2011-06-24T00:00:29 *** nullkuhl has joined #aichallenge 2011-06-24T00:01:01 *** nullkuhl_ has joined #aichallenge 2011-06-24T00:02:25 *** nullkuhl__ has joined #aichallenge 2011-06-24T00:05:29 *** nullkuhl_ has quit IRC (Ping timeout: 244 seconds) 2011-06-24T00:11:37 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-06-24T00:21:13 *** stocha has joined #aichallenge 2011-06-24T00:21:28 There is a problem in the way timeout are handle at the moment. (http://aichallengebeta.hypertriangle.com/visualizer.php?game=35074&user=226&turn=24) 2011-06-24T00:21:56 Timeout bot should rather be normal bot with the don't move order for every ant. 2011-06-24T00:22:01 than what is done now. 2011-06-24T00:22:42 (it also totally screw the stat on the bot page, the linked game is said to be decided at turn 29. Which is certainly wrong when you take into account the timeout) 2011-06-24T00:23:35 *** stocha has quit IRC (Client Quit) 2011-06-24T00:36:33 *** nullkuhl__ has quit IRC (Remote host closed the connection) 2011-06-24T00:44:02 *** eggs_ has quit IRC (Quit: Page closed) 2011-06-24T00:54:04 *** McLeopold has joined #aichallenge 2011-06-24T00:54:23 antimatroid-uni: so, you around? 2011-06-24T01:06:25 janzert: with our new cutoff code, I'm inclined to remove the food bonus 2011-06-24T01:43:30 *** McLeopold has left #aichallenge 2011-06-24T01:49:11 *** Chris_0076 has joined #aichallenge 2011-06-24T01:49:45 *** Chris_0076 has quit IRC (Client Quit) 2011-06-24T02:10:52 *** sigh has quit IRC (Ping timeout: 250 seconds) 2011-06-24T02:16:21 *** sigh has joined #aichallenge 2011-06-24T02:29:40 *** sigh has quit IRC (Ping timeout: 260 seconds) 2011-06-24T02:32:22 *** stocha has joined #aichallenge 2011-06-24T02:41:05 what is a bot "without matching opponent" ? There are always opponents arround any bot. If you use the ranking. Event the highest and lowest do have neighbors. 2011-06-24T02:42:53 You could have a bias that choose the best player a litle more often to offset that he has less neighbors. 2011-06-24T02:43:16 *** stocha has quit IRC (Quit: Page closed) 2011-06-24T03:01:20 *** gobang has joined #aichallenge 2011-06-24T03:10:05 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-24T03:10:58 *** gobang has joined #aichallenge 2011-06-24T03:18:36 *** aerique has joined #aichallenge 2011-06-24T03:29:36 *** amstan has quit IRC (Ping timeout: 255 seconds) 2011-06-24T03:29:47 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-06-24T03:50:22 @later tell McLeopold I'm still not clear whether we're using Python 2.6 or 2.7 to run playgame.py. 2011-06-24T03:50:22 aerique: Aye, aye, sir 2011-06-24T03:56:47 * antimatroid is writing a new map gen with more symmetries 2011-06-24T04:08:30 *** okayzed is now known as okay 2011-06-24T04:10:58 *** Palmik has joined #aichallenge 2011-06-24T04:28:56 *** okay is now known as okayzed 2011-06-24T04:41:40 *** mcstar has joined #aichallenge 2011-06-24T05:03:49 *** Accoun has quit IRC (Read error: Connection reset by peer) 2011-06-24T05:08:09 *** Accoun has joined #aichallenge 2011-06-24T05:39:49 so there is one way to do a "rotational" symmetry on a square grid with 8 players, 3 ways with 4 players and 7 ways with 2 players 2011-06-24T05:41:26 *** foRei has joined #aichallenge 2011-06-24T05:44:09 *** Accoun has quit IRC (Read error: Connection reset by peer) 2011-06-24T05:45:32 *** Palmik_ has joined #aichallenge 2011-06-24T05:45:51 *** Palmik has quit IRC (Read error: Connection reset by peer) 2011-06-24T05:58:18 *** choas has joined #aichallenge 2011-06-24T06:03:22 *** UncleVasya has joined #aichallenge 2011-06-24T06:03:46 lol: Your next game should be in about 13.233561645805967348000000000000 minutes. 2011-06-24T06:04:14 i like that everyone feels the need to comment on that :P 2011-06-24T06:05:04 *** choas has quit IRC (Quit: leaving) 2011-06-24T06:05:26 *** choas has joined #aichallenge 2011-06-24T06:05:49 4 of the types of rotational symmetry for 2 players require square maps 2011-06-24T06:05:54 3 can be rectangles 2011-06-24T06:06:13 happy midsummer festival for all of you! 2011-06-24T06:06:23 2 need to be squares for 4 player, 1 can be a rectangle, and needs to be a square for the one type of 8 player rotational symmetry 2011-06-24T06:07:23 antimatroid: going to read history and find another comments :) 2011-06-24T06:07:40 UncleVasya: (19:39:31) antimatroid: so there is one way to do a "rotational" symmetry on a square grid with 8 players, 3 ways with 4 players and 7 ways with 2 players 2011-06-24T06:09:47 antimatroid: I thought you'd answered on my comment about that big number. 2011-06-24T06:12:54 distance function is hard if player is very near but actually other side of the map 2011-06-24T06:13:12 my a* is crashing over there 2011-06-24T06:13:54 Rabidus: what do you mean? 2011-06-24T06:14:06 use the distance function given on the specs page 2011-06-24T06:14:16 although it can be worthwhile squaring it 2011-06-24T06:14:33 actually, you probably want to use manhatten for pathfinding 2011-06-24T06:14:36 like food coordinate is 1,1 and ant is at 1,195 (when map size is 300,300) 2011-06-24T06:14:46 nono he means to wrap on the edges 2011-06-24T06:14:50 i meant 200 * 200 2011-06-24T06:15:01 yes 2011-06-24T06:15:04 that's what i ment 2011-06-24T06:15:16 im educated myself in A* finally 2011-06-24T06:15:19 i really like it 2011-06-24T06:15:29 but i dont want to use graph searching at all 2011-06-24T06:15:40 Rabidus: http://pastebin.com/SCrz7kHH 2011-06-24T06:15:56 mcstar: how to you plan to find food? 2011-06-24T06:16:04 bfs is also graph searching 2011-06-24T06:16:09 i know 2011-06-24T06:16:25 i try to precompute some things and rely on geometric considerations 2011-06-24T06:17:15 bad idea :P 2011-06-24T06:17:25 thanks 2011-06-24T06:17:32 i use a* with multiple sources and multiple targets 2011-06-24T06:18:14 i hate this actually: i look at the map, i immediately know what goes where, why doesnt the computer too??? 2011-06-24T06:18:41 so i say to myself, come on, this can be done without searching, just use a proper representation 2011-06-24T06:19:02 path finding is a pretty common problem :P 2011-06-24T06:19:16 thank you Mr. Obvious :D 2011-06-24T06:19:16 do you think bfs and a* would be so common if they didn't work so well for the given problem? :P 2011-06-24T06:20:49 it cant hurt to view things from a different angle 2011-06-24T06:21:10 ok, thats not always true 2011-06-24T06:22:24 i'm not trying to be an ass, honest, just trying to make you see the light :) 2011-06-24T06:22:43 *** onensora has joined #aichallenge 2011-06-24T06:22:49 i guess i should consider using cpp too, right? 2011-06-24T06:22:58 i heard it is fast and glorious 2011-06-24T06:23:03 no, i just try to sway people on the edge there :P 2011-06-24T06:23:04 cause it's fun 2011-06-24T06:23:12 also people using java 2011-06-24T07:02:16 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-24T07:13:57 *** Zaphus has joined #aichallenge 2011-06-24T07:30:54 my next game will be in 12.464539009106156632000000000000 seconds, wait only 13.85829228413217986300000000000 2011-06-24T07:30:57 wth? 2011-06-24T07:30:58 *** Accoun has joined #aichallenge 2011-06-24T07:31:00 Oh and D is also working at a low level so you can optimize your ... innovative new represantation of the map that enables the computer to 'see' where ants must go 2011-06-24T07:31:42 ... innovative 2011-06-24T07:31:48 nice adjective 2011-06-24T07:33:11 of course I'm going to precompute everything which is even better :P 2011-06-24T07:33:37 yeah, sure, but you CANT precompute everything 2011-06-24T07:34:01 and you cant find a better representation -.- 2011-06-24T07:34:09 hehe 2011-06-24T07:34:09 *** Kingpin13 has joined #aichallenge 2011-06-24T07:34:16 lol 2011-06-24T07:34:22 there is always a better one 2011-06-24T07:34:23 i have 3 seconds for my precalculation 2011-06-24T07:34:32 as long as people's opinions differ 2011-06-24T07:34:46 that's true 2011-06-24T07:35:07 i we were the borg, everything would be nice, and we would have much less standards 2011-06-24T07:35:09 oh shit, I don't know the map from the beginning 2011-06-24T07:35:16 hehe 2011-06-24T07:35:20 only part of it 2011-06-24T07:35:28 didnt you know that? 2011-06-24T07:35:36 I forgot for a second 2011-06-24T07:35:45 i thought about pre-computing "highways" 2011-06-24T07:35:54 but that would have to be incremental too 2011-06-24T07:36:15 Yes, I'll incrementally precalculate everything if that makes any sense 2011-06-24T07:36:38 but still, pre-computing individual paths is not feasible 2011-06-24T07:37:34 only, if you really can find a way to make the ai "know" about obstacles, and do a representation of circle-graphs 2011-06-24T07:38:08 so you can make an ant join, or to make it go in the vicinity of a highway 2011-06-24T07:38:39 and then, when the destination is reached within a specified perimeter, it will deviate from it 2011-06-24T07:38:47 of course this is a bit too general 2011-06-24T07:39:12 because usually you dont want to send the ants to ANY distance, only in its viewradius 2011-06-24T07:39:44 thats one reason i dont want to fully implement an A* for example, i think its an overkill 2011-06-24T07:44:46 please, dont say anything, really 2011-06-24T07:45:02 im not interested, thats why i wrote 7 lines 2011-06-24T07:45:05 for the fun of it 2011-06-24T07:45:15 why not i say, i like typing 2011-06-24T07:45:22 it improves sexual performance 2011-06-24T07:45:33 or is it running, i cant recall 2011-06-24T07:46:27 mcstar: I was afk, no need to feel like an abandonned child, haha 2011-06-24T07:46:51 i was afraid, daddy 2011-06-24T07:47:05 I wonder if there isn't any case where you want to send an ant into the unexplored area of the map. 2011-06-24T07:47:27 but maybe you could just as well send it to the border of the view radius 2011-06-24T07:48:38 yeah, and again and again if nothing interesting is in view 2011-06-24T07:49:09 Anyway it may be a bad idea to cripple the pathfinding early on. Later you may find a use for longer routes and then you aren't motivated to go back to pathfinding 2011-06-24T07:49:20 but usually this is not the case, the map tends to be crowded 2011-06-24T07:50:11 actually i dont intend to cripple it 2011-06-24T07:50:15 in that case the focus should be on collision avoidance and optimal planning of what ant goes first 2011-06-24T07:50:25 it would remain stable and robust 2011-06-24T07:50:39 only difference is that it might not find the absolute shortest route 2011-06-24T07:51:29 so what else do you plan to precompute? 2011-06-24T07:52:05 nothing, I'm just trying to cache whatever is feasible 2011-06-24T07:52:20 i use the start up time to make my grid into a graph 2011-06-24T07:52:32 then as i receive water information i remove neighbour information from squares 2011-06-24T07:52:48 antimatroid: a struct with 4 legs? 2011-06-24T07:52:49 I estimate the map size and memory requirements for the data structures and if it is ~100 MB it can be done :p 2011-06-24T07:53:01 and an array of those of course? 2011-06-24T07:53:44 cause i beleive if i really wantod represent it as a "graph" i would use binary number system 2011-06-24T07:53:56 ____ for NWSE 2011-06-24T07:54:15 _ stands for 1,0 2011-06-24T07:54:33 i have have to grab the neighbour locations from the squares vector 2011-06-24T07:54:45 no need to use nsew 2011-06-24T07:54:48 *** FireFly has joined #aichallenge 2011-06-24T07:55:15 well, an integer would represent the conenctivity 2011-06-24T07:55:29 can be faster than always fetching nbors, and checking their values 2011-06-24T07:56:16 squares vector? you mean that you represent the board in 1D as me? 2011-06-24T07:56:39 no, i have a grid which is a 2-dimensional vector of Square structs 2011-06-24T07:56:51 then each square struct has a vector of Locations to neighbours that aren't water 2011-06-24T07:57:04 lol, isnt it an array? in cs a vector is always 1d 2011-06-24T07:57:26 yes, but i still have it as a 2-dimensional array, it's just easier 2011-06-24T07:57:28 antimatroid: I initially used a 'normal' distance function for A*, but using manhattan distance gave better results (less iterations to find the same route) 2011-06-24T07:57:40 i could probably change that actually, because I have index functions that would do it for me 2011-06-24T07:57:52 aerique: i use manhattne 2011-06-24T07:58:14 i have functions to switch: row,col <-> index 2011-06-24T07:58:17 aerique: do you use multiple targets? 2011-06-24T07:58:27 antimatroid: right, this was in reply to something you said at 12:14 :) 2011-06-24T07:58:28 mcstar: me too *g* 2011-06-24T07:58:45 *** choas has quit IRC (Ping timeout: 244 seconds) 2011-06-24T07:59:10 antimatroid: no, just running a* multiple times to multiple food sources and sorting the path lengths 2011-06-24T07:59:22 mleise: surely n-dimensional vectors in c++/D are just implemented in 1d? 2011-06-24T07:59:30 aerique: sounds messy :P 2011-06-24T07:59:47 euclidian distance is very different than manhattan, even more for large distances 2011-06-24T07:59:59 aerique: http://pastebin.com/Z1a9T7wu :P 2011-06-24T08:00:06 ignore the comment :P 2011-06-24T08:00:11 its nice that a rectangle has all the equally short paths 2011-06-24T08:00:32 that's my heuristic function for a* with multiple targets 2011-06-24T08:01:01 antimatroid: depends. static arrays of byte[256][256] are 1-D, but dynamic arrays of byte[][] are really 2-D 2011-06-24T08:02:07 antimatroid: cheers :) 2011-06-24T08:03:07 mcstar: you just present all the thoughts I had yesterday during a walk through the park 2011-06-24T08:03:21 :) 2011-06-24T08:03:32 i have themfor some time 2011-06-24T08:04:50 aerique: i'm not positive it's faster doing this than just bfs with that heuristic function 2011-06-24T08:05:07 I also don't think I'll use the multiple targets idea for finding the path to an enemy ant or food. I can think of something slightly faster, hmm... if it works? 2011-06-24T08:05:17 the search path is much smaller, but a bfs is probably faster if you have lots of food with some close to ants 2011-06-24T08:05:33 mleise: you can do it with bfs fairly easily 2011-06-24T08:05:59 with a* the heuristic function can be painful, you have to call it for every location added to your search queue 2011-06-24T08:05:59 nah, everyone uses it, i need something else,just because 2011-06-24T08:06:16 mleise: exactly 2011-06-24T08:06:54 antimatroid: the heuristic function is very cheap 2011-06-24T08:06:56 antimatroid: have you tried weighted a* yet? (http://en.wikipedia.org/wiki/A*#Weighted_A.2A) 2011-06-24T08:07:02 why not call it every time? 2011-06-24T08:07:21 maybe optimize it a bit, but its not a problem i guess 2011-06-24T08:07:21 antimatroid: gives almost the same routes but much less iterations 2011-06-24T08:07:52 aerique: i don't understand what weighted a* is? (i read the link...) 2011-06-24T08:08:18 aerique: still, it is for hard-maps and it is very general 2011-06-24T08:08:41 antimatroid: you just multiply your h by a value > 1 2011-06-24T08:08:53 aerique: who the hell would i do that? :P 2011-06-24T08:09:00 i have had a different idea for path finding 2011-06-24T08:09:14 Oo 2011-06-24T08:09:28 antimatroid: you didn't read the link then :) 2011-06-24T08:09:40 anyway, back to work for me 2011-06-24T08:09:41 make information about the shortest number of turns an enemy ant could get in battle range of each land square, then use that to disclude squares while searching to avoid enemies 2011-06-24T08:11:16 lol, thats a lot of calculation isnt it? 2011-06-24T08:11:49 just consider that possibly, you can see more enemies than your own ants 2011-06-24T08:11:59 it's one pass of the entire grid to get information like that 2011-06-24T08:12:23 but you want to keep that info up to date i guess 2011-06-24T08:12:27 you have to do it again 2011-06-24T08:12:32 yeah, you would have to do it once per turn 2011-06-24T08:12:33 fairly cheap 2011-06-24T08:12:54 idk 2011-06-24T08:13:19 compared to my iterative path finding to collect food and enemies, that is almost free 2011-06-24T08:14:09 you know, the problem is that it would not be stable 2011-06-24T08:14:30 what do you mean? 2011-06-24T08:14:43 consider that for every turn you precompute your path to a target, but it can change when an enemy ant changes position 2011-06-24T08:14:48 for a given turn, i can calculate the information based off the enemy ants i see then use that information with my path finding 2011-06-24T08:15:00 i don't store information across turns 2011-06-24T08:15:05 it can happen that your ant cant monotonically decrease its distance from the start 2011-06-24T08:15:06 i recalculate everything from scratch 2011-06-24T08:17:15 oh, and the new viewradius isnt that much larger than the battleradius 2011-06-24T08:17:21 right? 2011-06-24T08:17:24 what? 2011-06-24T08:17:34 attack2=5 and view2 = 77 yeah? 2011-06-24T08:17:52 77? 2011-06-24T08:17:55 it was 55 2011-06-24T08:17:58 yeah 2011-06-24T08:18:00 then it went up 2011-06-24T08:18:07 aaa who can follow this.... 2011-06-24T08:18:16 it's beta 2011-06-24T08:18:21 we're trying to work out what works best 2011-06-24T08:18:27 aaa i dont care aaaa 2011-06-24T08:18:30 :D 2011-06-24T08:20:28 shat, there are soo many variants of these pathf. algs. 2011-06-24T08:34:19 *** choas has joined #aichallenge 2011-06-24T08:37:43 ever considered stopping food after N turns, so it stops being area control and turns into a fight ? 2011-06-24T08:58:04 Zaphus: i had not 2011-06-24T08:59:05 i kind of like the idea of two cutoff points 2011-06-24T08:59:15 one where food spawning stops then another where the game is actually called off 2011-06-24T08:59:24 yep 2011-06-24T08:59:56 just my observation that games are area control right now 2011-06-24T09:00:27 there's no battle bots :P 2011-06-24T09:00:35 well sir_macelon and flag, but they're clearly winning :P 2011-06-24T09:01:26 or are you battling as well now? 2011-06-24T09:05:13 not yet, except when the stars align 2011-06-24T09:06:29 i'm working on a new map generator 2011-06-24T09:06:37 it's going to do rotational symmetry too 2011-06-24T09:07:04 get an empty map in there ! 2011-06-24T09:07:20 i've made maxLandProportion 0.99 2011-06-24T09:07:24 i can make it 1.0 2011-06-24T09:07:33 it'd still be quite unlikely i expect 2011-06-24T09:07:41 .99 is effectively empty 2011-06-24T09:09:36 zelark has improved 2011-06-24T09:10:41 wish we could put the snake, blobby and other 'interesting' bots into the official contest for people to see - but I'm not wastng my one login on them 2011-06-24T09:10:58 no code, just an entry 2011-06-24T09:19:25 *** gobang has joined #aichallenge 2011-06-24T09:22:26 *** Zaphus has quit IRC (Quit: Page closed) 2011-06-24T09:59:51 symmetric_2 in technicolor http://imgur.com/scUDd 2011-06-24T10:17:25 *** aerique has quit IRC (Quit: ...) 2011-06-24T10:31:51 *** davidjliu has joined #aichallenge 2011-06-24T10:41:51 *** foRei has quit IRC (Quit: Bye) 2011-06-24T10:41:57 *** dr- has quit IRC (Quit: WeeChat 0.3.0) 2011-06-24T10:44:55 *** Kingpin13 has quit IRC (Quit: this used to be the life, but I don't need another one) 2011-06-24T10:56:16 i have ze ultimate map generator 2011-06-24T10:57:51 or zo you zinnk, Herr Cat 2011-06-24T10:58:34 *** foRei has joined #aichallenge 2011-06-24T11:06:55 bah, the only problem i have left is that rotational maps seem to hang on not getting fully connected :S 2011-06-24T11:08:13 antimatroid: can you write a map editor for the web? 2011-06-24T11:09:13 antimatroid is turing complete, so yes, he can do it 2011-06-24T11:09:47 hmm so can css3 then 2011-06-24T11:10:02 yeah, i heard 2011-06-24T11:10:21 i meant this as a joke, turing completeness is overused, and isnt really funny anymore :( 2011-06-24T11:11:07 but i never really programmed in antimatroid, i beleive its a dialect of cpp 2011-06-24T11:11:50 oh is it better than cx00? 2011-06-24T11:12:35 sure, it has the capability of overloading everything, like >> or << 2011-06-24T11:12:38 does antimatroid work with templates? 2011-06-24T11:12:52 it gives an unprecedented flexibitlity to antimatroid 2011-06-24T11:12:57 ah yes and switch operator precedence as well 2011-06-24T11:12:58 yes sure that too 2011-06-24T11:13:02 mleise: i do use templates 2011-06-24T11:13:06 i don't know how to do web sutff 2011-06-24T11:13:32 antimatroid doesnt have those libraries it seems, in time it will evolve 2011-06-24T11:13:43 hehe 2011-06-24T11:16:50 *** stocha has joined #aichallenge 2011-06-24T11:17:57 mcstar: good idea. Everything original is always worth trying. Especially if it doesn't take too much time, OR if it's satisfying :) 2011-06-24T11:18:30 guten morgen dear stocha 2011-06-24T11:18:38 to what exactly are you referring to? 2011-06-24T11:18:39 ? he is french 2011-06-24T11:18:44 so? 2011-06-24T11:18:50 bon apetit then 2011-06-24T11:18:52 mleise: yeah, sorry about htat :) 2011-06-24T11:18:58 haha 2011-06-24T11:19:42 mcstar: mm, i don't know. I think you were saying you don't want to do this (path finding ?) and you want to do that (highway precalculated incrementally ?). And someone would tell you that pathfinding is important or so. 2011-06-24T11:19:47 once queens of the stoneage played in a studio in the netherlands and said "diamonds in scheisse" 2011-06-24T11:19:53 i didn't read it very carfully really :) 2011-06-24T11:20:02 ah the logs 2011-06-24T11:20:07 you read them? nice 2011-06-24T11:20:20 mcstar: alles goed. 2011-06-24T11:20:28 gut 2011-06-24T11:20:47 mcstar: i'm excited because i uploaded my new prototype. 2011-06-24T11:21:01 it's a slow bot if there is one. but still i want to see how things go with it :p 2011-06-24T11:21:28 whats the improvement? 2011-06-24T11:21:29 (more than 30 time slower than my fast ref - running under stoha02 at the moment) 2011-06-24T11:21:31 is it written in jave 2011-06-24T11:21:51 mcstar: it the trying of a brand new stuff i didn't have in my framework before. 2011-06-24T11:22:09 stocha improved his bot: now its 30 times slower, to give chance to the enemies :D 2011-06-24T11:22:14 mleise: yeah. now that java works, i' probably stick to it. unless i really have time on my hands :p 2011-06-24T11:22:25 mcstar: it's slower thinking :) 2011-06-24T11:22:51 remember to wash your hands if you have time onthem 2011-06-24T11:22:55 mcstar: i hope it'll be slightly stronger than my ref greedy. 2011-06-24T11:23:09 i didn't try it at home. i just wanted to see how it does on the real thing. 2011-06-24T11:23:40 then i'll also see how fast it goes on the server, and maps. 2011-06-24T11:23:48 because there an embeded time monitor. 2011-06-24T11:24:44 in about 5.3463 minutes. i'll go on an errand on come back in a few minute then. 2011-06-24T11:25:02 hey it's really exciting when you upload a bot ! 2011-06-24T11:25:12 :) 2011-06-24T11:25:22 so you basically want to test timeouts? 2011-06-24T11:26:49 *** sigh has joined #aichallenge 2011-06-24T11:28:15 mcstar: i have a security built in. If it take more than 300, it will try to do less. 2011-06-24T11:28:18 300ms 2011-06-24T11:29:16 from some test game i ran at home, i realize i should have added some more feature :p lool. I don't care, Yeah !!!! (it's been like a week i had wanted to see this new part of my framework working. And it does seem to work - slowly but works !) 2011-06-24T11:29:30 the bot is pretty raw. 2011-06-24T11:29:49 where are you now regarding sloc? 2011-06-24T11:31:07 mcstar: sloc ? 2011-06-24T11:31:07 common abbreviation: source line of code ( i dont like it but people use it) 2011-06-24T11:31:34 mcstar: i don't understand the question. 2011-06-24T11:31:48 mm, i should add ennemy ants avoidance probably 2011-06-24T11:32:03 how many lines do you have in all of your sorces that relates to the aichallenge bots? 2011-06-24T11:32:12 sources 2011-06-24T11:35:46 everything works except one of the types of symmetry :\ 2011-06-24T11:36:56 stocha: how many? im interested in knowing how big your bot is 2011-06-24T11:37:04 my loc went up since I started putting { on a new line :D 2011-06-24T11:37:05 http://pastebin.com/a0ebz8CE 2011-06-24T11:37:09 there's some rotational maps 2011-06-24T11:37:09 gimme an estimate 2011-06-24T11:37:19 some might not be symmetric as i mentioned above 2011-06-24T11:37:27 mleise: cheap trick 2011-06-24T11:37:33 *** astevenson has joined #aichallenge 2011-06-24T11:37:51 of course in lisp that is totally unacceptable 2011-06-24T11:38:16 to write opening/closing parantheses on new lines 2011-06-24T11:38:25 i can imagine 2011-06-24T11:39:12 why not count bytes. that seems to be as accurate 2011-06-24T11:39:43 our functions and variables tend to have long names :) 2011-06-24T11:42:14 hm, i think stocha started to count the lines by hand 2011-06-24T11:43:03 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-24T11:44:15 *** demerzel has joined #aichallenge 2011-06-24T11:46:21 *** sigh has joined #aichallenge 2011-06-24T11:48:40 *** choas has quit IRC (Ping timeout: 244 seconds) 2011-06-24T11:48:58 wc -l stocha! 2011-06-24T11:49:23 stocha is on the wc 2011-06-24T11:51:33 contestbot_: later tell McLeopold I'm still rather inclined to do away with scoring altogether and go strictly by inverse order of elimination ;) 2011-06-24T11:51:33 janzert: I come to serve. 2011-06-24T11:51:57 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-06-24T11:55:20 so all you have to do is survive? 2011-06-24T11:55:55 *** nux67 has joined #aichallenge 2011-06-24T11:57:32 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-24T12:03:10 *** sigh has joined #aichallenge 2011-06-24T12:06:05 yes, although I'm being rather facetious there 2011-06-24T12:07:04 janzert: I have new maps :) 2011-06-24T12:07:26 the last thing i need to do is be able to check that an nxn block can reach every land square 2011-06-24T12:07:33 great, I saw you had rotational symmetry working 2011-06-24T12:07:56 making some 'maze' maps too? 2011-06-24T12:08:05 i thought i'd be able to do 7 types of rotational symmetry with 2 players 2011-06-24T12:08:19 but two of the types don't seem to work with the way everything else does :\ 2011-06-24T12:08:35 <_flag> Will there even be games with two players? 2011-06-24T12:08:49 those being (r,c) -> (rows-c-1, r) and (r,c) -> (c, cols-r-1) 2011-06-24T12:08:58 those two break for some reason 2011-06-24T12:09:04 I'd like to see some maps that feel fundamentally different from the current 'random blob' maps rather than just different symmetries of 'random blob' ;) 2011-06-24T12:10:31 hopefully it's not too hard to adds more types of maps into this generator 2011-06-24T12:10:39 good 2011-06-24T12:10:49 it's all in one big class thing though 2011-06-24T12:10:57 which is a bit rank, but my python fu is lacking 2011-06-24T12:11:30 i'm going to translate the maps too 2011-06-24T12:11:40 i think that should make it even harder to work out the type of symmetry 2011-06-24T12:11:51 *** sigh has quit IRC (Ping timeout: 276 seconds) 2011-06-24T12:12:26 *** sigh has joined #aichallenge 2011-06-24T12:12:36 *** choas has joined #aichallenge 2011-06-24T12:13:07 *someone* should really write a map editor 2011-06-24T12:13:52 a really detailed one would be sweet 2011-06-24T12:14:06 like you say what type of symmetry you want, then it fills in symmetric squares as you edit one 2011-06-24T12:14:09 nonono, don't call for gfx and detail 2011-06-24T12:14:34 mleise: i don't expect you to do more dw :) 2011-06-24T12:14:43 ah yes, that's a good idea antimatroid. something i once started with, but I didn't fell like spending time on it 2011-06-24T12:15:13 *** McLeopold has joined #aichallenge 2011-06-24T12:15:35 McLeopold: I have a rotational map generator :) 2011-06-24T12:15:55 cool, are you including the ant starting orientation? 2011-06-24T12:16:13 not yet, what format do you want? 2011-06-24T12:16:18 that's one of the few things i have left 2011-06-24T12:16:27 s row col [nesw][nesw] 2011-06-24T12:16:30 http://pastebin.com/ayLMrWYa 2011-06-24T12:16:36 some drafts for you :P 2011-06-24T12:16:43 ant paste! 2011-06-24T12:16:50 ants.zeroviz.us 2011-06-24T12:17:44 bah, this is many at once :P 2011-06-24T12:17:47 i'm still working on it 2011-06-24T12:17:52 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-24T12:18:02 McLeopold: btw, are you still interested in the map viewing feature of the visualizer? 2011-06-24T12:18:43 yes, how do I use it? 2011-06-24T12:18:56 Just load the map as the replay file 2011-06-24T12:19:03 McLeopold: "s row col [n][e]" or "s row col n e"? 2011-06-24T12:20:12 http://ants.zeroviz.us/PW4xK/ 2011-06-24T12:20:35 mleise: I'll try that 2011-06-24T12:22:19 oh that map looks interesting 2011-06-24T12:22:33 *** stocha has joined #aichallenge 2011-06-24T12:22:40 "What do you see in this image?" 2011-06-24T12:23:05 mcstar: i have no idea how many line of code i have in my current bot framework. Bots won't use it all anyway. 2011-06-24T12:23:59 a week ago i attended a standard 10 card rorsach test 2011-06-24T12:24:34 so long guys, got to go now 2011-06-24T12:24:39 *** mcstar has left #aichallenge ("WeeChat 0.3.5") 2011-06-24T12:24:51 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-24T12:27:24 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-24T12:27:35 I like this map: http://ants.zeroviz.us/hzK3K/ 2011-06-24T12:33:28 *** okayzed is now known as okay 2011-06-24T12:34:48 *** onensora has quit IRC () 2011-06-24T12:35:39 *** sigh has joined #aichallenge 2011-06-24T12:36:16 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 4.0.1/20110413222027]) 2011-06-24T12:39:47 working out the food spawn stuff is a pita 2011-06-24T12:39:49 :) 2011-06-24T12:40:20 mleise: can you make it so we can still drag the map around? 2011-06-24T12:40:40 antimatroid: I need to find a pastebin for you... 2011-06-24T12:41:12 This explains the orientation stuff http://pastebin.com/yUBZTdpj 2011-06-24T12:42:29 McLeopold: I was wondering if I kept map dragging enabled 2011-06-24T12:42:39 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-24T12:43:10 oh, you did 2011-06-24T12:44:33 I did? I didn't? Don't confuse me 2011-06-24T12:44:43 I had interactive off 2011-06-24T12:44:46 aichallenge: McLeopold epsilon * r3ed23d6 / (3 files): added map viewing - http://bit.ly/j4KPp3 2011-06-24T12:44:55 ah ok 2011-06-24T12:46:18 aichallenge: McLeopold epsilon * re31be44 / website/map.php : added map visualizer page - http://bit.ly/ilTgwT 2011-06-24T12:46:41 *** gobang has joined #aichallenge 2011-06-24T12:48:29 aichallenge: McLeopold epsilon * re82b9ae / (website/index.php website/visualizer.php): fixed game visualizer pages - http://bit.ly/mzsTLd 2011-06-24T12:51:16 *** amstan has joined #aichallenge 2011-06-24T12:51:16 *** ChanServ sets mode: +o amstan 2011-06-24T12:59:18 McLeopold: this is going to be harder than i thought :P 2011-06-24T13:00:17 antimatroid: http://ants.zeroviz.us/latest 2011-06-24T13:00:28 yes i see that 2011-06-24T13:00:53 http://pastebin.com/J2Vszb57 2011-06-24T13:00:56 consider those examples 2011-06-24T13:01:01 i need to start 'a' in the same location' 2011-06-24T13:01:02 mleise: where's the ant build howto? 2011-06-24T13:01:48 antimatroid: your going to need a row *and* col :) 2011-06-24T13:01:58 yeah that's not the problem 2011-06-24T13:02:07 and make the last part just 'ne' 2011-06-24T13:02:10 the problem is that 'a' isn't always in the same relative symmetric place 2011-06-24T13:02:14 well, type "ant" and it tells you to type "ant -p" which tells you what commands are available 2011-06-24T13:02:31 so i can't always make him ne then etc. 2011-06-24T13:02:32 I wanted a copy paste from the wiki, but it's gone :( 2011-06-24T13:02:33 it breaks 2011-06-24T13:02:54 antimatroid: you could, they are all relative directions anyway 2011-06-24T13:03:14 that first map the directions don't work right? 2011-06-24T13:03:18 there was a wiki page on how to use ant? 2011-06-24T13:03:31 whereas for the second map they do, they are listed from a onwards 2011-06-24T13:04:00 McLeopold: have we got food spawn code for this? 2011-06-24T13:04:08 i would be interested in testing this once it's done... 2011-06-24T13:06:39 antimatroid: not yet 2011-06-24T13:06:55 McLeopold: you could copy past the relevent ant build command out of server_setup.py if you wanted 2011-06-24T13:07:08 I found it https://github.com/aichallenge/aichallenge/blob/1edebf6b27d557dca3eb3e775c1e5ece7be2968c/INSTALL.md 2011-06-24T13:10:13 @seen sigh 2011-06-24T13:10:13 McLeopold: sigh was last seen in #aichallenge 2 days, 19 hours, 8 minutes, and 11 seconds ago: yes, that seems likely 2011-06-24T13:11:32 btw, the workers have died a few times in the last day because they weren't able to connect to the main server 2011-06-24T13:12:01 I'm thinking I might try and make them a little more robust in that regard if they keep having a problem 2011-06-24T13:12:06 janzert: is there a python traceback? 2011-06-24T13:12:15 yes, there was 2011-06-24T13:12:48 If they can't get a task, they should fall back to the wait 20 seconds before trying code 2011-06-24T13:13:01 and maybe send an email or something 2011-06-24T13:13:06 latest one was http://pastebin.com/3BvbtWb4 2011-06-24T13:13:36 I never did try ant get the reposts to work 2011-06-24T13:15:33 janzert: are you going to fix it or should I? 2011-06-24T13:16:10 I'm not fixing it right now no, if it continues to be a problem I'll probably have time to look at it later today or tomorrow 2011-06-24T13:24:26 *** contestbot has joined #aichallenge 2011-06-24T13:25:35 *** mleise has quit IRC (Ping timeout: 250 seconds) 2011-06-24T13:35:24 McLeopold: http://pastebin.com/mijfed6L 2011-06-24T13:37:02 this is all set up with tessellated maps too :) 2011-06-24T13:41:49 why do you hate ant paste? 2011-06-24T13:42:43 i just find that quick :P 2011-06-24T13:42:50 sorry, these are really rough draft maps 2011-06-24T13:42:54 it's good if they get lost 2011-06-24T13:43:11 do you need a command line pastebinit? 2011-06-24T13:43:47 i just copy and paste to a web form 2011-06-24T13:43:57 ant paste has a web form 2011-06-24T13:44:14 it remembers your name 2011-06-24T13:44:24 what do you think about the maps though? 2011-06-24T13:44:25 2 clicks, and a paste 2011-06-24T13:44:30 i have a couple of things i want to do yet.. 2011-06-24T13:44:40 I can't really visualize them as raw text.... 2011-06-24T13:44:44 :p 2011-06-24T13:45:19 1. translate the rotationally symmetric maps to make it harder to work out the type of symmetry, 2. make is so i can check that an nxn block can traverse the land rather than a 1x1 block 2011-06-24T13:47:05 i have 5 types of rotation symmetry for 2 player games, 3 types for 4 player games and 1 for 8 player games 2011-06-24T13:47:42 can you fix maps if the nxn fails? 2011-06-24T13:48:14 i have an is_valid function that makes sure a 1x1 block 2011-06-24T13:48:24 the idea is to make that use an nxn block 2011-06-24T13:48:27 so yes 2011-06-24T13:48:47 i currently have nxn blocks carving land out of water 2011-06-24T13:49:10 http://pastebin.com/JG3kxFzG 2011-06-24T13:49:13 that's the generator atm 2011-06-24T13:49:16 a complete mess :) 2011-06-24T13:57:39 aichallenge: McLeopold epsilon * r7b77d92 / ants/playgame.py : Closes #202 - http://bit.ly/jLp3iq 2011-06-24T13:58:34 *** stocha has joined #aichallenge 2011-06-24T13:58:35 Your next game should be in about -2.7 minutes. 2011-06-24T13:58:42 now that one is good :) 2011-06-24T14:00:18 *** carlop has joined #aichallenge 2011-06-24T14:00:40 * antimatroid should never program in python 2011-06-24T14:00:41 Your next game should be in about -5.2 minutes. 2011-06-24T14:01:02 has the thing crashed ? or is there a very long game running ? 2011-06-24T14:01:27 McLeopold: a possible solution for relation map/food generator, is let the food generator find the type of simmetry 2011-06-24T14:02:37 fixed the orientation of a ant (say NE), you can choose another ant, try every possible orientation (at most 8) 2011-06-24T14:02:45 at most eight9 2011-06-24T14:02:47 ) 2011-06-24T14:03:30 so the food generation code can fast understand what type of symmetry has the map, without need to encoding 2011-06-24T14:04:39 (and the symmetry finder is O(number of total cell), so it shold run very fast) 2011-06-24T14:04:48 Stocha : Your next game should be in about -9.5 minutes. 2011-06-24T14:07:11 *** stocha has quit IRC (Quit: Page closed) 2011-06-24T14:09:29 *** choas has quit IRC (Read error: Operation timed out) 2011-06-24T14:14:50 *** Naktibalda has joined #aichallenge 2011-06-24T14:38:55 mleise: http://ants.zeroviz.us/sLuzI/ 2011-06-24T14:40:00 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-06-24T14:40:35 *** Null_01x has joined #aichallenge 2011-06-24T14:41:24 *** Null_01x has left #aichallenge 2011-06-24T14:47:02 *** UncleVasya has joined #aichallenge 2011-06-24T14:47:53 Does it means something if 2 different bots timeouted at the same tutrn? 2011-06-24T14:48:04 *turn 2011-06-24T14:57:41 maybe 2011-06-24T15:16:21 does it matter for me, if i'm fighting 2 red and 2 blue ants or 4 of the same color ? 2011-06-24T15:17:33 gobang: your ant die indipendentrly of composition of enemy ant 2011-06-24T15:18:37 say, i got 5 there 2011-06-24T15:19:12 but it matter for you cause if there are 2 team enemy ant die faster, so you can retract (leaving the two team fighting), or help, lose few ant for kill em all 2011-06-24T15:19:32 but i think, i got it, thanks 2011-06-24T15:20:33 antimatroid: there is a lot of fun in map generation, and i've found few more type of symmetry, eg: http://pastebin.com/X30Z50rf 2011-06-24T15:22:15 later i take a look at map generation code 2011-06-24T15:22:43 now chicken time, bye 2011-06-24T15:26:27 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-24T15:28:41 *** carlop has quit IRC (Ping timeout: 240 seconds) 2011-06-24T15:35:49 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-24T15:38:50 aichallenge: McLeopold epsilon * r76eb30e / (website/api_get_task.php worker/worker.py): improvements to worker - http://bit.ly/in603S 2011-06-24T15:44:01 is there a reason aren't you just doing a time.sleep(wait_time) at worker.py line 187? 2011-06-24T15:44:31 and rather than catching Exception there IOError or at least StandardError would probably be better 2011-06-24T15:52:12 janzert: I like to Ctrl-C the script with a fast response time 2011-06-24T15:52:32 And I'm bad at catching the proper exceptions, so feel free to improve that 2011-06-24T15:53:07 Ctrl-C breaks a time.sleep 2011-06-24T15:53:10 at least here 2011-06-24T15:53:16 then windows sucks 2011-06-24T15:53:32 and Ctrl-C is the main reason I don't want the a bare Exception catch ;) 2011-06-24T15:53:44 well I guess not the main reason but a big one 2011-06-24T15:53:49 aichallenge: McLeopold epsilon * rbf99ec0 / (website/profile.php website/sql.php): update to next game logic - http://bit.ly/iLv9Dz 2011-06-24T15:54:06 I guess I'm totaly wrong on that... 2011-06-24T15:55:19 wow. Am I missed some improvement to the server? Because just now my 14th version had a game instead of 16th and it makes me feel like "wow!" :) 2011-06-24T15:55:40 aichallenge: McLeopold epsilon * r1433c7e / worker/worker.py : removed stupid code - http://bit.ly/jbjeSo 2011-06-24T15:55:57 http://aichallengebeta.hypertriangle.com/profile.php?user=26 2011-06-24T15:56:01 on the next game stuff, how about changing the verbiage for the next game to "Your next game should be within" ? 2011-06-24T15:56:24 then try and make the calculation the maximum time till the game 2011-06-24T15:56:39 it might simplify the logic that way? 2011-06-24T15:56:44 ok 2011-06-24T15:57:10 I don't think the logic is too far off 2011-06-24T15:58:07 aichallenge: McLeopold epsilon * r4d64cad / website/profile.php : word smithing - http://bit.ly/mvB0sd 2011-06-24T15:58:31 oh, I thought you had mentioned doing something more complicated to try and predict the next game time 2011-06-24T15:58:54 looking at the code it appears not 2011-06-24T15:59:02 lol, version 14 keeps playing (already 3 games stolen from my newer version). Can somebody explain what's going on? 2011-06-24T15:59:12 I'm playing a game right now! http://aichallengebeta.hypertriangle.com/profile.php?user=11 2011-06-24T15:59:28 nice 2011-06-24T15:59:45 That should take care of the negative time stuff 2011-06-24T16:00:01 great 2011-06-24T16:00:11 UncleVasya: I'm messing with the database 2011-06-24T16:00:15 *** Accoun has quit IRC () 2011-06-24T16:01:14 McLeopold: yep, you like to play with it and make people worrying :) 2011-06-24T16:01:24 no offence, just a joke 2011-06-24T16:01:49 I'm a bit of a cowboy 2011-06-24T16:02:44 *** gobang has joined #aichallenge 2011-06-24T16:05:07 *** |UncleVasya| <|UncleVasya|!kvirc@178-133-162-3.dialup.umc.net.ua> has joined #aichallenge 2011-06-24T16:06:14 @later tell mleise http://ants.zeroviz.us/BqmlS/ can we have blank maps? 2011-06-24T16:06:14 McLeopold: Yes master! 2011-06-24T16:07:19 McLeopold: how can you have a map without any players though? 2011-06-24T16:08:01 I suppose it is just to look at the water patterns? 2011-06-24T16:08:16 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-24T16:10:11 *** choas has joined #aichallenge 2011-06-24T16:10:19 *** sigh has joined #aichallenge 2011-06-24T16:10:56 * janzert wonders how TRavary's bot managed to pass the functional test :( 2011-06-24T16:11:15 *** Accoun has joined #aichallenge 2011-06-24T16:14:41 aichallenge: janzert epsilon * r6b18102 / worker/worker.py : Narrow exception caught - http://bit.ly/igAU8u 2011-06-24T16:15:06 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-24T16:15:25 *** gobang has joined #aichallenge 2011-06-24T16:24:35 while(1): You are playing in a game right now. 2011-06-24T16:26:36 is there an english word for applying a fix and making it worse ? 2011-06-24T16:28:09 *** Palmik_ has quit IRC (Remote host closed the connection) 2011-06-24T16:47:54 *** amstan has joined #aichallenge 2011-06-24T16:47:54 *** ChanServ sets mode: +o amstan 2011-06-24T16:50:44 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-24T16:51:02 *** gobang has joined #aichallenge 2011-06-24T16:52:04 *** stocha has joined #aichallenge 2011-06-24T16:52:14 is aichallengebeta unvailable ? 2011-06-24T16:52:37 yeah, it's unresponsive for me too 2011-06-24T16:53:09 and it's back again 2011-06-24T16:53:24 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 5.0/20110622232440]) 2011-06-24T16:54:59 what is ant paste ? 2011-06-24T16:55:29 similiar to pastebin but for maps and replays 2011-06-24T16:55:36 it seems i'm back to 3rd rank :p 2011-06-24T16:56:08 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-24T16:56:14 see http://ants.zeroviz.us 2011-06-24T16:56:19 i doubt that all played game shows though 2011-06-24T16:57:00 do you mean that submiting a map, is not submiting the code, only the raw data of the static map ? 2011-06-24T16:57:12 or is this just a tool for people to share stuff ? 2011-06-24T16:57:18 *** sigh has joined #aichallenge 2011-06-24T16:57:30 not sure what the first question means, but yes to the second 2011-06-24T16:58:18 janzert: i think McLeopold wants people to submit maps. I though it was map code generation. But then, i wondered if it was juste the empty water/ground map in a paste. 2011-06-24T16:58:29 map example http://ants.zeroviz.us/ELUHK/ replay example http://ants.zeroviz.us/sLuzI/ 2011-06-24T16:58:56 not sure what McLeopold might have been refering to 2011-06-24T16:59:31 can you spool some of the pasted map directly to the contest ? 2011-06-24T16:59:31 but I would guess a map generator yes 2011-06-24T16:59:38 nope 2011-06-24T16:59:50 it's just a way to easily view things 2011-06-24T16:59:57 *** sigh has quit IRC (Client Quit) 2011-06-24T17:00:11 nice one. 2011-06-24T17:01:19 *** FireFly has quit IRC (Quit: swatted to death) 2011-06-24T17:07:45 *** UncleVasya has joined #aichallenge 2011-06-24T17:08:44 *** okay is now known as okayzed 2011-06-24T17:09:02 Les Fourmis - Boris Vian. - one part of me always tried to laugh, but was outvoted by the (shocked) majority 2011-06-24T17:11:51 *** |UncleVasya| <|UncleVasya|!kvirc@178-133-162-3.dialup.umc.net.ua> has quit IRC (Ping timeout: 260 seconds) 2011-06-24T17:12:40 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-24T17:23:18 I haven't read about new battle resolution much but is it possible that no one dies in the fight? 2011-06-24T17:23:38 don't think so 2011-06-24T17:23:48 I just saw it. 2011-06-24T17:23:54 wait a second 2011-06-24T17:24:13 http://aichallengebeta.hypertriangle.com/visualizer.php?game=36300&user=16 2011-06-24T17:24:16 turn 46 2011-06-24T17:24:30 row 39, col 46 2011-06-24T17:26:01 I believe spawning happens after the battle phase 2011-06-24T17:26:26 so they would only have fought if they hadn't moved apart during the next turn 2011-06-24T17:27:42 *** davidjliu has quit IRC (Ping timeout: 252 seconds) 2011-06-24T17:29:23 yes, it seems to be movement->collision->battle->spawn 2011-06-24T17:29:55 *** foRei has quit IRC (Quit: Bye) 2011-06-24T17:33:59 *** carlop has joined #aichallenge 2011-06-24T17:40:57 *** |UncleVasya| <|UncleVasya|!kvirc@178-133-202-76.dialup.umc.net.ua> has joined #aichallenge 2011-06-24T17:42:11 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-24T17:47:44 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-24T17:57:04 *** okayzed is now known as okay 2011-06-24T17:59:47 *** antimatroid has quit IRC (Ping timeout: 240 seconds) 2011-06-24T18:03:00 @later tell mleise some room for improvement in the attack display http://aichallengebeta.hypertriangle.com/visualizer.php?game=36300&user=16&turn=46&row=39&col=46 2011-06-24T18:03:00 McLeopold: As you wish. 2011-06-24T18:09:15 *** Zaphus has joined #aichallenge 2011-06-24T18:10:29 McLeopold: another thing with attack display is that when two ant battle from opposite side of the map, battle radius and animation have some problem 2011-06-24T18:16:36 *** |UncleVasya| <|UncleVasya|!kvirc@178-133-202-76.dialup.umc.net.ua> has quit IRC (Ping timeout: 260 seconds) 2011-06-24T18:17:22 carlop: if you can find a game with that, can you create a full link for mleise like I did? 2011-06-24T18:18:50 you can see the battle radius issue on any replay, just put the cursor on an ant that is on the border of the map and see that the circle doesnt draw on the other side of the line (game mechanics work fine, it's a display bug) 2011-06-24T18:19:58 *** okay is now known as okayzed 2011-06-24T18:25:39 @later tell mleise that replay have some problem when drawing battle near the border of the map: http://aichallengebeta.hypertriangle.com/visualizer.php?game=36300&user=16&turn=453&row=24&col=59 (+ sometimes the circle doesnt draw on the other side of the line) 2011-06-24T18:25:39 carlop: Job's done. 2011-06-24T18:28:22 *** nux67 has quit IRC (Quit: Page closed) 2011-06-24T18:33:20 *** demerzel has quit IRC (Ping timeout: 252 seconds) 2011-06-24T18:36:24 hi 2011-06-24T18:36:41 *** sigh has joined #aichallenge 2011-06-24T18:44:06 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-24T18:44:36 *** Zaphus has quit IRC (Quit: Page closed) 2011-06-24T18:55:50 *** sigh has joined #aichallenge 2011-06-24T18:56:45 *** sigh has joined #aichallenge 2011-06-24T18:59:44 *** sigh has quit IRC (Client Quit) 2011-06-24T19:00:43 *** sigh has joined #aichallenge 2011-06-24T19:05:16 @later tell antimatroid that the problem of draw ant map is know in the world as "wallpaper group"my suggestion on adding few more symmetries (http://pastebin.com/X30Z50rf) 2011-06-24T19:05:16 carlop: I think that worked... 2011-06-24T19:06:10 @later tell antimatroid that the problem of draw ant map is know in the world as "wallpaper group": http://en.wikipedia.org/wiki/Wallpaper_group, this page can help a lot on design few more symmetry class 2011-06-24T19:06:10 carlop: As you wish. 2011-06-24T19:09:25 *** mleise has joined #aichallenge 2011-06-24T19:09:26 i have some problem when i missclick on "enter"... sorry 2011-06-24T19:14:45 aichallenge: McLeopold epsilon * r254aec8 / (7 files in 2 dirs): improvements to game list and profile submissions - http://bit.ly/lhj425 2011-06-24T19:16:20 janzert: we have proof that the top bots are getting few games http://aichallengebeta.hypertriangle.com/rankings.php 2011-06-24T19:17:14 McLeopold: You really do want maps without player starts? What is the use case? 2011-06-24T19:18:38 mleise: just for pasting maps to ant paste 2011-06-24T19:18:58 I just want the error to be ignored. 2011-06-24T19:19:04 I suppose I could add the debug flag? 2011-06-24T19:19:59 yes, that should do the trick, but it is not very elegant to put it as a get parameter I guess 2011-06-24T19:21:36 how do I add the debug flag without a querystring? 2011-06-24T19:22:51 let me take a look at my source code. it is 1:20 am here. 2011-06-24T19:24:29 well, there is no way currently, to do this. only 'interactive' and 'dataDir' can be set through the get parameters 2011-06-24T19:26:03 McLeopold: then again you could create the visualizer and then add "vis.options['debug'] = true;" 2011-06-24T19:27:16 Since in JavaScript everything can be changed you don't need some getter or setter for that. 2011-06-24T19:27:29 visualizer.vis.options? 2011-06-24T19:27:47 vis would by the variable that you store the visualizer instance in. 2011-06-24T19:28:21 vis = new Visualizer(...); vis.options['debug'] = true; 2011-06-24T19:28:46 then load the map 2011-06-24T19:29:00 woot \o/ http://ants.zeroviz.us/BqmlS/ 2011-06-24T19:29:36 I would like to point out how nicely my cell maze tiles :) 2011-06-24T19:29:43 hey the water looks like rivers now :) 2011-06-24T19:30:24 does this map have any kind of symmetry? 2011-06-24T19:30:38 not yet 2011-06-24T19:31:01 I've half wrote the symmetry 2011-06-24T19:32:43 yeah, but it's looking good so far. very... organic 2011-06-24T19:34:41 McLeopold: you could probably also replace all '#fff' with your yellowish background color *g* 2011-06-24T19:35:16 what yellow? 2011-06-24T19:37:24 the page background 2011-06-24T19:37:29 McLeopond: maybe you want to take a look this link about wallpaper tessellation(that is the problem of design ant map): http://en.wikipedia.org/wiki/Wallpaper_group#Group_p4m 2011-06-24T19:38:19 carlop: The link you sent me, what is wrong there? It renders ok for me. 2011-06-24T19:39:05 carlop: yeah, I read that, it looks good 2011-06-24T19:39:20 i see only one the upper right ant attacking, and the bottom one do nothing 2011-06-24T19:42:08 McLeopold: tomorrow i try to find all possible symmetry group that give interesting map, and try to generate few map, there is a lot of fun on map generation 2011-06-24T19:44:03 aichallenge: McLeopold epsilon * r1e565ea / website/game_list.php : fixed missing td tag - http://bit.ly/lEKJir 2011-06-24T19:44:41 carlop: ah, now i see, thx 2011-06-24T19:47:09 mleise: and if you zoom out 2 times, you can see that the attack range circle are not full draw 2011-06-24T19:47:36 aichallenge: McLeopold epsilon * r9c69a70 / website/profile_submissions_widget.php : fixed missing submissions - http://bit.ly/ikwi8C 2011-06-24T19:48:41 carlop: I hoped noone would notice this, because this damn wrap-checking bloats my code with every item i draw :p 2011-06-24T19:50:01 I'll fix it tomorrow 2011-06-24T19:50:38 maybe you can draw every item at (r,c) near the border another time at (r-num_row, c), so the other part of circle is draw, or something similar 2011-06-24T19:50:53 aichallenge: McLeopold epsilon * rd761c39 / website/profile.php : added funny message - http://bit.ly/mdNBu1 2011-06-24T19:51:14 but i think this is a minor fix, that can be done after launch without problem 2011-06-24T19:52:23 carlop: sure, and the more general case is checking the bounding-box, use some modulo operation on it so it starts left most and top most and then redraw it while increasing x and y by width and height of map until the bounding-box would be outside the map 2011-06-24T19:53:35 (that way objects 5x the map size would still be repeated often enough, like a trace of an ant that moved several times from left to rigth through the map) 2011-06-24T19:54:26 I mean, I have the code for that, its just a dull copy and paste every time :p 2011-06-24T19:55:02 @later tell janzert have the workers gone on strike? 2011-06-24T19:55:02 McLeopold: Ready to serve. 2011-06-24T19:56:53 mleise: yep, almost my solution... have a good night, bye 2011-06-24T19:57:26 bye 2011-06-24T19:58:07 mleise: did you see my link to a game that had battle lines drawn for a spawned ant? 2011-06-24T20:00:09 yes I've seen that, will also fix that tomorrow 2011-06-24T20:00:54 McLeopold: workers are getting bad json from server during post result 2011-06-24T20:01:15 janzert: I'm running a game now on beta to see 2011-06-24T20:01:28 http://pastebin.com/jyGFWDAU 2011-06-24T20:02:05 when did it start? 2011-06-24T20:02:30 hmm, beta posted fine... 2011-06-24T20:02:42 45 minutes approximately 2011-06-24T20:02:57 or no 2011-06-24T20:03:36 *** carlop has quit IRC (Ping timeout: 258 seconds) 2011-06-24T20:04:11 connection timeout about 1.75 hours ago then the bad json error since 2011-06-24T20:04:35 maybe something with the retry isn't actually correct? 2011-06-24T20:05:47 this is the start of it http://pastebin.com/S2VpD0Qg 2011-06-24T20:08:32 gtg 2011-06-24T20:08:38 I'm not sure what the fix is 2011-06-24T20:09:15 *** McLeopold has left #aichallenge 2011-06-24T20:10:34 hmm, restarted the worker a minute ago and it did this http://pastebin.com/b7kTmXS4 2011-06-24T20:10:47 shortly before you said gtg... but I see you're gone 2011-06-24T20:11:29 and now it seems to be running again 2011-06-24T20:17:18 *** choas has quit IRC (Read error: Operation timed out) 2011-06-24T20:20:51 *** antimatroid has joined #aichallenge 2011-06-24T20:44:51 *** sigh has joined #aichallenge 2011-06-24T20:46:00 *** sigh has quit IRC (Client Quit) 2011-06-24T21:08:11 i can now make sure an nxn block can traverse all lands squares from any starting point on land 2011-06-24T21:12:46 *** Zaphus has joined #aichallenge 2011-06-24T21:17:44 *** stocha has joined #aichallenge 2011-06-24T21:18:03 there is simply to much parameters in my bot already :'( 2011-06-24T21:19:09 i have no idea how to sort them. On some map size it does better than my ref_greedy bot. On others it does far worse. Are the parameters good or not ? And then i have also the feature. Which one should i keep, and which one is not usefull ? I have no idea. 2011-06-24T21:19:22 i don't even have an idea about how to test it. 2011-06-24T21:19:43 Zaphus: do you have any suggestions ? :) 2011-06-24T21:20:01 stocha: what's the question ? 2011-06-24T21:20:15 oh 2011-06-24T21:20:21 Zaphus: how do i tune my parameters / sort my features. 2011-06-24T21:20:24 bah python fail 2011-06-24T21:20:31 how do i copy a list without it being the same copy? 2011-06-24T21:20:34 like "old_map = self.map_data" 2011-06-24T21:20:39 When i have no idea what the final map / bots will be like. 2011-06-24T21:21:15 Even if it was 1vs1 i'll be troubled. 2011-06-24T21:21:21 The truth is, on different maps/etc some bots will do better than others - you need to find the best compromise. I do this by playing a 'gauntlet' of matches and watching which one rises to the top (I'm using win percentage at the moment but may change that) 2011-06-24T21:21:24 but there, i just have no idea. 2011-06-24T21:21:46 what's a gauntlet ? 2011-06-24T21:22:02 stocha: i find it easiest to consider simple general cases, then build my strategy up as generally as possible from that 2011-06-24T21:22:23 stocha: the important thing I do is watch the rate that my bot finishes above each of my other bots, so if my top bot is losing 100% of the time to another bot it means it is strong but has a specific weakness 2011-06-24T21:22:25 antimatroid: it doesn't sound very optimized. 2011-06-24T21:22:52 stocha: it works surprisingly well 2011-06-24T21:22:54 Zaphus: you test it against your other bots ? 2011-06-24T21:23:11 antimatroid: i'm not convinced yet. But at least, you have a plan :) 2011-06-24T21:23:38 stocha: gauntlet... "an onlsaught or attack from all sides" (it has other meanings, but thats the one I mean) 2011-06-24T21:23:44 anyone here know python? 2011-06-24T21:23:54 this is like the last tweak of my map gen and meant to be an easy one :( 2011-06-24T21:24:03 amstan: you here? 2011-06-24T21:24:15 Zaphus: i have many bots now. Some are even quite different from others (but the most originals are sadly weak) 2011-06-24T21:24:20 stocha: yes, I keep the .exe versions of older bots to test against. means I tend to get 50-50 results until I make noticable improvements 2011-06-24T21:24:31 it also takes a long time to run the tests :-) 2011-06-24T21:24:32 Zaphus: but the map influences the result so much ... i don't know. 2011-06-24T21:25:21 stocha: right now I do 4-player maps. All maps, all combinations of 4 bots . I also track the win rate on the different maps. 2011-06-24T21:25:23 Zaphus: a yes, my not having any available test machines criples my habits also. I would simply play 2000 games against each of my references bots then. But i can't. 2011-06-24T21:25:37 Zaphus: you don't add 1vs1 ? 2011-06-24T21:25:48 1vs1 is my main line of information. 2011-06-24T21:25:50 i don't like testing with lots of games either 2011-06-24T21:25:54 stocha: not yet, but I have "sleeperbot" for that 2011-06-24T21:26:01 i prefer to test on examples i can decompose to work out what my bot is actually doing 2011-06-24T21:26:09 then find cases where it breaks from what i want to do 2011-06-24T21:26:16 antimatroid: that's important too. 2011-06-24T21:26:20 then go ahead and decompose it until i work out hwat is going wrong 2011-06-24T21:26:40 antimatroid: I prefer to start the tests, go and eat nachos and watch the footy :-) 2011-06-24T21:26:51 Zaphus: yeah i will do that on tcp 2011-06-24T21:26:59 Zaphus: that also is important :) 2011-06-24T21:27:01 but only to look for errors, then i need to go and replicate them 2011-06-24T21:27:35 antimatroid: I have no way to replicate at the moment, interested to hear how you do that 2011-06-24T21:27:36 i only need a few hundreds euros, and i would buy a cheap slient machine. then i can test on it. Sadly i don't have any money left for the time being. 2011-06-24T21:29:11 antimatroid: yes? 2011-06-24T21:29:41 amstan: i'm trying to translate my map, but when I do "old_map = self.map_data" which is a list (of lists), then it seems to be the same copy? 2011-06-24T21:29:50 so when i update self.map_data, it seems to change old_map 2011-06-24T21:30:02 how do i make an actual copy of self.map_data? 2011-06-24T21:30:46 http://pastebin.com/MF5aXcSw here's some sample maps, a 2x2 block can traverse all land squares 2011-06-24T21:30:57 yeah, my bot won two game in a row. Although i already want to update it .... i probably should go to sleep in fact. It's 3:30 am here. 2011-06-24T21:31:47 antimatroid: why not simply go with a 100% ground map ? then there is no blocking problems :p 2011-06-24T21:31:55 just realised I never saved the code for "MrBlobby" - doh! 2011-06-24T21:31:57 stocha: variation in maps is good 2011-06-24T21:32:04 what do you guys think of those maps? 2011-06-24T21:32:37 *** carlop has joined #aichallenge 2011-06-24T21:33:14 antimatroid: i think they are good as widely open maps. hardly as corridor maps though. 2011-06-24T21:33:31 antimatroid: python does shallow copies all the time 2011-06-24T21:33:35 yeah, random walks doesn't really do maze like maps well 2011-06-24T21:33:51 antimatroid: you want a new copy for sure? then you need to look at the deepcopy library 2011-06-24T21:33:54 antimatroid: I think they look good. although I would like to see some maze-like maps too 2011-06-24T21:34:20 Zaphus: yeah same, it needs to be done with something other than random walks though 2011-06-24T21:34:23 it's okay for both wide open map, and corridor maps. But here you only look like you can achieve wide open one. 2011-06-24T21:34:26 these are all just random walks with nxn blocks 2011-06-24T21:35:02 antimatroid: you should use both a nxn, then refine with a 1x1 block. 2011-06-24T21:35:19 so once you have connected each edge, you can then built tiny corridors. 2011-06-24T21:35:20 i have like 30 walks with min/max_block_size defined 2011-06-24T21:35:21 maybe. 2011-06-24T21:35:45 but i need min_block_size to be <= block_size required to traverse the grid, otherwise the map gen takes ages 2011-06-24T21:36:05 it's already taking up to a second or two for maps sometimes 2011-06-24T21:36:34 erm 2011-06-24T21:36:36 >=* 2011-06-24T21:41:41 amstan: i have lots of types of symmetry for you now 2011-06-24T21:42:09 i had tessellation, now for 2 players i have 5 types of rotational symmetry, 3 types for 4 players and 1 type for 8 players 2011-06-24T21:42:56 antimatroid: how do you feed test cases to your bot ? 2011-06-24T21:43:08 Zaphus: input file 2011-06-24T21:43:24 if you have an ifstream if, you can do if >> state 2011-06-24T21:43:48 or just paste it into the command window when running the bot 2011-06-24T21:44:05 but where does the state come from ? 2011-06-24T21:44:14 i make it? 2011-06-24T21:44:21 in a .txt document 2011-06-24T21:44:26 *** stocha has quit IRC (Quit: Page closed) 2011-06-24T21:44:33 the c++ starter bot has a state struct too 2011-06-24T21:44:50 hmm - I guess I don't know what it is I'd be making to feed in to it 2011-06-24T21:45:03 look at the sample input on the specs page 2011-06-24T21:45:19 just copy that and edit the ant/water info to meet your debug requirement 2011-06-24T21:45:44 the c+++ bot has operater>>(istream, state) defined for reading state information 2011-06-24T21:46:08 it uses cin >> state to get the state info from the engine, but you can just as easily pass pseudo state info from an input file 2011-06-24T21:46:15 oh, thats ugly... very manual process then figuring out the coords of each bit of water and ants just to make a test case ? 2011-06-24T21:46:42 i usually plonk an ant somewhere then work out the relative information for my debug case 2011-06-24T21:46:53 you could always write a separate input function which read a char map for you 2011-06-24T21:47:01 but i'd prefer not to messy things with that 2011-06-24T21:47:49 yeah, guess that makes sense... not sure if I'll be bothered :-) 2011-06-24T21:51:38 *** Zaphus has quit IRC (Quit: Page closed) 2011-06-24T21:57:21 <_flag> What do the testcases on the server look like? My bot crashed for some reason 2011-06-24T22:00:55 <_flag> Nevermind, I think I found it 2011-06-24T22:13:00 anyone got suggestions for how to actually make maze maps? 2011-06-24T22:13:34 antimatroid: oh? 2011-06-24T22:13:45 amstan: ? 2011-06-24T22:14:07 antimatroid: http://www.astrolog.org/labyrnth/algrithm.htm 2011-06-24T22:15:53 <_flag> antimatroid: How wide do you plan on making the corridors? 2011-06-24T22:16:02 _flag: not sure 2011-06-24T22:16:07 i can make them as wide as i want now 2011-06-24T22:16:36 i'm fairly tempted to make it so that a 3x3 block can traverse maos 2011-06-24T22:17:03 that doesn't remove the possibility of two regions being joined at one place with a 1 square opening, but it does mean there's a way around it 2011-06-24T22:18:22 <_flag> As long as it isn't absurdly narrow it should be fine 2011-06-24T22:18:41 yep 2011-06-24T22:18:57 although, if the water was all 1 width it could be quite interesting 2011-06-24T22:19:02 a full maze like that? 2011-06-24T22:19:02 :P 2011-06-24T22:19:15 1 width land corridors too 2011-06-24T22:19:21 that would be psycho 2011-06-24T22:20:27 <_flag> I'm pretty sure it would throw all the bots designed for regular games for a loop, my bot included 2011-06-24T22:20:34 yep 2011-06-24T22:20:49 it shall be added to the crazy maps tcp server later 2011-06-24T22:20:59 assuming we have one again 2011-06-24T22:21:18 <_flag> There was a semi-working tcp server posted a couple days ago, don't know what happened to it 2011-06-24T22:21:26 yeah it's still buggy i think 2011-06-24T22:21:36 rabidus and berak are the people to talk to i think 2011-06-24T22:24:10 _flag: http://pastebin.com/1zghWJVy lots more types of symmetry :) 2011-06-24T22:25:21 <_flag> Nice, looks a lot less random than your other ones 2011-06-24T22:26:10 i think the rotational maps do that 2011-06-24T22:26:17 all the old ones are tessellation ones 2011-06-24T22:28:45 <_flag> What do you think about making maps with well-defined shapes? For example a map made of circles, squares, triangles, etc. 2011-06-24T22:29:43 yeah mcleopold had a cool map generator where he'd place circles on the map then join them up to make a graph with lines 2011-06-24T22:30:08 i should be able to add more types of maps in with the different symmetry types 2011-06-24T22:30:30 i just need to write the functions to do it for one player, then i have functions to pick out the symmetrically equivalent squares for whatever symmetry i have 2011-06-24T22:30:34 <_flag> I think that way the maps would probably be more memorable 2011-06-24T22:33:33 <_flag> amstan: It would appear the worker is down, is this intentional? 2011-06-24T22:33:53 _flag: ask janzert/mcleopold 2011-06-24T22:34:32 <_flag> janzert: It would appear the worker is down, is this intentional? 2011-06-24T22:34:48 <_flag> McLeopold: same message as janzert 2011-06-24T22:35:40 nope and it's restarted 2011-06-24T22:41:47 contestbot: later tell McLeopold whenever the worker retries a post it always gets bad json back from the server (it seems to be just whitespace) 2011-06-24T22:41:47 janzert: Ready to serve, my lord. 2011-06-24T22:55:51 *** carlop has quit IRC (Ping timeout: 250 seconds) 2011-06-24T23:23:04 *** Birdie_ has joined #aichallenge 2011-06-24T23:29:57 *** Birdie_ has quit IRC (Quit: Birdie_) 2011-06-24T23:30:00 aichallenge: janzert epsilon * rcc19c14 / setup/worker_setup.py : Add base chroot install to package install part of worker_setup - http://bit.ly/l1fxK6 2011-06-24T23:30:00 aichallenge: janzert epsilon * ra8d66ee / (8 files in 2 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://bit.ly/jUATEV 2011-06-24T23:30:19 *** Birdie_ has joined #aichallenge 2011-06-24T23:38:53 I have a question. 2011-06-24T23:39:27 The main site says that the current tournament was back in 2010. What is the current tournament? 2011-06-24T23:41:35 <_flag> Birdie_: The beta is up for the upcoming tournament: http://aichallengebeta.hypertriangle.com/ 2011-06-24T23:43:34 I see. 2011-06-24T23:46:57 *** matrix has joined #aichallenge 2011-06-24T23:47:55 so? 2011-06-24T23:53:36 *** matrix has quit IRC (Quit: Leaving) 2011-06-24T23:58:27 *** Birdie_ has quit IRC (Remote host closed the connection)