2011-10-27T00:00:21 *** rac_ has joined #aichallenge 2011-10-27T00:01:02 hi 2011-10-27T00:01:52 *** ccc_ has quit IRC (Client Quit) 2011-10-27T00:02:09 *** cyphase has quit IRC (Quit: http://www.cyphase.com/) 2011-10-27T00:03:01 *** rac_ has quit IRC (Client Quit) 2011-10-27T00:04:19 *** GeorgeSebastian has quit IRC (Read error: Connection reset by peer) 2011-10-27T00:06:53 *** cyphase has joined #aichallenge 2011-10-27T00:07:00 hi, i am interested in using one of the more unorthodox languages (common lisp). is there API documentation out there, or do i just have to read the source? 2011-10-27T00:07:23 there are 2 starter packages for cl 2011-10-27T00:07:38 ccc_: unless you have very expensive path finding algorithms in there, there is no reason it should time out 2011-10-27T00:07:52 ccc_: is it saying "timed out"? 2011-10-27T00:07:58 *** McLeopold has left #aichallenge 2011-10-27T00:07:59 *** moondust has quit IRC (Ping timeout: 260 seconds) 2011-10-27T00:08:33 ah, i was planning on using the official one 2011-10-27T00:08:56 unless there is a good reason not to 2011-10-27T00:10:16 *** kornork has quit IRC (Ping timeout: 265 seconds) 2011-10-27T00:10:23 f4hy: do you limit the BFS to visible nodes for given ant? 2011-10-27T00:10:50 ccc_ cannot be helped 2011-10-27T00:11:20 *** chiku1188 has joined #aichallenge 2011-10-27T00:11:31 *** chiku1188 has left #aichallenge 2011-10-27T00:11:33 grom358: I would, or change the default map to mark them unpassable 2011-10-27T00:11:53 grom358: At least in java unexplored tiles are marked as land 2011-10-27T00:12:06 yeah I noticed that 2011-10-27T00:12:11 they need an Ilk.UNKNOWN 2011-10-27T00:12:43 as soon as I get time to make a path find algorithm that's the first thing i'm gonna change 2011-10-27T00:13:51 *** moondust has joined #aichallenge 2011-10-27T00:15:36 so, uh, are there general API docs out there, if not language specific ones, or is it just the source 2011-10-27T00:15:50 *** dvladim has quit IRC (Ping timeout: 260 seconds) 2011-10-27T00:16:03 jimi_hendrix: http://aichallenge.org/specification.php 2011-10-27T00:18:16 *** AntGotAChance has joined #aichallenge 2011-10-27T00:19:18 finally: [visualizer]GameID or GameURI or GameData(paste directly)[/visualizer] 2011-10-27T00:19:23 mleise: what do you think? 2011-10-27T00:20:15 mleise: To be honest I don't care much :) 2011-10-27T00:21:04 *** AntGotAChance has quit IRC (Client Quit) 2011-10-27T00:27:06 *** v_paul_v has joined #aichallenge 2011-10-27T00:27:50 *** Puj has quit IRC (Ping timeout: 276 seconds) 2011-10-27T00:29:08 *** Sunshowers has left #aichallenge 2011-10-27T00:29:36 *** oops has quit IRC (Ping timeout: 265 seconds) 2011-10-27T00:31:14 *** stryder199 has joined #aichallenge 2011-10-27T00:34:14 *** cyphase has quit IRC (Ping timeout: 260 seconds) 2011-10-27T00:34:37 bobbydroptable, thanks 2011-10-27T00:34:51 *** stryder199 has quit IRC (Client Quit) 2011-10-27T00:34:53 *** Areks|2 has joined #aichallenge 2011-10-27T00:35:15 *** Knekkebjoern has joined #aichallenge 2011-10-27T00:35:15 *** atetubou has joined #aichallenge 2011-10-27T00:36:22 *** atetubou has quit IRC (Client Quit) 2011-10-27T00:37:22 what am I doing wrong here, http://pastebin.com/gYdaRPQs 2011-10-27T00:38:11 *** choas has joined #aichallenge 2011-10-27T00:38:24 grom358: what are you doing wrong? do you have an error message? 2011-10-27T00:39:35 infinite loop 2011-10-27T00:40:05 grom358: you're not emptying the queue 2011-10-27T00:40:09 wait, nvm 2011-10-27T00:40:20 you are emptying, but then you're adding nodes to it 2011-10-27T00:40:21 *** Areks|2 has quit IRC (Ping timeout: 244 seconds) 2011-10-27T00:40:28 are you adding more than you are taking? 2011-10-27T00:40:43 oh.. i see woops forgot to tag the children as visited 2011-10-27T00:41:28 ah thats better 2011-10-27T00:41:35 *** SharkMonkey has joined #aichallenge 2011-10-27T00:42:16 sweet.. now I have vision to just walkable tiles 2011-10-27T00:46:28 @seen janzert 2011-10-27T00:46:28 amstan: janzert was last seen in #aichallenge 2 hours, 22 minutes, and 47 seconds ago: ChrisH_: historically it hasn't 2011-10-27T00:47:04 ? 2011-10-27T00:47:13 janzert: can we merge this? https://github.com/aichallenge/aichallenge/pull/324 2011-10-27T00:47:31 you can but it's broken ;) 2011-10-27T00:47:46 it is? 2011-10-27T00:47:50 if ants from two different players reach food simultaneously, is the food destroyed without spawning anything? 2011-10-27T00:47:50 I won't have a chance to fix it up till friday or the weekend probably 2011-10-27T00:47:55 yes, it is 2011-10-27T00:47:58 thx 2011-10-27T00:48:10 Q: What happens when a bot capture more than a piece of food in a turn? Are the 'newborn' queued or "fuck the 4 pieces you ate; just 1 newborn next turn anyway" ? 2011-10-27T00:48:25 a1k0n: oops I wasn't talking to you, but yes is still correct for you as well :) 2011-10-27T00:48:25 queued i assume 2011-10-27T00:48:31 oh ok. :) 2011-10-27T00:48:47 I also assume, but if we're wrong.... 2011-10-27T00:48:55 janzert: why is it broken? 2011-10-27T00:48:56 amstan: the compiler.py changes are almost certain not to work as is 2011-10-27T00:49:04 ic 2011-10-27T00:49:06 nor does it add the installation of ada to the worker 2011-10-27T00:49:23 should be pretty easy to fix, just have to do it 2011-10-27T00:49:38 well, it adds the installation just in the wrong spot 2011-10-27T00:49:57 *** rperkins has joined #aichallenge 2011-10-27T00:50:08 ugh.. i hate license headers in every file 2011-10-27T00:50:17 especially when the license is actually longer than the contents 2011-10-27T00:50:29 heh 2011-10-27T00:50:33 look at this: https://github.com/philxyz/aichallenge/blob/fbe41c2cfab6fbf911906dd806d1c25f4adf1574/ants/dist/starter_bots/ada/strategy.adb 2011-10-27T00:50:42 ouch 2011-10-27T00:50:49 *** rperkins has quit IRC (Client Quit) 2011-10-27T00:51:13 *** schnoid has joined #aichallenge 2011-10-27T00:51:19 yeah, full license should just be once in a separate file 2011-10-27T00:51:21 btw, i know we do apache, but do we have text explaining that anywhere? 2011-10-27T00:51:36 we never properly settled on apache even 2011-10-27T00:51:50 Also, if you have an ant right on a hill and a newborn arrives from this hill, are one of those two ants dead? I'm pretty sure as I don't see what could happen else, but, can anyone confirm? 2011-10-27T00:51:51 at the beginning of ants we were going to move to bsd or mit but it never happened 2011-10-27T00:52:11 we should fix that, some contributors easily get offended if they don't like the license 2011-10-27T00:52:14 including those in full in each file wouldn't be so bad :} 2011-10-27T00:52:25 meh 2011-10-27T00:52:31 I don't like to 'assume' anything in that kind of problem solving contest :P 2011-10-27T00:52:38 fixing it now technically means contacting each contributor 2011-10-27T00:52:39 Sakura_Away: ants on a hill are "bottled up" if you don't move the ant on top of the hill 2011-10-27T00:52:44 they don't kill the guy at the top 2011-10-27T00:52:56 janzert: do we have to include it in each file? 2011-10-27T00:53:01 shouldn't no 2011-10-27T00:53:03 a1k0n: bottled up = queued? 2011-10-27T00:53:06 yes 2011-10-27T00:53:19 *** rperkins has joined #aichallenge 2011-10-27T00:53:23 it'd be good to get philxyz to move it out of the source files at least 2011-10-27T00:53:25 so you could save up a bunch of ants and then spew them out one per turn if you really wanted 2011-10-27T00:53:30 and just include it in a LICENSE file 2011-10-27T00:53:41 janzert: ok, replying to him 2011-10-27T00:54:03 aik0n: great. Thanks 2011-10-27T00:54:07 *** choas has quit IRC (Ping timeout: 248 seconds) 2011-10-27T00:54:25 if you do that an enemy could just park two ants right next to the hill and kill them all in a row... 2011-10-27T00:54:47 aik0n: I guess you can also confirm that if your bot eats more than one piece in a turn, the newborns are bottled up; correct? 2011-10-27T00:55:16 Sakura_Away: i haven't personally checked it but that is how i understand the rules 2011-10-27T00:55:22 hehe right 2011-10-27T00:55:28 ok, thanks 2011-10-27T00:55:47 hi 2011-10-27T00:57:50 *** callahan has joined #aichallenge 2011-10-27T00:59:05 hey caleb 2011-10-27T00:59:12 you doing this contest? 2011-10-27T00:59:16 Looks like the game rate is low. Game servers having trouble? 2011-10-27T00:59:18 Don't think so 2011-10-27T00:59:37 building more robotic food-preparation devices instead? 2011-10-27T00:59:39 I am measuring interest in using OpenRA as next year's platform 2011-10-27T00:59:40 a1k0n: he wants to make the new contest http://openra.res0l.net/ 2011-10-27T00:59:43 *** rj_ has joined #aichallenge 2011-10-27T00:59:56 ohho 2011-10-27T01:00:00 *** he_the_great has quit IRC (Ping timeout: 240 seconds) 2011-10-27T01:00:08 *** Six has quit IRC (Read error: Connection reset by peer) 2011-10-27T01:00:17 so like the starcraft thing 2011-10-27T01:00:31 but more.. open. but i assume you still need to obtain the red alert assets? 2011-10-27T01:00:37 hi 2011-10-27T01:00:41 It downloads them from EA 2011-10-27T01:00:42 perhaps i should just read my scrollback 2011-10-27T01:00:44 RobotCaleb: i assume you know a great deal of that game. do you think the game rules can be used as is? 2011-10-27T01:00:48 i have checked in my code but 2011-10-27T01:01:17 rj_: hmm? 2011-10-27T01:01:29 since last 6 hrs it been showing Next game should be within 54.4 minutes. 2011-10-27T01:01:55 *** denysonique__ has quit IRC (Read error: Operation timed out) 2011-10-27T01:02:00 *** denysonique__ has joined #aichallenge 2011-10-27T01:02:06 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 256 seconds) 2011-10-27T01:02:12 only one worker's up 2011-10-27T01:02:26 chrisf (in #openra. He might come along here) is way more knowledgable than I am. He wrote most of the engine. I just contribute patches occasionally 2011-10-27T01:02:29 rj_: sorry about that, the algorithm for detecting that is kinda flaky 2011-10-27T01:02:32 i've seen as many as five. are they supposed to all be running 24/7? 2011-10-27T01:02:41 rj_: think of it like the windows progress bar 2011-10-27T01:02:44 *** maran has joined #aichallenge 2011-10-27T01:03:03 RobotCaleb: i think for this contest it might just be too damn complicated 2011-10-27T01:03:32 which isn't to say a redalert AI contest isn't a bad idea 2011-10-27T01:04:04 and it increases also 2011-10-27T01:04:06 er, is. too many negatives 2011-10-27T01:04:56 a1k0n: negatives? 2011-10-27T01:05:17 my last sentence was a double-negative and wasn't intended to be 2011-10-27T01:05:42 i parsed it alright 2011-10-27T01:06:13 *** Zaphus has joined #aichallenge 2011-10-27T01:06:20 *** DecodedCortex has quit IRC (Ping timeout: 265 seconds) 2011-10-27T01:06:58 *** rperkins has quit IRC (Quit: Peacing out) 2011-10-27T01:07:20 *** rperkins has joined #aichallenge 2011-10-27T01:07:41 *** rperkins is now known as TheDigitalNinja 2011-10-27T01:08:09 would be nice to have the next game not be rts style.. since this one and last one were 2011-10-27T01:08:43 burny: suggestions? 2011-10-27T01:08:57 burny: they're all going to boil down to strategy 2011-10-27T01:09:04 burny: I agree, we just have to come up with a good one 2011-10-27T01:09:48 rts of course lends itself to nice visualization besides all the other benefits it gives though 2011-10-27T01:09:49 can do an ai contest for basically ANY game 2011-10-27T01:10:07 weren't people suggesting go 2011-10-27T01:10:17 but for our contest we have lots of restrictions on ourselves for an ideal game 2011-10-27T01:10:18 how about team games 2011-10-27T01:10:31 I like the idea of team games also 2011-10-27T01:10:52 team game could be interesting, but that doesn't necessarily get us away from rts of course 2011-10-27T01:10:55 yeh because you don't know what your team mate will do 2011-10-27T01:10:57 yeh 2011-10-27T01:11:00 team based fps/capture the flag style could be neat 2011-10-27T01:11:28 pure bot fps is hard to make interesting actually 2011-10-27T01:11:40 since a big portion of the traditional fps is about reaction times 2011-10-27T01:11:46 i worry about all-starter-pack teams 2011-10-27T01:11:51 it can be fuzzy 2011-10-27T01:12:01 though with truskill at least the somewhat decent bots would fight together 2011-10-27T01:12:15 eg. if you are prone, then opponent is less likely to be given your infos 2011-10-27T01:12:21 sure you can artificially limit the bots but that's not very interesting 2011-10-27T01:12:36 if you're running, they'll be more likely to notice you, even if they're facing hte other way, etc. 2011-10-27T01:12:50 bzflag bot game :P 2011-10-27T01:13:05 team games might also have perverse incentives for players vying for "MVP" 2011-10-27T01:13:06 personally I would love go.. but I think it's WAY above the heads of most people 2011-10-27T01:13:35 and go is really too abstract for us as well 2011-10-27T01:14:06 a persistent game of some sort would be really sweet 2011-10-27T01:14:14 might be hard to pull off well though 2011-10-27T01:14:23 it's also pretty well-studied.. the answer will be to do whatever the best go bots do already 2011-10-27T01:14:24 persistent like? 2011-10-27T01:14:45 persistent world, not individual separate games 2011-10-27T01:14:50 rts's are pretty well studied also :P 2011-10-27T01:15:38 minecraft for bots :) 2011-10-27T01:15:50 * antimatroid1 is a big fan of simultaneous 2011-10-27T01:16:15 I can't think of anything that would work as persistent besides something like a stock market sim 2011-10-27T01:16:17 actually Eve for bots would be a better analogy I suppose 2011-10-27T01:16:48 delt0r was making what sounded like a pretty sweet game 2011-10-27T01:17:08 so you either have real time or turn based, not much difference 2011-10-27T01:17:44 how about tower defense lol 2011-10-27T01:17:51 *** gorogm has joined #aichallenge 2011-10-27T01:18:08 bobbydroptable: it's not vs anything 2011-10-27T01:18:13 randomly generated contra? 2011-10-27T01:18:34 well it could be a adverserial bot that decides what monsters to spawn 2011-10-27T01:18:39 lol 2011-10-27T01:19:06 *** Zaphus has quit IRC (Quit: Page closed) 2011-10-27T01:19:07 yeah a side-scroller would be fun 2011-10-27T01:19:14 though the answer to that is A* 2011-10-27T01:19:37 team based contra though.. 2011-10-27T01:20:03 so it's not just about "doing the best thing".. it's about doing what is best for the team 2011-10-27T01:20:03 as long as your bullets pass harmlessly through your teammates... 2011-10-27T01:20:18 for sure 2011-10-27T01:20:41 heh, so no jumping to the top on the waterfall level and killing your teammate at the bottom 2011-10-27T01:20:48 a1k0n: weren't all those mario videos done with minmax? 2011-10-27T01:20:55 A*, not minimax 2011-10-27T01:20:58 mario videos? 2011-10-27T01:21:06 there's no decisions for enemies ot make 2011-10-27T01:21:18 http://www.youtube.com/watch?v=DlkMs4ZHHr8 2011-10-27T01:21:20 oh, right 2011-10-27T01:22:21 i just realized in the shower this morning that i should have used a closed set for my A* in my entry in that contest, for states where mario's x velocity is 0 2011-10-27T01:22:46 *** xar0l has quit IRC (Ping timeout: 265 seconds) 2011-10-27T01:23:41 can all games have this mode? http://www.youtube.com/watch?annotation_id=annotation_512150&feature=iv&v=qYluZRwrw9w&src_vid=DlkMs4ZHHr8 2011-10-27T01:23:45 *** gorogm has quit IRC (Quit: Page closed) 2011-10-27T01:23:51 *** whe has quit IRC (Ping timeout: 252 seconds) 2011-10-27T01:23:52 if the enemys did some actions randomly(eg. shoot at you), it would massivly change that though 2011-10-27T01:24:14 *** cyphase has joined #aichallenge 2011-10-27T01:24:15 and, if there's potentially a team-mate not runnign as fast, to prevent you from landing wher eoyu expect, etc. 2011-10-27T01:24:18 burny: the ai would try to avoid it before it hits him 2011-10-27T01:24:40 unless there's no way to avoid, and in that case you're dead 2011-10-27T01:24:46 but you can't solve that at all 2011-10-27T01:25:02 unless you know you're in a dangerous spot and you should watch out 2011-10-27T01:26:09 amstan, it would have things like, deciding if you should shoot left or right.. and if a bullet is coming, should you jump over it, or back up before jumping 2011-10-27T01:26:41 yes, but that's all done with a* 2011-10-27T01:27:14 maybe the best thing in some situation, is to shoot left, and your partner shoot right.. but your partner is shooting left even if you do, so it's better if you take the sub-optimal strat, and shoot right 2011-10-27T01:27:30 hey speaking of RTS AI: http://graphics.stanford.edu/~mdfisher/GameAIs.html 2011-10-27T01:28:18 a1k0n: just before tron, we have a meeting at uwaterloo, we were going to make a team for that(with jeff and a bunch of others) 2011-10-27T01:28:35 mario AI? 2011-10-27T01:28:41 no, starcraft 2011-10-27T01:28:46 oh, yeah. 2011-10-27T01:28:58 this is SC2, so that's new 2011-10-27T01:29:00 the setup was a pain though, it relies on injecting dlls into starcraft 2011-10-27T01:29:31 it wasn't very linux friendly 2011-10-27T01:29:36 * cyphase is trying to make his algorithm work optimally no matter where he is 2011-10-27T01:29:55 *** jfs has joined #aichallenge 2011-10-27T01:30:16 jfs: is that like the filesystem? 2011-10-27T01:30:24 *** dr- has quit IRC (Ping timeout: 240 seconds) 2011-10-27T01:30:30 the ant is* 2011-10-27T01:30:50 amstan: no 2011-10-27T01:31:28 *** pgoldbr has quit IRC (Ping timeout: 265 seconds) 2011-10-27T01:36:18 *** r3b00tx has joined #aichallenge 2011-10-27T01:36:42 *** r3b00tx has left #aichallenge 2011-10-27T01:37:47 *** dr- has joined #aichallenge 2011-10-27T01:37:58 *** smiley1983 has joined #aichallenge 2011-10-27T01:38:37 *** maran_ has joined #aichallenge 2011-10-27T01:43:13 *** Garf has joined #aichallenge 2011-10-27T01:43:17 did the mario ai thing, end up with multiple 'perfect' solutions? 2011-10-27T01:45:56 three of us had essentially the same thing, with varying degrees of simulation fidelity (i didn't simulate the exact time cannons shoot, the winner did) 2011-10-27T01:47:30 *** replore__ has joined #aichallenge 2011-10-27T01:48:05 so, game types that could be neat, with and without teams.. side scroller, fps, fighter.. also tons of retro games could be modified to probably work well 2011-10-27T01:48:08 aichallenge: Alexandru Stan epsilon * r97195ef / (2 files): 2011-10-27T01:48:08 aichallenge: Merge pull request #334 from rat4/epsilon 2011-10-27T01:48:08 aichallenge: update java starter bot - http://git.io/undHqw 2011-10-27T01:48:10 *** replore_ has quit IRC (Read error: Connection reset by peer) 2011-10-27T01:53:02 *** ChrisH_ has joined #aichallenge 2011-10-27T01:55:05 antimatroid1: https://plus.google.com/u/0/118210384829172301796/posts/15ewiYkbGKC 2011-10-27T01:55:16 a1k0n: what i disliked about the mario challenge was that _every last detail_ could be accounted for. if you know exactly how your enemies are going to behave and even know their exact timings, the winner is the one with the best copy of an already existant game model, not the one who comes up with a good idea. (or at least that is what the challenge looked like from an outside perspective. _my_ outside perspective, at that ;)) 2011-10-27T01:56:26 *** oops has joined #aichallenge 2011-10-27T01:58:18 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-10-27T02:05:41 *** Seth1010 has joined #aichallenge 2011-10-27T02:10:08 *** Seth1010 has quit IRC (Ping timeout: 265 seconds) 2011-10-27T02:10:48 *** smiley1983 has quit IRC (Quit: Lost terminal) 2011-10-27T02:11:20 *** Palmik has joined #aichallenge 2011-10-27T02:14:05 :( my A* is so broken 2011-10-27T02:17:16 *** Sakura_Away has quit IRC (Quit: Page closed) 2011-10-27T02:17:32 *** aperiodic has joined #aichallenge 2011-10-27T02:20:51 *** djr__ has quit IRC (Quit: Leaving) 2011-10-27T02:23:41 *** Seth1010 has joined #aichallenge 2011-10-27T02:26:22 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-10-27T02:27:13 *** Rainydays has joined #aichallenge 2011-10-27T02:27:26 if there is an actual path.. A* should find it right? 2011-10-27T02:27:36 yeah 2011-10-27T02:27:38 yes 2011-10-27T02:28:17 what is the Ilk of a dead ant-hill? 2011-10-27T02:28:26 is it just treated as ground? 2011-10-27T02:28:54 *** kaemo has quit IRC (Read error: Operation timed out) 2011-10-27T02:29:36 *** cyphase has quit IRC (Ping timeout: 240 seconds) 2011-10-27T02:31:09 can you see where going wrong with this then.. http://pastebin.com/Qj0WEbtb 2011-10-27T02:32:40 its pathing into a wall and stops 2011-10-27T02:34:50 do you check to see if a tile is passable or not when adding it to the frontier? 2011-10-27T02:36:14 *** Puj has joined #aichallenge 2011-10-27T02:38:10 *** dominater has quit IRC (Ping timeout: 265 seconds) 2011-10-27T02:39:05 *** cybsy has joined #aichallenge 2011-10-27T02:39:58 *** nplus has quit IRC (Read error: Connection reset by peer) 2011-10-27T02:45:47 yeah map.getAdjacent(current.getTile()) only returns passable tiles 2011-10-27T02:46:16 *** nplus has joined #aichallenge 2011-10-27T02:47:35 *** kaemo has joined #aichallenge 2011-10-27T02:48:43 *** danielharan has joined #aichallenge 2011-10-27T02:50:04 *** grom358 has quit IRC (Quit: Ex-Chat) 2011-10-27T02:50:28 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T02:50:44 *** craklyn has quit IRC (Ping timeout: 265 seconds) 2011-10-27T02:50:59 *** danielharan has joined #aichallenge 2011-10-27T02:52:06 *** sdf has joined #aichallenge 2011-10-27T02:52:27 *** isbric has joined #aichallenge 2011-10-27T02:52:34 morning 2011-10-27T02:53:50 culd somone tell me why the starter pack of c++ crashes i change the Bot.cc to always move in direction 2 2011-10-27T02:54:16 why did they make it Row,Col? 2011-10-27T02:54:25 i'm so used to X,Y that i keep introducing bugs in my code 2011-10-27T02:54:49 ie: comment the "for(int d=4; d has quit IRC (Ping timeout: 258 seconds) 2011-10-27T02:55:39 isbric what's the error message 2011-10-27T02:55:40 bobbydroptable: i think that you more then wellcome to change any parts of the prewritten code. 2011-10-27T02:55:56 isbric the entire game engine etc is all in row,col 2011-10-27T02:56:02 so no point changing it 2011-10-27T02:56:17 I just invert the input I receive bobby 2011-10-27T02:56:18 it's just i keep getting confused when i put in x,y instead and that basically translate to col,row 2011-10-27T02:56:20 and treat it as x y 2011-10-27T02:56:47 and when i'm sending the comand to move I just invert it again 2011-10-27T02:56:48 bobbydroptable: "turn 6 bot 0 crashed" 2011-10-27T02:57:16 grom still there? 2011-10-27T02:57:20 oh 2011-10-27T02:57:21 he left 2011-10-27T02:57:21 =p 2011-10-27T02:57:57 isbric : use -E or -e on playgame.py to log error messages 2011-10-27T02:58:04 Draivin: if you index with [x][y] then you need to transpose the grid every time you input and output it 2011-10-27T02:58:06 bobbydroptable: yeah, that happend to me to so i changed them to xy, also change the move functionality sins im use to top,right,down,left order 2011-10-27T02:58:18 it's an awfully confusing way to do things 2011-10-27T02:58:25 not really 2011-10-27T02:58:32 I have a parser that takes care of input ant output 2011-10-27T02:58:40 so I don't even care about it 2011-10-27T02:58:42 if you index with [row][col] you never have to think about whether it needs transposing 2011-10-27T02:58:46 bobbydroptable: im using -verbose -e 2011-10-27T02:59:00 and* 2011-10-27T03:00:53 has anyone been using GAs on this contest yet? 2011-10-27T03:00:55 *** analyst74 has quit IRC (Ping timeout: 260 seconds) 2011-10-27T03:01:10 antimatroid1: believe it or not x,y is just much more natural to a programmer that has always dealt with it though :} 2011-10-27T03:01:12 "or rather is anyone not using a GA?" 2011-10-27T03:01:36 janzert: that doesn't stop having to transpose it from being confusing :P 2011-10-27T03:02:01 why it would be confusing if your actual code never has to deal with it? 2011-10-27T03:02:03 isbric then i'm not sure. you should be able to pick up the error messages in the game_log directory ( for python) 2011-10-27T03:02:12 basically just have to do it in two places and th rest of the time you can deal with what you're used to 2011-10-27T03:02:46 well i couldn't be stuffed to rewrite the starter kit so lol 2011-10-27T03:02:53 i'll just suck it up 2011-10-27T03:03:28 *** dmiles_afk has quit IRC (Ping timeout: 244 seconds) 2011-10-27T03:03:40 bobbydroptable: nope, no debug information what so ever. 2011-10-27T03:03:48 row, col is much more explicit for people that don't already have there mind stuck on x,y of course 2011-10-27T03:04:19 Option Explicit ftw 2011-10-27T03:04:33 guess so, I couldn't program intuitively based on it tough 2011-10-27T03:04:34 so I changed 2011-10-27T03:05:19 Draivin: if I were writing a bot I probably would as well 2011-10-27T03:05:22 isbric how about if you do this 2011-10-27T03:05:47 isbric, I know you. 2011-10-27T03:05:48 if you run playgame with -I 2011-10-27T03:05:54 it will generate input logs 2011-10-27T03:06:11 then ./MyBot < game_logs/0.bot1.input or whatever it's called 2011-10-27T03:06:29 note i put bot1.input, because that will have the inputs for the 2nd player (since yours crashed) 2011-10-27T03:06:55 or you could use bot0.input, and i guess it should replicate the crash 2011-10-27T03:07:10 isbric, jag känner dej 2011-10-27T03:07:48 *** grom358 has joined #aichallenge 2011-10-27T03:09:34 Oh hey grom 2011-10-27T03:09:36 *** kaemo has quit IRC (Ping timeout: 240 seconds) 2011-10-27T03:09:37 managed to solve your problem? 2011-10-27T03:12:31 *** oops has quit IRC (Quit: Page closed) 2011-10-27T03:14:34 *** aerique has joined #aichallenge 2011-10-27T03:15:12 http://grom.zeminvaders.net/files/pathfind.zip 2011-10-27T03:15:20 Draivin: ah no 2011-10-27T03:16:04 its really annoying me cause I have gotten A* to work in another demo I did once.. but can't get it working in Ants 2011-10-27T03:17:24 what isn't working ,exactly 2011-10-27T03:17:25 *** halligalli has joined #aichallenge 2011-10-27T03:17:30 i'll try and narrow it down for you 2011-10-27T03:17:41 it gets to wall and stops 2011-10-27T03:18:16 that file I just linked reads the maze map and prints it out with c marking closed nodes 2011-10-27T03:18:35 *** Bluedgis has joined #aichallenge 2011-10-27T03:19:08 where are you checking to see if you've reached goal? 2011-10-27T03:19:22 oh wait sorry saw it 2011-10-27T03:19:23 line 43 2011-10-27T03:19:34 are you sorting the frontier by lowest cost? 2011-10-27T03:19:50 what u mean by frontier? 2011-10-27T03:20:00 the frontier - openset 2011-10-27T03:20:07 the open sets.. yes.. using TreeSet 2011-10-27T03:20:12 *** nishant has joined #aichallenge 2011-10-27T03:20:31 and Node is sorted by fScore by implementing Comparable 2011-10-27T03:20:35 when you say it "stops" 2011-10-27T03:20:42 is that in the visualizer? 2011-10-27T03:20:47 hey guys having a problem at tutorial step 2 2011-10-27T03:21:10 during excutoio there is an error at 2011-10-27T03:21:12 TreeSet sortedFood = new TreeSet(ants.getFoodTiles()); 2011-10-27T03:21:15 bobbydroptable: for testing I got just command line utility 2011-10-27T03:21:21 ah ok 2011-10-27T03:21:30 http://grom.zeminvaders.net/files/pathfind.zip 2011-10-27T03:21:47 anyone can help? 2011-10-27T03:21:49 easier to test just the A* this way 2011-10-27T03:22:00 *** Seth1010 has quit IRC (Quit: Page closed) 2011-10-27T03:22:55 *** Blkt has joined #aichallenge 2011-10-27T03:23:14 hey guys having a problem at tutorial step 2 2011-10-27T03:23:15 during excutoio there is an error at 2011-10-27T03:23:15 TreeSet sortedFood = new TreeSet(ants.getFoodTiles()); 2011-10-27T03:23:24 this is java package 2011-10-27T03:25:32 *** schnoid has quit IRC (Ping timeout: 265 seconds) 2011-10-27T03:25:44 *** delt0r__ has quit IRC (Ping timeout: 260 seconds) 2011-10-27T03:26:30 *** Puj has quit IRC (Read error: Connection reset by peer) 2011-10-27T03:27:27 *** danielharan has quit IRC (Ping timeout: 260 seconds) 2011-10-27T03:28:27 it looks like its not adding adjacent tiles to openset 2011-10-27T03:28:29 grom358 your a* seems to be going in the wrong direction 2011-10-27T03:28:31 *** maran has quit IRC (Quit: Page closed) 2011-10-27T03:29:11 yeah.. I can't make sense of this at all 2011-10-27T03:29:16 *** sir_macelon has joined #aichallenge 2011-10-27T03:29:23 i have just followed the tutorial directions and am a bit of a newbie at this stuff 2011-10-27T03:29:25 good morning everyone 2011-10-27T03:30:20 nishant: you should download the latest tools package it was updated recently 2011-10-27T03:30:21 *** Palmik has quit IRC (Read error: Connection reset by peer) 2011-10-27T03:30:41 alright will try that 2011-10-27T03:31:02 it starts a [21,6] 2011-10-27T03:31:16 *** danielharan has joined #aichallenge 2011-10-27T03:31:19 bobbydroptable: I keep rereading it looking at the pseduocode off wikipedia and else where 2011-10-27T03:31:24 adds 20,6 21,7 22,6 21,5 2011-10-27T03:31:43 the closest one to 51 54 would be 22,6 2011-10-27T03:32:05 but it decides to go for 20,6 instead 2011-10-27T03:32:15 so seems that your sorting algorithm is backwards 2011-10-27T03:33:44 sir_macelon: I still get the same error 2011-10-27T03:33:59 what the.. i just swap the sort order around and it still goes in the same direction 2011-10-27T03:34:14 I think its a problem with the data types as it gves a Classcast Exception 2011-10-27T03:34:48 *** epicmonkey has joined #aichallenge 2011-10-27T03:34:49 nishant: that's impossible, maybe the tools are not updated, try pulling directly from github 2011-10-27T03:34:52 oh wait a minute 2011-10-27T03:34:53 lol 2011-10-27T03:35:34 sir_macelon: the tools seem to be updated, when extracted their size is 1.9 MB as opposed to 2.3 MB before this 2011-10-27T03:36:01 bah 2011-10-27T03:36:16 this.. 2011-10-27T03:36:28 nishant: check your content of Tile.java and compare it to this one: https://github.com/aichallenge/aichallenge/blob/epsilon/ants/dist/starter_bots/java/Tile.java 2011-10-27T03:36:58 bobbydroptable: see how the x doesn't get changed to an o .. 2011-10-27T03:37:06 gggrrr.. this is driving me insane.. 2011-10-27T03:37:10 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T03:37:31 *** Harpyon has joined #aichallenge 2011-10-27T03:37:46 *** Harpyon_ has joined #aichallenge 2011-10-27T03:37:46 *** delt0r__ has joined #aichallenge 2011-10-27T03:37:47 *** danielharan has joined #aichallenge 2011-10-27T03:38:20 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T03:38:28 sir_macelon: It worked like a charm....thanks a lot 2011-10-27T03:38:56 nishant: no problem, you're welcome 2011-10-27T03:38:57 *** danielharan has joined #aichallenge 2011-10-27T03:39:45 okay 2011-10-27T03:39:54 your neighbour nodes are never getting added more than once 2011-10-27T03:40:02 *** mithos28 has joined #aichallenge 2011-10-27T03:40:30 where? 2011-10-27T03:40:50 *** Harpyon_ has quit IRC (Client Quit) 2011-10-27T03:40:55 open.contains(neighbour) 2011-10-27T03:41:01 something is wrong with your hashing function 2011-10-27T03:41:18 on the first adj tile, open.contains(neighbour) is false 2011-10-27T03:41:26 on the 2nd adjTile, open.contains(neighbour) is true 2011-10-27T03:41:40 weird.. my hash just calls the Tile hash 2011-10-27T03:43:04 *** savaron has joined #aichallenge 2011-10-27T03:43:30 yeah.. that is where I just got.. the adjacents tiles are not getting added to openset 2011-10-27T03:45:09 suggestion, on Node remove the hashing function 2011-10-27T03:45:48 but isn't the object hashing based on the instance? or something in Java 2011-10-27T03:45:52 if u don't define one 2011-10-27T03:45:55 yeh 2011-10-27T03:46:13 *** denysonique__ has quit IRC (Quit: Connection closed for inactivity) 2011-10-27T03:46:38 kk.. that didn't fix it either 2011-10-27T03:47:41 okay then 2011-10-27T03:49:57 funny 2011-10-27T03:50:03 the hashes are returning different values 2011-10-27T03:50:11 but open.contains still thinks it has your 2nd neighbour 2011-10-27T03:50:33 yeah.. the hash functions look fine to me.. but something is wrong 2011-10-27T03:52:18 okay its not the hash 2011-10-27T03:53:00 yeah.. its not adding all the adjacent nodes to openset.. and for the life of me I don't see why not 2011-10-27T03:54:06 *** mviel has joined #aichallenge 2011-10-27T03:54:09 cant you just run a simple pathfinding from 2,2 to 4,4 for example, and log all the tiles through the process? 2011-10-27T03:54:25 so you can know where the problem is at 2011-10-27T03:54:38 *** RobotCaleb has quit IRC (Quit: Leaving.) 2011-10-27T03:54:55 Draivin: yeah I can already see the problem.. just don't understand it.. its not adding for the very first node.. its adjacent nodes to open set 2011-10-27T03:56:00 by first node you mean the first node the origin adds to the open set right? 2011-10-27T03:56:14 *** savaron has quit IRC (Quit: Page closed) 2011-10-27T03:56:20 yeah.. the origin.. its breaking on the very first time through the loop 2011-10-27T03:57:11 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T04:00:20 it freaking things tile 21,7 is in openset... WTF!! hashCode and equals looks correct to me 2011-10-27T04:02:27 open: [20 6] ... neighbour: 21,7; open.contains(neighbour) returns true 2011-10-27T04:02:38 seriously.. WTF! 2011-10-27T04:03:43 *** cbad has quit IRC (Ping timeout: 245 seconds) 2011-10-27T04:03:45 lol 2011-10-27T04:03:52 yeh that's what i'm looking at too 2011-10-27T04:04:03 mmm.. i think its todo with TreeSet 2011-10-27T04:04:10 *** nishant has quit IRC (Read error: Connection reset by peer) 2011-10-27T04:04:18 TreeSet instance performs all element comparisons using its compareTo (or compare) method 2011-10-27T04:05:01 i've put in some messages into equals and i'm getting messages 2011-10-27T04:05:12 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-10-27T04:05:18 for contains method it uses equals 2011-10-27T04:05:27 *** nishant has joined #aichallenge 2011-10-27T04:05:38 yeah I thought so.. I just can't explain why it things the openset has the neighbour when it does 2011-10-27T04:05:43 does NOT 2011-10-27T04:05:47 hehehe 2011-10-27T04:06:10 grom358: don't use the TreeSet for open set! 2011-10-27T04:06:25 sir_macelon: why not? 2011-10-27T04:06:41 grom358: when compareTo returns 0 then is assumes the objects are identical 2011-10-27T04:06:54 omg.. that explains this bug then 2011-10-27T04:06:54 grom358: and because it is a SET it won't add them 2011-10-27T04:07:17 all Nodes which are equidistant won't be added 2011-10-27T04:07:27 but is it actually using compareTo? 2011-10-27T04:07:29 use List instead 2011-10-27T04:07:31 so.. what data structure? 2011-10-27T04:07:43 HashSet? 2011-10-27T04:07:49 PriorityQueue? 2011-10-27T04:08:03 do u linear search for lowest score element then? 2011-10-27T04:08:06 PQ is go 2011-10-27T04:08:07 good 2011-10-27T04:08:10 bobbydroptable: yes because it is ordering the elements 2011-10-27T04:08:32 well grom, time to rewrite. 2011-10-27T04:08:39 you can use a List and Collections.sort(List) do sort it 2011-10-27T04:08:40 I implemented binary heap as described here 2011-10-27T04:08:41 http://www.policyalmanac.org/games/binaryHeaps.htm 2011-10-27T04:09:01 LOL 2011-10-27T04:09:08 man the whole time.. it was Java not my code 2011-10-27T04:09:13 so to speak 2011-10-27T04:09:19 I thought I had something wrong in my logic 2011-10-27T04:09:43 grom358: it was you - wrong choice of structure :) 2011-10-27T04:09:46 *** kaemo has joined #aichallenge 2011-10-27T04:09:59 yeah.. but I had no idea TreeSet did that 2011-10-27T04:10:02 grom358: aka design fail :) 2011-10-27T04:10:09 lol.. yes 2011-10-27T04:10:18 grom358: have you ever used it before? 2011-10-27T04:10:38 yeah I have 2011-10-27T04:11:01 TreeSet is a implementation of Set which by its nature doesn't allow duplicates 2011-10-27T04:11:08 I just thought it also checked equals method too 2011-10-27T04:11:37 yeah I realise can't have duplicates.. just thought it would use compareTo for order.. and then equals for contains 2011-10-27T04:11:45 TreeSet determines duplicates by compareTo method, if it returns 0, the elements are considered as duplicates 2011-10-27T04:11:59 still doesn't necessarily explain why contains() returns true for an element that isn't in it 2011-10-27T04:12:25 bobbydroptable: because is uses compareTo and it returns 0 2011-10-27T04:13:05 can't he just make it so that if the two are equal 2011-10-27T04:13:09 compareTo returns -1? 2011-10-27T04:13:43 Draivin: no, because it breaks his contract for comparing 2011-10-27T04:13:48 well the PriorityQueue fixed it 2011-10-27T04:14:09 contract? 2011-10-27T04:14:28 bobbydroptable: take a look at the TreeSet implementation it is backed up by NavigableMap and it overrides the contains method 2011-10-27T04:15:11 Draivin: contract is return -1 if one element has lower cost to compared one, the cannot be equal 2011-10-27T04:15:28 *they cannot be equal 2011-10-27T04:15:43 mmm.. still something wrong with this.. gah 2011-10-27T04:15:54 well, if he returns -1 wouldn't it just be inserted at a lower place in the list? 2011-10-27T04:16:52 Draivin: yes, but this a very dirty hack, noone should that 2011-10-27T04:17:07 very unintuitive 2011-10-27T04:17:40 Just wanted to know if it would work =p 2011-10-27T04:18:03 Draivin: I am not sure if it would work at all 2011-10-27T04:18:19 well.. that stopped it getting stuck but its still broken big time 2011-10-27T04:18:35 Draivin: how it would work if you would have a set of elements of all the same value? 2011-10-27T04:18:45 well, as I see it, if it is as I see it, it user the compareTo when inserting a new item on the tree right? 2011-10-27T04:19:31 http://pastebin.com/Yqw0icft 2011-10-27T04:19:38 so.. still don't work :'( 2011-10-27T04:20:03 if it is like that, it would say that this one has a lower value than the previous one, so it would be inserted at a lower index, and therefore be first in the list than their equals 2011-10-27T04:20:11 but when getting the first element from the list 2011-10-27T04:20:24 it doesnt really matter which of the equal distance elements you get first 2011-10-27T04:20:51 you just need them to be after the ones that have a lower distance on the list 2011-10-27T04:21:21 *** kaemo has quit IRC (Ping timeout: 258 seconds) 2011-10-27T04:21:29 *** barsandcat has joined #aichallenge 2011-10-27T04:22:36 Draivin: haven't tried that, but maybe it would work, anyway I don't intend to 2011-10-27T04:22:44 grom358: what's wrong? 2011-10-27T04:23:10 see how it drawing path all over the map :( http://pastebin.com/5i19Lh87 2011-10-27T04:23:18 x marks the path it took 2011-10-27T04:24:14 *** Fandekasp has joined #aichallenge 2011-10-27T04:24:56 LOL 2011-10-27T04:24:58 nvm 2011-10-27T04:25:02 ? 2011-10-27T04:25:25 I was making adjacent tiles as test with x as well I forgot to comment out 2011-10-27T04:26:51 *** master_ninja has joined #aichallenge 2011-10-27T04:27:12 how it looks now? 2011-10-27T04:27:23 grom358: how you are testing basically? 2011-10-27T04:27:42 with maze maps 2011-10-27T04:27:51 path from hill to hill 2011-10-27T04:28:28 *** cbad has joined #aichallenge 2011-10-27T04:28:28 *** cbad has joined #aichallenge 2011-10-27T04:29:26 grom358: I mean how you are performing the tests? 2011-10-27T04:29:40 manually? 2011-10-27T04:30:02 I made command line tool to read the maps, and do a path and write the result to screen 2011-10-27T04:30:33 grom358: but how do you verify the result? 2011-10-27T04:30:44 just visually 2011-10-27T04:31:23 grom358: let me give you a pro tip: start using jUnit before it won't be too late for you 2011-10-27T04:31:39 http://pastebin.com/h5uBwZWW 2011-10-27T04:32:01 yeah.. I been meaning to use JUnit.. I used it before 2011-10-27T04:32:17 *** reblue has joined #aichallenge 2011-10-27T04:32:50 grom358: it will be complicated when your bot logic grows, and I can't imagine how you would verify it visually 2011-10-27T04:33:14 yeah.. i only just started.. just been prototyping some ideas so far 2011-10-27T04:33:19 grom358: and when you will be making any significant changes 2011-10-27T04:33:20 i test with gratuitous amounts of stderr messages... oldskool 2011-10-27T04:33:24 gonna be using jUnit soon 2011-10-27T04:34:09 grom358: this will save you a lot of reading oldschool logs debugging 2011-10-27T04:34:20 so... for you gurus about A*.. I just looking at my old version of it.. which included comment out in.. http://pastebin.com/PyGdn4CG 2011-10-27T04:34:34 grom358: just be sure to design your bot in a way it can be easily unit tested 2011-10-27T04:35:06 why on earth did I ever do that closed.remove(neighbour); ?? on line 66 2011-10-27T04:35:20 who knows 2011-10-27T04:35:21 lol 2011-10-27T04:35:38 it seems to work but lol 2011-10-27T04:35:45 it does 2011-10-27T04:35:52 but it does unecessary stuff as well 2011-10-27T04:36:00 *** skyramp has joined #aichallenge 2011-10-27T04:36:03 the only node that is on the closed list 2011-10-27T04:36:09 that you will ever compare to 2011-10-27T04:36:10 is your parent node 2011-10-27T04:36:14 yeah.. i like the version below cause it follows the one on wikipedia 2011-10-27T04:36:16 and it will always be at a smaller distance 2011-10-27T04:36:41 just was trying to get my head around the old piece of code I had for this A* 2011-10-27T04:38:45 now I can maybe use A* for all my pathing now it fixed.. just not sure about performance doing it that way 2011-10-27T04:39:02 I have some wall hugging thing atm that sort of working now 2011-10-27T04:41:06 the code on wikipedia takes into account things like inconsistent heuristics and cost 2011-10-27T04:41:17 for the ants problem , alot of the heuristics can be assumed 2011-10-27T04:41:32 what u mean? 2011-10-27T04:41:34 *a lot 2011-10-27T04:41:43 he means that you can only move in four directions 2011-10-27T04:41:44 =p 2011-10-27T04:41:51 well that 2011-10-27T04:41:57 and movement cost is constant 2011-10-27T04:42:05 yeah.. I got + 1 in there 2011-10-27T04:42:22 I got +10 so if I want to weight some paths later I can 2011-10-27T04:42:32 without having to work with floats 2011-10-27T04:43:24 how you solved your exploration problem after all bobby? 2011-10-27T04:43:56 part of it was solved with a quadtree 2011-10-27T04:44:30 quad as in quadrants? 2011-10-27T04:44:33 yup 2011-10-27T04:44:41 hmm =) 2011-10-27T04:44:47 oops, secret's out 2011-10-27T04:44:49 lol 2011-10-27T04:45:08 I was thinking of doing it that way tough =p 2011-10-27T04:45:34 just not sure how I would implement it 2011-10-27T04:45:57 *** pedrosorio has joined #aichallenge 2011-10-27T04:47:35 and if I implement a quadtree I want to be able to re-use it for the A* too... 2011-10-27T04:47:39 *** TwistedLogic has joined #aichallenge 2011-10-27T04:48:29 well... i am doing that right now =P 2011-10-27T04:49:21 hi guys, i have a problem playing with the tcpserver. my bot doesn't seem to move yet it plays okay on my local machine. what seems to be the problem? here's what i'm using: python tcpclient.py3 ants.fluxid.pl 2081 "../twisted/TL/bin/Debug/TL" TwistedLogic password 2011-10-27T04:49:26 its only part of the puzzle though, i had to hack the tree quite a bit 2011-10-27T04:51:36 so only took 60ms to calculate the path.. sweet 2011-10-27T04:52:31 thats a lot of time 2011-10-27T04:52:45 you have 500ms per turn 2011-10-27T04:53:08 i thought its 1000ms per turn 2011-10-27T04:53:16 nop 2011-10-27T04:53:22 but u see.. I don't need to A* up every turn either 2011-10-27T04:53:29 only if don't have objective 2011-10-27T04:53:38 yeah, but you don't have the whole map upfront either 2011-10-27T04:53:45 as you discover more walls 2011-10-27T04:53:52 how will you do? 2011-10-27T04:54:16 yeah.. so I assume I can path to it.. and when I get to water.. I recalculate the path then 2011-10-27T04:57:33 *** eflajt has joined #aichallenge 2011-10-27T04:57:41 *** Rainydays has quit IRC (Quit: ChatZilla 0.9.87 [SeaMonkey 2.4.1/20110928161145]) 2011-10-27T04:58:20 *** rj_ has quit IRC (Ping timeout: 265 seconds) 2011-10-27T04:59:28 I want to thank you guys so much.. I could of spent in more time wasted not realising it was TreeSet causing it 2011-10-27T04:59:41 all your thanks goes to sir_macelon 2011-10-27T05:00:08 I really could not have helped you after all, as I do not know how the structures in java behave 2011-10-27T05:00:10 =p 2011-10-27T05:00:31 i haven't touched Java in ages 2011-10-27T05:00:41 out of the loop on it 2011-10-27T05:01:12 why did you bring it back from the dead then? 2011-10-27T05:01:28 *** ikaros has joined #aichallenge 2011-10-27T05:01:45 a mate is doing it in Java as well.. 2011-10-27T05:02:10 oh =) 2011-10-27T05:02:16 well, good luck 2011-10-27T05:02:28 i just want my bot to bet his lol 2011-10-27T05:02:33 beat* 2011-10-27T05:02:37 oh yeh... 2011-10-27T05:02:41 2nd on the ranking 2011-10-27T05:02:47 gonna beat him for sure 2011-10-27T05:02:47 :p 2011-10-27T05:03:07 cause my A* wasn't working you should see this other one I wrote lol 2011-10-27T05:03:08 *** onensora has joined #aichallenge 2011-10-27T05:03:40 each ant keeps track of visited nodes.. and just traces every tile basically 2011-10-27T05:03:51 like real ants then 2011-10-27T05:03:51 haha 2011-10-27T05:04:25 yeah.. but cause I didn't have proper pathing it only by chance kills enemy hills 2011-10-27T05:04:47 it has no logic.. it just walks around 2011-10-27T05:05:02 see, you was the better ant after all 2011-10-27T05:05:06 least it given me some ideas at lest.. wasn't a waste 2011-10-27T05:05:09 no one could distinguish it from a real ant 2011-10-27T05:05:34 i did it cause my first stab kept timing out 2011-10-27T05:05:46 with java? woah 2011-10-27T05:05:51 what were you ding? 2011-10-27T05:05:52 doing* 2011-10-27T05:05:55 yeah. cause I had broken A* in it 2011-10-27T05:06:16 *** smiley1983 has joined #aichallenge 2011-10-27T05:06:27 and it recalculate the path every turn too.. which I gonna fix 2011-10-27T05:06:37 I'm just in this for fun, and to learn a thing or two =p 2011-10-27T05:06:38 greetings everyone. I'm having some trouble with the java visualizer. Somebody else already described the problem here: http://aichallenge.org/forums/viewtopic.php?f=25&t=1630 . Does anyone know a solution? 2011-10-27T05:08:00 *** Vivitron has quit IRC (Ping timeout: 240 seconds) 2011-10-27T05:08:00 *** sir_macelon has quit IRC (Ping timeout: 265 seconds) 2011-10-27T05:08:01 *** g0llum has joined #aichallenge 2011-10-27T05:08:34 Draivin: same 2011-10-27T05:08:47 I always wanted to get into AI.. so starting with this challenge 2011-10-27T05:09:33 well, they are wanting to use a rts for next year, that is totally out of my league tough =p 2011-10-27T05:10:10 well.. the Ant thing is like a small RTS 2011-10-27T05:10:20 it's not real time to begin with 2011-10-27T05:10:27 and only has one unity time 2011-10-27T05:10:28 type* 2011-10-27T05:10:28 oh..lol yeah except that 2011-10-27T05:10:56 yeah.. real time makes it heaps harder 2011-10-27T05:11:51 *** TwistedLogic has quit IRC (Quit: Page closed) 2011-10-27T05:12:15 bobby going crazy there where I think how I would do the quadtree 2011-10-27T05:12:16 blergh 2011-10-27T05:12:31 while* 2011-10-27T05:12:47 *** mithos28 has quit IRC (Quit: mithos28) 2011-10-27T05:13:27 yeh? what's you thinking 2011-10-27T05:13:33 i don't mind sharing 2011-10-27T05:14:11 nah, just messing with you, are you changing the pathfind to make use of it? 2011-10-27T05:14:32 the ant's challenge is sort of real time, there is a quite low time limit anyway 2011-10-27T05:14:35 *** TheDigitalNinja has quit IRC (Remote host closed the connection) 2011-10-27T05:15:56 well, it does seem easier for me than a red alert AI 2011-10-27T05:17:38 *** sigh has joined #aichallenge 2011-10-27T05:17:43 *** ztfw has joined #aichallenge 2011-10-27T05:19:16 yeh i'm trying 2011-10-27T05:19:17 lol 2011-10-27T05:19:43 how would you go about determining if a quad is walkable or not? 2011-10-27T05:20:20 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-10-27T05:23:08 *** skyramp has quit IRC (Quit: Page closed) 2011-10-27T05:23:12 *** GeorgeJ has quit IRC (Remote host closed the connection) 2011-10-27T05:23:19 its just a flag 2011-10-27T05:23:33 obstacles get added to the tree 2011-10-27T05:25:21 bobbydroptable : so if a quad has a single obstacle it will traverse the whole quad normally in the pathfinding? 2011-10-27T05:25:34 *** GeorgeJ has joined #aichallenge 2011-10-27T05:25:34 *** GeorgeJ has joined #aichallenge 2011-10-27T05:26:15 mmm at the moment 2011-10-27T05:26:20 when i add an obstacle 2011-10-27T05:26:32 i partition the quadrant down to the lowest possible sisze 2011-10-27T05:26:39 1x1 2011-10-27T05:26:47 for that obstacle 2011-10-27T05:26:48 *** ananda has joined #aichallenge 2011-10-27T05:27:16 May I know where the old rankings page for Planetwars - AI challenge 2010 - is hosted? I see it removed from ai-contest.com 2011-10-27T05:27:18 i was considering a threshold with a % of obstacles to size of quadrant 2011-10-27T05:27:33 but decided to concentrate on getting ants moving first 2011-10-27T05:27:46 ananda no idea, sorry 2011-10-27T05:27:58 bobby, oh okay. 2011-10-27T05:29:38 *** ananda has quit IRC (Client Quit) 2011-10-27T05:32:18 *** sdf has quit IRC (Quit: Leaving.) 2011-10-27T05:33:19 *** boegel has joined #aichallenge 2011-10-27T05:35:29 *** craklyn has joined #aichallenge 2011-10-27T05:39:17 *** kaemo has joined #aichallenge 2011-10-27T05:43:03 question.. does the engine tell u there an enemy hill in vision if its been razed? 2011-10-27T05:44:52 nope 2011-10-27T05:47:04 *** nishant has left #aichallenge ("Ex-Chat") 2011-10-27T05:47:23 *** skunx has joined #aichallenge 2011-10-27T05:50:53 *** onensora has quit IRC () 2011-10-27T05:51:41 *** grandmaster has joined #aichallenge 2011-10-27T05:52:32 *** craklyn has quit IRC (Quit: Page closed) 2011-10-27T05:54:28 *** Murashka has joined #aichallenge 2011-10-27T05:57:42 *** mviel has quit IRC (Ping timeout: 260 seconds) 2011-10-27T05:59:26 *** replore__ has quit IRC (Remote host closed the connection) 2011-10-27T06:01:07 probably wrong channel.. but can u do foreach loop in java on the hashMap values and remove entries? 2011-10-27T06:01:25 I got HashMap 2011-10-27T06:01:44 so I can track when my ants die 2011-10-27T06:02:37 mmm.. why am I doing that.. could just do ants.getMyAnts().contains(ant.getLocation()); // stupid me 2011-10-27T06:03:13 *** cyphase has joined #aichallenge 2011-10-27T06:04:53 oh.. it was to find new ants.. now I remember 2011-10-27T06:06:53 hey grom 2011-10-27T06:07:18 you can just find you ants by location when you receive a "d" with owner 0 2011-10-27T06:07:28 and remove it from the list 2011-10-27T06:07:46 *** master_ninja has quit IRC (Quit: Page closed) 2011-10-27T06:07:48 yeah.. got that.. but what about new ants 2011-10-27T06:08:08 I have data structure I keep for each ant.. need to create it for my new ants 2011-10-27T06:08:16 well, I have a ant set with locations as index 2011-10-27T06:08:26 one I receive a ant 2011-10-27T06:08:31 I check if in my ant set 2011-10-27T06:08:34 there is one with that index 2011-10-27T06:08:43 it there is not I create a new ant 2011-10-27T06:08:57 yeah I keep the location of the ant too.. but can't look up my set of ants by location 2011-10-27T06:08:58 once* 2011-10-27T06:09:26 why not? 2011-10-27T06:09:46 can't you have location as index or strings as index? 2011-10-27T06:10:06 well can't figure out how in Java.. cause I got Tile .. and my ants in Set 2011-10-27T06:10:48 hashmaps? 2011-10-27T06:11:05 *** mviel has joined #aichallenge 2011-10-27T06:12:07 yeah.. that is what doing.. but I got to keep removing the entry and reinserting it 2011-10-27T06:12:55 just when you make a move, don't see a big deal in that, better than iterate through you whole array when you want a specific ant I guess 2011-10-27T06:12:59 you could even use simple 2d array since max size of maps is 200*200 and thats few memory 2011-10-27T06:14:18 sigh so difficult to find neighbours in a quadtree 2011-10-27T06:14:19 lol 2011-10-27T06:14:27 yeah 2011-10-27T06:14:32 was going to ask you about that 2011-10-27T06:14:35 I got stuck on that part 2011-10-27T06:14:36 hahaha 2011-10-27T06:14:44 lol 2011-10-27T06:14:52 thinking, or have you alerady done 2011-10-27T06:14:52 it 2011-10-27T06:14:57 thinking 2011-10-27T06:14:59 =) 2011-10-27T06:14:59 cuz that'd make me feel useless 2011-10-27T06:15:01 lol 2011-10-27T06:15:06 i've been coding this for 2 days already 2011-10-27T06:15:08 just this 2011-10-27T06:15:11 ==" 2011-10-27T06:15:52 the worst problem 2011-10-27T06:15:58 is that a quad can have multiple neighbours 2011-10-27T06:15:59 on one side 2011-10-27T06:16:13 exactly!!! *cries* 2011-10-27T06:16:26 i have an idea but it puts the UG in UGLY 2011-10-27T06:16:40 what? 2011-10-27T06:17:20 put everything into a 2d array 2011-10-27T06:17:21 lol 2011-10-27T06:17:24 well 2011-10-27T06:17:28 I have another idea 2011-10-27T06:17:31 which is? 2011-10-27T06:17:32 :p 2011-10-27T06:17:57 there could be a search by direction function in the quads 2011-10-27T06:18:05 bobbydroptable: pathfinging is fun :D 2011-10-27T06:18:06 so when you want to know the neighbours in a quad 2011-10-27T06:18:15 the top neighbours for example 2011-10-27T06:18:23 Do any of you store your map as layers? 2011-10-27T06:18:33 you would pass 'bottom' to that function 2011-10-27T06:18:40 of the quad that is on top 2011-10-27T06:18:41 the big one 2011-10-27T06:18:47 and it would iterate through the sub quads 2011-10-27T06:18:48 *** nishant has joined #aichallenge 2011-10-27T06:18:49 GeorgeJ: maybe maybe not. depends on how i need the data 2011-10-27T06:18:52 and get the bottom ones 2011-10-27T06:19:00 doesn't work for all cases 2011-10-27T06:19:01 *** Areks has quit IRC (Ping timeout: 240 seconds) 2011-10-27T06:19:10 sometimes a neighbour is a branch off a grandparent 2011-10-27T06:19:13 was thinking about that too 2011-10-27T06:19:14 or great grandparent 2011-10-27T06:19:23 if i have some pathfinder with nodes, I would make the nodes a separate layer from other data 2011-10-27T06:19:26 you could do a reverse tree lookup 2011-10-27T06:19:49 hey guys at the tutorial step 4, on implementing th code , i get duplicate orders when i run the test program 2011-10-27T06:19:52 but you would have to look for every possible neighbour? 2011-10-27T06:19:56 GeorgeJ: At least I can say that a single interger for a square is a bad idea 2011-10-27T06:19:57 and filter for duplicates 2011-10-27T06:20:00 any suggesions 2011-10-27T06:20:26 sets will do the filtering 2011-10-27T06:20:47 its the lookup that is bothering me at the moment 2011-10-27T06:20:48 lol 2011-10-27T06:20:53 yeah 2011-10-27T06:20:56 get every possible neighbour 2011-10-27T06:20:57 is a pain 2011-10-27T06:21:06 that varies with the size of the quad 2011-10-27T06:21:06 haha 2011-10-27T06:21:10 yeh 2011-10-27T06:21:22 but with yours, you'd have to check every node to test that they share edges 2011-10-27T06:21:53 aaaaaaargh 2011-10-27T06:22:02 going to bang my head on the wall a little more 2011-10-27T06:22:03 brb 2011-10-27T06:22:15 mleise: How so? I find that chars are enough for my map 2011-10-27T06:22:15 lol 2011-10-27T06:22:16 ok 2011-10-27T06:22:59 GeorgeJ: I want to store what I've already explored; just as an exaple 2011-10-27T06:23:05 *** nishant has quit IRC (Client Quit) 2011-10-27T06:23:18 And you need more than an int for that? 2011-10-27T06:23:36 With just one char per tile, you can have 8 layers 2011-10-27T06:24:15 *** Gussi has left #aichallenge ("WeeChat 0.3.2") 2011-10-27T06:24:20 Draivin : lets not even get started on the WRAPPING ==" 2011-10-27T06:24:25 OH FUCK 2011-10-27T06:24:28 forgot about that 2011-10-27T06:24:29 completely 2011-10-27T06:25:01 GeorgeJ: ok, I may want to store if it is visible, has ever been seen, has food, has ant+which color, has hill+which color, list of dead ants on the square, is water, how many turns this square has been unseen by my ants ... 2011-10-27T06:25:02 *** olexs has joined #aichallenge 2011-10-27T06:25:49 *** Murashka has quit IRC (Ping timeout: 265 seconds) 2011-10-27T06:25:52 GeorgeJ: some of that is just there to generate some data, that can be used by algorithms 2011-10-27T06:26:23 mleise: Well, ant_layer, enemy_layer, food_layer, hill_layer, enemy_hill_layer, water_layer, fog_layer 2011-10-27T06:26:33 I find it annoying to always switch between writing an algorithm and writing code that generates data for it to operate on :) 2011-10-27T06:26:36 But yeah, not as complex as what you are trying to achieve 2011-10-27T06:27:41 GeorgeJ: So you have practically more than a char, too. "Layers" or a single struct.. it comes down to the same in the end. 2011-10-27T06:27:57 mleise: Well, not really, a "layer" is a bit 2011-10-27T06:28:20 Draivin: glad to be of service lol 2011-10-27T06:28:21 oh so you don't care about which player an ant belongs to 2011-10-27T06:28:40 Not atm, no. But you raise a valid point 2011-10-27T06:28:48 There's no point in trying to conserve memory 2011-10-27T06:28:58 no absolutely not :) 2011-10-27T06:29:08 unless you want it all to fit into the CPU level 1 cache 2011-10-27T06:29:46 Well, tbh. Having stored maps as such, would make it rather easy for the pathfinding algo 2011-10-27T06:29:58 When you see 2 enemy parties fight, you can wait till they kill each other for example. 2011-10-27T06:31:44 Using (map[row*cols + col] >> layer) & 0x01 to check if I have an entity would be easier than map[row][col].is_water != 0 2011-10-27T06:32:07 Well, not easier, faster perhaps 2011-10-27T06:33:38 Then have a sepparate structure to keep all data needed by your algos perhaps 2011-10-27T06:34:02 *** UncleVasya has joined #aichallenge 2011-10-27T06:37:52 *** Areks has joined #aichallenge 2011-10-27T06:40:32 *** kiv has quit IRC (Quit: Leaving) 2011-10-27T06:45:51 *** jasperduba has joined #aichallenge 2011-10-27T06:46:40 why does my rank drop while i am waiting for the server to use my bot in a game 2011-10-27T06:47:29 *** stislav has joined #aichallenge 2011-10-27T06:48:52 jasperduba: because other bots' skill increases 2011-10-27T06:49:08 *** sdf has joined #aichallenge 2011-10-27T06:49:35 so it seems like a good strategy to move up ranks, is to never upload new code and just do it once 2011-10-27T06:50:04 so you don't drop to last place while you wait for the server to run your code 2011-10-27T06:50:28 jasperduba: that will change nothing for the final results 2011-10-27T06:50:42 ok maybe im confused then :) 2011-10-27T06:50:46 i was looking at my rank # 2011-10-27T06:51:02 jasperduba: the current ranking has nothing to do with the contest finals 2011-10-27T06:51:11 ahh ok 2011-10-27T06:51:17 it's just a kind of indicator 2011-10-27T06:51:34 so it doesn't matter when your rank drops after an upload 2011-10-27T06:51:55 but if you want a realistic rank for your current code, yeah you should wait a while after uploading 2011-10-27T06:52:37 *** skyramp has joined #aichallenge 2011-10-27T06:54:26 cool thanks for the info! 2011-10-27T06:54:42 looks like i am still getting a timeout, i need faster path finding :D 2011-10-27T06:54:43 http://aichallenge.org/visualizer.php?game=19563&user=816 2011-10-27T06:56:14 *** eflajt has quit IRC (Quit: Page closed) 2011-10-27T06:58:26 when does the challenge finish? 2011-10-27T07:00:36 the current phase finishes dec 18 2011-10-27T07:00:43 after that they will progress to finals, i presume 2011-10-27T07:02:53 *** kaemo has quit IRC (Ping timeout: 245 seconds) 2011-10-27T07:03:31 *** pedrosorio has quit IRC (Ping timeout: 265 seconds) 2011-10-27T07:03:52 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-10-27T07:05:18 *** nek has joined #aichallenge 2011-10-27T07:07:03 *** alc has quit IRC (Ping timeout: 245 seconds) 2011-10-27T07:10:36 i hope there will be some further server updates.. with the current state it's impossible to compare new versions to the field because it takes so long until a solid ranking is available 2011-10-27T07:11:42 tcp servers is your best bet 2011-10-27T07:12:26 yea i use it but the range of opponents alot smaller 2011-10-27T07:12:35 +is 2011-10-27T07:16:56 but you probably get higher calibre opponents too because most people will be doing the same things 2011-10-27T07:16:58 i think 2011-10-27T07:17:05 they'll be the ones with the weird quirky bots 2011-10-27T07:17:09 yea true 2011-10-27T07:17:29 *** arx has joined #aichallenge 2011-10-27T07:18:39 *** wer has joined #aichallenge 2011-10-27T07:18:46 test 2011-10-27T07:19:01 *** SharkMonkey has quit IRC (Ping timeout: 240 seconds) 2011-10-27T07:19:36 *** nek has quit IRC (Ping timeout: 258 seconds) 2011-10-27T07:19:46 Can you please add more gaming servers? 2011-10-27T07:24:18 *** wer has quit IRC (Ping timeout: 265 seconds) 2011-10-27T07:25:41 *** olexs1 has joined #aichallenge 2011-10-27T07:29:56 *** nnbot has joined #aichallenge 2011-10-27T07:31:11 *** fijiol has joined #aichallenge 2011-10-27T07:31:45 *** nnbot has quit IRC (Client Quit) 2011-10-27T07:32:44 *** stislav has quit IRC (Quit: Page closed) 2011-10-27T07:33:51 *** stislav has joined #aichallenge 2011-10-27T07:34:49 *** nielsrobin has joined #aichallenge 2011-10-27T07:36:18 Hi guys, I seem to be slamming my head against the wall here. Did i miss something, or is there no way of telling if a square is seen or unseen? 2011-10-27T07:36:29 there is 2011-10-27T07:36:36 unseen squares are marked ? 2011-10-27T07:36:45 nope you have to keep track of them yourself 2011-10-27T07:37:14 on setup, initialise all the squares as unmarked, then go through the vision and mark squares you see as marked 2011-10-27T07:38:00 Ok, i'm using the C# version, where, as far as i know there is no "get tile type" on a location. so where does this "?" come into play? in the game state? 2011-10-27T07:38:07 *** olexs has quit IRC (Quit: Leaving.) 2011-10-27T07:38:18 this is just a starting kit 2011-10-27T07:38:25 you should read the documentation 2011-10-27T07:38:27 look at the code where it reads w for water 2011-10-27T07:38:35 to see what input the server gives to you 2011-10-27T07:38:39 python starter has visible function 2011-10-27T07:38:41 and parse it properly 2011-10-27T07:38:54 that does the same as what bobbydroptable said 2011-10-27T07:39:59 alright, thanks i'll have a look at the function that keeps track of visible food and see if i can build a "unseen tiles" list using the same method 2011-10-27T07:45:25 ok, from looking at the code, it looks like i would have to cycle through every location on the map and figure out if it is seen or unseen by the distance to the nearest ant, does it sound like i'm on the right track? 2011-10-27T07:47:04 you should look at the tutorial 2011-10-27T07:47:08 thats how they do it 2011-10-27T07:47:12 *** stislav has joined #aichallenge 2011-10-27T07:48:46 Draivin: Ok thanks, good to know i'm not completely off, will have a look 2011-10-27T07:50:24 *** skyramp has quit IRC (Ping timeout: 265 seconds) 2011-10-27T07:52:55 ah, back to top 100 2011-10-27T07:55:14 *** reblue has quit IRC (Ping timeout: 265 seconds) 2011-10-27T07:56:23 *** goffrie__ has joined #aichallenge 2011-10-27T07:56:57 *** goffrie_ has quit IRC (Ping timeout: 255 seconds) 2011-10-27T07:58:24 *** olexs has joined #aichallenge 2011-10-27T07:59:13 Did the vision range of ants change? 2011-10-27T07:59:45 The tutorial videos seem to have ants with smaller vision that my local visualizer 2011-10-27T08:00:05 Garf: The vision range is specified at the start of every game. 2011-10-27T08:01:41 then what is this? http://aichallenge.org/game_settings.php 2011-10-27T08:02:20 *** yoden has joined #aichallenge 2011-10-27T08:02:37 the vision is now 77 2011-10-27T08:02:38 used to be 55 2011-10-27T08:03:27 ok, so the tutorial isn't exactly reproducible any more 2011-10-27T08:04:03 because it doesnt specify vision range in the map 2011-10-27T08:04:17 just use 55 when running 2011-10-27T08:05:00 *** tak0_n has joined #aichallenge 2011-10-27T08:06:30 *** UncleVasya has joined #aichallenge 2011-10-27T08:09:19 *** stislav has quit IRC (Quit: Page closed) 2011-10-27T08:10:18 *** george has joined #aichallenge 2011-10-27T08:11:39 just because the vision is a square larger doesn't make the tutorial any less valid lol 2011-10-27T08:12:18 thats not what I said 2011-10-27T08:13:27 any luck bobby? 2011-10-27T08:13:36 *** cyphase has quit IRC (Ping timeout: 240 seconds) 2011-10-27T08:16:26 *** kaemo has joined #aichallenge 2011-10-27T08:16:34 (Both HunterBot and the tutorial bot we will be making are deterministic, meaning if you give them the same input, they will produce the same set of moves. This means if you follow along exactly, you should see the tutorial games play out exactly the same on your machine.) 2011-10-27T08:16:47 My point was that the above is now probably no longer true. 2011-10-27T08:18:55 *** tak0_n has quit IRC (Ping timeout: 265 seconds) 2011-10-27T08:19:49 *** onensora has joined #aichallenge 2011-10-27T08:19:50 *** dwins has joined #aichallenge 2011-10-27T08:20:29 *** bigpete591 has joined #aichallenge 2011-10-27T08:21:43 Draivin: not much my array is doing weird things 2011-10-27T08:21:49 *** jasperduba has quit IRC (Ping timeout: 265 seconds) 2011-10-27T08:24:38 *** AGrishchenko has joined #aichallenge 2011-10-27T08:25:32 *** iris1 has quit IRC (Ping timeout: 260 seconds) 2011-10-27T08:27:23 *** cyphase has joined #aichallenge 2011-10-27T08:30:25 *** maran_ has quit IRC (Quit: Leaving) 2011-10-27T08:33:06 *** UncleVasya has quit IRC (Read error: Connection reset by peer) 2011-10-27T08:33:15 *** replore_ has joined #aichallenge 2011-10-27T08:35:27 *** skunx has quit IRC (Quit: leaving) 2011-10-27T08:42:24 *** mcstar has joined #aichallenge 2011-10-27T08:43:46 *** coramoor has joined #aichallenge 2011-10-27T08:46:44 *** invader has joined #aichallenge 2011-10-27T08:47:36 *** Harpyon_ has joined #aichallenge 2011-10-27T08:52:33 *** Harpyon_ has quit IRC (Remote host closed the connection) 2011-10-27T08:57:04 *** michaelpjohnson has quit IRC (Quit: Leaving) 2011-10-27T08:58:43 *** dubajj has joined #aichallenge 2011-10-27T08:58:52 *** paulwal has joined #aichallenge 2011-10-27T08:59:19 sup 2011-10-27T09:00:27 *** iris1 has joined #aichallenge 2011-10-27T09:02:06 *** dfgdfg has joined #aichallenge 2011-10-27T09:02:25 *** he_the_great has joined #aichallenge 2011-10-27T09:02:25 *** bobbydroptable has quit IRC (Ping timeout: 265 seconds) 2011-10-27T09:02:40 does anyone have a list of active tcp servers? thanks! 2011-10-27T09:05:34 i also would be interested in tcp servers 2011-10-27T09:05:46 342 minutes until my bot runs? 2011-10-27T09:06:22 *** avdg has joined #aichallenge 2011-10-27T09:07:18 can anybody point me to information on running my own tcp server for this? 2011-10-27T09:07:44 *** fijiol has quit IRC (Ping timeout: 265 seconds) 2011-10-27T09:09:26 *** amadoru has joined #aichallenge 2011-10-27T09:09:52 *** pgoldbr has joined #aichallenge 2011-10-27T09:11:47 *** sdf1 has joined #aichallenge 2011-10-27T09:11:52 *** tncardoso has joined #aichallenge 2011-10-27T09:13:40 *** sdf has quit IRC (Ping timeout: 260 seconds) 2011-10-27T09:15:56 *** bearoff_w has joined #aichallenge 2011-10-27T09:17:24 *** mellodroit has quit IRC (Ping timeout: 265 seconds) 2011-10-27T09:19:25 ants.fluxid.pl is one, i think 2011-10-27T09:20:11 *** RobotCaleb has joined #aichallenge 2011-10-27T09:22:27 is it embarrassing that this google sponsored contest can barely run on its servers, so people have to flock to other machines? 2011-10-27T09:22:35 seems embarrassing 2011-10-27T09:24:01 it sucks, i agree 2011-10-27T09:24:40 the slow server sucks. the contest is awesome 2011-10-27T09:24:47 yep 2011-10-27T09:26:22 thank you! 2011-10-27T09:26:29 *** Kingpin13 has joined #aichallenge 2011-10-27T09:27:59 *** v_paul_v has quit IRC (Remote host closed the connection) 2011-10-27T09:28:47 *** sdf1 has quit IRC (Quit: Leaving.) 2011-10-27T09:31:27 *** guest___ has joined #aichallenge 2011-10-27T09:32:45 *** mcstar has left #aichallenge ("WeeChat 0.3.5") 2011-10-27T09:36:44 *** paulwal has quit IRC (Ping timeout: 265 seconds) 2011-10-27T09:37:53 *** bigpete591 has quit IRC (Quit: Page closed) 2011-10-27T09:37:59 *** ajhager has quit IRC (Ping timeout: 258 seconds) 2011-10-27T09:42:26 iris1, accoun is runnung another on 213.88.39.97, and ash0d on ash.webfactional.com 2011-10-27T09:43:01 *** amadoru has quit IRC (Ping timeout: 265 seconds) 2011-10-27T09:46:41 cool, thanks! 2011-10-27T09:47:46 see ya there ! 2011-10-27T09:47:47 *** mviel has quit IRC (Ping timeout: 260 seconds) 2011-10-27T09:48:12 *** mviel has joined #aichallenge 2011-10-27T09:48:14 *** sunshowers has joined #aichallenge 2011-10-27T09:49:15 *** paulwal has joined #aichallenge 2011-10-27T09:52:23 *** RobotCaleb has quit IRC (Quit: Leaving.) 2011-10-27T10:03:25 *** sunshowers has quit IRC (Quit: Page closed) 2011-10-27T10:04:14 yes. your bot is good!! 2011-10-27T10:04:44 How is skill determined? 2011-10-27T10:07:13 *** tncardoso has quit IRC (Quit: bye) 2011-10-27T10:08:29 en.wikipedia.org/wiki/TrueSkill 2011-10-27T10:09:18 *** sdf has joined #aichallenge 2011-10-27T10:11:10 will parameters such as the ant's line of sight radius2 change a lot? or can we assume them to stay in the same ballpark? 2011-10-27T10:11:32 (if it changes to 1 for instance, a quite different type of AI will be needed) 2011-10-27T10:12:39 *** cbad has quit IRC (Read error: Connection reset by peer) 2011-10-27T10:12:49 I don't think there would be that many changes. 2011-10-27T10:13:08 *** cbad has joined #aichallenge 2011-10-27T10:13:08 *** cbad has joined #aichallenge 2011-10-27T10:13:09 *** sigh has quit IRC (Remote host closed the connection) 2011-10-27T10:13:28 I hope there won't - it will require a far more flexible AI if these parameters is to be expected to vary 2011-10-27T10:13:41 *** sigh has joined #aichallenge 2011-10-27T10:14:05 *** RobotCaleb has joined #aichallenge 2011-10-27T10:15:17 *** dubajj has quit IRC (Quit: Page closed) 2011-10-27T10:17:04 *** dd_123 has joined #aichallenge 2011-10-27T10:17:59 *** sigh has quit IRC (Ping timeout: 244 seconds) 2011-10-27T10:18:00 *** foRei has joined #aichallenge 2011-10-27T10:23:43 *** dd_123 has quit IRC (Quit: Page closed) 2011-10-27T10:24:13 *** pgoldbr_ has joined #aichallenge 2011-10-27T10:24:19 *** AGrishchenko has quit IRC (Quit: Page closed) 2011-10-27T10:24:35 *** pgoldbr has quit IRC (Ping timeout: 265 seconds) 2011-10-27T10:26:40 *** aerique has quit IRC (Quit: ...) 2011-10-27T10:27:35 *** sdf has quit IRC (Quit: Leaving.) 2011-10-27T10:30:57 *** GarfTop has joined #aichallenge 2011-10-27T10:30:57 *** ajhager has joined #aichallenge 2011-10-27T10:31:03 *** mviel has quit IRC (Read error: No route to host) 2011-10-27T10:33:08 *** mviel has joined #aichallenge 2011-10-27T10:34:09 *** Garf has quit IRC (Ping timeout: 265 seconds) 2011-10-27T10:35:11 *** dauryg has joined #aichallenge 2011-10-27T10:36:51 *** ajhager has quit IRC (Quit: ajhager) 2011-10-27T10:37:54 *** ajhager has joined #aichallenge 2011-10-27T10:39:43 *** oddmunds has quit IRC (Ping timeout: 252 seconds) 2011-10-27T10:40:43 *** upgrayeddd has quit IRC (Read error: Connection reset by peer) 2011-10-27T10:41:31 *** chaudeau has joined #aichallenge 2011-10-27T10:42:06 *** Knekkebjoern has quit IRC (Quit: Leaving.) 2011-10-27T10:42:52 My pathfinder does ~1300 paths per 500ms. Is that good or bad? 2011-10-27T10:43:16 depends on what paths I guess 2011-10-27T10:43:26 *** pgoldbr_ has quit IRC (Ping timeout: 265 seconds) 2011-10-27T10:44:24 *** paulwal has quit IRC (Ping timeout: 265 seconds) 2011-10-27T10:45:48 *** he_the_great has quit IRC (Ping timeout: 245 seconds) 2011-10-27T10:48:53 *** RobotCaleb has quit IRC (Quit: Leaving.) 2011-10-27T10:49:35 *** moises has joined #aichallenge 2011-10-27T10:50:00 An average over all the hill to hill paths in the maps that come with the tools 2011-10-27T10:50:53 *** RobotCaleb has joined #aichallenge 2011-10-27T10:50:56 Does anyone use cross-product-tie-breakers? It looks like it slows me down on average :( 2011-10-27T10:51:27 *** jcdny has quit IRC (Ping timeout: 252 seconds) 2011-10-27T10:51:57 *** ztfw has quit IRC (Ping timeout: 260 seconds) 2011-10-27T10:53:29 *** jcdny has joined #aichallenge 2011-10-27T10:54:04 *** coramoor has quit IRC (Ping timeout: 265 seconds) 2011-10-27T10:56:29 If you do use one, remember that you need to consider the map wrapping, otherwise the vectors will point in the wrong direction and A* will search the wrong tiles first. 2011-10-27T10:57:30 *** mithos28 has joined #aichallenge 2011-10-27T10:57:53 *** analyst74 has joined #aichallenge 2011-10-27T10:58:25 mleise, does it need to do 1300 paths per second? or are you just testing it's performance? 2011-10-27T10:58:39 per 500ms* 2011-10-27T11:00:12 *** mithos28 has quit IRC (Client Quit) 2011-10-27T11:00:18 cyphase: I am testing, but I want to know if that is a good number 2011-10-27T11:00:47 i don't know, i was just wondering if you actually had 1300 paths to compute 2011-10-27T11:00:48 if you have to calculate more than 1000 paths each step, then its probably not enough ;-) 2011-10-27T11:00:58 It sounds ok to me, but not like I'd say "hurray!" 2011-10-27T11:02:55 I think it should be possible to exploit the fact that the world is a subgraph of a regular grid to get much better performance than with a general A* 2011-10-27T11:02:59 is there a situation when you need to calculate more paths than your ants number (if it's not paths cashing)? 2011-10-27T11:03:02 I don't know yet how I would handle ants blocking each other or how to assign ants to their targets. It may mean that I want to calculate alternative paths. 2011-10-27T11:04:33 you can recalculate blocked ants when it happens 2011-10-27T11:06:10 I could well run into timing issues though. Is there anyone with a C++ or C bot that took the time of his path finding algorithm? 2011-10-27T11:07:04 *** upgrayeddd has joined #aichallenge 2011-10-27T11:07:26 *** Ishuah has joined #aichallenge 2011-10-27T11:08:11 hey 2011-10-27T11:08:39 anyone whose game ranking is > 50 2011-10-27T11:08:59 @top50 2011-10-27T11:09:00 *** analyst74 has left #aichallenge 2011-10-27T11:09:00 mleise: I do not know about 'top50', but I do know about these similar topics: 'topic' 2011-10-27T11:09:49 yesterday I was in 50 2011-10-27T11:10:05 before submitted new version ( 2011-10-27T11:12:48 @bearoff_w what language are you coding in? 2011-10-27T11:12:49 Ishuah: An error occured while trying to show the previous error. 2011-10-27T11:13:10 An error occured while trying to show the previous error. - Bingo! 0 2011-10-27T11:13:19 PHP :) 2011-10-27T11:13:25 *** Areks has quit IRC (Ping timeout: 240 seconds) 2011-10-27T11:14:26 now my average turntime on 200 ants is near 0.05..0.1 2011-10-27T11:14:47 *** Knekkebjoern has joined #aichallenge 2011-10-27T11:15:02 not all heavy logic realized still, but there are reserves to optimize too 2011-10-27T11:15:05 bearoff_w cool. I'm doing it in java 2011-10-27T11:15:44 it's kind of curiosity - how fast can me PHP bot be comparing to Java and C ) 2011-10-27T11:15:51 *my PHP 2011-10-27T11:16:00 I've just looked at a game with 460 ants for one player. That must be really heavy on pathfinding. Poor bearoff_w :p 2011-10-27T11:16:19 *** oddmunds has joined #aichallenge 2011-10-27T11:16:24 I already heared that php 5.4 is around 30% faster in basic operations 2011-10-27T11:16:52 bearoff_w that would be an interesting experiment 2011-10-27T11:16:59 I have 340 ants maximum, there was not problems - as not all paths are finded at one turn 2011-10-27T11:17:22 oh and if you have more, just block the hill :-) 2011-10-27T11:17:47 avdg yeah. Thats my plan 2011-10-27T11:18:13 my ants now are stupids in hill blocking ) 2011-10-27T11:18:17 *** Areks has joined #aichallenge 2011-10-27T11:18:43 bearoff_w: ah! you split the path finding over several turns? 2011-10-27T11:19:36 of course 2011-10-27T11:19:57 no need to recalculate the path you just found any turn 2011-10-27T11:21:34 bearoff_w: sounds like you calculate the path for all ants every turn that don't already have a path. sounds good. but what happens when you discover a water square? 2011-10-27T11:22:01 you are right 2011-10-27T11:22:15 if water hapenns, i simply recalculate path in this case 2011-10-27T11:22:17 suddenly there is a dead end in a maze! 2011-10-27T11:22:27 yes, it happens 2011-10-27T11:22:45 I plan to use free time to find such situations 2011-10-27T11:22:53 so you recalculate when you run against the water, or do you delete all paths when the map 'changes' 2011-10-27T11:23:18 but I see there is not a very needed task 2011-10-27T11:23:30 no, only affected 2011-10-27T11:23:52 especially for you I added some debug output: 2011-10-27T11:24:01 [27-Oct-2011 18:21:02] ----------------TURN 707 -------------- 2011-10-27T11:24:03 [27-Oct-2011 18:21:02] ants count: 171 2011-10-27T11:24:06 [27-Oct-2011 18:21:02] a searches: 23 2011-10-27T11:24:10 [27-Oct-2011 18:21:02] TURN TOTAL: 0.077973127365112 2011-10-27T11:24:28 77 ms ??? 2011-10-27T11:24:39 *** lavalamp has joined #aichallenge 2011-10-27T11:24:52 it's low or high value? 2011-10-27T11:25:01 *** kaemo has quit IRC (Read error: Operation timed out) 2011-10-27T11:25:09 *** delt0r__ has quit IRC (Ping timeout: 244 seconds) 2011-10-27T11:25:15 I think this is low, considering it is PHP :) 2011-10-27T11:25:20 a 2011-10-27T11:25:54 and I consider my code as not REALLY optimized ) 2011-10-27T11:26:06 see, there are only 23 a* searches 2011-10-27T11:26:16 and they are not throw all-the-map ) 2011-10-27T11:27:12 that makes ~13% new paths for your ants. is this a typical ratio? 2011-10-27T11:27:27 *** ztfw has joined #aichallenge 2011-10-27T11:28:00 yes, about this. may be 25% is more reliable - I saw this logs in first time too ) 2011-10-27T11:28:46 and this depends on food spawning rate 2011-10-27T11:29:01 as I need to reorder some ants to closest food 2011-10-27T11:29:59 http://aichallenge.org/visualizer.php?game=19781&user=4513 2011-10-27T11:30:04 does the rank really stabilise there? 2011-10-27T11:30:07 sure, that makes sense. hi antimatroid1! 2011-10-27T11:30:26 if given more time wouldn't sir macelon probably be ranked the same as necrys and momobot? 2011-10-27T11:30:27 *** boegel has quit IRC (Ping timeout: 260 seconds) 2011-10-27T11:30:46 ie. aren't the rankings other than 1st just as important for the overall rankings? 2011-10-27T11:31:47 hello 2011-10-27T11:31:48 antimatroid1: that's a good question. 2011-10-27T11:32:18 could someone maybe help me with running playgame.py on a map different from tutorial1? 2011-10-27T11:32:21 antimatroid1: aren't you one of those people who optimize their code? 2011-10-27T11:32:41 depends on the definition of optimization, not by any rigorous definition of the word 2011-10-27T11:33:11 i understand the importance of faster games and why we're cutting them short, but i think we need to make sure we do it only once rankings are pretty much finalised 2011-10-27T11:33:15 antimatroid1: do you code in C++ and use A*? 2011-10-27T11:33:24 i do 2011-10-27T11:33:48 do you have any numbers on it's speed? 2011-10-27T11:34:00 nope sorry 2011-10-27T11:34:11 talk to mcstar when he's about 2011-10-27T11:34:55 ok, will do. and you probably have to forward your question to McLeopold 2011-10-27T11:35:34 also, we've got over 3000 submissions i think 2011-10-27T11:35:49 chaudeau: start it with other --map_file parameter 2011-10-27T11:36:32 edit play_one_game.cmd 2011-10-27T11:36:38 @later tell McLeopold http://aichallenge.org/visualizer.php?game=19781&user=4513 has the rank really stabilised in this game? Wouldn't sir macelon probably also be ranked the same as necrys and momobot if given more time and aren't the rankings other than 1st just as important for the overall rankings? 2011-10-27T11:36:38 antimatroid1: I come to serve. 2011-10-27T11:38:21 *** delt0r__ has joined #aichallenge 2011-10-27T11:41:00 *** Guki has joined #aichallenge 2011-10-27T11:41:27 *** mithos28 has joined #aichallenge 2011-10-27T11:43:08 so i'm coding in C# and i seem to run into some performance issues (timeouts at around 400 ants) and i still want to add more complexity to the decision making process. I'm pretty sure that my core issue is not my A*, which seems to run fast enough but rather the input i'm giving it. Eg. lists of foods sorted by distance to the current Ant 2011-10-27T11:44:13 Should i be reusing them and not recreating them, should i drop generic lists all together and grab hashtables and hashmaps instead? or is my core problem that i recalculate every search for every ant every turn? 2011-10-27T11:44:50 As you can probably guess from my question, i'm not normally engaged in coding where performance is critical 2011-10-27T11:45:21 nielsrobin: why are you only "pretty sure" your A* implementation is not your problem? i for one am pretty _damn_ sure C# supports profiling. you should make use of it and find out what is causing your problems ;) 2011-10-27T11:45:26 *** kaemo has joined #aichallenge 2011-10-27T11:45:52 other than that, yes, caching paths does make sense. why recalculate something if there won't be anything different? 2011-10-27T11:46:46 *** smiley1983 has quit IRC (Quit: leaving) 2011-10-27T11:47:07 *** kapoc has joined #aichallenge 2011-10-27T11:47:29 bqf: except possibly new ants or food that change your chosen paths 2011-10-27T11:47:35 hi. i`m trying to run live simulation under ubuntu 11.10 and getting problem like there http://pastebin.com/1yUFrUHa any ideas? 2011-10-27T11:47:41 or loss of ants 2011-10-27T11:47:45 bqf: profiling ok, i'll have a google at that :) 2011-10-27T11:48:05 * antimatroid1 recalculates everything every move 2011-10-27T11:48:17 bearoff_w: thanks, I found the error now, /tools/ was missing in the map path :-/ 2011-10-27T11:48:29 oh wait, might not be a problem 2011-10-27T11:48:39 *** onensora has quit IRC (Ping timeout: 256 seconds) 2011-10-27T11:49:26 antimatroid1: is your bot stateless in terms of turns? 2011-10-27T11:49:50 i don't really have a bot atm but it was 2011-10-27T11:49:57 but i just realised it doesn't need to be the way i collect food 2011-10-27T11:50:15 i've got some pretty fun ideas to try out once i finish exams etc. 2011-10-27T11:50:31 *** olexs has quit IRC (Quit: Leaving.) 2011-10-27T11:51:18 *** master_ninja has joined #aichallenge 2011-10-27T11:51:19 damn, now i really want to start my bot aha 2011-10-27T11:51:55 *** Saulzar has joined #aichallenge 2011-10-27T11:52:37 *** Areks has quit IRC (Ping timeout: 240 seconds) 2011-10-27T11:54:00 *** frost_ has quit IRC (Ping timeout: 265 seconds) 2011-10-27T11:56:28 *** loktar has joined #aichallenge 2011-10-27T11:58:43 *** McLeopold has joined #aichallenge 2011-10-27T12:05:33 *** JamesMG has joined #aichallenge 2011-10-27T12:06:29 *** chuncho has joined #aichallenge 2011-10-27T12:06:33 antimatroid1: i'm facing the same dilemma. the new semester only just began and i really don't want to fall behind in course work because i'm taking care of my ants running around in mazes ;) 2011-10-27T12:06:55 bqf: i'm just finishing off my honours year then it's summer time :D 2011-10-27T12:07:04 antimatroid1: just took a look at that game 2011-10-27T12:07:33 McLeopold: i realise sir macelon might have taken some work to beat, but i imagine the game would have been stopped even if he was about to be taken? 2011-10-27T12:07:41 does anyone know what flags do i need to use to get the logger.debug to write to a file.. (using GreedyBot) ? in python :P 2011-10-27T12:08:15 yeah, we'd need to add logic that says, "if any bot can gain *or lose* rank, then the game continues" 2011-10-27T12:08:23 right now we just check for a gain in rank 2011-10-27T12:08:28 yep 2011-10-27T12:08:47 *** JamesMG_ has quit IRC (Ping timeout: 248 seconds) 2011-10-27T12:08:54 I'm also slightly okay with him having a higher rank than the other too, because he lasted longer 2011-10-27T12:09:03 i'm a bit iffy on that 2011-10-27T12:09:14 why not call it off before the middle bot got eliminated too then :P 2011-10-27T12:09:16 There are examples of it being more iffy than that 2011-10-27T12:09:40 because sir_macelon could have taken out the second and tied 2011-10-27T12:10:08 hmm yeah okay, i still think we should be concerned about rankings in general and not just first place in each game 2011-10-27T12:10:19 even if our only concern was finding the best bot, which it is not 2011-10-27T12:10:30 does skill calculations use anything but the top bot? 2011-10-27T12:10:44 it uses the ordinal rankings of bots 2011-10-27T12:10:53 at least i think it does? 2011-10-27T12:10:55 *** Ishuah has quit IRC (Ping timeout: 265 seconds) 2011-10-27T12:11:13 or total preordering if you will 2011-10-27T12:12:32 it does 2011-10-27T12:12:59 *** Areks|2 has joined #aichallenge 2011-10-27T12:13:25 *** kaemo has quit IRC (Read error: Operation timed out) 2011-10-27T12:13:29 then its a bug. if the winning bot has a big advantage, being the one he goes after first is not in your control 2011-10-27T12:13:55 GarfTop: we've established that :P 2011-10-27T12:14:06 Or rather, being last on his hitlist shouldn't give you an advantage 2011-10-27T12:14:19 the point is how important is it compared to the length of games, personally i think it is important 2011-10-27T12:14:47 *** Guki has quit IRC (Ping timeout: 265 seconds) 2011-10-27T12:15:12 *** onensora has joined #aichallenge 2011-10-27T12:16:37 The game rate *is* pretty horrible... 2011-10-27T12:17:02 *** Mooloo has joined #aichallenge 2011-10-27T12:17:04 In C++ starter package, there is a method named updateVisionInformation and a comment that says it's inefficient. Is it worth optimizing? How much time it could take to calculate if left unoptimized? 2011-10-27T12:17:11 *** tncardoso has joined #aichallenge 2011-10-27T12:17:13 lavalamp: yeah, more opponents in each game and way more turns 2011-10-27T12:17:17 tcp ftw 2011-10-27T12:17:47 master_ninja: i wrote that note and intend to just leave the code as is unless someone claims it actually does take a while with lots of ants 2011-10-27T12:18:12 there's a discussion in the forums 2011-10-27T12:18:18 i don't have time to read the entire thread though 2011-10-27T12:18:20 yeah, tcp. 2011-10-27T12:18:27 *** Glert has quit IRC (Read error: Operation timed out) 2011-10-27T12:19:08 antimatroid1: yeah, i profiled it and it was the second mos time consuming in some games with lots of ants.. but I guess I should optimize my path finding first ;) 2011-10-27T12:19:17 master_ninja: the point is don't just add all ant locations to one queue and search outwards, i can't remember what exactly but since the euclidean and manhatten metrics aren't equivalent that doesn't work correctly 2011-10-27T12:21:59 *** CptOatmeal has joined #aichallenge 2011-10-27T12:22:27 g'day folks, anyone getting an error with the visualizer.java ? just dled the package and the thing crashes as soon as I move my mouse over the visualizer window 2011-10-27T12:22:38 antimatroid1: btw, thanks for the started package. I find it quite easy to work with. 2011-10-27T12:22:54 rhinoException org.mozilla.javascript.EcmaError: TypeError: Cannot find default value for object. (CanvasElement.js#1038) 2011-10-27T12:23:02 master_ninja: it's basically just my planet wars bot made simple for ants :) 2011-10-27T12:23:18 structurally anyway 2011-10-27T12:23:24 antimatroid1: it gave incorrect results if a square was reached first from one ant rather than a closer ant 2011-10-27T12:23:32 *** bearoff_w has left #aichallenge 2011-10-27T12:24:36 you can add all the ants and search outwards, but when you reach a square that has already been searched, you need to track that furthur distance calcs must be done for both ants 2011-10-27T12:24:45 *** kapoc has left #aichallenge 2011-10-27T12:24:48 McLeopold: yeah but i can't remember why example wise, my go to example to show the metrics aren't equivalent is 2011-10-27T12:24:48 a.* 2011-10-27T12:24:48 .b. 2011-10-27T12:24:48 mdist(a,*) = mdist(b, *) = 2, edist(a,*) = 2 > sqrt(2) = edist(b,*) 2011-10-27T12:25:05 but even if you searched from a first, you could go right and up from b to find * 2011-10-27T12:25:10 it only appeared starting at distance 7, I believe 2011-10-27T12:25:17 ah yeah okay 2011-10-27T12:25:22 plus, you had to have a speacial offset 2011-10-27T12:25:41 yeah, like i said in the forums, the correct faster way is messy 2011-10-27T12:25:57 you'd end up with a few spots that are visible that you might think are not at the edges of you vision 2011-10-27T12:26:21 McLeopold: http://aichallenge.org/forums/viewtopic.php?f=24&t=1599 if you have any time you might want to explain some of that in there 2011-10-27T12:26:32 i probably shouldn't even be chatting in here atm :) 2011-10-27T12:26:38 I don't have time either :) 2011-10-27T12:27:06 oh god, someone in there is indexing with [x][y] :( 2011-10-27T12:27:21 i will never understand that convention 2011-10-27T12:28:15 i realise that's what computer graphics programmers do, but it means their notation is like reversed a lot of the time, like for cross products etc. 2011-10-27T12:29:00 *** kaemo has joined #aichallenge 2011-10-27T12:29:21 *** Blkt has quit IRC (Quit: ERC Version 5.3 (IRC client for Emacs)) 2011-10-27T12:30:12 they'd be the kind of people to integrate a line integral clockwise i bet :P 2011-10-27T12:30:31 around a closed loop* 2011-10-27T12:31:33 antimatroid1: i think it's just that when most people hear "coordinates", they automatically think "x, y" 2011-10-27T12:31:49 i think location, not coordinate :P 2011-10-27T12:32:31 loc and location are shorter than coord and coordinate, so i'm going to claim superiority :P 2011-10-27T12:32:44 :) 2011-10-27T12:33:36 and i think of the grid as being like a matrix just not over a field to form a ring 2011-10-27T12:34:07 and then there's the 1% of us that call it point 2011-10-27T12:34:21 shh you :P 2011-10-27T12:34:52 i'm just using unsigned int location and my "grid" is just single vector 2011-10-27T12:35:16 i turned my grid into a graph for pathfinding 2011-10-27T12:35:29 no need to calcualte neighbours all the time or check if a location is water 2011-10-27T12:35:42 brilliant 2011-10-27T12:36:27 just add a neighbours vector to each square and remove appropriate neighbours as you receive water inforamtion 2011-10-27T12:36:32 *** chuncho has quit IRC (Ping timeout: 265 seconds) 2011-10-27T12:36:32 *** chaudeau has quit IRC (Ping timeout: 265 seconds) 2011-10-27T12:36:54 time saving is probably negligible but it makes things a lot nicer anyway 2011-10-27T12:37:26 indeed 2011-10-27T12:38:06 antimatroid1: that's what i plan to do. it also solves the wrapping at the borders rather nicely - just connect the edges. 2011-10-27T12:38:08 hmmm i wonder how long it would take to calculate for each square the average length of the shortest path to every other square 2011-10-27T12:38:23 to give you some idea of which squares had more "escape" paths 2011-10-27T12:38:45 i imagine quite a long time unless i'm missing something 2011-10-27T12:39:06 antimatroid1: since the official info is that you have "about 1 gig of ram" available for your bot, you could just do this in the background if you have time left 2011-10-27T12:39:09 bqf: wrapping is solved by get neighbours or whatever :P 2011-10-27T12:39:24 *** dmiles_afk has joined #aichallenge 2011-10-27T12:39:24 it also changes as you get water information 2011-10-27T12:39:39 memory isn't an issue either 2011-10-27T12:40:07 well, i guess you would store them if you were going to that effort 2011-10-27T12:40:15 but there's no way you'd be able to get them all 2011-10-27T12:40:42 antimatroid1: the way i plan to implement things, memory might become scarce. then again, if other people are using java, then i guess i'll be fine ;) 2011-10-27T12:41:06 if you're planning a game tree think very carefully about just how ridiculously big the state space is 2011-10-27T12:41:18 even if you partition up ants that can battle into sets that can affect each other 2011-10-27T12:41:33 although i think olexs has had sucess with minimax 2011-10-27T12:41:36 so it's not completely out 2011-10-27T12:42:12 *** Knekkebjoern has quit IRC (Ping timeout: 260 seconds) 2011-10-27T12:50:00 *** CptOatmeal has quit IRC (Quit: Page closed) 2011-10-27T12:51:26 *** Macuyiko has joined #aichallenge 2011-10-27T12:52:00 *** guest___ has quit IRC (Ping timeout: 265 seconds) 2011-10-27T12:52:06 *** bhasker has quit IRC (Quit: bhasker) 2011-10-27T12:56:22 and so xathis is on top now 2011-10-27T12:56:35 with 13 games 2011-10-27T12:56:51 *** bishoco has joined #aichallenge 2011-10-27T12:57:19 and i still can't figure minmaxing 2011-10-27T12:57:35 i know how it works, but have no idea how to apply to ants 2011-10-27T12:57:42 fluxid: have you read the tron tutorial on minimax? 2011-10-27T12:57:50 antimatroid1: nope? 2011-10-27T12:58:00 http://www.sifflez.org/misc/tronbot/index.html 2011-10-27T12:58:43 thanks 2011-10-27T12:58:49 take two ants that can battle, the first layer is the moves your ant can make, the second layer is the trees your enemy can make then the terminal nodes have the outcome after battle resolution is run, then you need a function to evaluate the states 2011-10-27T12:58:55 and that gives you your game tree 2011-10-27T12:59:06 moves* your enemy can make 2011-10-27T12:59:30 but i don't have just one enemy 2011-10-27T13:00:10 but group of ants, and possibly group of my own ants which may have more killpower when coming closer to enemy 2011-10-27T13:00:18 *** Knekkebjoern has joined #aichallenge 2011-10-27T13:00:38 you can either treat it as an nplayer game tree and run backwards induction or treat all enemies as "one player ganging up on you" 2011-10-27T13:00:41 i plan to do the later 2011-10-27T13:01:19 when evaluating possible moves of one ants, can i simplify the case and try four same directions of all my ants, then four directions of all the enemy? 2011-10-27T13:01:26 also, i'm considering for a set of my ants that i want to consider moves together with as finding their set of moves, taking out collisions and then having a branch for each of those moves in one layer, rather than having a layer for each ant 2011-10-27T13:02:04 just ignore the fact that my or enemy ants can move in different directions in next turn 2011-10-27T13:02:22 i don't quite get what you mean? 2011-10-27T13:02:43 you can do more layers of the tree, but i think you'll be struggling enough already to do moves for you and your enemys 1 turn deep 2011-10-27T13:03:01 *** dmiles_afk has quit IRC (Ping timeout: 240 seconds) 2011-10-27T13:03:03 people were getting 10+ moves ahead with tron 2011-10-27T13:03:18 oh boy 2011-10-27T13:03:50 with tron we would recycle the relevant part of our game tree each turn then just iteratively generate it for as long as possible then make out decision based off that 2011-10-27T13:03:53 *** cybsy has joined #aichallenge 2011-10-27T13:05:58 preference of states was something like lose < separate with less squares available to us < same region with less squares closer to us < draw < separate with equal squares closer < together with equal squares closer < together with more squares closer < separate with more squares available < win 2011-10-27T13:06:27 and ordered relevant things in that based on the difference of squares which were closer to which bot path wise 2011-10-27T13:06:50 a1k0n: was your state evaluation in tron like that^^? 2011-10-27T13:08:08 *** dmiles_afk has joined #aichallenge 2011-10-27T13:09:11 Fluxid: when i write a game tree, i have a game tree struct with the root node and all functions to get moves etc. and a node struct and each node has a game state and state evaluation function (or maybe that's in state) etc. 2011-10-27T13:10:03 duh, so this collides with my bot design... i think 2011-10-27T13:10:10 :P 2011-10-27T13:10:19 always plan your bot before designing everything 2011-10-27T13:10:21 otherwise that happens 2011-10-27T13:10:44 antimatroid1: i have no experience at all 2011-10-27T13:10:53 i didn;t know what was minmaxing until yesterday 2011-10-27T13:11:04 yeah, i had no idea at the start of tron 2011-10-27T13:11:12 we all have to learn somewhere 2011-10-27T13:11:37 still i was surprised that my totally dumb bot which implemented only floodfiling to find directions of ants got to 30 place 2011-10-27T13:11:58 it's still very early in the contest 2011-10-27T13:12:28 more advanced players are coming, huh 2011-10-27T13:12:41 if past contests are anything to go by ;) 2011-10-27T13:13:16 *** dfgdfg has quit IRC (Ping timeout: 265 seconds) 2011-10-27T13:14:51 i'm curious, what do people think of the battle resolution rules? 2011-10-27T13:19:33 antimatroid1: just to clarify: i have group of four ants ants, each ant can move in four directions. do i try all possible moves of all ants? and then, four enemy ants... this tree would be huge! 2011-10-27T13:20:05 each ant has up to 5 moves actually, it can not move if you will 2011-10-27T13:20:11 even bigger 2011-10-27T13:20:21 yes the tree is massive, you need to use less ants or prune the tree somehow 2011-10-27T13:20:58 i'm also kind of iffy about how people are managing game trees with this 2011-10-27T13:21:07 i had pretty much ruled it out, but it appears people are 2011-10-27T13:21:58 *** cybsy is now known as Cybs 2011-10-27T13:22:49 *** Redgis has joined #aichallenge 2011-10-27T13:23:15 *** epicmonkey has joined #aichallenge 2011-10-27T13:23:22 *** Cybs is now known as Cybsy 2011-10-27T13:24:03 so, when considering moves and evaluating points, can i assume all of my ants move in same direction, (so it gives four nodes at first level) and then all of the enemy ants move in same direction, so second level would have sixteen nodes? 2011-10-27T13:24:19 sorry, five, twenty-five 2011-10-27T13:24:58 would accuracy of this be still okay? 2011-10-27T13:26:59 *** JamesMG_ has joined #aichallenge 2011-10-27T13:29:08 *** fenec has joined #aichallenge 2011-10-27T13:29:24 is anybody working on the game with RUBY? 2011-10-27T13:29:33 pls help 2011-10-27T13:29:37 ants on rails? 2011-10-27T13:30:01 that might make pathfinding easier 2011-10-27T13:30:17 *** zefyear has joined #aichallenge 2011-10-27T13:30:21 am trying to use ruby starter kit for the game 2011-10-27T13:30:24 *** JamesMG has quit IRC (Ping timeout: 255 seconds) 2011-10-27T13:30:34 but apprently a lot of features are missing 2011-10-27T13:30:45 like support for hills 2011-10-27T13:30:56 what language are you using? 2011-10-27T13:31:09 you can build everything from scratch easily, using different starting kit as starter 2011-10-27T13:31:17 *** HaraKiri has joined #aichallenge 2011-10-27T13:31:17 in different language 2011-10-27T13:31:27 as base* 2011-10-27T13:31:41 many people didn't use starter kit at all 2011-10-27T13:31:50 *** Larose has joined #aichallenge 2011-10-27T13:32:07 *** barsandcat has quit IRC (Ping timeout: 265 seconds) 2011-10-27T13:32:11 is there an API? 2011-10-27T13:32:27 protocol is described on page 2011-10-27T13:32:43 http://aichallenge.org/specification.php 2011-10-27T13:34:10 how can i find elaborated examples of a bot, my code is a mess 2011-10-27T13:34:16 yea, i glanced through the python starter kit while doing the tutorial, but apart from a few pieces of code that may or may not be eventually changed, i'm doing mine from scratch 2011-10-27T13:35:59 i made mine from scratch too, and glanced twice at game engine's ants.py (not the one from kit) to see how stuff actually work 2011-10-27T13:37:44 Any word on why the contest game processing is so slow? The current game rate is 1.5 games a minute which tells something is wrong with their servers. 2011-10-27T13:39:26 http://aichallenge.org/visualizer.php?game=19847 too bad this game ended so early 2011-10-27T13:39:29 there's nothing wrong with the servers, it's just 3 of them against 3000 of us .. 2011-10-27T13:40:20 *** loktar has quit IRC (Ping timeout: 265 seconds) 2011-10-27T13:41:33 *** Mayank has joined #aichallenge 2011-10-27T13:43:19 true. But, the games are only going to get slower as people run more sophisticated bots. I know it's not trivial to set up new servers, though. Hopefully, they can figure something out. 2011-10-27T13:43:20 Just out of interest, is anyone trying an autonomous agents approach as opposed to a top-down managed 2011-10-27T13:43:27 wow, bix0r4ever had 453 ants with 568 in his hive 2011-10-27T13:43:46 zefyear, AndrewBC is doing something like that 2011-10-27T13:44:15 indeed 2011-10-27T13:44:27 though I'm mostly only able to participate on weekends ;( 2011-10-27T13:45:20 I'm trying particle swarm optimization but with google moving so slow stochastic ants don't have much time to learn :( 2011-10-27T13:45:30 *** bhasker has joined #aichallenge 2011-10-27T13:45:47 how is google moving slow affecting you? 2011-10-27T13:45:50 zefyear: I'm trying to do this too (agent-based) 2011-10-27T13:46:08 *** invader has quit IRC (Quit: Page closed) 2011-10-27T13:46:32 Zannick: For PSO or any other GA to work, you have to have a lot of iterations so you can influence the weights in a markov chain 2011-10-27T13:47:06 If all of your ants 'die out' after each lost game and 'procreate' every won game, you have to lose a lot of games before you can converge to the local optima 2011-10-27T13:47:10 so there are a few things here i should point out 2011-10-27T13:47:47 first, you can't write anything on the contest servers. you have to run your learning or markov algorithm elsewhere 2011-10-27T13:48:20 there are unofficial tcp servers; check the forums or type @tcp 2011-10-27T13:48:22 i noticed 2011-10-27T13:48:32 you can pipe out traffic to the net 2011-10-27T13:48:42 that will get you disqualified 2011-10-27T13:49:00 second, google is not running the contest, only sponsoring it 2011-10-27T13:49:44 all that aichallenge gets from them is money and some publicity 2011-10-27T13:50:17 some googlers might be helping run in their spare time :) 2011-10-27T13:50:45 *** dwins has quit IRC (Ping timeout: 252 seconds) 2011-10-27T13:50:47 but yes, i recommend using the tcp servers to seed your learning algorithm 2011-10-27T13:51:06 yeah, you play games a lot more often with them 2011-10-27T13:51:28 check the tools section of the forum for a sticky about them 2011-10-27T13:53:13 gah, i hate typing when my hands are cold 2011-10-27T13:53:30 i just optimized my bot... best evening ever! 2011-10-27T13:53:50 it's still stupid though 2011-10-27T13:53:55 so no christmast yet 2011-10-27T13:54:01 it's now stupid faster 2011-10-27T13:54:23 yeah! 2011-10-27T13:54:47 lol 2011-10-27T13:55:46 *** heinrich5991 has quit IRC (Excess Flood) 2011-10-27T13:55:58 *** Knekkebjoern has quit IRC (Quit: Leaving.) 2011-10-27T13:56:15 *** Knekkebjoern has joined #aichallenge 2011-10-27T13:58:27 *** SITZ has joined #aichallenge 2011-10-27T14:00:23 *** heinrich5991 has joined #aichallenge 2011-10-27T14:00:30 *** halligalli has quit IRC (Quit: Page closed) 2011-10-27T14:01:07 *** nielsrobin has quit IRC (Ping timeout: 265 seconds) 2011-10-27T14:04:05 *** Bluedgis has quit IRC (Quit: leaving) 2011-10-27T14:04:54 hrm... can you not move onto a food square until the food is gone? 2011-10-27T14:05:22 e.g. if my ant is directly next to food, can I issue an order for that ant to move directly on top of the food locaiton - or do i have to wait for the next turn for the food to dissapear? 2011-10-27T14:05:36 *** Kingpin13 has quit IRC (Ping timeout: 260 seconds) 2011-10-27T14:05:39 i keep on getting # move blocked 2011-10-27T14:05:42 *** irchs has joined #aichallenge 2011-10-27T14:06:22 magiik: no, food blocks 2011-10-27T14:06:43 you need to wait 2011-10-27T14:06:43 hrm... i see, i guess the starter pack doesn't take that into account in it's .passable() routine 2011-10-27T14:06:57 aichallenge: janzert epsilon * r7dfbb40 / worker/compiler.py : Stop extraneous None's getting into the output - http://git.io/yTmObA 2011-10-27T14:06:58 aichallenge: janzert epsilon * r88fecd8 / (6 files in 4 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/764luw 2011-10-27T14:07:00 magiik: you want unoccupied 2011-10-27T14:07:00 possible 2011-10-27T14:07:16 passable means you could move there at some point 2011-10-27T14:07:17 i guess that makes sense 2011-10-27T14:07:32 for path finding, passable might be better because some of that stuff is temporary 2011-10-27T14:07:43 ok, thanks 2011-10-27T14:09:12 it's probably easier to just have your bot direct orders onto the food aye? 2011-10-27T14:09:31 *** Kingpin13 has joined #aichallenge 2011-10-27T14:09:36 antimatroid1: except you will fail the submission test 2011-10-27T14:09:46 that's possibly easier than putting in code to make sure it realises it's already adjacent to food 2011-10-27T14:09:52 oh why? 2011-10-27T14:09:59 if it blocks food, why fail submissions on it? 2011-10-27T14:10:03 because it is run in strict mode 2011-10-27T14:10:18 to weed out poorly written bots from participating 2011-10-27T14:10:30 but that'd be taking advantage of engine "features" :P 2011-10-27T14:11:53 *** modafinil has quit IRC (Remote host closed the connection) 2011-10-27T14:11:53 *** upgrayeddd has quit IRC (Remote host closed the connection) 2011-10-27T14:13:44 *** Redgis has quit IRC (Read error: Connection reset by peer) 2011-10-27T14:14:30 *** Redgis has joined #aichallenge 2011-10-27T14:16:41 http://aichallenge.org/visualizer.php?game=20676 <- meh, someone is collecting a big food supply 2011-10-27T14:17:20 its still only a fraction though 2011-10-27T14:17:58 *** modafinil has joined #aichallenge 2011-10-27T14:20:40 hmm, my latest submission has been 'compiling and running tests' for like 6 hours.. 2011-10-27T14:21:51 you're the one that broke the workers ;) 2011-10-27T14:22:00 oh, srry ;( 2011-10-27T14:22:03 it should have cleared now 2011-10-27T14:22:21 not really your fault just that submission happened to tickle the bug 2011-10-27T14:22:23 how did ido that ? 2011-10-27T14:22:31 haha. k 2011-10-27T14:23:12 *** choas has joined #aichallenge 2011-10-27T14:25:52 guess, i'll just watch the updates on github, to see, what happened there .. 2011-10-27T14:26:06 oh, yes it was the last commit 2011-10-27T14:26:29 Warning: Don't tickle the bugs 2011-10-27T14:26:30 basically it was allowing a None through as output that it shouldn't have and it was causing later code to choke 2011-10-27T14:26:45 janzert: do you want to merge the Ada starter? 2011-10-27T14:27:03 McLeopold: I will once I have a chance to fix it 2011-10-27T14:27:18 by it I mean the language support, not the starter bot itself 2011-10-27T14:27:23 janzert: I have an idea to speed up games. 2011-10-27T14:27:38 yes? 2011-10-27T14:27:49 I've already created a database field to support the idea a month ago :) 2011-10-27T14:28:19 Basically, a submission is deactivated after a certain amount of time. It will no longer be used as a seed, but can still be pulled in for matches. 2011-10-27T14:28:48 Each user profile page will allow you to reactivate any submission, if you want your current bot to continue to play. 2011-10-27T14:28:51 yeah, I'd like to see that 2011-10-27T14:28:58 :-) 2011-10-27T14:29:04 I was even discussing it with someone else the other day 2011-10-27T14:29:09 My assumption is that after awhile, starter bots will all start to drop off, and never be pulled into games. 2011-10-27T14:29:19 I think the time needs to be at least a week though 2011-10-27T14:29:26 (well, lets say you can only activate 1 bot?) 2011-10-27T14:29:28 * janzert nods 2011-10-27T14:29:45 I was thinking 3-5 days. 2011-10-27T14:29:47 avdg: it'll just be your latest 2011-10-27T14:30:17 on day 7, you'd still be pulled in if you had a serious submission. 2011-10-27T14:30:41 those that are checking everyday can just click the button to keep it activated 2011-10-27T14:30:52 yeah, but I can see it common that people work on their bot once a week. it'd be nice for it to run normally the rest of the week 2011-10-27T14:31:13 *** amstan has joined #aichallenge 2011-10-27T14:31:13 *** ChanServ sets mode: +o amstan 2011-10-27T14:31:15 After 3 days, you should be stable, and you'd still get some games. 2011-10-27T14:31:27 either way abandoned accounts will drop off 2011-10-27T14:31:31 yeah 2011-10-27T14:31:45 what to do if you activate an old submission? 2011-10-27T14:31:59 I don't want to reset the mu/sigma, because that gives an incentive to click it 2011-10-27T14:32:01 I think only let the latest be activated 2011-10-27T14:32:21 I'd really like them to be able to select any. 2011-10-27T14:33:25 I know we're doing it now anyway, but that means we are then committed to keeping all old submissions 2011-10-27T14:34:13 It makes the day before close so much easier for everyone 2011-10-27T14:34:37 Upload 5 bot candidates, pick the best performer at the end 2011-10-27T14:34:42 I'd rather people learn not to do last second uploads :P 2011-10-27T14:35:33 we could even allow the activate to occur a few hours after submissions are closed 2011-10-27T14:35:51 that's probably not going to happen 2011-10-27T14:35:59 anyway, what will happen after submission are closed? 2011-10-27T14:36:06 *** tncardoso has quit IRC (Quit: Leaving) 2011-10-27T14:36:15 So, if you fractically uploaded a crash bot at the last second, you have a few hours to watch its performance, and then decide to revert to the last submission 2011-10-27T14:36:15 it takes me a while to get the finals started and I end up staying up too late that night as it is 2011-10-27T14:36:25 adding a few hour delay in would be bad 2011-10-27T14:36:36 it gives you more time 2011-10-27T14:36:46 finals start 24 hours after submission deadline 2011-10-27T14:36:56 not really 2011-10-27T14:37:09 they can, we can say the can :) 2011-10-27T14:37:32 we're trying to shorten the time between close of submissions and the finals ending not lengthen it :P 2011-10-27T14:37:32 finals don't need to start right at the deadline 2011-10-27T14:37:40 why? 2011-10-27T14:37:40 janzert, sorry for the trouble, thanks for the help, works now ;) 2011-10-27T14:37:49 *** Ttech has quit IRC (Ping timeout: 240 seconds) 2011-10-27T14:37:57 g0llum: don't worry it wasn't your fault at all 2011-10-27T14:38:07 yeah, it was probably mine :) 2011-10-27T14:38:23 half of it was at least mine :) 2011-10-27T14:38:35 so, why do we need a quick final? 2011-10-27T14:38:38 the other half had been there for a very long time, I'm not sure who wrote it 2011-10-27T14:38:51 because people start to lose interest 2011-10-27T14:39:04 we want to wrap it up fast 2011-10-27T14:39:13 but, submission are closed? why do we need interest? 2011-10-27T14:39:31 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-10-27T14:39:59 *** g0llum has joined #aichallenge 2011-10-27T14:40:03 interest is kinda the point of all of it isn't it? 2011-10-27T14:40:50 for the start 2011-10-27T14:40:57 it would be bad luck if food spawned just next to my any in submission test... i guess i just test my luck for now :P 2011-10-27T14:41:00 it seems kinda mute for the finals 2011-10-27T14:41:21 *oops, wrong window 2011-10-27T14:41:25 master_ninja: no food spawning on submission test for now 2011-10-27T14:41:37 if you don't have your final hurrah, you aren't as likely to come back for the next one 2011-10-27T14:42:11 okay, so, the idea of activating an old bot seems to me a way to lessen the stress load just before finals start 2011-10-27T14:42:32 less people uploading their best bot, more people just activating their best bot 2011-10-27T14:43:22 *** chrisf has joined #aichallenge 2011-10-27T14:43:26 plus, for those that upload a bad performer, it's easier to just switch bad to the previous version then upload the same code again 2011-10-27T14:43:30 morning #aichallenge 2011-10-27T14:43:46 to me the commitment we have to make in keeping all submissions is still less than any advantage gained 2011-10-27T14:44:08 *** muha has quit IRC (Ping timeout: 265 seconds) 2011-10-27T14:44:29 less worker compiles? 2011-10-27T14:44:51 and I worry a bit about what sort of bugs we are going to run into by reactivating old submissions, seeing as the system hasn't been dealing with that 2011-10-27T14:44:53 we keep them now, we can discard them later for archiving 2011-10-27T14:45:10 like trueskill updates or matchmaking? 2011-10-27T14:45:13 not really, new workers basically recompile all active submissions all the time anyway 2011-10-27T14:45:16 *** Ephemera has quit IRC (Read error: Connection reset by peer) 2011-10-27T14:46:06 I don't really know where all problems could crop up just I won't be surprised to run into places that have it as an assumption 2011-10-27T14:46:29 *** Ephemera has joined #aichallenge 2011-10-27T14:46:42 *** dvladim has joined #aichallenge 2011-10-27T14:46:42 well, the system is capable of matching only the last successful submission today, so I think it can already handle this 2011-10-27T14:47:21 btw, I think we're pretty certain to surpass the number of active accounts in planetwars this weekend 2011-10-27T14:47:35 yep :) 2011-10-27T14:47:53 okay, I'll only do last submission for now, and continue to argue the point next week :) 2011-10-27T14:47:54 \o/ 2011-10-27T14:48:05 :) ok 2011-10-27T14:48:17 janzert: also, take a look at the pull request for the remember me option 2011-10-27T14:48:21 although this change will knock us back down a bit 2011-10-27T14:48:31 *** Rainydays has joined #aichallenge 2011-10-27T14:48:34 "down"? 2011-10-27T14:48:41 oh 2011-10-27T14:48:43 number of active accounts 2011-10-27T14:48:44 nvm 2011-10-27T14:49:06 *** the_real_crimper has quit IRC (Ping timeout: 276 seconds) 2011-10-27T14:49:10 *** McLeopold has left #aichallenge 2011-10-27T14:49:38 *** the_real_crimper has joined #aichallenge 2011-10-27T14:50:42 *** muha has joined #aichallenge 2011-10-27T14:51:41 *** Accoun has quit IRC (Ping timeout: 276 seconds) 2011-10-27T14:52:03 *** Accoun has joined #aichallenge 2011-10-27T14:52:50 *** Xeorr has quit IRC (Ping timeout: 263 seconds) 2011-10-27T14:53:35 *** ajf|offline is now known as ajf 2011-10-27T14:53:55 that 'remember me' patch looks good enough to me 2011-10-27T14:54:15 when does the challenge end? 2011-10-27T14:54:43 definitely don't want it using a hash of the password hash 2011-10-27T14:54:44 december ish 2011-10-27T14:55:02 ajf: see the bottom of the front page 2011-10-27T14:55:04 but a UUID for each user would be irritating 2011-10-27T14:55:14 since it'd only let you log in on one device at a time 2011-10-27T14:55:18 more ideal would be cookies 2011-10-27T14:55:23 a UUID isn't really needed 2011-10-27T14:55:25 it uses cookies 2011-10-27T14:55:36 just a large random number will be fine 2011-10-27T14:55:44 *** upgrayeddd has joined #aichallenge 2011-10-27T14:55:48 the activation code could probably even be reused 2011-10-27T14:56:13 *** sdf has joined #aichallenge 2011-10-27T14:56:22 *** upgrayeddd has left #aichallenge 2011-10-27T14:57:27 *** GIM has joined #aichallenge 2011-10-27T14:57:50 *** Saulzar has quit IRC (Ping timeout: 260 seconds) 2011-10-27T14:57:59 *** GIM has left #aichallenge 2011-10-27T14:58:11 *** xcrypt has joined #aichallenge 2011-10-27T14:58:22 hello 2011-10-27T14:59:27 I have a problem running 'play_one_game.cmd' from tools on windows 7 64 bit. I installed python but it still doesn't seem to work. Could anyone help me? 2011-10-27T14:59:37 *** danielharan has joined #aichallenge 2011-10-27T15:01:09 *** dwins has joined #aichallenge 2011-10-27T15:01:14 xcrypt: doesn't work how? 2011-10-27T15:01:21 xcrypt, still noone knows, what happened to you, what's the error ? 2011-10-27T15:01:52 it just opens up the console and then closes without doing anything 2011-10-27T15:01:54 no error 2011-10-27T15:02:02 xcrypt: run it from a terminal 2011-10-27T15:02:39 I'll try thx :) 2011-10-27T15:03:16 December 18th? 2011-10-27T15:03:20 whoa, I have loads of time 2011-10-27T15:03:36 btw, when does this challenge end? 2011-10-27T15:03:43 *** dmiles_afk has quit IRC (Ping timeout: 252 seconds) 2011-10-27T15:03:55 ^^ 2011-10-27T15:04:00 couldn't find the info on the site 2011-10-27T15:04:13 http://aichallenge.org/ 2011-10-27T15:04:21 >.> 2011-10-27T15:04:22 xcrypt: scroll down to bottom of front page 2011-10-27T15:04:26 like me, you missed it :) 2011-10-27T15:04:30 *** ChanServ sets mode: +o Zannick 2011-10-27T15:04:51 oh lol. Might want to put that somewhere more obvious :) Thanks 2011-10-27T15:05:14 *** Zannick changes topic to "AI Challenge (sponsored by Google): http://aichallenge.org || Channel Logs: http://contestbot.aichallenge.org || Code Repo: http://github.com/aichallenge/aichallenge || About the game (Ants): http://aichallenge.org/problem_description.php || Submission deadline: December 18 11:59 EST" 2011-10-27T15:05:26 *** Zannick changes topic to "AI Challenge (sponsored by Google): http://aichallenge.org || Channel Logs: http://contestbot.aichallenge.org || Code Repo: http://github.com/aichallenge/aichallenge || About the game (Ants): http://aichallenge.org/problem_description.php || Submission deadline: December 18 23:59 EST" 2011-10-27T15:05:41 *** Zannick sets mode: -o Zannick 2011-10-27T15:06:39 *** dwins has quit IRC (Ping timeout: 252 seconds) 2011-10-27T15:06:40 wow, a game on the front page had an ant in second place time out 2011-10-27T15:06:46 and all the ants stopped moving 2011-10-27T15:07:14 ... 2011-10-27T15:07:28 it was surreal 2011-10-27T15:07:30 oh, we don't get to watch the final contest as it happens? 2011-10-27T15:07:39 cyphase: don't we? 2011-10-27T15:07:42 I think we probably do 2011-10-27T15:07:43 cyphase: no, we do. 2011-10-27T15:07:47 just 2011-10-27T15:07:52 ALL results available afterwards 2011-10-27T15:07:53 you just can't submit new versions of your bot after the deadline 2011-10-27T15:07:54 "Upon completion the contest winner will be announced and all results will be publically available." 2011-10-27T15:07:57 or, the games will be preserved 2011-10-27T15:08:09 (at least I think that's their meaning) 2011-10-27T15:08:15 hmm 2011-10-27T15:09:32 *** dwins has joined #aichallenge 2011-10-27T15:10:20 why does that sound like a profet prediction? and fish will rain for the sky, everyone will rejoice 2011-10-27T15:12:17 in the past we have at least partially hidden the final results at some point, we'll probably do the same again in some form this time. Although I hope to add a little twist to what we've done before but we'll see what happens. 2011-10-27T15:12:55 *** sunshowers has joined #aichallenge 2011-10-27T15:13:54 *** SFX has joined #aichallenge 2011-10-27T15:14:16 oh, noo, that was pretty boring in PW! all of us 'others were kept out, and could not even SEE, what's going on there ! 2011-10-27T15:14:37 for all of ~12? hours :P 2011-10-27T15:14:45 *** sunshowers has quit IRC (Client Quit) 2011-10-27T15:14:46 hmm 2011-10-27T15:14:50 I need to up my game 2011-10-27T15:14:55 study path finding algorithms 2011-10-27T15:15:17 I want to make it to the top 100 damnit. Even if I have to cheat to get there 2011-10-27T15:15:22 well I'm not completely serious 2011-10-27T15:15:31 :) 2011-10-27T15:15:37 but I did have the idea of mimicking the behaviour of other bots 2011-10-27T15:15:52 (that is, the bot observes other bots and copies them) 2011-10-27T15:17:07 That's hard to do, since you already have limited information in the first place. 2011-10-27T15:17:38 and your presence and actions might change their actions 2011-10-27T15:17:41 heisenberg :P 2011-10-27T15:17:48 ^+1 2011-10-27T15:18:00 what if someone else has the same idea 2011-10-27T15:18:04 and they copy you copying them 2011-10-27T15:18:09 while you copy them copying you 2011-10-27T15:18:14 Zannick: it's bots all the way down 2011-10-27T15:18:27 stop copying us 2011-10-27T15:18:28 hehe 2011-10-27T15:18:56 stop copying us 2011-10-27T15:19:18 ajf: you'll always be a turn behind 2011-10-27T15:19:26 ajf: so he'll eat you first 2011-10-27T15:20:23 haha, 2011-10-27T15:20:34 haha 2011-10-27T15:20:35 true 2011-10-27T15:20:36 i remember making copybot way back at the beginning of planet wars 2011-10-27T15:20:41 *** spect_ has joined #aichallenge 2011-10-27T15:20:42 it was always a turn behind 2011-10-27T15:20:47 and occasionally it won 2011-10-27T15:22:33 *** dmiles_afk has joined #aichallenge 2011-10-27T15:26:37 ..ab.. 2011-10-27T15:26:44 can these move together? i.e. 2011-10-27T15:26:50 ..ab.. -> ...ab. 2011-10-27T15:26:57 *** SharkMonkey has joined #aichallenge 2011-10-27T15:27:02 GarfTop: yeas 2011-10-27T15:27:04 just 2011-10-27T15:27:09 move the right one first 2011-10-27T15:27:14 else they'll bump into eachother 2011-10-27T15:27:22 aha, so order ordering matters? 2011-10-27T15:27:28 i don't think it does 2011-10-27T15:27:34 GarfTop: how about trying it 2011-10-27T15:27:43 actually, good point, it might not 2011-10-27T15:28:13 I have a rule "dont move to occupied square", but this is too simple then 2011-10-27T15:28:53 http://aichallenge.org/visualizer.php?game=20724&user=280 <- meh, check that food queue 2011-10-27T15:30:00 interesting strategy, there 2011-10-27T15:30:06 lol 2011-10-27T15:30:22 "haha, he has no ants!" 2011-10-27T15:30:26 *swarmed* 2011-10-27T15:30:35 really, halfway through, when it was 1000 ants, it should just spam them and overrun everyone 2011-10-27T15:30:52 it would still take 500 turns to have 500 ants ;-) 2011-10-27T15:31:08 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T15:31:08 exactly 2011-10-27T15:31:43 *** danielharan has joined #aichallenge 2011-10-27T15:31:53 oh well, its enough to survive the game 2011-10-27T15:31:55 I've tried C, I've tried Python 2011-10-27T15:32:02 in case you are attacked or something 2011-10-27T15:32:04 I think I shall try C# again 2011-10-27T15:32:11 ajf: why? 2011-10-27T15:32:38 amstan: Speed 2011-10-27T15:32:43 I don't want to use C++ 2011-10-27T15:32:47 and C was just too low level 2011-10-27T15:32:52 :-) 2011-10-27T15:32:58 *** savaron has joined #aichallenge 2011-10-27T15:33:00 its still high enough 2011-10-27T15:33:06 and I don't like Java. So C# it is 2011-10-27T15:33:19 a lot of bots just go back and forth :/ 2011-10-27T15:33:23 you have to code much code in other languages as well 2011-10-27T15:33:26 just build layers 2011-10-27T15:33:57 ajf: LUA? 2011-10-27T15:34:06 janzert: lol 2011-10-27T15:34:11 :) 2011-10-27T15:34:16 avdg: I did 2011-10-27T15:34:23 i don't think we can handle multilanguage bots 2011-10-27T15:34:26 or polyglots 2011-10-27T15:34:48 if you want speed and not so low level though Lua might be a way to go 2011-10-27T15:34:57 it would probably use multiple threads anyway ;-) 2011-10-27T15:35:20 unless its a language in a language 2011-10-27T15:36:05 It should be possible to submit bots written in assembly language 2011-10-27T15:36:28 the only lua contestant in PW was actually a c one, supplying compiler and script.. 2011-10-27T15:38:36 I think a lot of people prototype in a higher level language (i.e. Python) and then do the final bot in a lower level language once they've optimized everything. 2011-10-27T15:38:58 not necessarily 2011-10-27T15:39:26 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-10-27T15:39:45 *** g0llum has joined #aichallenge 2011-10-27T15:39:58 meh, that would probably involve a lot of code, unless there is a way to do it with less effort 2011-10-27T15:41:47 Darrow: that's how I did it 2011-10-27T15:41:51 was not a good idea 2011-10-27T15:42:06 in fact, my bad C abilitiy meant it ran SLOWER 2011-10-27T15:42:29 Well, CPython has one of the most highly optimized list and hashtable implementations out there 2011-10-27T15:42:49 Rewriting your bot in C with your own lists will not end well for performance 2011-10-27T15:42:54 ajf: so.. use the cpython list 2011-10-27T15:42:58 some people may just start in a lower level language 2011-10-27T15:43:16 some people might not bother to optimize in the higher language 2011-10-27T15:43:48 or sometimes the optimization in the higher level language is fine 2011-10-27T15:43:50 I'm kinda relying that references are cheap 2011-10-27T15:44:01 maybe a bad assertion 2011-10-27T15:44:32 but oh well, most times probably better than full copie 2011-10-27T15:44:33 s 2011-10-27T15:44:38 I'm using D and just hoping it won't break too horribly 2011-10-27T15:44:48 *** dvladim has quit IRC (Read error: Operation timed out) 2011-10-27T15:45:23 I'm not that experienced in js, but thats what makes the implementation more fun :-) 2011-10-27T15:45:39 mleise will be glad ;) 2011-10-27T15:46:35 avdg: making full use of closures I hope? 2011-10-27T15:46:52 when I discovered they existed I went "fuck, I've being doing JS wrong all along" 2011-10-27T15:46:59 I currently have only a hand full of them 2011-10-27T15:47:10 (only for events) 2011-10-27T15:47:27 *** ccc has quit IRC (Ping timeout: 265 seconds) 2011-10-27T15:47:56 events are a gift, though over using it is very simple 2011-10-27T15:48:07 and was in fact my very first mistake in v1 2011-10-27T15:49:56 Ok guys I had a problem that the file "play_one_game.cmd" would run, open the console, and immediately close again. So I tried to run it from console, but then it said "Python is not recognised as an internal or external assignment, program or batchfile" So I tried copying the file to the folder where I installed python, and then tried this: c:\Python27>play_one_game.cmd python: can't open file 'c:\Python27\playgame.py': [Er 2011-10-27T15:50:01 So I copied that file into my python folder too, resulting in this: c:\Python27>play_one_game.cmd Traceback (most recent call last): File "c:\Python27\playgame.py", line 10, in import visualizer.visualize_locally ImportError: No module named visualizer.visualize_locally. Any ideas? 2011-10-27T15:50:35 *** McLeopold has joined #aichallenge 2011-10-27T15:50:39 xcrypt: don't copy the file in the python folder 2011-10-27T15:50:44 xcrypt: add python to your path instead 2011-10-27T15:52:37 how can I translate 2011-10-27T15:52:38 public class Prioritized

, T> implements Comparable> { 2011-10-27T15:52:40 to c# ? 2011-10-27T15:52:47 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-10-27T15:53:00 class Prioritized where P : IComparable

, but I cant add a interface to prioritized 2011-10-27T15:54:03 *** ChrisH_ has joined #aichallenge 2011-10-27T15:54:11 *** g0llum has joined #aichallenge 2011-10-27T15:56:36 no one? ;) 2011-10-27T15:57:43 amstan: I now copied "Python27" folder to C:\AI Challenge\tools 2011-10-27T15:58:14 amstan: still get the same error: python is not recognised as an internal or external assignment, program or batchfile 2011-10-27T15:58:50 *** Accoun has quit IRC () 2011-10-27T15:59:02 xcrypt: you shouldn't be moving any folders 2011-10-27T15:59:04 http://www.computerhope.com/issues/ch000549.htm 2011-10-27T15:59:05 xcrypt, leave everything as is, and open a cmdline, then CD to the ants folder, and restart from there .. 2011-10-27T15:59:13 g0llum: he doesn't have python in path 2011-10-27T15:59:25 ahh.. 2011-10-27T16:00:18 damn I suck at this :p I'll try this give me a min :) 2011-10-27T16:01:35 *** DarthJeos has joined #aichallenge 2011-10-27T16:01:36 aichallenge: McLeopold epsilon * r15df9db / (website/check_submit.php website/submit.php): changed some verbage - http://git.io/66ey4g 2011-10-27T16:01:48 hi all 2011-10-27T16:02:11 i have a question 2011-10-27T16:02:29 sorry for my bad english, i'm from russia 2011-10-27T16:02:48 DarthJeos: so far i couldn't tell 2011-10-27T16:02:53 DarthJeos: what do you need? 2011-10-27T16:03:25 i can't run my bot in sandbox. 2011-10-27T16:03:57 i use c++ and compile his in .exe 2011-10-27T16:06:17 how i should change file play_one_game.cmd in tools for this? 2011-10-27T16:06:38 yes, you can change play_one_game.cmd 2011-10-27T16:06:59 "When an e-mail begins "I'm from Germany, so please excuse my somewhat imperfect English," you're about to get schooled on proper grammar." -- https://twitter.com/#!/BenKuchera/status/128115671249846272 2011-10-27T16:07:05 *** dmiles_afk has quit IRC (Ping timeout: 276 seconds) 2011-10-27T16:08:38 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-10-27T16:08:47 how can I do this? 2011-10-27T16:08:48 @amstan: Okay. Remember my previous problem? I think it's solved now. However, I also think I've overwritten the default path, I suppose that's a bad thing? :p 2011-10-27T16:08:49 xcrypt: You have no gotten any error messages recently, so here's a random one just to let you know that we care. 2011-10-27T16:09:04 xcrypt: yes, that's very bad 2011-10-27T16:09:26 xcrypt: i suggest that you don't reboot until you fix it 2011-10-27T16:09:44 find a sample path for your os on google, and copy it 2011-10-27T16:09:59 thanks 2011-10-27T16:10:06 *** dmiles_afk has joined #aichallenge 2011-10-27T16:10:45 >.< 2011-10-27T16:11:16 I did it! 2011-10-27T16:11:25 grats! 2011-10-27T16:11:28 but python return error: 2011-10-27T16:11:55 *** dwins has quit IRC (Ping timeout: 252 seconds) 2011-10-27T16:12:42 Failed to start ['tools/csharp/myBot.exe'] 2011-10-27T16:13:56 Can not find the file specified 2011-10-27T16:14:41 is the path to myBot.exe correct? 2011-10-27T16:14:55 but file are exists 2011-10-27T16:14:58 yes, correct 2011-10-27T16:15:19 i run it in command line, its work 2011-10-27T16:15:38 it seems someone before said to remove the .exe extension, no idea if that's really necessary 2011-10-27T16:15:56 *** kapoc has joined #aichallenge 2011-10-27T16:16:33 in windows, should the path be separated with backslash (\) ? for example path\to\bot.exe 2011-10-27T16:17:02 hmm, I would expect that as well 2011-10-27T16:17:17 it did not help 2011-10-27T16:17:17 janzert: ec2 prices still high? 2011-10-27T16:17:35 again not still, but yes 2011-10-27T16:17:55 and not as high, but higher than normal 2011-10-27T16:21:01 okay, remove ".exe" help. thanks 2011-10-27T16:23:01 *** danielharan has quit IRC (Ping timeout: 240 seconds) 2011-10-27T16:23:05 *** Accoun has joined #aichallenge 2011-10-27T16:23:25 amstan: I think I restored my default system paths now and added the path to python27. Is there any 'safe' way to be sure? 2011-10-27T16:23:46 master_ninja: depends on where, you usually can't go wrong with backslash, but python commands will accept forward as well 2011-10-27T16:24:27 great I'm not the only one having problems with python :p 2011-10-27T16:24:39 xcrypt: you could send it to us 2011-10-27T16:25:27 Variable Name: Path 2011-10-27T16:25:38 *** DarthJeos has quit IRC (Ping timeout: 265 seconds) 2011-10-27T16:25:46 Value: C:\Windows\system32;C:\Windows;C:\Windows\System32\Wbem;C:\Python27 2011-10-27T16:26:21 but I'm running windows 7 64 bit, does it have the same default values than windows 7 32 bit? 2011-10-27T16:26:26 *** Fandekasp has quit IRC (Ping timeout: 258 seconds) 2011-10-27T16:26:33 xcrypt: i'm not sure 2011-10-27T16:26:42 http://en.wikipedia.org/wiki/Environment_variable#Default_Values_on_Microsoft_Windows 2011-10-27T16:26:53 doesn't seem to differentiate it 2011-10-27T16:27:03 *** olexs has joined #aichallenge 2011-10-27T16:28:03 *** Fandekasp has joined #aichallenge 2011-10-27T16:29:04 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-10-27T16:29:43 *** overburn has joined #aichallenge 2011-10-27T16:30:03 *** dwins has joined #aichallenge 2011-10-27T16:30:35 small question 2011-10-27T16:30:37 hey guys 2011-10-27T16:30:44 amstan: well I checked the paths and the all exist. Should be good now, no? 2011-10-27T16:30:48 does anyone know how i can run the php version of the bot? 2011-10-27T16:30:50 hey overburn :) 2011-10-27T16:30:55 i'm slightly at a loss 2011-10-27T16:31:30 xcrypt: you might need more for the 64bit version 2011-10-27T16:33:06 I'll do some more research then :) What's the worst thing that could happen? Forced to format my pc? =/ 2011-10-27T16:33:10 i mean , i can run the php but i have no idea how to integrate it with the python game code 2011-10-27T16:33:48 Here's what seems to be my system related path components on win7 x64 2011-10-27T16:33:50 %SystemRoot%\system32;%SystemRoot%;%SystemRoot%\System32\Wbem;%SYSTEMROOT%\System32\WindowsPowerShell\v1.0\; 2011-10-27T16:33:54 so any tips on what arguments to use on the playgame.py to use the php version? 2011-10-27T16:34:10 *** danielharan has joined #aichallenge 2011-10-27T16:34:15 overburn: php yourbot.php? 2011-10-27T16:34:20 overburn: assuming php is in your path 2011-10-27T16:34:24 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T16:35:04 aah path 2011-10-27T16:35:06 forgot about that 2011-10-27T16:35:08 thanks mate 2011-10-27T16:35:09 heh 2011-10-27T16:35:10 xcrypt: see above ^^ 2011-10-27T16:35:32 thanks! 2011-10-27T16:36:38 Do I have to replace %SystemRoot% by something? I'm kinda noobish when it comes to that sorry 2011-10-27T16:38:08 no that's expanded by windows itself 2011-10-27T16:38:33 ah I see thanks 2011-10-27T16:38:46 I'll restart my computer now and hope it doesn't explode :D 2011-10-27T16:38:54 good luck 2011-10-27T16:39:01 thx ^^ 2011-10-27T16:39:41 if all you did was change the PATH variable you should only need to restart the cmd window :/ 2011-10-27T16:40:19 cmd always worked 2011-10-27T16:40:27 even when I've overwritten the default path 2011-10-27T16:40:31 xcrypt: do you have a cmd window from before you overwrote it? 2011-10-27T16:40:37 *** chris_0076 has quit IRC (Quit: Leaving) 2011-10-27T16:40:41 no 2011-10-27T16:40:45 ok, nvm 2011-10-27T16:41:20 so, everythings fine right? :) 2011-10-27T16:41:33 also I've been assuming you changed the system wide path, under system properties->advanced->environment variables? 2011-10-27T16:41:51 *** Fandekasp has quit IRC (Read error: Operation timed out) 2011-10-27T16:42:30 *** Macuyiko has quit IRC (Read error: Connection reset by peer) 2011-10-27T16:42:34 *** kapoc has left #aichallenge 2011-10-27T16:42:42 if not then you're fine for sure because the system wide path won't have changed at all 2011-10-27T16:42:46 janzert: can't find a variable with such a name 2011-10-27T16:43:14 I'm describing a dialog -> tab -> button above :} 2011-10-27T16:43:30 *** tarac0 has joined #aichallenge 2011-10-27T16:43:36 I guess my question is how did you change the path? 2011-10-27T16:44:00 wow, you're still dealing with path issues? 2011-10-27T16:44:32 system properties->advanced->environment variables->Variables 2011-10-27T16:44:38 btw, workers are trying to come back up but ubuntu's package download is really slow at the moment :( 2011-10-27T16:44:43 I found 'Path' there and clicked 'edit' 2011-10-27T16:44:49 xcrypt: ok, yeah that's the system wide path then 2011-10-27T16:44:50 *** SFX has quit IRC (Quit: Page closed) 2011-10-27T16:44:58 *** spect_ has quit IRC (Ping timeout: 265 seconds) 2011-10-27T16:45:11 xcrypt: get this exe http://download.cnet.com/Redmond-Path/3000-2094_4-10811594.html 2011-10-27T16:45:17 janzert: it seems like some package servers throttle the downloads 2011-10-27T16:45:18 it makes path stuff easier to do 2011-10-27T16:45:25 Actually I clicked 'new...' because I assumed it would make a new path or something 2011-10-27T16:45:33 but it overwrote... 2011-10-27T16:45:41 :( 2011-10-27T16:45:41 janzert: are workers crashing, or is this just an automatic shutdown due to price? 2011-10-27T16:45:52 it was an automatic shutdown 2011-10-27T16:45:58 *** skyramp has joined #aichallenge 2011-10-27T16:46:24 *** tarac0 has quit IRC (Client Quit) 2011-10-27T16:46:38 xcrypt: you're probably as good to go as we can get you :/ 2011-10-27T16:47:07 okay thanks for all the help guys :) 2011-10-27T16:50:22 amstan: it's using us.archive.ubuntu.com which I assume is a dns roundrobin, but 3 different workers are all having the same slow speed 2011-10-27T16:50:42 one of which I restarted once already 2011-10-27T16:50:46 *** dmiles_afk has quit IRC (Ping timeout: 258 seconds) 2011-10-27T16:50:58 janzert: yeah, the us|ca.archive.ubuntu.com are usually slow, i change mine to mirror.csc.uwaterloo.ca as soon as i get the chance 2011-10-27T16:50:59 janzert: could we make a giant master tarball that they get from the server? 2011-10-27T16:51:28 McLeopold: I'll probably make a special AMI 2011-10-27T16:51:47 the setup script is already half way there to do that 2011-10-27T16:52:11 actually the setup script might be all the way there 2011-10-27T16:52:38 have to see if it's doing the right thing with the chroot 2011-10-27T16:53:55 *** master_ninja has quit IRC (Quit: the day ends but the code continues~~) 2011-10-27T16:55:44 *** Chris_0076 has joined #aichallenge 2011-10-27T16:57:34 *** thestinger has joined #aichallenge 2011-10-27T16:57:45 *** ztfw has quit IRC (Ping timeout: 252 seconds) 2011-10-27T16:58:16 *** bearoff has joined #aichallenge 2011-10-27T16:58:38 if i modify the ants.php script, will everything still run on the autocompile? 2011-10-27T16:58:52 *** paulwal has joined #aichallenge 2011-10-27T16:59:05 modifying anything is fine 2011-10-27T16:59:08 *** gfreferferferfr has joined #aichallenge 2011-10-27T16:59:15 ah perfect thanks Zannick :D 2011-10-27T16:59:18 from the start, you mean? 2011-10-27T16:59:22 starter * 2011-10-27T16:59:29 yeah 2011-10-27T16:59:35 only the starter 2011-10-27T17:00:02 yeah, the starter is just a reference point. you can start from scratch if you're so inclined, but the starter packs make it easy to get started 2011-10-27T17:00:18 i see 2011-10-27T17:00:19 ah \ 2011-10-27T17:00:22 umm anyone tried the php version? 2011-10-27T17:00:26 i have a weird problem 2011-10-27T17:00:35 that has nothing to do with my code and i can't pinpoint the cause :( 2011-10-27T17:00:41 hmm 2011-10-27T17:00:55 *** choas has quit IRC (Ping timeout: 260 seconds) 2011-10-27T17:01:15 aka i try to check if the next move tile is unoccupied and it always outputs -2 2011-10-27T17:01:19 even if there's an ant on it 2011-10-27T17:01:35 *** onensora has quit IRC (Ping timeout: 248 seconds) 2011-10-27T17:01:48 *** Ttech has joined #aichallenge 2011-10-27T17:02:03 overburn: maybe you should check the behavior of the map? 2011-10-27T17:03:21 oh 2011-10-27T17:03:27 which file is that? 2011-10-27T17:04:46 *** onensora has joined #aichallenge 2011-10-27T17:05:27 >> and it always outputs -2 what are you say about? 2011-10-27T17:06:14 *** gfreferferferfr has quit IRC (Ping timeout: 265 seconds) 2011-10-27T17:06:45 ok so the unoccupied function checks if the tile i want to move to is LAND or DEAD , LAND having the value -2 2011-10-27T17:07:21 and it always sees the -2 even if there's an ant on it 2011-10-27T17:07:22 grr 2011-10-27T17:08:10 unoccupied function returns boolean value 2011-10-27T17:08:21 yeah sorry 2011-10-27T17:08:37 i meant the internal value of the tile that is being checked 2011-10-27T17:08:49 ah 2011-10-27T17:09:11 and if i remove every condition from the issueOrder function 2011-10-27T17:09:16 it still avoids water 2011-10-27T17:09:17 O_o 2011-10-27T17:09:29 is it your ant? 2011-10-27T17:09:57 yeap 2011-10-27T17:10:35 have you modify update() function? 2011-10-27T17:10:43 amstan: can I put an embedded map in the forums? 2011-10-27T17:11:29 McLeopold: i'm not sure, depends on the visualizer 2011-10-27T17:11:48 By the way. I read in the rules that I cannot have two accounts. If I want an account for me personally, and one for my team, is that okay? 2011-10-27T17:11:56 McLeopold: you have to give it valid json i think 2011-10-27T17:12:04 amstan: or map text 2011-10-27T17:12:11 or the live stream 2011-10-27T17:12:30 visualizer.loadReplayData(data); 2011-10-27T17:12:35 so, if the [visualizer] tag could take a map file, that would be nice 2011-10-27T17:12:39 yes 2011-10-27T17:12:42 where data is between the visualizer tags 2011-10-27T17:12:46 yep 2011-10-27T17:12:53 or make a separate [map] tag 2011-10-27T17:12:57 or [ants] 2011-10-27T17:12:58 McLeopold: it's already done 2011-10-27T17:13:04 [visualizer]GameID or GameURI or GameData(paste directly)[/visualizer 2011-10-27T17:13:08 *** danielharan has joined #aichallenge 2011-10-27T17:13:27 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T17:13:55 http://aichallenge.org/forums/viewtopic.php?f=21&t=1636&p=9972#p9972 2011-10-27T17:14:00 *** danielharan has joined #aichallenge 2011-10-27T17:14:20 McLeopold: are you sure you can just give the map file to the visualizer? 2011-10-27T17:14:58 that what I do in the wiki 2011-10-27T17:15:25 McLeopold, are you an AI Challenge admin? 2011-10-27T17:15:32 yep 2011-10-27T17:15:43 McLeopold: what about addParam('replay_string', data);? 2011-10-27T17:15:54 umm.... 2011-10-27T17:17:05 oh, speaking of admins... 2011-10-27T17:17:36 nope, just vis.loadReplayData(data); 2011-10-27T17:17:43 https://github.com/aichallenge/aichallenge/blob/epsilon/website/visualizer_widget.php#L89 2011-10-27T17:18:04 well we had our first registration email marked as spam by the user receiving it 2011-10-27T17:18:15 *** amadoru has joined #aichallenge 2011-10-27T17:18:20 I just finished a starter bot for Tcl. Should I fork the repo, add the starter package, edit worker/compiler.py, then make a pull request? Then will it be added to the webpage of starter packages? 2011-10-27T17:18:34 janzert: oops :/ 2011-10-27T17:18:43 amstan: Is it okay to have one account for me personally, and one account for my team? If not: how do I delete my current account? 2011-10-27T17:18:53 xcrypt: link me 2011-10-27T17:19:27 I don't really want it deleted right now it was just a question 2011-10-27T17:19:44 xcrypt: ideally you should only have one 2011-10-27T17:19:54 guys, are the servers taking forever to do a game run or is it just me? 2011-10-27T17:19:56 the answer is no, you are either on a team of more than one person or on a team of one 2011-10-27T17:20:11 amadoru: nope, they are taking forever 2011-10-27T17:21:13 *** fenec has quit IRC (Ping timeout: 265 seconds) 2011-10-27T17:21:23 amstan: damn! i think it's been about 8 hours since the last run. This is heartbreaking and annoying. 2011-10-27T17:21:43 amstan: okay, I will ask an admin to delete my account when I get more info from my team partners. I will just keep it for now to experiment a bit :) 2011-10-27T17:21:47 Thanks 2011-10-27T17:21:51 amadoru: sorry about that, but we do have limited resources 2011-10-27T17:22:07 xcrypt: ok, that's fine, as long as you have your intentions clear 2011-10-27T17:23:04 @amstan: understandable. Thanks for the info. 2011-10-27T17:23:05 amadoru: User error -- Replace user. 2011-10-27T17:23:26 amstan: understandable. Thanks for the info. 2011-10-27T17:23:28 @tcp 2011-10-27T17:23:29 Zannick: tcp could be http://ants.fluxid.pl/howto. 2011-10-27T17:23:37 amadoru: you could try this ^ 2011-10-27T17:23:45 amadoru: yes, the tcp server is a great way to get a ton of games 2011-10-27T17:23:46 This project is sponsored by google though, how can you have problems with resources? 2011-10-27T17:24:24 xcrypt: they don't write blank checks, we have limited resources, and this is one of the games which is really a resource hog 2011-10-27T17:24:58 we don't use google's resources 2011-10-27T17:26:05 ok thanks. I'll use tcp server in future 2011-10-27T17:26:26 *** ltriant has joined #aichallenge 2011-10-27T17:26:53 McLeopold: so why doesn't it work? 2011-10-27T17:27:03 amstan: thanks. I'll give a try with the tcp server. 2011-10-27T17:27:19 wtf... 2011-10-27T17:27:25 *** ajhager has quit IRC (Quit: ajhager) 2011-10-27T17:27:29 the php version sets the map as all data equaling land 2011-10-27T17:27:33 :)) 2011-10-27T17:27:35 no wonder 2011-10-27T17:27:40 zannick: thanks. I'll give a try 2011-10-27T17:28:13 how do i get the map from the program? 2011-10-27T17:28:55 *** dauryg has quit IRC (Quit: Page closed) 2011-10-27T17:29:12 *** bishoco has quit IRC (Quit: Page closed) 2011-10-27T17:29:17 *** xcrypt has quit IRC (Quit: Page closed) 2011-10-27T17:30:10 Zannick, what does google sponsor if not resources? 2011-10-27T17:30:23 ikaros: i agree, money is considered a resource 2011-10-27T17:30:30 we don't use google's machines. that's what i meant 2011-10-27T17:30:44 hehe ok 2011-10-27T17:31:18 right now there are only 2 server active? do i get that page right: http://aichallenge.org/server_stats.php 2011-10-27T17:31:34 yes 2011-10-27T17:31:41 :/ 2011-10-27T17:31:49 ikaros: but they are 8 core servers with 12GB of ram each 2011-10-27T17:31:56 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T17:32:17 maybe but some hours back there were 5 or so right? 2011-10-27T17:32:29 depends on the instance price 2011-10-27T17:32:33 hmm, can that number raise above 5 if the price is dropped? 2011-10-27T17:32:37 *** danielharan has joined #aichallenge 2011-10-27T17:32:39 avdg: yes 2011-10-27T17:32:56 everywhere is a stockmarket... 2011-10-27T17:33:02 depressing 2011-10-27T17:33:52 well, as long your software and data scales, its easy to have a lot of workers 2011-10-27T17:33:59 and data is big business ;-) 2011-10-27T17:34:27 avdg: yes 2011-10-27T17:34:36 amstan: how huge is server load? 2011-10-27T17:35:04 Fluxid: hold on a sec 2011-10-27T17:35:04 *** Larose has quit IRC (Quit: Leaving.) 2011-10-27T17:35:53 i'm confused. if that's 2 server with 8 cores each.. how come they only play ~1 game per minute? shouldn't that be much more? 2011-10-27T17:35:56 i guess one cannot "donate server" because of ability to look into code of running codes? 2011-10-27T17:36:05 2 cores 2011-10-27T17:36:29 Fluxid: yes and the need for servers to match specs 2011-10-27T17:36:44 janzert: it is? i tought we were running each process on its own core 2011-10-27T17:36:52 janzert: or do we do it sequencially? 2011-10-27T17:36:54 amstan: we do 2011-10-27T17:36:58 2 at a time 2011-10-27T17:37:02 ic 2011-10-27T17:37:03 1 for each core 2011-10-27T17:37:15 hyperthreating? 2011-10-27T17:37:37 or am I confused? 2011-10-27T17:37:46 what on earth 2011-10-27T17:37:50 Fluxid: load average: 0.39 0.30 0.41 2011-10-27T17:37:56 the php starter seems to be useless 2011-10-27T17:37:57 O_O 2011-10-27T17:37:58 it's not that important but this page http://aichallenge.org/game_settings.php states turn limit is 1500 but i have the impression it's 1000 2011-10-27T17:38:19 amstan: that's low! 2011-10-27T17:38:28 maybe because its idling? 2011-10-27T17:38:30 Fluxid: yeah, i'm not sure why it's not higher 2011-10-27T17:38:44 janzert: shouldn't at least one core be at 100%? 2011-10-27T17:39:03 aichallenge: janzert epsilon * r42e0504 / (setup/chroot_configs/sources.list setup/worker_setup.py): Set chroot package source to ec2 mirror - http://git.io/e-PLeA 2011-10-27T17:39:11 amstan: generally it isn't 2011-10-27T17:39:13 the python bots seem to only use 20% when working 2011-10-27T17:40:01 most bots don't use cpu long enough to push it to 100%, especially the starter bots 2011-10-27T17:40:09 ic 2011-10-27T17:40:45 would it hurt to try running more games at once? 2011-10-27T17:40:46 it's why last contest danielvf decided to run 2 bots per core 2011-10-27T17:41:37 Fluxid: bots get the short end of the stick if they get stuck running with another high cpu usage bot 2011-10-27T17:41:49 so in order to be accepted, does the bot file need to be called MyBot, or does it get recognized automatically? 2011-10-27T17:42:00 has to be MyBot. 2011-10-27T17:42:11 okay thanks mate 2011-10-27T17:42:23 janzert: ah, i get the problem - timeouts 2011-10-27T17:42:26 2 bots per core wouldn't be fair with a round timeout i think 2011-10-27T17:42:26 the compiler's not sophisticated enough to find the entry point on its own 2011-10-27T17:42:36 janzert: but i don't get it, what the cpu doing for the other time when it's idle 2011-10-27T17:42:56 waiting for the engine or other bot 2011-10-27T17:43:24 janzert: but if it was waiting for the other bot, shouldn't the other core be at 100%? 2011-10-27T17:43:27 *** asdasd has joined #aichallenge 2011-10-27T17:43:28 being fair is suffering 2011-10-27T17:43:46 Fluxid: you should see what happens when things are unfair, it's like a war story 2011-10-27T17:44:02 #OccupyAIChallengeForums 2011-10-27T17:44:10 lol 2011-10-27T17:44:14 lol 2011-10-27T17:44:30 yeah, so probably waiting for engine/sandbox 2011-10-27T17:44:46 but engine is also doing nothing 2011-10-27T17:44:53 or only 20% 2011-10-27T17:45:06 umm guys 2011-10-27T17:45:07 question 2011-10-27T17:45:12 where do i find the format of the input data? 2011-10-27T17:45:25 overburn: specification.php 2011-10-27T17:45:32 overburn: http://aichallenge.org/specification.php#Bot-Input 2011-10-27T17:45:40 probably memory paging into cache 2011-10-27T17:45:43 http://aichallenge.org/specification.php 2011-10-27T17:45:46 ah 2011-10-27T17:46:02 ah umm 2011-10-27T17:46:05 so there it was 2011-10-27T17:46:08 : D thanks 2011-10-27T17:48:27 *** bearoff has left #aichallenge 2011-10-27T17:49:28 *** amadoru has quit IRC (Quit: Page closed) 2011-10-27T17:50:23 McLeopold: ping 2011-10-27T17:51:11 sup 2011-10-27T17:51:21 does anyone know what's happening to the games per minute rate ? 2011-10-27T17:51:36 do you think you'll use that other server for testing or should I just take it down? 2011-10-27T17:52:04 take it down 2011-10-27T17:52:10 thanks 2011-10-27T17:52:32 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T17:53:07 *** danielharan has joined #aichallenge 2011-10-27T17:55:03 *** george has quit IRC (Ping timeout: 265 seconds) 2011-10-27T17:55:09 *** SITZ has quit IRC (Quit: Leaving) 2011-10-27T17:55:20 *** teryret has joined #aichallenge 2011-10-27T17:55:43 have any of you guys had to use the toolkit's profiler? 2011-10-27T17:56:23 *** olexs has quit IRC (Quit: Leaving.) 2011-10-27T17:57:18 *** asdasd has quit IRC (Quit: Page closed) 2011-10-27T17:59:38 s 2011-10-27T17:59:43 @help 2011-10-27T17:59:43 paulwal: (help [] []) -- This command gives a useful description of what does. is only necessary if the command is in more than one plugin. 2011-10-27T17:59:43 *** Harpyon has quit IRC (Quit: Computer has gone to sleep.) 2011-10-27T18:00:13 *** Eruonen has joined #aichallenge 2011-10-27T18:00:29 *** toopp has joined #aichallenge 2011-10-27T18:00:41 hey contestbot, what can you tell me about how to interpret the results of running a game wiht the --profile flag 2011-10-27T18:01:00 *** onensora has quit IRC (Ping timeout: 260 seconds) 2011-10-27T18:01:06 contestbot will understand that ;-) 2011-10-27T18:01:07 avdg: User error -- Replace user. 2011-10-27T18:01:15 hahaha 2011-10-27T18:03:42 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-10-27T18:04:14 the maps have different kinds of symmetries, that's confusing. or at least very hard to take use of 2011-10-27T18:04:33 HaraKiri: don't rely on the map 2011-10-27T18:04:54 it only produces stupid bots 2011-10-27T18:05:03 on on those particular maps, but you can use the general information that maps have a symmetry 2011-10-27T18:05:27 you know what else produces stupid bots? A* searches that inexplicably take forever but don't show up on profiler runs... 2011-10-27T18:05:28 someone here considered using fourier tranformations and other method to predict enemy hives... 2011-10-27T18:06:07 if can't see how fourier would help. but hive positions are doable once you know the position of one other hive, and the map symmetry 2011-10-27T18:06:24 what is the meaning of game rate column in my profile statistics page? 2011-10-27T18:06:37 *** ajf is now known as ajf|offline 2011-10-27T18:06:43 that's the rate at which they are playing games on our behalf 2011-10-27T18:06:59 almost nothing, except how good you are compared with others in solving the same problem ;-) 2011-10-27T18:07:06 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T18:07:46 *** danielharan has joined #aichallenge 2011-10-27T18:08:53 oh, that was my other question 2011-10-27T18:09:12 are any of you guys using the time_remaining() function they provided in ants.py? 2011-10-27T18:09:34 you can use that 2011-10-27T18:09:48 I'm trying to, yes, but it's giving me wierd results 2011-10-27T18:10:09 telling me I have 800ms to go just before I time out 2011-10-27T18:10:24 *** xar0l has joined #aichallenge 2011-10-27T18:10:43 teryret: you can't rely on that 100%, since your bot is not the only thing executing on the cpu, keep into consideration the operating system 2011-10-27T18:11:06 teryret: so make your give up threshold to be much earlier 2011-10-27T18:11:29 ... my give up threshhold should be in the first 200 of 1000 ms? 2011-10-27T18:11:36 that doesn't sound right 2011-10-27T18:11:46 how about 700ms? 2011-10-27T18:11:47 the server sets it on 500 2011-10-27T18:11:51 oh.. that too 2011-10-27T18:11:52 lol 2011-10-27T18:11:54 400ms? 2011-10-27T18:12:13 * avdg is going for 300ms 2011-10-27T18:12:16 amstan: more likely is that the bot is doing something for 800ms before checking time_remaining again 2011-10-27T18:12:21 * avdg can't rely his gc 2011-10-27T18:12:23 but my problem is that it's showing 800+ms remaining just before it times out 2011-10-27T18:12:28 probably stuck in a loop 2011-10-27T18:12:42 I thought of that 2011-10-27T18:12:51 so I put the check deep in the heart of A* 2011-10-27T18:13:48 Question about the server 'compiler'. When running interpreted language bots, does the compiler load all *.py or *.rb files into the interpreter and just load MyBot.extension last, or is MyBot.extension expected to load any extra source files itself? 2011-10-27T18:13:49 *** danielharan has quit IRC (Ping timeout: 240 seconds) 2011-10-27T18:14:14 the latter is correct 2011-10-27T18:14:31 ok, thanks 2011-10-27T18:14:39 at least for python, you have to say "import x" anyway 2011-10-27T18:14:43 janzert: btw, are we compiling the python stuff yet? 2011-10-27T18:14:45 *** danielharan has joined #aichallenge 2011-10-27T18:16:04 @later tell mleise can we add a food count bar in the visualizer? maybe near the hive section? 2011-10-27T18:16:04 McLeopold: Yes master! 2011-10-27T18:16:28 janzert: I'd like the food start to be changed to 4+ 2011-10-27T18:17:03 maybe even up the lower food spawn rate 2011-10-27T18:17:41 if game rate of one bot is greater than others does that mean first bot will have higher rank? 2011-10-27T18:18:14 no 2011-10-27T18:18:20 not if it loses more! 2011-10-27T18:18:25 well, maybe 2011-10-27T18:18:47 *** dmiles_afk has joined #aichallenge 2011-10-27T18:18:51 if they truely should have the same mu, the sigma would be lower for the bot with more games 2011-10-27T18:19:01 thus, a higher skill 2011-10-27T18:19:29 *** dmiles_afk has joined #aichallenge 2011-10-27T18:19:38 @later tell janzert should we use the mu, not the skill, for the final rankings? 2011-10-27T18:19:38 McLeopold: Yes master! 2011-10-27T18:20:11 *** dmiles_afk has joined #aichallenge 2011-10-27T18:20:53 *** dmiles_afk has joined #aichallenge 2011-10-27T18:21:16 * avdg ports his poor ai on top of its shiny new core :-) 2011-10-27T18:21:18 *** sdf has quit IRC (Quit: Leaving.) 2011-10-27T18:21:35 *** dmiles_afk has joined #aichallenge 2011-10-27T18:21:38 *** dmiles_afk has quit IRC (Excess Flood) 2011-10-27T18:22:17 *** dmiles_afk has joined #aichallenge 2011-10-27T18:23:51 *** dmiles_afk has joined #aichallenge 2011-10-27T18:24:21 contestbot: @botdoc 2011-10-27T18:24:22 amstan: Run as fast as you can and don't look back. 2011-10-27T18:24:28 what? 2011-10-27T18:24:32 *** toopp has quit IRC (Ping timeout: 265 seconds) 2011-10-27T18:24:34 *** dmiles_afk has joined #aichallenge 2011-10-27T18:24:36 @botdoc 2011-10-27T18:24:36 *** ls1024 has left #aichallenge 2011-10-27T18:24:36 Zannick: botdoc could be (#1) http://ubottu.com/stdin/supydocs/plugins/, or (#2) http://supybook.fealdia.org/devel/, or (#3) http://sourceforge.net/apps/mediawiki/gribble/index.php?title=Supybot_Resources. 2011-10-27T18:24:37 *** dmiles_afk has quit IRC (Excess Flood) 2011-10-27T18:24:43 oh! ok 2011-10-27T18:25:04 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T18:25:16 *** dmiles_afk has joined #aichallenge 2011-10-27T18:25:31 *** amstan sets mode: +o contestbot 2011-10-27T18:25:51 contestbot: ban add dmiles_afk 3600 2011-10-27T18:25:51 amstan: Error: You don't have the #aichallenge,op capability. If you think that you should have this capability, be sure that you are identified before trying again. The 'whoami' command can tell you if you're identified. 2011-10-27T18:25:59 *** dmiles_afk has joined #aichallenge 2011-10-27T18:26:13 contestbot: ban add dmiles_afk 3600 2011-10-27T18:26:13 amstan: Error: 'dmiles_afk' is not a valid nick or hostmask. 2011-10-27T18:26:18 darn you! 2011-10-27T18:26:20 he was just in here 2011-10-27T18:26:28 hahaha 2011-10-27T18:26:40 *** dmiles_afk has joined #aichallenge 2011-10-27T18:26:41 contestbot: ban add dmiles_afk 3600 2011-10-27T18:26:41 amstan: Ready to serve. 2011-10-27T18:26:44 *** dmiles_afk has quit IRC (Excess Flood) 2011-10-27T18:26:44 perfect 2011-10-27T18:26:46 *** TheDigitalNinja has joined #aichallenge 2011-10-27T18:27:21 amstan: does that...do anything? 2011-10-27T18:27:23 *** dmiles_afk has joined #aichallenge 2011-10-27T18:27:30 *** ChanServ sets mode: +o Zannick 2011-10-27T18:27:36 Zannick: darn, come on! 2011-10-27T18:27:41 *** HaraKiri has quit IRC () 2011-10-27T18:28:06 *** dmiles_afk has joined #aichallenge 2011-10-27T18:28:48 *** dmiles_afk has joined #aichallenge 2011-10-27T18:28:48 *** Zannick sets mode: +b *!*@dsl-173-239-80-026.cascadeaccess.com 2011-10-27T18:28:49 *** dmiles_afk was kicked by Zannick (fix your internet) 2011-10-27T18:29:11 Zannick: i was trying to do it through contestbot, so it would be temporary 2011-10-27T18:29:14 but oh well 2011-10-27T18:29:20 McLeopold: I think the final ranking should still be skill 2011-10-27T18:29:21 i used /knockout which should undo that in 3600 seconds 2011-10-27T18:29:22 *** sdf has joined #aichallenge 2011-10-27T18:29:51 sigma should be pretty similiar though so won't make a big difference 2011-10-27T18:30:24 i probably have to stay opped though 2011-10-27T18:30:38 *** Guest03238 has joined #aichallenge 2011-10-27T18:30:58 "excess flood" means he's probably spamming or being spammed 2011-10-27T18:31:03 amstan: don't think so, but it'll make such a small difference I have no motivation to do it 2011-10-27T18:31:12 Zannick: probably a bouncer setting gone wrong 2011-10-27T18:31:19 * Zannick shrug 2011-10-27T18:31:25 janzert: other languages have it 2011-10-27T18:31:47 python compilation isn't like other languages 2011-10-27T18:31:59 the ones that is like don't do it 2011-10-27T18:32:07 where are the rules for how this game is played? 2011-10-27T18:32:26 janzert: i mean other interpreted languages 2011-10-27T18:32:27 all python compilation is, is basically pre-parsing the code 2011-10-27T18:32:30 yeah 2011-10-27T18:32:43 i guess it'll only make like .1s diff at startup 2011-10-27T18:32:48 janzert: what about food spawn rate, I'd like to increase the start and increase the lower limit 2011-10-27T18:33:16 do the ants attack other ants or..? 2011-10-27T18:33:26 increasing start doesn't matter either way to me 2011-10-27T18:33:29 hi! i have a question: i'm trying to build a bot with the python starter package... and as the game swallows all prints i just write debug messages in a file. so if i write a test debug message in do_turn() and run my program it is just written to the file ONCE, while the game has like 117 round or so... anyone has a clue why? 2011-10-27T18:33:32 ybit: in addition to the website, the best reference I can give you for how to play is pick up an AI textbook 2011-10-27T18:33:43 okay, I'll make the change and update the server message 2011-10-27T18:33:51 not so sure about lower limit though 2011-10-27T18:33:57 what server message? 2011-10-27T18:34:07 teryret: i'm looking at the tutorial, but it just mentions avoiding collisions, gathering food, not blocking hills, exploring the map, and attacking enemy hills 2011-10-27T18:34:08 server_message.html 2011-10-27T18:34:19 Guest: the -e and -E flags to the game will make it start logging things you print 2011-10-27T18:34:21 why? 2011-10-27T18:34:33 if you attack an enemy hill, does that destroy the entire colony automatically or can they then regenerate another hill somewhere 2011-10-27T18:34:33 uh, because? 2011-10-27T18:34:35 it's not something the bot is told at all 2011-10-27T18:34:36 ybit: don't look at the tutorial, follow it 2011-10-27T18:34:48 ture 2011-10-27T18:34:50 true 2011-10-27T18:34:51 just do a forum post, I wouldn't make it a front page thing 2011-10-27T18:34:54 ok 2011-10-27T18:35:12 suggestion: put the rules of how the game is played somewhere so i can read them 2011-10-27T18:35:23 ybit: they're on the website 2011-10-27T18:35:28 link? 2011-10-27T18:35:40 http://aichallenge.org/problem_description.php 2011-10-27T18:35:41 how did you get to this irc room except from the site 2011-10-27T18:35:47 http://aichallenge.org/specification.php 2011-10-27T18:35:49 link is the back button in the corner 2011-10-27T18:35:55 i just joined #aichallenge 2011-10-27T18:36:06 janzert: sorry, I meant increase food visible, the initial foods in view of the first ants 2011-10-27T18:36:07 i'm on freenode always 2011-10-27T18:36:07 umm shouldn't the output from the program be like "x coordX coordy" ? 2011-10-27T18:36:10 antimatroid, mleise: you're now ops 2011-10-27T18:36:10 did you look at the topic or no? 2011-10-27T18:36:32 thanks Zannick 2011-10-27T18:36:37 ybit: see game specifications 2011-10-27T18:36:42 McLeopold: yeah, that's what I thought you meant by start 2011-10-27T18:36:47 i did not, my bad 2011-10-27T18:36:49 amstan: did you ever make me an admin? 2011-10-27T18:37:02 i'm basically just reading the stream and 2011-10-27T18:37:05 after ready 2011-10-27T18:37:11 Zannick: on the forums? 2011-10-27T18:37:14 oh 2011-10-27T18:37:17 it outputs things like a 28 2011-10-27T18:37:21 Zannick: i don't think so 2011-10-27T18:37:26 Zannick: you can compete just fine 2011-10-27T18:37:27 i thought you meant on the website 2011-10-27T18:37:27 no y 2011-10-27T18:37:31 Zannick: oh.. 2011-10-27T18:37:34 right 2011-10-27T18:37:34 i'm not going to compete :P 2011-10-27T18:37:35 that stuff 2011-10-27T18:37:41 ok, what's your userid? 2011-10-27T18:37:47 Zannick. 2011-10-27T18:38:21 Zannick: ok, you should have a disable thing on each profile 2011-10-27T18:38:25 neat 2011-10-27T18:38:49 janzert, McLeopold: i think we should display the errors from profiles to admins 2011-10-27T18:38:57 the compile errors i mean 2011-10-27T18:39:50 oh, that would be brilliant 2011-10-27T18:40:56 we have that already: ($user_id == current_user_id() or logged_in_as_admin())) { 2011-10-27T18:40:58 hmm 2011-10-27T18:40:58 *** paulwal has quit IRC (Ping timeout: 265 seconds) 2011-10-27T18:41:47 darn, all top bots have only one success submission 2011-10-27T18:42:00 yeah, ok, i see them 2011-10-27T18:42:05 hmm I can't get it to work... this is the line i have in my do_turn() function: 2011-10-27T18:42:10 sys.stderr.write("##################TEST#################") 2011-10-27T18:42:14 and i call the thing with 2011-10-27T18:42:16 *** Bystroushaak has joined #aichallenge 2011-10-27T18:42:27 play_one_game.cmd -e 2011-10-27T18:42:43 try adding -E as well 2011-10-27T18:42:46 yep i see them 2011-10-27T18:42:59 the errors should show up in game_logs 2011-10-27T18:43:07 is there a diffrence from -e and -E ? 2011-10-27T18:43:10 yeah 2011-10-27T18:43:17 but damned if I know what it is 2011-10-27T18:43:22 TheDigitalNinja: did you try --help? 2011-10-27T18:43:26 no 2011-10-27T18:43:36 :P 2011-10-27T18:43:39 I just read the code 2011-10-27T18:43:47 that works too 2011-10-27T18:43:52 I was hunting for something else 2011-10-27T18:43:58 just noticed the two flags in passing 2011-10-27T18:44:27 *** tylerthemiler has joined #aichallenge 2011-10-27T18:44:47 the files in game_logs are created.. but they are empty 2011-10-27T18:45:11 tried print "##################TEST#################" too 2011-10-27T18:46:33 did you try with -E 2011-10-27T18:46:36 yes 2011-10-27T18:46:37 amstan: we do 2011-10-27T18:46:40 -e, -E and both 2011-10-27T18:46:46 janzert: yes, i noticed 2011-10-27T18:46:52 :) 2011-10-27T18:46:57 *** tylerthemiler has quit IRC (Client Quit) 2011-10-27T18:47:19 is your bot crashing or timing out guest? 2011-10-27T18:47:43 *** grom358 has quit IRC (Quit: Leaving) 2011-10-27T18:47:50 no 2011-10-27T18:47:58 to test it i just used the starter package again 2011-10-27T18:48:10 i just added that one line to the do_turn() function 2011-10-27T18:48:16 oh, then run it with --verbose and they should print to the terminal 2011-10-27T18:48:26 *** grom358 has joined #aichallenge 2011-10-27T18:48:35 morning 2011-10-27T18:50:20 *** replore_ has quit IRC (Remote host closed the connection) 2011-10-27T18:51:22 nope, nothing gets printed 2011-10-27T18:51:27 not with print, not with stderr 2011-10-27T18:51:50 planet wars had 7982 total users, we have 5365 users 2011-10-27T18:52:46 even though i can't print... i was able to write to a text file before... and i saw that the debug message in do_turn() was only written ONCE 2011-10-27T18:52:47 guest: http://pastebin.com/3aG30175 is the command I use 2011-10-27T18:52:51 shouldn 2011-10-27T18:53:00 it shows things I print both on the console and in the log file 2011-10-27T18:53:10 I don't know what combination of options is responsible 2011-10-27T18:53:12 't it be written as often as how many turns there are? 2011-10-27T18:53:22 or is do_turns() only called once??? Oô 2011-10-27T18:53:26 occam's razor says it's only being run once 2011-10-27T18:53:36 or you aren't flushing your output 2011-10-27T18:53:42 *** lavalamp has quit IRC (Quit: Page closed) 2011-10-27T18:53:51 thanx @teryret 2011-10-27T18:53:55 i'll try it out 2011-10-27T18:54:01 -e logs to stnd error and -E logs to standard out. -o will also output standard out 2011-10-27T18:54:23 oh, also 2011-10-27T18:54:35 if you're using the print keyword, your bot is outputting to stdout 2011-10-27T18:54:40 which is where your commands should be going 2011-10-27T18:55:03 so they'll be interpreted as invalid commands and your bot will be killed. 2011-10-27T18:55:23 it doesn't kill bots on my local box 2011-10-27T18:55:35 or something 2011-10-27T18:55:47 it logs them as input errors 2011-10-27T18:55:55 and ignores them 2011-10-27T18:56:08 okay, i must be thinking of planet wars 2011-10-27T18:56:40 yeah! 4 servers now 2011-10-27T18:56:40 or perhaps some other type of thing, I know there are strings you can output that will get your bot killed right off 2011-10-27T18:56:42 @rankings 2011-10-27T18:56:47 amstan: Top 10 players: sir_macelon(81.8), xathis(81.6), GarySWest(81.5), GreenTea(80.5), paratrechina(80.5), MomoBot(80.1), ThisIsNotABug(78.9), Ender(78.1), olex.s(78.1), FlagCapper(77.9) 2011-10-27T19:00:06 it's really weird... if i say my bot in do_turn() to walk to a random location on the map. if i let the program run it ALWAYS picks the same "random" location 2011-10-27T19:00:51 what do you seed your random number generator with? 2011-10-27T19:01:16 i use the python standard 2011-10-27T19:01:23 random.randint for row and column 2011-10-27T19:01:29 if you're using the script I pasted it seeds random to a constant 2011-10-27T19:01:47 (or alternatively if youre using the one from the tutorial, which mine is heavily based on) 2011-10-27T19:01:48 Guest03238: starter packages are supposed to use the seed given by the engine 2011-10-27T19:02:47 yes, and the engine gets it from a terminal argument 2011-10-27T19:03:03 i see 2011-10-27T19:03:13 but if i put nothing it should use the current time, or not? 2011-10-27T19:03:22 one way to find out 2011-10-27T19:04:10 Guest03238: just remember that having such control to the seed is a useful tool, you can test your bot in exactly the same situation when you have a bug 2011-10-27T19:04:25 *** xgunner has joined #aichallenge 2011-10-27T19:04:58 ah indeed the seed is given in the play_one_game.cmd 2011-10-27T19:04:59 *** Redgis has quit IRC (Ping timeout: 244 seconds) 2011-10-27T19:05:08 so that explains why it is not random 2011-10-27T19:05:24 *** Cybsy has quit IRC (Ping timeout: 260 seconds) 2011-10-27T19:05:35 never got my account registration email... 2011-10-27T19:05:48 can anyone explain to me how my bot is timing out given that the profiler indicates that the entire run of the game took 1.357 seconds? 2011-10-27T19:06:05 *** eduardo_jrh has joined #aichallenge 2011-10-27T19:06:24 hi, i need help please 2011-10-27T19:06:40 I'll summon the paramedics to your IP 2011-10-27T19:06:51 thanks a lot 2011-10-27T19:07:06 firemen will be ok too 2011-10-27T19:07:06 *** sjoerd_visscher has joined #aichallenge 2011-10-27T19:07:13 ooh, you're one of those types 2011-10-27T19:07:16 no problemo 2011-10-27T19:07:51 i'm having trouble with the play_one_game.sh file on mac 2011-10-27T19:08:09 error msg? 2011-10-27T19:08:26 i'm getting "You must specify a map file" even though there's one specified in the .sh file 2011-10-27T19:08:38 can't login, but can't reregister as username is taken as well as email 2011-10-27T19:09:04 *** Eruonen has quit IRC () 2011-10-27T19:09:23 i'm using the play_one_game.sh right from the tools folder, o changes made 2011-10-27T19:09:33 no changes* 2011-10-27T19:09:37 *** xar0l_ has joined #aichallenge 2011-10-27T19:09:43 eduardo_jrh: write a script that uses playgame.py ;-) 2011-10-27T19:10:02 play_one_game.sh is just an example 2011-10-27T19:10:24 but why is it not working? 2011-10-27T19:10:36 demons 2011-10-27T19:10:53 Guess I'll ask Siri... 2011-10-27T19:10:58 that flet me try with playgame.py 2011-10-27T19:11:14 eduardo_jrh: because the parameters are already set 2011-10-27T19:11:21 or wait 2011-10-27T19:11:36 hmm so it can't find the maps? 2011-10-27T19:11:49 (hmm, I hadn't this problem on mine) 2011-10-27T19:12:41 wait please 2011-10-27T19:12:41 you did not modify the script or added parameters while using it right? 2011-10-27T19:13:25 i didn't modify anything 2011-10-27T19:13:32 the example already uses playgame.py 2011-10-27T19:13:43 *** ademar has quit IRC (Remote host closed the connection) 2011-10-27T19:14:02 yeah 2011-10-27T19:14:10 ./playgame.py --player_seed 42 --end_wait=0.25 --verbose --log_dir game_logs --turns 1000 --map_file ./maps/maze/maze_04p_01.map "$@" "python sample_bots/python/HunterBot.py" "python sample_bots/python/LeftyBot.py" "python sample_bots/python/HunterBot.py" "python sample_bots/python/GreedyBot.py" 2011-10-27T19:14:18 most are heaving problems by using it with their own flags 2011-10-27T19:14:34 teryret what is the command you use to print an error message ? 2011-10-27T19:14:34 could you check the existence of the directory? 2011-10-27T19:14:44 since there's already a map specified and the file exists i don't get what may be the problem 2011-10-27T19:14:45 http://pastebin.com/3aG30175 2011-10-27T19:15:05 no i mean in the source code 2011-10-27T19:15:07 but that one won't solve his problem 2011-10-27T19:15:10 just print "..." 2011-10-27T19:15:16 ? 2011-10-27T19:15:17 try a full path instead of a relative (by doing "pwd" if you are in the map directory) 2011-10-27T19:15:27 yeah 2011-10-27T19:15:36 just print "blah" 2011-10-27T19:15:54 ok, i'm on it 2011-10-27T19:16:00 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-10-27T19:16:04 doesn't work for me. can't see why... Oô 2011-10-27T19:16:13 *** Areks|2 has quit IRC (Ping timeout: 240 seconds) 2011-10-27T19:16:41 Guest03238: are you using the stderr stream? 2011-10-27T19:17:01 no i just used print as teryret does 2011-10-27T19:17:14 this is the command i use 2011-10-27T19:17:14 python "%~dp0playgame.py" --engine_seed 42 --player_seed 42 --end_wait=0.25 --verbose --log_dir game_logs -e -E --turns 500 --map_file "%~dp0maps\maze\maze_04p_01.map" %* "python ""%~dp0sample_bots\python\GreedyBot.py""" "python ""%~dp0sample_bots\python\LeftyBot.py""" "python ""%~dp0sample_bots\python\HunterBot.py""" "python ""%~dp0submission_test\MyBot.py""" 2011-10-27T19:17:19 and in the souce code i just have a 2011-10-27T19:17:21 that will only create error messages ;-) 2011-10-27T19:17:22 print "blah" 2011-10-27T19:17:40 aha? so waht do i have to do? 2011-10-27T19:17:41 how do i print blah? what do you mean? 2011-10-27T19:17:48 *** sjoerd_visscher has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 9.0a2/20111023042022]) 2011-10-27T19:17:56 eduardo_jrh: not your part 2011-10-27T19:18:07 ah^^ 2011-10-27T19:18:49 eduardo_jrh: did you get the command working yet? 2011-10-27T19:18:57 MacBook-de-Eduardo-Ramirez:~ usuarioapple$ /Users/usuarioapple/aichallenge/tools/play_one_game.sh /Library/Frameworks/Python.framework/Versions/2.7/Resources/Python.app/Contents/MacOS/Python: can't open file 'tools/playgame.py': [Errno 2] No such file or directory 2011-10-27T19:19:02 janzert: what do you think about lowering the cutoff percent and raising the cutoff turns? say 75% and 200? 2011-10-27T19:19:19 meh 2011-10-27T19:19:27 again a relative path 2011-10-27T19:19:41 *** goffrie__ is now known as goffrie 2011-10-27T19:19:53 *** Guest03238 has quit IRC (Quit: Page closed) 2011-10-27T19:19:55 eduardo_jrh: are you using lion? 2011-10-27T19:20:48 Usage: playgame.py [options] map bot1 bot2 You must specify a map file. Options: -h, --help show this help message and exit -m MAP, --map_file=MAP Name of the map file -t TURNS, --turns=TURNS Number of turns in the game --serial Run bots in serial, instead of parallel. --turntime=TURNTIME Amount of time to give each bot, in milliseconds 2011-10-27T19:20:49 hmm, I'm not opposed in principle. I wonder if that percentage might be to low though? I guess we could try it and see. 2011-10-27T19:20:54 yeah, i'm on lion 2011-10-27T19:21:04 ah, that explains a lot 2011-10-27T19:21:08 no no 2011-10-27T19:21:11 I'm still on snow leopard 2011-10-27T19:21:11 i'm sorry 2011-10-27T19:21:20 i downgraded it 2011-10-27T19:21:24 it's snow leopard 2011-10-27T19:21:24 basically currently you have to have a 10 to 1 ratio currently that would only require outnumbering by 4 to 3 2011-10-27T19:21:45 meh, everything worked out of the box on mine 2011-10-27T19:21:45 *** sdf has quit IRC (Quit: Leaving.) 2011-10-27T19:22:09 that's what i foun weird 2011-10-27T19:22:13 found 2011-10-27T19:22:27 so what conclusion did we come to on stderr? 2011-10-27T19:22:35 4 to 1? 2011-10-27T19:22:46 I'm trying it a bit and anything I send there seems to disappear 2011-10-27T19:22:55 oops, yes 2011-10-27T19:22:57 (meh, I'm becoming confused) 2011-10-27T19:22:58 although stdout works 2011-10-27T19:23:05 The cutoff would start a lot sooner, but we give them more time to clean up 2011-10-27T19:23:19 i must open terminal, the write the address of play_one_game.sh right? 2011-10-27T19:23:31 double the clean up time but less than half the advantage 2011-10-27T19:23:34 eduardo_jrh: yeah 2011-10-27T19:23:44 how about 80% so it's double and half 2011-10-27T19:23:54 just copy that file, modify it to your own purposes 2011-10-27T19:24:10 that's what i've been doing 2011-10-27T19:24:24 ok 2011-10-27T19:24:50 unless you think not enough games are being cutoff 2011-10-27T19:24:55 but I think it's about right now 2011-10-27T19:25:15 *** thestinger has quit IRC (Quit: WeeChat 0.3.6) 2011-10-27T19:25:16 it is, but the ones that are should be cut off slightly sooner 2011-10-27T19:25:41 ? 2011-10-27T19:25:50 --map_file tools/maps/maze/maze_02p_01.map what may be the problem here? 2011-10-27T19:26:01 *** delt0r__ has quit IRC (Ping timeout: 258 seconds) 2011-10-27T19:26:03 the main effect is that food cutoff is going to take much longer I think 2011-10-27T19:26:26 eduardo_jrh: if its already pointing to the right map, try a full path instead 2011-10-27T19:26:29 maybe we shouldn't go for double right away but an extra third 2011-10-27T19:26:30 i'm gonna try it on win 2011-10-27T19:26:46 windows, doesn't seem to work 2011-10-27T19:27:12 87.5% and 150 turns 2011-10-27T19:27:30 :q 2011-10-27T19:27:36 oops, wrong window 2011-10-27T19:28:01 it worked!! full path on the map 2011-10-27T19:28:57 *** Zannick sets mode: -b *!*@dsl-173-239-80-026.cascadeaccess.com 2011-10-27T19:29:12 or rather more accurately 86.6% and 150 turns 2011-10-27T19:29:48 thanks a lot! 2011-10-27T19:29:58 *** eduardo_jrh has quit IRC (Quit: Page closed) 2011-10-27T19:32:11 *** zefyear has quit IRC (Quit: Page closed) 2011-10-27T19:33:12 @later tell McLeopold the workers do not have a low memory limit anymore, 1.5GB should be enough for anyone :P 2011-10-27T19:33:12 janzert: As you wish. 2011-10-27T19:33:28 even java 2011-10-27T19:33:47 never got my account registration email...anyone know how to reset? 2011-10-27T19:35:06 *** overburn has quit IRC (Ping timeout: 265 seconds) 2011-10-27T19:35:14 *** RobotCaleb has quit IRC (Ping timeout: 260 seconds) 2011-10-27T19:35:26 xgunner: the message hasn't bounced (at least not yet), have you checked your spam folder? 2011-10-27T19:35:38 why yes i have 2011-10-27T19:35:45 the subject will be "AI Challenge!" from staff@aichallenge.org 2011-10-27T19:36:48 *** amstan has quit IRC (Remote host closed the connection) 2011-10-27T19:37:06 once you submit, it will continually pay until you update? 2011-10-27T19:37:12 *play 2011-10-27T19:37:18 *** pgoldbr has joined #aichallenge 2011-10-27T19:38:08 xgunner: sent another message, this one will be from janzert@aichallenge.org 2011-10-27T19:39:10 *** delt0r__ has joined #aichallenge 2011-10-27T19:40:06 Mooloo: yes 2011-10-27T19:40:41 *** amstan has joined #aichallenge 2011-10-27T19:40:41 *** ChanServ sets mode: +o amstan 2011-10-27T19:41:08 k, found original network solutions decided its spam at their lovely server... thanks 2011-10-27T19:44:18 *** craklyn has joined #aichallenge 2011-10-27T19:44:50 *** Bystroushaak has quit IRC (Quit: Page closed) 2011-10-27T19:46:19 *** chrisf has quit IRC (Quit: Ex-Chat) 2011-10-27T19:46:19 *** Zannick sets mode: -o Zannick 2011-10-27T19:46:57 *** ChanServ sets mode: +o Zannick 2011-10-27T19:47:13 *** Zannick sets mode: +b *!*@S01060019db8e14ab.vc.shawcable.net 2011-10-27T19:47:34 *** Zannick sets mode: -b *!*nbusern@*.vc.shawcable.net 2011-10-27T19:47:55 *** Zannick sets mode: -o Zannick 2011-10-27T19:48:02 should have cleaned that up earlier 2011-10-27T19:48:41 this is crazy, my A* either takes <.003s or >30 seconds, but never in the middle 2011-10-27T19:49:20 time for caching I guess 2011-10-27T19:49:29 meh, strange things are happening, github encouraging notepad? 2011-10-27T19:49:44 teryret: or remodelling 2011-10-27T19:49:55 >30 seconds sounds like a bug 2011-10-27T19:50:08 too 2011-10-27T19:50:18 what is remodelling? 2011-10-27T19:50:33 *** amstan_ has joined #aichallenge 2011-10-27T19:50:33 *** ChanServ sets mode: +o amstan_ 2011-10-27T19:50:48 (lets just say remapping the map in an other representation) 2011-10-27T19:50:57 ah, gotcha 2011-10-27T19:51:07 notepad? eewwwwww 2011-10-27T19:51:16 *** amstan_ has quit IRC (Remote host closed the connection) 2011-10-27T19:53:16 aichallenge: McLeopold epsilon * r924a695 / (4 files in 2 dirs): update default game settings - http://git.io/12MAYw 2011-10-27T19:53:45 lol, the error log for my game is 15mb 2011-10-27T19:54:10 hooray for "when in doubt, print it out" 2011-10-27T19:55:26 ok, i just wrote the hardest 50 lines in my bot, i wonder how they run now 2011-10-27T19:55:40 avdg: where does github encourage notepad? 2011-10-27T19:55:49 check the blog 2011-10-27T19:56:13 McLeopold: you know antimatroid1 already wanted a lower food rate? 2011-10-27T19:56:16 (but yeah, I know that github just tries to help the noobs) 2011-10-27T19:56:28 did it go up? 2011-10-27T19:56:36 janzert: no I didn't 2011-10-27T19:56:41 https://github.com/blog/975-gitpad-write-your-commits-in-notepad 2011-10-27T19:57:05 i don't really care that much, i just think the games would be cooler if there were less ant and more strategy 2011-10-27T19:57:16 I use notepad2 for my git commits, when I'm not using a -m :) 2011-10-27T19:58:05 I'm already stuck with vim, but I want to learn emacs as well as the opportunity arrives 2011-10-27T19:58:17 or ed 2011-10-27T19:58:50 does anyone know why {time.sleep} is showing up as the big offender in my profile run? 2011-10-27T19:59:09 going to watch baseball be back later :) 2011-10-27T19:59:14 (and to preempt the first question, no, I'm not telling my ants to sleep at any point) 2011-10-27T19:59:14 cya :-) 2011-10-27T19:59:27 (and don't crash the server :-)) 2011-10-27T19:59:53 teryret: you're profiling the engine not your bot? 2011-10-27T20:00:11 oh, is that what the --profile flag does? 2011-10-27T20:00:13 sigh 2011-10-27T20:00:14 ok 2011-10-27T20:00:24 do you happen to know a good way to profile the bot? 2011-10-27T20:00:51 I tried -m cProfile in the run script 2011-10-27T20:00:56 but it got disabled by the engine 2011-10-27T20:01:27 mass suicide :) http://aichallenge.org/visualizer.php?game=21355 2011-10-27T20:03:11 teryret: run you game with -I, then pipe that back to your bot with cProfile 2011-10-27T20:03:46 *** Guest290825 has joined #aichallenge 2011-10-27T20:04:10 hi! i have a question. does the fog of war only count for food/enemies, or for the whole map? 2011-10-27T20:04:39 it counts for everything. AND water blocks are only reported the first time you see them 2011-10-27T20:04:39 "count" is a strange way of looking at things 2011-10-27T20:04:58 McLeopold: will do, thanks 2011-10-27T20:05:12 *** skyramp has quit IRC (Quit: Page closed) 2011-10-27T20:05:48 what we call fog of war is only a visualization of what the ants can currently see 2011-10-27T20:06:19 it's important to keep that separate in your mind from the notion of what the computer "knows" at any given time 2011-10-27T20:06:36 you can know a great deal more than you can see at any one moment 2011-10-27T20:07:04 *** eduardo_jrh has joined #aichallenge 2011-10-27T20:07:28 hi, need help with Makefile on Mac 2011-10-27T20:07:35 please 2011-10-27T20:07:49 * teryret hisses and runs off at the sound of the word Makefile 2011-10-27T20:08:42 i'm sorry 2011-10-27T20:08:54 but i don't know how to make it work 2011-10-27T20:08:56 *** Guest290825 has quit IRC (Ping timeout: 265 seconds) 2011-10-27T20:10:08 please 2011-10-27T20:10:32 ... you haven't asked a question yet (not that I know anything about makefiles) 2011-10-27T20:10:38 maybe it's a stupid question but i really don't know how to compile the jar file 2011-10-27T20:10:54 uh, don't you just run jar files? 2011-10-27T20:11:06 I tried running it on the Terminal 2011-10-27T20:11:15 how to compile the .java files on a .jar file? 2011-10-27T20:11:36 oh, how to make the jar file you mean 2011-10-27T20:11:52 sorry, that's what i meant 2011-10-27T20:11:57 *** Knekkebjoern has quit IRC (Quit: Leaving.) 2011-10-27T20:11:58 I don't know off hand, but I suspect the man page for either java or javac will be able to help 2011-10-27T20:12:10 jar is actually a utiity 2011-10-27T20:12:11 McLeopold: :-) 2011-10-27T20:12:14 utility 2011-10-27T20:12:29 ah, well that's good to know 2011-10-27T20:12:40 *** foRei has quit IRC (Quit: Bye) 2011-10-27T20:12:48 *** bhasker has quit IRC (Quit: bhasker) 2011-10-27T20:13:16 so how do i compile the bot into a jar file? 2011-10-27T20:13:54 man jar 2011-10-27T20:14:03 the tutorial says you should run makefile 2011-10-27T20:14:23 meh, js is an easy language to kill performance, but its still a relative fast language when these performance killers are solved 2011-10-27T20:14:26 figure out how to do it in the terminal, then use that as the command in your make file 2011-10-27T20:14:26 what does man jar mean? 2011-10-27T20:14:27 *** Guest29292929 has joined #aichallenge 2011-10-27T20:14:32 hi! i have a question. does the fog of war also hides the water, or only count for food/enemies? 2011-10-27T20:14:41 i'm sorry, feeling very retarded right now 2011-10-27T20:14:46 go to a terminal, and type "man jar" 2011-10-27T20:14:51 *** replore_ has joined #aichallenge 2011-10-27T20:15:09 if you mean running the makefile, you type "make" 2011-10-27T20:15:29 (the java starter pack comes with a makefile) 2011-10-27T20:15:35 i type that ont terminal... right? 2011-10-27T20:15:38 yes 2011-10-27T20:15:49 once you have the makefile written 2011-10-27T20:16:03 Guest29292929: you don't get a report of anything outside of the visibility radius of your ants 2011-10-27T20:16:25 but the first step to getting the make file written is figuring out the command that does the compiling 2011-10-27T20:16:39 i see. so i can't plan a path through unseen territorry , right? 2011-10-27T20:16:59 nothing wrong with planning it :) 2011-10-27T20:17:00 well, if you've seen it before you might be able to 2011-10-27T20:17:15 yes, true 2011-10-27T20:17:19 i wonder 2011-10-27T20:17:24 so i type make... make what? haha what's the target? 2011-10-27T20:17:26 but there may be new enemy ants there you don't know about 2011-10-27T20:17:32 just make 2011-10-27T20:17:33 if i call passable() on a cell i haven't seen before 2011-10-27T20:17:37 what is the return value? 2011-10-27T20:17:39 it has a default target 2011-10-27T20:17:51 it asks me what's the target 2011-10-27T20:17:53 the target is defined in the file called Makefile 2011-10-27T20:18:02 the default target is all 2011-10-27T20:18:04 MacBook-de-Eduardo-Ramirez:~ usuarioapple$ make make: *** No targets specified and no makefile found. Stop. 2011-10-27T20:18:30 you are in the wrong director 2011-10-27T20:18:55 oh, is the Makefile already written in the starter kit? 2011-10-27T20:18:59 mmm... so how do i go to the right director? sorry 2011-10-27T20:18:59 yes 2011-10-27T20:19:03 pwd 2011-10-27T20:19:09 to tell you where you are 2011-10-27T20:19:13 then cd to move around 2011-10-27T20:19:36 awesome 2011-10-27T20:19:39 that's it 2011-10-27T20:19:42 thanks a lot 2011-10-27T20:19:45 np 2011-10-27T20:19:57 *** eduardo_jrh has quit IRC (Quit: Page closed) 2011-10-27T20:20:15 can i access somehow the tiles which are visible? 2011-10-27T20:20:23 implicitly 2011-10-27T20:20:31 you know your ants and the view radius 2011-10-27T20:20:49 Guest: what language are you in 2011-10-27T20:20:53 python 2011-10-27T20:21:02 check out ants.py 2011-10-27T20:21:17 i'm looking at that right now 2011-10-27T20:21:38 and i tried to make a graph of tiles which are not water 2011-10-27T20:21:45 i tried to do that in update() 2011-10-27T20:21:46 the function is called visible 2011-10-27T20:22:00 you pass it a loc and it will tell you what's there, if it can see it 2011-10-27T20:22:18 with some refactoring you can pull a list of all visible spots 2011-10-27T20:25:06 *** McLeopold has left #aichallenge 2011-10-27T20:26:20 *** Guest29292929 has quit IRC (Ping timeout: 265 seconds) 2011-10-27T20:27:32 *** xgunner has quit IRC (Quit: Page closed) 2011-10-27T20:29:47 *** Guest29292929__ has joined #aichallenge 2011-10-27T20:29:59 sorry my internet keeps disconnecting ... 2011-10-27T20:30:19 but how do i use it? 2011-10-27T20:30:20 so i'm in python and i have the visible function 2011-10-27T20:31:19 what is the attackradius2 set to at the moment? 2011-10-27T20:32:48 what do i have to pass the visible function to use it? 2011-10-27T20:33:40 have you checked git.. they keep updating the starter kits. It might already have a visible function 2011-10-27T20:36:39 *** dwins has quit IRC (Ping timeout: 252 seconds) 2011-10-27T20:42:07 if ant (54, 60) goes to direction 3 it may kill something, CHAAARGE! 2011-10-27T20:42:16 an then it gets killed by six enemy ants 2011-10-27T20:42:27 i think i screwed up something 2011-10-27T20:42:53 *** Kingpin13 has quit IRC (Quit: quit) 2011-10-27T20:43:27 grom358: radius in which ants attack/kill each other 2011-10-27T20:43:55 grom358: http://aichallenge.org/specification_battle.php 2011-10-27T20:44:17 ah sorry 2011-10-27T20:44:29 yeah nvm.. I found out it is 5 2011-10-27T20:44:29 didn't see "set to" 2011-10-27T20:44:33 yeah, five 2011-10-27T20:44:46 isn't that a parameter somewhere? 2011-10-27T20:44:58 it comes at the beginning 2011-10-27T20:45:01 i was just buffering to avoid being near enemy ants when can't win a fight 2011-10-27T20:45:32 so I consider the attackradius2 to be 10 when avoiding fights 2011-10-27T20:45:40 *** nack has joined #aichallenge 2011-10-27T20:46:06 What is the last day for submission and when do the finals start? 2011-10-27T20:46:11 *** anty_m has joined #aichallenge 2011-10-27T20:48:01 nack: around 18dec 2011-10-27T20:48:19 nack: it is in topic 2011-10-27T20:48:35 *** xar0l_ has quit IRC (Ping timeout: 265 seconds) 2011-10-27T20:48:38 hey.. that first diagram on http://aichallenge.org/specification_battle.php is wrong 2011-10-27T20:48:41 Oh okay, thanks. I didn't see it. 2011-10-27T20:48:44 Bob can't reach Roy 2011-10-27T20:48:51 with attackradius2 at 5 2011-10-27T20:49:28 *** nack has quit IRC (Client Quit) 2011-10-27T20:50:15 *** savaron has quit IRC (Quit: Page closed) 2011-10-27T20:50:41 meh, I hope my new entry kills all timeouts and errors 2011-10-27T20:50:43 grom358: yeah, you're right 2011-10-27T20:50:54 B and R in the middle get killed 2011-10-27T20:51:07 avdg: me too... 2011-10-27T20:51:27 local times are 1ms each round 2011-10-27T20:51:31 Fluxid: I mean.. it says Bill and Roy attention is split 2011-10-27T20:51:38 so I was about to complain about how long my code is taking per turn, and then remembered that I'm using Python 2011-10-27T20:51:41 but its still a stupid bot 2011-10-27T20:52:19 but that isn't true if I'm correct.. Roy is not in Bob's attack radius 2011-10-27T20:52:26 grom358: it isn't split, but outcome is the same 2011-10-27T20:52:33 *** rj_ has joined #aichallenge 2011-10-27T20:52:36 no, wait 2011-10-27T20:53:10 yeah, it is not split, and only ants in the middle are killed, B at the top, and R at the bottom are nnot participating 2011-10-27T20:53:12 I'm pretty sure that the way to do this game right is to run inter-turn calculations 2011-10-27T20:53:55 oh wait.. I think I got this wrong then.. its when the attack radius overlap.. correct? 2011-10-27T20:54:11 but then, what do I know, I'm still struggling to figure out why A* is so slow 2011-10-27T20:54:20 no.. that's not right.. gah 2011-10-27T20:54:29 not sure if I get this 2011-10-27T20:54:43 when an enemy ant is within your attack radius you fire 2011-10-27T20:55:09 and since they have the same radius, when both ants are within that radius of each other they both fire 2011-10-27T20:55:25 teryret: yeah that is what I thought.. but why does on the page http://aichallenge.org/specification_battle.php is says Bob attacks Roy 2011-10-27T20:56:16 The pictures on the specification page all need neighborhood circles drawn on them. Several are unclear about who is within range of whom. 2011-10-27T20:56:53 hey, similar question but real-life: http://fluxid.pl/misc/2011-10-28-025217_428x346_scrot.png 2011-10-27T20:57:18 whis is the purple ant killed, and all three green ones survive? 2011-10-27T20:57:30 none of the three ants are in radius of 5... 2011-10-27T20:57:40 sqrt(5) 2011-10-27T20:57:49 The three ants are in radius of each other. Not the enemy. 2011-10-27T20:57:52 meh, I still have to wait for a long time to test mine on the site 2011-10-27T20:58:18 Because the three green ones each have 3 neighbors (they are within 3 circles). 2011-10-27T20:58:21 callahan: same, purple isn't in radius of those three 2011-10-27T20:59:40 grom358: because bob and ralph are too far apart 2011-10-27T20:59:44 What is the attack radius? 2011-10-27T20:59:53 sqrt(5) 2011-10-27T20:59:54 .###. 2011-10-27T20:59:54 ##### 2011-10-27T20:59:54 ##A## 2011-10-27T20:59:54 ##### 2011-10-27T20:59:54 .###. 2011-10-27T21:00:11 Fluxid: yes exactly.. that what I get 2011-10-27T21:00:23 that is why http://aichallenge.org/specification_battle.php seems wrong to me 2011-10-27T21:00:28 then why purple is killed >:( 2011-10-27T21:00:40 Fluxid: this is what I'm trying to figure out too 2011-10-27T21:00:53 Does he actually die on the subsequent turn? 2011-10-27T21:01:09 I think I heard someone said the attack drawing happens a turn early ? I don't know, I haven't looked at combat yet 2011-10-27T21:02:13 Step animation and see if they have made their move yet for the turn? 2011-10-27T21:02:14 oh wait 2011-10-27T21:02:17 waaaaait 2011-10-27T21:02:29 * avdg waits 2011-10-27T21:02:32 sorry 2011-10-27T21:02:40 this is attack of next turn 2011-10-27T21:02:45 yerp 2011-10-27T21:02:49 http://pastebin.com/pcqrWRnT 2011-10-27T21:03:04 what does it mean for a starter package to not support hills? 2011-10-27T21:03:23 It means that you have to add the code which reads the game's input to your bot 2011-10-27T21:03:25 http://fluxid.pl/misc/2011-10-28-025910_335x270_scrot.png this is the actual attack 2011-10-27T21:03:29 it doesn't recognize hills 2011-10-27T21:03:31 with respect to hill locations 2011-10-27T21:03:35 ~ is vision.. + is vision and last safe square.. # is attack 2011-10-27T21:03:56 * nods knowingly at Fluxid 2011-10-27T21:03:58 craklyn: thanks 2011-10-27T21:04:01 doh. How do I emote :P 2011-10-27T21:04:06 Fluxid: yeah so.. see the diagram on http://aichallenge.org/specification_battle.php 2011-10-27T21:04:13 its wrong right? 2011-10-27T21:04:36 AFAICT Roy is not in attack radius of Bob 2011-10-27T21:04:42 Which diagram? 2011-10-27T21:04:56 the Bob, Bill, Roy, Ralph one 2011-10-27T21:05:22 Yes, Bob and Ralph are out of range of one another 2011-10-27T21:05:25 Looks correct to me 2011-10-27T21:05:26 *** Apophis has quit IRC (Read error: Connection reset by peer) 2011-10-27T21:05:40 grom358: there is no said radius 2011-10-27T21:05:40 no.. I am saying Bob and Roy are out of range of each other 2011-10-27T21:05:50 so? 2011-10-27T21:05:52 Depends on the range 2011-10-27T21:05:56 grom358: my guess is on the time of writing radius was different 2011-10-27T21:06:04 I don't know if they said what the range was in this example, do they? 2011-10-27T21:06:07 grom358: don't rely on constant value 5 2011-10-27T21:06:10 oh, wait a sec, if you just scroll down it gives you how it works in python 2011-10-27T21:06:16 *** RobotCaleb has joined #aichallenge 2011-10-27T21:06:20 You are counting it wrong. It's not how many ants surround your enemies that matter. it's how many ants surround your friends. 2011-10-27T21:06:28 " After moving, Bill has stepped close enough to fight with both red ants while Bob only moved close enough to fight with Roy." 2011-10-27T21:06:40 So yes, by hypothesis, they're in range 2011-10-27T21:06:46 yeah I'm not relying on 5.. I just wanted to make sure I understood the radius right.. that my diagram is correct for the correct values 2011-10-27T21:07:49 Roy encircles bill and ralph. ralph encircles roy. Both roy and ralph have a score of 2 (in two friendly circles). 2011-10-27T21:07:58 Same for bob and bill. 2011-10-27T21:08:11 Thus bill and roy are in range of each other with same score. They blow up. 2011-10-27T21:08:18 Friendly circle? 2011-10-27T21:08:27 yeah, it counts friends. 2011-10-27T21:08:39 what? so you friends don't need to be in range? 2011-10-27T21:08:52 Your friends need to be in range of YOU, not your enemy. 2011-10-27T21:10:00 Uh, I don't think my understanding of ant combat agrees with your description callahan 2011-10-27T21:10:05 where does it say this? 2011-10-27T21:10:31 they talk about how many enemy ants. not friends 2011-10-27T21:10:48 Each ant splits its focus on all enemy ants within its range 2011-10-27T21:10:57 (enemies in range of enemy) in the pseudocode 2011-10-27T21:11:01 If an enemy ant within its range has greater focus than your ant has, then your ant dies 2011-10-27T21:11:48 yeah.. I see there shape matches mine for attackradius2 = 5.. That is what I was trying to confirm 2011-10-27T21:12:41 for all ants: count how many of their enemys are in range.. the higher that number is, the easier they are to die, and harder for them to kill 2011-10-27T21:13:00 if any enemy ant in range, has a smaller number: then it dies 2011-10-27T21:13:45 okay! time for some tcp testing! 2011-10-27T21:14:00 Which bot is yours Fluxid? 2011-10-27T21:14:02 tcp? 2011-10-27T21:14:10 so.. http://pastebin.com/pcqrWRnT . If if I move to + square then the enemy ant can kill me on their turn. 2011-10-27T21:14:11 we're allowed to talk to servers from our ants? 2011-10-27T21:14:26 NO 2011-10-27T21:14:35 there is no 'turn'.. all actions are at same time 2011-10-27T21:14:41 whew 2011-10-27T21:15:15 oh... 2011-10-27T21:15:25 callahan: it will be fluxid_v5 in a second 2011-10-27T21:15:27 @tcp 2011-10-27T21:15:28 Fluxid: tcp could be http://ants.fluxid.pl/howto. 2011-10-27T21:15:38 so the engine talks to all bots for actions.. then applys them all at same time.. got you 2011-10-27T21:16:23 Fluxid: That's a good naming scheme so you can tell your bots apart somewhat. 2011-10-27T21:17:31 callahan: yes, i'm connecting to tcp few of them at once, so i can check in ranking how they do 2011-10-27T21:18:43 i hope v5 will work better than v4... at least in close combat because this it the only change 2011-10-27T21:19:08 v4 just avoids enemy from far, v5 tries to get closer and eventually kill 2011-10-27T21:19:28 what is this tcp server u speak of? 2011-10-27T21:20:29 http://ants.fluxid.pl/howto 2011-10-27T21:21:02 yeah looked at that.. but what does it do differently to running a game locally? 2011-10-27T21:21:05 with playgame.py 2011-10-27T21:21:20 gives you different opponents 2011-10-27T21:21:46 It plays in a tournament against other people's bots. 2011-10-27T21:21:51 And ranks them. 2011-10-27T21:21:51 grom358: it does run it locally 2011-10-27T21:22:02 but with bots of other people 2011-10-27T21:22:08 http://ants.fluxid.pl/replay.9705 bah, fail 2011-10-27T21:22:21 oh.. is this to speed up waiting on where u upload it? 2011-10-27T21:22:23 *** he_the_great has joined #aichallenge 2011-10-27T21:22:35 Sort of 2011-10-27T21:22:40 cause at the moment I been uploading to test against others bots 2011-10-27T21:22:45 It's just a separate tournament that isn't so slow. 2011-10-27T21:22:58 grom358: no, just for testing purposes... 2011-10-27T21:23:04 ah.. okay. My bot not really ready for that 2011-10-27T21:23:14 its too stupid atm 2011-10-27T21:23:33 grrr... pathfinding is expensive. 2011-10-27T21:23:34 need battle logic.. then it will be ready 2011-10-27T21:23:46 gotta get smarter about storing results between searches 2011-10-27T21:23:48 anty_m: do it on queue 2011-10-27T21:23:57 that is what I did 2011-10-27T21:24:43 "on queue" ...breadth first search using a queue, you mean? 2011-10-27T21:25:11 My bot only does well in random walk maps so far. But it's really fast. 2011-10-27T21:25:19 apply on ants.fluxid.pl 2011-10-27T21:25:33 like put that you want to path find onto a queue.. then while timeRemaining > timeToPathfind (pop item off queue; pathfind for item) 2011-10-27T21:25:47 *** Guest29292929__ has quit IRC (Ping timeout: 265 seconds) 2011-10-27T21:26:27 ok, now I understand 2011-10-27T21:26:49 ok, i see some nice battle going on with my bot, can't wait.. for my bot to fail 2011-10-27T21:27:30 how did i not know about Ctrl+R in a terminal? 2011-10-27T21:27:33 yeah.. so I set a target for my ant.. but i haven't pathfind for it yet. Then I have loop over the queue to setup the path. If the ant has no path yet.. I currently just random walk 2011-10-27T21:27:46 I still don't, what's Ctrl+R? 2011-10-27T21:27:48 while it waits to get its path 2011-10-27T21:28:06 (and also, what OS/terminal emulator?) 2011-10-27T21:28:39 bah. I miss college... so much more free time to spend on stuff like this! :-P 2011-10-27T21:29:08 yeah, but no experience from which you could learn enough to be good... 2011-10-27T21:29:23 anty_m: seems to work out well here. http://aichallenge.org/visualizer.php?game=20807&user=4130 2011-10-27T21:30:10 pretty dumb bot atm. It has no attack logic. It just pathfinds to targets 2011-10-27T21:30:21 using A* 2011-10-27T21:30:42 none of the fluxid.pl replay links work for me 2011-10-27T21:31:18 have you tested it on maze maps yet? 2011-10-27T21:31:36 yes it works on maze maps 2011-10-27T21:31:52 hmmm did you have to do anything fancy to get it to be fast enough? 2011-10-27T21:32:10 nope.. I am writing in Java though 2011-10-27T21:32:30 the profiler run of my A* implementation has been going for 24 CPU minutes 2011-10-27T21:32:39 (for 20 turns) 2011-10-27T21:32:59 ouch 2011-10-27T21:33:05 mine is also dumb, no attacking, no defending, no food-collecting, no path-finding really... but not quite a random walk. http://aichallenge.org/visualizer.php?game=20277&user=2664 2011-10-27T21:33:30 teryret, i found it by accident, but then a search turned up this: http://lifehacker.com/278888/ctrl%252Br-to-search-and-other-terminal-history-tricks 2011-10-27T21:34:10 anty_m: hey if you don't want todo anything fancy I have a line walk to food I had for my random walk bot 2011-10-27T21:34:11 trying to add pathfinding to steer the ant to food if it's in easy reach... but having trouble doing so efficiently with many ants 2011-10-27T21:34:34 my bot isn't working currently, it's in pieces while i assemble the pathfinding 2011-10-27T21:34:58 *** craklyn has quit IRC (Ping timeout: 265 seconds) 2011-10-27T21:35:11 anty_m: really? hang on I'll show u code snippet 2011-10-27T21:35:27 *** pgoldbr has quit IRC (Ping timeout: 265 seconds) 2011-10-27T21:36:12 * avdg is about to start building his pathfinder 2011-10-27T21:36:48 http://ants.fluxid.pl/replay.9718 what a fail! 2011-10-27T21:36:56 i try to get to food behind the hill! 2011-10-27T21:37:14 i suppose it's not really, i could just take out the in-dev code and it would run 2011-10-27T21:37:40 Fluxid: do you block movements to your own hill? 2011-10-27T21:37:47 but avdg yes 2011-10-27T21:37:52 it sucks 2011-10-27T21:37:54 without "but " 2011-10-27T21:37:55 Fluxid: hehehe, that's the map I'm using for testing 2011-10-27T21:37:59 it's a killer 2011-10-27T21:38:03 I've doine it with mine 2011-10-27T21:38:05 *done 2011-10-27T21:38:14 if you get the first one trapped, game over 2011-10-27T21:38:26 i don't want to block hill at all... but think i would need to pathinf over the hill, lol 2011-10-27T21:38:49 just put a high penalty on that tile 2011-10-27T21:38:56 grom358: thanks, but I've gotta go 2011-10-27T21:39:14 if I catch you in here again, I'll try to remember to ask 2011-10-27T21:39:18 though penalties aren't easy to implement in basic bots or bots without penalty support 2011-10-27T21:39:35 *** anty_m has quit IRC (Quit: Page closed) 2011-10-27T21:39:43 avdg: i can implement penalty easily 2011-10-27T21:39:54 Fluxid, why does it keep moving so much 2011-10-27T21:41:33 why can't i look at those replays? all other pages there work 2011-10-27T21:41:35 cyphase: i made it so it doesn't move only if really needs so. 2011-10-27T21:42:11 anty_m: http://pastebin.com/qtk9iX5z 2011-10-27T21:42:21 ah crap.. I took too long 2011-10-27T21:43:02 raytrace :D 2011-10-27T21:43:09 do you support refractions? 2011-10-27T21:43:12 meh… "1611 players are ahead." I don't think I will get mine running before I need to sleep 2011-10-27T21:43:40 ah.. no. Its just a simple direct line 2011-10-27T21:44:29 ok, guesses as to why my bot is making a quarter billion calls to str.lower() ? 2011-10-27T21:44:46 teryret, stuck in parsing? 2011-10-27T21:44:51 my current version is using A* to food.. but I could replace it with that snippet again. It worked on my old version of the bot 2011-10-27T21:45:04 my only string operations are in a file with 321 loc 2011-10-27T21:45:10 nope, somehow str.lower() is being called from my A* code 2011-10-27T21:45:20 except debug messages 2011-10-27T21:46:02 ... as is file.readline?!? 2011-10-27T21:46:06 wtf? 2011-10-27T21:46:10 lol 2011-10-27T21:46:16 I didn't write that 2011-10-27T21:46:24 I know I didn't 2011-10-27T21:46:38 or. like do a hybrid. if can walk directly to it. set that as path.. if none of the food is in direct path.. then choose the closest one by distance and queue up A* for it 2011-10-27T21:47:18 if you do A* right it acts like DFS for direct paths 2011-10-27T21:48:09 so if it's direct it'll be just as fast as any other algorithm 2011-10-27T21:48:18 yeah.. I really should time it for food ones.. the 60ms I spoke about before was to travel across the entire map to enemy hill 2011-10-27T21:48:26 (or, more accurately, within a constant factor of any other algorithm) 2011-10-27T21:49:44 well.. my current bot.. does BFS limited to the ant's vision. Then picks the first food given by the BFS. Then queues an A* path for it 2011-10-27T21:50:14 yeah.. could do that better.. get the BFS to return the path 2011-10-27T21:51:09 or just not do the BFS at all 2011-10-27T21:51:21 you already have a collection of currently visible foods 2011-10-27T21:51:31 yeah but without the BFS how do I figure out the food that is closest? 2011-10-27T21:51:34 you can then find the nearest one in linear time 2011-10-27T21:51:49 loop over foods calculating distance for each 2011-10-27T21:51:57 teryret: how do u know distance? 2011-10-27T21:51:59 keeping track of the nearest as you go 2011-10-27T21:52:07 there's a function in ants.py that does it for you 2011-10-27T21:52:21 closest by vision is not closest by travelling 2011-10-27T21:52:24 it's just manhattan distance 2011-10-27T21:52:38 yeah.. thats no good. That is why I did BFS 2011-10-27T21:52:44 could be water in way 2011-10-27T21:53:12 that's very true, but it turns out to not matter 2011-10-27T21:53:42 you see the nearest food and start A*ing to it 2011-10-27T21:53:57 as you go you're continually checking what the nearest food is 2011-10-27T21:54:00 yeah.. but you might head to one that will take longer to get too 2011-10-27T21:54:12 that is why I was using BFS 2011-10-27T21:54:22 it will give u the closest one 2011-10-27T21:54:34 yes, provably optimal and all that 2011-10-27T21:54:46 *** arx has quit IRC (Ping timeout: 260 seconds) 2011-10-27T21:54:51 and its fast enough. Only looking at small area 2011-10-27T21:56:42 hmm, i keep putting off ai/ml/db -class to work on my bot 2011-10-27T21:56:47 idk.. by the time I combine looping over all food doing manhattan distance and then doing the A*.. I thinking BFS will be just as fast.. especially if the food happens to be right next to the ant 2011-10-27T21:57:03 cyphase: how is the db class 2011-10-27T21:57:07 I didn't sign up for that one 2011-10-27T21:57:38 teryret, it's fine. so far i've only watched what i've need to finish the first homework 2011-10-27T21:57:44 needed* 2011-10-27T21:58:06 the professor explains things fairly well 2011-10-27T21:59:00 there was some DTD stuff i thought was going to be tedious, but honestly it wasn't bad 2011-10-27T21:59:03 and the BFS means we travel to the closest one 2011-10-27T22:03:45 http://ants.fluxid.pl/replay.9753 massacre... 2011-10-27T22:07:32 *** TheGeek has joined #aichallenge 2011-10-27T22:09:45 Hey, I tried to register, and it said that my email was already in use... It directed me to this channel, so, what is the best way to get up and running? 2011-10-27T22:10:16 Fluxid, on your server, how easy is it for a new, better version of a bot to rise in the rankings given the previous version's ranking 2011-10-27T22:10:46 as hard as on official. i don't really know how trueskill work 2011-10-27T22:11:15 Fluxid, well on official, ranking and skill are scrubbed for a new upload 2011-10-27T22:11:46 cyphase: tcp doesn't detect if you use never version 2011-10-27T22:11:54 if you use same nick - same rank 2011-10-27T22:11:59 i know, that's my point 2011-10-27T22:12:07 different, new nick - from start 2011-10-27T22:12:16 as after upload on official 2011-10-27T22:12:22 Fluxid, your site serves the replays without any http response headers, it breaks some proxies 2011-10-27T22:12:38 any of the contest organizers online? 2011-10-27T22:12:52 notably mine :) please fix 2011-10-27T22:12:56 how hard would it be to have a reset flag sent when you connect your bot? 2011-10-27T22:13:03 Mooloo: its python http server behind nginx revproxy... 2011-10-27T22:13:28 all other pages work, just the replays are apparently just pinted 2011-10-27T22:13:31 printed 2011-10-27T22:13:39 Mooloo: gimme a link so i can check 2011-10-27T22:14:28 ants.fluxid.pl/replay.100 2011-10-27T22:15:00 link which doesn't work... okay, i'll look into broswer requests >_> 2011-10-27T22:15:23 i looked at it in wireshark. the response is the replay 2011-10-27T22:15:27 just no headers at all 2011-10-27T22:16:31 this exact url doesn't work? 2011-10-27T22:16:58 yes 2011-10-27T22:18:05 http://paste.pocoo.org/show/499334/ 2011-10-27T22:22:09 if it doesn't work with your proxy... did you check what proxy gets or what you get from proxy? 2011-10-27T22:22:09 *** smiley1983 has joined #aichallenge 2011-10-27T22:22:28 curl is lying 2011-10-27T22:22:39 that works here too, but with telnet, there is no header 2011-10-27T22:22:58 duh 2011-10-27T22:24:15 duh? it should 2011-10-27T22:24:26 here: http://paste.pocoo.org/show/499338/ 2011-10-27T22:24:32 yeah, i see 2011-10-27T22:25:10 *** TheGeek has quit IRC () 2011-10-27T22:25:43 Mooloo: no idea, this is probably some problem with python script 2011-10-27T22:25:59 did you try other's tcp servers? 2011-10-27T22:26:08 no 2011-10-27T22:26:36 is there some list? 2011-10-27T22:29:10 i don's remember the link, look at forum 2011-10-27T22:29:23 ok 2011-10-27T22:32:59 *** amstan_ has joined #aichallenge 2011-10-27T22:32:59 *** ChanServ sets mode: +o amstan_ 2011-10-27T22:34:03 *** amstan_ has quit IRC (Client Quit) 2011-10-27T22:34:17 *** amstan_ has joined #aichallenge 2011-10-27T22:34:17 *** ChanServ sets mode: +o amstan_ 2011-10-27T22:38:14 http://213.88.39.97:2080/ is another one, it does not have this problem 2011-10-27T22:38:28 but seems an older version. anyway, thx for the help. 2011-10-27T22:40:23 *** analyst74 has joined #aichallenge 2011-10-27T22:55:51 *** jesufer has joined #aichallenge 2011-10-27T22:56:29 *** Mayank has quit IRC (Quit: Leaving.) 2011-10-27T23:06:15 meh.. I'm 2999th 2011-10-27T23:06:54 *** moises has quit IRC (Remote host closed the connection) 2011-10-27T23:07:01 kinda good number 2011-10-27T23:08:02 wow 2011-10-27T23:08:04 http://ants.fluxid.pl/replay.9800 2011-10-27T23:08:10 look at A's movements! 2011-10-27T23:08:55 especially at the end 2011-10-27T23:09:12 wtf with this watery flowy ant flocking 2011-10-27T23:09:29 <_flag> haha 2011-10-27T23:09:32 <_flag> A is mine btw 2011-10-27T23:09:45 cool 2011-10-27T23:09:49 what do you do? 2011-10-27T23:10:04 <_flag> I send all my ants to kill enemies? 2011-10-27T23:10:25 *** Rainydays has quit IRC (Quit: ChatZilla 0.9.87 [SeaMonkey 2.4.1/20110928161145]) 2011-10-27T23:10:36 yeah, but until the end you don't know bearoff's hill location 2011-10-27T23:10:37 *** amstan_ has quit IRC (Ping timeout: 240 seconds) 2011-10-27T23:10:43 killing machiiiine 2011-10-27T23:11:04 <_flag> Sure but I've scouted the entire rest of the map 2011-10-27T23:11:28 <_flag> And I know where his ants are 2011-10-27T23:11:32 *** amstan_ has joined #aichallenge 2011-10-27T23:11:32 *** ChanServ sets mode: +o amstan_ 2011-10-27T23:11:40 hi 2011-10-27T23:11:43 and when you find the hill all your ants suddenly reorganize... ungroup 2011-10-27T23:12:17 when i check my logs i get invalid orders 2011-10-27T23:12:24 can anyone help me here 2011-10-27T23:12:35 wow 2011-10-27T23:12:41 that is some sick action 2011-10-27T23:12:42 did you try to move on water rj_? 2011-10-27T23:12:47 but i'm happy my stupid bot still can pwn yours somehow D: 2011-10-27T23:12:55 <_flag> :/ 2011-10-27T23:13:01 <_flag> First game I should have won 2011-10-27T23:13:11 rj_: food blocks too 2011-10-27T23:13:14 rj_: or did you tried to move ants which are not yours? 2011-10-27T23:13:28 _which one? 2011-10-27T23:13:32 _flag: which one? 2011-10-27T23:13:41 <_flag> This one: http://ants.fluxid.pl/replay.9814 2011-10-27T23:14:02 i get this 2011-10-27T23:14:04 turn 7 bot 0 invalid actions: 2011-10-27T23:14:11 dont know which one i tried 2011-10-27T23:15:15 rj_: you can log to stderr 2011-10-27T23:15:20 i hadn't considered that yet. even without taking advantage of symmetry, you can narrow down where the enemy hill(s) are 2011-10-27T23:15:20 and you know how man there are because it's the same as you 2011-10-27T23:15:20 many* 2011-10-27T23:15:28 log your movements to stderr 2011-10-27T23:17:02 i am just trying to do things from tutorial 2011-10-27T23:17:17 i ll try to log on stderr 2011-10-27T23:17:19 n see 2011-10-27T23:18:39 *** avdg has quit IRC (Quit: Leaving.) 2011-10-27T23:20:12 *** teryret has quit IRC (Quit: Page closed) 2011-10-27T23:20:32 *** avdg has joined #aichallenge 2011-10-27T23:23:36 *** amstan_ has quit IRC (Ping timeout: 240 seconds) 2011-10-27T23:26:13 *** danielharan has joined #aichallenge 2011-10-27T23:31:34 *** danielha_ has joined #aichallenge 2011-10-27T23:31:35 *** danielharan has quit IRC (Read error: Connection reset by peer) 2011-10-27T23:31:44 *** danielha_ has quit IRC (Remote host closed the connection) 2011-10-27T23:32:17 *** danielharan has joined #aichallenge 2011-10-27T23:35:23 *** bobbydroptable has joined #aichallenge 2011-10-27T23:35:49 Is this an EPIC battle or what!? http://aichallenge.org/visualizer.php?game=21067&user=284 2011-10-27T23:35:53 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-27T23:36:02 Will team submissions be ranked with individual submissions? Or are they separate? 2011-10-27T23:36:28 lot of mixed colors :-) 2011-10-27T23:36:34 *** danielharan has joined #aichallenge 2011-10-27T23:38:18 avdg yeh . watch r-bit. lol 2011-10-27T23:38:58 Wouldn't it be better to live in a world where every color got along? 2011-10-27T23:39:27 We have the power to make this happen 2011-10-27T23:39:53 the best battles are interrupted by turn limit :( 2011-10-27T23:40:13 meh, many bots are sucking in attacking on open fields 2011-10-27T23:40:21 i hope in finals turn limit will be 2x or 3x as long 2011-10-27T23:40:39 bots may be more interesting in the finals 2011-10-27T23:40:59 just a few running behind, and the team mates in front don't even care about these ants they hunting at 2011-10-27T23:41:03 hopefully bots will be a lot better by then 2011-10-27T23:41:06 My bot really doesn't throw down until round 2000 2011-10-27T23:41:41 *** Fandekasp has joined #aichallenge 2011-10-27T23:46:09 *** xar0l_ has joined #aichallenge 2011-10-27T23:50:17 *** jsmnsr has joined #aichallenge 2011-10-27T23:51:06 what python libraries are included with the contest interpreter 2011-10-27T23:51:44 *** danielharan has quit IRC (Ping timeout: 244 seconds) 2011-10-27T23:54:02 amstan: can we do anything about this http://aichallenge.org/forums/viewtopic.php?f=9&t=1402 ? 2011-10-27T23:54:06 i too would like to watch tron games 2011-10-27T23:55:06 I like this one better bobbydrop. http://aichallenge.org/visualizer.php?game=17360&user=4513 2011-10-27T23:56:20 nice 2011-10-27T23:57:37 hmm, some bot like to group them self, but are still scared to attack when they are big enough 2011-10-27T23:57:44 its actually funny 2011-10-27T23:58:04 1 big group of at least 50 ants, guarded by like 3 ants 2011-10-27T23:58:29 is there an ascii visualizer? 2011-10-27T23:58:56 avdg, You can't see the expressions on their faces. You wouldn't want to mess woth those 3 ants either. 2011-10-27T23:59:50 jsmnsr, So you can pretend to be working? :P