2011-11-21T00:08:21 *** u_ has quit IRC (Quit: u_) 2011-11-21T00:20:02 *** hacklash has quit IRC (Quit: hacklash) 2011-11-21T00:29:43 *** Areks has joined #aichallenge 2011-11-21T00:40:29 *** ibdknox has joined #aichallenge 2011-11-21T00:47:58 *** Palmik has joined #aichallenge 2011-11-21T01:00:23 *** ibdknox has quit IRC (Remote host closed the connection) 2011-11-21T01:01:52 *** chridi has joined #aichallenge 2011-11-21T01:01:52 *** chridi has joined #aichallenge 2011-11-21T01:04:19 *** Jak_o_Shadows has quit IRC (Remote host closed the connection) 2011-11-21T01:04:36 *** JorgeB_ has joined #aichallenge 2011-11-21T01:07:56 *** Jak_o_Shadows has joined #aichallenge 2011-11-21T01:08:13 *** Fandekasp has joined #aichallenge 2011-11-21T01:08:16 *** JorgeB has quit IRC (Ping timeout: 240 seconds) 2011-11-21T01:08:16 *** JorgeB_ is now known as JorgeB 2011-11-21T01:09:37 *** roflmao has quit IRC (Remote host closed the connection) 2011-11-21T01:10:20 *** Jak_o_Shadows has quit IRC (Client Quit) 2011-11-21T01:11:04 *** grom has joined #aichallenge 2011-11-21T01:13:50 *** Jak_o_Shadows has joined #aichallenge 2011-11-21T01:14:53 *** grom358 has quit IRC (Ping timeout: 252 seconds) 2011-11-21T01:19:28 *** twymer has quit IRC (Read error: Connection reset by peer) 2011-11-21T01:21:35 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-11-21T01:22:34 *** bmh has joined #aichallenge 2011-11-21T01:27:03 Are there any space filling curves appropriate for use on a rectangular torus? 2011-11-21T01:29:39 *** retybok_ has joined #aichallenge 2011-11-21T01:36:02 *** retybok_ has quit IRC (Ping timeout: 252 seconds) 2011-11-21T01:38:42 *** Harpyon has joined #aichallenge 2011-11-21T01:41:34 *** Harpyon_ has joined #aichallenge 2011-11-21T01:43:43 *** Harpyon has quit IRC (Quit: Computer has gone to sleep.) 2011-11-21T01:43:44 *** Harpyon_ is now known as Harpyon 2011-11-21T01:46:28 *** dlila has joined #aichallenge 2011-11-21T01:58:09 *** bhasker has joined #aichallenge 2011-11-21T01:58:19 wassup 2011-11-21T02:01:04 *** xar0l has quit IRC (Quit: Page closed) 2011-11-21T02:07:08 *** Conorach has quit IRC (Ping timeout: 252 seconds) 2011-11-21T02:17:51 *** Garf has joined #aichallenge 2011-11-21T02:19:45 *** speakeasy has joined #aichallenge 2011-11-21T02:26:55 *** Fandekasp has quit IRC (Quit: leaving) 2011-11-21T02:27:14 *** Fandekasp has joined #aichallenge 2011-11-21T02:29:02 *** CowTipperVirus has quit IRC (Ping timeout: 265 seconds) 2011-11-21T02:31:53 *** bmh has left #aichallenge 2011-11-21T02:34:39 *** speakeasy has quit IRC (Ping timeout: 248 seconds) 2011-11-21T02:34:53 *** ikaros has joined #aichallenge 2011-11-21T02:47:26 *** HaraKiri has joined #aichallenge 2011-11-21T02:48:26 *** dlila has quit IRC (Quit: Leaving) 2011-11-21T02:49:12 *** JorgeB has quit IRC (Ping timeout: 245 seconds) 2011-11-21T02:50:45 *** epicmonkey has joined #aichallenge 2011-11-21T02:52:25 *** g0llum has joined #aichallenge 2011-11-21T03:00:15 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-11-21T03:04:38 *** rajanaresh has left #aichallenge 2011-11-21T03:16:15 *** flag has quit IRC (Ping timeout: 248 seconds) 2011-11-21T03:22:28 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-21T03:25:53 *** RedTurtle has joined #aichallenge 2011-11-21T03:29:13 *** GeKKo_ has joined #aichallenge 2011-11-21T03:29:38 any admins present ? 2011-11-21T03:37:42 *** bhasker has quit IRC (Quit: bhasker) 2011-11-21T03:38:43 *** epicmonkey has quit IRC (Ping timeout: 252 seconds) 2011-11-21T03:39:04 *** sigh has joined #aichallenge 2011-11-21T03:46:04 *** chridi has quit IRC (Quit: leaving) 2011-11-21T03:46:16 *** mj41 has joined #aichallenge 2011-11-21T03:48:44 *** mviel has joined #aichallenge 2011-11-21T04:11:44 *** Fandekasp has quit IRC (Ping timeout: 244 seconds) 2011-11-21T04:11:44 *** dorisabayon has joined #aichallenge 2011-11-21T04:11:47 *** heinrich5991 has quit IRC (Ping timeout: 244 seconds) 2011-11-21T04:11:47 *** goffrie has quit IRC (Remote host closed the connection) 2011-11-21T04:11:48 *** goffrie has joined #aichallenge 2011-11-21T04:11:58 *** RedTurtle has left #aichallenge 2011-11-21T04:13:27 *** meem1029 has joined #aichallenge 2011-11-21T04:13:56 *** heinrich5991 has joined #aichallenge 2011-11-21T04:14:31 do any of you know wether it's possible to change your name in the challenge ? 2011-11-21T04:19:17 *** mj41 has quit IRC (Ping timeout: 260 seconds) 2011-11-21T04:22:32 *** ikaros has joined #aichallenge 2011-11-21T04:24:08 GeKKo_: As far as I know it's not, but you could try asking on the forums for help. 2011-11-21T04:25:15 *** hkraal has quit IRC (Quit: Leaving) 2011-11-21T04:26:13 *** mj41 has joined #aichallenge 2011-11-21T04:35:20 it's hard to make battle code that doesn't turn your bot into a pussy aha 2011-11-21T04:36:47 I find that sexist and offensive. 2011-11-21T04:39:47 *** Zaphus has joined #aichallenge 2011-11-21T04:43:53 *** raemde has quit IRC (Read error: Connection reset by peer) 2011-11-21T04:45:53 *** grwip has joined #aichallenge 2011-11-21T04:45:53 Hey look. Now that I started caching my A* results my bot actually goes at a reasonable speed! 2011-11-21T04:46:55 *** ALplus has quit IRC (Ping timeout: 248 seconds) 2011-11-21T04:54:08 *** Palmik_ has joined #aichallenge 2011-11-21T04:54:19 *** Palmik has quit IRC (Read error: Connection reset by peer) 2011-11-21T05:02:01 *** snthaoeu has quit IRC (Quit: Page closed) 2011-11-21T05:05:02 *** b0rder has quit IRC (Ping timeout: 245 seconds) 2011-11-21T05:05:39 *** mikewintermute has joined #aichallenge 2011-11-21T05:06:15 *** Euphony has joined #aichallenge 2011-11-21T05:11:00 *** avdg1 has joined #aichallenge 2011-11-21T05:13:35 *** avdg has quit IRC (Ping timeout: 248 seconds) 2011-11-21T05:15:18 *** replore has quit IRC (Remote host closed the connection) 2011-11-21T05:19:08 meem1029: thanks 2011-11-21T05:19:23 I wanted to pm one of the forum admins, but since I haven't got any posts, I am not allowed to do this 2011-11-21T05:25:58 *** Yexo_ has joined #aichallenge 2011-11-21T05:28:30 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-21T05:30:06 *** Yexo has quit IRC (Ping timeout: 276 seconds) 2011-11-21T05:33:31 *** Zaphus has quit IRC (Quit: Page closed) 2011-11-21T05:59:18 *** hkraal has joined #aichallenge 2011-11-21T06:00:47 *** UncleVasya has joined #aichallenge 2011-11-21T06:01:19 *** AntDroid has joined #aichallenge 2011-11-21T06:01:47 *** UncleVasya has quit IRC (Remote host closed the connection) 2011-11-21T06:06:57 *** ikaros has joined #aichallenge 2011-11-21T06:11:04 *** pairofdice has joined #aichallenge 2011-11-21T06:15:20 *** hkraal has quit IRC (Remote host closed the connection) 2011-11-21T06:16:45 *** Anilm3 has joined #aichallenge 2011-11-21T06:16:58 Hi 2011-11-21T06:17:25 *** hkraal has joined #aichallenge 2011-11-21T06:17:36 *** GeKKo_ has quit IRC (Quit: Page closed) 2011-11-21T06:26:04 *** liberforce has joined #aichallenge 2011-11-21T06:34:55 *** mikewintermute has quit IRC (Quit: mikewintermute) 2011-11-21T06:45:37 *** Euphony has quit IRC (Quit: Page closed) 2011-11-21T07:00:24 *** flag has joined #aichallenge 2011-11-21T07:02:06 *** zanir has joined #aichallenge 2011-11-21T07:02:09 *** Anilm3 has quit IRC (Quit: leaving) 2011-11-21T07:02:13 hello 2011-11-21T07:03:05 quick question: when an enemy hill is razed, is it still read in as an ant hill? 2011-11-21T07:03:26 i.e. do I need to check to see if the spot I'm on used to be an enemy ant hill? 2011-11-21T07:05:33 *** praveen_ has joined #aichallenge 2011-11-21T07:05:42 dead hills are gone iirc 2011-11-21T07:05:48 thanks 2011-11-21T07:10:23 *** mj41 has quit IRC (Ping timeout: 248 seconds) 2011-11-21T07:22:17 *** replore_ has joined #aichallenge 2011-11-21T07:22:25 *** praveen_ has quit IRC (Ping timeout: 265 seconds) 2011-11-21T07:25:59 *** mj41 has joined #aichallenge 2011-11-21T07:28:33 *** replore_ has quit IRC (Remote host closed the connection) 2011-11-21T07:31:11 *** ikaros has quit IRC (Ping timeout: 248 seconds) 2011-11-21T07:32:38 *** krikav is now known as frimend 2011-11-21T07:32:50 *** frimend has joined #aichallenge 2011-11-21T07:33:36 *** frimend has left #aichallenge 2011-11-21T07:34:45 *** Jak_o_Shadows has quit IRC (Remote host closed the connection) 2011-11-21T07:36:33 *** Antvolution has joined #aichallenge 2011-11-21T07:37:27 *** TheLinker has joined #aichallenge 2011-11-21T07:37:55 *** Yexo_ has quit IRC (Quit: Leaving) 2011-11-21T07:38:26 *** Yexo has joined #aichallenge 2011-11-21T07:42:36 *** Antvolution has quit IRC (Quit: Leaving.) 2011-11-21T07:45:27 *** flag has quit IRC (Ping timeout: 245 seconds) 2011-11-21T07:46:38 *** JamesMG_ has joined #aichallenge 2011-11-21T07:47:11 *** JamesMG has quit IRC (Read error: Connection reset by peer) 2011-11-21T07:47:15 *** emilk has quit IRC (Ping timeout: 276 seconds) 2011-11-21T07:47:40 *** emilk has joined #aichallenge 2011-11-21T07:47:53 *** knyppeldynan has quit IRC (Remote host closed the connection) 2011-11-21T07:48:07 *** knyppeldynan has joined #aichallenge 2011-11-21T07:50:49 *** liberforce has quit IRC (Quit: Leaving.) 2011-11-21T07:52:19 *** liberforce has joined #aichallenge 2011-11-21T07:52:24 *** Antvolution has joined #aichallenge 2011-11-21T07:54:27 *** praveen_ has joined #aichallenge 2011-11-21T07:59:17 *** kiv has joined #aichallenge 2011-11-21T08:06:04 *** ikaros has joined #aichallenge 2011-11-21T08:11:04 *** Antvolution has quit IRC (Quit: Leaving.) 2011-11-21T08:24:22 *** amstan has joined #aichallenge 2011-11-21T08:24:22 *** ChanServ sets mode: +o amstan 2011-11-21T08:33:32 *** bosie has joined #aichallenge 2011-11-21T08:38:45 games are slow :P 2011-11-21T08:39:47 *** Surya has joined #aichallenge 2011-11-21T08:51:42 *** Antvolution has joined #aichallenge 2011-11-21T08:52:10 *** bugnuts has left #aichallenge ("Leaving") 2011-11-21T08:52:44 *** bugnuts has joined #aichallenge 2011-11-21T08:55:51 *** u_ has joined #aichallenge 2011-11-21T08:56:18 *** u__ has joined #aichallenge 2011-11-21T08:59:43 *** amstan has quit IRC (Ping timeout: 248 seconds) 2011-11-21T09:00:15 *** u_ has quit IRC (Ping timeout: 248 seconds) 2011-11-21T09:00:15 *** u__ is now known as u_ 2011-11-21T09:01:22 *** bugnuts has quit IRC (Quit: Leaving) 2011-11-21T09:04:24 *** praveen_ has quit IRC (Ping timeout: 265 seconds) 2011-11-21T09:04:24 *** meem1029 has quit IRC (Ping timeout: 265 seconds) 2011-11-21T09:04:53 *** AntDroid has quit IRC (Ping timeout: 265 seconds) 2011-11-21T09:06:42 I wonder how good the ranking system is optimized in making games slower 2011-11-21T09:19:26 *** sigh has quit IRC (Remote host closed the connection) 2011-11-21T09:27:45 *** GeorgeSebastian has joined #aichallenge 2011-11-21T09:27:45 *** GeorgeSebastian has joined #aichallenge 2011-11-21T09:31:14 *** spartiant has joined #aichallenge 2011-11-21T09:32:55 *** xathis has joined #aichallenge 2011-11-21T09:33:26 *** raemde has joined #aichallenge 2011-11-21T09:33:28 *** spartiant has quit IRC (Client Quit) 2011-11-21T09:33:43 *** Conorach has joined #aichallenge 2011-11-21T09:35:10 *** seletskiy has quit IRC (Ping timeout: 260 seconds) 2011-11-21T09:35:49 *** seletskiy has joined #aichallenge 2011-11-21T09:36:40 *** olexs has joined #aichallenge 2011-11-21T09:49:37 *** olexs has quit IRC (Read error: Connection reset by peer) 2011-11-21T09:50:17 *** olexs has joined #aichallenge 2011-11-21T09:50:37 *** Antvolution has quit IRC (Read error: Connection reset by peer) 2011-11-21T10:01:22 *** bugnuts has joined #aichallenge 2011-11-21T10:02:47 *** Harpyon has joined #aichallenge 2011-11-21T10:04:26 *** Harpyon has quit IRC (Client Quit) 2011-11-21T10:09:01 *** Areks has quit IRC (Ping timeout: 240 seconds) 2011-11-21T10:13:13 *** twymer has joined #aichallenge 2011-11-21T10:45:15 *** UncleVasya has joined #aichallenge 2011-11-21T10:48:12 *** foRei has joined #aichallenge 2011-11-21T10:52:55 *** praveen_ has joined #aichallenge 2011-11-21T11:00:49 *** tobym has joined #aichallenge 2011-11-21T11:16:50 *** bmh has joined #aichallenge 2011-11-21T11:18:37 *** olexs has quit IRC (Quit: Leaving.) 2011-11-21T11:19:31 now that I have a grasp on foraging, all I need is fighting ;) 2011-11-21T11:23:14 I'll have one with everything 2011-11-21T11:28:43 Forums accounts are separate from bot accounts, 'eh? 2011-11-21T11:29:23 *** Antimony has joined #aichallenge 2011-11-21T11:29:33 am still struggling to forage, in very bad shape 2011-11-21T11:29:41 *** NightExcessive has quit IRC (Quit: Leaving) 2011-11-21T11:30:13 *** NightExcessive has joined #aichallenge 2011-11-21T11:32:11 praveen_: What are you trying to get your bot to do? 2011-11-21T11:32:20 NightExcessive: Did I mention that I've abandoned diffusion? 2011-11-21T11:32:34 am trying to distribute it while exploring 2011-11-21T11:33:00 praveen_: Your ants are clumping up? 2011-11-21T11:33:01 bmh: using BFS to unseen regions 2011-11-21T11:33:28 praveen_: "Go toward the nearest unseen square"? 2011-11-21T11:33:46 *** Blkt has joined #aichallenge 2011-11-21T11:33:58 bmh: Yes, as BFS to unseen node is working right but had issues 2011-11-21T11:34:48 praveen_: Could you try this -- Once an ant finds an unseen square, it claims that square and a bunch of other adjacent squares 2011-11-21T11:35:09 bmh: BFS to unseen square is the idea to distribute them as BFS to unseen node will not actually distribute 2011-11-21T11:35:32 sure. 2011-11-21T11:35:49 Where 'square' is some collection of nodes? 2011-11-21T11:35:50 bmh: square - square of tiles right? 2011-11-21T11:35:55 ok 2011-11-21T11:36:21 Once one ant decides to go to a square, can another ant go to the same square? 2011-11-21T11:36:25 bmh: are you using tracking? 2011-11-21T11:36:36 praveen_: I'm not. I'm going to implement it in the next few days. 2011-11-21T11:37:13 I've been busy getting my pathfinder to work at all :) 2011-11-21T11:37:21 only one ant will go for one square. But each turn, I recalculate all the squares and again find which ants is the best to reach which square and this is causing huge oscillations 2011-11-21T11:37:40 bmh: :) Ya No one can save if there is a bug in pathfinder 2011-11-21T11:38:15 aichallenge: janzert epsilon * r3172f6d / (3 files in 2 dirs): Add pypy support - http://git.io/nGc4PA 2011-11-21T11:38:15 aichallenge: janzert epsilon * r3bb139c / (manager/sql.py manager/worker_ssh.py): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/kytevw 2011-11-21T11:38:20 praveen_: I'm trying to do some silly fancy stuff to get fast paths so I can run a traveling salesman solver on the map. 2011-11-21T11:38:28 ant is getting assigned to different square each time I guess although I am trying to maintain the ordering by sorting 2011-11-21T11:38:37 i'm actually pretty happy with my battle code atm 2011-11-21T11:38:50 *** bosie has left #aichallenge 2011-11-21T11:38:51 still got to code up a few other things before the bot'll be any good 2011-11-21T11:38:56 >< 2011-11-21T11:38:56 . 2011-11-21T11:39:06 !$ 2011-11-21T11:39:11 bmh: TSP, so that one ant could collect all food within a square in the shortest manner? 2011-11-21T11:39:45 *** analyst74 has joined #aichallenge 2011-11-21T11:40:07 praveen_: Just for the early game where there are few enough ants that they can be considered to be independent 2011-11-21T11:40:07 antimatroid: maybe you will spend your free time thinking about Tron strategy instead of making bot for Ants? 2011-11-21T11:40:17 * UncleVasya is tying to kick out strong opponent 2011-11-21T11:40:34 This is way cooler than CoreWars 2011-11-21T11:40:38 UncleVasya: i've been programming pretty much non-stop the last few days :P 2011-11-21T11:40:48 about to climb into bed 2011-11-21T11:41:07 UncleVasya: What's the name of your bot? 2011-11-21T11:41:09 just kidding, I'm glad you found a time to write your entry to this challenge :) 2011-11-21T11:41:22 http://aichallenge.org/profile.php?user=53 2011-11-21T11:41:55 nice! I've sworn that I won't take any vacation time to work on this. 2011-11-21T11:42:29 i'm technically on indefinite vacation atm :P 2011-11-21T11:43:22 UncleVasya: your strategy looks like "get more ants then them. Then win" 2011-11-21T11:43:48 bmh: are you using the visualizer mentioned in forums which give more insights? 2011-11-21T11:44:00 *** bmh has quit IRC (Quit: gone) 2011-11-21T11:45:32 It looks like that because I haven't done a good prunning yet and my ant can't find even some normal solution for large battles so my bot just stay when it happens. 2011-11-21T11:46:12 my ant <--> my bot 2011-11-21T11:46:29 ant hivemind 2011-11-21T11:50:33 *** besh has joined #aichallenge 2011-11-21T11:51:34 antimatroid: I'd like to discuss your maze generation 2011-11-21T11:52:06 with tile symmetry, how did you ensure each separate tile was connected to the others? 2011-11-21T11:56:05 *** amstan has joined #aichallenge 2011-11-21T11:56:05 *** ChanServ sets mode: +o amstan 2011-11-21T11:56:34 janzert: I thought pypy had a ppa? 2011-11-21T11:56:44 janzert: http://morepypy.blogspot.com/2010/03/introducing-nightly-builds-and-ubuntu.html 2011-11-21T11:56:45 just a defunct one 2011-11-21T11:56:52 right 2011-11-21T11:56:55 oh, ok 2011-11-21T11:57:21 so, hows the matchups going? are we still playing lots of starter bots? 2011-11-21T11:57:32 ooh, someone actually did finally add 1.6 to it anyway 2011-11-21T11:57:53 not many I don't believe 2011-11-21T11:58:04 Pairings look pretty good 2011-11-21T11:58:21 I keep seeing forum posts on the "what's up with the game rate" thread 2011-11-21T11:58:41 we've got 2620 bots that did not have a game in the last 24 hours 2011-11-21T11:59:26 so over 1/3 2011-11-21T11:59:41 looks like the game rates are still bad though 2011-11-21T11:59:45 or, uneven 2011-11-21T12:00:04 that might be inactive bots... 2011-11-21T12:00:11 Mm, I've had 7? games in 24hours 2011-11-21T12:00:18 within about 3 or 4 I believe 2011-11-21T12:01:18 whoa, I get 863 active bots? 2011-11-21T12:01:38 that's about right 2011-11-21T12:01:50 811 that are older than 24 hours 2011-11-21T12:02:01 the account is older than 24 hours that is 2011-11-21T12:02:39 *** Antimony has quit IRC (Ping timeout: 248 seconds) 2011-11-21T12:02:45 http://pastebin.com/Y50UtUkk 2011-11-21T12:03:30 *** treeform has quit IRC (Remote host closed the connection) 2011-11-21T12:03:37 *** Antimony has joined #aichallenge 2011-11-21T12:04:23 janzert: so, those ~900 are pulling in about 5000 inactive? 2011-11-21T12:04:51 pulling in less than that 2011-11-21T12:05:39 ~900 pulling in ~3200-3300 2011-11-21T12:05:57 ~6800 total, ~2600 not playing = ~4200 playing - ~900 active = ~3500 getting pulled in 2011-11-21T12:06:14 ok, :) 2011-11-21T12:06:16 roughly :) 2011-11-21T12:06:26 so, what about another bias? 2011-11-21T12:06:36 maybe only pull in 1 inactive per match? 2011-11-21T12:06:44 or only 50% can be inactive? 2011-11-21T12:06:46 I'd really rather not 2011-11-21T12:07:24 I'm thinking that if 1 active pulls in only inactive, that isn't helpful to those participating actively 2011-11-21T12:07:34 how about every match must have 2 active? 2011-11-21T12:07:36 once we start doing that the ranking system will have the potential for more problems 2011-11-21T12:07:57 such as? 2011-11-21T12:08:07 if that 1 active happens to be a starter bot it should be playing starter bots 2011-11-21T12:08:22 and there is a good chance it should be inactive 2011-11-21T12:08:34 the opponents that is 2011-11-21T12:08:39 I don't think that matters. 2011-11-21T12:08:51 if the new starter plays real bots, it will lose and be pushed down faster 2011-11-21T12:09:01 farther than it should be 2011-11-21T12:09:04 *** lericson has quit IRC (Ping timeout: 240 seconds) 2011-11-21T12:09:33 so you think our pairing can drastically affect the skill? 2011-11-21T12:09:48 also given 10 bots in available range, there could be only 1 inactive bot, making that bot more likely to be chosen 2011-11-21T12:09:51 of course 2011-11-21T12:11:05 I'd like to see the games per 24 hours be at least 24 2011-11-21T12:11:08 extreme example, if you play one bot of pair only against xathis and one only against starters the skill result will be drastically different 2011-11-21T12:11:21 McLeopold: that'd be higher than planetwars 2011-11-21T12:11:34 I thought planetwars was around 48? 2011-11-21T12:11:49 roughly every 90 minutes I'm pretty sure 2011-11-21T12:12:03 so ~16 2011-11-21T12:12:13 we are still less than half that 2011-11-21T12:12:45 moar cpus 2011-11-21T12:12:48 yeah, and a way more computationally intensive game plus a whole lot more participants 2011-11-21T12:13:21 so, what about dropping starter bots? 2011-11-21T12:13:56 I'm more against it now 2011-11-21T12:14:16 you think the provide useful info? 2011-11-21T12:14:21 amstan and I discussed it for a while after you asked the other night and pretty much talked me out of it 2011-11-21T12:14:56 I think they provide a useful service more than useful info :) 2011-11-21T12:15:00 McLeopold: we have too many versions of the starter bots, it would be too painful to have hashes for all of them 2011-11-21T12:15:09 i don't even think we can get some of the older hashes 2011-11-21T12:15:19 amstan: that is not a problem for me, I can classify them 2011-11-21T12:15:22 but I need to run, I should be around later this evening 2011-11-21T12:15:40 McLeopold: how will you classify them? all you have a is a zip 2011-11-21T12:16:06 if someone unpacked them, then repacked them, it's likely that the dates inside the zip changed, so the hash changed 2011-11-21T12:16:30 amstan: I'll start with the most comming duplicate hashes and work down from there confirming starter bot behavior 2011-11-21T12:16:32 if we do hashes on the inside, we'll have to unpack them on the manager, right now they're just kept as zips in there 2011-11-21T12:16:32 *** jstrong has joined #aichallenge 2011-11-21T12:16:38 /common/ 2011-11-21T12:16:53 a hash of the zip should be good enough 2011-11-21T12:17:14 i'm pretty sure the zips contain modified date in them 2011-11-21T12:17:39 so we cann't get the hashes from the old starter pack versions, since any git checkout changes the modified date 2011-11-21T12:18:03 again, I just need to look at the most common duplicates in the database 2011-11-21T12:18:34 it may not be 100%, but if we stop playing them, it should make a big difference in game rate 2011-11-21T12:19:14 are you going to disable them like you did for the timeouts? 2011-11-21T12:19:15 McLeopold: I really don't think you're going gain that much anyway, since 1/3 of players all of which are ranked below 1590 aren't playing any games anyway. 2011-11-21T12:19:20 they'll play, but they won't be seeds anymore? 2011-11-21T12:19:32 McLeopold: right! that was another one of my points 2011-11-21T12:19:44 McLeopold: no begginer keeps their starter bot active intentionally 2011-11-21T12:19:59 McLeopold: they'll submit the starter bot and it'll play for a max of 3 days 2011-11-21T12:20:10 What's about make some low-end in strength sparring bot and run a few games against it for every new upload. If uploaded bot is weaker than that 'Guardian', do not accep it ('compiled but is weaker than we want.'?) 2011-11-21T12:20:13 okay, I may still attempt to classify, but I won't do any logic with it 2011-11-21T12:20:28 oh, ok, so just an experiment 2011-11-21T12:20:36 at least to get data 2011-11-21T12:21:41 UncleVasya: we still want to allow any unique bot to compete at any level 2011-11-21T12:22:02 ok 2011-11-21T12:22:04 restricting starter bots is more about restricting users who aren't actively participating 2011-11-21T12:22:18 *** emilk has quit IRC (Read error: Operation timed out) 2011-11-21T12:22:57 amstan: what about for next contest, would you be up for starter bots only getting ~10 per user? 2011-11-21T12:23:17 i'm not sure 2011-11-21T12:23:25 this goes to the core of the contest experience 2011-11-21T12:23:55 i think it's fine the way it is to be honest, as long as they get a decent ammount of games in that 3 day timeout window 2011-11-21T12:24:09 someone suggested since all starters are the same, everyone who uploaded a starter gets the same set of game results. I thought that was interesting 2011-11-21T12:25:08 *** emilk has joined #aichallenge 2011-11-21T12:25:08 what's the point of encouraging uploading pure starter bots? 2011-11-21T12:25:36 McLeopold: they're not all the same, some timeout earlier, some might have slight differences 2011-11-21T12:25:40 *** lericson has joined #aichallenge 2011-11-21T12:25:49 Garf: to test how you upload bots and stuff 2011-11-21T12:26:05 Garf: I believe to make people sure that it's easy to start participiating 2011-11-21T12:27:32 yeah, I think being able to upload a starter bot is cool 2011-11-21T12:27:35 I think the tutorial now says explicitly "now upload your starter bot". Maybe it should say 'now think about how to improve your bot, test locally balbla. If it works, upload your bot like this"' 2011-11-21T12:27:50 however if getting rid of the huge amount of inactive starter bots would make games go faster.. 2011-11-21T12:28:10 *** JorgeB has joined #aichallenge 2011-11-21T12:28:38 if the burden "make improvement to starter bot" is too big, well, should those people be in the contest? :P 2011-11-21T12:28:54 *** besh has quit IRC (Quit: Page closed) 2011-11-21T12:29:22 *** Antimony_ has joined #aichallenge 2011-11-21T12:29:37 Garf: yes, but if they take a week before they start working on their bot, it would be nice for them to see themselves in a few games 2011-11-21T12:30:16 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-11-21T12:30:17 *** Antimony_ is now known as Antimony 2011-11-21T12:31:59 i'm kind of surprised that my bot is timing out in production 2011-11-21T12:33:05 *** mceier has joined #aichallenge 2011-11-21T12:33:24 *** iglo has joined #aichallenge 2011-11-21T12:34:02 if a game has 9 participants or whatever, do they all run on the same worker just taking turns? 2011-11-21T12:34:59 a1k0n: I was lucky that you've got a timeout :) 2011-11-21T12:35:03 :| 2011-11-21T12:35:41 *** bhasker has joined #aichallenge 2011-11-21T12:36:22 try to check fo the time left not only before solving battle position, but also during generating moves ;) 2011-11-21T12:36:48 i leave myself a 25ms window to just output moves 2011-11-21T12:37:00 i think i need to use setitimer 2011-11-21T12:37:08 the question is: are we measuring wall clock time or CPU time? 2011-11-21T12:38:35 yep, you leave time. But the point is that you timeout between checks for the time. 2011-11-21T12:39:13 In the place that you think is always fast. 2011-11-21T12:39:47 agreed 2011-11-21T12:39:55 which is why setitimer is the solution 2011-11-21T12:40:40 *** Antimony has quit IRC (Ping timeout: 260 seconds) 2011-11-21T12:41:59 *** ThatSnail has joined #aichallenge 2011-11-21T12:42:11 a1k0n: yes, 2 at a time 2011-11-21T12:42:23 *** Surya has quit IRC (Ping timeout: 265 seconds) 2011-11-21T12:49:32 *** amstan has quit IRC (Ping timeout: 252 seconds) 2011-11-21T12:55:56 *** ulrik has joined #aichallenge 2011-11-21T12:56:20 *** jstrong is now known as roflmao 2011-11-21T12:59:26 *** mj41 has quit IRC (Ping timeout: 252 seconds) 2011-11-21T13:08:16 *** ikaros has quit IRC (Ping timeout: 240 seconds) 2011-11-21T13:09:46 *** mceier has quit IRC (Quit: leaving) 2011-11-21T13:10:25 *** praveen_ has quit IRC (Ping timeout: 265 seconds) 2011-11-21T13:10:30 *** bhasker has quit IRC (Quit: bhasker) 2011-11-21T13:11:34 *** jacob_strauss has joined #aichallenge 2011-11-21T13:16:33 guess i should resubmit. *sigh8 2011-11-21T13:16:58 *** ulrik has left #aichallenge 2011-11-21T13:17:04 i wish i could like, buy some games 2011-11-21T13:20:57 *** Redgis has joined #aichallenge 2011-11-21T13:23:34 *** treeform has joined #aichallenge 2011-11-21T13:28:38 *** Blkt has quit IRC (Remote host closed the connection) 2011-11-21T13:30:16 *** Antimony has joined #aichallenge 2011-11-21T13:30:16 *** Redgis has quit IRC (Read error: Connection reset by peer) 2011-11-21T13:30:59 *** Redgis has joined #aichallenge 2011-11-21T13:37:05 *** hacklash has joined #aichallenge 2011-11-21T13:40:13 *** Surya has joined #aichallenge 2011-11-21T13:45:48 *** Hexren has joined #aichallenge 2011-11-21T13:45:51 *** Hexren has left #aichallenge 2011-11-21T13:49:44 *** ikaros has joined #aichallenge 2011-11-21T14:00:43 *** choas has joined #aichallenge 2011-11-21T14:06:45 *** ALplus has joined #aichallenge 2011-11-21T14:09:28 *** retybok_ has joined #aichallenge 2011-11-21T14:09:51 aichallenge: janzert epsilon * r3b7d6b5 / setup/retrieve_languages.py : Use the right (64bit) version of pypy - http://git.io/5esyJA 2011-11-21T14:10:06 *** roflmao has quit IRC (Quit: Leaving.) 2011-11-21T14:17:30 *** Antimony has quit IRC (Ping timeout: 260 seconds) 2011-11-21T14:22:59 I just syntax errored by way to discovering a new python idiom... 2011-11-21T14:27:10 *** liberforce has left #aichallenge 2011-11-21T14:27:48 *** Antimony has joined #aichallenge 2011-11-21T14:30:23 *** bhasker has joined #aichallenge 2011-11-21T14:31:30 *** rajanaresh has joined #aichallenge 2011-11-21T14:37:44 anyone have a hunch what would make my java bot timeout on test_game.sh at the first step already, but it runs fine with the startup tutorial sh script ? 2011-11-21T14:39:09 *** Conorach_ has joined #aichallenge 2011-11-21T14:40:24 *** treeform has quit IRC (Remote host closed the connection) 2011-11-21T14:40:53 raemde, that tutorial map is very small, the 'real' ones much larger 2011-11-21T14:41:03 *** treeform has joined #aichallenge 2011-11-21T14:42:36 *** Conorach has quit IRC (Ping timeout: 276 seconds) 2011-11-21T14:43:40 ok 2011-11-21T14:49:20 *** praveen_ has joined #aichallenge 2011-11-21T14:54:51 *** Accoun has quit IRC () 2011-11-21T14:56:22 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-21T15:03:00 *** jstrong1 has joined #aichallenge 2011-11-21T15:03:06 *** jstrong1 is now known as roflmao 2011-11-21T15:05:42 *** speakeasy has joined #aichallenge 2011-11-21T15:06:02 *** zostay has quit IRC (Read error: Connection reset by peer) 2011-11-21T15:06:18 *** zostay has joined #aichallenge 2011-11-21T15:06:31 hi guys 2011-11-21T15:06:47 can i have a hand (again lol im useless)? 2011-11-21T15:07:01 hi 2011-11-21T15:07:58 hey uncle, i believe we have met here before lol, my bot is playing up, can you go through with me how to get from a few .java files to a game where they work? 2011-11-21T15:08:00 *** retybok_ has quit IRC (Ping timeout: 252 seconds) 2011-11-21T15:08:42 Yes, download the starter package from the website ;) 2011-11-21T15:09:11 done that lol 2011-11-21T15:09:22 im trying to follow the tut but its not working for me 2011-11-21T15:09:31 i think i may be doing something wrong 2011-11-21T15:09:46 i have the .java files i need i think 2011-11-21T15:09:51 sry, I do not use java 2011-11-21T15:10:09 what do you use, cos javas useless lol 2011-11-21T15:10:13 *** Accoun has joined #aichallenge 2011-11-21T15:10:14 I think you can try build a .jar-file 2011-11-21T15:10:15 i was thinking of going into python 2011-11-21T15:10:42 thats what the tut says but every time i do it springs loads of errors 2011-11-21T15:10:43 and launch it like 'java -jar mybot.jar' 2011-11-21T15:11:16 hmmm 2011-11-21T15:11:27 I have a question regarding the tests the server performs after uploading the code. I use lisp code and yesterday i uploaded my code with the prints in *debug-io* and it worked. Now it says the tests failed. Isn't *debug-io* supposed to be stderr and why did it accept before and not now? I had to remove all the prints to *debug-io* now. 2011-11-21T15:11:30 You can't build a jar or you can't launchit? 2011-11-21T15:11:30 is python easier? my friends seem to think so 2011-11-21T15:11:47 *** flag has joined #aichallenge 2011-11-21T15:11:52 build the jar itself 2011-11-21T15:14:13 yep, maybe python will be better for you. I think it's esier to start compete with that language than with java. 2011-11-21T15:15:21 lol i think your defo right there, java has always been a little line hungry for me 2011-11-21T15:15:31 thanks 2011-11-21T15:15:41 Also I think in terms of programming a bot for Ants c++ will be easy too. Easy to start and easy to write in. 2011-11-21T15:16:23 hmmm, i will take a look, thanks 2011-11-21T15:17:02 Because you will not feel the famous difficulty of c/c++ until you want to go deep. 2011-11-21T15:17:45 how do you mean? 2011-11-21T15:18:43 Yeah, you don't need many fancy features 2011-11-21T15:21:10 wow python has v few files verses the java starter pack 2011-11-21T15:21:16 java seems icky now 2011-11-21T15:21:46 what sort of information and instructions can you get givven and give out for the bot? 2011-11-21T15:22:01 thanks so much for the help guys, really appreciate it 2011-11-21T15:23:10 I think Java is easier 2011-11-21T15:23:15 c++ bot has quite a few files too, throws atleast me off a bit 2011-11-21T15:23:19 sure, py is easier as a language 2011-11-21T15:23:20 but it is slower 2011-11-21T15:23:27 much much slower 2011-11-21T15:23:42 so unless you have a very good grasp of python, you'll end up making a lot of mistakes that cause your code to run slow and will wonder while 2011-11-21T15:23:43 why* 2011-11-21T15:23:48 speakeasy: Here is detailed game specification: http://aichallenge.org/specification.php#Bot-Input 2011-11-21T15:23:53 well, pypy support was just added 2011-11-21T15:23:53 *** zanir has left #aichallenge 2011-11-21T15:24:00 since java is inherently faster, you can be a bit lazier with your code and it will still run pretty darn fast 2011-11-21T15:24:08 roflmao: On the other side, writing fast incorrect code doesn't help much either. 2011-11-21T15:24:35 not incorrect 2011-11-21T15:24:41 just unoptimized 2011-11-21T15:24:55 as in "oooh I had to use numpy with this custom lib to squeeze those extra 10 ms!" 2011-11-21T15:25:05 ^^ not nearly as much of a problem in Java as in Python 2011-11-21T15:25:24 10ms? I wish 2011-11-21T15:26:17 thats cool that link, thanks 2011-11-21T15:26:17 *** flag has quit IRC (Ping timeout: 252 seconds) 2011-11-21T15:26:18 *** delt0r_ has joined #aichallenge 2011-11-21T15:26:32 I'm just saying this from personal experience :P. I wasn't using Python but I was using ruby which is awfully similar. I was spending insane amounts of time trying to figure out what was the cause of my time-outs and was having to modify my ant behavior code to fit in my slow language. I switched to Java and that all just went poof :-\. 2011-11-21T15:27:22 *** flag has joined #aichallenge 2011-11-21T15:27:44 Yeah, I sqitched to c++ from Python 2011-11-21T15:27:59 *** ALplus has quit IRC (Ping timeout: 248 seconds) 2011-11-21T15:28:15 afaik the only person here who has been able to pull off a high level python bot is Fluxid, but I could be wrong 2011-11-21T15:28:26 everyone else is using either c++ (c# maybe) or java 2011-11-21T15:28:56 oh and some use Go 2011-11-21T15:29:10 Why would anyone use JAva 2011-11-21T15:29:16 C++ is faster and easier 2011-11-21T15:29:50 diff in speed is minimal 2011-11-21T15:29:52 Antimony: Familiarity, a will to live? 2011-11-21T15:29:58 some people know more java than c++ 2011-11-21T15:30:17 * avdg1 is getting frustrated by unpredicted behavior of the c++ streams 2011-11-21T15:30:22 diff is minimal for a project like this **** obviously once you get into heavy memory management c++ blows java out of the water because you have control over pointers 2011-11-21T15:30:35 and can do manual garbage collection 2011-11-21T15:30:57 * avdg1 doesn't think pointers are that worst 2011-11-21T15:32:28 and in case we have a language fight: learn to use each others language ;-) 2011-11-21T15:33:12 All languages suck, in different ways. 2011-11-21T15:33:28 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-11-21T15:33:33 I would use c++ if I was as good at it as I am at Java, but my experience with c++ is nowhere near my experience with Java, so I chose what's more comfortable I guess. 2011-11-21T15:33:37 I hate the language tbh, it sucks. 2011-11-21T15:33:40 (Java) 2011-11-21T15:34:23 well, I just blowed js as my focus language, I'm now trying c++ 2011-11-21T15:34:46 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-21T15:35:09 *** g0llum has joined #aichallenge 2011-11-21T15:35:44 avdg1: the JS runtime on the workers is not using the full JIT :( 2011-11-21T15:36:38 js was getting slow once I added a few more iterations, more than I could predict myself 2011-11-21T15:36:49 (and I know I had to add a lot of iterations on top of it) 2011-11-21T15:37:04 *** rajanaresh has quit IRC (Ping timeout: 240 seconds) 2011-11-21T15:37:20 I profiled my bot once and JSON.serialize was using 90% of the time 2011-11-21T15:37:25 turns out it isn't optimized :-/ 2011-11-21T15:37:39 you'd think that'd be a native code callout or something. guess not. 2011-11-21T15:37:51 Eveery programmers forum, every programmers chat room will have somewhere it's own language holywar. No exceptions :D 2011-11-21T15:38:16 oh well, I try to use as many language as my brain gets comfortable with 2011-11-21T15:38:28 I try not to judge one language, but its hard 2011-11-21T15:39:28 picking a good langauge is tough 2011-11-21T15:39:46 *** mcstar has joined #aichallenge 2011-11-21T15:39:49 If Python would run at C++ speed and be included in the browser, I wouldn't feel the need for language wars. 2011-11-21T15:39:54 As it is, things are more tricky. 2011-11-21T15:40:09 ive arrived at a bad time 2011-11-21T15:40:17 changing is hard as well, because you have to blow up something you know and things you could rely 2011-11-21T15:41:18 and something which is easy for someone is very hard for an other person 2011-11-21T15:41:19 *** Antvolution has joined #aichallenge 2011-11-21T15:41:35 *** mj41 has joined #aichallenge 2011-11-21T15:41:57 I tried a new language for this contest. Might try the same for the next one. 2011-11-21T15:42:15 "the same" meaning, again a new one. 2011-11-21T15:42:39 obviously ;) 2011-11-21T15:42:39 which one did you try now? 2011-11-21T15:42:52 "If Python would run at C++ speed and be included in the browser, I wouldn't feel the need for language wars." --> CoffeeScript 2011-11-21T15:42:54 I don't think I will be able to send my bot before the deadline 2011-11-21T15:43:00 *** avdg1 is now known as avdg 2011-11-21T15:43:10 *** epicmonkey has joined #aichallenge 2011-11-21T15:43:15 and I had a lot of plans 2011-11-21T15:45:55 im regretful that i dont write my bot in cl anymore, but honestly i feel more comfortable in c++, but nevertheless cl is nice, im just too much of a newbie 2011-11-21T15:48:10 oh, this contest isn't the hardest thing, including the programming problems, its just I am so much distracted because of the environment where I'm currently at 2011-11-21T15:48:46 Hehe, am I the one who decided to try a new language and haven't abandoned it? 2011-11-21T15:49:07 Because I am too lazy to rewrite the whole bot:) 2011-11-21T15:49:19 UncleVasya: well, c++ is as new to me as cl, almost, syntax not counted 2011-11-21T15:49:28 i did mostly c 2011-11-21T15:49:28 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-11-21T15:49:30 before 2011-11-21T15:49:34 UncleVasya: you aren't :P 2011-11-21T15:49:41 I'm not lazy not to do it, I just want to be able to do it 2011-11-21T15:49:47 UncleVasya: you aren't :P 2011-11-21T15:49:56 avdg: i find this contest pretty hard 2011-11-21T15:50:04 i have ideas but they dont manifest easily 2011-11-21T15:50:24 not me 2011-11-21T15:50:33 mcstar: ok, you are still in the club. 2011-11-21T15:50:36 i was here long before many of you guys 2011-11-21T15:50:46 and you still beat me easily 2011-11-21T15:50:47 the problem I have is that my concepts in my brain aren't that correct 2011-11-21T15:51:05 just a quick question, do the ants move without orders? 2011-11-21T15:51:10 :D 2011-11-21T15:51:14 yes, the good ones 2011-11-21T15:51:14 speakeasy: no 2011-11-21T15:51:21 right 2011-11-21T15:51:33 probably if you would see xanthis code you'd go "is that all??" 2011-11-21T15:51:38 im just trying to get to know the code in the tut to start with 2011-11-21T15:52:04 Garf: i think his code is a mess 2011-11-21T15:52:12 the just appeared, and got tothe top 2011-11-21T15:52:24 he must have spent two cofee-nights doing it 2011-11-21T15:52:33 the->he 2011-11-21T15:53:21 hmm, the problem with c++: using google :p 2011-11-21T15:53:38 avdg: what do you mean? 2011-11-21T15:53:47 the more I use it, the more questions I have 2011-11-21T15:53:53 really? 2011-11-21T15:53:58 yeah 2011-11-21T15:53:58 it is sort of simple 2011-11-21T15:54:07 I don't know the library details 2011-11-21T15:54:12 thats why I am struggling 2011-11-21T15:54:16 of course im not digging in too deep object orientation 2011-11-21T15:54:31 well, I *need* to understand things 2011-11-21T15:54:39 I'm forcing myself to do it 2011-11-21T15:54:40 can you share an example? 2011-11-21T15:54:50 avdg, http://www.cplusplus.com 2011-11-21T15:54:58 http://en.cppreference.com 2011-11-21T15:55:04 use this and the new standard 2011-11-21T15:55:08 *** dorisabayon has quit IRC (Ping timeout: 244 seconds) 2011-11-21T15:55:15 and, #geordi, of course 2011-11-21T15:55:15 no, I need better information 2011-11-21T15:55:28 *** bergmark has joined #aichallenge 2011-11-21T15:55:28 avdg: ok, so tell me what is that you need to know? 2011-11-21T15:55:37 i wont be able to help, i just want to know :D 2011-11-21T15:56:03 string > unsigned char convertion problem atm 2011-11-21T15:56:28 with respect your bot? or is that totally unrelated? 2011-11-21T15:56:32 to* 2011-11-21T15:56:37 yeah, its my bot 2011-11-21T15:56:41 *** praveen_ has quit IRC (Ping timeout: 265 seconds) 2011-11-21T15:56:47 so whats the problem there? 2011-11-21T15:56:58 https://gist.github.com/6d430a4050bd1b858b92 2011-11-21T15:57:02 failing to convert 2011-11-21T15:57:19 aaaaa 2011-11-21T15:57:20 and I almost didn't touch the code 2011-11-21T15:57:31 let me paste mine 2011-11-21T15:57:31 *** Fandekasp has joined #aichallenge 2011-11-21T15:57:32 the problem, I don't know what I've done 2011-11-21T15:57:40 before I ran in problems 2011-11-21T15:57:50 istream >> tmpString >> paramValue; 2011-11-21T15:57:56 gameParams[tmpString] = paramValue; 2011-11-21T15:58:11 unordered_map gameParams; 2011-11-21T15:58:14 thats it 2011-11-21T15:58:23 the last line goes to the top of course 2011-11-21T15:59:01 i can sort of understand if you do that kind of if/else in c 2011-11-21T15:59:05 but not in c++ 2011-11-21T15:59:06 *** Palmik_ has quit IRC (Remote host closed the connection) 2011-11-21T15:59:13 but i wouldnt do that way even in c 2011-11-21T15:59:24 meh, dad just entered my room :/ 2011-11-21T15:59:32 td::cerr << "Unknow parameter " << input; 2011-11-21T15:59:38 this is dangerous 2011-11-21T15:59:46 ok, not 2011-11-21T15:59:55 you just write ti stderr 2011-11-21T16:00:19 well, when i first did this, i exited the program when i encoutnered an unknown thing 2011-11-21T16:00:40 thats was bad, cause the engine changed its commands, and i didnt know first why my program stopped working 2011-11-21T16:02:11 *** delt0r_ has joined #aichallenge 2011-11-21T16:02:25 how rude i am 2011-11-21T16:02:30 Hello Dad! 2011-11-21T16:02:37 :p 2011-11-21T16:02:46 dad ruined my programming zone 2011-11-21T16:02:57 !slap dad 2011-11-21T16:03:36 oh well, its hard to live with an iq diff > 10 2011-11-21T16:04:50 *** JorgeB has quit IRC (Ping timeout: 260 seconds) 2011-11-21T16:05:02 that can be the difference between morning and afternoon 2011-11-21T16:05:23 *** amstan has joined #aichallenge 2011-11-21T16:05:32 *** ChanServ sets mode: +o amstan 2011-11-21T16:07:06 hm, my continuity check seems to break transitivity 2011-11-21T16:09:24 ive introduced self-destructing objects into my design 2011-11-21T16:09:33 i hope they will work out well 2011-11-21T16:10:16 :-) 2011-11-21T16:10:29 * avdg takes unwanted break 2011-11-21T16:11:05 yeah be a good boy, take out the garbage.. 2011-11-21T16:12:19 and bring me another cold one on your way back... 2011-11-21T16:13:24 btw, this is a nice resource http://www.parashift.com 2011-11-21T16:13:35 i think most c++ers are aware of this 2011-11-21T16:14:59 *** raemde_ has joined #aichallenge 2011-11-21T16:17:58 Awesome 2011-11-21T16:18:48 *** Antimony has joined #aichallenge 2011-11-21T16:18:51 to order an ant do you simply put its instructions in the order structure? 2011-11-21T16:18:57 *** raemde has quit IRC (Ping timeout: 252 seconds) 2011-11-21T16:19:21 wot arder structure? 2011-11-21T16:19:21 speakeasy: which starter pack? 2011-11-21T16:19:29 python 2011-11-21T16:19:35 sorry, probs should have said 2011-11-21T16:19:40 use the issue_order function 2011-11-21T16:19:47 the order array/dictionary thing 2011-11-21T16:19:48 ah ok 2011-11-21T16:19:51 thanks 2011-11-21T16:19:55 ants.issue_order(ant, direction) 2011-11-21T16:20:07 there is no saved order structure for the python starter 2011-11-21T16:20:53 that function outputs text on stdout for you 2011-11-21T16:22:02 thats weird 2011-11-21T16:22:10 how is ant and direction connected? 2011-11-21T16:22:27 yeah, ok 2011-11-21T16:22:30 they are not 2011-11-21T16:22:31 at all 2011-11-21T16:22:48 *** ltriant has joined #aichallenge 2011-11-21T16:24:05 *** ikaros has joined #aichallenge 2011-11-21T16:24:09 *** Fandekasp has quit IRC (Quit: leaving) 2011-11-21T16:25:27 mcstar: ant would be a tuple (2,3) and direction would be a char 'n' 2011-11-21T16:25:55 McLeopold: no, i read that "there is no saved order structure" after i asked my question 2011-11-21T16:26:02 oops, looks like they should be put together in a tuple 2011-11-21T16:26:10 ah 2011-11-21T16:26:19 They're in a dictionary AFAIK 2011-11-21T16:26:57 orders([ant_loc]) = 'n' 2011-11-21T16:27:19 *** Jak_o_Shadows has joined #aichallenge 2011-11-21T16:27:39 pairofdice: the starter package doesn't store the orders at all, it just outputs them right away 2011-11-21T16:27:49 Uhh, what 2011-11-21T16:28:18 Unless you do the modifications of the tutorial 2011-11-21T16:28:57 http://aichallenge.org/ants_tutorial_step_1.php 2011-11-21T16:29:05 that's still outputting them right away with issue_order 2011-11-21T16:29:11 it's just _also_ storing them in a dict 2011-11-21T16:29:16 Right 2011-11-21T16:31:47 wtf 2011-11-21T16:31:56 those tutorials give away too much 2011-11-21T16:32:11 Just enough 2011-11-21T16:32:33 Should be one for c++ too for me ;) 2011-11-21T16:32:39 *** Regis has joined #aichallenge 2011-11-21T16:32:46 mcstar: they don't really give away very much 2011-11-21T16:33:03 ;_; THIS IS HARD 2011-11-21T16:33:04 no pathfinding, no distance measurement with BFS, etc. 2011-11-21T16:33:10 hi guys, im on the unblocking the hills part of the tut and i get an error when i come to run it that ants.hill_loc() wont work as ants is not defined 2011-11-21T16:33:40 thestinger: lol, it seems they create more trouble than they solve 2011-11-21T16:34:10 yep 2011-11-21T16:34:21 and this is the second language that i tried lol 2011-11-21T16:34:23 Do the other ants functions work? 2011-11-21T16:34:28 yes 2011-11-21T16:34:34 which is weird 2011-11-21T16:34:48 it just doesnt like that line 2011-11-21T16:34:53 ants.my_hills() I thought it was 2011-11-21T16:35:02 speakeasy: you know indentation is how blocks of code are laid out in python, right? 2011-11-21T16:35:08 for hill_loc in ants.my_hills() 2011-11-21T16:35:19 sure 2011-11-21T16:35:21 yes that line 2011-11-21T16:35:57 @thestinger i sortof understand that but im not to great at python yet 2011-11-21T16:35:57 *** Redgis has quit IRC (Ping timeout: 244 seconds) 2011-11-21T16:35:57 speakeasy: I'm sorry Dave, err speakeasy; I cannot 'thestinger'. 2011-11-21T16:36:15 https://github.com/aichallenge/aichallenge/blob/epsilon/ants/dist/starter_bots/python/MyBot.py#L22 2011-11-21T16:36:20 ants is definitely defined 2011-11-21T16:36:33 Align the indents so it goes with the rest of the function 2011-11-21T16:36:41 let me check 2011-11-21T16:36:47 and make sure you're using consistent indents 2011-11-21T16:36:58 set your editor to put 4 spaces when you use tab 2011-11-21T16:36:59 Quick question, if I deactivate my bot will the timer reset? 2011-11-21T16:37:03 i.e. either tabs or spaces 2011-11-21T16:37:10 ThatSnail: No. 2011-11-21T16:37:16 which function? 2011-11-21T16:37:19 Cool, thanks 2011-11-21T16:37:42 *** raemde_ is now known as raemde 2011-11-21T16:39:27 *** hkraal_ has joined #aichallenge 2011-11-21T16:39:36 Also er how much can a bot take before it times out in a round? 2011-11-21T16:40:01 you're given the turn time on turn 0 2011-11-21T16:40:31 So an A* that processes each block for each piece of food individually per frame would be bad I'm guessing D: 2011-11-21T16:40:35 gottit, thanks guys 2011-11-21T16:41:41 *** retybok_ has joined #aichallenge 2011-11-21T16:42:08 do variables rollover to the next frame? 2011-11-21T16:42:14 speakeasy: Yes. 2011-11-21T16:42:18 awesome 2011-11-21T16:42:20 thanks 2011-11-21T16:42:34 speakeasy: Well, depending on the language and scope of course. But the program stays running between turns. 2011-11-21T16:43:40 so if i have a dictionary of all the known map then that will stay on for example between turns 2011-11-21T16:43:52 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-11-21T16:43:57 speakeasy: If it's in the correct scope then yes. 2011-11-21T16:44:13 cool 2011-11-21T16:44:16 thanks 2011-11-21T16:44:21 No problem. 2011-11-21T16:44:36 this competition has got me really inspired to go and learn to code properly 2011-11-21T16:44:42 i think i better advice is check it for yourself 2011-11-21T16:44:42 *** NightExcessive has quit IRC (Quit: brb) 2011-11-21T16:44:47 a* 2011-11-21T16:45:10 *** NightExcessive has joined #aichallenge 2011-11-21T16:48:10 *** amstan has joined #aichallenge 2011-11-21T16:48:10 *** ChanServ sets mode: +o amstan 2011-11-21T16:50:25 *** UncleVasya has quit IRC (Ping timeout: 244 seconds) 2011-11-21T16:51:27 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-21T16:52:13 http://pastebin.com/KZpZ3rFP 2011-11-21T16:52:33 stl: making error messages un-understandable 2011-11-21T16:52:53 just because i put a ) to where it wasnt supposed to be 2011-11-21T16:53:25 *** jhannah_ is now known as deafferret 2011-11-21T16:53:28 *** speakeasy has quit IRC (Ping timeout: 240 seconds) 2011-11-21T16:58:38 *** Anilm3 has joined #aichallenge 2011-11-21T17:03:02 Hi guys 2011-11-21T17:04:59 mcstar, happy template errors =) 2011-11-21T17:05:05 always good to see.. 2011-11-21T17:05:20 yeah... 2011-11-21T17:05:38 actually, the error messsage is helpful, once you get past the first 1 million lines 2011-11-21T17:05:50 *** twymer has quit IRC (Ping timeout: 252 seconds) 2011-11-21T17:06:41 =) 2011-11-21T17:07:25 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-11-21T17:09:37 *** hkraal_ has quit IRC (Quit: Page closed) 2011-11-21T17:10:43 *** goffrie has quit IRC (Read error: Connection reset by peer) 2011-11-21T17:14:43 *** goffrie has joined #aichallenge 2011-11-21T17:15:12 wow--template errors are *still* like that? 2011-11-21T17:16:18 *** foRei has quit IRC (Read error: Connection reset by peer) 2011-11-21T17:19:28 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-11-21T17:19:41 delt0r_: the error was a misplaced ) 2011-11-21T17:20:10 it was just the type of the thing that could not be converted to bool 2011-11-21T17:21:12 yea but templates where famous for 64k or more of error message 2011-11-21T17:21:16 *** Regis is now known as Redgis 2011-11-21T17:21:21 *** Antimony has joined #aichallenge 2011-11-21T17:21:42 but that was when compilers were still coming to grips with them 2011-11-21T17:22:05 *** iglo has quit IRC (Remote host closed the connection) 2011-11-21T17:22:15 delt0r_: how is your bot coming along? 2011-11-21T17:22:37 I am working on a game right now... 2011-11-21T17:22:44 I know what i will do however... 2011-11-21T17:23:00 all at once BFS for local stuff 2011-11-21T17:23:10 a clustering for combat 2011-11-21T17:23:28 and a hacked bit of "keep something close to the nest" 2011-11-21T17:23:32 for defense 2011-11-21T17:23:43 is that a winning combination? 2011-11-21T17:23:49 the potental map stuff was not better than just bfs 2011-11-21T17:23:56 doubt it... 2011-11-21T17:24:03 you tried potential maps? 2011-11-21T17:24:03 *** jacob_strauss has quit IRC (Quit: Page closed) 2011-11-21T17:24:13 trying to have something that is "good" without too much complications 2011-11-21T17:24:15 yea 2011-11-21T17:24:30 and cooperative influence whatevers 2011-11-21T17:24:35 and plain influence 2011-11-21T17:24:40 all kinda the same really 2011-11-21T17:24:48 yeah 2011-11-21T17:25:01 im at 2400 loc now 2011-11-21T17:25:04 work pretty well for local strat 2011-11-21T17:25:08 i have to search now 2011-11-21T17:25:11 but not global 2011-11-21T17:25:29 bleh, just fixed my string2char problem, put an int convertion inbetween :-) 2011-11-21T17:25:39 *** BenJackson has joined #aichallenge 2011-11-21T17:25:53 you dont like my solution? 2011-11-21T17:25:57 how much longer do we have? 2011-11-21T17:26:02 dec 18 2011-11-21T17:26:05 18th 2011-11-21T17:26:09 mcstar: what that at me? 2011-11-21T17:26:11 was 2011-11-21T17:26:20 avdg: 2011-11-21T17:26:25 right 2011-11-21T17:26:32 mcstar: I know it once worked 2011-11-21T17:26:41 why isn't tcpants.com more popular? 2011-11-21T17:26:52 BenJackson: critical mass 2011-11-21T17:27:02 more are at Fluxid 2011-11-21T17:27:09 BenJackson: because @tcp is not tcpants.com 2011-11-21T17:27:11 so more come/stay 2011-11-21T17:27:12 but I run bots on 3 tcp servers just as a community service... it's no harder than running just one 2011-11-21T17:27:20 mcstar: https://gist.github.com/6d430a4050bd1b858b92 2011-11-21T17:27:32 (most state variables are chars) 2011-11-21T17:28:31 *** speakeasy has joined #aichallenge 2011-11-21T17:29:22 hi guys, what does : in square brackets mean in python e.g. for loc in self.unseen[:]: blah blah 2011-11-21T17:29:44 if state.rows is an int, >> state.rows will correctly read an int 2011-11-21T17:30:09 speakeasy: element access 2011-11-21T17:30:13 rows => unsigned char since its at least 256 big 2011-11-21T17:30:17 ah fuck 2011-11-21T17:30:20 all elements 2011-11-21T17:30:30 basically a copy 2011-11-21T17:30:34 but I had to use a convertion between 2011-11-21T17:31:09 avdg: what is state? 2011-11-21T17:31:14 is this the c++ starter? 2011-11-21T17:31:16 state object 2011-11-21T17:31:17 no 2011-11-21T17:31:27 so rows is an int? 2011-11-21T17:31:36 no tmp is an int 2011-11-21T17:31:49 what type does state.rows have? 2011-11-21T17:31:55 uchar 2011-11-21T17:32:00 why on earth? 2011-11-21T17:32:13 because it fits 2011-11-21T17:32:16 you created a problem than you say you solved it? 2011-11-21T17:32:25 you created 2 problems 2011-11-21T17:32:28 well, it worked before 2011-11-21T17:32:48 hmm 2011-11-21T17:33:05 avdg: if rows is an uchar why would you think you should get a number?? 2011-11-21T17:33:16 like int? 2011-11-21T17:33:19 maan 2011-11-21T17:33:23 dunno 2011-11-21T17:33:31 do you know what >> or << does? 2011-11-21T17:33:39 yeah kinda 2011-11-21T17:33:43 these operators are overloaded by TYPE 2011-11-21T17:33:47 *** Surya has quit IRC (Quit: Page closed) 2011-11-21T17:34:01 they do different things depending what do you feed them 2011-11-21T17:34:11 char c; 2011-11-21T17:34:16 cin >> c; 2011-11-21T17:34:21 takes 1 char 2011-11-21T17:34:29 will read a charater from the stream, and place its ascii value in c 2011-11-21T17:34:34 I forgot my header declaration for one moment 2011-11-21T17:34:51 but int c; 2011-11-21T17:34:56 cin >> c; 2011-11-21T17:35:06 will expect an integer 2011-11-21T17:35:11 yeah 2011-11-21T17:35:23 i.e. a string that is recognizable as an integer number 2011-11-21T17:35:45 so why would you give uchar type to your state variables? 2011-11-21T17:36:24 because I was thinking it was acting like an int 2011-11-21T17:36:34 but yeah 2011-11-21T17:36:35 and that if/else forest can be thrown out the window too 2011-11-21T17:36:54 you simply dont need it 2011-11-21T17:37:47 avdg: nothing stops you from doing that with scanf or whatver, atoi or something 2011-11-21T17:37:54 but there is no reason to do that 2011-11-21T17:37:58 *** bergmark has quit IRC (Ping timeout: 258 seconds) 2011-11-21T17:37:59 mcstar: what's your name on the ranking? 2011-11-21T17:38:09 does it matter? 2011-11-21T17:38:13 agentsmith 2011-11-21T17:38:23 matters to me 2011-11-21T17:38:26 but pls dont break my bot with looking at it 2011-11-21T17:38:39 don't worry 2011-11-21T17:40:05 avdg: another advise: use stringstreams 2011-11-21T17:40:28 read in a whole line, and read from that line as a stream with the stream operators 2011-11-21T17:40:38 you have better control over what and how you read 2011-11-21T17:40:45 ty 2011-11-21T17:41:15 meh, I hate myself even more atm 2011-11-21T17:41:20 why? 2011-11-21T17:41:41 anyway, dont hate yourself 2011-11-21T17:41:51 parents doing stupid things -> I can't blame them, me doing stupid things -> oh well 2011-11-21T17:42:20 the conflicts are sometimes too big 2011-11-21T17:43:05 oh, i blame them too 2011-11-21T17:43:34 you're all traumatized 2011-11-21T17:43:58 yeah, and the "system" is making it worse for me atm ;-) 2011-11-21T17:44:37 I don't really follow 2011-11-21T17:44:40 * BenJackson misreads a replay for a second thinking "how did that ant score 3" and finally realizing that it had 3 ants left 2011-11-21T17:44:54 meh lets skip that subject 2011-11-21T17:45:10 he meant his operating system 2011-11-21T17:45:28 meh, not only os :p 2011-11-21T17:45:52 avdg: hang on whatever it is, there are ups and dows... 2011-11-21T17:46:10 my almost always go downs 2011-11-21T17:46:22 Yeah dude things eventually get better 2011-11-21T17:46:23 but its mostly not me causing it 2011-11-21T17:46:26 but your next live will be wonderful 2011-11-21T17:46:29 life* 2011-11-21T17:47:16 Don't hang to that if you want to get better... 2011-11-21T17:47:36 yeah, most people are sleeping now 2011-11-21T17:47:45 its about to get better atm 2011-11-21T17:48:26 anyone knows a bot name which uses heavy swarm behaviour/flocking? (to watch) 2011-11-21T17:48:51 ikaros: the reverse engineer mentality 2011-11-21T17:49:06 dont you want the binary too? 2011-11-21T17:49:09 does flocking really work well here? 2011-11-21T17:49:16 nah im just a little bored by seeing all the same strategies 2011-11-21T17:49:34 what do you consider flocking? 2011-11-21T17:49:44 cause there are some kind of flocking but thats useless 2011-11-21T17:50:01 The usual flocking based on boids I suppose 2011-11-21T17:50:06 for example: player notices a hill, drives all ants there 2011-11-21T17:50:09 its flocking but... 2011-11-21T17:50:54 boids are for better resolution space/time 2011-11-21T17:51:04 at least thats what i think 2011-11-21T17:51:14 what do you mean by resolution? 2011-11-21T17:51:33 well, here, much can happen in 1 turn or 1 tile next 2011-11-21T17:51:40 my ants flock only because they are mostly all responding to the same stimuli 2011-11-21T17:51:45 not because they're following each other 2011-11-21T17:51:45 and for boids you need some sort of cntinuity 2011-11-21T17:52:05 BenJackson: yeah, thats what i tried to illustrate 2011-11-21T17:52:10 that isnt really flocking 2011-11-21T17:52:22 I agree 2011-11-21T17:52:28 sorry had to go afk mcstar yea the classic boids thing 2011-11-21T17:52:30 BenJackson: so your flocking is an emergent behaviour, that's cool 2011-11-21T17:52:32 its just: same goal+exclusion principle 2011-11-21T17:52:46 flocking is emergent behavior by a bunch of agents with limited information (ie just about the birds around them) 2011-11-21T17:52:52 we have global knowledge so it's not likely to be useful 2011-11-21T17:53:06 I'm trying to achieve that but no luck 2011-11-21T17:53:19 Anilm3: you have to micor-manage 2011-11-21T17:53:24 boids go the opposite way 2011-11-21T17:53:34 i once saw a bot (it was still basic) by a guy from within this channel and it had very nice flocking behaviour 2011-11-21T17:53:54 (very nice flocking)++ 2011-11-21T17:54:04 BenJackson, thats just an assumption.. and was my first impression to 2011-11-21T17:54:06 too 2011-11-21T17:54:32 i just cant be sure without trying something like that.. especially because its so much fun :) 2011-11-21T17:54:50 so competition is clearly picking up 2011-11-21T17:54:57 my take on this: flocking is really an emergent phenomennon 2011-11-21T17:55:03 as people wrapped up the weekend and made new submissions I slipped about 10 spots 2011-11-21T17:55:05 you can always do better here 2011-11-21T17:55:13 *** choas has quit IRC (Ping timeout: 258 seconds) 2011-11-21T17:55:13 my bot is over a week old now 2011-11-21T17:55:14 yo dont have to wait for something to show up 2011-11-21T17:55:20 you program in the bhavior 2011-11-21T17:55:26 mcstar: I agree 2011-11-21T17:55:39 the alternative is also an interesting challenge but it's not *this* challenge 2011-11-21T17:55:57 its kiind of like with evolutionary approaches 2011-11-21T17:56:19 many years ago I wrote a program to evolve programs to make ants seek food in a grid world 2011-11-21T17:56:22 very similar to this 2011-11-21T17:56:30 except the ants had almost no sensory input 2011-11-21T17:56:33 BenJackson: i saw a java simulation of that 2011-11-21T17:56:42 (or soomething similar) 2011-11-21T17:56:44 just "is food ahead of me" and if it moved "did I hit a wall" 2011-11-21T17:56:48 well that's still quite different 2011-11-21T17:56:59 and it evolved programs in a stack language 2011-11-21T17:57:15 of course in this game you know the exact food and ant location and where the obstacles are and go straight there 2011-11-21T17:57:47 yes but you have to decide how many ants to send at targets and such 2011-11-21T17:58:05 and how do you know if you make good decisions 2011-11-21T17:58:18 well I'm not trying to evolve anything in THIS challenge 2011-11-21T17:58:28 so I'd know by watching replays, figuring out why things happened, tweaking the code 2011-11-21T17:58:30 i understood 2011-11-21T17:58:57 in the genetic programming case the ants had a cost for moving which they replenished by eating 2011-11-21T17:59:03 and by surviving they produced offspring 2011-11-21T17:59:12 as i said different problem 2011-11-21T17:59:26 I think there is an interesting meta-ai problem for this contest 2011-11-21T17:59:31 to write a program to analyze replays 2011-11-21T17:59:34 yea 2011-11-21T17:59:41 and determine what behaviors given ants have 2011-11-21T17:59:55 do they appear to have food seeking? combat? retained knowledge of enemy hills? 2011-11-21T17:59:56 BenJackson: isnt your bot supposed to do that in-game? 2011-11-21T17:59:57 would i have known of this when the contest was beta i would have tried something like that 2011-11-21T18:00:13 known of the contest 2011-11-21T18:00:31 mcstar: at the moment I don't actually try to learn anything about my enemies, although I have ideas along that line 2011-11-21T18:00:33 *** retybok_ has quit IRC (Ping timeout: 260 seconds) 2011-11-21T18:00:50 I have tons of ideas I'm not even using now 2011-11-21T18:00:57 for a week I was noodling on alternate ways to do global strategy 2011-11-21T18:01:03 what i dont like about that contest is that alot of maps just suck 2011-11-21T18:01:08 yesterday and today I fixed most of my combat bugs 2011-11-21T18:01:15 i had the notion of correlating enemies' decisions with the evaluation function i use to resolve a battle 2011-11-21T18:01:23 so later i can adopt to them 2011-11-21T18:01:32 yeah 2011-11-21T18:01:34 *** snthaoeu has joined #aichallenge 2011-11-21T18:01:36 alot of games are lots of random because of hills being so tight 2011-11-21T18:01:54 ikaros: maybe. my bot is better than average at multi-hill mazes 2011-11-21T18:02:03 its not multi hill 2011-11-21T18:02:05 what i mean 2011-11-21T18:02:11 thats ok since you have multi hills :) 2011-11-21T18:02:19 if you lose one.. ok 2011-11-21T18:02:40 oh, the single-hill mazes tend to have defensible hills 2011-11-21T18:02:49 but those random walks where every player has 5-8 ants and the first hills are razed.. its crazy 2011-11-21T18:03:06 ikaros: try making better maps 2011-11-21T18:03:28 am.stan is working on a new map generator i think 2011-11-21T18:03:31 you can decide to leave ants back early to defend hills but that stalls exploration heavily 2011-11-21T18:03:57 well i think for the bad ones it would be sufficent to be larger or have fewer players 2011-11-21T18:04:14 but whatever.. its just nitpicking 2011-11-21T18:04:15 my bot is usually the best food-gatherer in the games, but since it leaves its hill totally unguarded.... 2011-11-21T18:04:24 know that problem :) 2011-11-21T18:04:26 I think the real problem is that despite 'symmetry' any map with >2 players can be highly asymmetric based on player placement 2011-11-21T18:04:26 *** fefe has joined #aichallenge 2011-11-21T18:04:38 if I get a weak neighbor and you get a strong neighbor I will have a big advantage 2011-11-21T18:04:41 and i started to leave guards at hills since turn 40 2011-11-21T18:04:55 ikaros: which bot is you on aichallenge? 2011-11-21T18:04:56 that too BenJackson 2011-11-21T18:05:02 same name 2011-11-21T18:05:05 BenJackson: that should be averaged out 2011-11-21T18:05:08 just updated yesterday.. 2011-11-21T18:05:25 but its not so impressive :) 2011-11-21T18:05:39 yaaaaaay 2011-11-21T18:05:47 i have no real combat code so far.. only some masks which works kinda good for razing 2011-11-21T18:05:55 but for nothing else :) 2011-11-21T18:05:58 my old submsision peaked on rank due to sigma dropping and is now sliding down due to stronger opponents submitted recently 2011-11-21T18:06:11 peaked where? 2011-11-21T18:06:11 post combat analysis FOR THE WIN 2011-11-21T18:06:52 eh? 2011-11-21T18:07:12 what rank did you reach 2011-11-21T18:07:35 lol 2011-11-21T18:07:41 formidable painted a smiley 2011-11-21T18:07:44 ikaros: 17 2011-11-21T18:07:56 nice 2011-11-21T18:08:09 mcstar: I still haven't gotten around to drawing stuff with surplus ants 2011-11-21T18:08:24 my original defense code could have done it easily 2011-11-21T18:08:59 ikaros: I have a new version that krushes the old version but as long as I know about major bugs I keep not bothering to submit 2011-11-21T18:09:06 someone asked if a human could match against a bot 2011-11-21T18:09:09 well, noway sir 2011-11-21T18:09:46 yea it doesnt help anyways.. testing is better with tcp since the servers are so slow 2011-11-21T18:10:05 the real problem with aichallenge is that I don't even watch the replays for the most part 2011-11-21T18:10:09 if the human assists an ai, then perhaps 2011-11-21T18:10:11 that code is so old I don't care why it lost or won 2011-11-21T18:10:33 ai sends ants exploring and handles micro, human makes strategic decisions 2011-11-21T18:10:45 attack here, pull back from here, explore here etc 2011-11-21T18:10:49 look at top starcraft players.. would bet on them :) 2011-11-21T18:11:05 Some of the starcraft bots are pretty good 2011-11-21T18:11:07 they can't micro 200 ants in 500 ms successfully 2011-11-21T18:11:10 well 2011-11-21T18:11:26 they dont have to command them each one on its own 2011-11-21T18:11:30 *** Antimony_ has joined #aichallenge 2011-11-21T18:11:33 and all units are equal 2011-11-21T18:11:57 *** bhasker_ has joined #aichallenge 2011-11-21T18:12:08 that's what I'm saying, the human needs ai assistance for individual unit micro 2011-11-21T18:12:09 *** bhasker has quit IRC (Read error: Connection reset by peer) 2011-11-21T18:12:09 *** bhasker_ is now known as bhasker 2011-11-21T18:12:29 I have a pretty good idea of what I want my ants to do next 2011-11-21T18:12:30 dont think so.. just grouping 2011-11-21T18:12:44 *** Antvolution has left #aichallenge 2011-11-21T18:12:54 and what about pathfinding? the human can't do that quickly enough 2011-11-21T18:13:11 pathfinding is implicit in any strategy game 2011-11-21T18:13:36 right, and so is basic unit micro and many of the things we implement here 2011-11-21T18:14:01 well.. so the player vs bot contest is not well defined :) 2011-11-21T18:14:11 so I'm saying a human couldn't replace an ai in this game without a layer in between 2011-11-21T18:14:23 *** Antimony has quit IRC (Ping timeout: 248 seconds) 2011-11-21T18:14:25 *** Antimony_ is now known as Antimony 2011-11-21T18:14:29 thats pretty obvious i think so agreed =) 2011-11-21T18:14:29 that implements lots of the stuff we have in our ais 2011-11-21T18:14:56 hard to draw the line between what's a human's responsibility and what's the ai's 2011-11-21T18:14:58 that would be true for almost any RTS 2011-11-21T18:15:08 and 500ms is kinda RTS for a human lol 2011-11-21T18:15:59 hi guys, my bot isnt doing the same as the tutorial bot in step five, why should this be? 2011-11-21T18:16:41 speakeasy: probably a sorting difference 2011-11-21T18:16:48 ah 2011-11-21T18:16:54 how can i improve this? 2011-11-21T18:17:02 if you raze the hill, don't worry about it 2011-11-21T18:17:06 to make it the same as the tut 2011-11-21T18:17:09 i dont 2011-11-21T18:17:26 i wipe the enemy ants though but i dont know if my ants are razing hils 2011-11-21T18:17:32 try passing --viewradius 55 in the tutorial command 2011-11-21T18:17:40 *** bergmark has joined #aichallenge 2011-11-21T18:17:53 will try 2011-11-21T18:19:19 awesome, thanks so much 2011-11-21T18:20:01 McLeopold: can we get that fixed? 2011-11-21T18:20:27 *** mj41 has quit IRC (Ping timeout: 245 seconds) 2011-11-21T18:21:03 amstan: I'd have to redo all the tutorial replays, I think it is low priority 2011-11-21T18:21:26 ic, ok 2011-11-21T18:21:46 McLeopold: have you seen the new and improved worker_ssh? 2011-11-21T18:21:51 btw.. how often is this contest run? 2011-11-21T18:21:54 amstan: nope 2011-11-21T18:22:07 is there a specific interval 2011-11-21T18:22:09 McLeopold: try it, sudo su contest; worker_ssh 2011-11-21T18:22:33 ikaros: depends on a lot of things, but mostly how fast we dev 2011-11-21T18:22:42 ok 2011-11-21T18:22:44 ikaros: we try to minimize that interval 2011-11-21T18:22:55 thank you for doing this btw 2011-11-21T18:22:58 really like it 2011-11-21T18:23:01 down to the term though(4 months) 2011-11-21T18:23:10 i regret i had no time for planet wars :) 2011-11-21T18:23:12 http://ants.fluxid.pl/replay.9656 and he bites the dust 2011-11-21T18:23:32 amstan: sequencially -> sequentially :) 2011-11-21T18:23:39 McLeopold: warning though, -r is slow, that's why it's cached in /tmp/aliveworkers 2011-11-21T18:23:40 4 months?! 2011-11-21T18:23:43 >.> 2011-11-21T18:23:54 Zannick: rps was 4 months before tron 2011-11-21T18:23:57 soo time to sleep 2011-11-21T18:24:07 gnight 2011-11-21T18:24:17 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-21T18:24:32 rps was uwaterloo only 2011-11-21T18:24:34 :P 2011-11-21T18:24:39 iirc 2011-11-21T18:24:47 Zannick: same ammount of work required 2011-11-21T18:24:58 and tron was 6 months before planet wars, right? 2011-11-21T18:25:20 feb->sept 2011-11-21T18:25:32 7, then. 2011-11-21T18:25:34 i was close 2011-11-21T18:25:52 *** speakeasy has quit IRC (Ping timeout: 240 seconds) 2011-11-21T18:26:09 aichallenge: Alexandru Stan epsilon * r27806f0 / manager/worker_ssh.py : fixed typo - http://git.io/UJKAqA 2011-11-21T18:26:25 amstan: I'm getting closer on finishing my map generator http://paste.aichallenge.org/01466/ 2011-11-21T18:26:48 that looks cool 2011-11-21T18:27:03 what was wrong with my code? 2011-11-21T18:27:08 *** Redgis has quit IRC (Ping timeout: 252 seconds) 2011-11-21T18:27:15 holy fuck that map is terrifying 2011-11-21T18:27:17 i had coordonates as objects 2011-11-21T18:27:18 I haven't really looked at your code 2011-11-21T18:27:48 you can do stuff like map[Point(10,10)+Point(-1,-20)]=WATER 2011-11-21T18:28:18 it handles wrapping, adding values, symmetries 2011-11-21T18:28:26 does anyone here know how to fix the button sprite sheets? 2011-11-21T18:28:32 amstan: that's nice 2011-11-21T18:28:35 it would be nice if the scaled replay actually had images for the buttons 2011-11-21T18:29:04 McLeopold: that's a ton of hills! lol 2011-11-21T18:29:18 yeah :), but they are just markers for now 2011-11-21T18:29:19 and a bit more 2011-11-21T18:29:20 plus you can fight through all those walls 2011-11-21T18:29:26 oh, ic 2011-11-21T18:29:31 I'm working on selecting a good set for hills 2011-11-21T18:29:38 oh yeah, you can fight through the walls 2011-11-21T18:29:44 *water 2011-11-21T18:29:52 the cells are ok 2011-11-21T18:30:26 is there going to be a new wave of maps before the contest closes or are these new maps going to be used only for the final thign? 2011-11-21T18:30:52 *** fefe has quit IRC (Ping timeout: 265 seconds) 2011-11-21T18:30:52 *** amstan_ has joined #aichallenge 2011-11-21T18:30:52 *** ChanServ sets mode: +o amstan_ 2011-11-21T18:31:12 something's fishy with my internets 2011-11-21T18:31:19 contestbot: echo hi 2011-11-21T18:31:19 hi 2011-11-21T18:31:33 contestbot doesn't seem to think so 2011-11-21T18:31:37 *** Yexo has quit IRC (Quit: Leaving) 2011-11-21T18:31:55 *** amstan has quit IRC (Remote host closed the connection) 2011-11-21T18:32:09 *** Yexo has joined #aichallenge 2011-11-21T18:32:54 amstan_: the code can scale http://paste.aichallenge.org/KtWMl/ 2011-11-21T18:33:12 McLeopold: nice 2011-11-21T18:33:32 that is nice 2011-11-21T18:33:43 McLeopold: i'm certain that these maps/mapgens can be used beyond ants 2011-11-21T18:33:52 is that a 5-player? 2011-11-21T18:33:58 yes, 5 player 2011-11-21T18:34:40 *** xathis has quit IRC (Ping timeout: 240 seconds) 2011-11-21T18:35:08 McLeopold: that map seems inbalanced, at least with the current hill locations 2011-11-21T18:35:49 *** amstan_ is now known as amstan 2011-11-21T18:36:08 Yexo: hills are just markers for now 2011-11-21T18:36:23 ok. The map looks very nice :) 2011-11-21T18:37:18 McLeopold: i would put the hills where purple is 2011-11-21T18:38:52 Möebius maze 2011-11-21T18:42:36 McLeopold: neat! 2011-11-21T18:43:44 *** Antimony_ has joined #aichallenge 2011-11-21T18:46:29 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-11-21T18:46:39 *** Antimony_ is now known as Antimony 2011-11-21T18:49:50 *** speakeasy has joined #aichallenge 2011-11-21T18:50:40 McLeopold: that map is disjoint 2011-11-21T18:51:10 imagine 1 player is victorious in 1 region 2011-11-21T18:51:14 and another in the other one 2011-11-21T18:51:19 what now? 2011-11-21T18:51:50 *** fpcfan has quit IRC (Ping timeout: 244 seconds) 2011-11-21T18:52:00 It's perfectly joined 2011-11-21T18:52:11 Only 1 area 2011-11-21T18:52:24 no 2011-11-21T18:52:30 at least i dont see it that way 2011-11-21T18:52:36 mcstar: yep, all bases are connected 2011-11-21T18:52:46 no, it's one area. start at a red base and follow the path down 2011-11-21T18:52:53 *** seletskiy has quit IRC (Read error: Operation timed out) 2011-11-21T18:53:01 when you drop out the bottom, you appear in the "other" area 2011-11-21T18:53:03 mcstar: printscreen 2011-11-21T18:53:10 take it in paint, then floodfill it 2011-11-21T18:53:47 *** fpcfan has joined #aichallenge 2011-11-21T18:53:57 i gimped it 2011-11-21T18:54:43 interesting map 2011-11-21T18:54:45 nice 2011-11-21T18:54:52 i didnt see it 2011-11-21T18:55:28 It's in 3 layers so a bit trippy 2011-11-21T18:56:00 no, i wasnt following the edges 2011-11-21T18:56:00 several wraps 2011-11-21T18:56:32 i just fiddled with it, and thought, maan, this is disjoint 2011-11-21T18:56:41 i probably need to be more careful 2011-11-21T18:57:51 *** JamesMG_ has quit IRC (Read error: Connection reset by peer) 2011-11-21T18:57:57 *** JamesMG has joined #aichallenge 2011-11-21T18:58:40 *** speakeasy has quit IRC (Ping timeout: 240 seconds) 2011-11-21T19:02:51 *** speakeasy has joined #aichallenge 2011-11-21T19:04:59 *** fpcfan has quit IRC (Ping timeout: 258 seconds) 2011-11-21T19:05:21 i need a * that binds stronger than -> 2011-11-21T19:05:29 hi guys, is it possible to make my own maps to tes some things oiut 2011-11-21T19:05:36 yes 2011-11-21T19:05:41 woot 89% on the ai-class midterm 2011-11-21T19:05:43 open one up,and edit it 2011-11-21T19:06:08 amstan: does that count? 2011-11-21T19:06:13 mcstar: you don't like parens? 2011-11-21T19:06:14 do you get some certificate? 2011-11-21T19:06:37 Minthos: im tired of writing (*it)->field 2011-11-21T19:06:52 mcstar: i get a signed(ala crypto) certificate 2011-11-21T19:07:20 so its a + for job application? 2011-11-21T19:07:35 sure, everything is technically a + for job application 2011-11-21T19:07:42 mcstar: what you need is a better language then 2011-11-21T19:07:52 but yeah, i hope so 2011-11-21T19:07:59 gah, 84% ai-class midterm :< 2011-11-21T19:08:05 how can i open them? 2011-11-21T19:08:14 speakeasy: they are simple text files 2011-11-21T19:08:20 like everything on a computer 2011-11-21T19:08:27 silly me lol 2011-11-21T19:08:47 botched questions 1, 2 and 15 2011-11-21T19:08:48 Minthos: for example? 2011-11-21T19:09:12 mcstar: C#, java, D 2011-11-21T19:09:21 f#? 2011-11-21T19:09:21 obj-c 2011-11-21T19:09:29 *** fpcfan has joined #aichallenge 2011-11-21T19:09:30 maybe 2011-11-21T19:09:31 pairofdice: what was 15? 2011-11-21T19:09:49 i think c++11 lets me have new operators 2011-11-21T19:09:56 read something similar 2011-11-21T19:10:33 *** HaraKiri has quit IRC () 2011-11-21T19:10:49 what are the different symbols in map making? 2011-11-21T19:11:12 Laplace smoothing in Markov chain 2011-11-21T19:11:43 How did I get that wrong 2011-11-21T19:11:45 speakeasy: http://aichallenge.org/specification.php#Map-Format 2011-11-21T19:11:46 ;_; 2011-11-21T19:12:01 cool, thanks 2011-11-21T19:12:16 are those tears? 2011-11-21T19:12:22 Yeah 2011-11-21T19:12:27 cool 2011-11-21T19:12:39 i like sideways better :'( 2011-11-21T19:12:39 i nailed it for the first time 2011-11-21T19:12:54 thats just mucus in your nose 2011-11-21T19:13:14 once i thought <3 is actually an ass 2011-11-21T19:13:23 or ice-cream 2011-11-21T19:13:33 mcstar: apparently that's where the heard symbol was derived from 2011-11-21T19:13:42 heard? 2011-11-21T19:13:47 heart* 2011-11-21T19:13:51 oh, no 2011-11-21T19:13:54 tbbt? 2011-11-21T19:14:03 the big bang theory? no 2011-11-21T19:14:15 i remember there was some similar explanation there 2011-11-21T19:14:28 oh.. maybe, idk actually 2011-11-21T19:14:33 i could have remembered it from there 2011-11-21T19:14:55 penny said: so i actually painted little asses on my face in school? 2011-11-21T19:14:58 (or similar) 2011-11-21T19:16:32 I guess I should've checked my answers 2011-11-21T19:16:46 All user-defined literals are suffixes; defining prefix literals is not possible. 2011-11-21T19:16:54 tooo bad 2011-11-21T19:20:03 mcstar: is this from last week or something? i don't remember it 2011-11-21T19:20:19 no 2011-11-21T19:20:33 ive missed a couple of weeks 2011-11-21T19:20:46 but im up-to-date with dexter 2011-11-21T19:21:12 lol, i thought they gonna bust him, in last week's episode 2011-11-21T19:22:12 *** peyton has quit IRC (Quit: peyton) 2011-11-21T19:22:14 *** Antimony_ has joined #aichallenge 2011-11-21T19:23:11 *** grwip has quit IRC (Quit: Leaving) 2011-11-21T19:24:29 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-11-21T19:24:35 why is there no big bag theory transcript git anywhere? 2011-11-21T19:25:18 *** Antimony has quit IRC (Ping timeout: 258 seconds) 2011-11-21T19:25:24 *** Antimony_ is now known as Antimony 2011-11-21T19:25:42 *** peyton has joined #aichallenge 2011-11-21T19:25:45 *** ThatSnail_ has joined #aichallenge 2011-11-21T19:25:52 *** speakeasy has quit IRC (Ping timeout: 240 seconds) 2011-11-21T19:26:20 Amy Farrah Fowler: Did you know the iconic heart shape isn't based on an actual human heart, its based on what a womans ass looks like bending over. 2011-11-21T19:26:22 Penny: So in 8th grade, I was dotting my i's with little asses? That's cool. 2011-11-21T19:26:53 'The Alien Parasite Hypothesis' - Season 4, Episode 10 2011-11-21T19:27:04 *** ThatSnail has quit IRC (Ping timeout: 240 seconds) 2011-11-21T19:27:12 *** ThatSnail_ is now known as ThatSnail 2011-11-21T19:27:24 *** replore has joined #aichallenge 2011-11-21T19:33:08 *** NightExcessive has quit IRC (Quit: Leaving) 2011-11-21T19:33:38 Initialization follows the member declaration order, not the initializer order! 2011-11-21T19:33:43 nice pitfall 2011-11-21T19:36:48 *** netantho has quit IRC (Ping timeout: 258 seconds) 2011-11-21T19:37:03 mcstar: -Wreorder lets you know about that 2011-11-21T19:37:04 *** fpcfan has quit IRC (Ping timeout: 240 seconds) 2011-11-21T19:37:15 I think it's enabled if you're using '-Wall -Wextra' 2011-11-21T19:37:30 *** netantho has joined #aichallenge 2011-11-21T19:37:31 never turned on a W flag in my life 2011-11-21T19:37:44 (im not bragging) 2011-11-21T19:37:48 you should really have -Wall on 2011-11-21T19:37:53 k 2011-11-21T19:38:03 I always use -Wall 2011-11-21T19:38:05 and -Wextra too (shut off individual warnings if they annoy you) 2011-11-21T19:38:07 it's ocassionally helpful 2011-11-21T19:38:10 -Wno-missing-parameter 2011-11-21T19:38:12 etc. 2011-11-21T19:38:20 ill try 2011-11-21T19:38:34 you'll probably get a bunch of warnings with -Wextra since you haven't been using it 2011-11-21T19:38:42 the only warnign that commonly gives me problems is unsigned signed comparison 2011-11-21T19:38:43 everything in -Wall should be fixed 2011-11-21T19:38:46 lets see :D 2011-11-21T19:38:53 its too much work to put in the casts all the tiem 2011-11-21T19:39:06 Antimony: -Wconversion is usually too annoying 2011-11-21T19:39:14 it's even worse in C 2011-11-21T19:39:37 -Wreorder 2011-11-21T19:39:42 -Wsign-compare 2011-11-21T19:39:47 -Wunused-but-set-variable 2011-11-21T19:39:58 -Wunused-variable 2011-11-21T19:40:00 thats all 2011-11-21T19:40:19 -wreorder doesntmatter, i dont use that feature 2011-11-21T19:40:30 my initilizations dont depend on each other 2011-11-21T19:40:44 easy to fix tho 2011-11-21T19:41:07 -Wsign-compare 2011-11-21T19:41:11 this is more serious 2011-11-21T19:41:21 but still, no problem 2011-11-21T19:41:29 there are a few more useful warnings 2011-11-21T19:41:46 i dont see any more 2011-11-21T19:42:21 the unsets are annoying 2011-11-21T19:42:27 i know they are there 2011-11-21T19:42:31 *** NightExcessive has joined #aichallenge 2011-11-21T19:42:32 well, '-Wall -Wextra' enables a lot of stuff but not all the useful ones 2011-11-21T19:42:38 i just dont always use them since im developing that damn app 2011-11-21T19:43:13 most of them are much more useful in C, if you're heavily using the C++ standard lib you avoid a lot of the issues 2011-11-21T19:43:22 ants.cpp:94:1: warning: control reaches end of non-void function [-Wreturn-type] 2011-11-21T19:43:25 after -Wextra 2011-11-21T19:43:31 this was added only 2011-11-21T19:43:44 i know whats the problem 2011-11-21T19:43:51 if you know it never reaches the end, you can put '__builtin_unreachable()' 2011-11-21T19:43:59 but it's usually a possible edge case 2011-11-21T19:44:05 yes 2011-11-21T19:44:13 it cant really end wo a return 2011-11-21T19:44:19 but the code path is there 2011-11-21T19:44:42 __builtin-unreachable() lets the compiler optimize more (but it's undefined behaviour if it ever reaches there) 2011-11-21T19:44:58 that's a GCC extension, but you can just get rid of them with macros if you want portable code 2011-11-21T19:44:59 where do i put it? 2011-11-21T19:45:12 at the bottom 2011-11-21T19:45:16 (i wont since it doesn matter, not performance critical) 2011-11-21T19:45:40 its in a function i dont even use 2011-11-21T19:45:46 i wrote it up, and forgot abotu it 2011-11-21T19:46:06 -Wshadow is nice, but it doesn't really work in C++ 2011-11-21T19:46:11 due to function overloads being okay 2011-11-21T19:46:31 i dont get that 2011-11-21T19:46:37 what does it warn about? 2011-11-21T19:46:45 so if you have a for loop 2011-11-21T19:46:49 with i as the loop control variable 2011-11-21T19:46:55 and then you redefine i in an inner loop 2011-11-21T19:46:59 that's okay, but it will warn you 2011-11-21T19:47:16 *** fpcfan has joined #aichallenge 2011-11-21T19:47:16 ok, i get that 2011-11-21T19:47:19 it makes you use different variable names even though they are scoped basically 2011-11-21T19:47:25 but what does it have to do with function overlaoding? 2011-11-21T19:47:40 in C, it stops you from naming a function "abs" if you're including math.h already 2011-11-21T19:47:53 in C++, you're just making an overload of the function 2011-11-21T19:48:01 aham 2011-11-21T19:48:09 but its in a different namespace too 2011-11-21T19:48:16 yeah 2011-11-21T19:48:20 be careful about that though 2011-11-21T19:48:28 https://en.wikipedia.org/wiki/Argument-dependent_name_lookup 2011-11-21T19:49:22 namespaces in C++ are kind of broken, in that you can call 'abs(-5)' when you're including cmath 2011-11-21T19:49:25 even without a using statement 2011-11-21T19:49:33 it will use argument-dependent name lookup 2011-11-21T19:49:37 really bad feature imo 2011-11-21T19:49:50 lol 2011-11-21T19:49:52 yes 2011-11-21T19:49:56 i didint know this 2011-11-21T19:50:02 meh :P 2011-11-21T19:50:06 it's required for iostream's overload of >> and << to work 2011-11-21T19:50:17 i love the << and >> syntax :D 2011-11-21T19:50:19 thestinger: i thought friend is for that 2011-11-21T19:50:23 antimatroid: shut up 2011-11-21T19:50:26 it composes so nicel 2011-11-21T19:50:27 :D 2011-11-21T19:50:30 never :p; 2011-11-21T19:50:56 i've never seen a nicer syntax for io in a language 2011-11-21T19:50:56 antimatroid: yeah, but I think that's one area that bjarne views as a mistake 2011-11-21T19:51:07 because of the implications? 2011-11-21T19:51:09 it's so nice! 2011-11-21T19:51:11 no 2011-11-21T19:51:18 why? 2011-11-21T19:51:21 because it's an abuse of operator overloading that they only did as an example 2011-11-21T19:51:26 and it's unsafe 2011-11-21T19:51:29 it can't return an error to you 2011-11-21T19:51:30 an abuse? 2011-11-21T19:51:37 it can't say "end of file" or "broken pipe" 2011-11-21T19:51:39 it also frankly sucks compared to printf 2011-11-21T19:51:40 so? 2011-11-21T19:51:42 it doesn't even throw an exception 2011-11-21T19:51:47 so you can't handle errors properly 2011-11-21T19:51:47 a1k0n: how? :D 2011-11-21T19:51:49 :S* 2011-11-21T19:52:00 other languages do IO nicely too 2011-11-21T19:52:02 look at how Go does it 2011-11-21T19:52:03 thestinger: thats why i read into a stringstream 2011-11-21T19:52:20 i would like << and >> if C++ included a std::string format(fmt, ...) 2011-11-21T19:52:35 print(a, b, c) 2011-11-21T19:52:41 cout << a << b << c << endl; 2011-11-21T19:52:44 why is the 2nd better? 2011-11-21T19:52:52 print can be polymorphic and safe 2011-11-21T19:52:53 cause its chained 2011-11-21T19:53:05 theres a beauty to it 2011-11-21T19:53:08 you basically can't do the same precision stuff with ios, and even when you can it's retarded 2011-11-21T19:53:12 this pipelining paradigm 2011-11-21T19:53:16 how would you overload the print function for variable number parameters etc? 2011-11-21T19:53:19 that's just gross and messy 2011-11-21T19:53:22 no 2011-11-21T19:53:24 a1k0n: thats a pain to set it 2011-11-21T19:53:26 writing a composable output function rules 2011-11-21T19:53:34 *** rajanaresh has joined #aichallenge 2011-11-21T19:53:37 for formatted output id definitely go with printf 2011-11-21T19:53:46 antimatroid: well, haskell also has chaining of I/O like C++ 2011-11-21T19:53:59 it looks similar, but you can handle errors properly 2011-11-21T19:54:05 it's a great idea, it just wasn't executed properly 2011-11-21T19:54:10 like std::vector 2011-11-21T19:54:10 so the problem is error handling? :P 2011-11-21T19:54:19 i friggin hate vector 2011-11-21T19:54:24 they are going to fix it 2011-11-21T19:54:32 isn't it specialized to a bitset? 2011-11-21T19:54:37 yes, basically 2011-11-21T19:54:40 whats wrong with vector-bool? 2011-11-21T19:54:46 mcstar: it's not a container 2011-11-21T19:54:49 i can't write a Matrix operator for a vector thing and have an index function for a Location because you can't return a reference to an element of a vector of bools 2011-11-21T19:54:52 such a pita 2011-11-21T19:54:52 it doesn't hold bools 2011-11-21T19:55:05 holds ints right? 2011-11-21T19:55:05 it's a heap-allocated bitset that you can resize 2011-11-21T19:55:06 Matrix struct* 2011-11-21T19:55:11 but has the functinality of bool 2011-11-21T19:55:16 mcstar: no, it compresses them into bitfields basically 2011-11-21T19:55:17 whats wrong there? 2011-11-21T19:55:18 it's slow as hell 2011-11-21T19:55:22 oh, i see. it can'gt possibly return a bool& 2011-11-21T19:55:23 *** fpcfan has quit IRC (Ping timeout: 244 seconds) 2011-11-21T19:55:25 it's space-optimized 2011-11-21T19:55:25 it doesn't let you return references to its elements! that's annoying! 2011-11-21T19:55:33 aham i see 2011-11-21T19:55:33 they made a replacement 2011-11-21T19:55:42 that's also a heap-allocated bitset 2011-11-21T19:55:50 so now there are two things with identical functionality 2011-11-21T19:55:51 use a vector 2011-11-21T19:55:55 so people who need a bitset can use the new one 2011-11-21T19:56:03 does operator[] return some sort of proxy? 2011-11-21T19:56:04 and in 2015 or something they can make std::vector a real container 2011-11-21T19:56:22 returns what you tell it to 2011-11-21T19:56:23 C++: the world's biggest collection of ugly hacks 2011-11-21T19:56:23 you can still return a reference to a vector of bools so you can still index by [r][c], so i just live with that 2011-11-21T19:56:24 no? 2011-11-21T19:56:38 but grid[loc.row][loc.col] sucks compared to grid[loc] 2011-11-21T19:56:53 http://en.cppreference.com/w/cpp/utility/bitset they had this before, but it's fixed-size 2011-11-21T19:56:59 so it's not a replacement for std::vector 2011-11-21T19:57:02 yeah i end up using operator() for that 2011-11-21T19:57:12 the problem is now people are using it as a space-optimized way to store bits 2011-11-21T19:57:18 so they have to do a really slow transistion 2011-11-21T19:57:18 a1k0n: http://pastebin.com/F9uJKpGv those are my index operators for my grid 2011-11-21T19:57:19 i dont get this, why do you want to use bool then? 2011-11-21T19:57:33 mcstar: me? 2011-11-21T19:57:36 mcstar: you can use uint8_t and it's basically the same 2011-11-21T19:57:42 i didn't realize you could do that 2011-11-21T19:57:46 so there it is 2011-11-21T19:58:02 i did pretty much the same but using () 2011-11-21T19:58:10 i prefer the [] notation :) 2011-11-21T19:58:14 me too 2011-11-21T19:58:17 yeah 2011-11-21T19:58:26 template T& _array::operator[](pos p){ 2011-11-21T19:58:27 return m[wrapR(p.row)*_cols+wrapC(p.col)]; 2011-11-21T19:58:29 } 2011-11-21T19:58:47 I really like how "operators" work in haskell 2011-11-21T19:58:50 yeah, grid[pos] is nice 2011-11-21T19:59:04 *** amstan_ has joined #aichallenge 2011-11-21T19:59:04 *** ChanServ sets mode: +o amstan_ 2011-11-21T19:59:06 that was the first change I made to the bot 2011-11-21T19:59:07 *** peyton has quit IRC (Quit: peyton) 2011-11-21T19:59:08 you can call functions that take 2 arguments as infix 2011-11-21T19:59:17 and you can make them infix by default - so you can make your own operators 2011-11-21T19:59:33 Antimony: did you turn your grid into a graph? 2011-11-21T19:59:38 no 2011-11-21T19:59:40 mod and rem are functions, but you usually call them like '5 `rem` 3' 2011-11-21T19:59:42 why would I? 2011-11-21T19:59:55 and '5 + 3' is the same as '(+) 5 3' 2011-11-21T19:59:57 each square in my grid has a vector of land neighbours and a vector of all neighbours 2011-11-21T19:59:59 it's just infix by default 2011-11-21T20:00:00 thestinger: lets see some foldl magic! 2011-11-21T20:00:05 it makes code much nicer when pathfinding etc. 2011-11-21T20:00:06 that's just a waste of spce 2011-11-21T20:00:12 cause you just grab the land neighbours of a location 2011-11-21T20:00:16 space isn't an issue 2011-11-21T20:00:37 personally, I just use for (int i=0; i<5; ++i) {... 2011-11-21T20:00:41 mcstar: meh, haskell :P 2011-11-21T20:00:42 the lesson: everyone has their preferences and solutions 2011-11-21T20:00:51 i know im right 2011-11-21T20:01:00 I love the type system and lots of things about it, but I don't usually like trying to code purely functional crap 2011-11-21T20:01:09 too much of a fight with the compiler/language most of the time imo 2011-11-21T20:01:11 it's almost certainly faster as well 2011-11-21T20:01:18 thestinger: i love haskells type system 2011-11-21T20:01:20 *** NightExcessive has quit IRC (Quit: Leaving) 2011-11-21T20:01:26 i with c++ had type classes and type variables 2011-11-21T20:01:28 wish* 2011-11-21T20:01:30 Antimony: my tiles are structs with neigbor pointers 2011-11-21T20:01:39 this must be faster than a loop 2011-11-21T20:01:40 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-11-21T20:01:41 I really wish other languages let you call any function infix like an operator 2011-11-21T20:01:43 no 2011-11-21T20:01:44 since its always unrolled 2011-11-21T20:01:49 pointer dereferences are expensive 2011-11-21T20:01:59 i wish languages had operator precedence overloading 2011-11-21T20:02:05 special casing +, -, >>, <<, etc. is stupid 2011-11-21T20:02:06 but i dont have to traverse to avoid water 2011-11-21T20:02:08 and i wish c++ wasn't so restrictive on what you could overload as an operator 2011-11-21T20:02:15 still, it's only one lookup 2011-11-21T20:02:26 operator overloading is nice, and people wouldn't abuse it if they could make their own operators 2011-11-21T20:02:29 and i wish you could specify wether an operator is to be in/pre/post fix 2011-11-21T20:02:55 there comes lisp: with macros/functions/reader macros 2011-11-21T20:03:07 i want to keep the c++ syntax too :P 2011-11-21T20:03:13 *** peyton has joined #aichallenge 2011-11-21T20:03:14 i love c++'s syntax, despite all the hate it gets 2011-11-21T20:03:15 I want to keep msot of C++ syntax 2011-11-21T20:03:20 and distinct spaces for variables/function values 2011-11-21T20:03:30 but some things I'd take out 2011-11-21T20:03:53 I like the changes Go makes to the C-style syntax/statements 2011-11-21T20:04:06 i don't like how messy virtual functions etc. are 2011-11-21T20:04:15 i have to actually read what the hell they are every time i want to write one 2011-11-21T20:04:15 for i := 0; i < 10; i++ { } 2011-11-21T20:04:19 cause it's just so messy 2011-11-21T20:04:34 abstract virtuals? 2011-11-21T20:04:36 are you actually using inheritance in your bot? 2011-11-21T20:04:38 they are great 2011-11-21T20:04:46 no i'm not, i mean in general 2011-11-21T20:04:50 oh ok 2011-11-21T20:04:55 c++ woudl suck oo wise wo them 2011-11-21T20:05:01 it's useful sometimes, but not in a situation like this 2011-11-21T20:05:12 I use them a lot in Pickory, a big game I wrote 2011-11-21T20:05:18 but also writing multiple say order functions and using those for say a priority queue is also a pita 2011-11-21T20:05:27 definitely 2011-11-21T20:05:32 it would be cool if i could just pass an operator to the priority queue 2011-11-21T20:05:36 type classes for templates would be so awesome 2011-11-21T20:05:37 my priority queue needs work 2011-11-21T20:05:38 and that's the order operator 2011-11-21T20:05:44 i just use the stl one :P 2011-11-21T20:06:00 I couldn't use the stl one because I needed to be able to change keys 2011-11-21T20:06:03 and resort 2011-11-21T20:06:07 although if you just want to use one type of order function you can just overload operator< 2011-11-21T20:06:12 https://en.wikipedia.org/wiki/Concepts_(C++) 2011-11-21T20:06:26 *** fpcfan has joined #aichallenge 2011-11-21T20:06:27 really sad that they didn't include this in C++11 2011-11-21T20:06:40 I think some of the compiler vendors were opposed to implementing it 2011-11-21T20:06:52 (*cough* microsoft) 2011-11-21T20:06:56 is anyone using constraint satisfaction/optimization in their bots? 2011-11-21T20:07:01 gcc actually started to implement it, but I guess they won't now 2011-11-21T20:07:04 i want to make money on c++, after i hope it dies out 2011-11-21T20:07:12 ? 2011-11-21T20:07:27 *** JorgeB has joined #aichallenge 2011-11-21T20:07:30 another annoying thing is "priority_queue, greater >" i would intuitively expect to return the greatest elements with highest priority 2011-11-21T20:07:32 hm, after that* 2011-11-21T20:07:40 but it orders the vector by greater then returns the last element 2011-11-21T20:07:43 giving the minimum 2011-11-21T20:07:53 I just made my own prioirty queue class 2011-11-21T20:08:09 it calls std algorithms to maintain the heap 2011-11-21T20:08:41 I'm liking std::array, I'm replacing all my C arrays with it :) 2011-11-21T20:09:10 i haven't looked at that 2011-11-21T20:09:23 it's a stack allocated, fixed-size array 2011-11-21T20:09:27 so it replaces C arrays 2011-11-21T20:10:00 so you can use a.size() 2011-11-21T20:10:03 which is a constexpr 2011-11-21T20:10:08 it compiles to the same code as a C array 2011-11-21T20:10:24 (I tested that) 2011-11-21T20:10:39 I used boost array a lot in my old game* 2011-11-21T20:10:50 it also doesn't let you use it as a pointer without making it explicit 2011-11-21T20:10:52 I haven't gotten around to figuring out what parts of boost are now tandard 2011-11-21T20:10:56 and which ones are supported in my compiler 2011-11-21T20:11:13 http://gcc.gnu.org/projects/cxx0x.html 2011-11-21T20:11:20 gcc implements almost all of C++11 now 2011-11-21T20:11:43 yeah, proper arrays 2011-11-21T20:11:46 and libstdc++ seems to have almost all the new stuff (unordered_map, unordered_set, array, etc.) 2011-11-21T20:12:15 mcstar: i think i got half decent battle code going 2011-11-21T20:12:24 great 2011-11-21T20:12:27 it sucks other than that and food collecting though 2011-11-21T20:12:43 i greatly reduced the number of nodes in each tree 2011-11-21T20:12:47 im writing cooperation now 2011-11-21T20:12:55 oh 2011-11-21T20:13:08 so it is possible 2011-11-21T20:13:17 for each partition of the ants i just pick out defensive and offensive moves for my ants and the enemies ants then pick moves based off that 2x2 game 2011-11-21T20:13:27 unfortunately, I'm stuck with gcc 4.4.1 2011-11-21T20:13:33 sicne I'm a Windows noob 2011-11-21T20:13:52 Antimony: i compiled my app on windows with mingw 4.5.2-1 2011-11-21T20:14:09 and cygwin has gcc 4.6 too 2011-11-21T20:14:11 did you have to install mingw manually? 2011-11-21T20:14:12 so i dont get it 2011-11-21T20:14:19 with cygwin 2011-11-21T20:14:34 I've looked into it a couple times but I'm not brave enough to try reinstalling mingw 2011-11-21T20:14:36 <3 mingw 2011-11-21T20:14:51 Antimony: code::blocks comes with mingw compiler if you want 2011-11-21T20:14:56 argh 2011-11-21T20:15:02 but you're probably an emacs/vim guy :P 2011-11-21T20:15:08 Antimony: just try it, it is painless 2011-11-21T20:15:21 *** pairofdice has quit IRC (Quit: leaving) 2011-11-21T20:15:23 i did in a vm couple of days ago just to compile my app for windows users 2011-11-21T20:15:27 codeblocks comes with 4.4.1 2011-11-21T20:15:30 it wasnt a big deal 2011-11-21T20:15:38 ah okay 2011-11-21T20:15:50 I just installed the newest version of codeblocks and thats it 2011-11-21T20:16:11 <3 code::blocks :P 2011-11-21T20:16:27 try kdevelop <3 kdevelop 2011-11-21T20:16:33 lol, no idont 2011-11-21T20:16:33 I've never used it 2011-11-21T20:16:37 but its nice 2011-11-21T20:16:43 although i have a terrible habbit of hitting ctrl f11 instead of ctrl f9 and recompiling my entire project every time 2011-11-21T20:16:46 I was using clang to do autocompletion and stuff (you can use it with basically any editor I guess), but it doesn't have nearly the same C++11 support as gcc 2011-11-21T20:17:07 *** NightExcessive has joined #aichallenge 2011-11-21T20:17:10 the keys are closer together :P 2011-11-21T20:17:11 doesnt it popup a warning if you try to rebuild 2011-11-21T20:17:12 i still type make 2011-11-21T20:17:19 so now it underlines all kinds of stuff as syntax errors and fails to autocomplete for anything that touches the new stuff in the standard library 2011-11-21T20:17:21 I hate makefiles 2011-11-21T20:17:25 i probably got rid of it the first time it came up if it does :p 2011-11-21T20:17:34 especially the ones you get from large projects 2011-11-21T20:17:43 where you have 3 or 4 files recursive invoking each other 2011-11-21T20:17:47 with crazy macros 2011-11-21T20:18:17 makefiles are fine imo, it's the horrific ones that get generated with autotools that are a problem 2011-11-21T20:18:23 *** yoden has joined #aichallenge 2011-11-21T20:18:47 they are not human readable 2011-11-21T20:18:50 *** bmh has joined #aichallenge 2011-11-21T20:18:54 but thats the point 2011-11-21T20:19:06 you make a tool to make a makefile 2011-11-21T20:19:19 soon you will make a tool that will make a configure that makes a makefile 2011-11-21T20:19:25 except when tools go rogue and some poor bastard has to work out why :P 2011-11-21T20:19:53 yeah. I had trouble buildign a project with a makefile, ad couldn't figure out what hte problem was 2011-11-21T20:20:03 it's pretty pointless to use autotools imo 2011-11-21T20:20:06 and then I spent hours reading the documentation for make 2011-11-21T20:20:13 to find out what the stuff in the makefile even meant 2011-11-21T20:20:31 just capture what it really executes 2011-11-21T20:20:37 makefiles are for the birds. I enjoy typing 'ghc --make MyBot' and things just work. 2011-11-21T20:20:38 isn't Cmake bette than autotolls? 2011-11-21T20:20:56 bmh: definitely 2011-11-21T20:21:01 i enjoy just compiling a project file :P 2011-11-21T20:21:12 Antimony: I think cmake is for really big projects 2011-11-21T20:21:24 you don't need autotools/cmake/whatever if you have under 100 files 2011-11-21T20:21:29 i wouldnt know where to start with this game in a purely functional language 2011-11-21T20:21:33 id be dfunctinal in it 2011-11-21T20:21:35 cmake is for something like chromium with 100000 files 2011-11-21T20:21:37 For personal projects, I just hit f9 in codeblocks 2011-11-21T20:21:38 dis* 2011-11-21T20:21:50 the second least bad build system i've used was jam 2011-11-21T20:21:50 mcstar: I'll release my source code on December 18 or 19 2011-11-21T20:21:59 Antimony: that's rull it though yeah? 2011-11-21T20:22:11 bmh: are you the best haskeller? 2011-11-21T20:22:19 for small projects you can just write a makefile with some wildcards and variables and not worry about it until you add another library or language to your project 2011-11-21T20:22:23 mcstar: we're aiming to be. 2011-11-21T20:22:32 mcstar: that's jmcarthur as far as i'm concerned :P 2011-11-21T20:22:32 k 2011-11-21T20:22:36 or use an ide and worry about it even less :) 2011-11-21T20:22:38 oh, bot ranking? 2011-11-21T20:22:49 bmh: if you will be ranked high enough ill consider learning it 2011-11-21T20:22:54 antimatroid: is jmcarthur actually using haskell? 2011-11-21T20:23:02 i don't think he's writing a bot 2011-11-21T20:23:04 but he's a haskeller 2011-11-21T20:23:07 mcstar: Not a chance of being highly ranked :) That's why I chose haskell. 2011-11-21T20:23:12 my problem is that I don't think I'll ever be ranked high enough for anyoen to care about my source code 2011-11-21T20:23:29 Antimony: truism 2011-11-21T20:23:33 Antimony: except you then have to deal with the ide, I think that's worse by far 2011-11-21T20:23:35 yeah meh too 2011-11-21T20:23:37 Antimony: Sure, but if you wrote your bot in Erlang people would read it. 2011-11-21T20:23:46 Antimony: nobody cares about source code of others much except maybe first place 2011-11-21T20:24:01 actually the only thing that matters is how good your blog post about it is 2011-11-21T20:24:02 or it its in a really wird language 2011-11-21T20:24:03 like j 2011-11-21T20:24:07 I don't know. I think top 3 or 5 have a chance 2011-11-21T20:24:12 if they;re interesting 2011-11-21T20:24:14 a1k0n: lol 2011-11-21T20:24:17 a1k0n: you're not wrong, and i don't have a blog :P 2011-11-21T20:24:25 although someone has a blog who calls themself antimatroid 2011-11-21T20:24:36 I don't have a blog 2011-11-21T20:24:38 2:24 2011-11-21T20:24:40 set up jekyll and done 2011-11-21T20:24:41 good morning 2011-11-21T20:24:44 *** mcstar has quit IRC (Quit: WeeChat 0.3.6) 2011-11-21T20:24:47 guten morgen 2011-11-21T20:24:56 it's afternoon :P 2011-11-21T20:25:01 here it's evening 2011-11-21T20:25:07 well night really 2011-11-21T20:25:16 Antimony: i got confused for you the other day actually 2011-11-21T20:25:25 dimkadimon thought i just changed my name to antimony for ants 2011-11-21T20:25:25 lol 2011-11-21T20:25:30 I ended up writing my own Queue in Haskell, which I might throw up on Hackage given that I don't like the other implementations up there 2011-11-21T20:25:59 when i saw how awful the implementations are for prime seive thing in haskell i lost a lot of interest 2011-11-21T20:26:07 I remember one time every single active person on the channel had a name beginnign with a 2011-11-21T20:26:15 antimatroid: slow? 2011-11-21T20:26:25 bmh: just looks awful the code 2011-11-21T20:26:28 antimatroid: you can write a really sane, fast, simple version 2011-11-21T20:26:39 but you have to cheat and use state/mutable arrays 2011-11-21T20:26:56 thestinger: mutable arrays are too slow :( Use Vector 2011-11-21T20:27:01 you can't do it quickly with purely functional code without being a genius 2011-11-21T20:27:08 *** NightExcessive has quit IRC (Quit: Ciao.) 2011-11-21T20:27:47 ug, no matter what I do, I feel like I should be workign on something else instead 2011-11-21T20:27:53 I think someone used priorities queues and wrote a big paper on it xD 2011-11-21T20:28:01 priority queues* 2011-11-21T20:28:09 thestinger: Praise be to Oksaki? 2011-11-21T20:28:12 okasaki 2011-11-21T20:28:13 ? 2011-11-21T20:28:16 isn't that kind of like an embarrassment for the language though? 2011-11-21T20:28:20 jinx 2011-11-21T20:28:31 you can't easily find primes fast unless you know lots of the language? 2011-11-21T20:28:37 http://stackoverflow.com/questions/2221378/the-genuine-sieve-of-eratosthenes-algorithm-used-to-generate-prime-numbers 2011-11-21T20:28:55 a1k0n: do you have a copy of his PhD thesis? 2011-11-21T20:29:07 antimatroid: well, you can cheat and code it imperatively - haskell lets you do that 2011-11-21T20:29:13 i had his book, but gave it away as a gift 2011-11-21T20:29:27 you have to really know what you're doing to write fast code in haskell without making it impure 2011-11-21T20:29:31 that's not so bad then i guess 2011-11-21T20:29:43 perhaps that's an embarassment for functional programming in general? 2011-11-21T20:30:10 *** NightExcessive has joined #aichallenge 2011-11-21T20:30:11 yeah 2011-11-21T20:30:13 nah. I find the trouble with writing fast code in haskell doesn't stem from purity, but from laziness. 2011-11-21T20:30:29 profound 2011-11-21T20:30:42 a1k0n: touche. 2011-11-21T20:30:55 rank stabilized >.< http://ants.fluxid.pl/replay.9792 2011-11-21T20:31:11 I should have won that by extermination 2011-11-21T20:31:17 *** Conorach_ has quit IRC (Ping timeout: 259 seconds) 2011-11-21T20:32:19 hmmm i did point out a while ago that you need to give remaining ants time to alter the outcome of the game, i don't know if anyone did anything 2011-11-21T20:32:52 they seem happy to call the game off as soon as first place is decided, but the overall rankings care about the ordinal ranking of the bots 2011-11-21T20:33:10 not just remaining ants, I was the only one with an ant hill left in the game yet I lost 2011-11-21T20:33:24 ie. you shouldn't call the game off if a player has taken out 2 enemies and will likely take out the third 2011-11-21T20:33:24 it's not just first place when it ends early 2011-11-21T20:33:30 if I had gotten time to kill off the enemy ants, I woul have won from bonus points 2011-11-21T20:33:37 i've seen games at 40 3 3 drag on 2011-11-21T20:34:06 perhaps they should adopt something like Go endgame scoring to determine 'dead' ants in those caes 2011-11-21T20:34:31 * bmh punts on describe a scoring strategy 2011-11-21T20:34:33 McLeopold: are we ever going to fix up when "rank stabilises"? at the moment some games are called off before the ranks have actually been decided 2011-11-21T20:35:45 personally, I think that the penalty for timing out should be abolished as well 2011-11-21T20:36:03 that fluxid replay sure doesn't seem like "rank stabilized" BUT the reason I think is that hill-less ants are not given further opportunity to cap hills 2011-11-21T20:36:04 hehe, timing out was qisqualification in tron 2011-11-21T20:36:10 in planet wars it was an instant loss 2011-11-21T20:36:16 dis* 2011-11-21T20:36:18 it's practically an instant loss now 2011-11-21T20:36:58 i don't really have a strong opinion either way tbh 2011-11-21T20:36:59 the problem is that random variations in timing mean you have to be very conservative if you don't want to ever timeout 2011-11-21T20:37:18 on one side you did timeout, on the other we are running an ai contest not a programming contest 2011-11-21T20:37:30 really? 2011-11-21T20:37:33 yeah 2011-11-21T20:37:34 i intend to use most of my time 2011-11-21T20:37:39 and have timeout issues atm 2011-11-21T20:37:42 I've had bots time out when returning 30ms early 2011-11-21T20:37:43 although mostly on tcp i think 2011-11-21T20:37:57 Antimony: on tcp? that's different 2011-11-21T20:38:01 no local 2011-11-21T20:38:13 I submitted one that returns 25ms early to teh server 2011-11-21T20:38:13 What's the point of timing out bots? If a bot misses the deadline, just skip them for the turn. 2011-11-21T20:38:15 are you using a language with garbage collection? 2011-11-21T20:38:16 and it tiemd out twice 2011-11-21T20:38:23 no, C++ 2011-11-21T20:38:25 i wonder if we could get the engine to use an actual turnTime 50ms or so above the one specified 2011-11-21T20:38:26 Antimony: Have you read the engine code? It takes something like 10ms per hundred lines of bot outupt 2011-11-21T20:38:27 surely the problesm worse for java 2011-11-21T20:38:30 *** NightExcessive has quit IRC (Quit: Ciao.) 2011-11-21T20:38:34 so you may need to compensate that if you have a lot of orders to issue 2011-11-21T20:38:51 *** NightExcessive has joined #aichallenge 2011-11-21T20:39:20 hm, might even make sense to send orders early because those can "overlap" 2011-11-21T20:39:37 what do you mean? 2011-11-21T20:39:46 i output my moves as i make them 2011-11-21T20:39:47 so basically engine.timeRemaining + (engine.myAnts.length * 10) is what one should compare against? 2011-11-21T20:39:51 rather than all at the end 2011-11-21T20:40:13 oh never mind 2011-11-21T20:40:33 you need to return over 30ms early if you want to be safe 2011-11-21T20:40:40 with 300 ants that's still just 30 ms 2011-11-21T20:40:41 it's silly 2011-11-21T20:40:51 it's obviously not input that's the whole problem 2011-11-21T20:40:55 Antimony: The sandbox engine that runs the bots does "while sleep 10ms read 100 lines" 2011-11-21T20:41:03 wow 2011-11-21T20:41:09 but even with say 50 ants, it's a problem 2011-11-21T20:41:18 *** NightExcessive_ has joined #aichallenge 2011-11-21T20:41:19 I only know that because you have to hack that with the visualizer or you will time out a lot 2011-11-21T20:41:35 heh, so maybe I shouldn't just flush once at the end 2011-11-21T20:41:43 Antimony: i guess one can just go turntime-=50 after reading it 2011-11-21T20:41:44 yeah I just realized that myself 2011-11-21T20:41:44 *** NightExcessive has quit IRC (Disconnected by services) 2011-11-21T20:41:49 *** NightExcessive_ is now known as NightExcessive 2011-11-21T20:41:59 yeah i just flush every line 2011-11-21T20:42:06 me too 2011-11-21T20:42:08 i get too confused otherwise 2011-11-21T20:42:18 anyone hazard a guess where the cutoff line (in terms of ranking) is for "combat" these days? 2011-11-21T20:42:25 antimatroid: do you have example games where the rank is not stabalized? 2011-11-21T20:42:45 I think timeouts are too harsh. If a bot misses the deadline, just drop all the late orders. 2011-11-21T20:42:45 [17:30] rank stabilized >.< http://ants.fluxid.pl/replay.9792 2011-11-21T20:42:58 McLeopold: in that game Minthod could have still won 2011-11-21T20:42:58 BenJackson: my ancient version on the site has dropped down to 28 (it lacks combat) 2011-11-21T20:43:01 McLeopold: tie for 1st between two bots both of which razed the other's hill 2011-11-21T20:43:25 heh. that was awesome, i usually draw on this map: http://ants.fluxid.pl/replay.9804 2011-11-21T20:43:27 My current bot has no combat and it's #175 2011-11-21T20:43:28 thestinger: You don't fight at *all*? Your ants just blunder aroud? 2011-11-21T20:43:37 wait, no they couldn't? 2011-11-21T20:43:52 McLeopold: i'll get back to you :P 2011-11-21T20:43:52 *** bmh has quit IRC (Quit: back in a while) 2011-11-21T20:44:13 bmh: I like that timeout idea 2011-11-21T20:44:22 that would be nice 2011-11-21T20:44:27 antimatroid: lone survivor gives a point bonus, if I had gotten it I would have won 2011-11-21T20:44:33 but like 2011-11-21T20:44:43 N timeouts in a row should kill it 2011-11-21T20:44:53 you'd have to someout reset the bot, though 2011-11-21T20:44:55 oh, didn't realise there was a lone survivor bonus 2011-11-21T20:45:01 to give up on runaway procs 2011-11-21T20:45:04 McLeopold: that game then :P 2011-11-21T20:45:06 *** NightExcessive_ has joined #aichallenge 2011-11-21T20:45:32 *** NightExcessive has quit IRC (Disconnected by services) 2011-11-21T20:45:37 *** NightExcessive_ is now known as NightExcessive 2011-11-21T20:45:42 thestinger: re combat, seems like I see a lot of people here talking about combat but maybe everyone here is already in the top 30 or so 2011-11-21T20:45:52 McLeopold: http://aichallenge.org/visualizer.php?game=108275&user=4513 what about this game? given more time wouldn't orange probably also come in equal last place? 2011-11-21T20:45:57 I talk about combat because I wish I could get it working 2011-11-21T20:46:14 remember the ordinal rankings of bots matter, not just first place 2011-11-21T20:46:16 I have versions with combat on the tcp server now 2011-11-21T20:46:24 I just haven't uploaded a new version to the site for ~2 weeks 2011-11-21T20:46:41 @later tell janzert http://ants.fluxid.pl/replay.9792 are you up for changing the cutoff rules? 2011-11-21T20:46:41 McLeopold: OK 2011-11-21T20:46:42 *** flag has quit IRC (Ping timeout: 245 seconds) 2011-11-21T20:47:07 I wanted to see how high I would get with no combat, no defense and mediocre foraging 2011-11-21T20:47:13 BenJackson: i just got basic battle code going but i haven't finished other parts of my bot 2011-11-21T20:47:14 so it sucks :P 2011-11-21T20:47:22 I made it up to rank 15, but it's down to 28 now (people have finally reuploaded with better versions) 2011-11-21T20:47:22 but i like how it jives in combat situations 2011-11-21T20:47:46 antimatroid: we don't check for if a bot can lose and change rank, since he lasted the longest, he gets a little bonus 2011-11-21T20:48:08 McLeopold: why give them a bonus and not hte bot that lastest second longest? 2011-11-21T20:49:04 I can't give a good rational explaination, other than due to our cutoff rules, that's where we cutoff 2011-11-21T20:49:06 i realise we want faster games, but we're basically saying orange played better because it was the last hill white went after? 2011-11-21T20:49:08 meh, then you get a bonus just for having a few cautious ants that survive 2011-11-21T20:49:25 McLeopold: that's my point, shouldn't the cutoff rules be altered so that ranks can actually be finalised? 2011-11-21T20:49:30 the games could just be allowed to go on longer in the finals when speed won't matter so much 2011-11-21T20:49:49 I dunno 2011-11-21T20:49:50 thestinger: speed will matter most in the finals for more games :P 2011-11-21T20:49:57 people are basing strategy on turn # 2011-11-21T20:49:57 ah, cause all the bots will get to play 2011-11-21T20:50:01 if you go longer than 1500 that could change things 2011-11-21T20:50:04 or lead to timeouts 2011-11-21T20:50:14 i'm not talking about increasing hte max turns 2011-11-21T20:50:31 i'm talking about calling a game off earlier than the specified max turns when the ranks haven't been finalised 2011-11-21T20:50:50 *** trishume has joined #aichallenge 2011-11-21T20:51:17 i don't care THAT much, but i don't think what we're currently doing is really all that fair 2011-11-21T20:51:18 how about just make all games in finals run to maxturn or extermination? 2011-11-21T20:51:29 yes, you get a rank bonus if you last longer 2011-11-21T20:51:48 McLeopold: but the "lang longer bonus" is so horribly ill defined :P 2011-11-21T20:52:06 you get a bonus for lasting longer if no bots can increase in ranking? 2011-11-21T20:52:18 I haven't actually noticed that bonus 2011-11-21T20:52:27 just the lone survivor one 2011-11-21T20:52:29 which can basically come down to the strongest bot coming after your hill last 2011-11-21T20:52:40 thestinger: http://aichallenge.org/visualizer.php?game=108275&user=4513 2011-11-21T20:52:43 even though orange could be pushed down in rank, he can't gain, so the game ends 2011-11-21T20:52:44 it's a kind of implicit bonus 2011-11-21T20:53:13 thestinger: i would argue that orange should really have equal last ranking in that game, not second 2011-11-21T20:53:13 the other bots, without hills, lost their ability to extend the game 2011-11-21T20:53:24 ah, cause rank is stabilized and you haven't had the -1 for losing the hill :) 2011-11-21T20:53:41 that's enough reason for me to finally merge in this hill defense branch 2011-11-21T20:53:47 thestinger: that's why I said "rank bonus" 2011-11-21T20:53:58 ug, I'm being so distracted. I better get out of the irc 2011-11-21T20:54:12 McLeopold: but orange has the furthest hill from white at the start of the game? 2011-11-21T20:54:18 is it surprising that it's the last hill white went for? 2011-11-21T20:54:20 *** Antimony has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-11-21T20:54:28 did they really play any better to deserve a higher rank? 2011-11-21T20:54:41 aw, I was about to link him to http://tvtropes.org/pmwiki/pmwiki.php/Main/DistractedByTheSexy 2011-11-21T20:55:08 orange being ranked 2nd is more about luck than skill imo 2011-11-21T20:55:58 if tvtropes didn't exist my bot would be so much better :( 2011-11-21T20:56:09 antimatroid: I don't think it was a rush that caused it, white was able to spot blue, but not orange, so I could argue orange did a better job at protection 2011-11-21T20:56:21 there was plenty of time for white to spot either 2011-11-21T20:56:33 sure, but i would expect white to spot the closest hills first 2011-11-21T20:56:51 watches the text scroll of a game go by 2011-11-21T20:56:51 so, I'm not trying to justify that this is the best way to rank, just that is isn't bad enough to change 2011-11-21T20:56:55 and probably also take the closest hills first assuming equal skill of opponents 2011-11-21T20:56:58 I'm always rooting for the little ant counts 2011-11-21T20:57:46 Someone should build in some suicide code to prevent enemy ants from spotting the hills, and see if they do much better... 2011-11-21T20:58:19 I've considered how to deny hill visibility 2011-11-21T20:58:53 BenJackson: ChrisH had some defense code that did something like that 2011-11-21T20:59:04 try looking at his older replays 2011-11-21T20:59:27 he would put ants around his hill to block enemies from getting to it - but he could have put them farther out and blocked viewdistance 2011-11-21T20:59:34 McLeopold: planning on it :P 2011-11-21T20:59:47 I mean around his hill at ~5 radius 2011-11-21T20:59:52 defend and invade enemy hills are two of my next functions to write 2011-11-21T21:00:02 *** plr2 has joined #aichallenge 2011-11-21T21:00:04 and they would adapt to a maze map properly where there were only 2 ways to get to the hill 2011-11-21T21:00:31 it's really not that hard for me to add i think 2011-11-21T21:00:48 i just need to add into my outcomes if an enemy ant resides in view of one of my hills 2011-11-21T21:01:00 and add stuff into my outcome order function 2011-11-21T21:01:23 well, better yet just track which enemies have seen each hill 2011-11-21T21:01:36 and try to prevent an ant from seeing a hill it hasn't seen yet 2011-11-21T21:01:49 it's probably pretty safe to assume that any decent bot will track enemy hills outside view 2011-11-21T21:01:54 *** emiel_ has quit IRC (Quit: emiel_) 2011-11-21T21:02:07 *** twymer has joined #aichallenge 2011-11-21T21:04:08 you can discover hills without seeing them too 2011-11-21T21:04:17 :) 2011-11-21T21:04:34 antimatroid: probably has the most code to detect that :) 2011-11-21T21:06:56 i don't understand the scoring model 2011-11-21T21:07:02 how is score being computed 2011-11-21T21:07:14 antimatroid: actually, that first example game shouldn't have been cutoff... 2011-11-21T21:07:20 +2 for every hill you destroy, -1 for every hill you lose 2011-11-21T21:07:32 with a starting score of # of hills 2011-11-21T21:07:37 and if i don't lose any? 2011-11-21T21:07:40 till the end 2011-11-21T21:07:46 so nobody cries if they go negative 2011-11-21T21:07:51 at the end it's the number of hills you own or have captured 2011-11-21T21:08:12 er i guess not quite 2011-11-21T21:08:15 so ant counts don't matter at all? 2011-11-21T21:08:20 nope 2011-11-21T21:08:21 only hills? 2011-11-21T21:08:24 yep 2011-11-21T21:08:28 that makes for a wonky scoring 2011-11-21T21:08:34 a bot that kilsl another hill and then dies 2011-11-21T21:08:40 gets a higher score than a bot that survives? 2011-11-21T21:08:47 but didn't get to kill a hill? 2011-11-21T21:08:52 because surviving doesn't matter 2011-11-21T21:08:59 razing hills does 2011-11-21T21:09:01 that is why my bot is doing relatively well 2011-11-21T21:09:02 that makes no sense in a strategy game 2011-11-21T21:09:17 a survivor should always get a higher score than a dead bot 2011-11-21T21:09:33 if the bot is dumb enough to die 2011-11-21T21:09:48 then makes no sense that a survivor to be penalized 2011-11-21T21:09:51 but the bot may just be ganed up on 2011-11-21T21:09:58 doesn't make it better? 2011-11-21T21:10:06 any bot can be ganged up on 2011-11-21T21:10:06 so, survival length isn't always a measure of skill 2011-11-21T21:10:24 but if i end the game with like 30% of surviving ants 2011-11-21T21:10:25 it's a reasonable scoring system 2011-11-21T21:10:33 i should be better than a bot that ends game with like 1 2011-11-21T21:10:42 and probably got lucky enough to raze a hill early on? 2011-11-21T21:10:47 in a kamikaze system 2011-11-21T21:11:03 all maps are symmetric, kamikaze is a valid strategy 2011-11-21T21:11:38 sigh i guess i will have to rewrite my bot to be a bit more aggressive about attacking hills then 2011-11-21T21:11:50 though i still believe that the scoring is wonky 2011-11-21T21:11:58 *** Euphony has joined #aichallenge 2011-11-21T21:12:00 makes sense for the victor but not so for the middle bots 2011-11-21T21:12:13 if i just happen to be a bot who had strong neighbours 2011-11-21T21:12:18 i may never get to kill a hill 2011-11-21T21:12:35 but another one crappier bot might be able to go gaga on a weaker neighbor and then die 2011-11-21T21:12:42 but he will rank higher than my bot 2011-11-21T21:12:43 huh, I just had a game end at 1000 due to "turn limit reached" 2011-11-21T21:12:52 I could see "ants not razing hills" 2011-11-21T21:13:38 http://ants.fluxid.pl/replay.9811 yay for a 1497 turn game >.< 2011-11-21T21:14:12 like in this game 2011-11-21T21:14:13 http://ants.fluxid.pl/replay.9831 2011-11-21T21:14:20 holy shit what a wall of red 2011-11-21T21:14:59 http://aichallenge.org/visualizer.php?game=114044&user=3473 2011-11-21T21:15:06 did that game really run with a 1000 turn limit? 2011-11-21T21:15:13 i end the game with 21 ants but i end up 6th or something 2011-11-21T21:18:14 bhasker: so basketball games should be decided by time of posession? 2011-11-21T21:18:44 lets not make comparisons rules are very different. 2011-11-21T21:18:59 in basketball games you are not against 7 different opponents 2011-11-21T21:19:21 so chess should be decided by the number of pieces left at the end? 2011-11-21T21:19:28 again 1v1 2011-11-21T21:19:51 I don't think RTS-style FFA rules would be very fun 2011-11-21T21:20:01 *** emiel_ has joined #aichallenge 2011-11-21T21:20:02 there's already enough turtling/standoffs 2011-11-21T21:20:03 it is an RTS? 2011-11-21T21:20:06 for most purposes 2011-11-21T21:20:24 and i would be fine with the rules if this was a 1v1 2011-11-21T21:20:27 *** emiel_ has quit IRC (Client Quit) 2011-11-21T21:20:34 this is not a 1v1 2011-11-21T21:20:37 no, but I'm comparing the multiplayer maps to a multiplayer RTS free-for-all 2011-11-21T21:20:42 where the last player standing wins 2011-11-21T21:20:51 this game is very ffa'ey 2011-11-21T21:21:01 yes all the more reason to have a better scoring model 2011-11-21T21:21:03 excessively so, you can get screwed jsut by having the wrong opponent next to you 2011-11-21T21:21:04 so someone can just turtle and survive to the end 2011-11-21T21:21:11 I don't like that for this 2011-11-21T21:21:14 just razing hills is not a very good scoring model 2011-11-21T21:21:29 because combat already ends up creating big stalemates 2011-11-21T21:21:36 maybe next year we will have to program an RTS ai! 2011-11-21T21:21:39 situation: http://pastebin.com/7J1zr716 2011-11-21T21:21:40 with resource gathering and all :P 2011-11-21T21:21:55 any of you code code to make your ant try and have the enemy follow you around and take their hill? 2011-11-21T21:22:33 *** bmh has joined #aichallenge 2011-11-21T21:22:58 antimatroid: no, but I've considering pretending enemy attack radius isn't passable when ants are alone 2011-11-21T21:23:03 well, ignoring that he could probably get you around the edge with that example 2011-11-21T21:23:18 i personally think the current scoring model is rather broken in more than 2 player games 2011-11-21T21:23:20 or just making the cost for going through it really high 2011-11-21T21:23:28 i run battle code for all ants that might end up in a battle 2011-11-21T21:23:36 if in a 3 player game if 2 players decide to send ants at each other and nothing else 2011-11-21T21:23:41 the third player will lose 2011-11-21T21:23:50 if he doesn't get to either of the hills before the other 2 2011-11-21T21:23:53 makes no sense 2011-11-21T21:23:58 they kamikaze and win 2011-11-21T21:24:07 while the surviving player loses 2011-11-21T21:24:09 that is a bit weird 2011-11-21T21:24:25 increase the 'losing your hill' penalty? 2011-11-21T21:24:31 but he has no choice? 2011-11-21T21:24:36 there are only 2 hills? 2011-11-21T21:24:39 and they killed each other? 2011-11-21T21:24:42 what does he do now? 2011-11-21T21:24:51 there is nothing left to kill 2011-11-21T21:24:51 I think the game just needs to keep going instead of doing rank stabilized in that case 2011-11-21T21:24:55 bhasker: well .. killing hills is important, surviving less. 2011-11-21T21:25:00 but... 2011-11-21T21:25:00 there's a bonus for being the last survivor 2011-11-21T21:25:04 there are none left tokil? 2011-11-21T21:25:05 bots can't (lawfully) coordinate. In the top tier games I hope to see all permutations played out. 2011-11-21T21:25:11 i am not saying coordinate 2011-11-21T21:25:16 thestinger: the game would probably end before the other ants are all dead 2011-11-21T21:25:20 cause of rank stabilising 2011-11-21T21:25:22 i am just saying that lets say 2 bots were nearby and 3rd was a little further 2011-11-21T21:25:24 otherwise yeah it'd be a draw 2011-11-21T21:25:25 bhasker: the 3 player scenario you described is a bug :( 2011-11-21T21:25:26 so ... 2011-11-21T21:25:35 mcleopold it can happen with more than 3 too 2011-11-21T21:25:38 it can happen with 5 2011-11-21T21:25:42 it can happen with 7 2011-11-21T21:25:46 any odd number 2011-11-21T21:25:51 McLeopold: if you gave the last bot with hill the lone surviving bonus it'd be a draw? 2011-11-21T21:25:51 of players in a game 2011-11-21T21:25:52 it can happen with *any* number 2011-11-21T21:26:04 with even numbers with symmetrical maps less so 2011-11-21T21:26:06 A -> B -> C -> D -> E -> A, F is left out in the cold. 2011-11-21T21:26:11 bhasker: so you need to make your bot willing to rush at hills on the small maps 2011-11-21T21:26:17 thats not the point 2011-11-21T21:26:19 it's part of the challenge - balancing offense and defense 2011-11-21T21:26:21 the scoring model is broken 2011-11-21T21:26:26 i could lose even if i did the best thing 2011-11-21T21:26:39 thats where my problem is 2011-11-21T21:26:40 but you didn't do the *best* thing 2011-11-21T21:26:41 the best thing would be to ninja both their hills as they fight each other 2011-11-21T21:26:45 in a 3 player game 2011-11-21T21:26:50 if i am a little further away 2011-11-21T21:26:54 you aren't 2011-11-21T21:26:55 i will lose always 2011-11-21T21:26:56 it's symmetrical 2011-11-21T21:27:06 distances to every player to each other are not 2011-11-21T21:27:06 bhasker: i guess some rational could be that you need to get involved in the game 2011-11-21T21:27:21 those two ants got engaged into battle and killed each other while you were still sitting about setting up or something 2011-11-21T21:27:35 yes but …does that mean my bot is worse? 2011-11-21T21:27:37 yeah, if they're busy fighting each other they have their hills exposed 2011-11-21T21:27:43 but it makes no sense when you turn around and look at it from the point of view that you now have a hill and the entire world to yourself pretty much 2011-11-21T21:27:55 thestinger: they take out each others hills 2011-11-21T21:27:58 this scoring model only rewards one strategy 2011-11-21T21:28:05 antimatroid: ah 2011-11-21T21:28:06 in an RTS multiple strategies should be alowed 2011-11-21T21:28:17 bhasker: the scoring model is broken according to *your* perspective of what the game should be, it is not actually broken 2011-11-21T21:28:32 the game isn't what you want it to be, it is what it is 2011-11-21T21:28:33 you can do an early rush and then turtle 2011-11-21T21:28:38 there are many valid strategies 2011-11-21T21:28:40 the game is geared towards faster games 2011-11-21T21:28:49 why not make the hills respawn? 2011-11-21T21:28:49 protecting your hill is worth it, but you need to take risks 2011-11-21T21:28:51 it just isn't feasible otherwise with the number of competitors 2011-11-21T21:28:53 that would solve the problem 2011-11-21T21:28:59 while I'd love to see more tactical variety in this game, I think adding that tactical variety would make the programming a lot harder. 2011-11-21T21:29:03 respawn hills? 2011-11-21T21:29:21 strcat_camping :) 2011-11-21T21:29:22 hills don't respawn because then the games would take longer 2011-11-21T21:29:24 yes if they get razed after a few turns if no enemy ants are nearby respawn? 2011-11-21T21:29:31 no ants rather 2011-11-21T21:29:34 or add neutral hills 2011-11-21T21:29:38 that can be taken over 2011-11-21T21:29:42 neutral hills would be fun 2011-11-21T21:29:54 by competing bots and converted 2011-11-21T21:29:55 neutral hills would be like planet wars a bit 2011-11-21T21:30:02 All of the server code is out there. If people feel like playing this game, with a different scoring model, after the competition is over no one is going to stop them 2011-11-21T21:30:02 and also leads to longer games 2011-11-21T21:30:17 capturing hills instead of destroying them would be an interesting twist 2011-11-21T21:30:23 I'd like to see someone run a tcp server like that 2011-11-21T21:30:50 yeah that would be loads fo fun 2011-11-21T21:30:50 anyway i guess its too late to argue about the scoring model but it does seem a bit unfair in more than 2 player games 2011-11-21T21:30:50 is it possible to take a hill that is producing ants? 2011-11-21T21:30:51 if people are swapping ownership of hills the games could take ages 2011-11-21T21:30:59 but would require changing bot logic 2011-11-21T21:31:01 bmh: if they move off the hill right away, yes 2011-11-21T21:31:05 bmh: no 2011-11-21T21:31:17 animatroid: it could be a one time take thing 2011-11-21T21:31:19 well, if they move off and you kill that ant sure 2011-11-21T21:31:27 you take it and its yours, if someone destroys it it is gone forever 2011-11-21T21:31:29 but ants spawn between moves and battle resolution yeah? 2011-11-21T21:31:33 I have code to figure out whether they are staying on their hill or moving off 2011-11-21T21:31:37 so if the hive has enough ants there then no? 2011-11-21T21:31:44 antimatroid: no, you can move them the turn they spawn 2011-11-21T21:31:46 you could do tons of cool additions to this game 2011-11-21T21:31:53 add forts on map that give tactical advantages 2011-11-21T21:31:59 antimatroid: you can take a hill with 300+ ants ready to spawn if they don't stay on it when they spawn 2011-11-21T21:32:01 add different ant personalities 2011-11-21T21:32:01 yes, but then after it moves doesn't another ant spawn there before battle code is run? 2011-11-21T21:32:04 but the more you add the harder it gets 2011-11-21T21:32:07 antimatroid: nope 2011-11-21T21:32:14 I had to write code to deal with it 2011-11-21T21:32:20 i thought that was the case :\ 2011-11-21T21:32:24 because if you surround their hill 2011-11-21T21:32:28 and they are really moving off it right away 2011-11-21T21:32:31 they are colliding with you 2011-11-21T21:32:42 and it's just as bad as trying to take the hill while ants are spawning 2011-11-21T21:32:44 if a hills is spawning ants, it can't be razed 2011-11-21T21:32:48 so new ants spawn after battle code is run? 2011-11-21T21:32:55 *** b0rder has joined #aichallenge 2011-11-21T21:32:58 a valid defense is gathering >= 1 food per turn 2011-11-21T21:33:14 I can make an example scenario and paste it 2011-11-21T21:33:18 McLeopold: this is right yeah, ants move, ants spawn, ants collide, ants battle? 2011-11-21T21:33:58 http://aichallenge.org/problem_description.php move -> attack -> raze -> gather -> spawn 2011-11-21T21:34:12 then you can lose your hill 2011-11-21T21:34:25 yeah, so the next turn you can move the ant that spawned off your hill 2011-11-21T21:34:29 and an enemy can move there at the same time 2011-11-21T21:34:41 if you move the spawned ant off the hill then the hill can be razed before the next ant spawns 2011-11-21T21:34:43 you have to choose to sacrifice your hill, but lots of bots do it 2011-11-21T21:34:56 my battle code wont give the hill up 2011-11-21T21:35:17 an enemy razing my hills is the worst possible outcome 2011-11-21T21:35:20 according to my bot 2011-11-21T21:35:28 followed by an enemy razing an enemy hill 2011-11-21T21:36:56 crud, i guess enemies can be adjacent to your friends too 2011-11-21T21:37:00 antimatroid: I was thinking of counting the number of enemy ants seen and if my bot believes defeat is inevitable, give up my hill to the lesser bot 2011-11-21T21:37:04 i think i made the assumption that they couldn't 2011-11-21T21:38:04 if it went move, spawn, collide, battle then it would be impossible for bots to ever see enemy ants adjacent 2011-11-21T21:38:29 and would make it more worthwhile for people to track the size of their hive 2011-11-21T21:39:11 no, if a hills is spawning ants, it can't be razed 2011-11-21T21:39:32 McLeopold: yes it can :P 2011-11-21T21:40:57 you need to do spawning between moves and collisions if you want it to be impossible to raze hills while the hive has ants 2011-11-21T21:41:17 but it's not hard to make sure your hill can't be razed while you have a hive going on 2011-11-21T21:43:59 McLeopold: do you see why? if you move the newly spawned ant off and killed in battle, then an enemy can step onto your hill and raze it before another ant spawns 2011-11-21T21:45:10 *** yoden has quit IRC (Ping timeout: 265 seconds) 2011-11-21T21:45:14 thestinger: that makes things a bit messy aye :P 2011-11-21T21:45:31 I use dead ants to figure out if they're moving off or not 2011-11-21T21:45:42 yeah i figured as much 2011-11-21T21:45:50 but then you also need to go right move onto the hill you stupid bot 2011-11-21T21:46:22 normally my battle code will assume they'll stay put and presumably just use ants in other locations to take them out while they spawn 2011-11-21T21:46:32 but you need to get the bot to get the razing over and done with when possible 2011-11-21T21:49:59 McLeopold: http://paste.aichallenge.org/xOTnq/ 2011-11-21T21:51:01 my bot knows they are moving off because there aren't any dead ants in the hill tile 2011-11-21T21:51:27 if I just waited next to the hill, they could collide into my ants 2011-11-21T21:51:47 a smart bot could try to trick me by alternating, but that's just the same result as if they moved off every turn and I didn't deal with this 2011-11-21T21:51:50 *** treeform has quit IRC (Remote host closed the connection) 2011-11-21T21:52:01 (a collision each turn until the hill is empty) 2011-11-21T21:52:41 and that would be too risky for a good bot to do 2011-11-21T21:53:20 because lots of bots just try moving onto the hill no matter what (and others wait until it drains) 2011-11-21T21:53:51 it was a stupid little thing to think about but I think my solution to the problem is fine :) 2011-11-21T21:55:27 if they just stay on their hill, I do the standard wait til it drains http://paste.aichallenge.org/19rWQ/ 2011-11-21T21:55:52 antimatroid: ok, if you don't want the hill to be razed, you can prevent it by not stepping off 2011-11-21T21:56:00 *** b0rder has quit IRC (Ping timeout: 260 seconds) 2011-11-21T21:56:19 *** b0rder has joined #aichallenge 2011-11-21T21:56:21 McLeopold: yeah, the problem comes in determining whether an enemy bot sucks and you can raze it sooner or not :) 2011-11-21T21:56:37 well, not a "problem" 2011-11-21T21:56:44 just something that bot writers need to consider 2011-11-21T21:57:05 McLeopold: Of course you run the risk of not stepping off your hill and losing in the natural course of things due to not producing ants :) 2011-11-21T21:57:25 bmh: depends on how many ants they have :P 2011-11-21T21:57:42 if the enemy isn't right next to your hill you probably want to step off 2011-11-21T21:57:47 true :) 2011-11-21T21:57:56 ahhh games 2011-11-21T21:58:04 but then if this happens, or that happens, and then, or that 2011-11-21T21:58:13 *** snthaoeu has quit IRC (Ping timeout: 265 seconds) 2011-11-21T21:59:05 *** bhasker has quit IRC (Quit: bhasker) 2011-11-21T22:00:10 Is anyone doing interesting pathfinding? 2011-11-21T22:01:59 define interesting? 2011-11-21T22:02:09 if i don't have too many targets i use multi source and multi target a* 2011-11-21T22:02:40 *** amstan_ has quit IRC (Ping timeout: 240 seconds) 2011-11-21T22:03:35 *** plr2 has quit IRC (Quit: Page closed) 2011-11-21T22:03:47 *** amstan has joined #aichallenge 2011-11-21T22:03:47 *** ChanServ sets mode: +o amstan 2011-11-21T22:05:19 bmh: oh, you asked me a question before 2011-11-21T22:05:30 the bot I have uploaded on the site has no combat 2011-11-21T22:05:31 *** scribble has joined #aichallenge 2011-11-21T22:06:03 thestinger: It just gathers food like mad and actually does well? 2011-11-21T22:06:07 yeah 2011-11-21T22:06:10 it's 2 weeks old 2011-11-21T22:06:26 it gets crushed in a match with my new versions :) 2011-11-21T22:06:36 I think it will drop down to around rank 50 now 2011-11-21T22:06:39 I'll reupload soon 2011-11-21T22:06:47 perhaps I should submit the diffusion bot we had working. It's truly terrible. Loses to HunterBot on open maps. 2011-11-21T22:06:50 I was really surprised that it made it up to rank 15 like that though 2011-11-21T22:07:00 just by rushing early on 2011-11-21T22:08:20 my old version on the site just tries to pathfind around ants (diffusion I guess, but it was really primitive and broken) 2011-11-21T22:09:07 sometimes that makes it strafe around an enemy, but it's usually a negative thing when it comes to combat (it will turn a favourable 2v1 or 3v1 into 1v1, etc.) 2011-11-21T22:09:26 I did realize that if you spot an enemy ant on the nth turn you know that their hill is within n steps of that point. 2011-11-21T22:10:08 *** rajanaresh has quit IRC (Ping timeout: 244 seconds) 2011-11-21T22:10:24 bmh: http://paste.pocoo.org/show/bna62T0hnIWriF9VlKiL/ diff to the version I have on the site (I tag the commit when I upload) 2011-11-21T22:10:26 :) 2011-11-21T22:10:37 pretty much a completely different codebase now 2011-11-21T22:10:46 *** snthaoeu has joined #aichallenge 2011-11-21T22:11:02 I haven't touched my generic BFS, A*, etc. (I don't actually use A* anymore) 2011-11-21T22:11:16 but the bot logic is all in MyBot.cpp, and that's all changed 2011-11-21T22:19:41 *** yoden-cloud has joined #aichallenge 2011-11-21T22:25:36 bmh: nice observation :) 2011-11-21T22:31:48 *** trishume has quit IRC (Remote host closed the connection) 2011-11-21T22:33:59 *** yoden-cloud has quit IRC (Ping timeout: 265 seconds) 2011-11-21T22:34:40 *** JorgeB has quit IRC (Ping timeout: 258 seconds) 2011-11-21T22:35:56 *** zacharyd1nton has joined #aichallenge 2011-11-21T22:36:27 *** zacharydenton has quit IRC (Ping timeout: 276 seconds) 2011-11-21T22:48:29 *** Euphony has quit IRC (Ping timeout: 265 seconds) 2011-11-21T22:52:00 *** Jak_o_Shadows1 has joined #aichallenge 2011-11-21T22:53:24 *** treeform has joined #aichallenge 2011-11-21T22:54:03 *** Jak_o_Shadows has quit IRC (Ping timeout: 244 seconds) 2011-11-21T23:09:26 this map is just crap: map 2011-11-21T23:09:35 er http://ants.fluxid.pl/map/maze_04p_01.map 2011-11-21T23:10:15 *** u_ has quit IRC (Quit: u_) 2011-11-21T23:10:50 a1k0n: what's crap about it? 2011-11-21T23:11:03 it's utter chaos 2011-11-21T23:11:10 too many hills? 2011-11-21T23:11:13 not my problem 2011-11-21T23:11:16 anything else and hmph :P 2011-11-21T23:11:28 you're better off abandoning two of your hills and defending only one of them 2011-11-21T23:11:40 cuz you're only 10 moves away from enemy hills 2011-11-21T23:11:50 plus you can probably infer the entire map on the first turn 2011-11-21T23:12:11 *** avdg has quit IRC (Quit: Leaving.) 2011-11-21T23:12:20 so main problem is too many hills too close? 2011-11-21T23:12:21 grumpy a1k0n is grumpy. 2011-11-21T23:13:34 yes 2011-11-21T23:13:45 *** avdg has joined #aichallenge 2011-11-21T23:13:49 also my hill defense code seems to be broken anyway 2011-11-21T23:13:55 not my problem :P 2011-11-21T23:14:08 i have a dude defending the hill, who all of a sudden just goes "fuck it" and wanders away 2011-11-21T23:14:29 i will not tolerate disloyalty! 2011-11-21T23:14:30 what do you do for hill defense? 2011-11-21T23:14:43 nothing special, just try to keep at least one and closer to the hill than any enemy ants 2011-11-21T23:14:44 run away :p 2011-11-21T23:14:44 antimatroid: a1k0n casts bones. 2011-11-21T23:14:56 s/and/ant/ 2011-11-21T23:15:18 but yes, bone-casting is somewhat involved so it's possible i rolled a critical miss 2011-11-21T23:15:37 i was thinking of for any enemy ants within x distance of the hill make sure there is a unique friend that can get to my hill as fast as they could 2011-11-21T23:16:03 then possibly even trying to get them to meet them as soon as possible on one of the enemies shortest paths to my hill 2011-11-21T23:16:15 you only need them to be able to get within the battle radius too really 2011-11-21T23:16:24 not necessarily the hill at the same time 2011-11-21T23:16:27 we should modify truskill to infer a sigma factor for each map 2011-11-21T23:16:44 people wanted to evaluate the fairness of maps 2011-11-21T23:16:51 but imo that's just getting ridiculously messy 2011-11-21T23:16:55 i think they're all *fair* 2011-11-21T23:17:05 some are just more noisy than others 2011-11-21T23:17:07 yeah this was back when people wanted asymmetric maps 2011-11-21T23:17:22 and they're not all fair when you drop the assumption of equally skilled opponents 2011-11-21T23:17:44 because you don't have symmetrically equal access to each of your enemies 2011-11-21T23:17:51 true, yes 2011-11-21T23:18:08 you can get that for 5 players on the surface of a continuous torus 2011-11-21T23:18:13 not sure about discrete 2011-11-21T23:18:17 3 is obviously easy 2011-11-21T23:18:20 *** roflmao has quit IRC (Quit: Leaving.) 2011-11-21T23:18:35 but 4 is problematic 2011-11-21T23:18:36 in the finals it seems that they should run all permutations. 2011-11-21T23:18:49 yeah you wont get 4 2011-11-21T23:19:18 but you can tile the torus with 5 "squares" where each oppoent is an adjacent square 2011-11-21T23:19:39 i'm not sure if you can get it to match up properly in the discrete case, someone on #math pointed out the 5 player case to me 2011-11-21T23:20:10 bmh: incredibly inefficient use of server time 2011-11-21T23:21:23 antimatroid: Does the bot with the highest skill win, or is it tournament style? 2011-11-21T23:21:45 highest skill 2011-11-21T23:21:55 although all skills etc. are reset after entries close 2011-11-21T23:22:10 Does skill have a unit? 2011-11-21T23:22:22 and new maps are generated after entries close for the final contest too 2011-11-21T23:22:25 what do you mean? 2011-11-21T23:22:31 it's just what you see on the current rankings now 2011-11-21T23:22:51 it'll just run it like it is now but with a closed set of bots that all start with 0 games 2011-11-21T23:23:25 possibly "freezing" the lower ranked bots as time progresses to make sure the rankings near the top are better 2011-11-21T23:23:49 antimatroid: Does 'skill' have some meaning, like Elo? 2011-11-21T23:24:00 i don't know much about rankings sorry 2011-11-21T23:24:13 bmh: https://research.microsoft.com/en-us/projects/trueskill/details.aspx 2011-11-21T23:24:14 all i know is trueskill is somewhat like elo but allows us to rank n-player games :P 2011-11-21T23:24:30 and that we're only considering the ordinal rankings in each game for the overall rankings 2011-11-21T23:24:57 which is good because then it's score system independent 2011-11-21T23:25:07 well, assuming score system is ordinally equivalent 2011-11-21T23:25:47 i really hate that the euclidean and manhatten metrics aren't equivalent :P 2011-11-21T23:26:02 antimatroid: amen to that. 2011-11-21T23:27:14 Has anyone started thinking about the *next* competition? 2011-11-21T23:27:23 what do you mean? 2011-11-21T23:27:30 we want to rewrite the entire contest stuff 2011-11-21T23:27:34 A problem 2011-11-21T23:27:41 i don't think we could bear to just do it on the existing codebase 2011-11-21T23:27:58 there have been a few game ideas thrown around in the forums 2011-11-21T23:28:42 what we ideally want is a site that can handle multiple pages and official tcp pages 2011-11-21T23:28:57 then there could be tcp servers going to heaps of different games 2011-11-21T23:29:12 i really really look forward to writing a tron bot again 2011-11-21T23:29:43 plus we're always short on people to help out 2011-11-21T23:30:33 * antimatroid gets back to coding 2011-11-21T23:37:07 *** seletskiy has joined #aichallenge 2011-11-21T23:40:09 *** ChrisH_ has joined #aichallenge 2011-11-21T23:42:24 haha: http://ants.fluxid.pl/replay.9907 2011-11-21T23:43:19 :) 2011-11-21T23:43:55 thestinger: My personal paranoid theory is that the words after strcat_* are totally misleadign 2011-11-21T23:44:10 lol 2011-11-21T23:44:14 haha 2011-11-21T23:44:23 I am pretty lazy with some of the names 2011-11-21T23:44:29 strcat_dishwasher 2011-11-21T23:44:58 strcat_friendly_i_swear 2011-11-21T23:45:02 I reuse some of them for slightly different things sometimes :) 2011-11-21T23:45:12 strcat_comes_in_peace 2011-11-21T23:45:14 i like being anonymous on tcp 2011-11-21T23:45:50 i was gonna do that but couldn't help myself 2011-11-21T23:46:10 haha because it was doing so well? :P 2011-11-21T23:46:25 yeah 2011-11-21T23:46:32 sometimes I have 5 bots running so I need the suffixes to figure out what the hell is going on 2011-11-21T23:46:33 it was 'dumb' for a while 2011-11-21T23:46:53 i'll start running my next version as strcat_notstrcat 2011-11-21T23:47:05 haha 2011-11-21T23:47:12 I'll be strncat, it's safer 2011-11-21T23:47:24 good idea 2011-11-21T23:47:37 win. 2011-11-21T23:47:38 wait, my bot is C++, I'll have to be std::string::operator+ 2011-11-21T23:49:04 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-11-21T23:49:23 It's a shame bots don't get the names of their opponents 2011-11-21T23:49:28 strcat was taken on youtube so I used strlcat 2011-11-21T23:49:34 then I could be "DROP * FROM ANTS;" 2011-11-21T23:51:04 you do at the end 2011-11-21T23:52:01 meh, I'll have to make my foragers cautious 2011-11-21T23:52:16 I like aggressive foragers on 2-player maps though 2011-11-21T23:53:48 I guess I can count opponents as I see them... 2011-11-21T23:58:14 *** scribble has quit IRC (Ping timeout: 258 seconds) 2011-11-21T23:59:13 BenJackson: Is there any limit to bot name length? :) 2011-11-21T23:59:19 *** ChrisH_ has quit IRC (Quit: Page closed)