that? 2010-09-12T11:11:26 so this is closer to a nash equilibrium that solving a math problem 2010-09-12T11:11:42 jmcarthur: huh ? 2010-09-12T11:11:56 playing a nash equilibrium could somewhat be considered perfect play 2010-09-12T11:11:57 bartwe: perfect play in a simultaneous play game is assigning the exact best *probabilities* to each possible move 2010-09-12T11:12:30 jmcarthur: which is uncalculable if there are multiple nash peaks 2010-09-12T11:12:30 right, the nash equilibrium is almost certainly going to be mixed strategies 2010-09-12T11:12:53 or very boring rock/paper/shotgun solutions 2010-09-12T11:13:28 that just means that two perfect players will converge to a draw over an infinite number of games 2010-09-12T11:13:35 not necessarily uncalculable, although it is for this game and time constraints 2010-09-12T11:13:54 the existence of multiple nash equilbriums is not very significant 2010-09-12T11:14:01 i'm intending to do a small thesis next year on calculating mixed strategy nash equilibria in n-player finite normal form games 2010-09-12T11:14:05 danielfvf: if you're around, I am going to try restarting the mysql server to see if the site will come back up. 2010-09-12T11:14:11 antimatroid: its not calculable if you need to include the precise probs of other players 2010-09-12T11:14:24 it just means that perfect play may not be very reliable 2010-09-12T11:14:37 i think trying to calculate anything like that for this is not going to get someone very far though 2010-09-12T11:14:39 but i could be wrong 2010-09-12T11:14:44 bartwe: perfect play would assume that the opponent is also perfect 2010-09-12T11:14:48 j3camero: have you some slow queries running in mysql? or is it just too many visitors? 2010-09-12T11:15:00 jmcarthur: collections that contain themselves 2010-09-12T11:15:03 bartwe: there are known algorithms for this, they just have terrible complexities 2010-09-12T11:15:18 bartwe: i'm sorry, i don't see how russel plays into this 2010-09-12T11:15:25 *** rwa has joined #aichallenge 2010-09-12T11:15:33 Not really sure what's wrong right now. I just got out of bed. I am still poking around. 2010-09-12T11:15:38 this isn't about a system describing itself 2010-09-12T11:16:23 whats a website i can dump text to? 2010-09-12T11:16:24 I am actually going to try shutting down apache, killing all the remaining connections, then restarting. Hopefully the site won't tank immediately afterwards. 2010-09-12T11:16:26 bartwe: my point is you don't *have* to find all nash equilbriums. they are all equal. just find one and hope your perfect opponent picks a different one 2010-09-12T11:16:29 j3camero: wasn't a connection limit of apache increased yesterday? mysql connection limit must be higher than that + non-apache connections 2010-09-12T11:16:43 well, the hope is unnecessary, my bad 2010-09-12T11:16:49 *just pick one. 2010-09-12T11:16:49 Baphomet: http://pastebin.com/ 2010-09-12T11:17:01 thanks, thats the one i was thinking of 2010-09-12T11:17:15 for some reason i must be googling all the wrong words 2010-09-12T11:18:15 It's currently floating around the maxclients limit for apache. This is probably the problem. 2010-09-12T11:18:29 j3camero: restarting is just a temporary sollution with many unhappy users 2010-09-12T11:18:43 how about using lighty? 2010-09-12T11:18:46 anyway antimatroid i was saying earlier i have alil c++ experance, i found one of my programs i wrote about 2 years ago. http://pastebin.com/qg3yAn8w 2010-09-12T11:19:16 obviously not that complex 2010-09-12T11:19:41 should i just give up on this ai contest or do i have a chance of making something decent? 2010-09-12T11:19:55 j3camero: can you pastebin output of SHOW FULL PROCESSLIST; 2010-09-12T11:20:13 Baphomet: it's up to you 2010-09-12T11:20:16 Baphomet: it's definitely worth a try, I'm not a programmer by trade and I've gotten off to a decen start 2010-09-12T11:20:38 managed to beat MyBot with a few hours orientation 2010-09-12T11:20:44 omg i can hear something and i have a feeling it may be one of my hdds breaking >.< 2010-09-12T11:20:57 keep hearing a clicking 2010-09-12T11:21:03 i will be rage if i loose a hdd 2010-09-12T11:21:06 confirm it 2010-09-12T11:21:42 its would be hard to know what one when i have 4 all next to each other 2010-09-12T11:21:49 *** Mathnerd314 has joined #aichallenge 2010-09-12T11:22:41 canihazbacon: there is a push for nginix. not sure what the status is on that 2010-09-12T11:23:02 sif apache works fine 2010-09-12T11:23:10 just need a good webserv 2010-09-12T11:23:42 Baphomet: I'm probably not really the right person to ask, but I don't see why you couldn't get at least some kind of strategy going, learn a whole lot and have fun while you're doing it 2010-09-12T11:24:06 apache isn't as easy to configure for heavy loads. it can be done, but some of the other web servers are nicer out of the box 2010-09-12T11:24:32 Baphomet: quite sure you won't win but at least you'll learn some while attempting at it 2010-09-12T11:24:33 worst comes to worst you give up right? 2010-09-12T11:24:52 Baphomet: do this for learning, not for winning. that's all most of us are going to get anyway 2010-09-12T11:24:52 yah i never planned to win, just want to get some cool code going :) 2010-09-12T11:25:11 Baphomet: I won't win... but it will be fun ! 2010-09-12T11:25:14 i am looking for at least top 50, but not expecting to win 2010-09-12T11:25:16 would be nice though 2010-09-12T11:25:35 yeah i feel like i could have gotten top 50 last time if i had spent maybe 4 more hours on my bot :\ 2010-09-12T11:25:42 so when the site is up can you see previous matches and where ? :) 2010-09-12T11:25:45 but instead i only had this half broken thing uploaded 2010-09-12T11:25:55 antimatroid: Same here 2010-09-12T11:26:15 top 10 and i'll be stoked 2010-09-12T11:26:22 i almost uploaded a different version of my bot at the last hour and i'm so glad i didn't 2010-09-12T11:26:27 heh 2010-09-12T11:26:36 a1k0n: why not? 2010-09-12T11:26:36 i wish i had 2010-09-12T11:26:43 antimatroid: because he won? 2010-09-12T11:26:54 and the other bot's improvement was questionable 2010-09-12T11:27:26 question regarding nash equilibria: even if two players are using stuck in a nash equilibrium, that could still mean (in context of the contest) that one player could have a higher net ship growth per turn, and ultimately win the game? 2010-09-12T11:27:27 oh, i thought he was referring to like an hour ago :P 2010-09-12T11:27:33 oh haha 2010-09-12T11:27:43 yeah the last hour of the last contest, sorry 2010-09-12T11:27:56 Strepto: yes 2010-09-12T11:27:59 btw can we expect same symmetry in all maps ? 2010-09-12T11:28:05 good, thanks 2010-09-12T11:28:25 hi a1k0n 2010-09-12T11:28:32 hey bartwe 2010-09-12T11:28:33 helps to know if it's worth to evaluate starting conditions or not :) 2010-09-12T11:28:51 looks like i finally found a heuristic that can match you 0.2 at times :) 2010-09-12T11:29:00 haha 2010-09-12T11:29:09 Strepto: nash equilibrium can break just before 200 turns i guess :) 2010-09-12T11:29:37 wow a1k0n-0.2 and drewrobb0.31 are neck and neck 2010-09-12T11:29:54 though i'm pretty sure deepblue is better when it's playing 2010-09-12T11:30:09 a1k0n: how do you see that ? 2010-09-12T11:30:16 http://www.benzedrine.cx/planetwars/getratings 2010-09-12T11:30:18 a1k0n: your's is still strictly heuristic based yeah? 2010-09-12T11:30:25 bartwe: thanks 2010-09-12T11:30:33 yeah, i'm working on a rewrite. what's on there now is pretty dumb 2010-09-12T11:30:52 a1k0n: geh says mister 2 after deepblue 2010-09-12T11:31:02 well, it's an extremely lucky dumb bot 2010-09-12T11:31:04 i can't wait for show and tell at the end 2010-09-12T11:31:26 Is the source for PlayGame posted anywhere? 2010-09-12T11:31:31 ut true bartwe.2 doesn't plan into the future either 2010-09-12T11:32:22 yeah, a1k0n-0.2 is just arithmetic on incoming fleets. 2010-09-12T11:32:31 really ? 2010-09-12T11:32:34 really. 2010-09-12T11:32:43 intresting 2010-09-12T11:32:47 an O(N^2) "what should i do with my ships" search. 2010-09-12T11:32:54 not seeing that in its actions 2010-09-12T11:33:01 it is a bit passive at times 2010-09-12T11:33:07 phew, website is back up... at least for a few minutes. 2010-09-12T11:33:14 it thinks it's better to take neutrals all at once so it saves up for them 2010-09-12T11:33:27 word to that 2010-09-12T11:33:30 my search is alot less wide 2010-09-12T11:33:32 it's a waste otherwise 2010-09-12T11:33:57 yeah. except, it's definitely a waste if you send N+1 ships to an N-size neutral that's right next to your opponent... 2010-09-12T11:34:00 and has some biases like: take enemies over neutrals 2010-09-12T11:34:13 a1k0n: geh, right 2010-09-12T11:34:14 bartwe: I am already doing what I thought would be at least a bit and it's still at around 0.01 set up time 2010-09-12T11:35:31 *** delt0r_ has joined #aichallenge 2010-09-12T11:35:58 hah, i probably shouldn't have won this one http://www.benzedrine.cx/planetwars/canvas?game_id=1284304022|drewrobb0.31|a1k0n-0.2 2010-09-12T11:36:03 *** Strepto has quit IRC (Quit: Page closed) 2010-09-12T11:36:22 a1k0n: geh there should be some multiway combats 2010-09-12T11:36:40 *** delt0r___ has quit IRC (Ping timeout: 240 seconds) 2010-09-12T11:37:10 i can't figure out why your moving your units so random ;p 2010-09-12T11:37:12 between more than 2 players you mean? 2010-09-12T11:37:21 yes 2010-09-12T11:37:29 heh. it tries to defend against attacks, and potential attacks. 2010-09-12T11:38:03 I think your version of "dumb" is little different from others' 2010-09-12T11:38:12 *** Strepto has joined #aichallenge 2010-09-12T11:38:53 dumb i think just means that it has a single simple rule or something 2010-09-12T11:38:57 well, let's say it makes extermely rough guesses about the future 2010-09-12T11:39:03 I bet 95% of entries won't consider "potential" anything 2010-09-12T11:39:12 and doesn't search any alternative possibilities 2010-09-12T11:39:36 *** ihodes has joined #aichallenge 2010-09-12T11:40:25 Speaking of games that shouldn't have been won: http://ai-contest.com/visualizer.php?game_id=4443059 2010-09-12T11:40:36 i can't tell if McLeopold has something extremely smart or just sorta lucky right now 2010-09-12T11:40:39 that was what i was planning as a future step, predicted attacks 2010-09-12T11:40:57 *** Shachar has quit IRC (Read error: Operation timed out) 2010-09-12T11:41:13 it's kinda dominating on the games against me and drewrobb 2010-09-12T11:41:42 *** Shachar has joined #aichallenge 2010-09-12T11:42:54 dhartmei: you changed tactics ? 2010-09-12T11:43:17 *** gilaniali has quit IRC (Quit: gilaniali) 2010-09-12T11:43:38 dhartmei's bot is definitely a lot better than it was 2010-09-12T11:43:44 it occurs to me that it might be nice if the number of turns is a function of how far ahead either player is over the other as the turn limit approaches, but it might just be too ad hoc to justify 2010-09-12T11:43:53 it seems to pick a planet and just go wild now 2010-09-12T11:44:07 heh. lots of 10s. 2010-09-12T11:44:20 http://www.benzedrine.cx/planetwars/canvas?game_id=1284305484|bartwe.2|dhartmei 2010-09-12T11:44:23 got my ass handed to me 2010-09-12T11:44:28 like at 200 turns if one player has twice as many ships as the other it's a win, otherwise it continues, and the minimum ratio necessary to win gets smaller over time 2010-09-12T11:44:30 yes, 10 ships at a time 2010-09-12T11:44:53 i think it would do better with more predicition 2010-09-12T11:45:52 dhartmei's bot has my number. I squeak out wins against most other higher ranked bots, but not his 2010-09-12T11:47:06 which one's yours on benzedrine? 2010-09-12T11:47:07 i find the idea that A can beat B, B can beat C, and C can beat A interesting in tournaments like this. would be interesting to analyzing some game data to look for such cases 2010-09-12T11:47:18 *analyze 2010-09-12T11:47:39 jmcarthur: do PCA and find a N-dimensional Elo score 2010-09-12T11:47:52 a1k0n: that is actually exactly what i was thinking 2010-09-12T11:48:02 *** s_denchev has quit IRC (Ping timeout: 252 seconds) 2010-09-12T11:48:27 i wonder how significant the second factor would be 2010-09-12T11:48:58 (and whether we could distinguish it from the first) 2010-09-12T11:49:05 as in identify it 2010-09-12T11:49:21 *** FlagCapper has quit IRC (Quit: Page closed) 2010-09-12T11:49:27 *** sinan has quit IRC (Ping timeout: 265 seconds) 2010-09-12T11:51:15 wow 2010-09-12T11:51:42 the binary for my bot, stripped and zipped, is 5.3M... 2010-09-12T11:51:59 i wonder if ghc head is including a bunch of stuff that 6.12.3 wouldn't... 2010-09-12T11:52:09 a1kon: my bot is vortex (with variant numbers, but not vortex3636 - that's someone else) 2010-09-12T11:52:24 rwa: ahh 2010-09-12T11:53:43 if i wanted to lower someones elo, i'd just put up an example bot on their name 2010-09-12T11:53:47 bzip2 squeezes out 4.2M 2010-09-12T11:53:58 so don't put too much trust in names 2010-09-12T11:54:22 bartwe - you mean on the tcp server? 2010-09-12T11:54:29 would be nice to have some sort of authentication 2010-09-12T11:54:33 yes 2010-09-12T11:54:47 simple pass or hash 2010-09-12T11:54:54 pgp 2010-09-12T11:55:11 even a simple password would be nicer than nothing 2010-09-12T11:55:31 nah, something withouh accounts would be preferable 2010-09-12T11:55:35 no registration even needed. just use a password and as long as you keep using the same one it keeps updating the same bot stats 2010-09-12T11:55:45 ah, that works 2010-09-12T11:55:50 hey the site's fast again and i'm getting games again 2010-09-12T11:56:21 and if you use the same bot name with a different password it just shows as botname(2) or something 2010-09-12T11:56:30 Huh 2010-09-12T11:56:44 Can you just make one IssueOrder() a round ? 2010-09-12T11:57:00 *** Guest75371 has quit IRC (Ping timeout: 252 seconds) 2010-09-12T11:57:01 as many as you want 2010-09-12T11:57:06 or as few as you want 2010-09-12T11:57:14 you don't even have to issue any 2010-09-12T11:57:27 turtlebot ftw 2010-09-12T11:58:19 *** ihodes_ has joined #aichallenge 2010-09-12T11:58:30 *** ihodes has quit IRC (Read error: Connection reset by peer) 2010-09-12T11:58:31 *** ihodes_ is now known as ihodes 2010-09-12T11:59:05 a1k0n: Out of curiosity, how much time did you spend on the tron bot? 2010-09-12T11:59:20 people always ask me that, and i don't have a good answer 2010-09-12T11:59:51 i spent many evenings for two weeks or so 2010-09-12T12:00:47 but it kind of became an obsession and i couldn't stop thinking about it 2010-09-12T12:00:49 check your commit logs :P 2010-09-12T12:01:13 that reminds me 2010-09-12T12:01:14 yeah, those don't show how much time is pent on a given patch but it gives a rough guess 2010-09-12T12:01:17 i should commit 2010-09-12T12:02:02 a1k0n: so you just kept thinking about the contest all day long? 2010-09-12T12:02:30 you should compete in the icfpcontest 2010-09-12T12:02:38 shorter, easier to keep the focus for 72hours 2010-09-12T12:02:48 Mathnerd314: well, more or less 2010-09-12T12:02:54 the site is fast and games are working, but submission doesn't seem to be working now 2010-09-12T12:03:03 bartwe: i usually do. i can't *believe* i forgot about the last one 2010-09-12T12:03:15 geh 2010-09-12T12:03:24 it was an ok one, not one of the best 2010-09-12T12:03:41 i think you would have done well 2010-09-12T12:03:53 Mathnerd314: that's basically what happens to me 2010-09-12T12:03:54 Okay, I think I'll start working on a bot now. I'm not sure I'll have enough time to do much, but two weeks doesn't sound that bad. 2010-09-12T12:03:58 the thing is, i'm usually out spending time with the family on weekends these days 2010-09-12T12:04:06 i can't get it out of my head even as i try to do other things (like work) 2010-09-12T12:04:17 icfpcontest this year was a milestone for me, first one that i did in 100% pluk a compiler/language i wrote 2010-09-12T12:04:26 nice 2010-09-12T12:04:30 bartwe: awesome! i still haven't done icfp 2010-09-12T12:04:45 a1kon: if you're anything like me, family saves you a bit from being too obsessive about things 2010-09-12T12:04:46 that's what i want to do some day 2010-09-12T12:04:49 bartwe: any links for pluk? 2010-09-12T12:05:11 bartwe: oh were you the one wondering about being able to use pluk for the contest? 2010-09-12T12:05:23 i recall that language name vaguely, i think from that 2010-09-12T12:05:41 jmcarthur: yes, pluk-lang.org its not really for public consumption 2010-09-12T12:05:41 jmcarthur: I've found http://bitbucket.org/bartwe/plukc/wiki/Home 2010-09-12T12:05:57 rwa: yeah, exactly. and it seems better work comes from taking breaks and then focusing for short periods 2010-09-12T12:05:59 RainCT: check 2010-09-12T12:06:13 bartwe: cool, i'll peek at it 2010-09-12T12:06:19 i'm a PLT enthusiast 2010-09-12T12:06:34 blog link fail :\ 2010-09-12T12:06:35 bartwe: Is that like a Vala without gobject? 2010-09-12T12:07:14 RainCT: close enough 2010-09-12T12:07:28 bartwe: so what would you say is pluk's claim to fame? 2010-09-12T12:07:57 speaking of which I need to make food for my kids... hope submissions can be fixed soon 2010-09-12T12:08:22 jmcarthur: none, its a hobby project, it doesn't have a xsignificant company/communicty 2010-09-12T12:08:30 andi build compilers, not communities 2010-09-12T12:08:34 oh i found a page explaining it a little 2010-09-12T12:08:42 on bitbucket 2010-09-12T12:09:49 *** ratatata_ has joined #aichallenge 2010-09-12T12:09:57 sounds nice 2010-09-12T12:10:20 gotta love Chrome crashing when needed :/ 2010-09-12T12:10:22 looks like an imperative OO language? 2010-09-12T12:10:29 yeah 2010-09-12T12:10:30 it is 2010-09-12T12:10:36 java/C#-like 2010-09-12T12:10:45 basicly c# compiled to intel instructions 2010-09-12T12:10:53 with niceties like type inference 2010-09-12T12:11:10 ah, cool 2010-09-12T12:11:51 i can't make myself use OO languages anymore. i usually reach for either haskell or c, now 2010-09-12T12:11:52 *** ratatata has quit IRC (Ping timeout: 252 seconds) 2010-09-12T12:12:29 i do alot of c# at work, so i better stay comfortable with it 2010-09-12T12:13:50 a1k0n you work at yahoo? 2010-09-12T12:14:00 yep 2010-09-12T12:14:09 sf bay area? 2010-09-12T12:14:16 as of a month ago, yeah 2010-09-12T12:14:31 I'm in Livermore, if you know where that is (out east) 2010-09-12T12:15:01 how is the sf bay area? i've never been there 2010-09-12T12:15:08 ah, cool. i'm in sunnyvale. 2010-09-12T12:15:21 considering applying for some things there 2010-09-12T12:15:26 it's.. sunny and warm in the valley. all the time. 2010-09-12T12:15:40 expensive as hell to live here. 2010-09-12T12:15:42 I interviewed at Lockheed there once awhile back 2010-09-12T12:16:06 a friend of mine works there, though i haven't seen him in a long time 2010-09-12T12:16:27 I declined their offer when my senior manager, probably 65 years old, worked in a tiny, open, noisy cube 2010-09-12T12:16:35 i grew up in rural areas. whenever i interview in a big city i feel like i would miss the openness if i was to move and work there 2010-09-12T12:17:03 i lived in a pretty rural area until recently. there's plenty of openness in CA, it's just full of people 2010-09-12T12:17:24 full of people makes me feel kind of closed in :\ 2010-09-12T12:17:43 i live in tasmania, it's awesome 2010-09-12T12:17:49 but at least there is some openness, i guess 2010-09-12T12:17:51 i hate queues 2010-09-12T12:18:02 i don't mind using them in a program, but partaking in them makes me rage 2010-09-12T12:18:52 i like stacks because the later i arrive the sooner i leave 2010-09-12T12:18:57 even sitting at a traffic light in a car for 2 minutes makes me want to rip someones throat out 2010-09-12T12:19:07 :) 2010-09-12T12:19:16 heh. you sound like my two year old 2010-09-12T12:19:31 a1kon: you think it's expensive now, you should have seen things 5-6 years ago 2010-09-12T12:19:36 you're two year old talks about queues and stacks? 2010-09-12T12:19:45 &your 2010-09-12T12:19:49 we waited in line about a half hour for this awesome balloon thrill ride thingy at gilroy gardens and i thought he was going to die 2010-09-12T12:19:50 dangit 2010-09-12T12:19:57 heh 2010-09-12T12:20:32 i once sat in a crowded tennis arena for like 3 hours without water or sunscreen etc. just to watch anna kornikova play 2010-09-12T12:20:36 *** computerwiz_222 has joined #aichallenge 2010-09-12T12:20:45 rwa: yeah. and it kinda sucks now because we still have our house in WI to pay the mortgage on until it sells, and it ain't selling 2010-09-12T12:22:00 ouch 2010-09-12T12:22:51 (and rent on a little 3br ranch here costs more than the mortgage on a 2-story 4br 3ba in WI) 2010-09-12T12:24:27 I tried printing planet info to a file at the end of the first turn and ended up with some strange values 2010-09-12T12:24:35 planets with thousands of ships 2010-09-12T12:25:00 nvm, stupidity present 2010-09-12T12:25:00 *** ihodes has left #aichallenge 2010-09-12T12:27:32 *** mega1 has joined #aichallenge 2010-09-12T12:27:46 that's usually the case, and only after you say something about it publicly 2010-09-12T12:27:57 *** holmar has joined #aichallenge 2010-09-12T12:28:22 thought I fixed it, still getting strange ship numbers >.< 2010-09-12T12:28:35 that's what I love and hate about programming, it's always your fault 2010-09-12T12:28:41 only ship numbers? thats weird 2010-09-12T12:29:05 i got wrong coordinates for some time because somehow they considered it sane to just pass everything around as strings with who-knows what locale 2010-09-12T12:29:09 *** ademar has joined #aichallenge 2010-09-12T12:29:31 *** Shachar has quit IRC (Remote host closed the connection) 2010-09-12T12:29:36 anyone know what the turn limit on tcpserver is? 2010-09-12T12:30:28 *** ademar has quit IRC (Client Quit) 2010-09-12T12:30:38 300 2010-09-12T12:30:55 orly. thx. 2010-09-12T12:31:01 this is definitely strange, all planets show as owner 1 when I print it 2010-09-12T12:31:24 *** ademar has joined #aichallenge 2010-09-12T12:31:25 then you win! 2010-09-12T12:31:27 Hmm... so when does the player "die" On official server, you die if you have no planets and no fleets, on tcpserver you die if you have no planets 2010-09-12T12:31:32 lol 2010-09-12T12:31:51 Palmik: on tcp you die when your last fleet arrived and you have no planets. 2010-09-12T12:32:10 caytchen, it used to be like that, but was not when I last checked 2010-09-12T12:32:15 drewrobb: i don't understand why my bot suddenly pulled ahead of yours on benzedrine.cx 2010-09-12T12:32:33 do you do any game tree searching or anything btw? 2010-09-12T12:32:53 a1k0n: you didn't ninja switch to a new version? 2010-09-12T12:33:00 nope, haven't touched it 2010-09-12T12:33:10 no, my bot is just heuristics stuff for now 2010-09-12T12:33:52 hmm, ok 2010-09-12T12:34:19 Hehe, first page ends with player ranked 99th and second starts with player ranked 101st 2010-09-12T12:34:20 i was running a bunch of other tests, maybe the the CPU time it gets is less as the system is loaded? 2010-09-12T12:34:38 maybe it is a big statistical fluctiation? 2010-09-12T12:34:43 well, does that even matter? it wouldn't for mine, it's deterministic 2010-09-12T12:34:48 maybe deepblue was keeping me down? 2010-09-12T12:35:08 could be, but he was bringin' me down too. i guess it's just randomness. 2010-09-12T12:35:18 our confidence intervals overlap 2010-09-12T12:35:24 mine search random actions at depth 1 for .1 sec 2010-09-12T12:35:37 oh, cool. so you do do searching, you liar 2010-09-12T12:35:39 but yeah, for testing i keep it deterministic 2010-09-12T12:35:47 depth 1? 2010-09-12T12:36:00 i just consider random moves for myself 2010-09-12T12:36:00 heh, more than the depth 0 i do. :) 2010-09-12T12:36:09 ah, hm 2010-09-12T12:36:17 oh, i do depth 0 by your indexing convention 2010-09-12T12:36:29 *** gilaniali has joined #aichallenge 2010-09-12T12:36:41 i don't generate candidate moves per se, i just figure out how many ships i can afford to send and find places to send them 2010-09-12T12:37:06 with no real objective function to speak of 2010-09-12T12:37:16 ah 2010-09-12T12:37:45 i have an extremely naive way of finding sets of moves 2010-09-12T12:37:58 *** deepblue has joined #aichallenge 2010-09-12T12:37:59 but yeah, i also limit search to 1 move ammount 2010-09-12T12:38:04 the next generation of bots should be interesting 2010-09-12T12:38:10 speak of the devil 2010-09-12T12:38:15 morning :) 2010-09-12T12:38:26 next step: skynet 2010-09-12T12:39:14 after what time do the games get removed on the tcp server? 2010-09-12T12:39:34 it seems to be rolling 2010-09-12T12:39:52 yeah, not sure exactly how many games it holds onto but it does roll the out 2010-09-12T12:39:59 ah ok 2010-09-12T12:40:05 your absense has switched mine and drewrobb's ranking it seems 2010-09-12T12:40:09 so lots of games happens while I was asleep :) 2010-09-12T12:40:09 how much planets max in official tournament ? 2010-09-12T12:40:26 edcba: they didn't say, but they did say it won't be significantly more than the maps have now 2010-09-12T12:41:10 deepblue: i just have my machine at work cranking them out 2010-09-12T12:41:11 let's hope not too many more 2010-09-12T12:41:28 contestbot: rankings 2010-09-12T12:41:29 caytchen: Top 10 players: dmj111(2835), braddunbar(2815), albert(2792), seigen(2732), Kavish(2707), drewrobb(2695), Obeleh(2679), wtfbbq(2679), sighbot(2676), tinasky(2661) 2010-09-12T12:41:35 yeah i think a few O(N^2) operations are unavoidable 2010-09-12T12:41:35 I just started my bot again 2010-09-12T12:41:43 me too 2010-09-12T12:42:02 i need to start writing code and stop writing irc messages 2010-09-12T12:42:19 this random matching on the official server makes climbing so slow 2010-09-12T12:42:31 yes, this irc channel is very distracting 2010-09-12T12:42:34 and I have buggy code uploaded, takes so loooong 2010-09-12T12:42:42 *** retybok has joined #aichallenge 2010-09-12T12:43:06 yeah I think I'm gonna shut down the irc client after breakfast to get to implementing some changes :) 2010-09-12T12:44:38 i should just spam deepblue: every few minutes to keep you from getting focused 2010-09-12T12:44:52 ahhh ;) 2010-09-12T12:46:10 top10 I'm coming! already up to 25th place on the official 2010-09-12T12:46:50 i lost my totally unfair #1 place :( 2010-09-12T12:47:15 let's see if I can reserve it for you ;) 2010-09-12T12:47:21 i bet albert and dmj111 are afraid to resubmit 2010-09-12T12:47:44 I hope I get lucky and can play one game against them 2010-09-12T12:47:54 that would be a reason to resubmit 2010-09-12T12:48:02 i should probably resubmit 2010-09-12T12:48:07 *** virtualmike has joined #aichallenge 2010-09-12T12:48:10 there's a 1/90,000 chance of that every five seconds or so 2010-09-12T12:48:14 so far all the games I had on the official server were anywhere close to dangerous for me :/ 2010-09-12T12:48:16 is there an inofficial server? 2010-09-12T12:48:19 2010-09-12T12:48:31 hehe 2010-09-12T12:48:35 RobinB: http://www.benzedrine.cx/planetwars/ 2010-09-12T12:49:06 ah i see, thanks 2010-09-12T12:49:12 and that's faster, i presume? 2010-09-12T12:49:25 i guess with 58 players, it would be 2010-09-12T12:49:28 http://ai-contest.com/visualizer.php?game_id=4445926# 2010-09-12T12:49:34 well you run your own bot on your own computer 2010-09-12T12:49:39 a1k0n: albert hasn't been working on the bot the last few days due to work but i think today he is 2010-09-12T12:49:41 even with the bug still in there it wins hands down 2010-09-12T12:49:46 Hi, anyone know if regestration is still working? been waiting over 24 hours for my confirmation email (to a gmail address, its not in the spamfilter) 2010-09-12T12:50:02 hmz 2010-09-12T12:50:10 i see server errors on benzedrine 2010-09-12T12:50:11 a1k0n: but he has some incremental changes already so you are right a little bit ;) 2010-09-12T12:50:20 units that aren't moving 2010-09-12T12:50:43 phreeza: hehe, i changed just one number and totally gave up a top-3 bot 2010-09-12T12:50:52 s/bot/spot/ anyway 2010-09-12T12:51:38 yea... shouldn't read too much into the rankings right now i think 2010-09-12T12:51:45 Anyone using non-heuristic methods? 2010-09-12T12:51:58 not yet 2010-09-12T12:52:06 how? the state space is gigantic 2010-09-12T12:52:18 I'm trying to think how to implement one, but it seems incredibly hard 2010-09-12T12:52:19 you can still calculate some things 2010-09-12T12:52:27 yeah i might be doing a very narrowly directed search if any 2010-09-12T12:52:34 *** cobracom has quit IRC (Ping timeout: 252 seconds) 2010-09-12T12:52:36 yeah 2010-09-12T12:52:44 deepblue: with your name, thats what i'd expect you to do :D 2010-09-12T12:52:50 tron was so much nicer 2010-09-12T12:52:51 brute force 2010-09-12T12:52:59 i like this one 2010-09-12T12:53:03 tron was awesome too 2010-09-12T12:53:04 me too 2010-09-12T12:53:05 i know i know but with the 1sec timelimit 2010-09-12T12:53:18 'deepblue has to improvise' 2010-09-12T12:53:41 deepblue: i was wondering on the tcpserver if maybe you are running on some blue gene or sth ;) 2010-09-12T12:53:55 not really 2010-09-12T12:54:01 *** repson has joined #aichallenge 2010-09-12T12:54:14 you guys are still pantsing my heuristics 2010-09-12T12:54:29 * RainCT throws deepblue a '/me' ;) 2010-09-12T12:54:35 although they're being rather poorly thrown together 2010-09-12T12:54:53 right now it's running on a mac server at the university since my home internet is so sketchy 2010-09-12T12:55:16 Hey. The server do not want to run my bot. 2010-09-12T12:55:27 30 minutes 2010-09-12T12:55:29 it's al lot 2010-09-12T12:55:32 * deepblue calculates 2010-09-12T12:56:09 elijahbal be patient the server are slow 2010-09-12T12:56:22 btw 2010-09-12T12:56:33 deepblue : are you a bot ? 2010-09-12T12:56:55 * deepblue is alive 2010-09-12T12:57:03 flesh and blood and all 2010-09-12T12:57:04 deepblue : 1s is a lot for run this game 2010-09-12T12:57:06 antimatroid: is this some very nerdy way of talking smack? 2010-09-12T12:57:08 :D 2010-09-12T12:57:13 D: 2010-09-12T12:57:32 A lot of cpu time. Why dont you make the game a bit more challenging and say 0.1s ? 2010-09-12T12:57:34 elijahbal: i could need way more time then 1sec :) 2010-09-12T12:57:39 i was talking about my heuristics, not yours :) 2010-09-12T12:57:45 took me a minute to work out what you meant 2010-09-12T12:57:51 ah hehehe ok :) 2010-09-12T12:58:06 ah yea ok i misread, sorry 2010-09-12T12:58:13 missed the "my" 2010-09-12T12:58:20 or lets put it that way I will put 0.1s, 1s or 2hours all to good use :) 2010-09-12T12:58:26 just found this in the starter package http://codepad.org/TJSSxArB maybe the rwrite is not bad idea :) 2010-09-12T12:58:49 okay, the scheduler should be something like this: take the bots with large confidence intervals and match them against bots with maximally overlapping confidence intervals 2010-09-12T12:58:59 i think that should narrow down those confidence intervals the most quickly 2010-09-12T12:59:01 Palmik: i know, ridiculous isn't it? 2010-09-12T12:59:07 i can't believe i didn't get on tcp server last time, so much better 2010-09-12T12:59:33 * RainCT has just finished refactoring PlanetWars.py because the one in the start pack wasn't pythonic enough 2010-09-12T12:59:50 *** ratatata_ is now known as ratatata 2010-09-12T12:59:52 RainCT: there's some linked on the forums 2010-09-12T13:00:00 a1k0n, a bit indeed :) 2010-09-12T13:00:05 that are more pythonic 2010-09-12T13:01:02 highly recommend just throwing away all the starter code 2010-09-12T13:01:05 http://github.com/jaspervdj/planet-wars-haskell etc 2010-09-12T13:01:11 the input parsing is also terrible 2010-09-12T13:01:12 yay, first place again 2010-09-12T13:01:17 ack, the situation is the same for all the other packs, too 2010-09-12T13:01:22 deepblue: :| 2010-09-12T13:01:41 Wee back to too many connections 2010-09-12T13:01:42 a1k0n: yeah, I'll rewrite MyBot.py from scratch 2010-09-12T13:01:47 no offense to jaspervdj, but that haskell starter has some slow IO/parsing 2010-09-12T13:01:57 http://www.benzedrine.cx/planetwars/canvas?game_id=1284309764|deepblue|a1k0n-0.2 2010-09-12T13:01:58 oh, nice new names in the list of opponents 2010-09-12T13:02:01 i guess for most bots it's not a big deal 2010-09-12T13:02:12 err i meant to paste the python... 2010-09-12T13:02:20 a1k0n-0.2 focuses all effort on a lost cause, and then loses 2010-09-12T13:02:32 anyone think that this is optimal play? 2010-09-12T13:02:33 dan that is total annhilation 2010-09-12T13:02:33 mmh 2010-09-12T13:02:36 http://www.benzedrine.cx/planetwars/canvas?game_id=1284297611|McLeopold|wtfbbq 2010-09-12T13:02:37 *damn 2010-09-12T13:03:02 drewrobb: lol 2010-09-12T13:03:04 lol 2010-09-12T13:03:06 i doubt that is optimal 2010-09-12T13:03:10 i've seen a few of those now 2010-09-12T13:03:11 i want to play leopold! 2010-09-12T13:03:12 drewrobb: haha, a draw! 2010-09-12T13:03:14 actually my program do not take a lot of cpu time 2010-09-12T13:03:20 0.001 at his worst 2010-09-12T13:03:25 maybe not optimal, but would it draw with optimal play? 2010-09-12T13:03:36 Guest77645: Ah, found it. Thanks. 2010-09-12T13:03:55 the maps are symmetric so it should be a draw? 2010-09-12T13:04:08 if both play the same optimal strategie 2010-09-12T13:04:10 games like that should be a draw 2010-09-12T13:04:14 honestly i think a fair win in that match would have been for mcleopold to win it since he was the one laying on the pressure 2010-09-12T13:04:15 not if there is a mixed equilibrium 2010-09-12T13:04:21 i like the idea of extending it, then a certain radius would define a draw 2010-09-12T13:04:21 wow lulin4 is no pushover 2010-09-12T13:04:26 but of course that does satisfy a win criteria 2010-09-12T13:04:27 mmh 2010-09-12T13:04:30 *doesn't 2010-09-12T13:04:38 *** ademar has quit IRC (Remote host closed the connection) 2010-09-12T13:04:48 okay it was, but wasn't necessarily going to happen 2010-09-12T13:04:59 http://www.benzedrine.cx/planetwars/canvas?game_id=1284310123|deepblue|lulin4 2010-09-12T13:05:11 *** Strepto has quit IRC (Quit: Page closed) 2010-09-12T13:05:49 had a bad start there and hard work todo to win 2010-09-12T13:06:01 that was a good fight though 2010-09-12T13:06:27 why does everybody uses these streams of fleets 2010-09-12T13:06:33 no idea. i think it's horrible 2010-09-12T13:06:33 *** Strepto has joined #aichallenge 2010-09-12T13:06:34 taking the corner neut seems crazy 2010-09-12T13:06:35 deepblue: I don't 2010-09-12T13:06:57 deepblue: i think they are just thinking "hey i have spare ships. i should use them" 2010-09-12T13:06:57 it's only unneccesarly binding ships in flight before they are really needed 2010-09-12T13:06:59 lulin probably needs to adjust his attack.. always 1 ship short of actually taking the planet hehe 2010-09-12T13:07:15 mmh 2010-09-12T13:07:17 i think it's a waste 2010-09-12T13:07:33 deepblue: they could also be used to make the opponent focus time on the wrong thing, but i don't know whether that will be viable until i see some better bots anyway 2010-09-12T13:07:54 yeah 2010-09-12T13:08:12 I'm looking forward to the bot that beats the crap out of mine :) 2010-09-12T13:08:13 jmcarthur: mine starts to use streams in response to attack/defense of others that also turn into streams 2010-09-12T13:08:26 yeah 2010-09-12T13:08:27 same here 2010-09-12T13:08:28 sure, but that's defensive 2010-09-12T13:08:33 i think streams are unavoidable if both sides keep escalating their forces to the same point 2010-09-12T13:08:40 i'd say that is optimal defense, suboptimal offense 2010-09-12T13:09:05 *** FlagCapper has joined #aichallenge 2010-09-12T13:09:08 i guess a tweak that escalates the attack waves would force a solution 2010-09-12T13:09:08 assuming we are only considered small streams or big delayed chunks 2010-09-12T13:09:24 mine will stream, but not for the purpose of attacking or defending 2010-09-12T13:09:24 it is also a good way to make the growthrate the winning indicator 2010-09-12T13:09:41 Is anyone else having a problem with the official server where you're winning games but your rank doesn't increase? 2010-09-12T13:09:56 also, i have 0 'zero' defense code 2010-09-12T13:09:57 *** AJC has joined #aichallenge 2010-09-12T13:10:06 FlagCapper: maybe the ranking script isn't running? 2010-09-12T13:10:09 i just attack the predicted lost planets that i used to own 2010-09-12T13:10:13 and that way keep them 2010-09-12T13:10:17 do any of you use scripts to stress test your bot locally against the default bots on all maps? 2010-09-12T13:10:29 a simple question concerning the rules: can you abandon a planet? 2010-09-12T13:10:36 I don't know, but I've won all 6 of my games and I'm still at 602 2010-09-12T13:10:41 FlagCapper: somebody on the dev mailing list mentioned that the ranking updater is slowing things down a lot. maybe somebody is working on it 2010-09-12T13:10:48 ok, thanks 2010-09-12T13:10:53 j3camero: bug report; I've submitted bots 5 or 6 times this morning in the last couple hours and none of the entries show and I've got no email either (nothing in spam folder) 2010-09-12T13:11:02 FlagCapper: i wouldn't worry about it. your wins won't be lost. it's a batch updater, not incremental 2010-09-12T13:11:16 rwa: wait and don't resubmit again, just wait it will show up 2010-09-12T13:11:19 so if i have a planet with 5 ships and want to send all 5 somewhere else 2010-09-12T13:11:23 is that possible? 2010-09-12T13:11:30 or does one ship have to stay? 2010-09-12T13:11:31 canihazbacon 2010-09-12T13:11:32 canihazbacon: yes 2010-09-12T13:11:33 sure 2010-09-12T13:11:36 you can send all 5 2010-09-12T13:11:42 canihazbacon: you still own a planet with 0 ships 2010-09-12T13:11:42 and the planet will still be yours :) 2010-09-12T13:12:02 AJC: http://gist.github.com/574904 2010-09-12T13:12:09 canihazbacon: http://code.google.com/p/ai-contest/wiki/GameSpecification 2010-09-12T13:12:16 Naktibalda, thanks, taking a look now 2010-09-12T13:12:26 i guess i should go ahead and start that patch to put that on the main site 2010-09-12T13:12:40 deepblue: Pushed him over easier http://www.benzedrine.cx/planetwars/canvas?game_id=1284310776|ermau-1.0|lulin4 2010-09-12T13:12:59 thx for the spec! 2010-09-12T13:13:03 in my bot the streaming is sort of emergent behavior. I'm not quite sure why it does it, but it's satisfying my criteria for a good move somehow. 2010-09-12T13:13:06 *** Strepto has quit IRC (Ping timeout: 252 seconds) 2010-09-12T13:13:14 canihazbacon: no prob. i needed it too 2010-09-12T13:13:14 nice 2010-09-12T13:13:58 I think my attack calculation is off too.. growth happens before movement, right? 2010-09-12T13:14:17 http://www.benzedrine.cx/planetwars/canvas?game_id=1284310555|deepblue|vortex.6 2010-09-12T13:14:24 *** gilaniali has quit IRC (Quit: gilaniali) 2010-09-12T13:14:25 woho a loss I can analyse :) 2010-09-12T13:14:33 It seems graph representations could be useful for this game. Anyone using them? 2010-09-12T13:14:48 retybok, what for? 2010-09-12T13:15:01 finally beat you deepblue? ;) 2010-09-12T13:15:16 retybok: probably most bots are treating it as a graph, even if not explicitly 2010-09-12T13:15:23 but still directly 2010-09-12T13:15:27 relationship between the planets 2010-09-12T13:15:30 *** xkpe has quit IRC (Ping timeout: 255 seconds) 2010-09-12T13:15:32 right 2010-09-12T13:15:50 most bots are probably precomputing the distances as some sort of adjacency matrix already 2010-09-12T13:15:54 or will be, anyway 2010-09-12T13:15:58 :) 2010-09-12T13:16:26 jmcarthur, a fully connected matrix isn't really a graph though :-) 2010-09-12T13:16:26 deepblue: tell me how I did it when you figure it out... 2010-09-12T13:16:33 I'm still designing my bot, I don't have a definite idea yet 2010-09-12T13:16:44 AJC: why not? 2010-09-12T13:16:46 *** vikhyat has quit IRC (Ping timeout: 245 seconds) 2010-09-12T13:16:58 a graph is a graph 2010-09-12T13:17:34 my bot is taking a lot of up front work as i create some data structures 2010-09-12T13:17:35 AJC: you could prune some of the longest links... 2010-09-12T13:17:40 *** virtualmike has quit IRC (Quit: Page closed) 2010-09-12T13:17:46 retybok, then yeah it's a graph. 2010-09-12T13:17:53 if it's fully connected it makes no sense to think of it as a graph 2010-09-12T13:18:08 it's still a graph if it's fully connected, even if it's not really useful to see it as a graph then :) 2010-09-12T13:18:10 seriously, someone was saying yesterday about pointers 2010-09-12T13:18:17 AJC: yeah, that's why i said "even if not explicitly" a graph 2010-09-12T13:18:22 AJC: but it's still technically a graph 2010-09-12T13:18:26 jmcarthur, gotcha. 2010-09-12T13:18:26 *** jaspervdj has joined #aichallenge 2010-09-12T13:18:49 and i bet a lot of the analyses that people try are graph algorithms in disguise, too 2010-09-12T13:18:49 *** td has joined #aichallenge 2010-09-12T13:19:56 what happened to deepblue's 800~ ELO? O_o 2010-09-12T13:19:59 "graph algorithm" is just a point of view, really 2010-09-12T13:20:08 DAE keep watching a game thinking you're a completely different colour to what you are? 2010-09-12T13:20:25 antimatroid: that happend a few times :) 2010-09-12T13:20:44 rwa: the problem is finding the point of you that will help you design the best bot :) 2010-09-12T13:20:45 rwa: indeed. most (all?) algorithms are kind of graph algorithms "in disguise" i guess 2010-09-12T13:20:58 *point of view 2010-09-12T13:21:10 antimatroid: yes! 2010-09-12T13:21:26 antimatroid: i especially get confused when watching somebody like deepblue who has a color in his name 2010-09-12T13:21:32 antimatroid: especially when 'deepblue' is in red :) 2010-09-12T13:21:37 lol 2010-09-12T13:21:39 :) 2010-09-12T13:21:40 would be nice if the protagonist was always blue or red 2010-09-12T13:21:40 haha 2010-09-12T13:21:46 hehe 2010-09-12T13:22:00 could just assign each username a different colour? 2010-09-12T13:22:16 then split conflicts with different colours to what either is used to 2010-09-12T13:22:23 antimatroid: wouldn't work when blue played blue, for example 2010-09-12T13:22:29 ^^ 2010-09-12T13:22:39 do you guys think i should integrate the spec into the problem description page or should it be a separate page? 2010-09-12T13:22:41 that canvas thing generally could be beefed up.. 2010-09-12T13:22:56 hmz, color from username hash 2010-09-12T13:22:57 the canvas thing is awesome 2010-09-12T13:22:59 make the protagonist the player who's profile you clicked from? 2010-09-12T13:23:04 antimatroid: yeah, a hash-generated color would be nice 2010-09-12T13:23:16 drewrobb: right 2010-09-12T13:23:25 that could work 2010-09-12T13:23:26 hashcolor can also turn out horrible 2010-09-12T13:23:33 bartwe: you just have to be smart 2010-09-12T13:23:57 please tweak the has that way that I get blue :D 2010-09-12T13:24:08 or have "home" and "away" colors for each player 2010-09-12T13:24:27 its not football D: 2010-09-12T13:24:29 then hash into a very small color space 2010-09-12T13:24:32 how do you guys do a/b testing on bot versions ? 2010-09-12T13:25:35 oh god, it's 3 20am and i have class at 10 2010-09-12T13:25:58 this competition is far too addictive 2010-09-12T13:26:00 i always round robin my bots (or equivalently run my newest against all others and keep a log) 2010-09-12T13:26:00 heh, why take class when you could be improving your bot ;p 2010-09-12T13:26:23 i'm using tcp atm, i will set up something like that at some point 2010-09-12T13:26:51 i might upload my current one to official site and leave it at that for the night 2010-09-12T13:26:54 *** RobinB has quit IRC () 2010-09-12T13:27:10 the solution is to work on bot during class 2010-09-12T13:27:18 i'll just make a separate page for the spec and if we want to merge it with problem description later we can 2010-09-12T13:27:28 i don't have a laptop, and my maths lecturers teach off the whiteboard 2010-09-12T13:28:11 unfortunate 2010-09-12T13:29:14 *** ath88 has quit IRC (Ping timeout: 252 seconds) 2010-09-12T13:29:16 i thought of a dumb idea for an ai contest 2010-09-12T13:29:22 finding good parking spots 2010-09-12T13:29:39 unfortunate? 2010-09-12T13:29:51 i much prefer being taught that way 2010-09-12T13:30:06 ok, i'm off to implementing some changes :) 2010-09-12T13:30:27 *** deepblue has quit IRC (Quit: Page closed) 2010-09-12T13:30:48 meh, i always learned more reading 2010-09-12T13:33:16 depends, i tend to only go to class if it's worth going 2010-09-12T13:33:27 but if it is, i'd rather have the lecturer up the front talking to us 2010-09-12T13:33:46 mmk, time to start on v2 2010-09-12T13:35:12 yeah, now we're talking, i finally beat someone in the top area 2010-09-12T13:35:21 I wish challenge n°3 was a tetrinet challenge :) 2010-09-12T13:36:28 even though this is a pretty neat challenge :) 2010-09-12T13:37:53 morning 2010-09-12T13:38:25 *** braddunbar has quit IRC (Quit: leaving) 2010-09-12T13:39:14 @where tcpserver 2010-09-12T13:39:14 ratatata: tcpserver = http://www.benzedrine.cx/planetwars/ 2010-09-12T13:40:33 alright, v0.31 is obsolete 2010-09-12T13:40:58 *** braddunbar has joined #aichallenge 2010-09-12T13:41:04 no more cputime for him 2010-09-12T13:41:20 when I was undergrad, if the class had a good textbook I did not waste my time with lectures. Terrible waste of an hour generally. 2010-09-12T13:41:29 is boost available on the server? 2010-09-12T13:41:42 *** braddunbar has quit IRC (Client Quit) 2010-09-12T13:42:58 *** ratatata has quit IRC (Quit: Page closed) 2010-09-12T13:44:18 broke into the top 100 on the server oooo 2010-09-12T13:44:42 congrats :) 2010-09-12T13:44:58 rwa: you should have gone to a liberal arts school 2010-09-12T13:45:16 fucawb: whys that 2010-09-12T13:45:55 because small class size will make even a crappy lecturer decent 2010-09-12T13:46:47 I went to smallish school. I just don't like being held hostage to the lecture material. It's like the difference between watching TV and surfing the web for information, kinda. 2010-09-12T13:47:01 the mysql things is back 2010-09-12T13:47:14 I can read 5x faster than a lecturer can speak 2010-09-12T13:47:29 Warning: mysql_connect() [function.mysql-connect]: Too many connections in /var/www/vhosts/ai-contest.com/www/mysql_login.php 2010-09-12T13:47:51 maybe it's just a difference of learning style 2010-09-12T13:47:59 I never opened half my books 2010-09-12T13:48:02 if already know something, I can skip that. If I find it new and difficult, I can re-read it several times and not lose anything 2010-09-12T13:48:27 fucawb: I'm sure it's a learning style thing 2010-09-12T13:49:16 is there a python hook for that TCP server? 2010-09-12T13:50:16 Hey... 2010-09-12T13:50:40 Are you sure your architecture is not at 100% of cpu 2010-09-12T13:50:41 hmm, I guess I can just compile the .c file and it should work.. 2010-09-12T13:51:04 elijahbal: you mean the server? 2010-09-12T13:51:07 yes 2010-09-12T13:51:11 you read like 700 words a minute? 2010-09-12T13:51:22 I'm nearly certain it is 2010-09-12T13:51:23 what kind of hardware have you to keep running 1400 bot ? 2010-09-12T13:51:39 elijahbal: not nearly enough yet 2010-09-12T13:51:55 do you plan to make a distributed architecture ? 2010-09-12T13:52:15 Well, add two-three servers could help I think. 2010-09-12T13:52:38 there is some talk of that, but nothing concrete so far as I know 2010-09-12T13:52:57 all my info is 2nd and 3rd hand from asking similar q's 2010-09-12T13:52:58 my bot was submited more than 1h ago 2010-09-12T13:53:06 should just make a package that works like the tcp server and runs the bot in a limited environment 2010-09-12T13:53:07 and no news 2010-09-12T13:53:28 elijahbal: I'm having the same problem. Something has gone wrong with submission IMO. Even when the server was slammed, it was compiling them within 10 minutes. 2010-09-12T13:53:50 AJC: yes, you just pass an exacutable as the last parameter 2010-09-12T13:54:21 Naktibalda, thanks! 2010-09-12T13:54:22 you need to stop the contest and address scaling issues 2010-09-12T13:54:46 it wouldn't be hard to write some kind of distributed architecture using python 2010-09-12T13:54:55 the number of participant has been multiplied by 2 in 24h 2010-09-12T13:54:57 then I can see your bots source? 2010-09-12T13:54:59 yes indeed. 2010-09-12T13:55:02 bobng, stopping the contest wouldn't do much. people would still be coding :-) 2010-09-12T13:55:16 How do I increase the timeout time for my bot? 2010-09-12T13:55:16 bobng: you're right 2010-09-12T13:55:21 yes, but the site obviously doesn't scale at the moment 2010-09-12T13:55:24 they need to remove the bottleneck which is the cpu time for bots and offload it to us 2010-09-12T13:55:34 keeping it running and causing downtime will only drive people away 2010-09-12T13:55:42 just run it like the tcp server 2010-09-12T13:55:44 I dont know what is the bottleneck 2010-09-12T13:55:53 bobng, the online rankings are for informational purposes. there are other ways to test it... 2010-09-12T13:56:07 just advertise on the upload code that you are overloaded would be fine. 2010-09-12T13:56:09 so what AJC? 2010-09-12T13:56:10 elijahbal: 1 cpu at google, each bot fighting another bot for upto 200 turns? 3 mins and using a ton of cpu 2010-09-12T13:56:27 @where tcpserver 2010-09-12T13:56:27 Cyndre: tcpserver = http://www.benzedrine.cx/planetwars/ 2010-09-12T13:56:30 bobng, it's not a show stopper. 2010-09-12T13:56:33 elijahbal, do not worry, I have waited more then 5 hours 2010-09-12T13:56:50 *** ratatata has joined #aichallenge 2010-09-12T13:56:52 what, the site not scaling? of course it is, its a pretty horrible showstopper 2010-09-12T13:57:09 bobng: the site scale fines, running the games doesn't 2010-09-12T13:57:11 400 turns = 400 s if you are at the limit 2010-09-12T13:57:12 bobng, the site or the bot evaluations? 2010-09-12T13:57:22 I define the site as the frontend and the backend 2010-09-12T13:57:27 3mn= 200s. I dont think the majority of bots are so slow 2010-09-12T13:57:51 for a google sponsored game, we could expect something better. 2010-09-12T13:57:55 I've had some mysql connection issues while using the site to do with too many connections 2010-09-12T13:58:05 Google could lend us many computers I think 2010-09-12T13:58:13 so I guess thats not scaling too well, but the backend is the issue. 2010-09-12T13:58:20 bobng, it's open source. simple html caching could fix the front-end. 2010-09-12T13:58:52 So shut down the site, say you have had lots of feedback and are working hard to improve the backend to streamline the submission, compiling and test systems 2010-09-12T13:59:19 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-12T13:59:22 bobng, yeah, shutting down the backend seems sensible if it's that bad. i've not noticed any problems. 2010-09-12T13:59:38 or chage it so the bots are treated as a plugin for a common remote client that handles the bots like the tcp server 2010-09-12T13:59:46 *** danielvf has joined #aichallenge 2010-09-12T14:00:34 keep the current architecture, just create a way to make it distributed and then distribute it across trusted servers. I could supply a vps or two. 2010-09-12T14:00:46 and the only argument against it is peole cheating so far - and if they cheat and use to much resources or play it themselves it will fail when the final testing begins 2010-09-12T14:00:58 and then not participate in the contest bobng? 2010-09-12T14:01:10 and the only one that will be harmed will be them 2010-09-12T14:01:42 I would still participate, the vps would be donated and not used by me. 2010-09-12T14:01:48 ^^ 2010-09-12T14:02:04 would love to find what amstand was posting - if you donate server time you cant competre 2010-09-12T14:02:14 already planning it all out bobng 2010-09-12T14:02:14 ah crud, i didn't meant that 1. increment on my username 2010-09-12T14:02:25 oh well 2010-09-12T14:02:28 I might sketch out some code later 2010-09-12T14:02:30 :D 2010-09-12T14:02:35 twisted.spread seems perfect for this. 2010-09-12T14:02:36 bobng: its already coded 2010-09-12T14:02:52 Cyndre: a distributed backend? 2010-09-12T14:02:53 bobng: they are working on the rules for donating a server to run part of it 2010-09-12T14:03:05 bobng: regretably yes 2010-09-12T14:03:26 Hmm I didn't see that in the svn 2010-09-12T14:04:04 its on the server side sandboxing stuff 2010-09-12T14:04:30 but in my eyes thats not solving the problem as well as the tcp server does 2010-09-12T14:05:47 nusers using x servers to simulate games (distributed system attacking with hardware) nusers using nusers_Comp to connected to remote server and run games (removes constraints on cpu for simulating games) 2010-09-12T14:05:50 a1k0n: :) I cannot resubmit ... mine needs a lot of work. maybe i should just to quit screwing up the rankings (from the lucky draw to start with) 2010-09-12T14:05:54 Cyndre, the TCP server is cool to get an idea of what's going on without actually submitting :-) 2010-09-12T14:06:36 AJC: now imagine if the package was improved and auth added to it 2010-09-12T14:07:19 Cyndre, I haven't tried it yet. what do you need auth for? make sure someone doesn't steal your nick? 2010-09-12T14:07:20 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-12T14:07:26 dmj111: hah, well, i wouldn't. :) the rankings are screwy anyway, there's just not enough games to go around 2010-09-12T14:07:26 AJC: instead of wasting time attacking an exponential scalling problem with linear solutions why not remove the exponential problem 2010-09-12T14:07:46 *** danielvf has joined #aichallenge 2010-09-12T14:08:14 whats the load on the contest server? 2010-09-12T14:08:19 maaxed 2010-09-12T14:08:35 yeah, I'm just enjoying the notoriety while it lasts. Plus, if I don't resubmit, I am less tempted to spam the server. 2010-09-12T14:08:40 I noticed this: http://code.google.com/p/ai-contest/source/browse/trunk/experimental/python_subprocess_cpu_time/master.py 2010-09-12T14:08:51 it looks like a horrible way to do it. forking is not a good way. 2010-09-12T14:08:53 dang, i'm gonna have to do some new css to get the spec to look right on the site 2010-09-12T14:09:05 the reason my bot is still on the tcp server is just because i know it is more fun to have more that 5 bots active over there. 2010-09-12T14:09:12 what is cameron.jp's handle? 2010-09-12T14:09:26 bobng: j3camero 2010-09-12T14:09:27 j3camero I think 2010-09-12T14:09:41 j3camero: ping 2010-09-12T14:10:10 *** TheMadH4x0r has joined #aichallenge 2010-09-12T14:10:20 Cyndre, so as part of the contest entry, you'd be required to run your bot a couple hundred times from your own machine? 2010-09-12T14:10:41 AJC: Start a program and leave it running to help build stats - you need x games to qualifyh 2010-09-12T14:11:49 AJC: and then the final results are done on googles main machine 2010-09-12T14:12:56 AJC: 300 bots /1 comp(maybe running 5 games) for 30 games every 2 mins or so - or 300 bots connecting in like a mmorpg player and running 150 games every 2 mins 2010-09-12T14:13:15 or require personal identification so programs can be run on bare metal, trusting the programmers 2010-09-12T14:13:20 the average game time is 3 mins? 2010-09-12T14:13:25 thats quite a lot... 2010-09-12T14:13:32 max turns 2010-09-12T14:13:40 is the limiting factor there 2010-09-12T14:13:47 if each bot takes 1 second per turn and it lasts 200 turns, yes. so far bots have been much faster than that 2010-09-12T14:14:08 No bot should take a second of CPU time a turn 2010-09-12T14:14:22 bobng, those are the rules. 2010-09-12T14:14:36 with tron the smallest maps were 15*15 so there were always 200 turns and the bots that were any good used their whole second to search the game tree 2010-09-12T14:14:42 AJC: exactly - build the system around the rules, not the rules around the system 2010-09-12T14:15:14 (well, more like 150 squares were open but still) 2010-09-12T14:15:15 aik0n: yea, when I started utilizing my time properly I went from 300's to to 100 - 200 2010-09-12T14:15:28 Cyndre, 1s makes sense. either go for 30/60 turns per second, game like, or set it to something reasonably high -- like 1s :-) 2010-09-12T14:16:00 or just run more games in parallel. less cpu cycles/second, but the rules are still fullfilled... 2010-09-12T14:16:25 dhartmei: but then the games take 2000 seconds instead of 200 and we run 10 times as many at once 2010-09-12T14:16:27 yeah, that ended up happening last time and kinda degraded the effectiveness of a lot of the bots 2010-09-12T14:16:37 real time, not cpu time 2010-09-12T14:17:11 like I said - next step should be defining environment bot is placed in 2010-09-12T14:17:26 you get this much cpu and this much ram 2010-09-12T14:17:40 so, if you have 500 programs, how many games per minute do you have to do to make them happy? then calculate backwards :) 2010-09-12T14:18:01 one game every 10 minutes makes everyone happy? 2010-09-12T14:18:08 You are prematurely optimizing tbh 2010-09-12T14:18:10 thats designing to fail though - what happens at 1000 programs? 2010-09-12T14:18:15 one game per program, that is 2010-09-12T14:18:24 most likely, though once you have 100 games or so the returns diminish 2010-09-12T14:18:58 i didn't know (last time) that everything was done on a single PC-type server 2010-09-12T14:19:04 As I see it each game has 3 processes - a manager and two competitors? 2010-09-12T14:19:08 is that correct? 2010-09-12T14:19:11 yes 2010-09-12T14:19:17 i thought they had a couple last time 2010-09-12T14:19:22 at any rate it was an 8-core xeon 2010-09-12T14:19:23 server and 2 clients 2010-09-12T14:19:30 So why not reduce the number of managers. 2010-09-12T14:19:38 bobng: because then you get less games 2010-09-12T14:19:40 You can do that without reducing the number of clients 2010-09-12T14:19:54 bobng: the problem is you only get x games/10 minutes no matter how you arrange things 2010-09-12T14:19:54 Is it a comparable maching to the 8-core xeon this year? 2010-09-12T14:19:55 in other words, use benzedrine.cx's server as the tournament server 2010-09-12T14:19:57 move the database and web server onto a different server, then check performance again, that's what i would do 2010-09-12T14:19:59 no 2010-09-12T14:20:07 thats what Im pusshing for 2010-09-12T14:20:27 no, but like, the clients run on their end, they are just managed by a single server which hosts the games 2010-09-12T14:20:39 yes 2010-09-12T14:20:40 i don't know how much overhead the engine is though in comparison to everything else 2010-09-12T14:20:42 if the database read queries and web server are slow, that's not as bad if there's still games being played 2010-09-12T14:20:48 The db and web server should really be on another server reguardless 2010-09-12T14:20:56 greghaynes: yes indeed 2010-09-12T14:21:14 Why not use python's twisted library to manage the clients. Twisted has asynchronous stdout packagers that might be very useful in a situation like this. You could spawn one manager and have it handle as many client pairs as you like 2010-09-12T14:21:28 Would easily handle the timeout as well 2010-09-12T14:21:43 bobng: sure, or just use the server dhartmei already built which does all that 2010-09-12T14:21:48 bobng: organizers are against us running our own client 2010-09-12T14:21:56 So then why do you need 1 server per 2 clients 2010-09-12T14:22:03 bobng: at any rate, it decreses some of the overhead but not much compared to clients that (in theory) use one CPU-second per turn 2010-09-12T14:22:16 Worst case scenario. 2010-09-12T14:22:41 Make the manager time the average time taken per turn and submit into a db 2010-09-12T14:22:50 and that solves another issue - multiple bot submitting 2010-09-12T14:23:01 because I handle the cpu time used by my bots I could run 30 if I wanted 2010-09-12T14:23:03 so you get more games if your bot is faster or what? 2010-09-12T14:23:07 bots that consistently use a lot of CPU time can have a reduced number of games 2010-09-12T14:23:16 I mean use a *lot* of CPU time 2010-09-12T14:23:31 You could also make some interesting stats using that data 2010-09-12T14:23:39 what if you use 1 CPU-second per turn, but trounce every opponent in like 30 turns 2010-09-12T14:23:52 a1k0n: then take that into account 2010-09-12T14:24:03 heh, fear complexity 2010-09-12T14:24:03 vs my bot that uses 100 ms per turn and wins after 300 turns 2010-09-12T14:24:07 bots that use a lot of CPU, loose all the time or take a long time to win 2010-09-12T14:24:17 *** hamstah has joined #aichallenge 2010-09-12T14:24:27 hey 2010-09-12T14:24:31 will not be preferable to bots that win fast and use a small amount of CPU time 2010-09-12T14:24:40 hi hamstah 2010-09-12T14:24:48 you could even email the owner asking him to optimize his code a bit. 2010-09-12T14:25:06 a1k0n: ^ 2010-09-12T14:25:07 bobng: whats the point of rules if they get changed after they have been put out? 2010-09-12T14:25:08 bobng, wanna ban loosy code? 2010-09-12T14:25:11 Is the server box much weaker than last years? 2010-09-12T14:25:22 bobng: we could also limit the bots to 10ms per turn and that would solve every problem 2010-09-12T14:25:24 hamstah: not ban, just try and avoid it. 2010-09-12T14:25:31 greghaynes: it seems it's about 1/4th the machine as last year 2010-09-12T14:25:41 (and yet jeff wanted 20x as much traffic) 2010-09-12T14:25:52 You guys are fitting all of your assumptions around the worst case scenario of each bot using 1 second per turn 2010-09-12T14:26:01 bobng: did you compete in the tron contest? 2010-09-12T14:26:08 bobng, everyone who knows what they're doing are designing to use 1s 2010-09-12T14:26:28 *** coventry` has quit IRC (Remote host closed the connection) 2010-09-12T14:26:28 Rather than make a huge complexity increase, why not find someone in the area willing to lend a box with some reasonable power and stick it on the same lan as the current server, use both for running the system 2010-09-12T14:26:31 AJC: not necessarily 2010-09-12T14:26:32 on a quater of the hardware. 2010-09-12T14:26:44 it may make sense to just make the rule 100ms instead 2010-09-12T14:26:57 100ms instead of 1s? 2010-09-12T14:26:59 greghaynes: someone willing to do that, and willing to be disqualified 2010-09-12T14:26:59 google is sponsoring the contest, so why not ask them to provide free google app-engine ressources? 2010-09-12T14:27:04 You are telling me that you expect to have a responsive server that can handle a lot of games when each game is going to take a second per turn 2010-09-12T14:27:08 jesus, that's a big reduction 2010-09-12T14:27:10 bobng: no, its called designing the system around the rules - MaxTurnTime = 1, MaxTurns = 200, MaxTime = MaxTurnTime * MaxTurns 2010-09-12T14:27:13 at least 0.25 would be nice 2010-09-12T14:27:16 antimatroid, exactly 2010-09-12T14:27:31 you can't have it all. either lower the turn limit or penalize the people who use a lot of CPU 2010-09-12T14:27:34 bobng: i don't think anyone here honestly expects that with the things being as they are 2010-09-12T14:27:39 or scale up your hardware. 2010-09-12T14:27:44 i have a better solution 2010-09-12T14:27:49 nobody actually running the contest is here 2010-09-12T14:27:56 run the final tournament for however long it takes with a live updating scoreboard 2010-09-12T14:28:00 problem solved 2010-09-12T14:28:17 If I wasnt doing o-k in the tourney id be up for donating a 4 core box... 2010-09-12T14:28:25 and top 50 round robin to top it off 2010-09-12T14:28:34 Maybe sabotoge my bot 2010-09-12T14:28:42 can somebody share tcp server plug built for Win? 2010-09-12T14:28:45 ggreghaynes: which only makes the problem somewhat go away 2010-09-12T14:28:47 Anyone here using C# that could answer a few questions? :D 2010-09-12T14:28:54 In any case collecting CPU usage info for each bot would be interesting for data mining purposes 2010-09-12T14:29:14 greghaynes, now if everyone had client that connects to a central server the problem is gone - how many games is the tcp server handling vs the contest server? 2010-09-12T14:29:29 Cyndre: Kinda, theres some new issues that come up with that system 2010-09-12T14:29:37 greghaynes, like? 2010-09-12T14:29:50 network issues, first of all 2010-09-12T14:29:54 *** ratatata has quit IRC (Quit: Page closed) 2010-09-12T14:30:04 The time limit is not really enforcable then 2010-09-12T14:30:17 People with shoddy connections are also dq'd 2010-09-12T14:30:17 if you don't have stable bandwidth - you're out 2010-09-12T14:30:18 it is, up to some large constant factor 2010-09-12T14:30:21 greghaynes: time limit is hadled via the program you download 2010-09-12T14:30:24 if you have 8 cores, and you want to provide cpu cycles equal to what the programmers have at home, it follows you can only run 8 games every 200 seconds, that's 1 game per minute. now you have 500 participants. how long is the average time between games of one participant? 2010-09-12T14:30:44 just not enough :) 2010-09-12T14:30:53 dhartmei: 4 games every 200 seconds 2010-09-12T14:30:56 greghaynes: ping running in background and ping time added in (if you change the program your side to cheat you will just lose durring finals and be replaced by next in list) 2010-09-12T14:30:58 right 2010-09-12T14:31:26 and then I can solve the problem for $20 a month by getting a dedicated server online as well 2010-09-12T14:31:31 People who live across continents from the box are very liable to have massive lag spikes 2010-09-12T14:31:59 greghaynes: it doesn't matter because you can run games after they are finished and the timeout is handled client side 2010-09-12T14:32:17 Yea, but your ping precision is > 1 sec 2010-09-12T14:32:31 but the client enforces timeout, the ping time is leniency 2010-09-12T14:32:49 if your running 150 games wasting 2 secs a turn doesnt matter 2010-09-12T14:33:05 Ah, thats pretty easy to circumvent, and that would also require a closed source client 2010-09-12T14:33:11 like I said 2010-09-12T14:33:20 if they circumvent it looks good going into finals 2010-09-12T14:33:31 then fails miserably - the next person in line is moved up and they are dqed 2010-09-12T14:33:40 I would not be too happy with needing to run a closed source client 2010-09-12T14:33:47 closed source client? 2010-09-12T14:33:52 why would it be closed source? 2010-09-12T14:34:07 Well, if its open source circumventing is hilariously trivial for anyone who can program 2010-09-12T14:34:12 Making it closed source wouldn't really help that much anyway, just an annoyance 2010-09-12T14:34:22 ermau: Yea, its all you can do though 2010-09-12T14:34:24 greghaynes, you mean some kind of phylosophical complaint with closed source? here it would make entirely sense. 2010-09-12T14:34:24 greghaynes: that doesnt give them any advantage in the finals which is what counts 2010-09-12T14:34:54 greghaynes: in fact screwing with the software to make their bot better would pretty much guarentee dq when finals start 2010-09-12T14:35:02 *** Queue29 has joined #aichallenge 2010-09-12T14:35:19 greghaynes: and I would sooner run 30 games an hour compared to 1 every 2 hours currently 2010-09-12T14:35:40 *** DirtyKeyboard has joined #aichallenge 2010-09-12T14:35:49 What about people who are using 1sec cpu at home on a more powerful box than the server for finals? 2010-09-12T14:36:05 the client sandbox's the bot and only gives it x ram and x cpu 2010-09-12T14:36:14 guarenteed to run on final box 2010-09-12T14:36:24 AJC: Thats one reason, its also a security issue among other things 2010-09-12T14:36:26 provide some benchmark code so people can _voluntarily_ synchronize, only run the finals on internal servers? 2010-09-12T14:36:34 exactly 2010-09-12T14:36:37 greghaynes: your program should check the time periodically if you're worried about this 2010-09-12T14:36:42 allow people to cheat during pre-finals, they only deceive themselves 2010-09-12T14:36:46 yes 2010-09-12T14:36:50 dhartmei gets it 2010-09-12T14:37:00 eek, coding a sandbox for all platforms would be a fairly difficult task 2010-09-12T14:37:04 DirtyKeyboard: Good point 2010-09-12T14:37:12 no, the sandbox is only on the internal servers 2010-09-12T14:37:27 well I would want a client that limits the bots ram and cpu time for benchmarking 2010-09-12T14:37:27 people don't need to sandbox their own code 2010-09-12T14:37:40 I would want to limit the avaliable resources to that of the final sandbox 2010-09-12T14:37:41 exactly, voluntarily, so you measure with equal sticks 2010-09-12T14:37:54 anyone have problems with tcp.c? execv: python ../python_starter_package/MyBot.py: failed 2010-09-12T14:38:10 chmod +x MyBot.py ? 2010-09-12T14:38:12 make a wrapper shell script 2010-09-12T14:38:23 the execv argument must be a single file name 2010-09-12T14:38:36 dhartmei, ah. and with the full path? 2010-09-12T14:38:49 relative path is ok 2010-09-12T14:38:58 relative works, just give it something you can ./ 2010-09-12T14:39:00 AJC: make sure the python file is executable and the first line is this: #!/usr/bin/env python 2010-09-12T14:39:14 *** mog has left #aichallenge ("Leaving") 2010-09-12T14:39:20 but yea, why dont they allow multiple bots per person? Problem with processing games - with a remote client you could run 5 bots 2010-09-12T14:39:25 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-12T14:39:38 *** danielvf has joined #aichallenge 2010-09-12T14:39:42 AJC: and then the command would be "./MyBot.py" instead of "python ./MyBot.py" 2010-09-12T14:40:16 Given that we have a 1s limit for this game the idea seems to be that you dont really have a hard time limit, just dont use up all the time in the world 2010-09-12T14:40:23 DirtyKeyboard, greghaynes, Cyndre thanks. 2010-09-12T14:41:01 greghaynes: true, but as a backend designer the system should expect people to use every second of time 2010-09-12T14:41:03 Which is a big reason the tcp server/client system would work 2010-09-12T14:41:08 Yea 2010-09-12T14:41:40 and then with 3 servers they can simulate identical conditions to run finals in, except clients are trusted 2010-09-12T14:41:51 Someone should really add at least a little authentication to that thing ;) 2010-09-12T14:42:01 the tcp server/client system would not benefit those who cannot leave their systems on 24/7 to compete. 2010-09-12T14:42:18 I was thinking a private key generated on download and inserted into directory 2010-09-12T14:42:34 bobng: if you run your home comp for 1h you get 20 games at 3 minutes a game 2010-09-12T14:42:44 bobng: I got 10 games in 20 hours yesterday 2010-09-12T14:43:12 Maybe the official contest runners could just be a client that accepts uploads and runs games periodicly for each person uploading 2010-09-12T14:43:13 bobng: and it would - within 10 minutes you could start work on your next bot after watching it perform in realtime 2010-09-12T14:43:39 They would lose a ton of their load because many with faster machines would want to run at home to get more games in 2010-09-12T14:43:45 bobng: currently my uploaded bot broke top 100 but I have no idea what version it even is now 2010-09-12T14:43:48 So, I don't really like ShowGame.jar, does anyone have a better alternative? 2010-09-12T14:44:20 ulope, did you try your python bots on the TCP server? apparently hamstah says the default bot issues invalid orders... 2010-09-12T14:44:22 dhartmei: any chance of working on a client for yours that can browse for a file, open it, play it etc and visualize in rt? 2010-09-12T14:45:14 Wow, now that im actually getting to play some games my rank went way up :) 2010-09-12T14:45:19 would be os dependant, no? 2010-09-12T14:45:32 dhartmei: Use Qt :) 2010-09-12T14:45:33 developers, review this patch please http://code.google.com/p/ai-contest/issues/detail?id=135 2010-09-12T14:45:58 dhartmei: even a windows client that they can easily add a bot and test with it repeatedly (a exe file they can just run) <- would probably help alot with amount of bots 2010-09-12T14:46:25 I'm working on a windows version of the tcp server client 2010-09-12T14:46:41 are you going to make it graphical or cml? 2010-09-12T14:46:44 last time someone posted one to the forums (not graphical, though, iirc), myself i don't even have a windows box... :) 2010-09-12T14:46:44 If you want me to make a Qt dialog for that I can, will take all of 2 seconds 2010-09-12T14:46:45 cl 2010-09-12T14:49:16 *** danielvf_ has joined #aichallenge 2010-09-12T14:49:16 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-12T14:49:16 *** danielvf_ is now known as danielvf 2010-09-12T14:50:07 greghaynes: sorry for my noobness but Qt? (think its an interface maker for python but not sure) 2010-09-12T14:50:41 Its a widget toolkit, in linux land everything is either Gtk or Qt (besides some others rarely used) 2010-09-12T14:50:57 Qt has the advantage of also working on the major other OS's 2010-09-12T14:51:01 okay, so you could make like a bot loader, runner, game visualizer for the tcp server? 2010-09-12T14:51:08 greghaynes: Erm, so does Gtk 2010-09-12T14:51:10 *** mietek has joined #aichallenge 2010-09-12T14:51:13 Hello 2010-09-12T14:51:18 *** Vio has joined #aichallenge 2010-09-12T14:51:24 Anyone using Haskell? 2010-09-12T14:51:41 yes 2010-09-12T14:51:49 *** JCS^ has joined #aichallenge 2010-09-12T14:51:51 hi mietek 2010-09-12T14:52:15 Is it possible to submit a Haskell solution yet? 2010-09-12T14:52:37 not sure on that myself - jmcarthur? 2010-09-12T14:52:49 Interesting, gtk on win works just seems a little scary 2010-09-12T14:53:02 mietek: it is. it must work with ghc 6.8.2, and your main must be called MyBot.hs 2010-09-12T14:53:05 stick with what your comfortable with 2010-09-12T14:53:20 6.8.2, oh my. 2010-09-12T14:53:24 yeah i know... 2010-09-12T14:53:38 i tried to get a newer version working, but ubuntu 8.04 just won't take to it :\ 2010-09-12T14:53:59 if we get our sandboxes fixed up we might be able to just do arbitrary binaries though 2010-09-12T14:54:11 i'm pushing for that, but i don't know if it will work 2010-09-12T14:54:32 Oh, you're part of the team? 2010-09-12T14:54:32 hehe: Im pushing for client server..... 2010-09-12T14:54:53 I don't suppose anyone has looked at upgrading Mono? 2010-09-12T14:55:40 mietek: the team got pretty big when they decided to open source it :) 2010-09-12T14:55:42 jmcarthur, would not it be easier to persuade them to update to newer version of ubuntu? 2010-09-12T14:55:55 Palmik: dist-upgrades tend to break things 2010-09-12T14:56:09 would have to be a fresh install 2010-09-12T14:56:13 Palmik: we might have to go to a distributed architecture for the bots anyway, and we could do newer stuff then 2010-09-12T14:56:22 Ive tried 1 dist-upgrade and I will not try one agani 2010-09-12T14:56:28 Cyndre: same here 2010-09-12T14:56:31 *** TheMadH4x0r has quit IRC (Ping timeout: 252 seconds) 2010-09-12T14:56:43 Cyndre: actually, the next time i need to upgrade my htpc, i'm switching it to arch. i'm sick of fighting ubuntu 2010-09-12T14:56:54 jmcarthur: I recently went to centOs 2010-09-12T14:56:54 Given that we seem to have finally started playing some games I would hope they dont decide to break the server.... 2010-09-12T14:57:01 i only used ubuntu in the first place so i could use the (then) newer version of xbmc 2010-09-12T14:57:07 Cyndre: ugh, i hate centos too 2010-09-12T14:57:11 I always fresh install, it's no bother thanks to partitions :) 2010-09-12T14:57:18 the entire red hat family needlessly changes things 2010-09-12T14:57:19 jmcarthur: thanks for putting in the work. 2010-09-12T14:57:37 jmcarthur: so far it hasnt given me grief, but havent done anything major with it yet (want to play with the clustiner/vm stuff one day) 2010-09-12T14:57:41 Cyndre: i use arch on my personal machines and love it 2010-09-12T14:57:59 jmcarthur: Ive heard a lot of good things about arch, and not just recently talking years ago 2010-09-12T14:58:29 There has to be some function describing the time a meeting of programmers on irc devolves into a distro discussion 2010-09-12T14:58:41 lmao 2010-09-12T14:58:45 Cyndre: it's basically like gentoo but actually stays out of your way, is way faster to set up, and does binaries-by-default 2010-09-12T14:58:53 heh 2010-09-12T14:59:03 slackware for the win! 2010-09-12T14:59:13 arch has a lot of inspiration from slackware, actually 2010-09-12T14:59:18 we have a machine at work still running slackware 2010-09-12T14:59:21 the guy that started arch came from slackware 2010-09-12T14:59:33 my wife and daughter use slack 2010-09-12T14:59:40 my wife uses arch 2010-09-12T14:59:48 I can't complain about my openSUSE either :) 2010-09-12T14:59:55 the guy I work with used to outsource linux work, and the guy that did the outsourcing used slackware so the guy I work with could never get over his dependancy on said linux guy 2010-09-12T15:00:06 None of my relatives uses linux xD 2010-09-12T15:00:58 Palmik, change you relatives. 2010-09-12T15:01:03 +r 2010-09-12T15:01:10 Palmik: Well i have quite a few that don't too. And they say things like --"but windows is so much easer to use. btw can you come round and fix my computer" 2010-09-12T15:01:12 but yea I need to work on my bot :) - it now abandons when theres no chance of reinforcements reaching in time, or not enough reinforcements 2010-09-12T15:01:21 Or I could ninja swap their OSes :D 2010-09-12T15:01:32 I've heard about people pulling that off 2010-09-12T15:01:50 i'm more sad than happy when i hear about that working 2010-09-12T15:01:50 time to get the reinforcements being dispatched 2010-09-12T15:02:05 ? 2010-09-12T15:02:14 jmcarthur: I can see it working if it was a slow process 2010-09-12T15:02:57 I've heard of people doing it in the middle of the night, setting the desktop and aliases the same, and having it work 2010-09-12T15:02:58 i'm talking about when people just don't even notice that their machines are completely different all of a sudden 2010-09-12T15:03:09 it makes me lose some faith in humanity 2010-09-12T15:03:12 javagamer_Cpp, hehe 2010-09-12T15:03:18 as long as the icons are the same place and it works 80% the same I can see it working 2010-09-12T15:03:23 the reactions must have been priceless :) 2010-09-12T15:03:26 and the same back ground 2010-09-12T15:03:34 Palmik, only if they notice 2010-09-12T15:03:59 One of my friends did it to his parents, it took a while before one noticed, and he noticed b/c it was working smoother 2010-09-12T15:04:02 people are so dumb - it could be an entirely new process but if you ensure them thats how they always did it before...... 2010-09-12T15:04:03 They might notice the speed up :P 2010-09-12T15:04:32 His parents made him change it back, but after a day or two, they had him switch it to Linux again 2010-09-12T15:04:51 Also, solitaire is much nice here on my machine then the pixelated one on XP :D 2010-09-12T15:04:55 Im pretty sure I clicked microsoft works before, no no, its always been openOffice Calc - a little to much partying on the weekend? 2010-09-12T15:04:55 *nicer 2010-09-12T15:04:56 bah, my parents use linux 2010-09-12T15:05:16 I told them I dont do win tech support ;) 2010-09-12T15:05:43 same 2010-09-12T15:06:02 greghaynes: Me too. "i will help, but only if its unix" --however no one taken me up on the offer, which is also fine 2010-09-12T15:06:13 I was actually thinking of starting a computer buisness selling computers that included 10 hours of class based instruction based on a linux distro ( have it like $1500 for comp and get $100 per hour for training) 2010-09-12T15:06:20 Talking about support :P http://theoatmeal.com/comics/computers I'm sure you know this :) 2010-09-12T15:06:24 *** kozlovsky has joined #aichallenge 2010-09-12T15:06:30 Cyndre: Google for freegeek, its a place that does that here 2010-09-12T15:06:39 greghaynes: cool 2010-09-12T15:06:47 *** Obeleh has joined #aichallenge 2010-09-12T15:06:59 Although they give the machines out for free, the training involves you building your own computer from spare parts ;) 2010-09-12T15:07:00 *** troutwine has joined #aichallenge 2010-09-12T15:07:05 greghaynes: wouldnt pick a geeky name - would go with computer center or something drab so normal people come 2010-09-12T15:07:06 Hey all 2010-09-12T15:07:22 *** ratatata has joined #aichallenge 2010-09-12T15:07:36 greghaynes: mine would be 10 hours in classroom and you get the machine with ubunto or whatever on it 2010-09-12T15:08:02 thats what linux needs to kick windows ass 2010-09-12T15:08:17 The bots run on a Posix system, right? 2010-09-12T15:08:29 Yep 2010-09-12T15:08:36 Well, as posix as linux is 2010-09-12T15:08:39 your in vancouver? 2010-09-12T15:08:45 Portland 2010-09-12T15:08:52 *** deepblue has joined #aichallenge 2010-09-12T15:09:00 Close enough. 2010-09-12T15:09:00 top 3 google links are for vancouver for freegeek 2010-09-12T15:09:01 back :) 2010-09-12T15:09:05 heh 2010-09-12T15:09:10 I miss Portland. 2010-09-12T15:09:16 Cyndre: Whoa, I didnt realize they had one 2010-09-12T15:09:25 troutwine: :) You lived here? 2010-09-12T15:09:29 ty - looks cool 2010-09-12T15:09:37 i think i would like to live in or near portland 2010-09-12T15:09:52 Its nice, ill be moving soon though 2010-09-12T15:09:59 greghaynes: I did, for six years. Took a degree there and everything. I moved away to Rochester, NY in the past month. 2010-09-12T15:10:10 see, google should give the aicontest 20 computers and donate to freegeek at the end of the contest 2010-09-12T15:10:22 jesus christ 2010-09-12T15:10:26 my vps is still down 2010-09-12T15:10:30 way to go steadfast 2010-09-12T15:10:33 Ah, sounds like me, im here until my degree is done then moving off for work 2010-09-12T15:10:35 *** gilaniali has joined #aichallenge 2010-09-12T15:10:42 What exactly is Google sponsoring with this? 2010-09-12T15:10:52 its name, hehe 2010-09-12T15:10:56 its like $400 a box, whats $8k? 2010-09-12T15:10:58 https://support.steadfast.net/index.php?_m=news&_a=viewnews&newsid=285 2010-09-12T15:11:30 its like the secratarys lunch budget for googleplex x for 1 day 2010-09-12T15:11:31 maybe someone should make a suggestion to Google 2010-09-12T15:11:52 Im sure it has come up 2010-09-12T15:12:01 There are a couple google competitors 2010-09-12T15:12:04 sometimes these things don't happen because nobody in the right place thinks about it 2010-09-12T15:12:27 Yeah, could be everyone's assumed someone else has mentioned it to them 2010-09-12T15:12:27 don't just give them the email addresses for the recruiters, trade them for some nice hardware :) 2010-09-12T15:12:28 Ill write a blog about how awsome the contest is, if it had the resources having a name like google behind it makes you think it would 2010-09-12T15:12:43 amazon would have nice clusters for this, no? 2010-09-12T15:13:30 Cyndre: Just keep in mind they could just as easily distance themselves from the contest if they think they are getting bad PR as a resuly 2010-09-12T15:13:41 greghaynes: was thinking that myself to 2010-09-12T15:14:33 google also has the technology to run untrusted i386 binary code natively, if you haven't read that paper. this would be an awesome showcase for it... 2010-09-12T15:14:43 but really if all they are doing is providing a name and a computer I wouldnt give my kids to play on...... 2010-09-12T15:15:00 really? thats an intresting blog post 2010-09-12T15:15:02 Im not sure they are providing the computer... 2010-09-12T15:15:07 problem + solution = win 2010-09-12T15:16:12 http://google-code-updates.blogspot.com/2008/12/native-client-technology-for-running.html 2010-09-12T15:16:25 ty, saved me some googling 2010-09-12T15:16:56 thats a recipe for disaster heh 2010-09-12T15:16:59 *** td123 has quit IRC (Quit: WeeChat 0.3.3) 2010-09-12T15:17:01 they provide a patched gcc, which produces binaries that can be verified to adhere to some sandboxing standards 2010-09-12T15:17:12 along the lines of Google ai contest a win again. More backend power. Maybe if they used your article (courtesy of dhartmei or your real name if so inclined) was used it would be a win for both 2010-09-12T15:17:13 hah, security risks are a great way to create a bad name for yourself 2010-09-12T15:17:18 so you can run such binaries without risk 2010-09-12T15:17:24 weee too many connections 2010-09-12T15:17:31 *** ratatata has quit IRC (Quit: Page closed) 2010-09-12T15:17:36 *** td123 has joined #aichallenge 2010-09-12T15:17:53 r u aware the website is down? 2010-09-12T15:18:17 wooo, php 2010-09-12T15:18:17 need to increase the mysql connections in php.ini 2010-09-12T15:19:24 Cyndre, more to the point, cache common pages and those viewed by people not logged in. should be no DB access there. it's a simple .htaccess trick. 2010-09-12T15:20:59 i hope you didn't just include mysql_login in every single php page.. 2010-09-12T15:21:52 *** Palmik has quit IRC (Remote host closed the connection) 2010-09-12T15:21:59 Cyndre: http://nativeclient.googlecode.com/svn/data/docs_tarball/nacl/googleclient/native_client/documentation/nacl_paper.pdf 2010-09-12T15:22:08 that's the juicy details 2010-09-12T15:22:17 oh wait, you did :( 2010-09-12T15:22:28 read that with this scenario in mind, i think it might fit perfectly together :) 2010-09-12T15:22:55 caytchen: whats wrong with that? 2010-09-12T15:23:21 Cyndre, hehe the site goes down. 2010-09-12T15:23:36 *** troutwine has left #aichallenge 2010-09-12T15:23:39 AJC: should have code making sure the db is avaliable 2010-09-12T15:24:04 greghaybes: I think donating boxes to FreeGeek is a great idea! They are partically a franchise now: http://www.freegeekarkansas.org/ (I run that site.) 2010-09-12T15:24:09 *** bss03-AFK is now known as bss03 2010-09-12T15:24:23 wow 2010-09-12T15:24:28 They are taking over 2010-09-12T15:24:57 Thats awesome to hear 2010-09-12T15:25:23 *** dhartmei has quit IRC (Quit: good night) 2010-09-12T15:25:42 http://ai-contest.com/visualizer.php?game_id=4448691 woot 2010-09-12T15:25:44 gota new game 2010-09-12T15:27:46 http://www.benzedrine.cx/planetwars/canvas?game_id=1284319077|deepblue-test|dhartmei 2010-09-12T15:27:59 :) 2010-09-12T15:28:41 deepblue: do you search for planets based on dist/growthRate/costToTakeOver 2010-09-12T15:29:12 well 2010-09-12T15:29:22 * deepblue deepblue just knows 2010-09-12T15:29:24 ;) 2010-09-12T15:29:36 in that one, I see you send just enough ships to take planet x, and enough ships to take planet y with enough for planet z left over 2010-09-12T15:29:48 So, how can I use that TCP thing? The example bots suck big time 2010-09-12T15:29:53 pretty cool eh :) 2010-09-12T15:30:03 Hes just turned the game into a modified version of chess 2010-09-12T15:30:16 Pointers shouldn't move around on their own, right? 2010-09-12T15:30:37 RainCT: Read the fine source 2010-09-12T15:30:38 btw, why are you guys sending around so many small fleets at once and to the same planet? 2010-09-12T15:30:39 very - its like an iterative search for 5 planets with x planets within 3 distance or something 2010-09-12T15:30:44 I suck at chess btw, get to excited when things seem to work out and then I lose 2010-09-12T15:30:51 RainCT: Can you beat RageBot all the time? And DualBot? 2010-09-12T15:31:04 RainCT: They actually perform fairly well in practice. 2010-09-12T15:31:11 deepblue: am I close? 2010-09-12T15:31:24 kind of :) 2010-09-12T15:32:03 bss03: DualBot yes, haven't tried Rage. /me will try the "play 100 times against every example bot" script now 2010-09-12T15:32:31 I saw an interestin trick by a couple people on the top, since everyone seeems to send numfleets + 1, some people are sending numfleets + 2 2010-09-12T15:32:36 deepblue: saw this one where you lose against 1 ship : http://www.benzedrine.cx/planetwars/canvas?game_id=1284317604|OminousBirthdayCake|deepblue-test ? :) 2010-09-12T15:33:13 RainCT, try to improve your average turns then on every bot 2010-09-12T15:33:16 @topic 2010-09-12T15:33:16 rogue780: Official Google AI Challenge: http://ai-contest.com/ || http://contestbot.hypertriangle.com/ || No stupid questions. Don't ask to ask. || Developers wanted, contact j3camero. || Sorry about the slowness. || Latest Idea, decentralization: http://tinyurl.com/decentralizedai 2010-09-12T15:33:24 hehe 2010-09-12T15:33:39 ouch 2010-09-12T15:33:43 edcba: my bot didn't care to finish him off :D 2010-09-12T15:34:00 Mine likes to do that also 2010-09-12T15:34:12 Your algo seems similar to mine, just tweaked better 2010-09-12T15:34:15 especially a zero growth planet... why care? 2010-09-12T15:34:22 *** jerkface03 has quit IRC (Ping timeout: 276 seconds) 2010-09-12T15:34:32 Thats definitely what your algo is thinking ;) 2010-09-12T15:34:53 he's like 'let him come to me!' 2010-09-12T15:34:58 That multiply by 0 issue can mess you up 2010-09-12T15:35:07 *** jerkface03 has joined #aichallenge 2010-09-12T15:35:11 more like when you do some division :) 2010-09-12T15:35:20 heh, especially then 2010-09-12T15:35:20 was going to say 2010-09-12T15:35:43 could be a stratedy to take that planet just to see if the opponent will crash ;) 2010-09-12T15:35:44 Is the tcp prog compiled for Windows available anywhere? I really don't want to have to download and install gcc/Cygwin/etc :-) 2010-09-12T15:36:09 need to make that easier and more accessable 2010-09-12T15:36:15 Doctor: use a better OS. :P I kid, I kid 2010-09-12T15:36:18 then I wont lose so much on the tcp server 2010-09-12T15:36:33 yea yea. im working on a qt client as we speak 2010-09-12T15:36:40 cool 2010-09-12T15:36:47 Well, VMS stopped booting on this machine. 2010-09-12T15:36:58 RMS is raging as I create a Qt dialog in Gnome 2010-09-12T15:37:11 Doctor: I haven't tried that java client that's in that forum thread, have you tried that out? 2010-09-12T15:37:18 see the reason my bot loses isnt that I suck at this, its that we dont have enuogh people with lower skills then me on the tcp server 2010-09-12T15:37:24 http://www.benzedrine.cx/planetwars/canvas?game_id=1284319319|a1k0n-0.2|deepblue-test 2010-09-12T15:37:33 seems like my new one needs some more tweaking 2010-09-12T15:37:39 Cyndre: idk, my bot sucks at the moment. :P 2010-09-12T15:37:42 Cyndre: don't worry I'll join sooner or later :-D 2010-09-12T15:38:00 Cyndre: My only working bot has some serious bugs and logic flaws.. I'll run it some more if you want :P 2010-09-12T15:38:10 Pointer to the forum thread? 2010-09-12T15:38:21 DirtyKeyboard: mine on the contest serers is doing good top 100, but I have no idea what version that is 2010-09-12T15:38:28 retybok: yay 2010-09-12T15:38:35 Oh yeah, fine, put me against wtfbbq 2010-09-12T15:38:36 ermau: lmao 2010-09-12T15:38:39 * ermau rolleyes 2010-09-12T15:38:51 Doctor: http://ai-contest.com/forum/viewtopic.php?f=18&t=424 2010-09-12T15:39:02 I need to finish sending reinforcements, I think that will mke a huge difference 2010-09-12T15:39:04 DK: ty 2010-09-12T15:39:07 Just tried with the script to play 100 times against each example bots, and I've won all the games. The surprising thing is that I haven't even seriously started writing anything, I'm just indiscriminately conquering anything I can conquer sending the exact number of ships 2010-09-12T15:39:22 RainCT: Yup 2010-09-12T15:39:25 RainCT: Example bots suck 2010-09-12T15:39:41 Hell, even put it on the real contest server, most of the bots running there suck too 2010-09-12T15:39:46 RainCT: I have testing like that, but I put my old bot versions in there too. ;) 2010-09-12T15:39:48 including that damn ragebot and dualbot? 2010-09-12T15:40:09 Cyndre: Yes 2010-09-12T15:40:15 as I have been adding things I went from a 90% success to a 65% success 2010-09-12T15:40:19 DirtyKeyboard: That's a good idea, I guess I should do the same 2010-09-12T15:40:27 so.. is the world burning yet? 2010-09-12T15:40:56 but removed a bug that would reinforce any planet that was being attacked.... 2010-09-12T15:41:49 so do the contest managers have safeguards in place for when these bots get too smart and try to take over the world? 2010-09-12T15:41:50 *** braddunbar has joined #aichallenge 2010-09-12T15:42:29 danielvf: sandbox 2010-09-12T15:42:35 DirtyKeyboard: sandbox 2010-09-12T15:42:37 jmreardon: I'd say you have a bug http://www.benzedrine.cx/planetwars/canvas?game_id=1284312934|jmreardon.2|ermau-1.0 2010-09-12T15:42:37 DirtyKeyboard: Then you we have to write new bots that attempt to disable our previous ones. 2010-09-12T15:42:42 Did everyone remember their rules of robotics 2010-09-12T15:42:43 DirtyKeyboard: but we cannot guarantee they don't escape 2010-09-12T15:42:50 greghaynes: +1 2010-09-12T15:42:50 No killing humans programmed in? 2010-09-12T15:43:05 Ah crap, I have to add that in. ;) 2010-09-12T15:43:15 ai-contest.com is dead, MySQL says "too many connections" 2010-09-12T15:43:19 I used a roulette wheel selection for that greghaynes 2010-09-12T15:43:21 Turn 999 Player 2 Wins! 2010-09-12T15:43:28 heh 2010-09-12T15:43:41 the turn limit is 1001, correct? 2010-09-12T15:43:43 ermau: ya, I noticed 2010-09-12T15:43:47 DirtyKeyboard: 200 2010-09-12T15:43:55 what's up ? 2010-09-12T15:44:15 I'd like to confirm: ai-contest.com ist dead^^ 2010-09-12T15:44:31 what time is it in US ? 2010-09-12T15:44:31 i'm restarting mysql now 2010-09-12T15:44:39 + 9 ? 2010-09-12T15:44:42 So ? 2010-09-12T15:44:43 i think 200 is also really enough 2010-09-12T15:44:46 contestbot: time 2010-09-12T15:44:46 amstan: 03:44 PM, September 12, 2010 2010-09-12T15:44:53 200 is enough yes. 2010-09-12T15:45:02 should change my starter package, that would be alot nicer 2010-09-12T15:45:03 ty 2010-09-12T15:45:07 *** Appleman1234 has joined #aichallenge 2010-09-12T15:45:08 200 instead of 1000 2010-09-12T15:45:12 site is back 2010-09-12T15:45:14 200 is enough 2010-09-12T15:45:18 0.1 s is enough 2010-09-12T15:45:20 oh, this silly site is validating username and password on each request 2010-09-12T15:45:44 contestbot: rank 2010-09-12T15:45:44 elijahbal: Error: The command "rank" is available in the Alias, ChannelStats, and Factoids plugins. Please specify the plugin whose command you wish to call by using its name as a command before "rank". 2010-09-12T15:45:52 elijahbal: use rankings 2010-09-12T15:45:56 contestbot: rankings 2010-09-12T15:45:58 elijahbal: Top 10 players: dmj111(2835), braddunbar(2815), albert(2792), seigen(2732), Kavish(2707), drewrobb(2695), Obeleh(2679), wtfbbq(2679), sighbot(2676), tinasky(2661) 2010-09-12T15:46:12 *** evgenius has joined #aichallenge 2010-09-12T15:46:57 http://www.benzedrine.cx/planetwars/canvas?game_id=1284319077|deepblue-test|dhartmei <-- thats some crazy cool game 2010-09-12T15:46:58 Well, in the worst case a match would take up in 40 s 2010-09-12T15:47:02 yay, http://www.benzedrine.cx/planetwars/canvas?game_id=1284320318|cwbowron|Cyndre reatreat is working good I see 2010-09-12T15:47:10 you can tell from its moves that its working really clean and concise. 2010-09-12T15:47:20 *** TheMadH4x0r has joined #aichallenge 2010-09-12T15:47:23 cpu time is an arbitrary measure. 2010-09-12T15:47:32 why don't you count in cpu cycles ? 2010-09-12T15:47:48 contestbot: later tell j3camero mysql is erroring that there's too many connections, java bots returned 2010-09-12T15:47:48 amstan: The operation succeeded. 2010-09-12T15:48:05 If you have any scheduler with 3-4 matches it doesnt seem the same than 1s in one match / one computer. 2010-09-12T15:48:10 Cyndre: lol, did you build some kind of flee heuristic? ;p 2010-09-12T15:48:13 Cyndre: lol.. and I won against you 2010-09-12T15:48:38 caytchen: yea - just on the part where it dispatch's reinforcements 2010-09-12T15:49:19 I want to fix some things in site? do I have to create issue and attach patch or is it possible to become commiter? 2010-09-12T15:49:29 Cyndre: I want to see your Bot vs. Mine. 2010-09-12T15:49:31 caytchen: so if absolutely no chance of saving planet (ships wont make in time or unavaliable it abandons said planet) 2010-09-12T15:49:44 Cyndre: Since I never "cut off" fleets, I image it would take a while. 2010-09-12T15:49:44 bss03: my bot sucks 2010-09-12T15:50:08 bss03: nah, cause I will be constantly fleeing to neutral planets and weekening them for you 2010-09-12T15:50:15 that might not be optimal.. your enemy will start on that planet with much more ships than expected 2010-09-12T15:50:23 Naktibalda: It is possible to become a developer. 2010-09-12T15:50:27 finally! i finished the standalone port of the game engine to c++ 2010-09-12T15:50:41 Naktibalda: j3camero is someone to talk to about it, there are probably others. 2010-09-12T15:50:44 so I tried the tcp client. I get a few INFO lines and then nothing... is that expected? (Does it take a while...?) 2010-09-12T15:50:44 caytchen: but if the ships are lost? and its 10% of my fleet and 1% of his fleet? 2010-09-12T15:51:17 Depends on if you can recover the planet quickly 2010-09-12T15:52:06 ermau: it would seem my boss needs to take into account projected growth rate based on your dispatched ships as well (I was ahead so started attacking your planets, but then your ships landed and put you ahead) 2010-09-12T15:52:09 http://www.benzedrine.cx/planetwars/canvas?game_id=1284320400|Cyndre|ermau-1.0 2010-09-12T15:52:34 my bot - lolz at boss 2010-09-12T15:53:49 Cyndre: The fact that you didn't go for the 8 ship large planet early is probably why you lost 2010-09-12T15:54:12 yeah you went for that massive 89.. 2010-09-12T15:54:16 Cyndre: I'd gained 45 ships from mine by the time you took yours 2010-09-12T15:54:25 Cyndre -- how are you getting the game_ids for games on the tcp server? 2010-09-12T15:54:41 http://www.benzedrine.cx/planetwars/getplayer?player=Cyndre 2010-09-12T15:54:47 ah thanks 2010-09-12T15:54:50 but still I managed to catch up in production 2010-09-12T15:55:11 http://www.benzedrine.cx/planetwars/getratings 2010-09-12T15:55:13 Yup 2010-09-12T15:55:22 Like I said.. this bot has huge logic flaws 2010-09-12T15:55:32 but then I gave it up 2010-09-12T15:56:00 so now when Im ahaead Im going to make my bot do this > if ahead in growth rate attack neutral planets the enemy is sending ships to instead of enemy planets directly 2010-09-12T15:56:07 For example, I could of captured a larger planet easier than the smaller one (turn 44, that 41 ship fleet) 2010-09-12T15:56:14 Which is a double fail on prioritization 2010-09-12T15:57:01 Naktibalda: you can create patches for the meantime, j3camero can add you as a comitter as well 2010-09-12T15:58:34 game uses floats for planet positions ?? 2010-09-12T15:59:02 *** gilaniali has quit IRC (Ping timeout: 252 seconds) 2010-09-12T15:59:08 ermau: in my todo list -> better planet selection involving closeness and ammount of ships compareable to my total ships instead of just closeness 2010-09-12T15:59:11 edcba: I think it is IEEE double-precision. 2010-09-12T15:59:19 edcba: But, yes. 2010-09-12T15:59:25 is that specified in the rules ? 2010-09-12T15:59:35 Cyndre: I'm working on a whole new bot from scratch 2010-09-12T15:59:46 edcba: IEEE-double? No. Floating-point? Yes. 2010-09-12T15:59:58 Somehow my bot is timing out now :-\ 2010-09-12T16:00:18 add me in on the "total rewrite" :/ 2010-09-12T16:00:19 *** dolphin278 has quit IRC (Ping timeout: 252 seconds) 2010-09-12T16:00:25 because how you will handle those floats may change 'everything' :) 2010-09-12T16:00:28 *** dolphin278 has joined #aichallenge 2010-09-12T16:00:43 and not reinforcing costs me a huge ammount of ships 2010-09-12T16:00:46 the planet positions aren't even all that relevant; it's the graph formed by the distance function that matters 2010-09-12T16:00:47 *** hornairs has joined #aichallenge 2010-09-12T16:01:14 a1k0n: the distance may gives you some 14.0000000001 instead of 13.999999999 2010-09-12T16:01:17 http://github.com/albertz/planet_wars-cpp 2010-09-12T16:01:41 edcba: That should happen in so few cases as to be ignorable. 2010-09-12T16:01:50 edcba: have you seen any maps where it's close enough to matter? 2010-09-12T16:02:26 a1k0n: i didn't try anything serious yet 2010-09-12T16:02:31 i'm studying the game 2010-09-12T16:02:42 fair enough 2010-09-12T16:02:45 *** hornairs is now known as dorpairs 2010-09-12T16:02:51 * a1k0n lunch 2010-09-12T16:05:13 *** AJC has quit IRC (Quit: Leaving) 2010-09-12T16:06:19 albert: if I understand correctly, your engine is a native replacement for PlayGame? 2010-09-12T16:06:37 *** gefunkt has quit IRC (Remote host closed the connection) 2010-09-12T16:06:43 *** gefunkt_ has joined #aichallenge 2010-09-12T16:06:47 If I want to prevent my unordered_map from rehashing should I just initialize the size to something greater than the number of items I want to store in it? 2010-09-12T16:06:58 *** Appleman1234 has quit IRC (Ping timeout: 255 seconds) 2010-09-12T16:07:04 *** gefunkt_ has quit IRC (Client Quit) 2010-09-12T16:07:21 retybok: yea 2010-09-12T16:07:29 and now i am going to rewrite the showgame 2010-09-12T16:07:39 *** jukkerknott has joined #aichallenge 2010-09-12T16:07:46 in c++? ;) 2010-09-12T16:07:49 yes 2010-09-12T16:07:54 *** bobng has quit IRC (Ping timeout: 240 seconds) 2010-09-12T16:08:09 qt or some other abnomination? 2010-09-12T16:08:12 is the "too many connections" error std knowledge? 2010-09-12T16:08:30 i was thinking about sdl 2010-09-12T16:08:41 rwa: I belive it has been reported. 2010-09-12T16:08:55 neat :) 2010-09-12T16:09:02 sometimes the tcp thing tells me I lost the game but in reality I won 2010-09-12T16:09:17 Is it still legal to continue thinking when you're not asked to? 2010-09-12T16:09:25 *** Appleman1234 has joined #aichallenge 2010-09-12T16:09:33 retybok: i did last time 2010-09-12T16:09:33 (i believe it was for the previous challenge) 2010-09-12T16:09:38 [14:42] contestbot: later tell j3camero mysql is erroring that there's too many connections, java bots returned 2010-09-12T16:09:56 ? 2010-09-12T16:10:13 *** kozlovsky has quit IRC (Ping timeout: 252 seconds) 2010-09-12T16:10:27 amstan: That was to rwa. He wanted to know if "too many connections" had been reported. 2010-09-12T16:10:31 right 2010-09-12T16:10:37 is it still doing it? 2010-09-12T16:10:47 I just got it twice in a row 2010-09-12T16:10:53 Website doesn't work for me at all right now. 2010-09-12T16:10:54 rwa: after i sent that message? 2010-09-12T16:11:23 now it's "loading..." but no error yet 2010-09-12T16:11:35 amstan: is max clients still250? 2010-09-12T16:11:45 dstufft_: for apache, yes 2010-09-12T16:11:52 ok.. 2010-09-12T16:11:58 one core is at 100% , kernel use 2010-09-12T16:12:00 wtf? 2010-09-12T16:12:01 *** raffi has joined #aichallenge 2010-09-12T16:12:06 amstan: mysql defaults to 151 max connections 2010-09-12T16:12:14 *** s_denchev has joined #aichallenge 2010-09-12T16:12:25 Cyndre: what the heck is this? http://www.benzedrine.cx/planetwars/canvas?game_id=1284321284|Cyndre|caytchen 2010-09-12T16:12:27 :o 2010-09-12T16:12:29 *** raffi is now known as Guest56798 2010-09-12T16:12:38 default was 100 if I remember correctly 2010-09-12T16:12:39 or 100 on older versions of mysql 2010-09-12T16:12:55 default changed in 5.1.15 2010-09-12T16:13:19 http://dev.mysql.com/doc/refman/5.1/en/server-system-variables.html#sysvar_max_connections 2010-09-12T16:15:39 anybody know what happened to submission? Is it down entirely? 2010-09-12T16:16:36 *** Guest56798 has left #aichallenge 2010-09-12T16:16:54 http://www.dontevenreply.com/view.php?post=76 2010-09-12T16:16:55 lol 2010-09-12T16:17:17 *** rafsoaken has joined #aichallenge 2010-09-12T16:18:56 is the source to ShowGame.jar in svn? 2010-09-12T16:19:02 fcking website 2010-09-12T16:19:29 is there really a need to do some db lookup for static pages ? :) 2010-09-12T16:19:49 *** GordonQuad_ has joined #aichallenge 2010-09-12T16:20:15 dstufft_: rofl 2010-09-12T16:20:29 It's safe to assume planets won't just appear, right? 2010-09-12T16:20:41 javagamer_Cpp: hope so! 2010-09-12T16:20:48 could anyone here help me out? my bot keeps timing out, and i don't understand why it's doing it :s 2010-09-12T16:20:51 *** kozlovsky has joined #aichallenge 2010-09-12T16:20:57 rwa, me too because that could probably seriously throw my code off 2010-09-12T16:20:59 td: running where? 2010-09-12T16:21:26 from my extensive experiance watching Doctor Who, plants can appear from no where !! 2010-09-12T16:21:42 yeah plants really just need some sun... 2010-09-12T16:22:25 as soon as i start up MyBot to play it against an example bot, before the first move it says WARNING: player 1 timed out 2010-09-12T16:22:55 running on your local machine? 2010-09-12T16:23:01 yes, sorry 2010-09-12T16:23:03 http://downforeveryoneorjustme.com/ai-contest.com 2010-09-12T16:23:35 td: what lang are you using? 2010-09-12T16:23:40 C# 2010-09-12T16:23:54 ok, so my bot easily defeats all the example bots. now what do I do? How do I try and compete against other peoples' bots? Once I submit, can I update my bot, or am I stuck with what I submitted? 2010-09-12T16:24:01 I would sanity check your setup by commenting out your entire DoTurn() function... make sure that runs... 2010-09-12T16:24:16 bah.. alright D: 2010-09-12T16:24:23 I did try that though 2010-09-12T16:24:29 td, I had this problem. It ended up being an unhandled exception 2010-09-12T16:24:31 that works or no? 2010-09-12T16:24:36 ended up getting to after turn 1, then it times out there 2010-09-12T16:24:37 rogue780: You can resumit as often as you like (and can get the fsckin' website to work). 2010-09-12T16:24:45 rogue780: try the tcp server 2010-09-12T16:25:01 td, are you making sure that you aren't sending more ships from a planet than that planet has? 2010-09-12T16:25:14 rogue780: Re-sumbitting resets your calculated rating, but does not otherwise "penalize" you. 2010-09-12T16:25:18 if it times out with a no-op DoTurn(), I can't really help you... something's screwy 2010-09-12T16:25:28 caytchen, tcp server? 2010-09-12T16:25:40 rogue780: However, using the tcp server often gets quicker game results. 2010-09-12T16:25:43 @tcp-server 2010-09-12T16:25:44 bss03: tcp-server = tcpserver = http://www.benzedrine.cx/planetwars/ 2010-09-12T16:25:59 but be wary, you will be blown away ;) 2010-09-12T16:26:05 most of the examples are.. well examples 2010-09-12T16:26:11 rogue780: yes 2010-09-12T16:26:17 Is there a way to watch tcpserver games outside of the server (locally on my machine)? 2010-09-12T16:26:26 uhm how can I get tcp.c to work for Python? 2010-09-12T16:26:33 I can't beat all the examples yet, and I'm above 70th. 2010-09-12T16:26:39 yeah. how do I get tcp.c to work with java? 2010-09-12T16:26:49 So, if you can beat all the examples, I'm betting top 50. 2010-09-12T16:27:00 tcp.c doesn't care about your specific language. just compile it into a binary and use it instead of PlayGame.jar 2010-09-12T16:27:00 Above that, things get really tough. 2010-09-12T16:27:02 bss03, really? 2010-09-12T16:27:32 rwa: do you know how to use C#? 2010-09-12T16:27:33 caytchen, I don't have a C compiler presently 2010-09-12T16:27:47 javagamer_Cpp: I go 99/100 vs. Dual, 47/100 vs. Rage, and 100/100 against the rest, IIRC. 2010-09-12T16:27:55 rogue780: do you run *nix or on windows? 2010-09-12T16:27:59 td: I'm not too familiar with C#, no. I'm a C++ guy. 2010-09-12T16:28:06 bah, alright ;( 2010-09-12T16:28:29 caytchen, windows xp 2010-09-12T16:28:42 rogue780: if you can, get a linux distro for the challenge, it will be easier for you 2010-09-12T16:28:59 you don't need a lot of space, even a virtual machine is good enough 2010-09-12T16:29:05 mhm ok then you won't be able to run tcp.c since it was written for unix systems, but i think theres a java version on the forums 2010-09-12T16:29:09 (do the forums work?) 2010-09-12T16:29:16 yup forums work 2010-09-12T16:29:22 retybok, at home I have 3 linux boxes. I'm away right now and just have my under-powered netbook 2010-09-12T16:29:23 you can compile tcp.c using cygwin. worked for me 2010-09-12T16:29:44 can someone review my patch? http://code.google.com/p/ai-contest/issues/detail?id=137 2010-09-12T16:29:50 *** xkpe has joined #aichallenge 2010-09-12T16:31:02 http://downforeveryoneorjustme.com/http://benzedrine.cx 2010-09-12T16:31:04 wtf... 2010-09-12T16:31:25 is there that many ppl checking those sites ? :) 2010-09-12T16:31:54 www.benzedrine.cx/planetwars/ 2010-09-12T16:31:56 works 2010-09-12T16:32:06 hmm ok 2010-09-12T16:32:09 this room is very active and still growing 2010-09-12T16:32:16 edcba: missing www. 2010-09-12T16:32:28 yes dunno how i ended up eating www 2010-09-12T16:33:12 woo!, going to milk this elo score by not playing any more games 2010-09-12T16:33:17 I've never had any succes with cygwin 2010-09-12T16:33:32 greghaynes, that's my strategy! 2010-09-12T16:34:04 ok got tcp.c working hardcoding /usr/bin/python, bot file name and arguments into bpopen 2010-09-12T16:34:09 uh, forums are down? 2010-09-12T16:34:29 indeed on windows players timeout before even beginning 2010-09-12T16:34:38 rogue780, so is ai-contest.com 2010-09-12T16:34:45 gah! 2010-09-12T16:35:10 tcp server is working great though 2010-09-12T16:35:11 http://www.benzedrine.cx/planetwars/getratings 2010-09-12T16:35:11 would have be wised to have included some hello message in the protocol... 2010-09-12T16:35:17 I wish I could help with bandwidth w/o forfeiting 2010-09-12T16:36:03 greghaynes: you're 3rd 2010-09-12T16:36:04 Utkarsh, I'm downloading cygwin. would you be willing to help me get tcp.c compiled if I have any questions? 2010-09-12T16:36:06 rogue780: bandwith isn't an issue 2010-09-12T16:36:23 amstan: Which server? 2010-09-12T16:36:27 tcp 2010-09-12T16:36:31 someone got in front 2010-09-12T16:36:34 they're milking too 2010-09-12T16:36:34 aww 2010-09-12T16:36:37 rogue780: it should be very easy, just make sure you install gcc 2010-09-12T16:36:48 well I have a sever with over 20TB storage and even more bandwidth that allows more than 1040 mysql connections 2010-09-12T16:36:50 (i mean select it when installing cygwin) 2010-09-12T16:36:55 isn't tcp.c swappable by some netcat command ? 2010-09-12T16:36:59 and even more postres connections 2010-09-12T16:39:23 Oh wow, for the first time ever I've beat dual bot. Though I've really spent most of my time working on organizing data rather than actual logic. 2010-09-12T16:39:56 rogue780: as retybok said, install gcc from devel tools. 2010-09-12T16:40:04 *** Appleman1234 has quit IRC (Ping timeout: 276 seconds) 2010-09-12T16:40:42 I rarely ever use gcc. I never code in it. Always Java, Python or PHP. Only in the last month have I been learning C++. 2010-09-12T16:41:02 better late than never 2010-09-12T16:41:57 *** pacotabaco has joined #aichallenge 2010-09-12T16:42:08 Who hosts the tcp planetwars? 2010-09-12T16:42:18 *** Appleman1234 has joined #aichallenge 2010-09-12T16:42:32 dhmatei, hope i spelled that right 2010-09-12T16:42:39 should I install GCC 3.x or GCC 4? 2010-09-12T16:42:46 ah...hes not here 2010-09-12T16:43:14 i've got 4.3.4 2010-09-12T16:45:31 *** aiaiaiaiaia has joined #aichallenge 2010-09-12T16:45:35 off topic, but anyeone else here addicted to reddit? 2010-09-12T16:45:49 hah 2010-09-12T16:46:12 <- 2010-09-12T16:46:34 <- 2010-09-12T16:46:47 glad to know I'm not alone 2010-09-12T16:46:49 but not recently. recently it's been planet wars 2010-09-12T16:47:13 jmcarthur: when the site is up... 2010-09-12T16:47:17 Is it normally super-slow to login? I haven't tried up until now b/c my code wasn't in a state worth submitting. 2010-09-12T16:47:30 j3camero: heh 2010-09-12T16:47:33 oh is it down again? 2010-09-12T16:47:36 The following words in your search query were ignored because they are too common words: timeout. 2010-09-12T16:47:39 ah no 2010-09-12T16:47:47 edcba: i know it's weird 2010-09-12T16:47:50 I think its calculating new rankings 2010-09-12T16:47:58 it's working fine for me 2010-09-12T16:48:02 i was looking for timeout issues on windows :) 2010-09-12T16:48:02 j3camero: can we restart the server? it's acting weird 2010-09-12T16:48:05 rogue780: yes 2010-09-12T16:48:16 j3camero: btw, i put up a patch for review to get the game spec on the site 2010-09-12T16:48:30 i tested it in html form 2010-09-12T16:48:34 just not with php 2010-09-12T16:48:41 but i'm pretty sure i did that fine. it's just header and footer 2010-09-12T16:49:06 Warning: mysql_connect() [function.mysql-connect]: Too many connections in /var/www/vhosts/ai-contest.com/www/mysql_login.php on line 17 2010-09-12T16:49:06 Array ( [type] => 2 [message] => mysql_connect() [function.mysql-connect]: Too many connections [file] => /var/www/vhosts/ai-contest.com/www/mysql_login.php [line] => 17 ) cannot connect: 1 2010-09-12T16:49:13 *** Obeleh has quit IRC (Quit: Page closed) 2010-09-12T16:49:22 that's maybe a little too much information 2010-09-12T16:49:31 Ah. now it's just telling me my account isn't registered, much better 2010-09-12T16:49:52 ok. who would write a bot in PHP? Seriously. I like php quite a bit (like that scruffy stray dog that has adopted you), but I've always found it ill-suited for anything but web projects 2010-09-12T16:50:00 rogue780: dunno 2010-09-12T16:50:07 rogue780: i've even seen people write cocoa apps in it 2010-09-12T16:50:18 * rogue780 facepalm 2010-09-12T16:50:33 what ever happened to the right tool for the right job? 2010-09-12T16:50:35 ok v1.2 fixes timeouts problems 2010-09-12T16:50:47 when was 1.2 released? 2010-09-12T16:50:49 nothing should be written in php unless it's a web app where you need to the ability to isntall it in almost any web environment imaginable 2010-09-12T16:50:56 rogue780: http://ai-contest.com/forum/viewtopic.php?f=18&t=534 2010-09-12T16:51:50 edcba, awesome, thanks 2010-09-12T16:51:50 *** GordonQuad_ has quit IRC (Ping timeout: 252 seconds) 2010-09-12T16:52:42 I love the way PHP manages arrays 2010-09-12T16:53:16 *** semyglass has joined #aichallenge 2010-09-12T16:53:29 odds that the PHP bot wins it all? 2010-09-12T16:54:04 is there a PHP starter pack? 2010-09-12T16:54:29 is there was, I would rather use it instead of C++ 2010-09-12T16:54:42 there was one submitted to the issue tracker iirc 2010-09-12T16:54:50 rogue780: I have a PHP bot 2010-09-12T16:55:00 i just don't understand 2010-09-12T16:55:18 I spent a lot of time fighting with SÒL 2010-09-12T16:55:28 Someone please make a php bot, that would be hilarious 2010-09-12T16:55:35 Naktibalda, evgenius, why? Why not python or Java? 2010-09-12T16:55:39 if you know php why not just use python or ruby? 2010-09-12T16:55:44 greghaynes: too late 2010-09-12T16:55:51 It exists? 2010-09-12T16:55:53 yes 2010-09-12T16:55:54 Link? 2010-09-12T16:55:56 I don't know python or Java 2010-09-12T16:56:07 sounds like a good time to learn python 2010-09-12T16:56:10 evgenius, this would be a great opportunity to learn 2010-09-12T16:56:24 You know python, you just dont realize it yet 2010-09-12T16:56:26 python is pretty great/easy tbh 2010-09-12T16:56:29 :0 2010-09-12T16:56:31 :) 2010-09-12T16:56:33 it's almost psuedo code at times 2010-09-12T16:56:47 Naktibalda, have you submitted your bot? 2010-09-12T16:56:48 because I know PHP, I'm not interested in other top languages. if I write a bot in another language, it will be Lua 2010-09-12T16:56:50 eh, python is not psuedocode to me 2010-09-12T16:57:01 Then use lua 2010-09-12T16:57:12 ruby > python :P 2010-09-12T16:57:18 ew ruby 2010-09-12T16:57:20 *** s_denchev has quit IRC (Quit: Page closed) 2010-09-12T16:57:21 ruby vs. python is a moot argument 2010-09-12T16:57:21 ew 2010-09-12T16:57:26 * rogue780 Slaps Utkarsh with an ISO compliant trout. 2010-09-12T16:57:30 Yes, because python rules all 2010-09-12T16:57:32 ;) 2010-09-12T16:57:36 they are basically the same, just with different syntactic philosophies 2010-09-12T16:57:42 well, when you learn a language, you spend a lot of time on simple problems. I'd rather get some results as quick as possible 2010-09-12T16:57:50 one python to rule them all 2010-09-12T16:57:54 and the ruby community seems to like magic a little more 2010-09-12T16:57:55 Actually, they are pretty opposite design IMO 2010-09-12T16:58:07 has ruby been added to the server yet? 2010-09-12T16:58:10 evgenius: http://code.google.com/p/ai-contest/issues/detail?id=104 2010-09-12T16:58:13 opposite along a certain dimension 2010-09-12T16:58:14 couldn't find a single ruby entry 2010-09-12T16:58:20 but along all others they are pretty much the same 2010-09-12T16:58:26 ruby likes everything and the kitchen sink, python likes you having to import as much as possible 2010-09-12T16:58:28 evgenius, agreed. Although it seems to me that with c++, fast results is almost a never 2010-09-12T16:59:13 ruby is convention over configuration, python is explicit over implicit 2010-09-12T16:59:34 Well said 2010-09-12T16:59:39 that's the main difference in my book. all others are insignificant 2010-09-12T16:59:40 For this challenge i think "slow" languages have a chance, computation time doesn't seem so important (since the problem is so overwhelming anyways) 2010-09-12T16:59:53 Now to finish the qtcpcient... 2010-09-12T16:59:54 retybok: dunno about that. there are some computation heavy methods that still apply 2010-09-12T17:00:01 i've never tried ruby. I've wanted to, but it seems too gimicky to me. Kinda like java did back in the day (in my strange mind anyway) 2010-09-12T17:00:06 of course running fast is always an advantage, but less so than for tron I believe 2010-09-12T17:00:38 Naktibalda, I'm really curious about your bot. I've never made a CLI app with python before. Have you submitted your bot yet? 2010-09-12T17:00:41 hi guys. What's the ip and port to test our bots with tcp.c? Can we use any username without passwd? 2010-09-12T17:00:42 bss03: Turns out the test script wasn't working correctly. I've tried again now and the results are pretty much the same, except for RageBot where I only got 18/100, and against stuff on TCP it also sucks (as I had expected) 2010-09-12T17:00:48 well with tron there were things like pathfinding, with this challenge everything that counts is basically lookahead and heuristics 2010-09-12T17:01:02 pacotabaco: 213.3.30.106, 9999, yes 2010-09-12T17:01:07 tnks 2010-09-12T17:01:17 rogue780: is it possible to submit a bot written in unsupported language? 2010-09-12T17:01:42 my bot is pretty dumb yet, I need some good strategy 2010-09-12T17:01:51 Naktibalda, I...I suppose not. 2010-09-12T17:01:57 *** semyglass has quit IRC (Quit: Page closed) 2010-09-12T17:02:07 *** Queue29 has quit IRC (Ping timeout: 276 seconds) 2010-09-12T17:02:10 Naktibalda: you can always submit an issue detailing how to support that particular language and i think they'll try to realise it 2010-09-12T17:02:21 Naktibalda, have you pitted it against the example bots? 2010-09-12T17:02:27 ok c# starter pack is bugged 2010-09-12T17:02:52 I do this: ./tcp_command 213.3.30.106 9999 pacotabaco "/usr/bin/python p/MyBot.py" and get "execv: /usr/bin/python p/MyBot.py: failed" :( 2010-09-12T17:02:53 very nice to put // Owned in catch part... 2010-09-12T17:02:53 caytchen: I did that 2010-09-12T17:02:54 greghaynes: it isn't submitted or anything (i think still not supported on the server) http://code.google.com/p/ai-contest/issues/detail?id=104 2010-09-12T17:03:08 so System.FormatException: Le format de la chaîne d'entrée est incorrect. 2010-09-12T17:03:08 pacotabaco: chmod 755 MyBot.py 2010-09-12T17:03:14 hahaha 2010-09-12T17:03:15 Naktibalda: so, if I upload a PHP file, the server will understand how to run it? is PHP starter package enabled on server already? 2010-09-12T17:03:25 no 2010-09-12T17:03:43 guys.. unless the starter packs are on the main website, your languages will not work 2010-09-12T17:03:44 they haven't added support for any new language yet 2010-09-12T17:03:48 no Environment.NewLine, no CultureInfo.InvariantCulture... 2010-09-12T17:03:57 doesn't look like c# program :) 2010-09-12T17:03:57 we have much bigger issues than to add random languages 2010-09-12T17:04:08 except for haskell, but that's just because it happens to have been added for something i was testing earlier 2010-09-12T17:04:16 What could be more important than allowing people to write AI's in php!? 2010-09-12T17:04:18 edcba: maybe this is due to the server using mono? 2010-09-12T17:05:12 anybody that wants to try their obscure technologies (i am among them) can do so with the tcp server 2010-09-12T17:05:13 greghaynes: after all php is an all-time ai devs favourite language... unlike kiddie stuff like lisp or haskell 2010-09-12T17:05:31 ;-) 2010-09-12T17:05:44 Yes, everyone knows lisp was made for web hacks, whilst php was designed by phd's for AI 2010-09-12T17:05:44 of course, *eventually* other languages can be added 2010-09-12T17:05:47 tcp server sadly doesn't work for me. I keep getting timeouts 2010-09-12T17:05:59 evgenius: bad internet connection? 2010-09-12T17:06:05 maybe 2010-09-12T17:06:06 *** Kastra has quit IRC (Ping timeout: 240 seconds) 2010-09-12T17:06:08 canihazbacon: no it's language thingie 2010-09-12T17:06:08 :\ 2010-09-12T17:06:13 evgenius: check error log 2010-09-12T17:06:20 i use a 'french' computer 2010-09-12T17:06:25 question: any other python dev here that would be interested in writing a python port of the TCP client? 2010-09-12T17:06:27 fix is to put Thread.CurrentThread.CurrentCulture = CultureInfo.InvariantCulture; 2010-09-12T17:06:32 evgenius: have you tried to run it against local bots? 2010-09-12T17:06:37 edcba: i highly recommend writing your own version of the game framework 2010-09-12T17:06:38 canihazbacon: there is a java one on the forums somewhere 2010-09-12T17:06:53 java :( 2010-09-12T17:06:59 caytchen: i will have to some time if i want to do some real bot :) 2010-09-12T17:07:03 what's wrong with the C one? 2010-09-12T17:07:05 sure, with local bots it works like a charm 2010-09-12T17:07:53 Im working on a gui one... 2010-09-12T17:07:55 edcba: try http://rapidshare.com/files/418688029/aiChallenge.rar 2010-09-12T17:08:05 *** Vio has quit IRC (Read error: Connection reset by peer) 2010-09-12T17:08:06 oh god, rapidshare 2010-09-12T17:08:12 weren't there a bunch of people complaining that they didn't get it to work on windows? 2010-09-12T17:08:16 edcba: its full of linq so probably won't compile on official but in the meantime... 2010-09-12T17:08:20 canihazbacon: just use cygwin 2010-09-12T17:08:22 greghaynes: gui what? 2010-09-12T17:08:27 canihazbacon: the complainers were too lazy to try, mainly 2010-09-12T17:08:30 ok. were the fudge is cygwin.dll 2010-09-12T17:08:33 zerd: http://github.com/greghaynes/QTcpPlanetWars 2010-09-12T17:08:34 just might be a bit of an understatement ;-) 2010-09-12T17:08:46 understatement? 2010-09-12T17:08:58 just as in just use cygwin 2010-09-12T17:09:16 cygwin i easy. you just install it and build with the gcc that comes with it 2010-09-12T17:09:19 *is 2010-09-12T17:09:24 and run from there 2010-09-12T17:09:33 edcba: it still sucks though, i'm looking to integrate UIDs for planets and fleets especially 2010-09-12T17:09:46 canihazbacon: you could also use a linux vm if you'd rather 2010-09-12T17:10:00 actually i am on archlinux right now ;-) 2010-09-12T17:10:03 canihazbacon: or you could, like, write your own client :) 2010-09-12T17:10:10 ah, then what's the deal? :P 2010-09-12T17:10:13 that was what i was suggesting 2010-09-12T17:10:29 sounded like you were suggesting that somebody else do it 2010-09-12T17:10:49 ok what's what port and ip address again? 2010-09-12T17:10:52 that others might have problems with the code... 2010-09-12T17:10:53 and that being paranoid i am not particularly fond of c code... 2010-09-12T17:11:02 213.3.30.106 9999 2010-09-12T17:11:25 jmacarthur: i was wondering whether anybody would be interested in doing it WITH me 2010-09-12T17:11:30 aha 2010-09-12T17:11:45 then i misunderstood 2010-09-12T17:11:52 np 2010-09-12T17:11:55 wtf 2010-09-12T17:12:07 off-topic: where are you guys from? 2010-09-12T17:12:13 canihazbacon, you know how you can tell an arch linux user? 2010-09-12T17:12:14 The internet 2010-09-12T17:12:15 man, i'm looking at this php starter now. this is bringing back memories of when i actually used php... 2010-09-12T17:12:18 not great memories 2010-09-12T17:12:23 :D 2010-09-12T17:12:33 nope? 2010-09-12T17:12:35 rogue780: they tell everybody 2010-09-12T17:12:36 how? 2010-09-12T17:12:40 :D 2010-09-12T17:12:49 jmcarthur, you get a prize! 2010-09-12T17:12:58 YAY I'M AN ARCH USER 2010-09-12T17:13:13 *ahem* i mean YAY 2010-09-12T17:13:47 you kind of asked ;-) 2010-09-12T17:13:48 but i admit that some bragging on my part might have been involved as well ;-) 2010-09-12T17:13:49 *** Queue29 has joined #aichallenge 2010-09-12T17:13:55 jmcarthur: is there a good starter package for Haskell somewhere? 2010-09-12T17:14:02 Mathnerd314: jaspervdj has one on github 2010-09-12T17:14:12 haven't used it, dunno how good it is 2010-09-12T17:14:23 jmcarthur: so you wrote your own? 2010-09-12T17:14:29 Mathnerd314: basically 2010-09-12T17:14:40 Mathnerd314: it's not really in shape for public consumption 2010-09-12T17:14:57 and it works very differently from how most people might expect 2010-09-12T17:15:09 so i don't think it's worth trying to share 2010-09-12T17:15:17 i could share a small part of it, perhaps... 2010-09-12T17:15:21 the IO, at least 2010-09-12T17:15:29 jmcarthur: that part sounds good 2010-09-12T17:15:29 *** TheMadH4x0r has quit IRC (Ping timeout: 252 seconds) 2010-09-12T17:15:42 jmcarthur: I hate the IO monad :p 2010-09-12T17:15:48 well, this is still IO monad 2010-09-12T17:15:56 caytchen, ok so I ran ./tcp.exe 213.3.30.106 9999 rogue780 "java ErrBot" 2010-09-12T17:15:57 but i could factor it out and throw it up somewhere 2010-09-12T17:16:07 sometime later tonight 2010-09-12T17:16:13 *** pacotabaco has quit IRC (Ping timeout: 252 seconds) 2010-09-12T17:16:27 Mathnerd314: do you not like jaspervdj's? 2010-09-12T17:16:32 caytchen, it said "execv: java ErrBot: failed" 2010-09-12T17:16:44 java -jar ErrBot.jar 2010-09-12T17:16:53 caytchen, It's not in a jar 2010-09-12T17:17:09 does it work with just java ErrBot? 2010-09-12T17:17:35 caytchen, yes 2010-09-12T17:17:57 I'm back 2010-09-12T17:18:11 yay! 2010-09-12T17:18:22 does the forum login work? because it does not for me 2010-09-12T17:18:41 i'm afraid to log out to try 2010-09-12T17:19:21 ah it works now 2010-09-12T17:19:45 jmcarthur: not really. it seems unwieldy. 2010-09-12T17:20:06 wb danielvf 2010-09-12T17:20:08 Yeah, server is IO wait'ed out 2010-09-12T17:20:27 caytchen, I can use my normal testing script with ShowGame.jar and PlayGame.jar but the tcp.exe doesn't like me 2010-09-12T17:20:28 Trying to run 11 tournaments at the same time 2010-09-12T17:20:29 uhm, then i suppose it should work.. I don't run tcp using cygwin so I can't really help you there :/ 2010-09-12T17:20:40 *** sigh has joined #aichallenge 2010-09-12T17:21:32 nice 2010-09-12T17:21:39 got the strongest bot on benzedrine 2010-09-12T17:21:41 Mathnerd314: mine basically just spits out a [Either Planet Fleet] for you to use, and gives you an Order structure with an issueOrders :: [Order] -> IO () function 2010-09-12T17:21:49 caytchen, I had to stick "java Err" in a .bat file 2010-09-12T17:21:54 mostly because other bots are down it seems ;p 2010-09-12T17:22:18 jmcarthur: so maybe I'm better off writing my own 2010-09-12T17:22:22 rogue780: yeah seems execv doesn't like arbitrary command lines 2010-09-12T17:22:23 Mathnerd314: where Planet and Fleet are exactly the information in the file format 2010-09-12T17:22:38 but needs program and command line clearly seperated but tcp.c won't do that 2010-09-12T17:22:42 Mathnerd314: dunno about "better off". you'd have to write basically the same functionality and then whatever else you want on top of it 2010-09-12T17:22:50 so, does tcp give the same output as playgame so I can pipe it to showgame? 2010-09-12T17:23:14 you can watch game in site 2010-09-12T17:23:16 bartwe: i'll let loose my new version in a minute or so 2010-09-12T17:23:20 Mathnerd314: my I/O and parsing is probably faster than jaspervdj's, but really, his has more conveniences 2010-09-12T17:23:23 Naktibalda, where? 2010-09-12T17:23:24 afaik it only ever prints out information like "Your opponent is.." 2010-09-12T17:23:29 deepblue: pheer 2010-09-12T17:23:37 dmj111, you just beat me 2010-09-12T17:23:40 :) 2010-09-12T17:23:56 rogue780: http://www.benzedrine.cx/planetwars/getplayer?player=bartwe.3.2 2010-09-12T17:23:57 i don't even have a 1st version yet! 2010-09-12T17:24:16 rogue780: but you can check http://www.benzedrine.cx/planetwars/getplayer?player=rogue780 for your games 2010-09-12T17:24:22 next up: chain killer algo 2010-09-12T17:24:39 your bot is fun to watch bartwe 2010-09-12T17:24:42 caytchen, this tracks the tcp games as well? 2010-09-12T17:24:50 'caues it says no games 2010-09-12T17:24:51 nice forward chains :) 2010-09-12T17:24:56 it only tracks tcp games 2010-09-12T17:25:03 and it can take a minute for a game to show up 2010-09-12T17:25:09 and it takes a minute or so until your latest shows up 2010-09-12T17:25:10 yep, but not chains towards enemies, unles the enemy provokes it 2010-09-12T17:25:11 ^^ 2010-09-12T17:25:39 jmcarthur: or I could just stick with Java. My code is taking a decidedly imperative sort of structure. 2010-09-12T17:26:45 first dinner and debugging than new version on the tcp :) 2010-09-12T17:26:50 later guys 2010-09-12T17:27:03 *** deepblue has quit IRC (Quit: Page closed) 2010-09-12T17:27:05 btw, isn't it possible that google could provide some servers for the challenge? 2010-09-12T17:27:29 this question should be in FAQ 2010-09-12T17:27:30 *** coventry has joined #aichallenge 2010-09-12T17:27:40 yeah, we all wonder what googles role is here.. 2010-09-12T17:27:51 *** rafsoaken has quit IRC (Quit: Page closed) 2010-09-12T17:28:06 google is a part of title :) 2010-09-12T17:28:23 well, it is said that google just sponsors this, so i guess they dont do much more yet. but maybe they could, at least for some servers 2010-09-12T17:29:00 *** sinan has joined #aichallenge 2010-09-12T17:29:04 Mathnerd314: tsk tsk 2010-09-12T17:29:33 LOL 2010-09-12T17:29:34 Is this room logged somewhere? 2010-09-12T17:29:46 i just read the reason that the server has been funky today 2010-09-12T17:30:10 coventry: http://contestbot.hypertriangle.com/channellogs/%23aichallenge/ 2010-09-12T17:30:58 jmcarthur: thank you 2010-09-12T17:31:04 wow this channel has 100 users 2010-09-12T17:31:53 yay 2010-09-12T17:32:33 hey guys 2010-09-12T17:33:05 jmcarthur: to do it functionally would, I think, require FRP - which I don't know how to do 2010-09-12T17:33:13 *** hornairs has joined #aichallenge 2010-09-12T17:33:19 Mathnerd314: it wouldn't require it 2010-09-12T17:33:47 my bot is certainly not using FRP 2010-09-12T17:33:49 at least not directly 2010-09-12T17:33:59 i guess if you squint it has some resemblances 2010-09-12T17:34:06 How often do rankings update? 2010-09-12T17:34:28 FlagCapper: i think 15 minutes, right now, but i'm only inferring from other information i've seen 2010-09-12T17:34:51 I've just read the decentralized concept page on the wiki and wondered why the calculation assumed 100 turns? is that the average for games played until now? 2010-09-12T17:34:51 Well then there's probably something wrong with mine, because it hasn't changed for 19 hours 2010-09-12T17:35:04 wow bartwe on the top 2010-09-12T17:35:09 wait, no 2010-09-12T17:35:19 FlagCapper: things have been really really wonky over the last 24 hours 2010-09-12T17:35:27 Sorry, 8 hours. 2010-09-12T17:35:29 ulope: no idea 2010-09-12T17:36:11 *** Appleman1234 has quit IRC (Ping timeout: 240 seconds) 2010-09-12T17:36:30 jmcarthur: Yeah, I realize things are going slow, I just wanted to check to see if there's something wrong with my account or it's a general problem. 2010-09-12T17:36:47 *** dorpairs has quit IRC (Ping timeout: 245 seconds) 2010-09-12T17:36:51 *** SavageCamel2 has joined #aichallenge 2010-09-12T17:37:23 i'm sure it's quite general 2010-09-12T17:37:45 *** rafsoaken has joined #aichallenge 2010-09-12T17:37:47 are the competition servers down right now? I can't load ranking and I never received an email from my submission 2010-09-12T17:38:49 e.g. ai-contest.com is taking too long to respond 2010-09-12T17:39:13 SavageCamel2: yeah there's a lot of work being done to it right now to try and hold things together 2010-09-12T17:40:55 on the plus side 2010-09-12T17:41:01 atleast it's not hosted at steadfast 2010-09-12T17:42:19 Is this the reason: http://highscalability.com/blog/2010/9/5/hilarious-video-relational-database-vs-nosql-fanbois.html 2010-09-12T17:42:39 bartwe's bot amazes me 2010-09-12T17:42:41 *** RainCT has quit IRC (Ping timeout: 240 seconds) 2010-09-12T17:42:51 SavageCamel2: Site should be running again 2010-09-12T17:43:01 http://howfuckedismydatabase.com/mysql/ 2010-09-12T17:43:15 I think we aren't sending out submission emails. :( 2010-09-12T17:43:26 ha ha ha ha 2010-09-12T17:43:30 I'm trying to figure out why it says my game against bhasker was a loss. here's my page http://www.benzedrine.cx/planetwars/getplayer?player=rogue780 and here's the game http://www.benzedrine.cx/planetwars/canvas?game_id=1284326418|bhasker|rogue780 2010-09-12T17:43:50 rogue780: its just the tcp server fucking up, don't worry too much 2010-09-12T17:44:02 saw that game by the way, cool bot 2010-09-12T17:44:18 I think the server needs a process monitor 2010-09-12T17:44:19 just some harakiri decisions at times heh 2010-09-12T17:44:36 to preform harakiri when shit hangs 2010-09-12T17:44:57 dstufft_: lol, yeah 2010-09-12T17:45:00 caytchen, cool bot at me, or someone else? 2010-09-12T17:45:05 you :o 2010-09-12T17:45:09 sweet 2010-09-12T17:45:17 I can't beat dmj.4 though 2010-09-12T17:45:27 apparently TM get's restarted quite frequently, though im not sure why 2010-09-12T17:45:50 they are just batch runners. i think they terminate on their own 2010-09-12T17:45:55 in normal operation 2010-09-12T17:46:18 but there are strange things happening right now... 2010-09-12T17:46:19 jmcarthur: yea i know, but that assumes that they terminate on their own correctly 2010-09-12T17:46:24 yeah 2010-09-12T17:46:34 instead of having a master process that stays running 2010-09-12T17:46:37 the sandbox is completely screwed up anyway. maybe we should rewrite the TM 2010-09-12T17:46:39 that can handle everything 2010-09-12T17:46:57 especially if we are thinking of doing this distributed thing 2010-09-12T17:46:57 constantly firing up new tm's via cron seems very hackish to me 2010-09-12T17:47:01 agreed 2010-09-12T17:47:18 i'd rather than a single TM process and another process to watch it and restart it if necessary 2010-09-12T17:47:18 i think it's done to reload data from mysql 2010-09-12T17:47:22 *rather have 2010-09-12T17:47:27 jmcarthur: I agree 2010-09-12T17:47:29 see, that's silly to me 2010-09-12T17:47:38 but you can handle that by perodically rereshing 2010-09-12T17:47:44 in the code itself 2010-09-12T17:47:46 it should just grab minimal data from mysql for each match 2010-09-12T17:47:51 rather than batch it up 2010-09-12T17:48:34 and if the next complaint is that you need all that data to process properly then we should be using a fancier database than mysql anyway 2010-09-12T17:50:11 *** RainCT has joined #aichallenge 2010-09-12T17:50:11 *** RainCT has joined #aichallenge 2010-09-12T17:50:44 jmcarthur: I don't think mysql is the problem 2010-09-12T17:51:05 i didn't mean to imply that directly 2010-09-12T17:51:20 *** aiaiaiaiaia has quit IRC (Quit: Page closed) 2010-09-12T17:51:26 tutorials say: "Have fun laughing at how much RandomBot and RageBot suck". But RageBot beats me on 36/100 maps, while DualBot only on 8/100 2010-09-12T17:51:28 i mean we are using it improperly 2010-09-12T17:51:33 *** Appleman1234 has joined #aichallenge 2010-09-12T17:51:34 either way, i doubt that's the issue with the TM 2010-09-12T17:51:48 danielvf: Yes, bactch getting all games is wrong 2010-09-12T17:51:49 danielvf: did you see the kernel usage? 2010-09-12T17:51:59 talking to youself again :P 2010-09-12T17:52:04 lawal 2010-09-12T17:52:08 neutral planets are not growing, isn't it ? 2010-09-12T17:52:11 amstan: The server was paging badly 2010-09-12T17:52:17 elijahbal: they do not grow, correct 2010-09-12T17:52:25 danielvf: huh? it was using 300mb or ram 2010-09-12T17:52:26 oh* 2010-09-12T17:52:28 of* 2010-09-12T17:52:29 elijahbal: http://code.google.com/p/ai-contest/wiki/GameSpecification 2010-09-12T17:52:49 elijahbal: there's a patch in the queue to put that on the main site, but there are other fish to fry first, so that will have to do for now 2010-09-12T17:52:54 amstan: Sorry, was talking about a half hour ago 2010-09-12T17:53:24 amstan: Looking at the graphs, I never see the system go over 10% 2010-09-12T17:53:34 danielvf: not the graphs.. 2010-09-12T17:53:37 danielvf: look in htop 2010-09-12T17:53:52 the 1 and 2 should be all green 2010-09-12T17:53:59 *** Zaphus has joined #aichallenge 2010-09-12T17:54:01 but they're all red, the whole cpu is used by the kernel now 2010-09-12T17:54:20 *** RobinB has joined #aichallenge 2010-09-12T17:54:35 interesting 2010-09-12T17:54:39 hi 2010-09-12T17:55:02 I don't know what that is 2010-09-12T17:55:09 danielvf: i suggest a restart 2010-09-12T17:55:15 there's something fishy going on here 2010-09-12T17:55:22 guys from #ubuntu-server suggest it's iowait 2010-09-12T17:55:25 amstan: why a restart? 2010-09-12T17:55:27 *** Strepto has joined #aichallenge 2010-09-12T17:55:44 idk.. there's something weird going on with the kernel 2010-09-12T17:55:50 vmstat 3 2010-09-12T17:56:06 eek, iowait? 2010-09-12T17:56:12 we swapping again? 2010-09-12T17:56:46 No, not swapping 2010-09-12T17:57:01 And it's not the db 2010-09-12T17:57:28 time for lsof 2010-09-12T17:57:34 iotop 2010-09-12T17:57:34 i have no experience with IRC, is it possible to view archive of aichallenge for the last two days? 2010-09-12T17:57:46 *** sigh has quit IRC (Remote host closed the connection) 2010-09-12T17:57:57 kozlovsky: of the channel? 2010-09-12T17:58:02 yes 2010-09-12T17:58:17 kozlovsky: http://contestbot.hypertriangle.com/channellogs/%23aichallenge/%23aichallenge.log <-- 1.3M warning 2010-09-12T17:58:27 thanks 2010-09-12T17:59:47 It took me a while to understand bartwe's strategy of chaining through a string of planets, but it's pretty clever. It gives him more flexibility in re-routing. 2010-09-12T18:00:10 *** mega1 has quit IRC (Ping timeout: 276 seconds) 2010-09-12T18:00:57 Exactly 2010-09-12T18:01:40 is the chaining defined explicitly or implicitly? 2010-09-12T18:01:51 contestbot: learn spec as http://code.google.com/p/ai-contest/wiki/GameSpecification 2010-09-12T18:01:51 jmcarthur: The operation succeeded. 2010-09-12T18:02:00 contestbot: learn specification as http://code.google.com/p/ai-contest/wiki/GameSpecification 2010-09-12T18:02:00 jmcarthur: The operation succeeded. 2010-09-12T18:02:03 Strepto: vm stat output: http://pastebin.com/eLN66sUx 2010-09-12T18:02:06 SavageCamel2: ^^ 2010-09-12T18:02:14 stupid autocompletion.. 2010-09-12T18:02:35 SavageCamel2: iotop, no such command 2010-09-12T18:02:40 c# starter pack should use *properties* 2010-09-12T18:02:45 damn java coders :) 2010-09-12T18:02:51 edcba: fix it? 2010-09-12T18:02:54 amstan: can you get the bot to split up the contest logs by date or something? 2010-09-12T18:03:13 umm 2010-09-12T18:03:19 Strepto: We won't know unless bartwe shows us his code. :-) 2010-09-12T18:03:37 amstan: doesn't worth it, if i want some good bot i'll have to scrape the whole thing i think 2010-09-12T18:03:44 amstan: I already have an updated C# starter pack ready to go.. just as soon as the Mono version gets updated 2010-09-12T18:03:45 I wasn't really interested in the code 2010-09-12T18:03:49 Hi all, I assume the rankings are disabled at the moment - maybe a posting on the front page is in order ? Perhaps there needs to be a blog page for latest news from the admins ? 2010-09-12T18:04:13 i agree 2010-09-12T18:04:14 *** coventry has quit IRC (Remote host closed the connection) 2010-09-12T18:04:15 I'm working on something similar to chaining, but it's not defined explicitly, more like a result of the algorithm 2010-09-12T18:04:43 contestbot: help channellogger 2010-09-12T18:04:43 amstan: Error: There is no command "channellogger". 2010-09-12T18:04:48 contestbot: list 2010-09-12T18:04:48 amstan: AIChallenge, Admin, Alias, BadWords, Channel, ChannelLogger, ChannelStats, Config, Ctcp, Dict, Dunno, Factoids, Filter, Format, Games, Google, Herald, Internet, Later, Limiter, Math, Misc, Network, News, Note, Owner, Plugin, Quote, QuoteGrabs, RSS, Reply, Scheduler, Seen, Services, ShrinkUrl, Status, String, Success, Time, Todo, Topic, URL, Unix, User, Utilities, and Web 2010-09-12T18:04:55 @contestbot time 2010-09-12T18:04:56 Zaphus: Error: "contestbot" is not a valid command. 2010-09-12T18:04:57 i.e. the bot won't know it's chaining, it just moves ships according to where it thinks it'll need them 2010-09-12T18:05:01 contestbot: help ChannelLogger 2010-09-12T18:05:01 amstan: Error: There is no command "channellogger". 2010-09-12T18:05:28 how do I find out the time on the server ? can't remember the timediff between us :-) 2010-09-12T18:05:47 *** arik181 has quit IRC (Ping timeout: 255 seconds) 2010-09-12T18:05:49 Zaphus, +1 2010-09-12T18:05:57 GMT+1? 2010-09-12T18:05:59 had the same problem 2010-09-12T18:06:14 Zaphus: added a note to the rankings page 2010-09-12T18:06:22 *** arik181 has joined #aichallenge 2010-09-12T18:06:57 Mathnerd314: i think it rotates them now 2010-09-12T18:07:06 Zaphus: @time 2010-09-12T18:07:13 @time 2010-09-12T18:07:13 Zaphus: 06:07 PM, September 12, 2010 2010-09-12T18:07:14 Zaphus: either contestbot: or @ 2010-09-12T18:07:17 Zaphus: not both.. 2010-09-12T18:07:28 amstan: yes 2010-09-12T18:07:31 wow the ending is so wrong http://www.benzedrine.cx/planetwars/canvas?game_id=1284327683|bartwe.4|bartwe.4.1 2010-09-12T18:07:43 that doesnt look like GMT+1, but at least I know 2010-09-12T18:07:54 *** gzurl has joined #aichallenge 2010-09-12T18:08:03 absolutely not GMÒ+1 2010-09-12T18:08:16 GMT+5 2010-09-12T18:08:20 err -5 2010-09-12T18:08:26 Mathnerd314: now.. it would be cool if we could split the other file too 2010-09-12T18:08:43 amstan: that sounds like work :p 2010-09-12T18:10:37 Hello could you help me with it ? 2010-09-12T18:10:39 http://pastebin.ca/1939119 2010-09-12T18:10:45 just copy past in log.txt 2010-09-12T18:10:52 *** rafsoaken has quit IRC (Ping timeout: 252 seconds) 2010-09-12T18:11:25 jmcarthur: got an idea for you, for "fun" - what would a ranking of just games in the last 24h look like ? I was thinking a rolling 24h rankings might be interesting for those that want to see something more volatile :-) 2010-09-12T18:13:00 Zaphus: it shouldn't be any more interesting than normal since old submissions aren't counted anyway 2010-09-12T18:13:06 it would just have wider confidence intervals 2010-09-12T18:13:11 *** Zeiris has joined #aichallenge 2010-09-12T18:13:53 i mean a single submission doesn't change in skill over time anyway 2010-09-12T18:14:19 *** gzurl has quit IRC (Remote host closed the connection) 2010-09-12T18:14:32 elijahbal: you're sending negative ships on the turn you get dropped 2010-09-12T18:14:41 jmcarthur: no, but who you play changes - still, you are prob. right. 2010-09-12T18:14:46 elijahbal: assuming you're p2? 2010-09-12T18:14:50 then http://pastebin.ca/1939121 2010-09-12T18:14:58 contestbot: quit 2010-09-12T18:19:58 *** contestbot has joined #aichallenge 2010-09-12T18:19:59 *** ChanServ sets mode: +o contestbot 2010-09-12T18:20:03 @yay 2010-09-12T18:20:08 Double rainbow all the way! 2010-09-12T18:20:11 Zaphus: the problems with the rankings is that the script asks the DB for every game that's ever been played. 2010-09-12T18:20:23 Zaphus: It just needs to be more incremental about things 2010-09-12T18:20:24 elijahbal: You don't own the planet 2010-09-12T18:20:36 elijahbal: YOu are trying to send from a neutral planet. 2010-09-12T18:20:38 http://pastebin.ca/1939128 2010-09-12T18:20:43 danielvf: bayeselo can't do that i don't think 2010-09-12T18:20:45 ok 2010-09-12T18:20:46 danielvf: fair enough! can a tarball be made for jmcarthur to play with ? 2010-09-12T18:20:51 we'd need to change scoring systems i think 2010-09-12T18:21:02 *** Theobon has joined #aichallenge 2010-09-12T18:21:04 Zaphus, danielvf: i don't need a tarball right now 2010-09-12T18:21:12 this stuff can wait 2010-09-12T18:21:24 i'm a bit more eager to get the specifications linked from the web site though 2010-09-12T18:21:32 jmcarthur: I think we just to incrementaly add data from the database to a file containing all games 2010-09-12T18:21:32 http://code.google.com/p/ai-contest/issues/detail?id=135 2010-09-12T18:21:39 that bayslo can work from 2010-09-12T18:21:44 danielvf: ah, that could work 2010-09-12T18:22:00 danielvf: or have the rankings process be persistent 2010-09-12T18:22:04 bayeselo will still continue to get slower, but i guess that's a small load compared to the query 2010-09-12T18:22:07 and on intiial load pull everything up 2010-09-12T18:22:28 and every 15 minutes or whatever add new data 2010-09-12T18:22:32 might be nice for the next competition to switch to an incremental scoring system 2010-09-12T18:22:38 but a better one than what we had for most of tron 2010-09-12T18:22:47 i still vote for trueskill 2010-09-12T18:22:52 http://pastebin.ca/1939129 here ? 2010-09-12T18:22:58 What s the problem ? 2010-09-12T18:23:58 elijahbal: you are always playerID = 1 2010-09-12T18:24:02 well dunno it works 2010-09-12T18:24:18 no matter which player you are in the viewer 2010-09-12T18:24:29 dude, if we were able to record some error logs on the server from our bots and retrieve them later i would totally record the server load when my bots starts so i can show the correlation between it and how bad my bot performs :P 2010-09-12T18:25:09 of course if we go back to VMs then that won't work 2010-09-12T18:26:13 *** AegNuddel has joined #aichallenge 2010-09-12T18:26:30 will be difficult to remove all this stuff. 2010-09-12T18:26:33 time to sleep 2010-09-12T18:26:35 thank you 2010-09-12T18:27:01 *** cYmen has joined #aichallenge 2010-09-12T18:27:04 * cYmen pokes in. 2010-09-12T18:27:06 Bonsoir. 2010-09-12T18:27:24 *** AegNuddel has left #aichallenge ("Ex-Chat") 2010-09-12T18:28:14 *** sinan has quit IRC (Ping timeout: 240 seconds) 2010-09-12T18:30:32 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-09-12T18:31:41 Zaphus: i will wait to ask for tarballs until the server issues are a bit more under control 2010-09-12T18:32:04 could be several days 2010-09-12T18:33:32 jmcarthur: indeed, I hope things get sorted out. end users aren't going to have this much info as to whats going on and why their bot only gets one game every X hours (if at all) - be good to get things rolling again 2010-09-12T18:33:47 yeah 2010-09-12T18:34:09 So which one of you can make me a VM'd bot jail? 2010-09-12T18:35:06 bartwe: wow dude 2010-09-12T18:36:20 apparently it's time to step up my game 2010-09-12T18:36:30 danielvf: is this to be distributed or on the one server? 2010-09-12T18:36:44 jmcarthur: either 2010-09-12T18:37:57 *** tebele has joined #aichallenge 2010-09-12T18:38:01 Hi 2010-09-12T18:38:39 I've uploaded 30 minutes ago a new pack, and it doesn't appear into my "recent submissions". 2010-09-12T18:39:44 tebele: did you get the "Success" page after upload? 2010-09-12T18:39:56 Yes, but I didn't recieve any email 2010-09-12T18:40:51 *** pierce_ has joined #aichallenge 2010-09-12T18:43:10 *** phreeza has joined #aichallenge 2010-09-12T18:44:43 danielvf: most secure without a vm is probably to use ptrace to filter system calls. the problem is it kills the cpu, even makes the java entries pretty much useless. 2010-09-12T18:45:45 wouldn't be a huge problem if we are willing to run fewer matches for longer periods of time (longer time limits) 2010-09-12T18:45:49 but that might suck 2010-09-12T18:49:27 *** PabloM has joined #aichallenge 2010-09-12T18:49:35 hi 2010-09-12T18:51:37 I'm having troubles running the simulations, sometimes for no aparent reason the game ends with a completly different state that the previous turn... Is anyone having this problem? 2010-09-12T18:52:01 PabloM: as in all your planets suddenly go dark? 2010-09-12T18:52:30 yes 2010-09-12T18:52:44 that means you were kicked for making an invalid move 2010-09-12T18:52:51 PabloM: you've timed out, crashed, or sumbitted an illegal move 2010-09-12T18:53:13 jmcarthur: What is the long term, right thing to do 2010-09-12T18:53:14 or is it just illegal moves? 2010-09-12T18:53:24 the simulator really should, like, print to stderr "player 2 did x which is illegal because y" 2010-09-12T18:53:38 i think timeouts might do it too, actually 2010-09-12T18:53:51 *** computerwiz_222 has left #aichallenge 2010-09-12T18:54:27 It the log.txt file supposed to have the moves? because mine is empty :S 2010-09-12T18:54:47 *** hornairs has quit IRC (Remote host closed the connection) 2010-09-12T18:55:14 *** dstufft_ has quit IRC (Ping timeout: 252 seconds) 2010-09-12T18:55:20 so I'm trying to log to my own file every move, but I'm wathching very weird behaivours, like ignoring moves 2010-09-12T18:55:51 *** dstufft_ has joined #aichallenge 2010-09-12T18:56:08 PabloM: if you have an old starter pack you may want to update, they fixed some bugs that sound like that after the initial release 2010-09-12T18:56:22 http://ai-contest.com/forum/viewtopic.php?f=18&t=546 2010-09-12T18:56:26 I have that problem too 2010-09-12T18:56:33 danielvf: our options are basically: ptrace, bsd jail, or vm. each has some tradeoffs, and none are perfect even from a pure security standpoint 2010-09-12T18:56:35 jmreardon: I downloaded it today 2010-09-12T18:56:55 PabloM: probably not then 2010-09-12T18:57:04 ptrace is probably the closest to perfect, but the speed hit is killer 2010-09-12T18:57:11 What about vm + jail? 2010-09-12T18:57:52 it would reasonable 2010-09-12T18:58:04 the existing qemu code was probably on the right track. i forget why that was abandoned 2010-09-12T18:58:11 but qemu+kvm would be good 2010-09-12T18:58:14 or headless virtualbox 2010-09-12T18:58:24 we have to be careful with vms though 2010-09-12T18:58:33 *** tebele has quit IRC (Quit: Page closed) 2010-09-12T18:58:41 it's easy to introduce holes if you aren't thorough 2010-09-12T18:59:05 even the lease secure vm would probably be more secure than what we have now though 2010-09-12T18:59:07 *least 2010-09-12T18:59:26 What are multiple threads prohibited? 2010-09-12T18:59:45 mietek: yes 2010-09-12T18:59:54 s/What/Why/ 2010-09-12T18:59:55 :) 2010-09-12T18:59:57 mietek: too hard on server 2010-09-12T19:00:26 Does that mean POSIX threads? 2010-09-12T19:00:30 also a good way to starve other bots 2010-09-12T19:00:33 yeah no pthreads 2010-09-12T19:00:38 you can use green threads 2010-09-12T19:00:45 but only if they all run in a single os thread 2010-09-12T19:00:52 OK, cool. 2010-09-12T19:01:36 *** sigh has joined #aichallenge 2010-09-12T19:01:38 isn't there some bug with central planet and starterpack ? 2010-09-12T19:02:02 danielvf: when you said vm + jail, did you mean chroot or did you mean bsd jail? 2010-09-12T19:02:12 *** Zaphus has quit IRC (Ping timeout: 252 seconds) 2010-09-12T19:02:37 danielvf: if you meant the latter, the vm would only be necessary if we just can't run bsd natively 2010-09-12T19:02:48 if the former, yeah the vm is very necessary 2010-09-12T19:02:56 and the chroot would still be called for 2010-09-12T19:03:09 *** Zaphus has joined #aichallenge 2010-09-12T19:03:30 *** rogue780 has quit IRC (Ping timeout: 265 seconds) 2010-09-12T19:03:55 Ah 2010-09-12T19:04:05 I'm guessing not not BSD 2010-09-12T19:04:21 Hm.. what's the standard way to get some debug output when running locally? (Python) 2010-09-12T19:04:46 *** dstufft_ has quit IRC (Ping timeout: 252 seconds) 2010-09-12T19:05:47 i've never made a very secure jail before though. only basic chroots on machines that only i should have had access to in the first place and stuff like that 2010-09-12T19:06:02 i have some experience with... crazier things, but i can't share that here 2010-09-12T19:06:15 Ah. Okay 2010-09-12T19:06:16 but it's not very applicable 2010-09-12T19:06:23 it's kernel-level security rather than userspace security 2010-09-12T19:07:03 cYmen: I don't know python, but if it can redirect/duplicate stderr to a file, that works 2010-09-12T19:07:19 *** AlliedEnvy has quit IRC (Read error: Connection reset by peer) 2010-09-12T19:07:37 *** dstufft_ has joined #aichallenge 2010-09-12T19:07:51 jmreardon: So, just log to a file is what you're saying? :) 2010-09-12T19:07:57 w00t, GUI tcp client is working 2010-09-12T19:08:29 cYmen: oh, ya, you could do that too :) 2010-09-12T19:08:38 hm :) 2010-09-12T19:08:38 greghaynes: So a human can play games? 2010-09-12T19:09:11 No, just to make it easier to use the tcpclient 2010-09-12T19:09:22 It should also compile on win if anyone want to try 2010-09-12T19:10:15 Heres, the source: http://github.com/greghaynes/QTcpPlanetWars 2010-09-12T19:10:56 *** retybok has quit IRC (Quit: leaving) 2010-09-12T19:11:36 greghaynes: cool 2010-09-12T19:12:54 greghaynes: how can I build it? 2010-09-12T19:13:05 cmake .;make 2010-09-12T19:13:27 You want to run it from a terminal and leave the terminal open 2010-09-12T19:13:33 * edcba tries tcpserver :) 2010-09-12T19:13:50 or wait about 20 or 30 mins until I hook up the user friendly debug info 2010-09-12T19:14:03 i lost of course ... 2010-09-12T19:14:16 *** kozlovsky has quit IRC (Quit: Page closed) 2010-09-12T19:14:18 *** Strepto has quit IRC (Ping timeout: 252 seconds) 2010-09-12T19:14:36 greghaynes: Object::connect: No such signal QTcpSocket::error(QAbstractSocket::SocketError socketError) in /tmp/QTcpPlanetWars/game.cpp:22 Object::connect: No such signal QProcess::error(ProcessError error) in /tmp/QTcpPlanetWars/game.cpp:34 2010-09-12T19:14:40 theortically, a 200 turn game can last what? 6, 7 minutes or so? 2010-09-12T19:14:45 greghaynes: (Trying it with a .py file) 2010-09-12T19:14:55 dstufft_: 3:20 2010-09-12T19:15:02 RainCT: Yea, thats just debug info 2010-09-12T19:15:09 ok, 3:20 + server overhead 2010-09-12T19:15:18 Naktibalda: oh right the turns are in parellel not in sequence 2010-09-12T19:15:19 RainCT: It should still work with that though 2010-09-12T19:15:39 Naktibalda: 200 seconds each bot, gives me 400s 2010-09-12T19:15:43 greghaynes: well I guess it doesn't handle Python, does it? 2010-09-12T19:15:52 jmreardon: I think they are concurrent 2010-09-12T19:15:52 It does...I ran mine using pythin 2010-09-12T19:16:07 I think I know why yours may not work, let me add some debug stuff to make sure 2010-09-12T19:16:08 *** Theobon has quit IRC (Ping timeout: 252 seconds) 2010-09-12T19:16:28 greghaynes: is it changing the working directory into that of the script? 2010-09-12T19:16:55 Dont think so, im testing that atm 2010-09-12T19:16:56 RainCT: from the bots point of view they move simultaneously but I'd be surprised if they were actually executed in parallel 2010-09-12T19:17:09 *** rogue780 has joined #aichallenge 2010-09-12T19:17:26 jmreardon: no they are actually in parallel 2010-09-12T19:17:57 Well, simply adding output redirection to the call doesn't work.. 2010-09-12T19:18:26 jmcarthur: finished extracting the IO parts from your code? 2010-09-12T19:18:45 jmcarthur: thanks 2010-09-12T19:23:35 *** Ported has joined #aichallenge 2010-09-12T19:24:32 *** yuwei has joined #aichallenge 2010-09-12T19:25:28 Java question - I have a bash script that runs my bot against all the sample bots on all maps and will extend this to run against other versions of my bot that I create... This is way slower than it ought to be! I think its the JVM start up time that kills the performance, has anyone got anything like this running where it does not need to start a new vm for each bot every game? 2010-09-12T19:27:24 Ported: Are you using the default start pack MyBot? 2010-09-12T19:27:39 because the input is really really slow 2010-09-12T19:28:33 yeah i am using that as my base 2010-09-12T19:28:43 So its writen like C 2010-09-12T19:28:45 do you mean the way it reads stdin? 2010-09-12T19:29:02 the slowest way to write java is to write it like c 2010-09-12T19:29:03 yea 2010-09-12T19:29:14 You should use a LineNumberReader 2010-09-12T19:29:20 i had not really looked at that, I see what you mean 2010-09-12T19:29:50 not that dirty while((c=system.out.read())==whatever){} 2010-09-12T19:30:13 Its make quite a difference --at lest for me 2010-09-12T19:30:46 However JVM startup time is definatly going to make the run all slow 2010-09-12T19:31:06 why a LineNumberReader? And not just a BufferedReader? 2010-09-12T19:31:15 takes about 10min for mine right now... but i don't use a second for a turn like i should 2010-09-12T19:31:28 Ported: About the same 2010-09-12T19:31:49 Buffers help... buts thats probably already done by the system 2010-09-12T19:32:08 I doubt one is slower than the other 2010-09-12T19:32:17 * cYmen grunts. How hard can it be to output something? :) 2010-09-12T19:32:29 LOL: http://www.benzedrine.cx/planetwars/canvas?game_id=1284333492|UloPe-04.6|dwc-002 2010-09-12T19:32:34 delt0r_: ok thanks ill change that now 2010-09-12T19:34:30 ulope: haha 2010-09-12T19:34:34 sweet :) 2010-09-12T19:34:48 ulope: ha! 2010-09-12T19:35:13 dwc-002 clearly savoring its victory :P 2010-09-12T19:35:26 RainCT: I just ran mine without being in the same dir as my bot 2010-09-12T19:35:31 I was kind of wonderung waht was going on looking at the tcp output... a solid minute of go go 2010-09-12T19:35:35 *** delt0r___ has joined #aichallenge 2010-09-12T19:35:36 RainCT: I did add some debug into to the terminal, see if it helps 2010-09-12T19:35:57 seems it happens with mine too ulope ... 2010-09-12T19:36:10 hm..I think if I try to write to a file my python gets so slow it times out... 2010-09-12T19:36:17 Its cool, I can just keep clicking play game and it adds a new game, does both simultaneously 2010-09-12T19:36:53 dwc-002 needs to savor every victory 2010-09-12T19:36:56 i won against nah22 lol 2010-09-12T19:37:15 RainCT: You may just need to +x your py file 2010-09-12T19:37:27 *** delt0r_ has quit IRC (Ping timeout: 265 seconds) 2010-09-12T19:37:29 *** rogue780_ has joined #aichallenge 2010-09-12T19:37:56 I am not getting email notification. any idea? 2010-09-12T19:38:36 yuwei: 2010-09-12T19:38:36 http://ai-contest.com/forums/viewtopic.php?f=18&t=479 2010-09-12T19:38:44 greghaynes: Ah. Yup, with +x it works. 2010-09-12T19:39:34 greghaynes: but I timeout after a couple turns 2010-09-12T19:39:43 *** evgenius has left #aichallenge 2010-09-12T19:39:46 Must be your bot... 2010-09-12T19:39:51 Apophis: what's that country setting to do with my question? 2010-09-12T19:40:01 I am running two simultaneously and its working great 2010-09-12T19:40:14 Doesn't happen with tcp.c 2010-09-12T19:40:18 Even better, I won both my games 2010-09-12T19:40:20 hrmmm 2010-09-12T19:40:23 heh, nice 2010-09-12T19:40:24 country setting? its the bug trhead for email... 2010-09-12T19:40:34 ah, probably 2010-09-12T19:40:34 yuwei, ^^ 2010-09-12T19:40:45 *** preetum has joined #aichallenge 2010-09-12T19:40:46 amstan: the iowait is due to mysql having to seek alot, even when checking just a hundred games or so, it has to seek all over the disk because the gamestates are so huge 2010-09-12T19:40:48 oh nice, I've just won my first game on tcp :P 2010-09-12T19:41:07 Apophis: thanks, miss out the last 9 2010-09-12T19:41:08 danielvf: don't we have an index? 2010-09-12T19:41:13 we do 2010-09-12T19:41:26 RainCT: It worked? 2010-09-12T19:41:26 But it still has to seek to each record that matches the index 2010-09-12T19:41:32 amstan: I'm adding two indexes to cover the entire fields that profile code uses 2010-09-12T19:41:52 This will let it work entirely in the index 2010-09-12T19:41:53 greghaynes: nope, I played another game with tcp.c to verify it wasn't my connection being stupid (and won that one :)) 2010-09-12T19:42:00 hrm 2010-09-12T19:42:29 Try again, use the username testbot (which is what ive been testing with), see if it happens in the same number of turns 2010-09-12T19:42:45 is the engine down for everyone, or is it just me? 2010-09-12T19:43:08 i can't access my user page for the tcp server 2010-09-12T19:43:16 *** SavageCamel2 has quit IRC (Ping timeout: 252 seconds) 2010-09-12T19:43:34 I'm having trouble on the tcp site 2010-09-12T19:43:42 rogue780_: i'm sure you will! i'm leaving dmj.4 running until I can get something else to work. 2010-09-12T19:44:07 dmj111, your bot destroys me 2010-09-12T19:44:11 I uploaded code ~15 hours ago, but I haven't seen any results on my profile page yet 2010-09-12T19:44:41 you are too kind. I haven't followed it very close this weekend, but it looks like there are several very strong bots up there now. 2010-09-12T19:44:50 tcp server webinterface down? 2010-09-12T19:44:59 hmpf..ok does anybody have a sample bot for me that outputs stuff to a file for debugging? 2010-09-12T19:45:00 ulope, it appears to be 2010-09-12T19:45:18 cYmen: what language? 2010-09-12T19:45:42 Oh, python. 2010-09-12T19:45:45 cYmen, I have a Log() method for my java bot 2010-09-12T19:46:06 cYmen: I'm using http://github.com/RainCT/planetwars-python-kit 2010-09-12T19:46:16 Naktibalda, you there? 2010-09-12T19:46:23 no 2010-09-12T19:46:35 I have not checked that out yet, but there is also a post in the forum where someone shows how to use the logging module. 2010-09-12T19:46:36 if you use the python engine source, you can get some actual debugging working 2010-09-12T19:46:37 ok nevermind 2010-09-12T19:46:46 :))) what's a question? 2010-09-12T19:47:00 RainCT: Thanks, I'll check it out. 2010-09-12T19:47:01 http://twitter.com/DeadfastNet/status/24325670567 2010-09-12T19:47:03 lol 2010-09-12T19:47:11 is the uploader/engine down? 2010-09-12T19:47:15 Naktibalda, have you thought of using Quercus for your bot at all? 2010-09-12T19:47:21 no 2010-09-12T19:47:27 IIRC, it allows you to put your php files in a jar and then run them with java. 2010-09-12T19:47:49 It might be a good temporary workaround since it would allow you to officially submit your bot 2010-09-12T19:48:15 I'm not in a hurry 2010-09-12T19:48:58 dstufft_: some angry customer eh lol 2010-09-12T19:48:58 *** preetum has left #aichallenge 2010-09-12T19:49:26 *** dukenukem has joined #aichallenge 2010-09-12T19:50:08 greghaynes: hmm, actually it's at turn 1 :P 2010-09-12T19:50:31 odd... 2010-09-12T19:50:31 greghaynes: so it may well be because of the locale 2010-09-12T19:50:37 It's really quite mesmerizing to watch those chains of bartwe and similar bots 2010-09-12T19:50:57 RainCT: hrm im pretty sure I convery all text to ascii 2010-09-12T19:51:01 yeah i wonder if he's actually computing minimum spanning tree for a supply chain 2010-09-12T19:51:27 greghaynes: No, I mean I have to export LANG=en_US for stuff to work since else it gets commas instead of dots for float separators 2010-09-12T19:51:48 greghaynes: and your GUI may not be forwarding that environment variable to the process it launches 2010-09-12T19:52:04 hrmm 2010-09-12T19:52:27 *** craigferguson has joined #aichallenge 2010-09-12T19:52:29 a1k0n: sometimes after a disturbance the flows change direction... that was quite a surprise to me 2010-09-12T19:52:46 ah I meant change paths not direction per se 2010-09-12T19:52:55 that makes sense 2010-09-12T19:53:12 RainCT: quick intro on how to use that with the visualizer for local experiments? 2010-09-12T19:53:29 contestbot: time 2010-09-12T19:53:29 Zaphus: 07:53 PM, September 12, 2010 2010-09-12T19:53:43 *** SavageCamel2 has joined #aichallenge 2010-09-12T19:54:08 dang, only one game in the last 6 hours :-( random numbers 2010-09-12T19:55:27 cYmen: eg. java -jar tools/PlayGame.jar maps/map7.txt 500 500 log.txt "java -jar example_bots/RageBot.jar" "python stupidbot.py --log output.log" | java -jar tools/ShowGame.jar 2010-09-12T19:56:18 cYmen: see stupidbot.py, loggingbot.py, etc. for examples on how to use it (although you'll have to look into the source to see what other methods are available) 2010-09-12T19:57:46 sweet 2010-09-12T19:59:40 ok, i don't know how to fix my code for email. my LANG=en_NZ.UTF-8 do i need to change it to en_US? 2010-09-12T20:01:40 what is the host and port to connect to the tcp version? http://www.benzedrine.cx/planetwars/ 2010-09-12T20:01:50 *** xkpe has quit IRC (Ping timeout: 264 seconds) 2010-09-12T20:01:51 213.3.30.106 9999 2010-09-12T20:02:50 if anybody needs the tcp client, I have a precompiled package in a zipfile for you at http://www.errgaming.com/tcp.zip 2010-09-12T20:03:16 all necessary files are included to run on windows, except, of course your bot 2010-09-12T20:03:26 It has been tested on Windows XP and Windows 7 home premium 2010-09-12T20:04:18 a1k0n, destroyed my bot 2010-09-12T20:05:15 good night all 2010-09-12T20:05:15 oo, rogue780 : do you know how to compile Qt apps on win? 2010-09-12T20:05:19 gn 2010-09-12T20:05:34 *** RainCT has quit IRC (Remote host closed the connection) 2010-09-12T20:05:45 wow, played my first game against the tcp client, very cool --- the tcp version rocks 2010-09-12T20:05:59 *** AJC has joined #aichallenge 2010-09-12T20:06:08 ulope: hi, you around? 2010-09-12T20:06:08 greghaynes, sorry, no. I never do C/C++ programming if I can help it, and then it is only text-based programs 2010-09-12T20:06:16 aw 2010-09-12T20:06:32 I made a GUI for tcp version that should work on win, I just dont have a win box 2010-09-12T20:06:34 AJC: Yes. 2010-09-12T20:06:53 tcp.c should do optional compression maybe... 2010-09-12T20:06:53 ulope: I think I found a bug in the my_fleets vs. enemy_fleets. they all seem to be enemy. 2010-09-12T20:07:15 It should have user/pass 2010-09-12T20:07:49 edcba: not as much compression as just transmitting binary data.. 2010-09-12T20:08:05 *** fitzgen has joined #aichallenge 2010-09-12T20:08:10 greg, give me the source and I'll see what I can do. if it's just a matter of compiling/linking, I can probably figure it out 2010-09-12T20:08:15 AJC: Sure? How are you noticing that? 2010-09-12T20:08:21 rogue780_: http://github.com/greghaynes/QTcpPlanetWars 2010-09-12T20:08:24 caytchen: easier to add compression i think 2010-09-12T20:08:25 rogue780_: Thanks for trying 2010-09-12T20:08:26 debug logging :-) 2010-09-12T20:08:41 ulope: I mean the planet's functions. 2010-09-12T20:09:04 planet.reinforcement_fleets is empty. and planet.attacking_fleets contains my own fleets. 2010-09-12T20:09:14 AJC: Ah. did you read the docstrings? They are working from the point of view of the planet 2010-09-12T20:09:53 ulope: no if the planet is neutral they're all attacking? 2010-09-12T20:09:59 ulope: s/no/so 2010-09-12T20:10:25 ulope: hmm, I can't think of a use for it being that way. 2010-09-12T20:10:32 but I can work around i.t 2010-09-12T20:10:39 AJC: Yeah thats a bit unfortunate. I will probably put a warning in there or even change it. a neutral planet cant have reinf. fleets because there are no neutral fleets 2010-09-12T20:11:14 ulope: how about my_fleets and enemy_fleets (so ditch attacking / reinforcements) 2010-09-12T20:11:17 what are your experiences against rage-bot ? 2010-09-12T20:11:19 *** rogue780 has quit IRC (Quit: Leaving) 2010-09-12T20:11:39 canihazbacon: hehe, I won 3% of the maps in my latest tournament :-) 2010-09-12T20:11:51 :D 2010-09-12T20:12:02 was testing against other more sensible bots, lol. 2010-09-12T20:12:15 *** caytchen has quit IRC (Ping timeout: 265 seconds) 2010-09-12T20:12:15 ? 2010-09-12T20:12:19 AJC: ;) I had those before switching to attacking/reinforcement but then I had the problem to differentiate between my_outgoing and my_incomming and enemy_incomin getc 2010-09-12T20:12:50 i easily beat all other bots, but rage-bot is a different story... :D 2010-09-12T20:13:20 ulope: OK, with good naming it could work. but i'll work around it. thanks. 2010-09-12T20:13:43 anyone know what the two numbers are in the PlayGame.jar command? java -jar tools/PlayGame.jar maps/map7.txt 500 500 log.txt "java -jar example_bots/RageBot.jar" "python stupidbot.py\ 2010-09-12T20:14:01 timeout (ms) and move limit 2010-09-12T20:14:10 thanks! 2010-09-12T20:14:11 can't remember what order 2010-09-12T20:14:18 whats a move limit? 2010-09-12T20:14:44 AJC: I agree that those methods are not the hight of API design ;) But I continue to work on it (and some other people on github too, you should pull in the lateste changes btw.) 2010-09-12T20:14:44 if the game isn't over after that many moves, then the winner is decided by hte number of ships 2010-09-12T20:14:56 ahh 2010-09-12T20:15:05 thanks 2010-09-12T20:15:14 ulope: not used git much before, it's weird. it doesn't like merging much... 2010-09-12T20:15:19 is there a new python starter package that fixes bugs? 2010-09-12T20:15:33 ulope: git seems to like refusing to do work more than it helps :-) 2010-09-12T20:15:49 hah, git loves merging! 2010-09-12T20:15:53 You just have to ask nicely 2010-09-12T20:16:10 AJC: I guess you have to get over teh initial learning "hump" then it becomes quite pleasant 2010-09-12T20:16:12 how do I do it? 2010-09-12T20:16:16 git merge 2010-09-12T20:16:22 :) 2010-09-12T20:16:28 is the starter package on the site the newest or should I head to forums to grab it? 2010-09-12T20:16:32 greghaynes: :D 2010-09-12T20:16:36 error: Your local changes to 'planetwars/game.py' would be overwritten by merge. 2010-09-12T20:16:42 If you want it to automerge your changes with upstream, just commit before you pull 2010-09-12T20:16:47 AJC: Always commit first 2010-09-12T20:17:45 git won't even commit when I ask it to. 2010-09-12T20:17:53 just gives me a long list of stuff I changed 2010-09-12T20:17:54 hehe 2010-09-12T20:17:56 it's weird. 2010-09-12T20:18:17 git add each file you want to add to the commit 2010-09-12T20:18:30 Thats where it differs from some other similar dcs's 2010-09-12T20:18:33 I don't want to add, just commit. 2010-09-12T20:18:35 AJC: Which OS are you using? 2010-09-12T20:18:41 the latest Ubuntu 2010-09-12T20:18:49 *** oh_cripes has joined #aichallenge 2010-09-12T20:18:56 AJC: git commit -a does what you expect 2010-09-12T20:18:57 I use bzr with pleasure, but this git thing is weird. 2010-09-12T20:19:05 a1k0n: apparently, thanks! 2010-09-12T20:19:10 AJC: Then you can at least use something like git-gui to make it abit more friendly 2010-09-12T20:19:33 AJC: if you have stuff in your tree you don't want to commit before merging you can git stash; ; git stash apply 2010-09-12T20:19:36 a1k0n: But is also the sledgehammer of commit options 2010-09-12T20:20:03 ulope: yes indeed. but that's what most other source control things do 2010-09-12T20:20:18 don't use git commit -a 2010-09-12T20:20:22 ulope: ah, you added the turn count... explains the conflict. 2010-09-12T20:20:25 the index will confuse the hell out of you until you realize that it exists 2010-09-12T20:20:30 git add is your friend D: 2010-09-12T20:20:56 git uses a staging directory between your working copy and the repo 2010-09-12T20:20:56 AJC: Its just explicit about what to commit rather than bzr which assumes you want to add all your changes to the commit 2010-09-12T20:21:01 dstufft_: Even more the index is your friend 2010-09-12T20:21:14 you "add" a file to your staging area, then you commit your staging area 2010-09-12T20:21:47 ...I have multiple bots playing on the tcp server right now...and I think two of them are playing each other 2010-09-12T20:22:00 You can even stage (and commit if you want) single lines out of a big changed file 2010-09-12T20:22:11 *** canihazbacon has left #aichallenge 2010-09-12T20:22:13 greghaynes: with bzr you can specify on the command line, with git you just need a cryptic -a flag to access the sub-menu... "commit" is apparently just there for cosmetic reasons :-) 2010-09-12T20:22:38 AJC -a means ALL 2010-09-12T20:22:46 Time to grab some sleep. Night all 2010-09-12T20:22:58 dstufft_: I'm glad git commit told me then :-) 2010-09-12T20:23:02 dstufft_: oh wait. 2010-09-12T20:23:02 * rogue780_ facepalm 2010-09-12T20:23:06 I forgot to recompile my bot 2010-09-12T20:23:26 that's one advantage of python 2010-09-12T20:23:26 AJC: git commit was commiting what you added to your staging area (in this case nothing, since you didn't git add anything) 2010-09-12T20:23:42 why is he complaining about git? git is awesome 2010-09-12T20:23:52 contestbot: seen j3camero 2010-09-12T20:23:52 amstan: j3camero was last seen in #aichallenge 3 hours, 36 minutes, and 38 seconds ago: jmcarthur: when the site is up... 2010-09-12T20:23:54 I like svn, but that's just me 2010-09-12T20:23:57 amstan: i can only assume brain damage 2010-09-12T20:23:59 what 2010-09-12T20:24:03 someone likes svn 2010-09-12T20:24:04 ?? 2010-09-12T20:24:15 ouch 2010-09-12T20:24:15 rogue780_: dude.. gtfo, lol 2010-09-12T20:24:25 You must love pain 2010-09-12T20:24:27 beep boop does not compute 2010-09-12T20:24:28 Give a cheer for jmcarthur! 2010-09-12T20:24:29 yeah, it's fine. it does what you want to do. 2010-09-12T20:24:31 My bot just lost against SUPER_EXTRA_AWESOME_BOT 2010-09-12T20:24:36 Game specs are live http://ai-contest.com/specification.php 2010-09-12T20:24:40 @rankings 2010-09-12T20:24:41 amstan: Top 10 players: dmj111(2835), braddunbar(2815), albert(2792), seigen(2732), Kavish(2707), drewrobb(2695), Obeleh(2679), wtfbbq(2679), sighbot(2676), tinasky(2661) 2010-09-12T20:25:12 indent for the ul is way too high 2010-09-12T20:25:48 well, in my defense, all I ever do is commit or checkout 2010-09-12T20:25:52 never merged or anything 2010-09-12T20:25:57 with one person? 2010-09-12T20:26:02 just me 2010-09-12T20:26:13 i've never had the opportunity to code with anyone else 2010-09-12T20:26:14 rogue780_: still a good idea to use git 2010-09-12T20:26:24 No local commit are such a PITA 2010-09-12T20:26:28 rogue780_: Even in that case i find git very usefull 2010-09-12T20:26:29 er, is 2010-09-12T20:26:41 I like my free private repo. I have to pay for a private repo with github 2010-09-12T20:26:50 rogue780_: git is private.. 2010-09-12T20:26:56 as private as it can ever be 2010-09-12T20:27:02 and being an unemployed college student, free is awesome 2010-09-12T20:27:03 for svn you need a server 2010-09-12T20:27:11 amstan, and I have one 2010-09-12T20:27:12 All you need is ssh to host git somewhere 2010-09-12T20:27:19 So use a lab computer at school 2010-09-12T20:27:27 greghaynes: meh... http 2010-09-12T20:27:43 http is le suck for git hosting 2010-09-12T20:27:45 my problem is that I have 3 computers that I develop on at any given time and I want to keep things synchronized. svn server does that for me 2010-09-12T20:28:02 so does git if you configure it 2010-09-12T20:28:04 You know what else does that.... ;) 2010-09-12T20:28:10 just use whatever 2010-09-12T20:28:12 rogue780_: so... git pull from all your comps before starting to work. 2010-09-12T20:28:13 and write code 2010-09-12T20:28:20 rogue780_: you don't need a central place 2010-09-12T20:28:45 What sucks is if youre somewhere without internet and using svn 2010-09-12T20:28:53 meh. I like being able to pull my netbook out at school and sync it then if I forgot. 2010-09-12T20:29:02 greghaynes, phone tether :) 2010-09-12T20:29:03 what sucks is if you're .. using svn 2010-09-12T20:29:07 fixed 2010-09-12T20:29:16 rogue780_: This is what i use git --keeping my 5 computers "synced" 2010-09-12T20:29:17 jeez. all the hate 2010-09-12T20:29:17 rogue780_: Doesnt work on plane, I have that too ;) 2010-09-12T20:29:25 rogue780_: you can do all these things with git, by default 2010-09-12T20:29:25 svn: atleast it's not cvs 2010-09-12T20:29:33 Ill support that 2010-09-12T20:29:33 git hadels the case where your forgot to "commit" way better than svn 2010-09-12T20:29:44 Its definitely not cvs, thank god 2010-09-12T20:29:50 svn > cvs > nothing 2010-09-12T20:29:56 > rcs 2010-09-12T20:29:59 lol 2010-09-12T20:30:00 danielvf: noting>floppy 2010-09-12T20:30:00 svn > cvs > rcs 2010-09-12T20:30:00 well, actually 2010-09-12T20:30:08 svn > rcs > cvs 2010-09-12T20:30:08 sorcesafe? 2010-09-12T20:30:16 sourcesafe 2010-09-12T20:30:16 paper? 2010-09-12T20:30:19 paper! 2010-09-12T20:30:20 lol 2010-09-12T20:30:32 yea yea, svn > rcs > paper > cvs 2010-09-12T20:30:45 write it in stone.. and use carbon dating for revision control 2010-09-12T20:31:18 amstan: Have you seen my pretty gui tcp client? 2010-09-12T20:31:23 http://tom.preston-werner.com/2009/05/19/the-git-parable.html 2010-09-12T20:31:28 the git parable is ftw 2010-09-12T20:31:33 greghaynes: no, qt? 2010-09-12T20:31:39 Of course 2010-09-12T20:31:43 http://github.com/greghaynes/QTcpPlanetWars 2010-09-12T20:32:07 greghaynes: GNUQTcpPlanetWars++ 2010-09-12T20:32:11 *** AJC has left #aichallenge 2010-09-12T20:32:13 lol 2010-09-12T20:32:15 greghaynes: can you add any more acronyms? 2010-09-12T20:32:20 Yes 2010-09-12T20:32:24 it's C though.. ew 2010-09-12T20:32:28 QTcpAIPW 2010-09-12T20:32:31 After KDE ive become a pro at adding acronyms 2010-09-12T20:32:40 my cat has decided to take this opportunity to lay in my lap and make typing next to impossible 2010-09-12T20:32:44 *** FlagCapper has quit IRC (Quit: Page closed) 2010-09-12T20:32:58 solution: kill the cat 2010-09-12T20:33:00 QTcpKPlanetWars 2010-09-12T20:33:19 rogue780_: git commit the cat 2010-09-12T20:33:23 OpenQTcpBotPlanetWarsAuto 2010-09-12T20:33:37 greghaynes: you forgot the K 2010-09-12T20:33:40 and the ++ 2010-09-12T20:33:44 and pro at the end 2010-09-12T20:33:52 I havent hooked it into KDE yet 2010-09-12T20:33:56 and enterprise 2010-09-12T20:34:00 So im not allowed to have any K's in the name 2010-09-12T20:34:03 enterprise edition 2010-09-12T20:34:19 No, SE, I have to charge for the EE 2010-09-12T20:34:30 Also, it needs a random number in the name 2010-09-12T20:34:31 GNUOpenQTcpBotKPlanetWars++ Pro -- Enterprise Edition 2010-09-12T20:34:37 5 2010-09-12T20:34:44 4000 2010-09-12T20:34:50 Even better 2010-09-12T20:34:51 wait.. no, 9000 is better 2010-09-12T20:34:53 9001 2010-09-12T20:35:00 it needs to be over 9000 obv 2010-09-12T20:35:15 sounds like something microsoft would release 2010-09-12T20:35:26 how dare you 2010-09-12T20:35:36 microsoft would never release anything with GNU in it's name 2010-09-12T20:35:45 Sure they would. 2010-09-12T20:35:48 *** shawnshekari has joined #aichallenge 2010-09-12T20:35:54 "GNU for Windows(R)" 2010-09-12T20:35:57 :) 2010-09-12T20:36:19 It sounds like an oracle product 2010-09-12T20:36:25 oh snap 2010-09-12T20:36:26 Help > Coming Soon 2010-09-12T20:36:49 :) 2010-09-12T20:36:59 Its the FOSS crito 2010-09-12T20:37:50 Ok, so ill rename it to something more simple... 2010-09-12T20:37:55 How about 'Fred' 2010-09-12T20:38:13 Then youll have no idea what it does and be forced to try it and see 2010-09-12T20:38:21 so.. how does it work? 2010-09-12T20:38:26 i press play game, nothing happens 2010-09-12T20:38:31 Did you put a username in? 2010-09-12T20:38:37 yeah 2010-09-12T20:38:44 maybe i have to chmod +x the python bot 2010-09-12T20:38:48 It should say INFO blah blah blah 2010-09-12T20:38:50 Yea 2010-09-12T20:38:55 greghaynes: yeah.. make it error 2010-09-12T20:39:08 kinda ugly.. 2010-09-12T20:39:11 The error is in there, I just havent hooked it up to a GUI dialog yet 2010-09-12T20:39:12 *** jerkface03 has quit IRC (Ping timeout: 265 seconds) 2010-09-12T20:39:21 make it show the game in real time 2010-09-12T20:39:27 yea yea, in due time 2010-09-12T20:39:33 This was a 1hr job using two file 2010-09-12T20:39:37 er, classes 2010-09-12T20:40:06 pyamstan timed out 2010-09-12T20:40:13 contestbot, tcp 2010-09-12T20:40:14 amstan: tcp = http://www.benzedrine.cx/planetwars/ 2010-09-12T20:40:24 so, would it be against the rules to send fleets with 0 ships? 2010-09-12T20:40:42 Decoys? 2010-09-12T20:40:55 DirtyKeyboard: so... when we fix this, you'll either do something useless, or do an illegal instruction 2010-09-12T20:41:45 I'm just asking because it works on my PlayGame.jar and ShowGame.jar but it's a little.. odd. 2010-09-12T20:42:00 *** Mathnerd314 has quit IRC (Ping timeout: 272 seconds) 2010-09-12T20:42:12 DirtyKeyboard: The server's now ignore it. 2010-09-12T20:42:12 greghaynes: javagamer_Cpp is a sauer guy too 2010-09-12T20:42:20 anyone in here own the bhasker bot? 2010-09-12T20:42:25 Oh, who? 2010-09-12T20:42:29 my bot accidentally did it and I was thinking someone might use it to trick certain bots 2010-09-12T20:42:40 danielvf: alright, cool 2010-09-12T20:43:15 DirtyKeyboard: Yeah. The good bots do ship math, so it's would be more just spam than useful 2010-09-12T20:43:17 ok... that's it.. i'm going spying 2010-09-12T20:43:27 define spying 2010-09-12T20:43:33 lol 2010-09-12T20:43:34 see how ppl are doing this 2010-09-12T20:43:40 define this 2010-09-12T20:43:44 @rankings 2010-09-12T20:43:45 amstan: Top 10 players: dmj111(2835), braddunbar(2815), albert(2792), seigen(2732), Kavish(2707), drewrobb(2695), Obeleh(2679), wtfbbq(2679), sighbot(2676), tinasky(2661) 2010-09-12T20:43:46 that ^^ 2010-09-12T20:43:54 what is that? 2010-09-12T20:44:00 how people are doing ranking? 2010-09-12T20:44:06 amstan: Profiles pages are way faster. 2010-09-12T20:44:20 danielvf: for submissionid? 2010-09-12T20:44:20 Yes, but its been many hours since ive played a match 2010-09-12T20:44:22 I'm guessing amstan is going to see what strats the top bots are using 2010-09-12T20:44:30 amstan: eh? 2010-09-12T20:44:47 I don't have a bot on the server and can't get one apparently. :( 2010-09-12T20:45:33 oh, cool, dmj111's using python 2010-09-12T20:45:37 greghaynes, almost got it to compile on windows 2010-09-12T20:45:54 woo! 2010-09-12T20:46:00 sweet 2010-09-12T20:46:24 wtg greghaynes!!! someone should goto http://awsomenessreminders.com and buy him a subscription 2010-09-12T20:46:56 w00t 2010-09-12T20:46:58 eugh... 2 spaced indents 2010-09-12T20:47:05 i can never understand how ppl can use spaces 2010-09-12T20:47:05 Or at least kudos on ohloh 2010-09-12T20:47:19 amstan: Me either, I figured ill give it a try though 2010-09-12T20:47:31 amstan: starter pack has 2 space indents 2010-09-12T20:47:32 it's horrible 2010-09-12T20:47:41 Yea, tab ftw 2010-09-12T20:47:44 blame the starter pack :P 2010-09-12T20:47:45 those things about how editors can manage spaces.... it's a lie! 2010-09-12T20:47:47 *** ratatata has joined #aichallenge 2010-09-12T20:47:55 *** braddunbar has quit IRC (Quit: leaving) 2010-09-12T20:47:56 Im regularly given crap at python meetings for using tabs 2010-09-12T20:48:28 greghaynes: python meetings? 2010-09-12T20:48:31 tabs are evil 2010-09-12T20:48:32 danielvf, my bot doesn't to ship math....but it's also at version 0.0.2a 2010-09-12T20:48:37 Yea, pycon was just here for example 2010-09-12T20:48:41 er, djangocon, sorry 2010-09-12T20:48:48 rogue780_: Then rock on! 2010-09-12T20:49:00 what 2010-09-12T20:49:04 tabs are the worst 2010-09-12T20:49:05 sorry 2010-09-12T20:49:09 *** rogue780_ is now known as rogue780 2010-09-12T20:49:11 hehe 2010-09-12T20:49:13 4 spaces supremacy 2010-09-12T20:49:15 tabs vs spaces, go! 2010-09-12T20:49:30 oh man.. he debugs a lot.. 2010-09-12T20:49:39 tab/space agnostic here 2010-09-12T20:49:41 debug ftw 2010-09-12T20:50:16 You should at least be mildly impressed at the size 2010-09-12T20:50:18 I mean, I am 2010-09-12T20:51:27 yay - my bot was doing great, but not sure wtf happened 2010-09-12T20:51:30 *** sfsgagi has joined #aichallenge 2010-09-12T20:51:34 http://www.benzedrine.cx/planetwars/canvas?game_id=1284334970|Cyndre|retybok 2010-09-12T20:52:11 to many ships in the air at once it looks like 2010-09-12T20:52:48 Cyndre: epic.. 2010-09-12T20:52:57 retybok is easy to beat (no offense) 2010-09-12T20:53:26 You look like a ragebot 2.0 2010-09-12T20:53:45 benzedrine removes players it hasnt seen in a while, is that right ? (my scores from last night are gone so I can't compare!) 2010-09-12T20:53:51 Yep 2010-09-12T20:54:10 rogue780: shut down^H^H^H^H^H send everything! 2010-09-12T20:54:16 *** rwa has quit IRC (Quit: rwa) 2010-09-12T20:54:31 yea there was some juicy targets you were ignoring there... 2010-09-12T20:54:36 wow 2010-09-12T20:54:40 that visualizer murders my battery 2010-09-12T20:54:42 wow, very epic game 2010-09-12T20:54:47 note to self do not view games on battery o.O 2010-09-12T20:55:13 btw, my bot is the ErrBot 2010-09-12T20:55:49 *** pacotabaco has joined #aichallenge 2010-09-12T20:55:57 ohai, ErrBot master 2010-09-12T20:56:41 danielvf: yay! 2010-09-12T20:56:42 *** deepblue has joined #aichallenge 2010-09-12T20:56:51 say i came late to my own party 2010-09-12T20:56:52 hi again 2010-09-12T20:57:23 Hey deepblue 2010-09-12T20:57:44 seems like ladder is not working 2010-09-12T20:57:57 aw man, cache keeps the old css until you manually hit refresh so the spec looks bad until you do that 2010-09-12T20:58:01 oh well 2010-09-12T20:58:07 bartwe number 4 seems quite good 2010-09-12T20:58:14 jmcarthur: add a ?v=1 to the css 2010-09-12T20:58:15 looks like the whole site is afk atm 2010-09-12T20:58:27 dstufft_: hey thanks! 2010-09-12T20:58:30 * jmcarthur does it 2010-09-12T20:58:50 infact if you do that for all static files, and increment the number everytime it changes 2010-09-12T20:58:57 you can cache shit for like 30 days or even more 2010-09-12T20:59:08 and when a static file changes, you just increase the v=X 2010-09-12T20:59:26 jmcarthur: lol 2010-09-12T21:00:14 greghaynes, do you know where I can get the source for qmake? 2010-09-12T21:00:28 rogue780: Itll be in the Qt dev package 2010-09-12T21:00:32 dstufft_: yeah, i knew the trick but never had to use it, so it didn't come to mind 2010-09-12T21:00:43 rogue780: There should be a win version 2010-09-12T21:00:57 I'm compiling with cygwin, so i need to build the source 2010-09-12T21:01:01 ah 2010-09-12T21:01:03 jmcarthur: ah 2010-09-12T21:01:05 unless there's a cygwin package out there 2010-09-12T21:01:18 i <3 that trick, i have my django projects setup to do it automatically 2010-09-12T21:01:19 Its a very common package... 2010-09-12T21:03:40 *** RobinB has quit IRC () 2010-09-12T21:04:26 Tell me. Notice the site going anyfaster? 2010-09-12T21:04:31 is the upload code broken? for testing, i uploaded starter package untouched from website. and nothing happened at all 2010-09-12T21:04:53 yuwei: try it again? 2010-09-12T21:04:59 greghaynes, I would think. but the only qt packages that cygwin has don't contain qmake. very frustrating 2010-09-12T21:05:01 yuwei: Sometimes you have to do it twice 2010-09-12T21:05:12 I'm downloading the Visuald Studio package right now and see if I can compile it there 2010-09-12T21:05:16 yuwei: and I don't think it's sending emails right now even if it works 2010-09-12T21:05:28 danielvf: woah, yeah - much faster ?! 2010-09-12T21:05:32 *** panchi has joined #aichallenge 2010-09-12T21:05:39 rogue780: Theres some links on this page: http://qt.nokia.com/downloads/ 2010-09-12T21:05:51 Zaphus: Added compression to the pages served by the website 2010-09-12T21:06:06 Cuts the page sizes in 1/4 2010-09-12T21:06:24 *** SavageCamel2 has quit IRC (Quit: Page closed) 2010-09-12T21:06:24 danielvf: nice... now if you could just arrange for my bot to get another game... it's been a while :-) 2010-09-12T21:06:32 danielvf: small patch for immediate upload to server http://code.google.com/p/ai-contest/source/branch?spec=issue138&branch=/branches/css-param 2010-09-12T21:06:52 danielvf: not sending email is fine. Not updating profile page? 2010-09-12T21:07:08 yuwei: It does take about five minutes - it compiles your bot 2010-09-12T21:07:33 jmcarthur: doing it 2010-09-12T21:07:37 danielvf: thanks 2010-09-12T21:08:36 *** xkpe has joined #aichallenge 2010-09-12T21:08:53 lol, the rankings are "temperarily" disabled 2010-09-12T21:09:00 hmm...I really should make my bot send more than one fleet per turn 2010-09-12T21:09:44 Gah. Evil pacotabaco 2010-09-12T21:09:53 jmcarthur: lol 2010-09-12T21:10:06 jmcarthur: my signatuure - the misspelled word 2010-09-12T21:10:11 rogue780: What's happen? 2010-09-12T21:10:20 you always beat me 2010-09-12T21:10:31 my bot is ErrBot 2010-09-12T21:10:35 I just lost to you...again 2010-09-12T21:10:42 ok :) 2010-09-12T21:10:50 danielvf: iswydt 2010-09-12T21:10:57 i'll see the game 2010-09-12T21:11:34 Your css change is in 2010-09-12T21:13:53 does anyone know if submissions are broken at the moment? 2010-09-12T21:14:27 *** Naktibalda has quit IRC (Remote host closed the connection) 2010-09-12T21:14:42 Ported: I think submissions get processed every five minutes 2010-09-12T21:14:55 Ported: and it doesn't send you an email. :( 2010-09-12T21:15:24 does it list your submission in your profile? 2010-09-12T21:16:06 i did one at 2 past but cant see it... not sure if i clicked the button propperly now 2010-09-12T21:18:27 waiting for ai-contest.com... 2010-09-12T21:18:48 You can't get to it? 2010-09-12T21:18:55 apparently not 2010-09-12T21:18:57 It's up and super fast from here. 2010-09-12T21:19:00 Weird 2010-09-12T21:19:34 where is "here"? 2010-09-12T21:19:36 now it works 2010-09-12T21:19:37 Oh, nvm 2010-09-12T21:19:49 jmpespxoreax: try now? 2010-09-12T21:19:54 works now 2010-09-12T21:20:16 how often tcp server adds new game playback to web server? 2010-09-12T21:20:44 hey, I actually won on tcp! :D 2010-09-12T21:20:55 )) 2010-09-12T21:21:09 woo, I moved up on tcp 2010-09-12T21:21:49 *** kersny has joined #aichallenge 2010-09-12T21:25:13 *** rwa has joined #aichallenge 2010-09-12T21:26:37 can I let my bot all the night running on www.benzedrine.cx? or may I stop it? 2010-09-12T21:27:03 pacotabaco: you may do whichever you want 2010-09-12T21:27:35 ok, i just don't want to overload, i'll let it running 2010-09-12T21:27:46 see you tomorrow guys! 2010-09-12T21:28:27 seeing versions of my bot fight it self is epic 2010-09-12T21:28:58 don't you hate losing to an old version? 2010-09-12T21:29:19 I do! 2010-09-12T21:29:27 my old version is so mean 2010-09-12T21:29:29 :( 2010-09-12T21:29:33 teachable moment 2010-09-12T21:29:39 *** hornairs has joined #aichallenge 2010-09-12T21:29:44 ;) 2010-09-12T21:30:22 dwchandler: yes it sucks a lot! i hate it over loose with deepblue or bartwe :P 2010-09-12T21:30:37 yeah 2010-09-12T21:30:51 working on defeating bartwe more regularly 2010-09-12T21:31:05 I always lose to bartwe, but that just let's me see why my bot is so stupid ;-) 2010-09-12T21:31:22 *lets 2010-09-12T21:31:33 dang, I never do that 2010-09-12T21:31:39 maybe my connection is bad here. I am "waiting on benzedrine.cx" now too 2010-09-12T21:32:24 unknown host benzedrine.cx... 2010-09-12T21:32:53 WIN against dhartmei! woo hoo 2010-09-12T21:32:56 how do you pass a java bot to the ./tcp ? I tried any combination that came to my mind 2010-09-12T21:33:37 ratatata: make a wrapper script with "java -jar ..." and call that on the tcp command line 2010-09-12T21:33:53 Cyndre, I have defeated the 2010-09-12T21:34:01 thee* 2010-09-12T21:34:22 dwchandler: will do, thanks 2010-09-12T21:34:26 how are you running this on tcp? is there any link with instructions? 2010-09-12T21:35:11 PabloM, are you running windows? 2010-09-12T21:35:15 nop 2010-09-12T21:35:19 ubuntu 2010-09-12T21:35:24 ok 2010-09-12T21:35:42 did you get tcp.c? 2010-09-12T21:35:55 mmm no 2010-09-12T21:36:02 PabloM, go here and get tcp.c and compile it using "gcc tcp.c" 2010-09-12T21:36:44 the ip and port are 213.3.30.106 9999 2010-09-12T21:37:06 *** kay has quit IRC (Ping timeout: 276 seconds) 2010-09-12T21:37:11 www.benzedrine.cx/planetwars/tcp.c 2010-09-12T21:37:12 got the tcp going with Cygwin on a Win64 using C# bots, just FYI since it is prob a uncommon setup. 2010-09-12T21:37:16 *** kay has joined #aichallenge 2010-09-12T21:37:34 Now if I can just stop loosing matches. :P 2010-09-12T21:38:32 nice 2010-09-12T21:39:15 lol bartwe.2 reminds me of a spider tank setup in Eve 2010-09-12T21:42:06 i see the method behind bartwe's madness 2010-09-12T21:42:07 gah! greghaynes, qt doesn't have support for vs2010 yet. I'm getting quite frustrated. maybe I'll use mingw 2010-09-12T21:42:22 use a real operating system ;) 2010-09-12T21:42:35 i jest i jest 2010-09-12T21:42:41 heh 2010-09-12T21:42:48 jmcarthur: tsk, tsk 2010-09-12T21:42:48 rogue780: They have an IDE that may work 2010-09-12T21:42:50 *** oh_cripes has quit IRC (Quit: Page closed) 2010-09-12T21:42:58 (but only because I wanted to day it) 2010-09-12T21:43:08 say it 2010-09-12T21:44:48 my last version should be named "pulverizer"..... 2010-09-12T21:44:54 Good night 2010-09-12T21:50:05 every two fight with bhasker i have 1 win, 1 lose :) 2010-09-12T21:50:09 *** caezar has joined #aichallenge 2010-09-12T21:50:14 hey guys 2010-09-12T21:50:20 anyone have c# uploads working? 2010-09-12T21:50:57 caezar: I haven't gotten my upload working (python) 2010-09-12T21:50:59 uploads have not worked for me all day 2010-09-12T21:51:05 C++ 2010-09-12T21:52:02 @later tell j3camero Are uploads working? Coders in (at least) 3 different languages have been reporting issues uploading throughout the day. 2010-09-12T21:52:02 bss03: The operation succeeded. 2010-09-12T21:53:16 *** panchi has quit IRC (Quit: Page closed) 2010-09-12T21:53:39 *** jerkface03 has joined #aichallenge 2010-09-12T21:53:54 ahhh it is so nice being able to run my log4net logging on the tcp server. 2010-09-12T21:55:15 gah, bartwe's bot is rediculous 2010-09-12T21:55:57 *** jmcarthur has quit IRC (Quit: WeeChat 0.3.0) 2010-09-12T21:56:03 ridiculous like a fox 2010-09-12T21:56:11 or something 2010-09-12T21:59:17 *** jmcarthur has joined #aichallenge 2010-09-12T22:00:33 *** xkpe has quit IRC (Ping timeout: 276 seconds) 2010-09-12T22:01:34 danielvf: woot, cpu usage is back to none, the kernel is not hogging anymore 2010-09-12T22:02:31 ty bss 2010-09-12T22:03:32 awesome 2010-09-12T22:03:41 Does this mean we may get to play games again? 2010-09-12T22:04:14 where are the docs for tcpserver and/or getting logs from other bots? 2010-09-12T22:04:59 caezar: www.benzedrine.cx/planetwars/ 2010-09-12T22:05:06 @tcp=server 2010-09-12T22:05:07 bss03: I do not know about 'tcp=server', but I do know about these similar topics: 'tcp-server', 'tcpserver' 2010-09-12T22:05:09 @tcp-server 2010-09-12T22:05:10 bss03: tcp-server = http://www.benzedrine.cx/planetwars/ 2010-09-12T22:05:44 Oh 2010-09-12T22:05:49 *mistype 2010-09-12T22:08:18 *** Doctor has quit IRC (Quit: Page closed) 2010-09-12T22:09:45 found the bot's wiki, tyvm 2010-09-12T22:14:19 *** JCS^ has quit IRC (Quit: \x00) 2010-09-12T22:14:26 anyone help a Win nerd with a script for cgywin to keep running my bot on the tcp server? 2010-09-12T22:16:58 amstan: is there a firewall besides iptables? 2010-09-12T22:19:17 *** albert has quit IRC (Quit: Page closed) 2010-09-12T22:22:10 shawnshekari: use java-client for tcp server - it's easier 2010-09-12T22:22:48 Use Qt client! 2010-09-12T22:24:24 I'll go take a look at it, thanks 2010-09-12T22:25:43 lol 2010-09-12T22:26:02 so, I'm having a hard time debugging in Java with this whole thing. 2010-09-12T22:27:18 *** smhanov has quit IRC (Read error: Connection reset by peer) 2010-09-12T22:28:43 *** computerwiz_222 has joined #aichallenge 2010-09-12T22:31:01 *** PabloM has quit IRC (Quit: bye) 2010-09-12T22:31:16 epic battle until turn 300 with dhartmei :P http://www.benzedrine.cx/planetwars/canvas?game_id=1284344081|dhartmei|pacotabaco 2010-09-12T22:31:49 i lose for -8 ships :'( 2010-09-12T22:32:03 * shawnshekari so tired of paco kicking my ass. lol 2010-09-12T22:33:08 *** kersny has quit IRC (Quit: I'm Out) 2010-09-12T22:34:42 That was an interesting match pacotabaco 2010-09-12T22:35:07 *** jmreardon has quit IRC (Ping timeout: 245 seconds) 2010-09-12T22:35:58 shawnshekari: seems we have similar bots (dhartmei & me) 2010-09-12T22:37:13 it looks like heaven vs hell 2010-09-12T22:38:15 I can't figure out why my bot is timing out 2010-09-12T22:38:26 *** sfsgagi has quit IRC () 2010-09-12T22:38:52 *** ermau has quit IRC (Quit: ermau) 2010-09-12T22:39:56 perhaps java is to finiky for this and I really should just go python. 2010-09-12T22:40:19 *** smhanov has joined #aichallenge 2010-09-12T22:40:35 shawnshekari, 2010-09-12T22:40:45 go to errgaming.com and download tcp.zip 2010-09-12T22:40:47 I'm trying to uphold Mr. Gates honor and so am using c# 2010-09-12T22:41:16 it's got a compiled tcp.c for windows and all the cygwin dlls in it as well as a few batch files to help you out 2010-09-12T22:41:31 nice thanks 2010-09-12T22:41:55 I got it all working with the exception of a script to keep running the bot after a match ends 2010-09-12T22:42:05 *** Appleman1234 has quit IRC (Ping timeout: 255 seconds) 2010-09-12T22:42:35 *** Appleman1234 has joined #aichallenge 2010-09-12T22:42:37 greghaynes, I'm trying one more thing to get that to compile under windows 2010-09-12T22:42:51 it better be an epic app for all the work I'm putting into it 2010-09-12T22:43:42 haha 2010-09-12T22:43:52 Its a learning experience ;) 2010-09-12T22:44:52 lol 2010-09-12T22:45:05 *** computerwiz_222 has left #aichallenge 2010-09-12T22:45:13 you could just use mingw and the qt stuff bundled with qtcreator lol 2010-09-12T22:45:25 that's my newest attempt 2010-09-12T22:45:31 *** AlliedEnvy has joined #aichallenge 2010-09-12T22:45:33 I would think the Qt creator would 'just works' 2010-09-12T22:45:37 er, just work 2010-09-12T22:46:25 Muhahahah........ "You WIN against bartwe.3" 2010-09-12T22:46:45 Since theres a win version of it 2010-09-12T22:47:16 *** antimatroidUni has joined #aichallenge 2010-09-12T22:47:24 oh that's just one of his old bots huh? boo 2010-09-12T22:49:34 gah. how the heck do I add files to a project? 2010-09-12T22:49:47 nevermind 2010-09-12T22:50:04 If you open the CMakeLists.txt file in Qt Creator, magic may happen 2010-09-12T22:50:07 *** jmreardon has joined #aichallenge 2010-09-12T22:50:09 is the server on port 9999? 2010-09-12T22:51:18 can't tell if I'm off to a lucky start with my latest changesm 2010-09-12T22:52:19 is the main site even running games? 2010-09-12T22:52:31 CMake Error: Could not create named generator CodeBlocks - MinGW Makefiles 2010-09-12T22:52:32 i uploaded last night before going to sleep and still haven't had one 2010-09-12T22:52:56 the main server is not doing uploads today 2010-09-12T22:53:09 I'm talking about tcp server - only option right now 2010-09-12T22:53:16 *** computerwiz_222 has joined #aichallenge 2010-09-12T22:53:31 *** tebele has joined #aichallenge 2010-09-12T22:53:36 what happens with the submissions? 2010-09-12T22:53:46 rwa: what username are you using on tcp? 2010-09-12T22:53:53 vortex.N 2010-09-12T22:54:04 oh, i finally got one game against you last night 2010-09-12T22:54:14 i might actually start on my proper bot later 2010-09-12T22:54:28 what's happening with submissions? 2010-09-12T22:54:38 tebele: they are being ejected into deep space 2010-09-12T22:54:50 so.. ? 2010-09-12T22:55:53 antimatroidUni: we've had a fair number of issues today, and are getting them worked out 2010-09-12T22:56:23 dstuff_: all good, just curious 2010-09-12T22:57:11 woohoo, can finally beat dhartmei with my new mods 2010-09-12T22:57:12 greghaynes, ok it did a bunch of stuff, but no exe file (well, two a.exe files, but neither work( 2010-09-12T22:58:11 does tcp require a specific line ending? 2010-09-12T22:58:27 hrm 2010-09-12T22:58:32 caezar: huh? 2010-09-12T22:59:16 does it freak out if you send crlf? 2010-09-12T22:59:32 trying to figure out why my client doesn't read anything 2010-09-12T22:59:57 its just \n delimited, not \r 2010-09-12T23:00:49 rogue780 is that working 2010-09-12T23:01:27 have we convinced people to change the diqualification rule on time outs yet? 2010-09-12T23:01:40 Cyndre, on. I still can't get it to work 2010-09-12T23:04:11 damn 2010-09-12T23:04:20 *** xtyger has joined #aichallenge 2010-09-12T23:04:23 *** Mathnerd314 has joined #aichallenge 2010-09-12T23:05:49 antimatroidUni: you've convinced me. not that it matters... 2010-09-12T23:06:47 *** tebele has quit IRC (Quit: Page closed) 2010-09-12T23:07:48 antimatroidUni: what rule on time outs? 2010-09-12T23:08:06 dstufft_: that timeout should be a lost game, not a DQ for the tournament 2010-09-12T23:08:13 fuck, school tomorrow 2010-09-12T23:08:50 amstan: I know, and I have homework too o_O 2010-09-12T23:08:58 amstan: fuck! 2010-09-12T23:09:08 amstan: i stayed up to 5 last night, terrible ideda 2010-09-12T23:09:12 I need to make a demo presentation for a work that I haven't touched since I heard about this contest :/ 2010-09-12T23:09:28 ratatata, focus on that ... 2010-09-12T23:09:36 ... ai bot 2010-09-12T23:09:37 antimatroidUni: same here: http://contestbot.hypertriangle.com/channellogs/%23aichallenge/%23aichallenge.2010-09-12.log 2010-09-12T23:09:39 ratatata: I'm worse; I won't elaborate 2010-09-12T23:10:10 * Appleman1234 is liking his dilligence, he has barely started on his ai bot, and has focused on his job applications and other projects 2010-09-12T23:10:48 Appleman1234: bullshit, not possible 2010-09-12T23:10:48 Appleman1234: yeah, but *you* don't have to win the contest 2010-09-12T23:10:49 Appleman1234: we'll see how tomorrow goes.. I doubt they have high expectations so I can slack on that front and focus here :) 2010-09-12T23:10:50 best of luck with all of your RL (real life) everyone :) 2010-09-12T23:11:21 I don't have to lol ...., but if actually submit something I will see how I go ..... 2010-09-12T23:11:22 working for yourself is A+ 2010-09-12T23:11:26 Appleman1234: oh man.. we really need such a thing for this contest 2010-09-12T23:11:28 *** Theobon has joined #aichallenge 2010-09-12T23:11:36 Appleman1234: a notice of somekind: remember the RL 2010-09-12T23:11:43 if you die in RL you die for real 2010-09-12T23:11:50 exactly 2010-09-12T23:11:50 I just spent most of today working on AI and i'll sleep through tommarow 2010-09-12T23:11:59 and no one will yell at me so /hell yea 2010-09-12T23:12:00 I will add some stuff on the wiki or something 2010-09-12T23:12:40 *** braffert has joined #aichallenge 2010-09-12T23:13:36 who needs to feed the kids when you're at #1 on the tcp server 2010-09-12T23:13:42 j/k, my kids are well fed 2010-09-12T23:14:14 *** tapwater has joined #aichallenge 2010-09-12T23:14:21 just tell your kids to bask in the glow of your ranking 2010-09-12T23:15:03 lol 2010-09-12T23:15:40 rwa, you have used your ai skills to build a rune goldberg machine to feed people ? 2010-09-12T23:15:40 jmcarthur: ping 2010-09-12T23:15:55 rwa: o rly? 2010-09-12T23:15:57 *** Theobon has quit IRC (Ping timeout: 252 seconds) 2010-09-12T23:16:18 *** antimatroidUni has quit IRC (Quit: Page closed) 2010-09-12T23:16:26 my 4 y/o only wants to watch the games I lose 2010-09-12T23:16:45 Appleman1234: no, but I did build a floor washing robot 2010-09-12T23:17:13 lol 2010-09-12T23:17:24 floor washing robot is cool too 2010-09-12T23:17:26 and I often daydream about how to design a toilet cleaning robot 2010-09-12T23:17:35 Mathnerd314: why are you pinging him>? 2010-09-12T23:17:53 * Appleman1234 goes to work on an ai perl irc chat bot for another irc channel 2010-09-12T23:18:21 *** Theobon has joined #aichallenge 2010-09-12T23:18:54 I actually don't have any ai skills 2010-09-12T23:19:01 never took an ai course in my life 2010-09-12T23:19:09 amstan: jmcarthur said he *might* have some interesting IO code for Haskell 2010-09-12T23:19:33 at least, I think he did :p 2010-09-12T23:19:41 Mathnerd314, what is wrong with monads for io ? 2010-09-12T23:19:57 so I assume my bot will get overtaken by people who know what they are doing eventually 2010-09-12T23:20:00 *** computerwiz_222 has left #aichallenge 2010-09-12T23:20:21 The IO in the unofficial starter kit should be fine for most purposes. 2010-09-12T23:20:51 Appleman1234: nothing, he just said that his way might be more efficient than jaspervdj's 2010-09-12T23:21:51 lol 2010-09-12T23:22:13 rwa, all my ai skills are self taught 2010-09-12T23:22:32 Mathnerd324: jaspervdj's isn't too bad. There's a few things I would change, but it works well. 2010-09-12T23:22:51 Appleman1234: cool. do you do software only or do you tinker with hardware? 2010-09-12T23:22:54 Mathnerd314: I wouldn't worry about it unless you are running real close to 1000ms already. 2010-09-12T23:22:58 I didn't take any ai courses in my undergraduate studies, but did help other students pass theirs 2010-09-12T23:23:18 bss03: haven't done any timing yet, so no idea how close I am 2010-09-12T23:23:32 dang.. misallocation fail.. http://www.benzedrine.cx/planetwars/canvas?game_id=1284347237|ratatata-0.1|ademar 2010-09-12T23:23:32 rwa, I do everything, but I haven't done much hardware yet, I have the ambition of making an working hard ai android 2010-09-12T23:24:46 __init__ (GAIC) / grep0 (GitHub) is using jaspervdj's IO. He's still in the Top 40, I think. 2010-09-12T23:24:46 rwa all my skills ever are self taught 2010-09-12T23:24:56 well 2010-09-12T23:25:03 anything after high school anyways 2010-09-12T23:25:08 dstufft_: nice 2010-09-12T23:25:32 not sure how much highschool had to do with it either, sleeping in class was my favorite school activity :/ 2010-09-12T23:25:37 I have too man years in school but mostly that was an excuse to not have a job 2010-09-12T23:25:40 too many 2010-09-12T23:26:05 I was a serious slacker in HS too 2010-09-12T23:26:38 * Mathnerd314 is in HS :p 2010-09-12T23:26:49 tbh i ended up getting a GED 2010-09-12T23:26:59 mostly because i failed to go to gym class 2010-09-12T23:27:03 lol 2010-09-12T23:27:04 and i didnt feel like making it up 2010-09-12T23:27:10 they were like 2010-09-12T23:27:17 well you can come backa 5th year and take gym 2010-09-12T23:27:27 im like, no thanks, *signs up for GED* 2010-09-12T23:28:54 yeah 2010-09-12T23:28:58 I can see that 2010-09-12T23:29:19 you work for yourself now? 2010-09-12T23:30:30 yea 2010-09-12T23:30:44 nice, software? 2010-09-12T23:30:56 yea, Mostly my customers need internal business shit 2010-09-12T23:31:01 web apps mostly 2010-09-12T23:31:07 + managing servers for them 2010-09-12T23:31:08 contestbot time 2010-09-12T23:31:08 Zaphus: 11:31 PM, September 12, 2010 2010-09-12T23:31:34 oddly enough i never made a site for myself 2010-09-12T23:31:43 *** PlanetWarrior has joined #aichallenge 2010-09-12T23:31:47 i keep meaning to do it, but then more projects get dumped in my lap 2010-09-12T23:31:51 from referals 2010-09-12T23:31:57 always the way 2010-09-12T23:32:05 like the plumber with leaky taps 2010-09-12T23:33:01 j3camero: hows that new scheduler coming along - still on the backburner with the other issues ? 8 hours since my last game, and it was a loss just to rub it in :-) 2010-09-12T23:33:24 Zaphus: how many games do you have? 2010-09-12T23:33:28 Sorry to cut in with a quick question - There was some talk about the conflict resolution when let's say the opposing fleets arrive on a neutral planet at the same time. Has that been changed to something intelligent or is it still the original? 2010-09-12T23:34:10 I don't know what the original was, or how intelligent the resolution is, but we do have an official method. 2010-09-12T23:34:18 amstam: 11 games, first 10 in the first 4-5 hours, one since then 2010-09-12T23:34:21 It was changed, the draft game specification explains the new method http://code.google.com/p/ai-contest/wiki/GameSpecification 2010-09-12T23:34:25 biggest force gets the planet 2010-09-12T23:34:34 reduced by second largest fleet size 2010-09-12T23:34:36 It's the method 3 a.k.a. all fleets fight at once. 2010-09-12T23:34:42 ^^ 2010-09-12T23:34:44 amstam: but, as expected with randomness, other bots have played many games in that time just by the luck of the dice 2010-09-12T23:35:01 Zaphus: if others have have around 11 games too, there's not much that can be done 2010-09-12T23:35:20 Ok so the two arriving fleets arrive in mid-air, duke it out and then the winner fights the fleet on the planet? 2010-09-12T23:35:39 nope :) 2010-09-12T23:36:13 basically they all fight simultaneously but read the spec it gives a better explanation 2010-09-12T23:36:37 and it boils down to biggest fleet wins with it's number reduced by second largest 2010-09-12T23:36:38 i think the smallest force basically get's ignored 2010-09-12T23:36:56 amstam: yes, it could prioritize people who havent played in a long time (after dealing with the people who have very small numbers of games) - e.g. someone who has played more games and had 5 games in that time period, should have shared some of those slots with others who had none :-) 2010-09-12T23:37:38 Biggest - 2nd Biggest is just an easy way to calculate the results. 2010-09-12T23:37:59 * Janzert agrees 2010-09-12T23:38:00 The algorithim is eliminate 1 ship from each fleet until 1 or less fleets remain. 2010-09-12T23:38:04 Zaphus: stop calling me amstam, lol, it doesn't highlight me then 2010-09-12T23:38:28 amstan: lol, sorry 2010-09-12T23:38:32 Ok I understand I was wrong in what I said and thanks for the GameSpecification link. It spells it out very well. I hadn't seen obvious links to it while browsing even though it seems to be a very important document. 2010-09-12T23:38:55 amstan: maybe I didnt really want you to see my reply, just talking to myself like those guys in Hyde Park in London 2010-09-12T23:38:56 If one fleet reamins it takes the planet, if 0 fleets reamin the planet stays under control of the current owner. 2010-09-12T23:39:22 yeah, it's just a draft that was written up. I don't know that it has been officially approved, that's why it's not linked prominently I assume 2010-09-12T23:39:28 Zaphus: lol, ok 2010-09-12T23:39:58 The spec is available on the main site: http://www.ai-contest.com/specification.php 2010-09-12T23:40:17 dunno if that means that it is officially approved 2010-09-12T23:40:21 ahh, and I see the sidebar also links to it now 2010-09-12T23:40:24 amstan: still, I want another game for my bot :-) are we still on the random/random scheduler, or now on the one that prioritizes people with low games ? 2010-09-12T23:40:43 I would assume yes that does make it official now :) 2010-09-12T23:40:46 :) 2010-09-12T23:41:17 *** DirtyKeyboard has quit IRC (Quit: DirtyKeyboard) 2010-09-12T23:44:13 *** braffert has quit IRC (Quit: Page closed) 2010-09-12T23:45:54 Another quick question - The fleets are sent, advance one step and arrival is then processed. Doesn't this mean that in the hypothetical case where two planets are at a distance of 1 away from each other then fleets sent between the two planets never actually show up in the "fleets" portion of the game state string. They just result in changes to the planet populations? 2010-09-12T23:46:55 @whatis tcpserver 2010-09-12T23:46:55 greghaynes: tcpserver = http://www.benzedrine.cx/planetwars/ 2010-09-12T23:49:23 Umm... has my client crashed...... 2010-09-12T23:49:48 Oh apparently not. Hopefully this signifies that you're all deep in thought instead then..... 2010-09-12T23:50:53 Umm..... hello? Am I alone in here all of a sudden? 2010-09-12T23:51:11 PlanetWarrior: no… 2010-09-12T23:51:14 PlanetWarrior, no you are not alone 2010-09-12T23:51:21 PlanetWarrior, you are one of us ..... 2010-09-12T23:51:34 contestbot will always talk to you if no one else will 2010-09-12T23:51:35 sigh: Error: "will" is not a valid command. 2010-09-12T23:51:40 Oh thanks! It just went awfully quiet.! 2010-09-12T23:51:48 see 2010-09-12T23:52:17 why didn't anyone else join me in the one of us, one of us, chant ? 2010-09-12T23:52:19 oh well 2010-09-12T23:54:05 *** other_donkey has joined #aichallenge 2010-09-12T23:57:10 A fellow pdxer! 2010-09-12T23:57:34 Shhh 2010-09-12T23:57:44 :) 2010-09-12T23:57:51 cbeck: Have you made a bot yet? 2010-09-12T23:58:19 *** xtyger has quit IRC (Quit: Page closed) 2010-09-12T23:58:23 Nope, working on fixing Go's %^%^@#@ html parsing lib 2010-09-12T23:58:31 ah 2010-09-12T23:58:36 MAKE A BOT ALREADY! 2010-09-12T23:58:41 *** AlliedEnvy has quit IRC (Ping timeout: 252 seconds) 2010-09-12T23:58:51 We need more portland stater's on the front page