2011-09-23T00:00:50 *** Keth has joined #aichallenge 2011-09-23T00:00:51 *** Keth_ has quit IRC (Read error: Connection reset by peer) 2011-09-23T00:00:57 *** amstan has joined #aichallenge 2011-09-23T00:00:57 *** ChanServ sets mode: +o amstan 2011-09-23T00:09:05 *** xathis has quit IRC () 2011-09-23T00:11:51 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 245 seconds) 2011-09-23T00:21:38 *** dvladim has quit IRC (Ping timeout: 260 seconds) 2011-09-23T00:22:26 *** McLeopold has joined #aichallenge 2011-09-23T00:52:19 *** McLeopold has left #aichallenge 2011-09-23T00:53:39 *** eashoka has joined #aichallenge 2011-09-23T01:10:19 *** okayzed is now known as okay| 2011-09-23T01:11:30 *** okay| is now known as okayzed 2011-09-23T01:35:56 *** McLeopold has joined #aichallenge 2011-09-23T01:47:55 *** Palmik has joined #aichallenge 2011-09-23T01:49:15 *** detryo has joined #aichallenge 2011-09-23T02:08:23 *** detryo has quit IRC (Ping timeout: 260 seconds) 2011-09-23T02:23:41 *** chris__0076 has joined #aichallenge 2011-09-23T02:27:55 *** chris___0076 has quit IRC (Ping timeout: 276 seconds) 2011-09-23T02:29:25 *** detryo has joined #aichallenge 2011-09-23T02:44:10 *** Palmik has quit IRC (Remote host closed the connection) 2011-09-23T02:46:52 *** onensora has joined #aichallenge 2011-09-23T03:02:07 *** detryo has quit IRC (Ping timeout: 248 seconds) 2011-09-23T03:25:01 *** ltriant has quit IRC (Quit: Get MacIrssi - http://www.sysctl.co.uk/projects/macirssi/) 2011-09-23T03:33:34 *** aerique has joined #aichallenge 2011-09-23T03:37:11 *** McLeopold has quit IRC (Quit: Leaving.) 2011-09-23T03:47:42 *** TTE has joined #aichallenge 2011-09-23T03:47:50 *** amstan has quit IRC (Read error: Operation timed out) 2011-09-23T03:48:15 *** TTE has quit IRC (Client Quit) 2011-09-23T04:08:35 *** mcstar has joined #aichallenge 2011-09-23T04:10:33 *** ibdknox has quit IRC (Remote host closed the connection) 2011-09-23T04:11:05 *** ibdknox has joined #aichallenge 2011-09-23T04:15:43 *** ibdknox has quit IRC (Ping timeout: 248 seconds) 2011-09-23T04:23:17 *** berak has joined #aichallenge 2011-09-23T04:35:16 *** sigh has joined #aichallenge 2011-09-23T04:47:59 *** sigh has quit IRC (Remote host closed the connection) 2011-09-23T04:56:43 @later tell McLeopold you only need to do x_m/(U^{1//alpha}) to generate draw from a pareto distribution. 2011-09-23T04:56:43 delt0r_: I come to serve. 2011-09-23T04:56:52 x_m is the mode 2011-09-23T04:57:11 or scale 2011-09-23T04:57:23 and alpha is the shape 2011-09-23T04:57:44 U is a uniform random variable 2011-09-23T04:58:51 you should probably catch NaN --even thou the probability is 1/(2^56) since real random numbers are not totally random 2011-09-23T05:04:44 i dont get it, if u were to be 0, the expression would yield inf 2011-09-23T05:04:52 how did you come up with 2^56? 2011-09-23T06:10:31 *** cichociemno has joined #aichallenge 2011-09-23T06:20:01 these spectral differentiation matrices are fing unbelieveable 2011-09-23T06:46:55 *** olexs has joined #aichallenge 2011-09-23T06:47:56 *** eashoka has quit IRC (Ping timeout: 252 seconds) 2011-09-23T06:50:19 *** FireFly has joined #aichallenge 2011-09-23T07:27:41 *** delt0r_ has quit IRC (Ping timeout: 245 seconds) 2011-09-23T07:40:45 *** delt0r_ has joined #aichallenge 2011-09-23T07:51:18 hmm, does compiler.py consider any output on stderr a failed compile? 2011-09-23T07:53:41 *** sigh has joined #aichallenge 2011-09-23T07:56:00 *** olexs1 has joined #aichallenge 2011-09-23T07:57:48 *** olexs has quit IRC (Ping timeout: 260 seconds) 2011-09-23T08:03:17 (setf *error-output* *standard-output*) 2011-09-23T08:03:32 that is needed for lisp to compile for the same reason 2011-09-23T08:09:56 *** xathis has joined #aichallenge 2011-09-23T09:02:05 *** superflit_ has joined #aichallenge 2011-09-23T09:03:05 *** superflit has quit IRC (Read error: Operation timed out) 2011-09-23T09:03:06 *** superflit_ is now known as superflit 2011-09-23T09:14:57 *** Effilry has joined #aichallenge 2011-09-23T09:20:56 *** FireFly has quit IRC (Quit: FireFly) 2011-09-23T09:21:11 *** Effilry has quit IRC (Changing host) 2011-09-23T09:21:12 *** Effilry has joined #aichallenge 2011-09-23T09:21:16 *** Effilry is now known as FireFly 2011-09-23T09:31:50 *** mceier has joined #aichallenge 2011-09-23T09:50:01 *** SimonT has joined #aichallenge 2011-09-23T09:51:54 *** nplus has quit IRC (Ping timeout: 260 seconds) 2011-09-23T09:52:00 *** SimonT is now known as nplus 2011-09-23T09:56:47 *** smkitct_ has joined #aichallenge 2011-09-23T09:57:51 *** Momo___ has joined #aichallenge 2011-09-23T09:59:47 hi, I would like to know what is the current turntime in ants 2011-09-23T10:01:54 Momo__: http://aichallengebeta.hypertriangle.com/game_settings.php 2011-09-23T10:02:37 thanks so it is 500 2011-09-23T10:03:22 *** smkitct_ has quit IRC (Ping timeout: 252 seconds) 2011-09-23T10:04:18 ms 2011-09-23T10:10:43 *** aerique has quit IRC (Quit: ...) 2011-09-23T10:11:01 *** eashoka has joined #aichallenge 2011-09-23T10:11:06 ah. so that's why my bot sometimes times out... 2011-09-23T10:12:23 it's tuned to 1sec initially. it'll try and keep the turn within the given time limit, but sometimes the very first timely operation is too long already before it can cut it off... hm 2011-09-23T10:30:58 *** McLeopold has joined #aichallenge 2011-09-23T10:31:53 *** McLeopold has left #aichallenge 2011-09-23T10:32:34 thanks bye 2011-09-23T10:32:45 *** Momo___ has quit IRC (Quit: Page closed) 2011-09-23T10:33:31 *** xathis has quit IRC () 2011-09-23T10:45:51 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-09-23T10:47:55 *** cichociemno has quit IRC (Quit: Page closed) 2011-09-23T10:49:04 olexs1: The time limit is sent a the start of the game 2011-09-23T10:49:18 mleise: I know, and I use that info 2011-09-23T10:49:41 thing is, my bot determines how long his "complex" part needs only after running it a few times, then it can tweak it to take less time 2011-09-23T10:49:58 if the first run is longer than the 500ms given, there's no mechanism to prevent a timeout 2011-09-23T10:50:37 Wow, the first turns you see nothing much, just your ant and some of the map. How can you possibly time out? :) 2011-09-23T10:51:01 it'll be the first battle simulation 2011-09-23T10:51:02 well... there should also be a food item 2011-09-23T10:51:39 I build a search tree, first time with a "maximum" depth. afterwards, I can limit the depth to shorten the search 2011-09-23T10:51:52 *** ChrisH has joined #aichallenge 2011-09-23T10:52:11 but on the first run, I measure how long the maximum depth search took, and use that number later :) 2011-09-23T10:52:57 You should do anything as artificial as that (not depending on what is actually on the map) in the loading phase - unless you use a VM language 2011-09-23T10:53:14 You get 3 seconds there to do your stuff 2011-09-23T10:53:33 I could try that, indeed 2011-09-23T10:54:26 or I could just check the time in the tree building algorithm and cut it off dynamically, instead of only shortening the depth at search initialization 2011-09-23T10:54:41 *** sigh has quit IRC (Remote host closed the connection) 2011-09-23T10:54:45 *** ChrisH has quit IRC (Client Quit) 2011-09-23T11:04:29 *** onensora has quit IRC () 2011-09-23T11:05:04 cant you just build the search tree once? 2011-09-23T11:05:15 for example at compile time? 2011-09-23T11:05:37 not really, since it's specific for every battle situation encountered 2011-09-23T11:06:06 but you said you generate it in the first turn, so you have no information at all 2011-09-23T11:06:20 no, in the first turn where I encounter a battle situation 2011-09-23T11:06:23 *** berak has quit IRC (Ping timeout: 248 seconds) 2011-09-23T11:06:32 i see 2011-09-23T11:06:50 it's classic minimax state-space search with alpha-beta cutoff 2011-09-23T11:07:09 how many ants do you consider? 2011-09-23T11:07:16 n vs. m 2011-09-23T11:08:07 as many as there are in a certain area. I do a fast recursive search around the first pair of my_ant-enemy_ant, looking for any more ants in a reasonable radius 2011-09-23T11:08:42 those are then treated as an isolated situation, I simulate possible moves and evaluate the results down to a certain depth 2011-09-23T11:09:24 so it seems its pretty costly 2011-09-23T11:09:36 do you consider many turns? 2011-09-23T11:09:50 initial search depth is currently 6 plies, which is 3 turns 2011-09-23T11:09:52 im new to this game-tree stuff 2011-09-23T11:09:59 i dont really know the practical limits 2011-09-23T11:10:40 yes, it is pretty costly, and I'm looking for places to optimize it. alpha-beta cutoff was a good one 2011-09-23T11:10:47 so you pick the branch that minimizes your maximum loss? 2011-09-23T11:11:08 yes 2011-09-23T11:11:53 evaluation function is based around "enemy_ants_killed - my_ants_lost", with a few heuristic additions to that 2011-09-23T11:12:14 but this kind of logic can only lead to good local decisions 2011-09-23T11:12:20 do you have global strategies? 2011-09-23T11:12:40 I do, but it's not quite finished yet 2011-09-23T11:12:41 i mean for example ants far apart collaborating 2011-09-23T11:12:46 ok 2011-09-23T11:13:30 I have a sort of a grid covering the area that is "secure" to see and fetch spawned food asap, and ants that would otherwise idle go join ongoing battles or explore unseen terrain 2011-09-23T11:14:11 initial exploration needs some work, I've written some stuff down and will code it in the next few days 2011-09-23T11:14:40 for the inflation phase i use a time-delyed influence map 2011-09-23T11:14:46 delayed* 2011-09-23T11:17:22 olexs1: you've been working on your bot a whole week through? 2011-09-23T11:17:36 i hope you attended classes though :) 2011-09-23T11:18:09 don't have any classes right now, and the bot is actually related to uni work, so yes, I've put some time into it 2011-09-23T11:19:04 is it a term-assignment? or what?(if you dont mind my asking) 2011-09-23T11:19:17 it's part of my bachelor thesis 2011-09-23T11:19:48 what is the required final ranking for a succesful thesis? 2011-09-23T11:19:52 I'm doing some AI stuff, and this should be the practical application to demonstrate :) 2011-09-23T11:20:27 i see, so you must have a secret weapon 2011-09-23T11:20:32 no requirements, my prof is very liberal with the whole thing. But a high ranking should help with good marks :) 2011-09-23T11:21:07 good luck then 2011-09-23T11:21:13 thanks. 2011-09-23T11:21:38 the planetwars challenge last year nearly cost me my maths-3 exam, this time I'm putting the time invested to good use 2011-09-23T11:22:33 a tcp server like we had back then would be really nice, though 2011-09-23T11:22:57 deltor is working on it 2011-09-23T11:23:01 sort of 2011-09-23T11:23:04 at least will be 2011-09-23T11:23:23 evaluating small changes or tweaks is not easy now, battling my own old versions, examples and a few other bot binaries can only give so much information 2011-09-23T11:23:30 <_flag> Wasn't there an almost ready tcp server made by berak that just had a few bugs to work out? 2011-09-23T11:23:54 _flag: fetch his repo, and try it how it works 2011-09-23T11:24:02 he worked on it agai 2011-09-23T11:24:04 n 2011-09-23T11:24:30 <_flag> ah okay, I'll try it out 2011-09-23T11:24:40 https://github.com/flagcapper/Berak-s-TCP-Server ? 2011-09-23T11:25:18 or more like https://github.com/berak/ants-tcp, last update 3 days ago 2011-09-23T11:25:35 <_flag> Yeah, I mirrored it awhile ago because he took his down for awhile 2011-09-23T11:25:38 https://github.com/berak/ants-tcp 2011-09-23T11:25:46 <_flag> But he's working on it again so his is more recent 2011-09-23T11:42:49 olexs1: my marks definitely suffered from planet wars :P 2011-09-23T11:42:49 He says, that 1000 games take up 60 MB disk space, but he could gzip them up, like on the beta server. 1000 replays there use up 16 MB 2011-09-23T11:43:48 i was on facebook the other day and it gave a post from a year ago "new google ai challenge is up, goodbye productivity for anything else" oh how right i was 2011-09-23T11:45:02 *** nux67 has joined #aichallenge 2011-09-23T11:47:17 *** Keth has quit IRC (Read error: Connection reset by peer) 2011-09-23T11:47:21 *** Keth has joined #aichallenge 2011-09-23T11:48:23 jskills - JavaScript kills ? 2011-09-23T11:48:47 java skills, more likely :) 2011-09-23T11:50:35 and what did he do with the visualizer, lol. he added back obsolete files 2011-09-23T11:51:29 mleise: he who? 2011-09-23T11:51:40 the boogeyman? 2011-09-23T11:51:50 berak, on his copy for the tcp server 2011-09-23T11:51:58 not the official repo :) 2011-09-23T11:53:13 <_flag> mleise: I was about to ask, the visualizer doesn't seem to be working and I'm not sure if it's my problem or his 2011-09-23T11:54:35 As it is a cracked version, I cannot give support for it. Really, I have no idea what would be wrong there. You better ask berak. 2011-09-23T11:54:44 Or is there a distinct error message? 2011-09-23T11:54:56 <_flag> http://flagcapper.com:2080/replay.1 2011-09-23T11:55:14 <_flag> "CanvasElementMap is not defined" 2011-09-23T11:55:27 *** Accoun has quit IRC (Ping timeout: 248 seconds) 2011-09-23T11:55:38 http://www.0.2*fftinu/ 2011-09-23T11:55:45 why was this in my paste buffer? 2011-09-23T11:57:24 _flag: that loads as plain text for me 2011-09-23T11:58:02 <_flag> mleise: As in, it shows the actual replay file? 2011-09-23T11:58:08 ReferenceError 2011-09-23T11:58:13 thats for me in red 2011-09-23T11:58:37 _flag: it shows the html source in Opera. I bet the http headers are missing 2011-09-23T11:58:59 it shows a page in firefox 2011-09-23T11:59:08 with the aforementioned error 2011-09-23T11:59:52 *** nplus has quit IRC (Quit: ChatZilla 0.9.87-rdmsoft [XULRunner 1.9.2/20100222071121]) 2011-09-23T12:00:13 The script file CanvasElement.js is missing 2011-09-23T12:00:23 but oddly, when i click show source, it looks empty 2011-09-23T12:00:43 *** Accoun has joined #aichallenge 2011-09-23T12:02:08 _flag: Does the tcp-bot come with an integrated web server? 2011-09-23T12:02:15 sorry, tcp-server 2011-09-23T12:03:09 <_flag> mleise: yes, it does 2011-09-23T12:04:08 that could explain missing headers 2011-09-23T12:04:29 <_flag> Probably 2011-09-23T12:04:35 did it lock up? http://flagcapper.com:2080/replay.1 2011-09-23T12:05:22 <_flag> Well my computer isn't overloaded, but maybe it's his code 2011-09-23T12:05:40 <_flag> I'll restart it 2011-09-23T12:06:04 thread dead-locks are also explained by homebrew web servers ^^ 2011-09-23T12:06:33 <_flag> I suppose all the errors could be explained by homebrew web servers ^^ 2011-09-23T12:07:47 _flag: are you trying to get the server running on your pc? if you need clients to connect for testing, I'd be glad to help :) 2011-09-23T12:08:45 <_flag> oles1: You can connect now, but you won't be able to watch your games 2011-09-23T12:09:01 <_flag> olexs1: ^ 2011-09-23T12:09:37 ok, I'll give it a go 2011-09-23T12:09:45 you should really give berak some time to fix the server. right now it doesn't look like anything that could be released 2011-09-23T12:13:04 well, it seems to at least connect and play, which is something 2011-09-23T12:13:38 <_flag> I think it's mostly just the webserver that has problems 2011-09-23T12:13:56 <_flag> But I have no problem waiting for berak 2011-09-23T12:13:57 the game is taking a while, but that's to be expected I guess 2011-09-23T12:14:19 <_flag> The turn limit is 750 2011-09-23T12:14:38 yes, I see. 2011-09-23T12:14:54 <_flag> Although I remember the visualizer working in previous versions, which is somewhat odd 2011-09-23T12:15:21 <_flag> Especially considering the commit history has comments like "fixed visualizer" 2011-09-23T12:15:22 ok, I don't quite understand the result 2011-09-23T12:16:07 it finished before game end, my bot survived (which is odd, considering yours is the #1 currently if I'm right), but my score is lower than yours. does it mean some sort of cutoff? 2011-09-23T12:16:31 ^before game end = before turn 750 2011-09-23T12:17:13 <_flag> I'm not sure why it says it survived, but I seem to have gotten a higher score than you and we didn't reach 750 2011-09-23T12:17:18 <_flag> Mine says it survived too 2011-09-23T12:18:55 survived will be right though, the data I got on the last turn has a lot of ants with player ids 0 and 1 2011-09-23T12:19:11 so it must be some sort of "no progress cutoff" 2011-09-23T12:20:26 trueskill doesn't seem to work, either, no skill is recorded in "Rankings" 2011-09-23T12:21:10 <_flag> Yeah, it seems that the server is incomplete 2011-09-23T12:21:23 <_flag> I'll probably take it down after this game 2011-09-23T12:21:31 ok 2011-09-23T12:21:57 *** nplus has joined #aichallenge 2011-09-23T12:22:11 <_flag> Okay, I added the missing js file and replays work 2011-09-23T12:22:23 ah, nice 2011-09-23T12:22:27 <_flag> So you can watch our games 2011-09-23T12:22:32 <_flag> So i'll probably leave it up for a bit 2011-09-23T12:25:19 <_flag> Yeah, it looks like a no progress cutoff is killing our games 2011-09-23T12:25:25 yep 2011-09-23T12:25:54 I'm gonna AFK for a while, should I leave the bot client on? 2011-09-23T12:26:17 <_flag> It's up to you -- If you want to play some games, sure 2011-09-23T12:26:29 <_flag> Maybe someone else will show up 2011-09-23T12:27:21 I wouldn't advertise it too much though, until the server is figured out properly 2011-09-23T12:27:56 <_flag> Yeah, that's probably a good idea 2011-09-23T12:28:15 <_flag> I'm a little wary of running it on my home computer anyway 2011-09-23T12:32:13 *** gattschardo has joined #aichallenge 2011-09-23T12:32:32 hi 2011-09-23T12:32:50 <_flag> hello 2011-09-23T12:33:23 any of the "officials" here? 2011-09-23T12:33:37 <_flag> Not at the moment 2011-09-23T12:33:38 i've been working on a starter package for erlang these days 2011-09-23T12:33:41 okay 2011-09-23T12:34:01 i would complete it, if they say it's okay to add erlang as a language 2011-09-23T12:35:15 should i ask about that here or rather open an issue on github? any ideas? 2011-09-23T12:35:21 <_flag> I doubt they'll mind, most of the starter packages written so far were from people outside the main developers 2011-09-23T12:35:47 http://antsbeta.janzert.com/visualizer.php?game=205262 interesting, FlagCapper has an end game kamikaze mode 2011-09-23T12:35:51 :} 2011-09-23T12:36:10 <_flag> janzert: :} 2011-09-23T12:36:25 <_flag> Actually gattschardo, you can ask janzert 2011-09-23T12:36:49 is that new or have I just not watched enough of your games? 2011-09-23T12:36:56 *** ibdknox has joined #aichallenge 2011-09-23T12:36:59 <_flag> It's part of the newest version 2011-09-23T12:39:06 gattschardo: if the package at http://packages.ubuntu.com/natty/erlang-base is new enough then the only sticking point would be adding it to the automatic compiler script at worker/compiler.py 2011-09-23T12:39:41 okay, i'll have a look at what ubuntu ships 2011-09-23T12:39:56 or if there's a better package in ubuntu natty to use 2011-09-23T12:40:14 *** onensora has joined #aichallenge 2011-09-23T12:40:27 <_flag> Actually, GreenTea seems to have a kamikaze mode too, his is just weaker: http://antsbeta.janzert.com/visualizer.php?game=204074&user=7 2011-09-23T12:40:43 <_flag> Notice the dips in both our scores, mine is more towards the end 2011-09-23T12:40:45 <_flag> And sharper 2011-09-23T12:41:01 yep 2011-09-23T12:41:08 janzert: erlang-base looks fine 2011-09-23T12:41:57 the bots run from the same directory they are compiled in? 2011-09-23T12:42:06 basically yes 2011-09-23T12:43:24 the other part is https://github.com/aichallenge/aichallenge/blob/epsilon/worker/compiler.py needs erlang added in the array of languages starting at line 264 2011-09-23T12:44:10 k 2011-09-23T12:45:40 one other thought is I know erlang is known for it's parallel programming, contest submissions are supposed to restrict themselves to using a single main thread of processing, is there an option for erlang to not use multiple OS threads or processes? 2011-09-23T12:45:42 *** berak has joined #aichallenge 2011-09-23T12:46:52 afaik that depends on how the erlang vm is compiled 2011-09-23T12:47:33 so there is basically no way to stop the user from using user-level-threads, but depending on how the vm is compiled they will all run in the same OS thread 2011-09-23T12:47:49 <_flag> berak: I'm running your tcp server here: http://flagcapper.com:2080/, it seems to work fairly well so far although the visualizer was missing a file 2011-09-23T12:48:00 ;) 2011-09-23T12:48:15 is this message talking about the same VM as Ubuntu uses? http://erlang.org/pipermail/erlang-questions/2010-September/053328.html 2011-09-23T12:48:26 if so it looks like a +S option might do it 2011-09-23T12:48:31 <_flag> Also I think the webserver part seems to hang sometimes 2011-09-23T12:49:02 anyway that can be played with once the basics are implemented 2011-09-23T12:49:43 the manpage says "-smp disable" does it 2011-09-23T12:49:52 maybe +S is a short form for that 2011-09-23T12:50:04 http://www.erlang.org/doc/man/erl.html 2011-09-23T12:50:17 nice to know it doesn't need to be recompiled 2011-09-23T12:50:39 +S sets the amount of scheduler threads, which seem to be erlang speak for OS level threads :} 2011-09-23T12:51:06 yep, you're right 2011-09-23T12:51:15 <_flag> berak: Also the rankings don't work (disregard this if you already know it) 2011-09-23T12:51:35 also +hms might be useful to limit the heap size 2011-09-23T12:51:45 yes, probably 2011-09-23T12:54:51 _flag, is that on linux ? since you're using jskills there, have to change 'cp_separator' in main.py to ':' 2011-09-23T12:56:04 *** Qua has joined #aichallenge 2011-09-23T12:56:21 <_flag> berak: Well that's what I get for not actually reading the file :) 2011-09-23T12:56:29 ;) 2011-09-23T12:56:50 <_flag> If I restart the server will connected players reconnect automatically? 2011-09-23T12:57:52 <_flag> I don't want to lose the one player I have but I'd also like to see rankings 2011-09-23T12:58:16 Game specification is gone!? 2011-09-23T12:58:17 guess not., but it's just 2 players now. 2011-09-23T12:58:19 I can reconnect manually :) 2011-09-23T12:58:28 wait, trying remote admin.. 2011-09-23T12:58:34 *** amstan has joined #aichallenge 2011-09-23T12:58:34 *** ChanServ sets mode: +o amstan 2011-09-23T12:58:35 <_flag> olexs1: okay, restarting 2011-09-23T12:58:41 Qua: it got broken in the recent server move 2011-09-23T12:58:47 it'll be back eventually ;) 2011-09-23T12:59:18 any place to find it? 2011-09-23T12:59:19 *** okayzed is now known as okay| 2011-09-23T12:59:22 looking for attack rules 2011-09-23T12:59:57 https://github.com/aichallenge/aichallenge/wiki/Ants-Game-Specification 2011-09-23T12:59:59 <_flag> olexs1: Okay, try reconnecting 2011-09-23T13:00:16 _flag: done 2011-09-23T13:00:34 *** Palmik has joined #aichallenge 2011-09-23T13:01:19 hmm, there was a better page than that about the battle resolution I though 2011-09-23T13:01:51 https://github.com/aichallenge/aichallenge/wiki/Ants-focus-battle-resolution-method 2011-09-23T13:01:56 there it is 2011-09-23T13:07:06 _flag, thanks for trying that ! ranking works now ;) 2011-09-23T13:07:19 <_flag> berak: no problem 2011-09-23T13:08:14 <_flag> berak: Is the server running 2 player maps intentionally because of low user count? 2011-09-23T13:08:29 *** amstan has quit IRC (Ping timeout: 256 seconds) 2011-09-23T13:09:01 yes, not more than 1/2 of all players online. 2011-09-23T13:09:51 there's a cheesy /online page, too 2011-09-23T13:11:43 how much cpu/ram does the server take? I could put it up on a proper-ish vps, if it's not too much 2011-09-23T13:12:28 *** UncleVasya has joined #aichallenge 2011-09-23T13:13:24 it's not much. server is waiting most of the time 2011-09-23T13:13:34 <_flag> It's currently using about 2% of my CPU and 40MB of memory, but it did spike to 12% CPU for a moment (maybe it was doing rankings?) 2011-09-23T13:13:38 janzert: i think this should work for the compiler https://gist.github.com/1237908 2011-09-23T13:13:56 probably. starting java for that 2011-09-23T13:15:33 <_flag> The webserver seems to have hanged 2011-09-23T13:17:01 behh ;( 2011-09-23T13:17:32 <_flag> :( 2011-09-23T13:18:02 <_flag> The games are still playing though, so maybe it's just a particular thread? 2011-09-23T13:18:28 yes, same here, and yes, a different thread 2011-09-23T13:18:44 why is a standalone webserver used anyway? 2011-09-23T13:19:33 olexs1, ? 2011-09-23T13:19:41 running a simple web app written in python through apache/lighttpd/nginx&co should be more reliable 2011-09-23T13:21:09 oh, shure, nginx would be much better, than pythons SimpleHttpServer. but it would need a dedicated database instead of just sharing memory then. 2011-09-23T13:21:42 _flag, something in the logs ? 2011-09-23T13:23:46 <_flag> berak: Where are the logs? Or do you mean the output to the terminal? 2011-09-23T13:24:32 yes, terminal output 2011-09-23T13:25:44 does it hang up when no games are played? 2011-09-23T13:25:56 no, why ? 2011-09-23T13:26:37 maybe there is a database transaction deadlock when games are stored and listed at the same time or something 2011-09-23T13:27:06 ah, threading/locking, thats more likely 2011-09-23T13:27:42 actually most likely 2011-09-23T13:28:31 <_flag> This happened awhile ago, not sure if it conincided with the hang or not: http://pastebin.com/KJVje7UY 2011-09-23T13:30:35 <_flag> berak: ^ 2011-09-23T13:30:37 that's just a broken socket, client hung up early 2011-09-23T13:30:49 <_flag> Okay then, it looks normal otherwise 2011-09-23T13:36:24 *** dr0b3rts has joined #aichallenge 2011-09-23T13:43:34 but thanks for the stacktrace, line 1 is really puzzling me.. 2011-09-23T13:49:56 someone tried to download /home/flag/Code/git/Ants-TCPServer/ants-tcp/../../../../../../bin/cp ;( 2011-09-23T13:54:12 *** okay| is now known as okayzed 2011-09-23T13:55:58 <_flag> berak: Yeah... hopefully that wasn't possible 2011-09-23T13:57:44 *** okayzed is now known as okay| 2011-09-23T13:59:58 hmm, can't go up from / in my local ubuntu 2011-09-23T14:00:27 gattschardo: thanks, I'm going to change my_bot to MyBot just so it matches the rest of the languages, unless that causes a problem with Erlang somehow 2011-09-23T14:01:12 <_flag> berak: I can get /../main.py 2011-09-23T14:01:55 that's definitely a security flaw, the rcon password and other settings should not be downloadable 2011-09-23T14:02:14 *** okay| is now known as okayzed 2011-09-23T14:02:30 not to speak of files outside the tcp server directory, those can be anything 2011-09-23T14:02:36 yes, you're right 2011-09-23T14:10:56 *** UncleVasya has quit IRC (Ping timeout: 244 seconds) 2011-09-23T14:13:45 *** Qua has quit IRC (Quit: Page closed) 2011-09-23T14:14:22 <_flag> berak: Does your server automatically cache all file requests? 2011-09-23T14:16:44 the static ones, yes 2011-09-23T14:16:50 ont the replays 2011-09-23T14:17:20 *** Kingpin13 has joined #aichallenge 2011-09-23T14:17:26 <_flag> Okay, I'm trying to think if I would have seen it if someone had tried to access something else 2011-09-23T14:18:21 like remote access passwords, ssh keys etc? 2011-09-23T14:18:24 gattschardo: why are there or what do the two -s arguments to erl do? reading the option description I would have expected just the first one 2011-09-23T14:18:25 janzert: it's against the usual naming convention for modules, but it'll work 2011-09-23T14:18:40 ok 2011-09-23T14:18:42 first one executes the start function in my_bot (which needs to be changed) 2011-09-23T14:18:48 init stop stops the vm 2011-09-23T14:18:53 <_flag> mleise: Things like that, yes 2011-09-23T14:18:54 after the bot exits 2011-09-23T14:19:10 ahh, so basically they get run in sequence 2011-09-23T14:19:13 yep 2011-09-23T14:20:39 *** boegel has joined #aichallenge 2011-09-23T14:22:24 For the attribution on the commit gattschardo good or would you prefer a different name? 2011-09-23T14:22:48 also is the starter bot available anywhere publically yet? 2011-09-23T14:22:50 gattschard is fine 2011-09-23T14:23:15 i'm still working on some of the suggested helper functions for the starter bot 2011-09-23T14:23:20 ok 2011-09-23T14:23:28 direction and visible aren't done yet 2011-09-23T14:24:14 and i'm not sure about remaining time 2011-09-23T14:24:44 *** dvladim has joined #aichallenge 2011-09-23T14:24:54 i'll let you know when i'm done, probably later tonight or some time tomorrow 2011-09-23T14:25:08 aichallenge: janzert epsilon * r5251c00 / (setup/worker_setup.py worker/compiler.py): Initial Erlang support thanks to gattschardo - http://git.io/8EYmdw 2011-09-23T14:25:08 aichallenge: janzert epsilon * r4dcca12 / (7 files in 5 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/19Cg4g 2011-09-23T14:25:23 ok, I'll roll out the support to the workers so you can test if everything is working 2011-09-23T14:27:18 thanks :) 2011-09-23T14:27:26 <_flag> berak: It seems I can't go further back than the server directory, so /../../../../../../../main.py is the same as /../main.py 2011-09-23T14:28:46 contestbot later tell McLeopold there were some local changes to sql/opponent.sql sitting on the server, I moved the file to /home/contest/opponent_local.sql 2011-09-23T14:28:46 janzert: Yes master! 2011-09-23T14:29:05 _flag, that's what i thought 2011-09-23T14:29:26 <_flag> berak: Any idea how /bin/cp may have been accessed? 2011-09-23T14:31:28 *** eashoka has quit IRC (Quit: Page closed) 2011-09-23T14:32:30 not yet, that one fell through. if it showed up in that 'added static' line, it was accessed ;( 2011-09-23T14:34:12 _flag, it tried to read it(fopen), not execute 2011-09-23T14:35:17 <_flag> Yeah, I assumed that 2011-09-23T14:50:13 *** dvladim has quit IRC (Ping timeout: 260 seconds) 2011-09-23T15:02:35 *** Eruonen has joined #aichallenge 2011-09-23T15:04:13 *** onensora has quit IRC (Ping timeout: 260 seconds) 2011-09-23T15:11:53 <_flag> berak: I've reproduced the error 2011-09-23T15:12:18 oh, i couldn't 2011-09-23T15:12:25 <_flag> berak: If you try to access the file with your browser or wget or some other tool it will remove the /../../../, however if you manufature your own http request 2011-09-23T15:12:32 <_flag> You can access whatever you want 2011-09-23T15:12:52 <_flag> Luckily they only stole my copy 2011-09-23T15:14:04 i don't quite get it, 'manufacture your own http request' here, all those /../ just get eaten 2011-09-23T15:14:18 <_flag> Write a program to send your own http request 2011-09-23T15:14:26 use telnet ;) 2011-09-23T15:14:30 ahh ;) 2011-09-23T15:14:32 <_flag> Or that % 2011-09-23T15:14:36 <_flag> that ^ 2011-09-23T15:14:37 <_flag> * 2011-09-23T15:15:10 the only use for telnet anymore, manually playing with text based protocols ;P 2011-09-23T15:15:20 <_flag> true 2011-09-23T15:15:28 haha, shure.. 2011-09-23T15:17:20 <_flag> I think the reason I didn't see anymore requests is because the potential attacker doesn't know the directory structure, and you can assume that after a certain number of ../../ you will end up at / 2011-09-23T15:18:05 *** Qua has joined #aichallenge 2011-09-23T15:18:24 <_flag> Also, failed requests aren't logged 2011-09-23T15:24:07 ok, http logging is back in, added some code to catch that '..' stuff, but still got no idea, why it stalled 2011-09-23T15:24:39 You working on a TCP server? 2011-09-23T15:24:52 telnet can be used to configure routers 2011-09-23T15:25:29 *** delt0r_ has quit IRC (Read error: Operation timed out) 2011-09-23T15:25:38 Qua, yeah, but it didn't fly far in the first run ;( 2011-09-23T15:26:25 What language are you writing in it? Is the source available somewhere? 2011-09-23T15:27:26 python, https://github.com/berak/ants-tcp 2011-09-23T15:38:41 [f(x) for x in range(n)] 2011-09-23T15:38:52 x has global value afterwards 2011-09-23T15:38:57 incredible, why?? 2011-09-23T15:39:06 this isl ame 2011-09-23T15:40:49 *** delt0r_ has joined #aichallenge 2011-09-23T15:42:20 janzert: actually, MyBot is impossible as module name. myBot will work, but the first letter can't be uppercase 2011-09-23T15:43:14 other than that the starter package is basically done 2011-09-23T15:43:30 i didn't do all the optional functions, though 2011-09-23T15:45:05 *** McLeopold has joined #aichallenge 2011-09-23T15:45:49 janzert: I've cloned to repo on the test server to the /home/mcleopold dir, so I won't be doing that with local changes anymore, sry 2011-09-23T15:48:06 gattschardo: ok, if we're going to have to change it anyway we might as well follow the language convention. I'll change it to my_bot 2011-09-23T15:48:15 McLeopold: np 2011-09-23T15:48:59 ok 2011-09-23T15:49:25 *** asdf_ has joined #aichallenge 2011-09-23T15:51:17 *** okayzed is now known as okay| 2011-09-23T15:56:48 McLeopold: since youre the python guy, could you answer me why the list comprehension leaks variables? 2011-09-23T15:56:52 *** Palmik has quit IRC (Remote host closed the connection) 2011-09-23T15:57:09 *** amstan has joined #aichallenge 2011-09-23T15:57:09 *** ChanServ sets mode: +o amstan 2011-09-23T16:01:21 they don't in python3 anymore ;) it's mostly an artifact of the way they were initially implemented then they couldn't get rid of it in python2 because of backwards compatibility concerns 2011-09-23T16:01:49 i see, i was about to abandon python completely 2011-09-23T16:01:56 heh 2011-09-23T16:03:13 *** Accoun has quit IRC () 2011-09-23T16:03:39 does numpy/ipython/matplotlib exists/stable for python3? 2011-09-23T16:03:45 -s 2011-09-23T16:03:49 aichallenge: janzert epsilon * r1c27ea2 / worker/compiler.py : Use Erlang convention for bot name - http://git.io/Sewn0Q 2011-09-23T16:04:26 mcstar: no idea really since I don't use any of those (modules or py3) :} 2011-09-23T16:04:54 i guess i will have to look it up myself :) 2011-09-23T16:05:10 but i dont expect sage anytime soon switching to python 3 2011-09-23T16:10:59 janzert: compiling still does not work: "Output file /home/contest/compiled/tmpQbeEaN/bot/MyBot.beam was not created. 2011-09-23T16:11:22 i don't see why it looks for MyBot.beam, though? 2011-09-23T16:11:37 is MyBot hardcoded im some place? 2011-09-23T16:11:41 yes 2011-09-23T16:11:45 I just saw that go by 2011-09-23T16:12:30 the compiler output is expected to produce a Mybot.extension file 2011-09-23T16:12:42 where .extension is given as the second Language parameter 2011-09-23T16:12:52 err, MyBot.extension 2011-09-23T16:12:58 hmm 2011-09-23T16:13:42 is that because of the targetCompiler? 2011-09-23T16:14:25 no, the check is at line 416 and 417 of compiler.py 2011-09-23T16:14:41 ah, i see 2011-09-23T16:15:04 we could make another compiler pass that simply touches MyBot.beam 2011-09-23T16:15:13 not sure why it was done that except that every other language can produce a MyBot.something file :( 2011-09-23T16:15:17 *** Accoun has joined #aichallenge 2011-09-23T16:15:50 well the right thing to do would be to look for a my_bot.beam file, right? 2011-09-23T16:16:02 yes, that would be the real fix 2011-09-23T16:17:21 pretty sure that is the only place Language.extension is used, think I'll just change it to Language.out_file 2011-09-23T16:17:56 touches a fair number of lines but is a fairly simple change in logic 2011-09-23T16:21:23 easier than changing all the other languages to "myBot" for sure ;) 2011-09-23T16:21:31 hehe 2011-09-23T16:27:30 the comments in there are quite outdated as well 2011-09-23T16:27:40 fixing a few of them at least 2011-09-23T16:28:27 i found them quite helpful, the big blurp at the top at least 2011-09-23T16:29:14 yeah, the general idea is still accurate they just still refer to a languages dictionary and the examples show code for that 2011-09-23T16:31:10 *** detryo has joined #aichallenge 2011-09-23T16:31:53 aichallenge: janzert epsilon * rda8d1bc / worker/compiler.py : Have languages specify an output file instead of just an extension, also update some of the comments - http://git.io/WyB99w 2011-09-23T16:33:40 try that 2011-09-23T16:34:22 k, thanks 2011-09-23T16:35:58 hm... 2011-09-23T16:38:44 *** superflit has quit IRC (Remote host closed the connection) 2011-09-23T16:38:56 *** superflit has joined #aichallenge 2011-09-23T16:40:00 *** boegel has quit IRC (Quit: Leaving) 2011-09-23T16:53:19 gattschardo: what language are you writting in? 2011-09-23T16:54:26 I guess it's erlang 2011-09-23T16:54:38