2011-11-27T00:00:57 *** raemde has joined #aichallenge 2011-11-27T00:03:05 McLeopold: here now 2011-11-27T00:04:57 *** Euphony has quit IRC (Quit: Page closed) 2011-11-27T00:05:12 I'll give you $1 to bump me up in the queue 2011-11-27T00:05:37 :) 2011-11-27T00:06:57 *** delt0r_ has joined #aichallenge 2011-11-27T00:08:58 what queue? 2011-11-27T00:09:07 I resubmitted on aichallenge 2011-11-27T00:09:17 so I'm in like 7000th 2011-11-27T00:09:22 but with a solid win over a bot in 2500th! 2011-11-27T00:11:18 *** AntDroid_ has joined #aichallenge 2011-11-27T00:11:34 yeah you definitely killed that dude's ant 2011-11-27T00:11:57 also haha i'm in 1st on fluxid, yay trueskill 2011-11-27T00:12:09 *** fpcfan has quit IRC (Ping timeout: 245 seconds) 2011-11-27T00:13:54 *** amstan has quit IRC (Remote host closed the connection) 2011-11-27T00:15:05 *** fpcfan has joined #aichallenge 2011-11-27T00:15:17 *** amstan has joined #aichallenge 2011-11-27T00:15:17 *** ChanServ sets mode: +o amstan 2011-11-27T00:15:34 what's really annoying is how on the website it says 2011-11-27T00:15:47 it may take a few mintues for your new bots ranking to stablize 2011-11-27T00:15:59 overoptimistic much? 2011-11-27T00:17:34 wow, me and pguillory had some epic battles on fluxid 2011-11-27T00:17:40 sadly I lost them all 2011-11-27T00:18:06 oh, wait, here's one where he scored 0 2011-11-27T00:18:08 must be a fluke 2011-11-27T00:19:42 MAHAHA *this* (endgame) is what I spent all my time trying to achieve: http://ants.fluxid.pl/replay.327 2011-11-27T00:20:42 almost 500 ants in the hive, wow 2011-11-27T00:24:20 holy cow 2011-11-27T00:24:43 It's a nice problem to have when your ants gather food too fast 2011-11-27T00:24:47 I also like the massive pile I bury his hill in 2011-11-27T00:24:57 I can't think of any easy way to detect that though 2011-11-27T00:25:10 maybe penalize food gathering when you have more food in teh hive? 2011-11-27T00:25:40 Did you see the game where one bot got over 700 ants? 2011-11-27T00:26:12 Antimony: Do you have a link? 2011-11-27T00:26:30 *** ThatSnail has quit IRC (Ping timeout: 240 seconds) 2011-11-27T00:26:31 nope sorry. But it was posted in the Cool Games thread somewhere 2011-11-27T00:26:51 ok time to write a gcd function 2011-11-27T00:27:06 *** asorman has joined #aichallenge 2011-11-27T00:27:12 OK, I'll see if I can dig it out of there 2011-11-27T00:30:41 ug, I feel tired 2011-11-27T00:30:48 but I'm running out of time to work on my bot 2011-11-27T00:30:51 decisions, decision 2011-11-27T00:31:03 hah, watching this game with me and pguillory I realize we're using the same trick in one case 2011-11-27T00:31:11 what trick? 2011-11-27T00:32:13 I'll be glad to explain on Dec 19 :) 2011-11-27T00:32:37 *** AntDroid_ has quit IRC (Ping timeout: 265 seconds) 2011-11-27T00:32:48 :P 2011-11-27T00:32:52 I wonder how far I can get with crappy combat 2011-11-27T00:33:03 pretty far 2011-11-27T00:33:15 my bot is good at gathering, and I bet that symmetry will givei t a huge edge at exploration 2011-11-27T00:33:18 just search around the rankings and eyeball whether you think bots have combat 2011-11-27T00:33:47 *** besh has joined #aichallenge 2011-11-27T00:34:49 Hmm, my bot's currently at mu=81.95, so it will probably be top 100 once ranking stablizes 2011-11-27T00:35:04 How do you stop the tester from opening a browser when game ends? I want to do silent testing with many maps 2011-11-27T00:35:05 and symmetry should push it up even farther 2011-11-27T00:35:08 your sigma will drop to about 1.6 2011-11-27T00:35:12 there's an option 2011-11-27T00:35:15 so 81.95 - 1.6*3 2011-11-27T00:35:21 --nolaunch I think 2011-11-27T00:35:30 Ok thanks 2011-11-27T00:35:41 *** roflmao has quit IRC (Read error: Connection reset by peer) 2011-11-27T00:35:53 77.15 is 85th right now 2011-11-27T00:35:57 Of course, my ranking will probably drop like a stone at the end when all the super awesome final combat bots get submitted 2011-11-27T00:36:16 you have no combat now? just avoidance? 2011-11-27T00:36:24 well, I sort of have comba 2011-11-27T00:36:37 I assume all enemies are stationary, and then minimize losses 2011-11-27T00:37:27 wow, at around 100th I don't see significant combat 2011-11-27T00:37:33 nothing like you see all the time on fluxid 2011-11-27T00:37:43 it's ok when I'm attacking and the defenders are approximately static, but it doesn't work well for defence or 1v1 2011-11-27T00:38:08 have you watched my games? 2011-11-27T00:38:15 a couple 2011-11-27T00:38:22 because what I'm doing is not that far from what you're doing 2011-11-27T00:38:47 oh 2011-11-27T00:38:59 you don't try to predict enemy movements or consider different cases? 2011-11-27T00:39:09 I didn't say that... 2011-11-27T00:40:28 for each batttle partition i do defensive and offensive moves for my ants and opponents ants and pick my moves from a 2x2 game 2011-11-27T00:40:45 i'm not sharing all the info i use for picking those moves though 2011-11-27T00:40:50 *** AntDroid has quit IRC (Ping timeout: 265 seconds) 2011-11-27T00:40:54 a 2x2 game as in prisoner's dilemma style? 2011-11-27T00:41:20 *** AntDroid has joined #aichallenge 2011-11-27T00:41:25 well, i treat it more like a game tree where i have two moves and the opponents have 2 moves 2011-11-27T00:41:35 but i think you could pretty much represent it as either and get similar results 2011-11-27T00:41:39 I wonder if anyone is using the payoff matrix approach 2011-11-27T00:41:49 i don't give states numerical payoffs though, i return an Outcome which has an order operation 2011-11-27T00:42:02 Someone here earlier claimed they oculd do an exact solution over all 5^7+ cases of 6 enemies 2011-11-27T00:42:07 i store my outcomes in a payoff matrix 2011-11-27T00:42:15 but make the moves like it's a game tree 2011-11-27T00:42:35 i did full 1 ply minimax but it was fairly slow 2011-11-27T00:42:43 for battle partitions 2011-11-27T00:43:03 although i didn't try and reduce the size of partitions, each parition just had ants that could land in combat with each other 2011-11-27T00:43:18 if i wanted to get the attack radius squared of the current attack radius plus 2 is it 17? 2011-11-27T00:43:22 or land in battle with an ant that can land in battle with another ant and so on 2011-11-27T00:43:54 AntDroid: careful, the euclidean and manhatten metrics aren't equivalent 2011-11-27T00:44:12 antimatroid: that still makes me a little bit sad 2011-11-27T00:44:20 What you should do is rad2 + 4*rad1 + 4 2011-11-27T00:44:25 where rad1 is floor of sqrt 2011-11-27T00:44:39 bmh: me too, i really wish we'd have just gone with manhatten but euclidean does give nicer "circles" 2011-11-27T00:44:51 i don't like doing stuff like that 2011-11-27T00:44:56 well I think euclidena leads to move interseting tactics 2011-11-27T00:45:02 more interesting 2011-11-27T00:45:04 my advice: don't fuck around getting it wrong 2011-11-27T00:45:06 i just generate translation amounts during turn 0 for what i want 2011-11-27T00:45:14 I half-assed it on my combat and then when I fixed it had to totally re-trune 2011-11-27T00:45:36 my bot is 100% half-assed. I need to get something working again 2011-11-27T00:45:44 mine is a mess atm 2011-11-27T00:45:51 i'm not sure trying to generalise path finding was a good idea 2011-11-27T00:46:00 antimatroid: amen to that. 2011-11-27T00:46:04 there's so many little things that need to change depending on the type of path you want to find 2011-11-27T00:46:29 I probably should have just gone whole-hog and done APSP 2011-11-27T00:46:36 i'm currently using this: DistanceEuclidianSquared(row, col, newrow, newcol, Info) <= attRadius + 2 but I cant seem to get the nice circle, it includes some extra squares on the edges 2011-11-27T00:47:11 it looks fine 2011-11-27T00:47:15 except +2 is wrong 2011-11-27T00:47:27 speaking of things I half-assed 2011-11-27T00:47:57 doing it that way is also dangerous because you don't account for situations like ...a..%b.... 2011-11-27T00:48:01 they can't land in battle 2011-11-27T00:48:16 a.%.b can fight 2011-11-27T00:48:19 there's also some other bits of information you can take into account 2011-11-27T00:48:24 a..%b can't though 2011-11-27T00:48:31 but +2 is wrong because you want +2 manhattan not +2 euclidean 2011-11-27T00:49:04 you don't want to waste time considering that a as in a battle when it's not 2011-11-27T00:49:09 he could do other stuff if he's not needed there 2011-11-27T00:49:13 that a is in battle 2011-11-27T00:49:21 not this turn it's not 2011-11-27T00:49:36 BenJackson: I bounced this off a1k0n, but I'd like your feedback. My first thought was to find rows and columns of ants and treat each ant as a polya urn with a uniform prior. For a collection of ants (row/column/group/whatever) give them a common goal (East/West, toward a point, etc), reweight their urns and ask if the move is good under minimax 2011-11-27T00:49:43 if a steps one step closer to the water a and b will die 2011-11-27T00:49:54 If the move is good, update the urn, otherwise revert it. Rinse and repeat until it converges (if it ever goes) 2011-11-27T00:50:15 BenJackson: you just gave me a good idea 2011-11-27T00:50:24 randomly order my neighbours in each square 2011-11-27T00:50:30 so path finding doesn't favour a particular direction 2011-11-27T00:50:50 i can do it with a seed so debugging doesn't become impossible 2011-11-27T00:50:55 * BenJackson learns about polya urns :) 2011-11-27T00:51:05 rows and columns are advantageous because you only need to consider East/West and North/South, but I think (combatRadius + 1) might be the best organizing technique 2011-11-27T00:51:38 I haven't had any trouble writing solvers for this challenge 2011-11-27T00:51:52 the tricky thing is correctly expressing the "score" in an efficient way 2011-11-27T00:52:00 do you guys have one path finding function? 2011-11-27T00:52:06 or do you just rewrite for each use? 2011-11-27T00:52:22 antimatroid: yes, but it's broken. Hierarchical A* 2011-11-27T00:52:35 I don't pathfind except in the limited case of food 2011-11-27T00:52:43 which I do with plain bfs 2011-11-27T00:52:56 collecting food is a pita 2011-11-27T00:52:57 there's never really that much food visible to you 2011-11-27T00:53:04 because you don't actually want to move your ant onto the targets 2011-11-27T00:53:49 anyway, i'm going to program 2011-11-27T00:54:01 haha, strcat_flock tries to jump me, loses hill to drbraskbot: http://ants.fluxid.pl/replay.400 2011-11-27T00:54:07 these multi-multi-player games are so random 2011-11-27T00:55:30 Is it possible to store only results with nolaunch option ? 2011-11-27T00:57:17 I use BFS for finding distance to my hill and A* for distance to food 2011-11-27T00:57:31 I want to automate self-testing between multiple versions of my bot. Maybe there is already a tool for that ? 2011-11-27T00:59:43 besh: some people set up the website 2011-11-27T00:59:47 just like the aichallenge.org site 2011-11-27T00:59:53 or the tcp servers 2011-11-27T01:01:09 Yes. But I want to do some local testing. I am already playing on the servers 2011-11-27T01:01:22 you misunderstand 2011-11-27T01:01:29 *** asorman has quit IRC (Quit: Page closed) 2011-11-27T01:01:38 some people clone the git repo, set up the db and webserver, and run the whole thing local 2011-11-27T01:02:10 Oh I see. But I am not that knowledgabe to do that unfortunately. 2011-11-27T01:06:52 BenJackson: now that you've wikipedia'd urns -- does that strategy seem reasonable? 2011-11-27T01:07:51 (interesting similarity to particle filters, btw) 2011-11-27T01:08:40 your idea is vaguely along the lines of what I was trying to do for almost the entire time between my last two aichallenge submissions 2011-11-27T01:08:53 I came up with some mathematically clever things 2011-11-27T01:09:05 but then when I implemented it I realized that ants can only go 4 directions 2011-11-27T01:09:22 and at the single-step granularity it's impossible to attribute the actions of any ant to any overall goal 2011-11-27T01:09:45 so if you want one ant to go toward food, eg, it's likely that 1/4 of your ants will move "toward" that food (or even more) 2011-11-27T01:09:58 so you can't really weight *against* that because what, is everyone else going to stand still?? 2011-11-27T01:10:29 I'm going to use separate logic for foraging and fighting. Keep my fighters and foragers separate as best I can 2011-11-27T01:11:07 my conclusion was that if you wanted to do something (like a maximum weight matching or whatever) of goals and ants you'd have to be doing that optimization on many steps ahead of "now" 2011-11-27T01:11:22 where distictions in the actions of individual ants began to tell 2011-11-27T01:11:39 and I noodled about ideas for how to express that 2011-11-27T01:12:05 if I follow your "east/west/towards a point" idea then it would be to refine that to a point N steps away rather than just "east" or "towards" 2011-11-27T01:12:37 then maximize where I want everyone to be in N steps 2011-11-27T01:12:42 then have them all take the 1st step 2011-11-27T01:12:47 of course the search space goes through the roof 2011-11-27T01:12:55 oh, you look ahead more than one turn? wow! 2011-11-27T01:13:08 no, no, this was what I was trying to do 2011-11-27T01:13:14 I never got those ideas working 2011-11-27T01:13:43 I do look ahead in some ways, but I didn't do a "future goal" solver like I wanted to 2011-11-27T01:14:40 back in a while 2011-11-27T01:15:03 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-11-27T01:15:37 haha in multi-hill mazes I always end up staring at one of my hills waiting to see how it gets taken 2011-11-27T01:15:43 when it does I feel bad for however I messed up 2011-11-27T01:15:52 then I glance at the score and have 15 points 2011-11-27T01:15:58 *** Antimony has joined #aichallenge 2011-11-27T01:16:13 shiiiiiiiiiiiiiiiiiiit. not again. http://aichallenge.org/visualizer.php?game=133645&user=432 2011-11-27T01:18:09 this like never happens on the tcp server 2011-11-27T01:18:30 a1k0n: what arae you doing? lol 2011-11-27T01:18:40 my guess is he has code to avoid stepping on his own hill 2011-11-27T01:18:45 but wants to go south for that food 2011-11-27T01:19:00 probably something like that 2011-11-27T01:19:03 yes, but any kind of pathfinding would fix that problem 2011-11-27T01:19:04 and those two things cancel out 2011-11-27T01:19:19 if your anti-self-hill code is low level it will do that 2011-11-27T01:19:29 maybe.. 2011-11-27T01:19:33 eg most of my pathfinding considers my hills to be like water 2011-11-27T01:19:39 but i guess only a1k0n would know his own bot, lol 2011-11-27T01:19:47 however there are maps where you can step off your hill into a 1 square niche 2011-11-27T01:20:18 a1k0n: fix it and resubmit now: we can climb the ranks together 2011-11-27T01:20:38 i'm not sure why it didn't just go around the hill. it's perfectly capable. my guess is the extra food symmetrically placed to the left and right is somehow cancelling out the ability to go either way (and thus away from either food) 2011-11-27T01:21:13 it's shit like this 2011-11-27T01:21:18 and the crazy importance of the first few food 2011-11-27T01:21:23 that led me to do a bunch of work 2011-11-27T01:21:33 someone was talking about a TSP solver for the initial game 2011-11-27T01:21:35 it's not a bad idea 2011-11-27T01:21:55 1st turn, observe all food, fully plan everything (including expected ants) and make sure you get an optimal start 2011-11-27T01:22:01 nice, when i reproduce locally with the same bot it's sending the bot randomly east and west 2011-11-27T01:22:28 I had a test game at home where the only real difference between the winner and loser was the loser stepped "back" farther into his cave for the 1st food 2011-11-27T01:22:37 the winner went out and a later ant got the back food 2011-11-27T01:22:48 winner bot hit all the chokepoints first 2011-11-27T01:23:07 a1k0n: the replay data has the random seed 2011-11-27T01:24:03 *** Jak_o_Shadows has joined #aichallenge 2011-11-27T01:26:48 yeah its prolly worth optimising the very early game food collection 2011-11-27T01:27:33 haha watch all my ants at the verrry end of this: http://ants.fluxid.pl/replay.438 2011-11-27T01:27:49 you can tell without even watching any specific ants when I finally spot your hill 2011-11-27T01:28:10 haha 2011-11-27T01:28:17 that map alone has me seriously considering symmetry detection 2011-11-27T01:28:22 I'd have know where your hill was farr sooner 2011-11-27T01:28:36 yeah otherwise they take forever 2011-11-27T01:28:41 *** treeform has quit IRC (Remote host closed the connection) 2011-11-27T01:30:47 http://ants.fluxid.pl/replay.455 turn 306 around 65,13 2011-11-27T01:30:54 static defense you say 2011-11-27T01:30:55 hah! 2011-11-27T01:32:05 hmm, too many of my ants suicided there 2011-11-27T01:32:24 where? 2011-11-27T01:32:49 defending my first lost hill 2011-11-27T01:33:22 *** TheLinker has quit IRC (Quit: Bye) 2011-11-27T01:34:11 mmm, yes 2011-11-27T01:34:21 you have to watch the multi-hill maps so many times to see everything 2011-11-27T01:35:37 starting around 480 in the upper left you help me out 2011-11-27T01:35:54 suicide into yellow's defenders 2011-11-27T01:38:00 starting at 565 check out my dance with yellow at 26,133 2011-11-27T01:38:08 I had to go back and slow it down to see how I won 2011-11-27T01:38:24 that's awesome 2011-11-27T01:42:08 mostlly yellow killed himself? 2011-11-27T01:45:17 a blue ant sneaks in behind for the final kill 2011-11-27T01:45:41 there's a blue ant in the hallway 2011-11-27T01:46:32 why do you stop your ants from walkign on the hill? They might get stuck 2011-11-27T01:46:43 I only prevent walking on the hill if my bot is sure that it has stored food 2011-11-27T01:48:04 I never got around to estimating food storage 2011-11-27T01:49:29 anyway, I'm going to bed. Hopefully I can figure out how to get symmetry working tomorrow morning 2011-11-27T01:49:37 bye 2011-11-27T01:54:07 *** Antimony has quit IRC (Ping timeout: 248 seconds) 2011-11-27T02:21:55 *** besh has quit IRC (Quit: Page closed) 2011-11-27T02:22:59 *** epicmonkey has joined #aichallenge 2011-11-27T02:25:00 hmm... it probably isn't too late to abandon haskell 2011-11-27T02:25:18 * bmh considers violating the first commandment of Joel Spolsky 2011-11-27T02:28:28 thestinger: wow neither of us doing so hot at the end there http://ants.fluxid.pl/replay.472 2011-11-27T02:39:41 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-11-27T02:40:44 *** chris__0076 has quit IRC (Read error: Connection reset by peer) 2011-11-27T02:41:13 *** Chris_0076 has joined #aichallenge 2011-11-27T02:53:45 *** Palmik has joined #aichallenge 2011-11-27T02:55:07 *** sf17k has joined #aichallenge 2011-11-27T03:11:06 *** choas has quit IRC (Ping timeout: 244 seconds) 2011-11-27T03:22:54 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-11-27T03:22:54 *** Fandekasp has joined #aichallenge 2011-11-27T03:23:19 *** delt0r_ has joined #aichallenge 2011-11-27T03:33:12 http://code.recollect.com/post/13372390010/mechanical-turk-unshredder 2011-11-27T03:33:37 yay, we can watch tron games again 2011-11-27T03:34:24 http://tron.aichallenge.org/profile.php?user_id=2788 that was my awesome bot (that i wrote with a friend) which got disqualified :( 2011-11-27T03:39:56 why? 2011-11-27T03:40:18 timeout, 'eh? rough 2011-11-27T03:42:01 yeah 2011-11-27T03:42:03 super crap rule 2011-11-27T03:42:12 didn't even know of the rule until that happened 2011-11-27T03:44:09 it was barbaric to disqualify bots people spent weeks/months on just because of a timeout 2011-11-27T03:44:27 it was only weeks for tron 2011-11-27T03:45:20 iterated bfs to collect targets is quite slow 2011-11-27T04:04:59 *** Euphony has joined #aichallenge 2011-11-27T04:05:20 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-11-27T04:07:09 http://tcpants.com/replay.7578 2011-11-27T04:07:11 I love that map 2011-11-27T04:07:21 except that it ends like that a lot 2011-11-27T04:25:34 *** sigh has quit IRC (Remote host closed the connection) 2011-11-27T04:27:04 *** sigh has joined #aichallenge 2011-11-27T04:30:00 *** sigh has quit IRC (Remote host closed the connection) 2011-11-27T04:37:18 *** cyphase has quit IRC (Quit: http://www.cyphase.com/) 2011-11-27T04:40:46 *** cyphase has joined #aichallenge 2011-11-27T04:41:51 *** bmh has quit IRC (Quit: Page closed) 2011-11-27T04:55:19 *** antimatroid has left #aichallenge 2011-11-27T04:58:43 *** antimatroid has joined #aichallenge 2011-11-27T05:07:54 *** mj41 has joined #aichallenge 2011-11-27T05:13:10 *** UncleVasya has joined #aichallenge 2011-11-27T05:16:39 *** antimatroid has quit IRC (Ping timeout: 244 seconds) 2011-11-27T05:20:00 *** antimatroid has joined #aichallenge 2011-11-27T05:25:46 *** choas has joined #aichallenge 2011-11-27T05:28:31 *** kilae has joined #aichallenge 2011-11-27T05:49:18 *** Redgis has joined #aichallenge 2011-11-27T05:50:44 *** ikaros has joined #aichallenge 2011-11-27T05:52:06 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-11-27T06:02:46 *** Anilm3 has joined #aichallenge 2011-11-27T06:03:19 wow the servers are on fire today 2011-11-27T06:05:18 *** Conorach has joined #aichallenge 2011-11-27T06:07:53 *** pairofdice has joined #aichallenge 2011-11-27T06:12:47 *** HaraKiri has joined #aichallenge 2011-11-27T06:13:11 *** mcstar has joined #aichallenge 2011-11-27T06:24:35 *** g0llum has joined #aichallenge 2011-11-27T06:26:34 *** AlliedEnvy has joined #aichallenge 2011-11-27T06:30:35 *** olexs has joined #aichallenge 2011-11-27T06:31:27 *** Jak_o_Shadows has quit IRC (Remote host closed the connection) 2011-11-27T06:48:08 *** choas has quit IRC (Read error: Operation timed out) 2011-11-27T06:48:56 *** UncleVasya has quit IRC () 2011-11-27T06:51:19 *** grwip has joined #aichallenge 2011-11-27T07:20:38 *** Euphony has quit IRC (Quit: Page closed) 2011-11-27T07:48:06 *** kire has joined #aichallenge 2011-11-27T07:50:53 *** sigh has joined #aichallenge 2011-11-27T07:51:49 *** mj41 has quit IRC (Ping timeout: 244 seconds) 2011-11-27T07:53:23 whoa, game rates are high, there are 9 workers 2011-11-27T07:53:25 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-11-27T07:54:41 did anyone make png to map converter or something? 2011-11-27T07:55:57 you want to pixel-draw a png and convert it to a map? 2011-11-27T07:56:00 yeah 2011-11-27T07:56:15 if not i'll make one, but meyba someone already did this 2011-11-27T07:56:30 *** sf17k has quit IRC (Quit: Page closed) 2011-11-27T08:04:08 Fluxid: http://i.imgur.com/EX0z8.png 2011-11-27T08:04:14 whats wrong with this? 2011-11-27T08:04:58 mcstar: i want to see all 200x200 at once ;) 2011-11-27T08:05:21 use smaller font 2011-11-27T08:05:33 yep 2011-11-27T08:06:05 *** delt0r_ has joined #aichallenge 2011-11-27T08:06:14 ... 2011-11-27T08:06:49 Fluxid: of course do it if you want to, i just dont want you to waste your time 2011-11-27T08:07:05 pixel-editing and color-coding a map is harder than using vim i think 2011-11-27T08:07:29 it would be cool a map creator that automatically applied symmetry 2011-11-27T08:07:37 Editing a picture is pretty easy 2011-11-27T08:07:54 pairofdice: but you have to color-code the objects 2011-11-27T08:08:02 Not a problem 2011-11-27T08:08:03 so you have to set up a palette... 2011-11-27T08:08:14 i wouldnt do it, thats what im saying 2011-11-27T08:08:29 I would rather paint a map than edit one in text 2011-11-27T08:08:35 *** AntDroid_ has joined #aichallenge 2011-11-27T08:08:51 i won't color code. i just want to draw water and the rest do in editor 2011-11-27T08:09:17 Fluxid: code a map creator that applied different levels of symmetry/players 2011-11-27T08:10:00 do it online so everybody can create maps 2011-11-27T08:10:01 :D 2011-11-27T08:12:39 http://www.libpng.org/pub/png/libpng.html 2011-11-27T08:12:47 those guys take security seriously 2011-11-27T08:13:44 that's a lot of warnings 2011-11-27T08:13:57 yeah 2011-11-27T08:17:03 *** Antimony has joined #aichallenge 2011-11-27T08:21:42 http://paste.pocoo.org/show/513533/ here, working code of what i wanted ;P 2011-11-27T08:22:42 '' 2011-11-27T08:22:48 dont you need a ' ' 2011-11-27T08:22:50 ? 2011-11-27T08:22:54 *** AntDroid_ has quit IRC (Ping timeout: 265 seconds) 2011-11-27T08:23:12 or 'm ' 2011-11-27T08:23:29 i think theres a space after the m 2011-11-27T08:24:11 mcstar: believe me 2011-11-27T08:24:33 print 'a', 'b', in python 2 prints "a b 2011-11-27T08:24:37 " 2011-11-27T08:24:46 oh 2011-11-27T08:24:48 and ''.join() just joins an iterator with blanks 2011-11-27T08:24:52 so it concatenates chars 2011-11-27T08:25:18 thats weird, i mean the first one 2011-11-27T08:25:27 yeah, a bit 2011-11-27T08:25:44 python3 replaced print statement with function 2011-11-27T08:25:49 and you can define a joiner 2011-11-27T08:25:55 which is still a default ' ' 2011-11-27T08:31:56 whoa, suddenly my bot works horribly slow... 2011-11-27T08:32:04 no idea why 2011-11-27T08:38:02 or maybe it's because i pulled yesterday and something in engine slows down... 2011-11-27T08:38:28 *** smiley1983 has joined #aichallenge 2011-11-27T08:40:19 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-11-27T08:45:16 http://paste.aichallenge.org/9urYG/ lol 2011-11-27T08:46:23 *** foRei has joined #aichallenge 2011-11-27T08:48:30 something is bad, my current bot gives up on keeping area visible and wants to go to the other half of map 2011-11-27T08:48:47 *** kire has quit IRC (Remote host closed the connection) 2011-11-27T08:49:36 nice map 2011-11-27T08:49:40 where can I download it? 2011-11-27T08:50:12 one moment 2011-11-27T08:50:49 http://paste.aichallenge.org/zgCpr/ 2011-11-27T08:50:58 i've drawn it in gimp, lol 2011-11-27T08:51:03 thanks 2011-11-27T09:00:11 the single best idea in programming languages in general is the ability to comment out code 2011-11-27T09:02:32 *** Antimony has joined #aichallenge 2011-11-27T09:08:58 i still use my right hand to push B, soo annoying 2011-11-27T09:14:07 *** TheLinker has joined #aichallenge 2011-11-27T09:17:16 *** sigh has quit IRC (Remote host closed the connection) 2011-11-27T09:20:09 Ug, I hate implicit signedness comversions 2011-11-27T09:20:19 I did -75 % 150u and got 121 2011-11-27T09:33:49 *** mj41 has joined #aichallenge 2011-11-27T09:34:53 has anyone seen UncleVasya lately? a while ago he had a useful testing tool for Windows, I wonder if it is still being developed 2011-11-27T09:37:14 he is here often 2011-11-27T09:37:25 @seen unclevasya 2011-11-27T09:37:25 mcstar: unclevasya was last seen in #aichallenge 1 day, 22 hours, 19 minutes, and 32 seconds ago: on PlanetWars early rush was effective too. 2011-11-27T09:39:05 do you need this tool? 2011-11-27T09:39:36 if it works with current playgame.py, yes 2011-11-27T09:40:57 it should work 2011-11-27T09:41:00 email 2011-11-27T09:42:50 Dlayne: sent in pm 2011-11-27T09:42:50 but i faced a problem with his tool and created my onw, with blackjack and.. 2011-11-27T09:43:03 what problem? 2011-11-27T09:43:41 it crashes when i press "playgame" 2011-11-27T09:43:58 okay, that is a problem all right :) 2011-11-27T09:44:09 =) 2011-11-27T09:44:35 I used it a few times back in the beta, there were problems when playgame.py was changed a few times, and I don't think I ever used it since the start 2011-11-27T09:45:56 I should probably stop being lazy and write my own, too... 2011-11-27T09:49:43 UncleVasya dont have this issue. 2011-11-27T09:51:10 *** xathis has joined #aichallenge 2011-11-27T10:02:58 *** avdg has joined #aichallenge 2011-11-27T10:05:36 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-11-27T10:10:12 *** Anilm3 has joined #aichallenge 2011-11-27T10:11:58 *** sEagle has joined #aichallenge 2011-11-27T10:12:27 hello? 2011-11-27T10:12:36 *** sEagle has quit IRC (Client Quit) 2011-11-27T10:14:51 *** choas has joined #aichallenge 2011-11-27T10:15:11 *** Hexren has joined #aichallenge 2011-11-27T10:15:13 *** epicmonkey has joined #aichallenge 2011-11-27T10:15:13 *** Hexren has left #aichallenge 2011-11-27T10:18:32 *** NoxiaZ^ has joined #aichallenge 2011-11-27T10:27:36 *** Ruuhkis has joined #aichallenge 2011-11-27T10:29:36 Anyone feeling like giving a eye on my bots pathfinder that i cant figure whats wrong with (path gets generated (tried on another test class) but ant doesn't seem to move through the path no matter what :( I've been debugging it for a day or so, if someone smarter could have a eye on it? 2011-11-27T10:30:37 Its a* btw, I have had experience with it inbefore and it works as i said but my ants doesn't like it :( 2011-11-27T10:37:06 you can try the new visualizer 2011-11-27T10:37:12 http://forums.aichallenge.org/viewtopic.php?f=25&t=1633 2011-11-27T10:37:16 it may help you debug 2011-11-27T10:37:49 I maybe got it 2011-11-27T10:38:29 It seems like every turn the path gets reset.. So the tile in myAnts() ins't the ant actually, its just the tile that the ant is on? so if i give tile a path its not on the ant, its on the tile? 2011-11-27T10:38:44 *** mj41 has quit IRC (Ping timeout: 260 seconds) 2011-11-27T10:42:36 ooh, debug visualizer 2011-11-27T10:49:19 I dont undestand, even if i generate new path, in end of the turn its null.. :l 2011-11-27T10:49:23 *** Antimony has quit IRC (Ping timeout: 244 seconds) 2011-11-27T10:51:41 *** olexs1 has joined #aichallenge 2011-11-27T10:52:24 Is anybody going to watch final f1 gp? 2011-11-27T10:52:53 *** olexs has quit IRC (Ping timeout: 252 seconds) 2011-11-27T11:04:31 *** Antimony has joined #aichallenge 2011-11-27T11:09:07 not atm, but its just a channel hop away here 2011-11-27T11:10:02 Darn it, my symmetry detection isn't working 2011-11-27T11:11:46 :/ 2011-11-27T11:11:55 darn my pathfinding isnt working 2011-11-27T11:12:03 every turn the path magically sets to null 2011-11-27T11:12:19 It's not magical if you set it to null 2011-11-27T11:12:24 Ruuhkis: which pathfinder are you using? 2011-11-27T11:12:52 avdg I am using a* 2011-11-27T11:13:03 and i dont know how to pm here if that was a PM you sent 2011-11-27T11:13:10 hmm, did you implement bfs before? 2011-11-27T11:13:19 bfs? ;o 2011-11-27T11:13:24 pfs? 2011-11-27T11:13:30 breath first search 2011-11-27T11:13:35 ooh, nah 2011-11-27T11:13:48 if you can implement that one, extend a little bit, you have a, a* 2011-11-27T11:13:52 *an 2011-11-27T11:13:54 but the path got generated etc. but all the sudden its null 2011-11-27T11:14:11 How do you know if it got generated 2011-11-27T11:14:15 debug 2011-11-27T11:14:22 it prints clearly that it got the path 2011-11-27T11:14:33 and then when its the next turn 2011-11-27T11:14:34 it says its null :( 2011-11-27T11:14:39 and generates new.. 2011-11-27T11:14:45 obviously you dont understand the starter bot if you are using it 2011-11-27T11:14:57 so whats the problem ?:( 2011-11-27T11:15:08 we don't have any code or debug information 2011-11-27T11:15:10 and yes I am using it 2011-11-27T11:15:10 It's a bit hard to say 2011-11-27T11:15:23 i show you debug info and i can upload code too :) 2011-11-27T11:15:24 second 2011-11-27T11:15:44 * avdg doesn't know if it would be smart to upload the code 2011-11-27T11:16:01 keep it small :p 2011-11-27T11:16:03 http://pastebin.com/e6Ax4S59 2011-11-27T11:16:06 Well, if it's not a top100 bot 2011-11-27T11:16:17 LOL ;) 2011-11-27T11:16:31 the ( number ) is the "id" of ant 2011-11-27T11:16:43 typo overlaod 2011-11-27T11:16:59 lol, i wrote ramdom glibbish 2011-11-27T11:17:10 gibberish* 2011-11-27T11:17:59 hmm, is it fixed? 2011-11-27T11:18:04 nope :( 2011-11-27T11:18:09 if you read the debug 2011-11-27T11:18:12 it says in end of turn 2011-11-27T11:18:15 that it has like 3 tiles on path 2011-11-27T11:18:16 (well, maybe it would be ok if you add the path in your debug information) 2011-11-27T11:18:22 but on next turn of that ant it says its null 2011-11-27T11:18:38 but how can it turn into null anyway? but i can ad :3 2011-11-27T11:18:40 add* 2011-11-27T11:19:19 it smells like you still using the wrong variables 2011-11-27T11:19:33 as return information 2011-11-27T11:20:13 http://pastebin.com/XyPKwAm0 2011-11-27T11:20:23 i can show you code 2011-11-27T11:20:25 ALL the code 2011-11-27T11:20:27 its nothing much 2011-11-27T11:20:32 just basic 2011-11-27T11:20:59 how did you get the number of steps? 2011-11-27T11:20:59 its java code anyway 2011-11-27T11:21:00 http://pastebin.com/UyLz7YvL 2011-11-27T11:21:06 list.size 2011-11-27T11:21:20 hmm 2011-11-27T11:21:24 http://pastebin.com/Bjry7Sji 2011-11-27T11:21:27 have a look 2011-11-27T11:21:53 I don't directly see any return value 2011-11-27T11:22:08 "return path;" 2011-11-27T11:22:15 and it also prints it was there :o 2011-11-27T11:22:19 not in the paste 2011-11-27T11:22:23 that debug info is useless 2011-11-27T11:22:27 what are you looking at? 2011-11-27T11:22:34 what should i debug mr mcstar? 2011-11-27T11:22:51 "name of variable" - "value of variable" 2011-11-27T11:22:56 Ruuhkis: just do a search on "return" 2011-11-27T11:22:59 and no nonsense 2011-11-27T11:23:10 avdg - where you made that search? 2011-11-27T11:23:18 on the pastebin 2011-11-27T11:23:21 avdg the path finding code is on the second pastebin i sent 2011-11-27T11:23:23 pathfinding* 2011-11-27T11:23:30 ah 2011-11-27T11:23:34 lol :) 2011-11-27T11:23:34 srr 2011-11-27T11:23:42 all good 2011-11-27T11:24:17 and sorry for messy code, bad debug etc. I only have 2yr of programming experience :e trying to learn through this :3 2011-11-27T11:26:17 it looks fine 2011-11-27T11:26:21 yeah.. 2011-11-27T11:26:26 maybe an issue with the code that uses it 2011-11-27T11:26:33 hmm? 2011-11-27T11:26:37 how could it null my path :@ 2011-11-27T11:26:37 lol 2011-11-27T11:26:50 maybe if i made list of pathes? 2011-11-27T11:27:01 what if you write a small test? :p 2011-11-27T11:27:16 what kind of test you're talking about ?:o 2011-11-27T11:27:25 (meh, putting a map into a test would be not easy) 2011-11-27T11:27:27 unit test 2011-11-27T11:27:50 load map, mark 2 points, get path 2011-11-27T11:29:06 but the map works, the problem is somehow it seems to "forget" the path.. :( I am trying to give each ant unique id and then make map that contains path for each id and then.. I'll see what happens and report back! :D 2011-11-27T11:29:30 Half of that code is error printing :) 2011-11-27T11:29:32 yeah, you just have to test if the path is longer than 1 2011-11-27T11:29:48 dependent on what you expect 2011-11-27T11:31:30 are you trying to assign paths permanently? 2011-11-27T11:31:42 are you assigning id's to ants each turn? 2011-11-27T11:35:51 Ruuhkis: instead of passing a new object, what would be the result if you put that piece of code in 2 lines? 2011-11-27T11:36:16 mcstar, yes I am 2011-11-27T11:36:26 what is happening to ants in each turn for real? :o 2011-11-27T11:36:30 O___O 2011-11-27T11:36:35 I already got in problems in c++ by passing an object with a new keyword, but there could be other issues as well (and you can add more debug information) 2011-11-27T11:36:35 they cease to exist 2011-11-27T11:36:37 I am lost *taking a look at superclass* 2011-11-27T11:36:58 Ruuhkis: the starter packs dont preserve state 2011-11-27T11:37:00 oh my god.. 2011-11-27T11:37:05 at least i dont think so 2011-11-27T11:38:41 oh noes.. 2011-11-27T11:39:28 Doesn't stop you from saving state though 2011-11-27T11:39:45 Ug, my bot decided to hallucinate immaginary food 2011-11-27T11:40:11 this explains a lot 2011-11-27T11:40:12 Antimony: i think it hallucinates REAL food no? 2011-11-27T11:40:21 glad your bot is thinking positive Antimony 2011-11-27T11:40:22 lol 2011-11-27T11:40:46 what did you feed your bot with? wow 2011-11-27T11:40:53 shrooms 2011-11-27T11:41:00 ants 2011-11-27T11:41:03 why do it clear the state every turn? just wondering? 2011-11-27T11:41:12 Ruuhkis: why not? 2011-11-27T11:41:21 :( because i lose their states too! :D 2011-11-27T11:41:23 why dont you read the source before you are going mad? 2011-11-27T11:41:27 because its more complicated to track the changes between every turn? 2011-11-27T11:41:39 doesn't mean you don't have to do that 2011-11-27T11:42:37 hmm.. 2011-11-27T11:43:04 what should i do? what should i do? *tears hair out of head* 2011-11-27T11:43:38 throw things you don't need away until you have 1 thing to focus at? 2011-11-27T11:44:06 and with throwing away things, I don't mean code 2011-11-27T11:44:32 Girlfriend, job 2011-11-27T11:44:45 :p 2011-11-27T11:44:51 ohh wow 2011-11-27T11:44:57 All this time it turns out there was a real food there 2011-11-27T11:45:01 my bot isn't crazy after all 2011-11-27T11:45:02 :D 2011-11-27T11:45:26 I never realized it was possible for food to spawn right next to the hill like that 2011-11-27T11:46:37 sooo.. should i make it so each tile has its path and take the path from the tile map which state doesnt get cleared to the and on the tile or just modify how they're updated ? :( 2011-11-27T11:47:08 Wow 2011-11-27T11:47:23 what have you guys done with ur pathfinders? :o 2011-11-27T11:47:26 My current bot has already beaten my previous bot's skill, and it's only played 18 games instead of 80 2011-11-27T11:47:37 I don't really do pathfinding 2011-11-27T11:47:42 well 2011-11-27T11:47:46 I don't store paths 2011-11-27T11:47:59 Me neither, also, mine is broken 2011-11-27T11:48:03 you generate new path each turn? 2011-11-27T11:48:12 I don't generate a path at all 2011-11-27T11:48:17 It's implicit in the distance map 2011-11-27T11:48:30 You only need the first step after all 2011-11-27T11:49:12 I don't understand how can you avoid obstacles without having to walk there and here 100 times without generating path? 2011-11-27T11:49:26 What 2011-11-27T11:49:44 A journey of 20000 squares begins with a single step :) 2011-11-27T11:49:55 lol 2011-11-27T11:50:16 true :) 2011-11-27T11:50:29 so pathfinder is a bad idea? 2011-11-27T11:50:37 Well, it's still a path 2011-11-27T11:50:39 bad and good 2011-11-27T11:50:40 Antimony: wtf? 2011-11-27T11:50:51 i should just let my ant guess? 2011-11-27T11:51:07 well, mine gets rank 1600 for doing that 2011-11-27T11:51:17 I'm just pointing out that there's no need to store paths. You only need the distance, and then you can use graident descent 2011-11-27T11:51:20 it may actually perform a little bit better 2011-11-27T11:51:38 (which is still my js implementation) 2011-11-27T11:51:47 My ants are too stupid to calculate partial derivates 2011-11-27T11:51:51 *looks at wikipedia article* 2011-11-27T11:52:30 Antimony: you need the path if you are looking for an optimal path, notwithstanding how many step you are using from it 2011-11-27T11:52:34 Ug, my symmetry detector still refuses to work 2011-11-27T11:52:48 IF you have the distance to each square, the path is implicit. 2011-11-27T11:53:07 On each step, you go the direction that decreases distance 2011-11-27T11:53:08 ? 2011-11-27T11:53:27 you have all the minimal distance from every point to every point? 2011-11-27T11:53:29 yeah but what if you're next to looon wall 2011-11-27T11:53:31 looong* 2011-11-27T11:54:25 What Antimony is saying that pepper-spray is a food product, essentially 2011-11-27T11:54:41 no 2011-11-27T11:54:54 you only need the minimal distance from any dest to your ant 2011-11-27T11:55:03 how much you store is up to you 2011-11-27T11:55:11 I know, I'm doing the same thing, but it's still pathfinding 2011-11-27T11:55:21 Antimony: you have to do a bfs for that 2011-11-27T11:55:24 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-11-27T11:55:26 OR A* 2011-11-27T11:55:29 if you backtrack or not thats your own business 2011-11-27T11:55:46 but it IS pathfinding 2011-11-27T11:55:56 I know 2011-11-27T11:56:03 I'm jsut telling how I handle pathfinding 2011-11-27T11:56:20 then dont confuse Ruuhkis he's got enough problems on his own 2011-11-27T11:56:33 lol :D 2011-11-27T11:56:53 Like crazy variable names 2011-11-27T11:56:54 I am going to make my very own mixture :) 2011-11-27T11:58:21 *** raemde_ has joined #aichallenge 2011-11-27T12:01:25 *** raemde has quit IRC (Ping timeout: 240 seconds) 2011-11-27T12:04:02 *** besh has joined #aichallenge 2011-11-27T12:05:23 *** treeform has joined #aichallenge 2011-11-27T12:06:00 *** NoxiaZ^ has quit IRC (Ping timeout: 260 seconds) 2011-11-27T12:08:26 *** Antimony has quit IRC (Ping timeout: 244 seconds) 2011-11-27T12:09:06 *** ThatSnail has joined #aichallenge 2011-11-27T12:13:56 *** Antimony has joined #aichallenge 2011-11-27T12:18:15 *** choas has quit IRC (Ping timeout: 244 seconds) 2011-11-27T12:24:01 cell_maze_p02_05.map is hard 2011-11-27T12:26:28 My ants cluster on that map for no reason. At first I supsected a bug in my path finding but later I saw other bots get in the same problem. 2011-11-27T12:28:44 it's just hard to discover enemy's hill 2011-11-27T12:28:54 because you need to enter the cave to see it 2011-11-27T12:30:31 yes with those thick walls 2011-11-27T12:30:36 waters* 2011-11-27T12:33:55 Hopefully the symmetry I'm working on will help with that 2011-11-27T12:34:27 Antinomy, what are you trying to do about symmetry ? 2011-11-27T12:35:37 I'm trying to add symmetry detection to my bot 2011-11-27T12:35:42 so it can find hills without ever having seen them 2011-11-27T12:35:56 I've been debugging it all mornign but it still isn't working 2011-11-27T12:36:23 is it even possible to have rotational symmetry on a nonsquare map? 2011-11-27T12:37:21 i don't think so 2011-11-27T12:38:07 a erctangle can be just a projection of a square, some transformation inbetween 2011-11-27T12:38:44 so you can't guess from the shape 2011-11-27T12:39:58 hmm 2011-11-27T12:40:10 I think it is impossible to have rotation on rectanuglar maps in normal circumstances 2011-11-27T12:40:15 that will simplifiy things immensly 2011-11-27T12:42:21 Is food also placed symmetrically ? 2011-11-27T12:42:27 *** Ruuhkis has quit IRC (Ping timeout: 265 seconds) 2011-11-27T12:42:43 *** amstan has joined #aichallenge 2011-11-27T12:42:43 *** ChanServ sets mode: +o amstan 2011-11-27T12:43:37 *** jstrong has joined #aichallenge 2011-11-27T12:44:46 *** Antimony has quit IRC (Ping timeout: 244 seconds) 2011-11-27T12:50:26 besh: depends on map, but usually yes 2011-11-27T12:50:49 so each player has food in same locations around the hive 2011-11-27T12:57:18 *** cyphase has quit IRC (Ping timeout: 240 seconds) 2011-11-27T13:05:28 My new combat works. Here is how you beat green_tee http://ants.fluxid.pl/replay.1137 2011-11-27T13:06:22 *** magnum696 has joined #aichallenge 2011-11-27T13:08:32 he had 700 ants in the buffer 2011-11-27T13:09:04 besh: that wasn't really beating 2011-11-27T13:09:27 Can't a poor bot enjoy himself :) 2011-11-27T13:09:49 besh: nice otherwise, but the forgot to raze hills somehow 2011-11-27T13:09:55 the-yher 2011-11-27T13:09:55 *** cyphase has joined #aichallenge 2011-11-27T13:10:00 the -> he 2011-11-27T13:10:29 *** dr- has quit IRC (Ping timeout: 245 seconds) 2011-11-27T13:10:36 I so wanted to have a code like top bots use for confrontation. I think it now has a jist of it. 2011-11-27T13:11:35 Earlier it just goes in based on analysis of a single enemy ant. do or die. 2011-11-27T13:18:43 *** tassadarq has joined #aichallenge 2011-11-27T13:22:30 *** dr- has joined #aichallenge 2011-11-27T13:22:59 *** glhf has joined #aichallenge 2011-11-27T13:24:28 *** pvpv4001 has joined #aichallenge 2011-11-27T13:26:07 *** choas has joined #aichallenge 2011-11-27T13:26:15 is it just me or are some tcp games timing out randomly? 2011-11-27T13:26:42 it's not just you, i've seen that often 2011-11-27T13:27:07 my bot flushes his output with well more than 4s time remaining, and still gets a time out message 2011-11-27T13:27:33 didn't happen yesterday, happening all the time since about an hour now 2011-11-27T13:27:55 huh. seems to be working ok for me lately 2011-11-27T13:28:24 <_flag> olexs1: Happens to me all the time, I'm pretty sure it's something wrong with the tcp server 2011-11-27T13:28:38 oleks1, that's tcp ? and like sunday evening on germany, everyone and his mothere online ? 2011-11-27T13:28:53 seems to be the connection, either on my end or server end, since pings are rubbish as well 2011-11-27T13:29:40 and that'll be mine, since pings to google are just as bad... hm. 2011-11-27T13:29:49 *** glhf has quit IRC (Ping timeout: 265 seconds) 2011-11-27T13:30:28 g0llum: I'm in germany too, pings to ants.fluxid.pl are 20-30 ms usually, but now about 400 and with some packet loss even :) 2011-11-27T13:32:09 at least we both agree, it's more a local networking problem 2011-11-27T13:32:13 ok, very wtf, google pings on another pc in my house on the same connection - 25ms. I think I'll just reboot 2011-11-27T13:32:26 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-11-27T13:32:28 oh 2011-11-27T13:32:38 <_flag> Have the food spawn rates been made much higher recently? 2011-11-27T13:32:52 <_flag> I'm getting more and more games that end in stalemates due to really high ant counts 2011-11-27T13:35:25 *** Antimony has joined #aichallenge 2011-11-27T13:35:36 *** magnum696 has quit IRC (Quit: Page closed) 2011-11-27T13:36:21 woah, pguillory switched languages (C -> Go) 2011-11-27T13:36:42 hehe 2011-11-27T13:38:06 *** tassadarq has quit IRC (Quit: Page closed) 2011-11-27T13:39:00 hehe he will lose his edge 2011-11-27T13:39:16 you know, his go source was compiled as c source up to know 2011-11-27T13:39:24 that was his "bug" 2011-11-27T13:39:44 yeah, another strong one in that league, that's the down thing ;) 2011-11-27T13:41:08 if i end up whithin the top ten go bots, i'm happy. 2011-11-27T13:44:09 http://en.wikipedia.org/wiki/Gobots 2011-11-27T13:46:43 a1k0n: holy shit, i had one of these!!! 2011-11-27T13:49:05 haha. i had some tranformers. gobots were the lame knockoff 2011-11-27T13:49:05 *** mj41 has joined #aichallenge 2011-11-27T13:50:53 *** Surya has joined #aichallenge 2011-11-27T13:51:23 http://www.myspace.com/arcadeactiongraphics/photos/11447760, passing by there every day 2011-11-27T13:51:53 i read books when i was little, and fixed anything that wasnt broken 2011-11-27T13:52:47 *** Garf has joined #aichallenge 2011-11-27T13:52:54 Hmmm 2011-11-27T13:53:06 My test implementation of symmetry isn't terribly effective 2011-11-27T13:53:13 even after 100 turns it still has no clue 2011-11-27T13:53:57 a1k0n: yeah, that wasn't a great toy, but haven't heard of tranformers back then 2011-11-27T14:00:56 i never had "cool" toys, probably thats why im hung up on this ants thing 2011-11-27T14:05:00 society and toys < computers 2011-11-27T14:06:18 i didn't really have a lot of toys either, once i got a computer capable of running BASIC 2011-11-27T14:06:33 my kid, OTOH, is spoiled rotton 2011-11-27T14:07:25 i lost interest in toys when i found out about qbasic in msdos 6 on our computer 2011-11-27T14:09:59 *** gazpachoking has joined #aichallenge 2011-11-27T14:10:09 i was almost grown up when i had my computer 2011-11-27T14:10:13 *** nathanr has joined #aichallenge 2011-11-27T14:10:20 ~16-17 2011-11-27T14:10:56 did fluxid get reset again? 2011-11-27T14:11:04 so no surprise if you beat me guys :D 2011-11-27T14:11:15 Garf, yes, yesterday 2011-11-27T14:12:17 *** kire has joined #aichallenge 2011-11-27T14:14:34 a friend on my school gaved me qbasic books 2011-11-27T14:14:49 *** NoxiaZ^ has joined #aichallenge 2011-11-27T14:14:50 and I had a broken computer (which helped as well) 2011-11-27T14:16:10 avdg, as in: having low expectations, and high pleasure, whe something atually works ? 2011-11-27T14:16:42 well, it had so many problems nobody want to use it, but I use it for experimenting 2011-11-27T14:18:41 well, my high-school had a book explaining all the circuits of the C64, i borrowed it multiple times, that complexity just amazed me 2011-11-27T14:19:16 i wanted a pc, but it was expensive, and my mom thought that she must buy a real good one for me 2011-11-27T14:19:27 :p 2011-11-27T14:19:31 so i got mine years after i started wanting one, but it was real good 2011-11-27T14:19:39 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-27T14:19:47 my parents are too lazy atm, they think that they get better by sleeping 2011-11-27T14:20:02 *** g0llum has joined #aichallenge 2011-11-27T14:20:21 (but it was going instead leading to more boring things) 2011-11-27T14:20:31 i've got my own computer, because i was too annoying to my parents after they banned computers, because i formatted the disk of computer dad used for business 2011-11-27T14:20:43 :p 2011-11-27T14:20:49 Fluxid: lucky you 2011-11-27T14:21:26 i think my dad had some idea about what a computer was 2011-11-27T14:21:30 i still wonder why did they do that, i think i would go back to my toys eventually 2011-11-27T14:21:51 can you ask them? 2011-11-27T14:21:56 1992 some friend had no further use for borland c++, 16 disks + a 1m60 pile of books 2011-11-27T14:21:59 my dad can't work with a computer, even after 20 years :p 2011-11-27T14:22:29 (and is still taking big effort in learning to) 2011-11-27T14:22:40 sweet '92, my first year in elementary 2011-11-27T14:22:53 my parents still mistake double/single/right clicks 2011-11-27T14:23:21 my dad can't remember passwords/locations 2011-11-27T14:23:28 (file locations) 2011-11-27T14:23:33 I can't remember passwords either 2011-11-27T14:23:38 ok, my python bot has 600 lines of C code 2011-11-27T14:23:44 i wonder is that okay 2011-11-27T14:23:52 well, he confuses them 2011-11-27T14:23:52 my C++ bot has 0 lines of python ;-) 2011-11-27T14:24:11 Fluxid: is it more than the python locs? 2011-11-27T14:24:25 mcstar: not yet 2011-11-27T14:24:57 mcstar: i have 1196 lines of python code... oh boy 2011-11-27T14:25:09 Fluxid: is it all in one file? 2011-11-27T14:25:17 so in my game against ifdef he got ahead of me while we mopped up the weaker bots (7-5) but then I was about to take his hill and killed every one of his ants. The visualizer shows a bonus of "2" (so, 7-7) but I think the engine properly deducted his hill (so 7-6) because I'm alone in 1st 2011-11-27T14:25:20 amstan: fortunately not 2011-11-27T14:25:22 600 of c is pretty much id say 2011-11-27T14:25:24 Fluxid: good, lol 2011-11-27T14:25:25 is that just a bug in the visualizer? or in the replay output? 2011-11-27T14:25:29 *** choas has quit IRC (Ping timeout: 245 seconds) 2011-11-27T14:26:30 2611 lines of C++ 2011-11-27T14:26:56 and pretty dense, too, not full of my usual explanatory comments 2011-11-27T14:27:06 5290 wtf 2011-11-27T14:27:20 I wrote a C++ game that has python in it 2011-11-27T14:27:45 im doing a cat *.cpp *.hh|wc -l 2011-11-27T14:27:55 same here 2011-11-27T14:28:02 it was ~2600 yesterday and 5290 now 2011-11-27T14:28:03 well, not counting unit tests 2011-11-27T14:28:16 My bot decided that there can't possible be any enemies as soon as it sees it's first hill :( 2011-11-27T14:28:24 mcstar: how did you write that much loc? 2011-11-27T14:28:29 1699 here 2011-11-27T14:28:35 avdg: not sure 2011-11-27T14:28:40 mine is much less :/ 2011-11-27T14:28:54 close to 1k lines? 2011-11-27T14:29:03 a1k0n: is your fluxid bot in sync with your aichallenge submit? 2011-11-27T14:29:16 since resubmitting I'm trying to guess where I'll end up :) 2011-11-27T14:29:33 BenJackson: what do you mean? 2011-11-27T14:29:49 nope 2011-11-27T14:29:54 it's ahead 2011-11-27T14:30:08 "ahead", ah, I'm all to familiar with that ;-) 2011-11-27T14:30:12 Fluxid: ? 2011-11-27T14:30:16 well, it's about 1 skill point better 2011-11-27T14:30:22 (on fluxid's rankings) 2011-11-27T14:30:24 huh, ok, i have a base.1.cpp and base.off.cpp aka base.cpp, poor man's version control.... 2011-11-27T14:31:08 i'm really frustrated with how terrible my combat is, and at the same time surprised such a stupid approach works so well 2011-11-27T14:31:28 a1k0n: your bot takes the initiative 2011-11-27T14:31:46 yeah, and bots like pguillory just wait for me to kill myself on their wall of ants 2011-11-27T14:31:54 my ant is in denial about the existence of enemies 2011-11-27T14:32:05 probably better off that way 2011-11-27T14:33:43 yeah, my previous submit (v7) was aggressive/blundering sort of combat 2011-11-27T14:33:52 smart enough to beat or trade evenly in most cases 2011-11-27T14:34:05 but against "real" combat it would suicide 2011-11-27T14:34:35 after I fixed that I had to re-tune everything to regain any aggression 2011-11-27T14:34:55 i haven't tried tweaking my constants in a while 2011-11-27T14:35:06 still I just watched a game against 500ish ranked bots on aichallenge where I carefully got out of the way of a blundering ant 2011-11-27T14:35:16 which I probably should have just suicided to get out of my way 2011-11-27T14:36:09 aham 2011-11-27T14:36:12 "constants" 2011-11-27T14:36:17 thats whats missing from my bot 2011-11-27T14:36:31 *** nathanr has quit IRC (Ping timeout: 265 seconds) 2011-11-27T14:36:41 * pairofdice gives mcstar some pi 2011-11-27T14:36:53 i love pi 2011-11-27T14:36:56 the movie too 2011-11-27T14:37:02 i'd be amazed if your code has no constants whatsoever 2011-11-27T14:37:10 a1k0n: not like yours 2011-11-27T14:37:11 a1kOn: So true about pguillory. That is what I just implemented and it is starting to play like it. 2011-11-27T14:37:37 besh: you implemented the patient wall of ants? 2011-11-27T14:37:42 and no i dont have constants, only those that the engine provides 2011-11-27T14:37:55 yes I just form a wall of ants at attack_radius + 2 2011-11-27T14:38:01 ah 2011-11-27T14:38:08 I was thikning my bot nearly broke this seige: http://aichallenge.org/visualizer.php?game=135711&user=3473 2011-11-27T14:38:13 then I saw how many ants yellow had in his hive 2011-11-27T14:38:17 and they come and collide. But before I used to take the initiative 2011-11-27T14:38:18 no wonder my assault failed 2011-11-27T14:39:25 a1k0n: you said you can tune things, i cant 2011-11-27T14:39:42 i guess you have some sort of linear combination of things 2011-11-27T14:39:52 i dont want to go that way 2011-11-27T14:40:14 have you put that q-learning stuff into your bot? 2011-11-27T14:40:22 no. 2011-11-27T14:40:48 i wondered if i could use it to teach my bot whether an enemy is behaving retionally in battle 2011-11-27T14:41:05 a* 2011-11-27T14:43:05 theres this other thing i came across 2011-11-27T14:43:10 random forests 2011-11-27T14:43:40 pretty interesting, but i havent figured out how to put that into good use 2011-11-27T14:43:53 another thing that might be interesting is the SVM 2011-11-27T14:44:22 i guess you know about these 2011-11-27T14:44:23 *** nathanro11 has joined #aichallenge 2011-11-27T14:44:45 hidden markov models or particle filters would be good ways to measure enemy behavior 2011-11-27T14:45:08 *** Accoun has quit IRC () 2011-11-27T14:45:15 svm can do that i think 2011-11-27T14:45:31 How about particle filters to predict enemy hills 2011-11-27T14:45:32 and theres some ultra-flexible implementation out there with non-linear kernels 2011-11-27T14:45:38 I think svm is overkill 2011-11-27T14:45:54 SVM would be able to learn about patterns of moves and quickly classify a bot (given that you had lots of prior knowledge) 2011-11-27T14:46:08 you can use much simpler stuff if you just want to reason from nothing in each new game 2011-11-27T14:46:29 so you are saying that i could train it on tcp 2011-11-27T14:46:56 and it would be faster to actually upload the training data, then letting it train itself in every new game 2011-11-27T14:47:09 faster->better idk 2011-11-27T14:47:39 a SVM learns to classify examples 2011-11-27T14:48:09 a HMM would be a simple matter of the hidden probability of rationality of the other bot 2011-11-27T14:48:14 yeah, its a binary classifier in a many dimensinal space 2011-11-27T14:48:14 which you observe as moves 2011-11-27T14:48:38 in a HMM you would look at how rational a bot is each turn as you encounter it 2011-11-27T14:48:57 and the model assumes that your observation of rationality is a product of the hidden *actual* rationality 2011-11-27T14:49:01 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-27T14:49:03 idk about that, my idea was to see if the enemy follows my rational evaluation function or not 2011-11-27T14:49:08 I think it would be pretty easy to use particle filter to predict enemy hill locations 2011-11-27T14:49:17 *** g0llum has joined #aichallenge 2011-11-27T14:49:20 oh YOUR function 2011-11-27T14:49:40 pairofdice: well you can just evaluate map symmetry 2011-11-27T14:49:51 I implemented "hill intuition" if you like 2011-11-27T14:50:01 the problem was I would "discover" hills too early 2011-11-27T14:50:11 pairofdice: you mean guessing the hill positions from the motion of the enemy? 2011-11-27T14:50:25 my bot was actually better off only knowing about hills as it expanded and in the order it found them 2011-11-27T14:50:49 knowing about a hill on the far end of the map just meant my ants wanted to cross enemy territory to make a remote assault befroe taking care of business at home 2011-11-27T14:51:01 :) 2011-11-27T14:51:23 it would make sense on big maps with few players 2011-11-27T14:51:35 heh 2011-11-27T14:53:30 pairofdice: would you pls answer 2011-11-27T14:56:30 No, not from movement 2011-11-27T14:57:34 pairofdice: im just not familiar with these methods so im wondering how would you use it... 2011-11-27T14:57:53 ofc i understand if its a state secret 2011-11-27T14:59:25 http://www.youtube.com/watch?v=H0G1yslM5rc 2011-11-27T14:59:43 From ai class 2011-11-27T15:00:11 yeah, i recognize sebastian's voice 2011-11-27T15:00:48 the vovot is ... 2011-11-27T15:02:09 k thanx, im gonna watch it 2011-11-27T15:05:32 pairofdice: aham 2011-11-27T15:05:43 what was the purpose of the robot 2011-11-27T15:05:59 he says it "doesnt know" where it is 2011-11-27T15:06:11 and it gets into a room, so it now does? 2011-11-27T15:06:25 Slight variations from symmetry in the room 2011-11-27T15:06:56 *** Palmik has quit IRC (Remote host closed the connection) 2011-11-27T15:07:20 pairofdice: can you be more verbose? 2011-11-27T15:07:32 i dont see what did the robot try to accomplish 2011-11-27T15:07:46 It just tried to locate itself with sonar 2011-11-27T15:08:05 doesnt make sense to me 2011-11-27T15:08:08 And the hallway was perfeclty symmetrical so it had to move to a room to resolve it's position 2011-11-27T15:08:09 sorry :D 2011-11-27T15:08:28 It's not that complicated 2011-11-27T15:08:53 theres something that is a triviality for you, and not for me 2011-11-27T15:09:00 and you think i know that but i dont 2011-11-27T15:09:58 ill search for this localization problem 2011-11-27T15:10:22 Well the robot knows the map and it's dropped at a random poisitoon 2011-11-27T15:10:30 It only has sonar range sensors 2011-11-27T15:10:33 aham 2011-11-27T15:10:39 pairofdice: thats it so it knows the map 2011-11-27T15:10:43 Yes 2011-11-27T15:10:45 now makes sense 2011-11-27T15:10:46 *** NoxiaZ^ has quit IRC (Ping timeout: 260 seconds) 2011-11-27T15:11:59 *** SlashV has quit IRC (Quit: Leaving.) 2011-11-27T15:12:23 *** besh has quit IRC (Quit: Page closed) 2011-11-27T15:13:58 *** Taelia has joined #aichallenge 2011-11-27T15:15:46 *** Taelia has quit IRC (Client Quit) 2011-11-27T15:26:15 it seems im looking forward to a willy fun class 2011-11-27T15:26:19 *** Antimony has quit IRC (Ping timeout: 245 seconds) 2011-11-27T15:30:33 *** mj41 has quit IRC (Ping timeout: 258 seconds) 2011-11-27T15:32:27 *** epicmonkey has quit IRC (Ping timeout: 252 seconds) 2011-11-27T15:44:42 *** Anilm3 has joined #aichallenge 2011-11-27T15:45:31 Hey 2011-11-27T15:46:58 *** kire has quit IRC (Quit: Leaving) 2011-11-27T15:48:24 *** Antimony has joined #aichallenge 2011-11-27T15:50:24 *** thestinger has quit IRC (Quit: WeeChat 0.3.6) 2011-11-27T15:53:28 *** ltriant has joined #aichallenge 2011-11-27T15:53:35 *** delt0r_ has quit IRC (Ping timeout: 248 seconds) 2011-11-27T15:54:45 *** Accoun has joined #aichallenge 2011-11-27T15:56:16 *** thestinger has joined #aichallenge 2011-11-27T15:59:06 *** nathanro11 has quit IRC (Quit: Page closed) 2011-11-27T16:00:46 Wow! On multi_hill_maze/maze_04p_01.map, my bot was able to find all the hills by turn 3 through symmetry, and yet it still lost! 2011-11-27T16:02:03 I would avoid the hills of others 2011-11-27T16:02:05 *** kire has joined #aichallenge 2011-11-27T16:02:48 Because that's where most ants are going to be and the least amount of food 2011-11-27T16:03:01 multihill maps are so crazy 2011-11-27T16:03:04 Antinomy, multi_hill_mazes are no no more 2011-11-27T16:03:13 you start out right next to each other and there are hills everywhere 2011-11-27T16:03:17 oh 2011-11-27T16:03:20 Yeah 2011-11-27T16:03:21 too bad, I was good at those 2011-11-27T16:03:31 BenJackson: yeah :) 2011-11-27T16:03:37 woo hoo, top of the queue in 6 minutes 2011-11-27T16:03:41 but they're still on tcp for some reason 2011-11-27T16:03:48 will my rank improve from 163rd? only time will tell 2011-11-27T16:03:50 but my bot sucks at the random walks with hills right near each other 2011-11-27T16:04:09 i just noticed on ants servers maze maps are actually multihill ones 2011-11-27T16:04:18 oh damn 2011-11-27T16:05:53 *** besh has joined #aichallenge 2011-11-27T16:07:09 fixed 2011-11-27T16:07:18 *** delt0r_ has joined #aichallenge 2011-11-27T16:15:47 343 minutes until next? ow 2011-11-27T16:16:41 hmmm 2011-11-27T16:16:55 knowing enemy hill locatiosn doesn't appear to help my bot 2011-11-27T16:17:29 correcting bugs doesn't appear to help mine 2011-11-27T16:17:45 I corrected a "major" bug and now it loses against the buggy one... 2011-11-27T16:17:57 been there 2011-11-27T16:18:01 in both cases! 2011-11-27T16:18:07 haha 2011-11-27T16:19:31 *** TheLinker has quit IRC (Remote host closed the connection) 2011-11-27T16:20:02 I think I'm going to have to redesign my goal selection 2011-11-27T16:21:08 *** Garf has quit IRC (Quit: Make a new plan, Stan!) 2011-11-27T16:22:14 *** mj41 has joined #aichallenge 2011-11-27T16:27:41 *** BenJackson has quit IRC (Ping timeout: 265 seconds) 2011-11-27T16:27:45 Ok, I've put the symmetry bot up on tcp, so you can see it for yourself 2011-11-27T16:28:25 ok, link? 2011-11-27T16:32:36 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-11-27T16:33:45 *** pvpv4001 has quit IRC () 2011-11-27T16:34:47 *** kilae has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-11-27T16:35:36 *** Kaji has joined #aichallenge 2011-11-27T16:37:06 *** antimatroid has quit IRC (Read error: Connection reset by peer) 2011-11-27T16:37:19 *** antimatroid has joined #aichallenge 2011-11-27T16:37:49 wtf! fluxid timeouts :( 2011-11-27T16:38:11 haha a1k0n 2011-11-27T16:42:28 *** jab_bott has joined #aichallenge 2011-11-27T16:43:50 http://ants.fluxid.pl/player/Parasprites 2011-11-27T16:44:00 wow, two consecutive games on same map, and completely opposite outcomes 2011-11-27T16:44:17 *** u_ has joined #aichallenge 2011-11-27T16:50:00 McLeopold: Can you help me to get the stand-alone visualizer running? http://forums.aichallenge.org/viewtopic.php?f=25&t=1633&p=12134#p12134 2011-11-27T16:52:13 McLeopold: I'm probably doing something very stupid - I don't use it myself but I want to provide working instructions for anyone who wants to use it with the AI-state visualization commands I've added on my fork 2011-11-27T16:53:16 *** ltriant has quit IRC (Read error: No route to host) 2011-11-27T16:54:06 *** Conorach has quit IRC (Ping timeout: 276 seconds) 2011-11-27T16:56:56 *** ltriant has joined #aichallenge 2011-11-27T16:58:12 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-27T17:00:23 Anyone wants to see my latest AI-state visualization: http://dl.dropbox.com/u/9957688/imgbin/ai-state-vis007.png 2011-11-27T17:01:58 It looks like an.. elephant 2011-11-27T17:03:35 i see an angry bear 2011-11-27T17:03:56 I see a fat dancing ghost 2011-11-27T17:03:59 hit by few plasma gun beams 2011-11-27T17:04:33 or perhaps the ghostbusters ghost doing a boogie 2011-11-27T17:04:47 yeah - what Minthos said 2011-11-27T17:05:23 is that a rorschach test? :D 2011-11-27T17:05:33 Overlayerer brings my computer to it's knees 2011-11-27T17:05:49 Possible user error 2011-11-27T17:06:02 Or probable rather 2011-11-27T17:06:16 it rather brings java to knees 2011-11-27T17:06:53 pairofdice: are you using the vis_overlay fork? 2011-11-27T17:06:56 java visualizer works few times slower than inside normal browser 2011-11-27T17:07:03 i still find it interesting that the thing can draw thousands of colored tiles, but as soon as you use the text strings, it becomes borderline unusable :( 2011-11-27T17:07:04 jab_bott: yeah 2011-11-27T17:07:37 pairofdice: yeah - it's very easy to get carried away and put too much stuff in there, especially since I added the 'i' commands for map-info 2011-11-27T17:07:44 I had it just filling tiles with color 2011-11-27T17:08:04 pairofdice: you just have to stick to visualizing a few things at a time and keeping the output small so you're not there all day 2011-11-27T17:08:14 Yeah 2011-11-27T17:08:51 HaraKiri: yeah - I don't fully understand that either 2011-11-27T17:10:09 HaraKiri: the data is massive though 2011-11-27T17:10:31 hmm but it doesn't get slow after thousands if i's, it gets slow after 2 or 3 2011-11-27T17:10:45 Fluxid: can you help me to get the stand-alone visualizer running? I posted about it on the ai state thread on the forums 2011-11-27T17:11:27 *** jstrong has quit IRC (Ping timeout: 248 seconds) 2011-11-27T17:12:08 HaraKiri: hrm, interesting - what have you noticed is slowest - while running your bot, or while writing out the data at the end, or while loading it into the browser? 2011-11-27T17:12:16 HaraKiri: I mean affected the most? 2011-11-27T17:12:51 *** lhb__ has joined #aichallenge 2011-11-27T17:13:47 uhm after the last round, before the browser pops up 2011-11-27T17:14:17 the running itself or looking at the resulsts are always fine 2011-11-27T17:14:28 jab_bott: not today, maybe tomorrow. i should be in bed hour ago... 2011-11-27T17:15:58 *** besh has quit IRC (Quit: Page closed) 2011-11-27T17:16:25 HaraKiri: it could be I'm doing something silly in python (I've never used python before, so I only know what I've learned in doing this vis mod) - have you looked at the code at all in ants.py? 2011-11-27T17:16:26 there's a weird bug I observe sometimes in rankings.php... 2011-11-27T17:16:35 *** raemde_ has quit IRC (Ping timeout: 252 seconds) 2011-11-27T17:16:46 HaraKiri: maybe there's a rate-limit (like there was in worker.py for reading bot output) on writing out the data too? 2011-11-27T17:16:53 sometimes, instead of the total rankings, it will only display the rankings for "Other" (organisation) 2011-11-27T17:17:14 oh no i haven't looked at the new code at all 2011-11-27T17:17:31 Fluxid: could you paste me a playgame.py command that works in the java visualizer on epsilon branch? 2011-11-27T17:18:00 where's the "new" stuff it in jab_bott? 2011-11-27T17:18:05 O_o 2011-11-27T17:18:44 Fluxid: I honestly think it's something really stupid that I'm just missing 2011-11-27T17:19:03 http://paste.pocoo.org/show/513756/ 2011-11-27T17:19:15 but it plays a game ant doesn't launch anything.. 2011-11-27T17:19:24 it works both in epsilon and yours 2011-11-27T17:19:41 HaraKiri: what new stuff? The last thing I added was the i command - nothing since then (although I made lines and arrows go the shortest wrapped distance - can't remember if that was before or after) 2011-11-27T17:20:17 yeah i was more thinking where in the file, but i found it :D 2011-11-27T17:20:45 *** antimatroid has quit IRC (Read error: No route to host) 2011-11-27T17:20:54 *** antimatroid has joined #aichallenge 2011-11-27T17:21:44 *** Fandekasp has quit IRC (Ping timeout: 252 seconds) 2011-11-27T17:22:00 *** iglo has joined #aichallenge 2011-11-27T17:22:57 Fluxid: how does that stream in real-time to the Java visualizer? 2011-11-27T17:23:22 i don't stream 2011-11-27T17:23:41 because java kneels when streaming 2011-11-27T17:23:50 HaraKiri: ahh right - if you go to github you can see all the changes nice and easily 2011-11-27T17:23:56 no idea, i just run visualizer from console each time after game 2011-11-27T17:24:19 *** lhb__ is now known as lhb 2011-11-27T17:24:22 Fluxid: oh right - is there any benefit to that? I was thinking the only real benefit to the java version was that you could stream it 2011-11-27T17:24:24 *** lhb is now known as raemde 2011-11-27T17:25:11 Fluxid: are you a brit? 2011-11-27T17:25:26 pole 2011-11-27T17:26:26 jab_bott: it's just easier for me. it's not as slow as when streaming, and it starts in small pretty window, i don't have to switch desktops to browser 2011-11-27T17:26:35 ahh cool - yeah get some sleep mate :) 2011-11-27T17:26:49 after i debug one segfault ;_; 2011-11-27T17:27:16 *** jstrong has joined #aichallenge 2011-11-27T17:27:33 *lol* just one more feature.. I promise dear... then I'll come to bed... (oh crap, it's 06:00 - I'd best try not to wake her or she'll be mad) 2011-11-27T17:27:44 this is horrible, one time it segfaults, the other time python crashes with so obscure error i couldn't ever imagine 2011-11-27T17:28:00 ;) 2011-11-27T17:28:24 fortunately i don't have a girlfriend/wife, but work early in the morning 2011-11-27T17:29:00 actually - she's waiting for me know as we're supposed to watch a film tonight... I'd best be off. 2011-11-27T17:29:16 cu! 2011-11-27T17:29:22 later all - HaraKiri: enjoy the code - let me know if you find the problem 2011-11-27T17:29:28 i will! bye 2011-11-27T17:29:32 *** jab_bott has left #aichallenge 2011-11-27T17:29:53 *** HaraKiri has quit IRC () 2011-11-27T17:37:04 *** kire has quit IRC (Remote host closed the connection) 2011-11-27T17:41:09 *** Dlayne has quit IRC (*.net *.split) 2011-11-27T17:41:09 *** welterde has quit IRC (*.net *.split) 2011-11-27T17:41:09 *** Raimondi has quit IRC (*.net *.split) 2011-11-27T17:41:09 *** johnsonav has quit IRC (*.net *.split) 2011-11-27T17:41:09 *** krishna has quit IRC (*.net *.split) 2011-11-27T17:41:09 *** Minthos has quit IRC (*.net *.split) 2011-11-27T17:41:09 *** jmcarthur has quit IRC (*.net *.split) 2011-11-27T17:41:09 *** spacebat has quit IRC (*.net *.split) 2011-11-27T17:41:09 *** Fluxid has quit IRC (*.net *.split) 2011-11-27T17:41:16 *** mcstar has quit IRC (*.net *.split) 2011-11-27T17:41:16 *** Cyndre has quit IRC (*.net *.split) 2011-11-27T17:41:16 *** rexou has quit IRC (*.net *.split) 2011-11-27T17:41:16 *** jbroman has quit IRC (*.net *.split) 2011-11-27T17:41:16 *** Kingpin13 has quit IRC (*.net *.split) 2011-11-27T17:41:16 *** bugnuts2 has quit IRC (*.net *.split) 2011-11-27T17:41:16 *** j3camero has quit IRC (*.net *.split) 2011-11-27T17:41:17 *** Zepp has quit IRC (*.net *.split) 2011-11-27T17:41:17 *** dai-ra has quit IRC (*.net *.split) 2011-11-27T17:41:17 *** lnx has quit IRC (*.net *.split) 2011-11-27T17:41:17 *** juuso_ has quit IRC (*.net *.split) 2011-11-27T17:41:17 *** klutometis has quit IRC (*.net *.split) 2011-11-27T17:41:19 *** amstan has quit IRC (Ping timeout: 248 seconds) 2011-11-27T17:41:21 *** grwip has quit IRC (*.net *.split) 2011-11-27T17:41:21 *** Chris_0076 has quit IRC (*.net *.split) 2011-11-27T17:41:21 *** a1k0n has quit IRC (*.net *.split) 2011-11-27T17:41:21 *** hkraal has quit IRC (*.net *.split) 2011-11-27T17:41:21 *** sorki has quit IRC (*.net *.split) 2011-11-27T17:41:21 *** MikeyG has quit IRC (*.net *.split) 2011-11-27T17:41:22 *** Sir_Ragnarok has quit IRC (*.net *.split) 2011-11-27T17:41:22 *** johnsonav has joined #aichallenge 2011-11-27T17:41:22 *** a1k0n_ has joined #aichallenge 2011-11-27T17:41:25 *** Zepp has joined #aichallenge 2011-11-27T17:41:26 *** klutometis has joined #aichallenge 2011-11-27T17:41:31 *** bugnuts2 has joined #aichallenge 2011-11-27T17:41:34 *** jbroman has joined #aichallenge 2011-11-27T17:41:39 *** Minthos has joined #aichallenge 2011-11-27T17:41:39 *** MikeyG has joined #aichallenge 2011-11-27T17:41:39 *** sorki has joined #aichallenge 2011-11-27T17:41:39 *** krishna has joined #aichallenge 2011-11-27T17:41:40 *** Fluxid has joined #aichallenge 2011-11-27T17:41:40 *** sorki has quit IRC (Changing host) 2011-11-27T17:41:40 *** sorki has joined #aichallenge 2011-11-27T17:41:40 *** hkraal has joined #aichallenge 2011-11-27T17:41:40 *** dai-ra_ has joined #aichallenge 2011-11-27T17:41:41 *** klutometis is now known as Guest39521 2011-11-27T17:41:43 *** jmcarthur has joined #aichallenge 2011-11-27T17:41:45 *** lnx has joined #aichallenge 2011-11-27T17:41:46 *** mcstar has joined #aichallenge 2011-11-27T17:41:52 *** j3camero has joined #aichallenge 2011-11-27T17:41:53 *** Chris_0076 has joined #aichallenge 2011-11-27T17:42:00 *** Cyndre has joined #aichallenge 2011-11-27T17:42:05 *** grwip has joined #aichallenge 2011-11-27T17:42:10 silly freenode 2011-11-27T17:42:12 *** lnx is now known as Guest92330 2011-11-27T17:42:12 *** Kingpin13 has joined #aichallenge 2011-11-27T17:42:23 duh 2011-11-27T17:42:24 *** Sir_Ragnarok has joined #aichallenge 2011-11-27T17:42:45 *** iglo has quit IRC (Read error: Operation timed out) 2011-11-27T17:43:34 *** Raimondi has joined #aichallenge 2011-11-27T17:44:15 I get seriously creepy spam 2011-11-27T17:44:34 they address me with the correct first name, and a last name starting with the same letter as mine (but never actually my last name) 2011-11-27T17:45:10 if they know my email, they know my name - so I don't understand where the spambot is failing 2011-11-27T17:45:43 *** welterde has joined #aichallenge 2011-11-27T17:45:53 it knew too much 2011-11-27T17:46:09 i dont read email 2011-11-27T17:46:17 *** rexou has joined #aichallenge 2011-11-27T17:46:18 *** spacebat has joined #aichallenge 2011-11-27T17:46:36 i don't have email 2011-11-27T17:46:40 *** juuso_ has joined #aichallenge 2011-11-27T17:47:25 thestinger: your first name is easily guessed 2011-11-27T17:47:28 The 2011-11-27T17:47:31 soo common 2011-11-27T17:47:44 only a programmer could be worried about a poor spam bot not correctly recognising names 2011-11-27T17:48:00 a programmer that actually writes spam bots!!!! 2011-11-27T17:48:02 well, they're sending me fake job offers 2011-11-27T17:48:09 youre busted sir! 2011-11-27T17:48:11 they should just hire me and have me fix their crappy name guessing 2011-11-27T17:48:48 arrrgh the timeouts on fluxid are getting bad 2011-11-27T17:48:59 I get ~3 fake job offers a day, and occasionally a russian rolex/viagra one 2011-11-27T17:49:15 0 spam on gmail 2011-11-27T17:49:20 i like the site with the chinese ladies 2011-11-27T17:49:23 well, it goes to my spam folder in gmail :) 2011-11-27T17:49:29 but I'm curious so I always read them 2011-11-27T17:49:33 Oh 2011-11-27T17:50:00 sometimes I respond to the scams because pretending to fall for them is amusing 2011-11-27T17:50:44 antimatroid: are you ready for tcp play? 2011-11-27T17:50:58 not even programming atm 2011-11-27T17:51:06 but no not really :P 2011-11-27T17:51:49 i was perfectly inefficient today 2011-11-27T17:52:15 are any of you doing combat prisoner's dilema style? like generate 2-4 moves for yourself and your opponents and just picking the least bad one 2011-11-27T17:52:20 *** materi has joined #aichallenge 2011-11-27T17:52:30 that's what I'm trying atm 2011-11-27T17:52:40 2-4 moves into the future? 2011-11-27T17:52:43 no 2011-11-27T17:52:52 just possible moves for the current turn 2011-11-27T17:53:02 an ant has 5 2011-11-27T17:53:11 when not blocked by water 2011-11-27T17:53:21 I mean 2-4 sets of moves 2011-11-27T17:53:30 hm 2011-11-27T17:53:36 I'm not worried about 2v1 or anything atm, just big battles 2011-11-27T17:53:44 *** Redgis has quit IRC (Ping timeout: 244 seconds) 2011-11-27T17:53:49 how do you chose the set? 2011-11-27T17:53:55 choose* 2011-11-27T17:54:24 well right now I'm just doing 3 possible sets of moves - running away, staying still or moving towards the enemy 2011-11-27T17:54:51 so basically you enumerate, but prune a little in front 2011-11-27T17:55:56 thestinger: i am 2011-11-27T17:56:00 my old "combat" (current bots on tcp) just cancels moves for ants that are going to die if the enemy doesn't move 2011-11-27T17:56:15 i partition ants up into those that can battle an enemy in the same partition 2011-11-27T17:56:20 yeah 2011-11-27T17:56:22 then do defensive and offensive moves, and pick them off that 2011-11-27T17:56:30 I divide up everything into battles 2011-11-27T17:56:45 a battle is just a set of enemies, and a set of allies 2011-11-27T17:56:53 based on +2 to enemy attackradius 2011-11-27T17:56:59 thestinger: but ive been telling this for quite some time 2011-11-27T17:57:17 with the exception that i explore all possibilities 2011-11-27T17:57:49 well, I plan on adjusting the number of possibilities I' 2011-11-27T17:57:57 I'll try based on the # of ants in the battle* 2011-11-27T17:59:27 *** yoden has joined #aichallenge 2011-11-27T18:01:26 mcstar: are you actually doing a game tree though? 2011-11-27T18:01:40 *** antimatroid1 has joined #aichallenge 2011-11-27T18:01:42 *** antimatroid has quit IRC (Read error: No route to host) 2011-11-27T18:01:51 i really dont want to call that a game-tree 2011-11-27T18:01:56 but it would be it 2011-11-27T18:02:04 but with generating only 1 turn.... 2011-11-27T18:02:17 is anyone doing attacking an enemy ant from just one direction or in all directions? 2011-11-27T18:02:34 i try from any direction 2011-11-27T18:02:47 i.e. my ants could be anywhere in relation the the enemy 2011-11-27T18:02:56 i have a new grouping method in mind 2011-11-27T18:03:06 but im nowhere near finished 2011-11-27T18:03:30 is it better that just attacking in one direction? top bots eems to attack in just one direction (pguillory, xathis) 2011-11-27T18:03:37 seems* 2011-11-27T18:05:39 the attacks coming from all directions i think will only work well with static enemy ants but not when they are moving 2011-11-27T18:05:56 i dont agree 2011-11-27T18:06:31 I just do a BFS from the current enemy ant locations and mark off distances - then use that to move towards/away from or strafe/encircle 2011-11-27T18:06:42 it doesn't really matter if they move 2011-11-27T18:06:49 they are still in the same direction for the most part 2011-11-27T18:07:12 I really need to finish off this code and start testing it :P 2011-11-27T18:08:13 the problem i could see is that when you encircle an enemy ant and it decides to move in any direction, it will get into attackRadius of one of your ants, so instead of killing it, it and your ant is killed 2011-11-27T18:08:58 not if you outnumber them everywhere 2011-11-27T18:09:19 while attack from a single direction will make sure that enemy ant will be crushed in your wall of ant when they couldn't get away 2011-11-27T18:09:37 I plan on dealing with stuff like that with tiebreakers 2011-11-27T18:09:55 so if an ant could move in 2+ directions to get closer to a target, I'll figure out which will be better 2011-11-27T18:09:57 a1kon: you have a point 2011-11-27T18:10:28 i path find to enemies while blocking path finding with my ants 2011-11-27T18:10:34 so it'll slowly surround the buggers 2011-11-27T18:10:44 that's what strcat_flock does on the tcp server - the tie breaker is based on allies in attack range 2011-11-27T18:10:59 it doesn't actually seem to work though :) 2011-11-27T18:11:46 it will also divert ants around to enemies from all paths if you block up one way 2011-11-27T18:11:56 instead of just piling up ants where they get blocked up 2011-11-27T18:12:04 antimatroid1: yeah, I do something similar (I increase the cost of a path going through a location one of my ants is at) 2011-11-27T18:12:24 but I'm pretty conservative about it atm because it makes my ants spread out too far 2011-11-27T18:13:42 thats a good idea. my current implementation just counts if my possible attacks to an enemy ants is greater the enemy ant can attack me. been stuck with it for 2 weeks now, i relly need to make my battle code better. :) 2011-11-27T18:14:07 really* 2011-11-27T18:14:27 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-27T18:16:25 *** cyphase has quit IRC (Ping timeout: 260 seconds) 2011-11-27T18:18:27 *** mcstar has quit IRC (Quit: WeeChat 0.3.6) 2011-11-27T18:20:49 *** Cowbandit has joined #aichallenge 2011-11-27T18:21:18 *** ikaros has joined #aichallenge 2011-11-27T18:24:54 *** Cowbandit has quit IRC (Client Quit) 2011-11-27T18:25:50 *** BenJackson has joined #aichallenge 2011-11-27T18:28:10 *** Jak_o_Shadows has joined #aichallenge 2011-11-27T18:30:23 *** a1k0n_ has quit IRC (Ping timeout: 248 seconds) 2011-11-27T18:31:25 *** a1k0n has joined #aichallenge 2011-11-27T18:34:13 thestinger: holy shit at the end that's quite an attack: http://ants.fluxid.pl/replay.1479 2011-11-27T18:34:54 I still don't have my real combat code done :) 2011-11-27T18:35:57 the reason they back away from you sometimes is basically this thought process: 2011-11-27T18:36:32 "oh crap, there are a bunch of enemies. I'm just going to go forage, the hive mind is an idiot for telling me to attack their hill." 2011-11-27T18:36:38 it's sort of a result of a "bug" 2011-11-27T18:37:46 will be fixed and done properly in my latest combat attempt 2011-11-27T18:38:35 I make my foragers avoid enemies if there is likely more than one enemy left 2011-11-27T18:38:38 I meant the vast pouring of ants toward my base 2011-11-27T18:38:43 oh :) 2011-11-27T18:38:50 well, if orange had died they would have killed you faster 2011-11-27T18:38:53 due to that foraging bug 2011-11-27T18:38:56 yeah 2011-11-27T18:39:00 I did sort of ninja orange 2011-11-27T18:39:10 stinger's da boss 2011-11-27T18:39:13 well, see the three orange ants at the end? 2011-11-27T18:39:16 *** jstrong is now known as roflmao 2011-11-27T18:39:19 my bot considers it a 3-player game still 2011-11-27T18:39:34 and the ants at the battle line are sometimes trying to forage, so they run away 2011-11-27T18:39:55 they are trying to search for food near your hill basically 2011-11-27T18:40:04 because they are closest to that area 2011-11-27T18:40:22 once I get within 10 range of your hill that bug stops 2011-11-27T18:40:36 *** materi has quit IRC (Ping timeout: 265 seconds) 2011-11-27T18:40:39 wait, 3 orange ants aren't you 2011-11-27T18:41:03 yeah, it's someone else 2011-11-27T18:41:10 so my bot has estimated_enemies set to 2 2011-11-27T18:41:16 so my foragers are cautious 2011-11-27T18:41:26 interesting 2011-11-27T18:41:32 and the ants closest to you (the battle line) are chosen for foraging until they get close to you 2011-11-27T18:41:39 I've considered doing something with enemy count but never did 2011-11-27T18:41:41 so there's a bug where they will retreat 2011-11-27T18:42:16 with my new combat, I won't just naively avoid enemies with the foraging ants, so that shouldn't happen 2011-11-27T18:42:36 I lost that one due to too many of my ants getting stuck in that cul-de-sac 2011-11-27T18:42:57 why do they keep going in there? 2011-11-27T18:43:01 to search for food? 2011-11-27T18:43:19 *** jnwhiteh has quit IRC (*.net *.split) 2011-11-27T18:43:28 I revised one of my exploration metrics to favor going away from hills 2011-11-27T18:43:34 ah 2011-11-27T18:43:44 so the desire to explore is not making it past my hill 2011-11-27T18:44:00 that's my guess 2011-11-27T18:44:04 emergent behavior is hard to be sure of 2011-11-27T18:44:22 yep 2011-11-27T18:44:23 even when that doesn't happen I still overpopulate those areas 2011-11-27T18:44:30 which didn't exist much in the "classic" maps 2011-11-27T18:44:43 *** jstrong has joined #aichallenge 2011-11-27T18:44:45 my food searching is essentially emergent behavior that I had to tune for ages 2011-11-27T18:44:47 *** jnwhiteh has joined #aichallenge 2011-11-27T18:45:28 it would have been easier just to code it directly after all the work I had to do 2011-11-27T18:45:28 my combat is emergent, in the rare cases where it happens 2011-11-27T18:46:19 *** roflmao has quit IRC (Ping timeout: 245 seconds) 2011-11-27T18:46:23 strcat_dense is doing poorly for some reason 2011-11-27T18:46:42 it's just a simple bug fix for my movement, and I can't figure out what good behavior the bug is causing 2011-11-27T18:47:10 what was the bug? 2011-11-27T18:47:42 they would basically never swap places because I had an off-by-one error 2011-11-27T18:47:55 *** mj41 has quit IRC (Ping timeout: 260 seconds) 2011-11-27T18:48:19 they still rarely swap places, but that's intended 2011-11-27T18:49:07 I think the bug causes some completely unrelated behavior I've been relying on, no idea what the hell that would be though :\ 2011-11-27T18:53:23 after rewriting pathfinding in C my bot takes half the cpu time it needed before... 2011-11-27T18:53:37 feels like useless effort 2011-11-27T18:53:47 good news is I played 2 quick games on aichallenge. bad news is in the second I got double teamed 2011-11-27T18:54:01 Fluxid: wow, only half? 2011-11-27T18:54:08 I bet having to use the python types slows it down 2011-11-27T18:54:27 i need to profile it 2011-11-27T18:54:32 yeah, if you're refcounting and etc you should translate entirely into C, do the calc, then entirely back out 2011-11-27T18:54:56 thestinger: internally i use native types 2011-11-27T18:54:58 this is why I used C++: I can write gnarly code for the tight inner loop stuff 2011-11-27T18:55:08 i mean, whole pathfinding is pure C 2011-11-27T18:55:11 but the high level code can still deal with lists/sets/etc of ants 2011-11-27T18:55:16 Fluxid: are you using malloc()? 2011-11-27T18:55:37 yeah, but not often 2011-11-27T18:55:44 that's probably more expensive than how I imagine python allocates memory 2011-11-27T18:55:57 I know that java just mallocs a massive pool and then uses that 2011-11-27T18:56:00 I don't optimize away mallocs at all and it's not hurting me 2011-11-27T18:56:07 I use std::vector with abandon 2011-11-27T18:56:22 same with me :) 2011-11-27T18:56:23 ok, I do optimize them away a bit by re-using some things turn to turn 2011-11-27T18:56:31 but I do create a shitton of vectors 2011-11-27T18:56:53 my pathfinding code only allocates when it pushes stuff onto a std::queue though 2011-11-27T18:57:00 I really did end up using that vector,vector>> 2011-11-27T18:57:06 most of my vectors are allocated once at the start and reused 2011-11-27T18:57:13 well 2011-11-27T18:57:38 *** Antimony has quit IRC (Ping timeout: 255 seconds) 2011-11-27T18:57:47 the funny thing is that I balked at using that vector 2011-11-27T18:57:59 until I realized that I'd have no qualms in python about doing [([...], [...]), ....] 2011-11-27T18:58:05 which is exactly what it is 2011-11-27T18:58:41 thestinger: you need to get on tcpants 2011-11-27T18:58:44 *** cyphase has joined #aichallenge 2011-11-27T18:58:44 it's lonely 2011-11-27T18:58:50 I will 2011-11-27T18:59:06 I want to finish this combat code :) 2011-11-27T18:59:39 I don't know how std::queue is implemented, but the priority queue I wrote is backed by a dynamic array and I got a huge performance boost by preallocating the array to hold a few thousand nodes 2011-11-27T19:00:02 std::queue is a FIFO 2011-11-27T19:00:10 I bet it's just a linked list 2011-11-27T19:00:24 well profile it anyway 2011-11-27T19:00:32 it's the best way to find out what's going on 2011-11-27T19:00:33 it's fast enough :) 2011-11-27T19:00:55 I can stick 7000 targets into a distance map in way under 1ms 2011-11-27T19:01:17 ah I mean fluxid should profile his 2011-11-27T19:01:32 my suicidal bot with crappy battle code gets lucky here: http://tcpants.com/replay.8545 2011-11-27T19:01:53 lol, I actually hate that map 2011-11-27T19:02:08 I hate random walks :( 2011-11-27T19:02:48 I guess bots with symmetry detection could do really well on random walks, because you can rush a hill 2011-11-27T19:02:57 but for me, it's just luck 2011-11-27T19:02:58 i hate random walk too especially my hill defense is not that great. 2011-11-27T19:03:06 I hate that my bot isn't super awesome 2011-11-27T19:03:16 hill defense won't help on a random walk imo, at least until turn 100+ 2011-11-27T19:03:27 if someone actually finds your hill really fast, you're probably screwed 2011-11-27T19:03:57 after the initial rushes, it definitely helps 2011-11-27T19:04:28 if your bot is suicidal about attacking yes 2011-11-27T19:04:52 if it just chills out keeping ants away from its territory it doesn't matter much 2011-11-27T19:05:09 then you're screwed on a random walk anyway :P 2011-11-27T19:05:21 *** xathis_ has joined #aichallenge 2011-11-27T19:05:27 some bot will rush hills and get 9 points before your bot starts attacking 2011-11-27T19:05:44 or die trying 2011-11-27T19:06:04 yeah 2011-11-27T19:06:30 random walks are pretty weird though 2011-11-27T19:06:40 you can win by just rushing and getting points, because it doesn't matter if you die 2011-11-27T19:07:11 *** xathis has quit IRC (Ping timeout: 248 seconds) 2011-11-27T19:07:23 that's how my old extremely crappy version is doing well on the official site 2011-11-27T19:07:31 http://aichallenge.org/profile.php?user=2255 2011-11-27T19:08:34 it has horrible movement and no combat 2011-11-27T19:08:41 it just rushes 2011-11-27T19:09:00 and it has half-working foraging so it can sort of sustain itself sometimes 2011-11-27T19:09:25 I haven't reuploaded because my new version will probably be worse, because it's better :) 2011-11-27T19:10:45 tcp is different because all the bots are really good 2011-11-27T19:10:59 rate of play seems faster in tcpants compared to fluxid? 2011-11-27T19:13:08 *** grwip has quit IRC (Remote host closed the connection) 2011-11-27T19:13:59 http://paste.aichallenge.org/sq8OH/ a bit better... now to make it more aggresive 2011-11-27T19:14:47 @thestinger: i'm sure if you upload the same bot again it would have a hard time climbing to the top20 again. bots in the top200 right now is much stronger than the time you've uploaded it. 2011-11-27T19:14:48 AntDroid: No! 2011-11-27T19:15:16 AntDroid: it recently dropped down to ~50 and climbed back to 20 2011-11-27T19:15:21 I'm in the top 200 so they can't be that good 2011-11-27T19:16:09 AntDroid: mu was 85 a few days ago, and now 86 2011-11-27T19:16:48 so if I reuploaded it would climb up to ~30 after ~3 days I think 2011-11-27T19:16:48 Minthos: std::priority_queue is a container adapter. you can use any random access container as the storage 2011-11-27T19:16:51 *** Surya has quit IRC (Ping timeout: 265 seconds) 2011-11-27T19:16:52 I mean that same version 2011-11-27T19:16:59 by default I think it's a std::deque 2011-11-27T19:17:28 *** foRei has quit IRC (Read error: Connection reset by peer) 2011-11-27T19:19:19 *** gnutella has joined #aichallenge 2011-11-27T19:26:28 *** Antimony has joined #aichallenge 2011-11-27T19:28:35 *** replore has joined #aichallenge 2011-11-27T19:31:11 *** Antimony has quit IRC (Ping timeout: 258 seconds) 2011-11-27T19:33:13 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-27T19:38:50 http://ants.fluxid.pl/replay.1577 my pseudo-combat does some silly stuff :) 2011-11-27T19:39:23 My IRC client thinks STL templates contain smileys 2011-11-27T19:39:49 :d ? 2011-11-27T19:40:37 bot :p and :d 2011-11-27T19:40:41 pidgin? 2011-11-27T19:40:42 *both 2011-11-27T19:40:44 yes 2011-11-27T19:40:51 use bitlbee + a good client :) 2011-11-27T19:41:21 no thx, I have IRC, ICQ and MSN in one client + all the chat logs 2011-11-27T19:41:35 that's what bitlbee does 2011-11-27T19:41:40 for any irc client 2011-11-27T19:41:48 https://en.wikipedia.org/wiki/BitlBee 2011-11-27T19:41:49 *** NightExcessive has joined #aichallenge 2011-11-27T19:42:01 it can be built with libpurple support 2011-11-27T19:42:09 which is pidgin's backend lib 2011-11-27T19:43:52 ok, I may take a look at it. Although I don't see the benefit over my current setup. 2011-11-27T19:44:02 Fluxid: the formation in the center looks like a hadouken if you go fast enough 2011-11-27T19:44:29 mleise: I just turned smilies off completely in pidgin, fairly useless anyway 2011-11-27T19:44:30 *** jstrong has quit IRC (Ping timeout: 260 seconds) 2011-11-27T19:44:44 well, I like pidgin because of libpurple, but I hate the UI 2011-11-27T19:44:48 weechat forever <3 2011-11-27T19:45:14 *** ltriant_ has joined #aichallenge 2011-11-27T19:47:58 http://ants.fluxid.pl/replay.1609 daaaaamn 2011-11-27T19:48:16 what a big map :D 2011-11-27T19:50:11 http://ants.fluxid.pl/replay.1287 this makes me think that flocking works 2011-11-27T19:50:25 but then it completely fails to do anything in some games... 2011-11-27T19:52:01 it definitely seems to make my ants avoid spreading out too much 2011-11-27T19:52:04 sometimes. 2011-11-27T19:52:22 http://ants.fluxid.pl/replay.1612 daaaaaaaaaaaamn! 2011-11-27T19:53:29 are you increasing the density when you get more ants or is that just emergent? 2011-11-27T19:53:45 i try to 2011-11-27T19:57:01 *** rajanaresh has quit IRC (Ping timeout: 240 seconds) 2011-11-27T20:06:24 *** Antimony has joined #aichallenge 2011-11-27T20:11:13 where's my next game 2011-11-27T20:11:16 come on... 2011-11-27T20:13:17 looking for janzert... 2011-11-27T20:14:21 does tcpants limit to one connection per ip or something similar? 2011-11-27T20:15:25 I'm testing two versions in parallel, one is currently playing, and another one is constantly waiting at "connected to tcpants.com"... 2011-11-27T20:15:36 ew, that sucks 2011-11-27T20:16:06 on fluxis everything works as it should 2011-11-27T20:16:10 *fluxid 2011-11-27T20:20:12 ok, hopefully my bot will be halfway decent now 2011-11-27T20:38:10 *** jstrong has joined #aichallenge 2011-11-27T20:38:19 *** jstrong is now known as roflmao 2011-11-27T20:40:31 *** AntDroid_ has joined #aichallenge 2011-11-27T20:48:57 *** treeform_ has joined #aichallenge 2011-11-27T20:52:37 *** treeform has quit IRC (Ping timeout: 240 seconds) 2011-11-27T20:54:17 Here's a cool game. http://ants.fluxid.pl/replay.1655 2011-11-27T20:54:23 My bot sucks at combat, except when it comes to defence 2011-11-27T20:54:31 It always manages to barely regroup and fight them off 2011-11-27T21:02:54 is anyone here? 2011-11-27T21:03:57 Maybe 2011-11-27T21:04:28 Antimony: what's the name of your bot? 2011-11-27T21:04:29 *** fpcfan has quit IRC (Ping timeout: 260 seconds) 2011-11-27T21:10:50 *** NightExcessive has quit IRC (Ping timeout: 255 seconds) 2011-11-27T21:11:17 *** fpcfan has joined #aichallenge 2011-11-27T21:12:57 *** TheLinker has joined #aichallenge 2011-11-27T21:15:13 My bot's name is Parasprites 2011-11-27T21:15:32 @seen BenJackson 2011-11-27T21:15:32 Antimony: BenJackson was last seen in #aichallenge 1 hour, 31 minutes, and 29 seconds ago: Fluxid: the formation in the center looks like a hadouken if you go fast enough 2011-11-27T21:16:47 *** fpcfan has quit IRC (Ping timeout: 248 seconds) 2011-11-27T21:17:03 *** smiley1983 has quit IRC (Quit: leaving) 2011-11-27T21:17:16 *** xathis_ has quit IRC () 2011-11-27T21:20:24 *** olexs1 has quit IRC (Quit: Leaving.) 2011-11-27T21:21:24 *** fpcfan has joined #aichallenge 2011-11-27T21:24:01 does anyone know how to let a bot output a file when it is run by the tcp client? 2011-11-27T21:25:41 just open the file and write? 2011-11-27T21:32:55 *** gcflymoto has joined #aichallenge 2011-11-27T21:36:35 ahoy 2011-11-27T21:36:37 *** NightExcessive has joined #aichallenge 2011-11-27T21:36:43 yeah, normal debug works for me 2011-11-27T21:37:33 ahhh shit 2011-11-27T21:37:43 my next aichallenge game was a stupid setup 2011-11-27T21:37:58 7 player multi-hill with almost no food 2011-11-27T21:38:53 Yeah, you always loe hills in those and it's pretty random 2011-11-27T21:38:55 lose 2011-11-27T21:41:28 150 turns in everyone had like 10 ants 2011-11-27T21:42:24 *** amstan has joined #aichallenge 2011-11-27T21:42:24 *** ChanServ sets mode: +o amstan 2011-11-27T21:44:05 *** Guest39521 has quit IRC (Changing host) 2011-11-27T21:44:05 *** Guest39521 has joined #aichallenge 2011-11-27T21:44:09 *** Guest39521 is now known as klutometis 2011-11-27T21:47:31 http://ants.fluxid.pl/replay.1689 2011-11-27T21:47:47 quite a bottleneck at the end 2011-11-27T21:48:40 *** jj_ has joined #aichallenge 2011-11-27T21:49:08 oh hi ben 2011-11-27T21:49:23 I implemented symmetry but it doesn't appear to have helped me yet 2011-11-27T21:49:32 *** dom7b5 has joined #aichallenge 2011-11-27T21:49:57 Anyway, my bot can normally write to files just fine 2011-11-27T21:50:06 but the same executable run in the tcp client doesn't output a file 2011-11-27T21:53:41 playgame runs your bot in its directory 2011-11-27T21:53:42 it might be in the directory of the script 2011-11-27T21:53:44 tcp just runs it wherever 2011-11-27T21:54:05 *** Extrarius has quit IRC () 2011-11-27T21:54:13 oh I didn't think to look in the tcp directory 2011-11-27T21:54:17 I think you're right 2011-11-27T21:54:52 have you seen any improvement from symmetry? 2011-11-27T21:56:00 *** Jak_o_Shadows1 has joined #aichallenge 2011-11-27T21:56:01 *** treeform_ has quit IRC (Remote host closed the connection) 2011-11-27T21:56:26 *** treeform has joined #aichallenge 2011-11-27T21:57:13 wow, I didn't realize C++11 had tuples 2011-11-27T21:57:48 now I feel stupid for making all these structs I use in 1 place 2011-11-27T21:58:23 *** Jak_o_Shadows has quit IRC (Ping timeout: 248 seconds) 2011-11-27T21:58:45 thestinger: I was always confused if I should pass locations as Location structs or row and col to functions 2011-11-27T21:59:19 std::pair 2011-11-27T21:59:30 std::tuple t; 2011-11-27T21:59:33 t proof (1, 5, 3); 2011-11-27T21:59:38 that's the tuple syntax I think 2011-11-27T21:59:44 pairs just use .first and .second 2011-11-27T21:59:44 std::pair is the poor man's tuple 2011-11-27T22:00:13 pair only existed for set/map 2011-11-27T22:00:37 *** ThatSnail has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-11-27T22:01:35 *** treeform_ has joined #aichallenge 2011-11-27T22:01:43 heh, so I have lambdas and tuples now 2011-11-27T22:02:30 Tuples are ok, but I usually defien structs anyway, since tuple acess syntax is so hideous 2011-11-27T22:02:39 pos.x is better than get<1>(pos) 2011-11-27T22:03:20 *** jj_ has quit IRC (Quit: Page closed) 2011-11-27T22:03:36 pos.get<1> 2011-11-27T22:03:41 pos.get<1>() 2011-11-27T22:03:43 * 2011-11-27T22:04:27 hmm 2011-11-27T22:04:50 *** treeform has quit IRC (Ping timeout: 255 seconds) 2011-11-27T22:04:56 I think get<1>(pos) is the real syntax now 2011-11-27T22:06:23 *** dom7b5 has quit IRC (Ping timeout: 248 seconds) 2011-11-27T22:09:43 Wtf? When I weight hill razing over food more strongly, my bot takes longer to raze the hill! 2011-11-27T22:10:45 "weight" 2011-11-27T22:11:31 ew, the tuple syntax is really bad :\ 2011-11-27T22:11:46 yep 2011-11-27T22:11:56 Python has spoiled me 2011-11-27T22:12:00 C++ looks ugly now 2011-11-27T22:12:20 python doesn't have proper anonymous functions though :) 2011-11-27T22:12:23 C++11 does 2011-11-27T22:12:45 * p_l used Common Lisp. Neither Python nor C++ make the bar anymore ^^; 2011-11-27T22:12:55 p_l: haskell! 2011-11-27T22:13:19 lisp with syntax :) 2011-11-27T22:13:27 thestinger: don't know it well enough at the moment, though I plan on returning to Haskell at some point (probably next semester) 2011-11-27T22:13:58 *** sf17k has joined #aichallenge 2011-11-27T22:14:45 std::tuple t(1, 5, 2); 2011-11-27T22:14:47 std::cout << std::get<0>(t) << std::endl; 2011-11-27T22:14:49 yeah, that sucks 2011-11-27T22:15:02 better than a struct though imo 2011-11-27T22:16:31 *** Antimony has quit IRC (Ping timeout: 248 seconds) 2011-11-27T22:16:41 ... looks like a job for a macro to me. Or a 3-element simple-vector 2011-11-27T22:16:57 well, I was just figuring out the syntax 2011-11-27T22:17:13 that's not actually in my code :P 2011-11-27T22:17:27 std::array a = {1, 5, 2}; would be better than that 2011-11-27T22:18:02 (type-of #(1 2 3)) => (SIMPLE-VECTOR 3) 2011-11-27T22:18:05 :) 2011-11-27T22:18:16 though I'll admit accessing members can be annoing 2011-11-27T22:18:25 D has tuples for return types, too. You access them with t[0], t[1] and so on. 2011-11-27T22:19:54 ... I feel like an old lisper who has urge to say "it's all been done to hell and back, show me something new" 2011-11-27T22:20:31 >_> 2011-11-27T22:20:55 at least Haskell does 2011-11-27T22:21:22 (embedding a prolog interpreter is old lisp trick, though) 2011-11-27T22:21:34 more often than not it's a compiler 2011-11-27T22:21:44 *** JorgeB has joined #aichallenge 2011-11-27T22:23:35 If I was allowed to drink, I'd drink myself under the table for falling into that stereotype 2011-11-27T22:25:40 it would be nice if new languages actually learned something from the old ones 2011-11-27T22:25:41 *** redhosth has joined #aichallenge 2011-11-27T22:26:22 thestinger: actual comprehension necessary to learning would made C++ committee commit suicide from depression 2011-11-27T22:26:37 hi 2011-11-27T22:26:53 (and lead to a lot of alcoholism among other language designers) 2011-11-27T22:28:02 *** redhosth has left #aichallenge 2011-11-27T22:28:22 it's ridiculous that C++ lets you overload operators but not define your own 2011-11-27T22:28:25 I just don't get it 2011-11-27T22:28:49 that's very hard for postfix languages 2011-11-27T22:29:44 the only attempts I've ever seen at that had a set predefined overridable operators, with pre-decided precedence, commutativeness and associative rules 2011-11-27T22:30:06 oelewapperke: Haskell gets around that by removing operators :) 2011-11-27T22:30:16 yeah, operators are functions 2011-11-27T22:30:17 and once you go that route, you might as well make those operators do something ont he privitive types 2011-11-27T22:30:29 at which point you're back at square 1, in a language with slightly more operators 2011-11-27T22:30:36 no 2011-11-27T22:30:38 *** Antimony has joined #aichallenge 2011-11-27T22:30:53 Prelude> rem 5 2 2011-11-27T22:30:55 1 2011-11-27T22:30:57 Prelude> 5 `rem` 2 2011-11-27T22:30:59 1 2011-11-27T22:31:08 Prelude> 5 + 2 2011-11-27T22:31:10 7 2011-11-27T22:31:12 Prelude> (+) 5 2 2011-11-27T22:31:14 7 2011-11-27T22:31:35 yeah, iirc, there are no operators in Haskell 2011-11-27T22:32:04 cool. So does "+" have special syntax rules then ? 2011-11-27T22:32:04 gmp uses operator overloads in C++ 2011-11-27T22:32:15 oelewapperke: well functions can be infix by default 2011-11-27T22:32:20 oelewapperke: no 2011-11-27T22:32:25 there's nothing special about + 2011-11-27T22:32:35 Prelude> let (+++) x y = x + y 2011-11-27T22:32:37 Prelude> 5 +++ 2 2011-11-27T22:32:39 7 2011-11-27T22:33:06 they should have added that to C++11 imo 2011-11-27T22:33:12 backticks aren't used for anything afaik 2011-11-27T22:33:33 thestinger: ... don't encourage them 2011-11-27T22:33:36 lol 2011-11-27T22:34:10 that language already looks like a stew of "let's accept anything that comes this way as long as it doesn't break compatibility in spec, but let's accept screwed-up templates that break older code" 2011-11-27T22:34:54 they almost added something like type classes to C++11 2011-11-27T22:35:10 thestinger: so why doesn't "let (boem) x y = 2 * x + 3 * y", "2 boem 3" work ? 2011-11-27T22:35:18 If not for OpenCV, I wouldn't have touched it again since that time GCC managed to make if(pointer_to_class) into a template hell 2011-11-27T22:35:39 ah, it was if(!pointer_to_class_instance) 2011-11-27T22:36:02 apparently, someone got the bright idea to implement ! on pointers to classes with templates 2011-11-27T22:36:19 I don't know what all the fuzz about defining your own operators is about ^^. Just call a function. 2011-11-27T22:36:50 mleise: well, it's more about C++ missing the point on either side 2011-11-27T22:36:57 std::cout << 5 << std::endl; 2011-11-27T22:37:00 sensible languages made it into functions long time ago 2011-11-27T22:37:03 hideous abuse of operator overloading 2011-11-27T22:37:04 5 ** 2 2011-11-27T22:37:13 can't do that in C++ with gmp, they can only overload operators 2011-11-27T22:37:37 also, C++ somehow managed to become the only language I know where you can "overload" "operators" that ended with that being horribly abused 2011-11-27T22:37:54 oelewapperke: they need to be symbols like + * etc. 2011-11-27T22:37:59 <<- <<| 2011-11-27T22:38:15 C++ could be so much better if someone would just rewrite the standard library into a more modern syntax 2011-11-27T22:38:18 haskell doesn't waste << >> etc. on bit operators 2011-11-27T22:38:29 thestinger: so there is special syntax 2011-11-27T22:38:54 me hates: ofstream(osstream.str().c_str()) :) 2011-11-27T22:40:12 one thing i don't like is how python doesn't have function overload 2011-11-27T22:40:23 thestinger: when it's all "function", it's easier to figure out rather than the horrible operator<< 2011-11-27T22:40:59 amstan: that prevents you from wondering about the "now which function does this call" conundrum that you (sometimes) have in C 2011-11-27T22:40:59 amstan: overloading sucks, type classes is the way to go :P 2011-11-27T22:41:13 anyway python does pretty much have function overloading 2011-11-27T22:41:29 yes, but you want to do stuff like add(a,b) and add(list) 2011-11-27T22:41:42 you can do that 2011-11-27T22:41:43 and you need this ugly code to do it 2011-11-27T22:41:45 how do you do multiple constructors without overloading? 2011-11-27T22:42:14 In Delphi you have named constructors, but in other languages...? 2011-11-27T22:42:15 def add(*args), if len(args)==2 return arg[0]+arg[1] else return sum(args) 2011-11-27T22:42:32 it gets more fun when you want to name arguments properly 2011-11-27T22:42:40 p_l: blast you!, << is wonderful! 2011-11-27T22:42:55 wait.. no.. sum(args[0]) 2011-11-27T22:43:14 antimatroid1: chaining I/O is nice, but abusing operator overloading and not letting you handle errors properly isn't nice :P 2011-11-27T22:43:34 thestinger: you can check the error bit 2011-11-27T22:43:48 you say that like i ever handle errors :P 2011-11-27T22:44:11 no one handles errors in languages like python or C++ 2011-11-27T22:44:12 i wish there was a language which was as cool as all other languages combined 2011-11-27T22:44:34 *** antimatroid1 has quit IRC (Read error: Connection reset by peer) 2011-11-27T22:44:46 *** antimatroid has joined #aichallenge 2011-11-27T22:45:32 hmm, I wonder how to test with gdb if a bot needs to be executed by a python program 2011-11-27T22:45:53 GNU debugger? 2011-11-27T22:46:58 yeah 2011-11-27T22:47:16 then I don't understand your problem 2011-11-27T22:47:34 Go is really disappointing imo 2011-11-27T22:47:43 bots are executed by a python interpreter, right? 2011-11-27T22:47:49 amstan: Common Lisp probably does that to all imperative languages out there, with slight competition from Smalltalk, I guess :P 2011-11-27T22:47:54 avdg: gdb debugs a native executable. you cannot debug python with it 2011-11-27T22:48:02 sad 2011-11-27T22:48:11 unless you want to debug the python interpreter itself 2011-11-27T22:48:11 mleise: you can 2011-11-27T22:48:21 I like how they reinvented the Maybe monad using multiple return values and called it the "comma ok" idiom 2011-11-27T22:48:26 *** treeform_ has quit IRC (Remote host closed the connection) 2011-11-27T22:48:30 p_l: does it have an extension for that? 2011-11-27T22:48:40 p_l: yes, but lisp has syntax issues 2011-11-27T22:48:44 mleise: no. I call such stuff "low-level debugging" :) 2011-11-27T22:48:50 amstan: so, haskell 2011-11-27T22:48:58 ouch ;) 2011-11-27T22:49:02 amstan: only for the casual reader who never actually tries to use it 2011-11-27T22:49:10 *** bmh has joined #aichallenge 2011-11-27T22:49:22 avdg: look for something like http://docs.python.org/library/pdb.html 2011-11-27T22:49:37 thestinger: i would, but i guess i could say this about anything besides the major ones, they're not really used by people 2011-11-27T22:50:33 but i guess nothing's stopping me then, except sometimes support/libraries 2011-11-27T22:50:35 the widely used languages got there by being pragmatic and just re-using syntax/concepts from old languages to make them easy to learn 2011-11-27T22:50:48 thestinger: good point 2011-11-27T22:51:15 thestinger: then D must be super popular, but it isn't :( 2011-11-27T22:51:18 but what's the reason lisp is not as used? it's older than c/c++ 2011-11-27T22:51:41 mleise: if gcc and llvm had frontends for the latest version of the language, people might use it 2011-11-27T22:51:43 just the different way of thinking? 2011-11-27T22:51:45 amstan: everything is a list 2011-11-27T22:51:45 dmd is crap 2011-11-27T22:51:52 you need a good implementation too :) 2011-11-27T22:52:04 amstan: lisp was never used 2011-11-27T22:52:07 thestinger: gcc is only one revision behind 2011-11-27T22:52:21 but it's still around, and still has big advantages over other languages 2011-11-27T22:52:27 well _widely used_ is what I should have said 2011-11-27T22:52:39 amstan: the reason Lisp is not used more is mainly... bad publicity. And the fact that it doesn't look like or feels like algol 2011-11-27T22:52:51 lisp will be around in 50 years 2011-11-27T22:52:59 look at perl and awk 2011-11-27T22:53:04 python, ruby, etc. will be the same in 10 years imo 2011-11-27T22:53:16 python can't break backwards compatibility over and over to stay modern 2011-11-27T22:54:19 why learn perl 6 when you can learn python? 2011-11-27T22:54:39 why port code to python4 when a new, much shinier and nicer language exists? 2011-11-27T22:54:52 lisp has been around longer than C :P 2011-11-27T22:56:09 thestinger: as a side note, gcc and LLVM actually use the dmd front-end with modifications. So you get the same level of hints/warnings and compile-time features there. 2011-11-27T22:56:47 can I fetch a web page in D with the standard library? 2011-11-27T22:57:52 that's what people want in a new language (and a big reason why haskell isn't widely used) 2011-11-27T22:58:04 http://golang.org/pkg/ 2011-11-27T22:58:12 people don't use haskell because we can't understand its performance characteristics 2011-11-27T22:58:18 as I have so abruptly learned 2011-11-27T22:58:29 bmh: well, I don't mean instead of C or C++ 2011-11-27T22:58:44 *** dom7b5 has joined #aichallenge 2011-11-27T23:01:55 *** treeform has joined #aichallenge 2011-11-27T23:01:58 bmh: people don't use haskell because the mindfuck 2011-11-27T23:02:26 well, the reason I'll use python instead of haskell is because I don't want to search around for libraries 2011-11-27T23:02:43 I just want to get stuff done, and that's what python is good at 2011-11-27T23:02:50 downloading a web page in both haskell and Common Lisp is a case of short googling to find apropriate library, and one command to install it 2011-11-27T23:03:06 yeah, but then you have 5 implementations of something like strings 2011-11-27T23:03:14 (installing libs in both Haskell and Common Lisp is IMHO easier than Python) 2011-11-27T23:03:32 p_l: I use an OS with proper package management so that doesn't matter :) 2011-11-27T23:03:35 thestinger: sure you can download a web page (using curl bindings) 2011-11-27T23:03:38 thestinger: that's because the "standard" Strings turned out to be nice in theory, horrible disappointment in practice 2011-11-27T23:04:24 strcat@arch i ~/projects/ai-challenge-bot % cower -sq haskell- | wc -l 2011-11-27T23:04:26 1743 2011-11-27T23:04:28 strcat@arch i ~/projects/ai-challenge-bot % pacman -Ssq haskell | wc -l 2011-11-27T23:04:30 49 2011-11-27T23:04:51 and there's cabal2arch which just turns any cabal package into an Arch one :) 2011-11-27T23:04:59 I guess that's why so many libs are packaged... 2011-11-27T23:05:03 thestinger: http://freeze.steamwinter.com/D/web/phobos/etc_curl.html <- this is in review and will hit the std lib soonish 2011-11-27T23:05:33 that was just a silly example though 2011-11-27T23:05:43 haskell and D mostly just have language features in the standard lib 2011-11-27T23:05:45 not practical stuff 2011-11-27T23:06:09 http://docs.python.org/py3k/library/ I mean this is why I use python 2011-11-27T23:06:13 not because I like the language 2011-11-27T23:06:19 and CL's standard lib is frozen, with few semi-standard extensions (that are normalized with few popular libs) 2011-11-27T23:08:14 thestinger: you are right. it is understood, but the lack of consense and man power led to the std lib D has now. The same is true for a few other languages. 2011-11-27T23:08:31 that's why I linked the Go standard lib 2011-11-27T23:08:43 it's super new and basically has like 3 devs working part time on it 2011-11-27T23:09:39 but you can make an http server in one line without a library, etc. 2011-11-27T23:10:43 yes, it has all the stuff that is asked for in the news groups from time to time. most of that actually exist (for example a fast cgi impl.), but has to be searched for on the net. 2011-11-27T23:11:11 it's different when it's a third party lib 2011-11-27T23:11:21 standard libraries are usually very consistent, and the documentation is great 2011-11-27T23:12:02 and the docs are ofc all in one place 2011-11-27T23:12:14 "great" is a bit of a stretch. 2011-11-27T23:13:02 well, python's docs are great :) 2011-11-27T23:13:03 godoc is nice though 2011-11-27T23:13:04 A web server is not something you write every day. I think it doesn't need to be in the std lib just to have it all in one place 2011-11-27T23:13:16 go has a long way to go before the docs are nice 2011-11-27T23:13:54 mleise: xml, sql, csv, image encoding/decoding, etc. 2011-11-27T23:13:58 json 2011-11-27T23:15:22 json-c, json-glib, qjson, jansson, yajl are all libs in the Arch repos for json in C/C++ :P 2011-11-27T23:16:24 if I want to contribute to a python program using json, I won't have to learn the API for the json lib 2011-11-27T23:16:24 *** testingthis has joined #aichallenge 2011-11-27T23:17:15 sql is worked on, csv is in review, image decoding is missing. There are std.json and std.xml. See also: https://github.com/D-Programming-Language/deimos for C bindings. Arch repos or a language specific package manager are a matter of preferrence I think. 2011-11-27T23:17:22 *** testingthis has left #aichallenge ("Konversation terminated!") 2011-11-27T23:18:21 I'll be fair, most of what you want isn't there yet or in a bad condition. But it is not because people don't see that it is needed. 2011-11-27T23:19:35 that's why people use perl/ruby/python though imo 2011-11-27T23:19:56 including everything under the sun in the standard lib might not be a good idea, but it gets people to use the language 2011-11-27T23:20:38 yeah it makes powerfull scripting possible 2011-11-27T23:20:53 if you rely on a third party to a lib there are lots of consequences, like relying on them to update it 2011-11-27T23:20:58 for a lib* 2011-11-27T23:21:23 like all the third party python libs that aren't ported to py3 yet 2011-11-27T23:21:48 like PIL 2011-11-27T23:21:59 scipy and twisted 2011-11-27T23:22:01 Has PIL even been ported to 2.7 yet? 2011-11-27T23:22:03 on the other hand you are better of not maintaining 10000 features in the std lib with a hand full of volunteers 2011-11-27T23:22:29 mleise: yeah, but if people are maintaining a third party library they could be maintaining it in the standard library 2011-11-27T23:22:39 I think that's how python's standard lib has grown 2011-11-27T23:22:59 once it's standard, more people will use it and then more people will develop it 2011-11-27T23:23:17 *** rajanaresh has joined #aichallenge 2011-11-27T23:23:18 thestinger: I thought you referred to C libs used in Python where you need to maintain the lib and the bindings 2011-11-27T23:23:38 I meant standard library stuff 2011-11-27T23:23:45 a lot of that just uses C libs though 2011-11-27T23:24:01 like zlib, sqlite, etc. 2011-11-27T23:24:08 it is faster anyway ;) 2011-11-27T23:24:21 with cpython it is 2011-11-27T23:25:25 you forgot to mention container classes btw in you list 2011-11-27T23:26:01 that's an area where Go sucks, because they don't have nice polymorphism 2011-11-27T23:26:37 *** bhasker has quit IRC (Quit: bhasker) 2011-11-27T23:27:16 they decided to build dynamic arrays and hash tables into the language 2011-11-27T23:27:48 *** TheLinker has quit IRC (Quit: Bye) 2011-11-27T23:27:49 your language has issues if you can't implement containers in it :) 2011-11-27T23:27:50 thestinger: I think the situation is not different in D. 2011-11-27T23:28:01 doesn't D have templates? 2011-11-27T23:28:15 Go doesn't have static polymorphism 2011-11-27T23:28:30 it has something like duck typing, but only for types you define 2011-11-27T23:28:36 *** bhasker` has joined #aichallenge 2011-11-27T23:28:46 I mean: you can implement generic, template based container classes, but the lead library designer had a bias towards structs 2011-11-27T23:28:52 ah 2011-11-27T23:28:54 *** Jak_o_Shadows has joined #aichallenge 2011-11-27T23:29:04 you can't implement a generic container in Go 2011-11-27T23:29:14 crappy language imo, it just has nice concurrency :) 2011-11-27T23:29:33 crappy if it weren't for the cute mascot 2011-11-27T23:29:52 they just copypasted the plan9 mascot 2011-11-27T23:30:02 they made plan9, so I guess they can get away with that 2011-11-27T23:30:21 *** bmh has left #aichallenge 2011-11-27T23:30:50 http://glenda.cat-v.org/friends/ 2011-11-27T23:30:57 *** Jak_o_Shadows1 has quit IRC (Ping timeout: 252 seconds) 2011-11-27T23:31:02 same artist 2011-11-27T23:31:44 yes, I recognize the style ;) 2011-11-27T23:32:27 *** McLeopold has quit IRC (Quit: Leaving.) 2011-11-27T23:32:36 Go is just a better C 2011-11-27T23:32:47 D tries to be a better C++ 2011-11-27T23:32:53 *** avdg has quit IRC (Quit: Leaving.) 2011-11-27T23:33:13 I don't think either will succeed :) 2011-11-27T23:33:33 *** u_ has quit IRC (Quit: u_) 2011-11-27T23:33:37 correct. I want iterators, but then Andrei wants: http://www.digitalmars.com/d/archives/digitalmars/D/Iterators_Must_Go_89524.html 2011-11-27T23:34:36 Go is very nice for networking stuff, and I guess that's what google is going to use it for 2011-11-27T23:34:41 definitely less painful than erlang :P 2011-11-27T23:34:46 there are endless discussions in the community about every little bit and how it is done right or better. how should const work, how should concurrency work, do we need more keywords or less 2011-11-27T23:35:18 http://www.rust-lang.org/ yet another C++ :) 2011-11-27T23:35:30 it will be nicer than D imo 2011-11-27T23:35:47 and Mozilla actually makes software people use, so it will actually get used 2011-11-27T23:37:11 at least their own software projects xD 2011-11-27T23:37:56 hehe, Mozilla has a browser that is used, although it uses more memory than any other. they sure must do something right with their marketing 2011-11-27T23:38:09 it doesn't anymore 2011-11-27T23:38:24 it uses much lower memory than chrome per tab 2011-11-27T23:38:25 mleise: Chrome actually tends to eat more resources 2011-11-27T23:38:36 evening 2011-11-27T23:38:40 *** johnsonav has quit IRC (Remote host closed the connection) 2011-11-27T23:38:57 not to mention that Chrome can often be slower in actual use compared to benchmarks 2011-11-27T23:39:18 firefox 3 used a LOT of memory 2011-11-27T23:39:28 firefox 4 had a completely new js engine, so there were lots of leaks 2011-11-27T23:40:27 firefox just has a slow UI, it renders pages and runs javascript as fast as chrome 2011-11-27T23:40:39 that's what you get when you implement your UI in XML and javascript 2011-11-27T23:41:22 thestinger: Chrome implements its own UI in HTML+JS 2011-11-27T23:41:35 except it has much smaller UI to cover 2011-11-27T23:41:38 running in a different process though 2011-11-27T23:41:50 "A lot of thing that are statements in C are expressions in Rust. This allows for useless things" - lol 2011-11-27T23:42:02 firefox has 1 thread for JS iirc 2011-11-27T23:42:20 thestinger: actually it's multithreaded 2011-11-27T23:42:49 afaik by default two concurrent at once, modifiable through about:config, if I understood it correctly 2011-11-27T23:42:58 ah 2011-11-27T23:43:05 but it's still not separated yet 2011-11-27T23:43:10 of course not for one and the same tab 2011-11-27T23:43:23 https://wiki.mozilla.org/Electrolysis 2011-11-27T23:43:39 mleise: try running a crazy js loop on a page in ffx 2011-11-27T23:43:54 the UI will freeze until firefox shows a message asking if you want to kill the script 2011-11-27T23:44:19 thestinger: when I wrote the visualizer I had a lot of those ^^ 2011-11-27T23:44:24 thestinger: Usually running a crazy JS loop in ffx means to me "hmm, it got slower, but still works", while I get tab freezing up in Chrome 2011-11-27T23:44:47 yeah chrome has UI in a separate process (along with each tab) 2011-11-27T23:45:08 webkit can now do a process for the UI and one for content iirc 2011-11-27T23:45:16 chrome is multi process 2011-11-27T23:45:20 each tab is a separate process 2011-11-27T23:45:24 Oo 2011-11-27T23:45:34 why not a thread 2011-11-27T23:45:40 isolation 2011-11-27T23:45:42 plugins 2011-11-27T23:45:43 mleise: it's also sandboxed 2011-11-27T23:45:48 *** besh has joined #aichallenge 2011-11-27T23:45:49 faulty plugins don't screw up each other 2011-11-27T23:45:51 mleise: their whole UI is sandboxed too 2011-11-27T23:45:56 flash can't crash the browser 2011-11-27T23:46:01 only the active tab 2011-11-27T23:46:06 that's why chrome can't properly use your GTK theme/engine and font settings on linux 2011-11-27T23:46:07 chromium has a task manager actually 2011-11-27T23:46:10 flash is the reason for too many workarounds imo ^^ 2011-11-27T23:46:13 i can use it to kill flash 2011-11-27T23:46:24 flash is in a separate process in firefox now too 2011-11-27T23:46:30 same with google talk plugin etc. 2011-11-27T23:46:32 thestinger: works for me, with the theme 2011-11-27T23:46:41 amstan: look at the scrollbars 2011-11-27T23:46:43 and checkboxes 2011-11-27T23:46:47 it just copies the colors 2011-11-27T23:46:53 bhasker`: their multi-process separation also meant that for a long time, a runaway JS loop wouldn't ground the browser, it would kill the system 2011-11-27T23:46:54 it doesn't use the GTK engine 2011-11-27T23:47:01 becaus there was *no* resource management 2011-11-27T23:47:08 thestinger: the point is that the plugins crashin in one tab, don't crash content in other tabs i think 2011-11-27T23:47:14 thestinger: copies everything for me 2011-11-27T23:47:24 thestinger: it's all dark and stuff like my theme 2011-11-27T23:47:25 hmm a runaway js loop shouldn't kill the system 2011-11-27T23:47:27 *** rajanaresh has quit IRC (Ping timeout: 252 seconds) 2011-11-27T23:47:32 at most the active tab hogs the cpu 2011-11-27T23:47:37 i don't see how it can crash the system 2011-11-27T23:47:43 amstan: aren't the scrollbars different? 2011-11-27T23:47:48 it copies the color, but not the style 2011-11-27T23:47:53 oh.. wait. 2011-11-27T23:47:54 you're right 2011-11-27T23:47:58 they're not the same.. 2011-11-27T23:48:03 bhasker`: having loadavg a multiple of available cpus can ground the OS to death in practice 2011-11-27T23:48:03 yeah, it's just a hack 2011-11-27T23:48:07 they read the colors from the theme 2011-11-27T23:48:19 here's some rough visual notes I took to try to understand battle mechanics, drawn by hand so there could be mistakes, players are *not* differentiated by color http://i.imgur.com/w1DBU.png 2011-11-27T23:48:20 chromium also uses ffmpeg, fontconfig, freetype etc. 2011-11-27T23:48:22 bhasker`: especially when there's *zero* i/o 2011-11-27T23:48:26 but it can't be dynamically linked against them 2011-11-27T23:48:35 thestinger: it looks fairly good though 2011-11-27T23:48:38 yeah 2011-11-27T23:48:45 they copy the colors properly 2011-11-27T23:48:55 it probably looks good with default themes on ubuntu 2011-11-27T23:49:03 there shouldn't be a load average of multiple cpu's unless you are running the same JS in multiple tabs 2011-11-27T23:49:14 thestinger: the gtk theme that i'm using is also fake, since it copies the kde theme 2011-11-27T23:49:21 a single tab can't really be worse than a single process runaway JS in firefox 2011-11-27T23:49:23 amstan: the thing that gets me is that it doesn't use fontconfig settings 2011-11-27T23:49:25 so that's 2 levels of redirection, and it's still nice 2011-11-27T23:49:47 I use the infinality patches for freetype2, so it also has fonts that are ugly 2011-11-27T23:51:21 *** besh has quit IRC (Quit: Page closed) 2011-11-27T23:51:55 amstan: except that Chrome used to achieve that without malicious/broken JS code 2011-11-27T23:52:02 s/amstan/bhasker`/ 2011-11-27T23:52:42 http://ompldr.org/vYmhoZA ugly chromium fonts :( 2011-11-27T23:53:08 sadly that can't be fixed due to sandboxing, there's a bug report but I think they closed it 2011-11-27T23:53:43 probably hard to tell since you're not using my monitor :) 2011-11-27T23:54:07 *** delt0r_ has quit IRC (Ping timeout: 248 seconds) 2011-11-27T23:55:54 has anyone gotten an actual improvement from symmetry? 2011-11-27T23:56:06 Antimony: what? 2011-11-27T23:56:09 well that would be a bug in the browser :) not really in the concept of multiprocess sandboxing 2011-11-27T23:56:20 symmetry detection? 2011-11-27T23:56:39 i have been using chrome for the last couple of years now and its been far more stable than any of the latest firefox releases 2011-11-27T23:56:45 bhasker`: I don't think they can read my fontconfig settings due to the sandbox 2011-11-27T23:56:48 firefox has been getting slower and buggier with every new release 2011-11-27T23:56:57 thestinger possible 2011-11-27T23:57:11 bhasker`: what's the last ffx version you tried? 2011-11-27T23:57:31 4.x and since then they release new versions every 2-3 weeks 2011-11-27T23:57:41 so haven't bothered and i usually use the dev channel chrome 2011-11-27T23:57:42 4.x was terrible 2011-11-27T23:57:53 yea 2011-11-27T23:57:56 they have the same 6-week release cycle as chrome now 2011-11-27T23:58:05 well, 4.x was like a year late and they rushed it out near the end 2011-11-27T23:58:27 btw any of you guys using pledge algorithm for maze line maps? 2011-11-27T23:58:30 maze like? 2011-11-27T23:58:56 Antimony: try hardcoding hill locations for a specific map and see if you actually gain any benefit from knowing where they are 2011-11-27T23:58:57 anyway.. back to ants then 2011-11-27T23:59:01 I don't on sane maps 2011-11-27T23:59:34 No, my bot can deduce locatiosn but doens't seem to gain any benefit to it 2011-11-27T23:59:53 yeah, that's what I mean 2011-11-27T23:59:55 bhasker`: Chrome's networking code, however, seems to utterly suck. At least that's one of my guesses regarding why it locks up all the time