2011-12-14T00:01:03 it's battle of the best bugs sometimes 2011-12-14T00:03:14 *** Raimondi has quit IRC (Quit: Get MacIrssi - http://www.sysctl.co.uk/projects/macirssi/) 2011-12-14T00:04:13 all my improvements came from bugfixes 2011-12-14T00:04:18 which then added more bugs I had to fix 2011-12-14T00:04:37 and some of those bugs were good, so I kept them and turned them into features 2011-12-14T00:05:20 I understand each bit of code by itself, but some of the stuff my bot does I can't comprehend 2011-12-14T00:05:48 like my food search code is somehow a _huge_ factor in combat 2011-12-14T00:06:21 if I tweak it, my bot no longer retreats, etc. 2011-12-14T00:11:16 *** u_ has quit IRC (Quit: u_) 2011-12-14T00:19:49 *** praveen has joined #aichallenge 2011-12-14T00:24:39 *** Areks has joined #aichallenge 2011-12-14T00:29:29 my tron bot taught me a few things about tron 2011-12-14T00:31:45 does anyone know if it's possible for multiple sets of food tiles to spawn on a single turn? 2011-12-14T00:32:38 *** TheLinker has quit IRC (Remote host closed the connection) 2011-12-14T00:33:47 Antimony: what do you mean? 2011-12-14T00:34:35 "The entire map is divided into sets of squares that are symmetric. The sets are shuffled into a random order. When food is spawned, the next set is chosen. When all the sets have been chosen, they are shuffled again." 2011-12-14T00:34:51 oh, yes it should be possible 2011-12-14T00:34:57 but i doubt the spawn rate is set high enough 2011-12-14T00:35:08 you are aware of what happens when an ant resides on a spawn location right? 2011-12-14T00:35:23 the food queues up and waits to spawn there the next turn an ant doesn't reside there 2011-12-14T00:35:29 so statit ants == bad 2011-12-14T00:35:53 yes 2011-12-14T00:43:00 aichallenge: janzert epsilon * r32edc91 / manager/matchup_buffer.py : Fix null check and sleep in matchup buffer script - http://git.io/_Ko1iw 2011-12-14T00:50:53 ug, how long do these games take? 2011-12-14T00:51:26 my bot has been "playing a game" for like 40 minutes on the server 2011-12-14T00:51:42 it might be hours 2011-12-14T00:51:49 it probably isn't really in a game 2011-12-14T00:53:02 the profile page thinks you're in a game when a matchup has been created, but a whole bunch of pairings got pre-created by accident (because of the bug fixed in that last checkin) 2011-12-14T00:53:37 actually it should be about an hour for all the excess matches to clear out 2011-12-14T00:53:48 oh 2011-12-14T00:55:36 thestinger: did you ever do anything about moving dying ants out in battle? 2011-12-14T00:57:38 not really, I haven't really touched my code for a while 2011-12-14T00:57:42 I've just merged some stuff 2011-12-14T00:58:00 I might make one final improvement to combat so that my bot actually uses up all the turntime trying stuff 2011-12-14T00:58:07 i might just add a bunch more sets of heuristically chosen moves into my tree 2011-12-14T00:58:12 it's too messy to do the former 2011-12-14T00:58:42 *** andrei_c has joined #aichallenge 2011-12-14T00:59:20 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-12-14T01:00:57 *** praveen has quit IRC (Ping timeout: 258 seconds) 2011-12-14T01:01:16 aichallenge: janzert epsilon * rd43fd63 / website/sql.php : Don't count a player in a match until a worker has claimed the matchup - http://git.io/OLmhCg 2011-12-14T01:12:30 what does this mean? 2011-12-14T01:12:31 HILL_ANT = string = 'ABCDEFGHIJ' 2011-12-14T01:12:58 the ant above the hill is on Capital letter? 2011-12-14T01:13:12 *** Jak_o_Shadows1 has quit IRC (Remote host closed the connection) 2011-12-14T01:13:50 I've found it on ants.py line 23 2011-12-14T01:14:58 from the map format 2011-12-14T01:15:03 A is 0's hill etc 2011-12-14T01:15:06 AntDroid: it's the syntax for maps. since those are plaintext with one character per tile, there is no way to say "there is a hill AND an ant on this tile" other than using a different symbol. 2011-12-14T01:15:39 21:31 < Antimony> does anyone know if it's possible for multiple sets of food tiles to spawn on a single turn? 2011-12-14T01:15:51 Antimony: interesting question ;-) 2011-12-14T01:16:18 so on the one hand I'm pleased to win 8 games in a row 2011-12-14T01:16:35 on the other hand I could have done with matchups that would have counted for more than 639th! 2011-12-14T01:17:38 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-12-14T01:17:39 bqf: thanks. my current bot doesnt recognize hills when there is an ant over it. there are lots of games where it abandoned attacking the hill as the enemy ant covered it. 2011-12-14T01:17:52 AntDroid: the server does *send* those 2011-12-14T01:18:23 *doesn't 2011-12-14T01:18:47 oh, sorry. you meant "it send a hill AND an ant info", not "it send the uppercase letters". carry on then. 2011-12-14T01:19:05 yeah there might be a starter package out there that "loses" that info 2011-12-14T01:19:10 why both those "send" are missing an "s", i will never know. 2011-12-14T01:19:14 since I've seen comments about ants "covering" hills 2011-12-14T01:20:21 BenJackson: i'm confused. so it is not sending the Capital letters instead a separate input like "h 1 1 0" and "a 1 1 0" 2011-12-14T01:21:18 i've used the C starter package by the way and it has no hills support 2011-12-14T01:21:38 AntDroid: yes. as i said, the capital letters are from the map format - you don't need to care about that unless you want to create scenarios for your bot or work on the map generators. 2011-12-14T01:21:39 right, it sends h 1 1 0 and a 1 1 0 2011-12-14T01:23:22 okay. thanks guys. need to fix this bug. my bot assumed the hill is razed because the enemy ant covered it. 2011-12-14T01:24:58 *** epicmonkey has joined #aichallenge 2011-12-14T01:28:53 *** Fandekasp has joined #aichallenge 2011-12-14T01:29:21 *** ccarbonix has joined #aichallenge 2011-12-14T01:29:56 once a hill is razed you wont receive information from the engine about it anymore 2011-12-14T01:33:58 *** dr- has quit IRC (Ping timeout: 248 seconds) 2011-12-14T01:35:17 *** andrei_c has quit IRC (Quit: Page closed) 2011-12-14T01:35:40 janzert: problem is that my map overwrites the hill with ant, so when i loop over the map the hill can't be seen. the C starter package by FlagCapper which I'm using has 2 levels of initialization. one is for the map table and one is for game data. 2011-12-14T01:36:34 the hill is overwritten at the first initialization 2011-12-14T01:39:15 <_flag> AntDroid: have you checked the newest starter package? 2011-12-14T01:41:13 _flag: not yet. do you have hill support for it now? i have changed the base code a lot now so your changes may not be compatible with my bot anymore. 2011-12-14T01:45:23 *** Palmik has joined #aichallenge 2011-12-14T01:45:41 janzert: we're doing a live scoreboard during the final contest again yeah? 2011-12-14T01:45:48 <_flag> AntDroid: It got updated to support hills awhile ago, but yes it may not be compatible 2011-12-14T01:47:07 how hard is it to set up a tcp server? anyone have any ideas? 2011-12-14T01:47:12 i want to set up a tron one :P 2011-12-14T01:47:48 *** avdg has quit IRC (Quit: Leaving.) 2011-12-14T01:49:04 aichallenge: janzert epsilon * re14eb5f / (manager/matchup_buffer.py manager/matchup_buffer.py): Make matchup_buffer executable - http://git.io/Z8N6og 2011-12-14T01:55:20 *** Kurnevsky has joined #aichallenge 2011-12-14T01:55:40 so i've finally watched a few xathis games 2011-12-14T01:55:43 what makes his bot so good? 2011-12-14T01:55:51 there's nothing particularly amazing about his battle code 2011-12-14T01:56:39 he must be controlling/exploring the map better than everyone? 2011-12-14T01:58:36 <_flag> antimatroid1: His battle code is the best I've seen so far actually 2011-12-14T01:58:48 <_flag> Because it focuses on controlling space very well and not giving up any area 2011-12-14T01:58:59 <_flag> It's simple but effective 2011-12-14T01:59:07 <_flag> And yes, he does explore well, but he did that before 2011-12-14T01:59:17 <_flag> The change in battle code seems to be the most major improvement 2011-12-14T02:02:37 hmmm okay 2011-12-14T02:02:41 i'll take your word for it 2011-12-14T02:02:48 i am pretty happy with my battle code 2011-12-14T02:02:53 i just need to get the rest working alright 2011-12-14T02:07:57 I didn't really watch his battles that closely 2011-12-14T02:08:05 I think his main strength is the efficiency of his ant usage 2011-12-14T02:08:10 and the timely delivery where they're needed 2011-12-14T02:09:21 the secret to a strong bot is having no weaknesses ;) 2011-12-14T02:11:08 haha I reloaded my profile because it had said "in a game" 2011-12-14T02:11:15 my rank went from 640 to 639 in the reload 2011-12-14T02:11:16 <_flag> janzert: I know that replacing bots directly into their spot in the rankings was removed because some people go to used to playing only top bots, so I was wondering (earlier on irc) if maybe there should instead be an optional reset feature? 2011-12-14T02:11:23 I was really wondering wtf 2011-12-14T02:11:26 but of course it was nothing 2011-12-14T02:11:50 janzert: or an optional "please seed me against skill=n" 2011-12-14T02:12:17 this round has been particularly slow for me because even my 6th game was against ~1000 ranked players 2011-12-14T02:12:30 <_flag> BenJackson: Well, then everyone would want to be seeded against top players 2011-12-14T02:12:43 I think the effort to implement an individual reset is more than the people that would take advantage of it since the people that want it can just resubmit now 2011-12-14T02:12:45 not people who didn't expect to get there. 2011-12-14T02:12:59 <_flag> You should be seeded wherever your mu dicates you should be seeded imo 2011-12-14T02:13:07 _flag: your OLD mu? 2011-12-14T02:13:13 I'm just talking about after a reset 2011-12-14T02:13:13 allowing people to choose opponents is bad for rankings 2011-12-14T02:13:20 <_flag> janzert: I mean for people who complain that their bots don't get practice with new bots 2011-12-14T02:13:23 *** raemde_ has joined #aichallenge 2011-12-14T02:13:31 <_flag> I would like to be placed in the same spot in the rankings like in pw 2011-12-14T02:13:33 tell 'em to resubmit :) 2011-12-14T02:13:45 <_flag> But if that is really the problem, then give those people an option 2011-12-14T02:13:50 woo hoo, I got a 10 player game out of the way 2011-12-14T02:13:50 *** Gnome has joined #aichallenge 2011-12-14T02:14:43 <_flag> I should've been more clear, I'm saying that we should go back to the pw system and give people who want to climb through the stats again that option 2011-12-14T02:14:46 a new submission should prove that it still does well against the range of players coming up through the ranks 2011-12-14T02:14:58 *** McLeopold1 has joined #aichallenge 2011-12-14T02:14:58 <_flag> But it will have to do that anyway in the final contest 2011-12-14T02:15:11 http://aichallenge.org/visualizer.php?game=226969&user=3473 2011-12-14T02:15:15 oh my cap of dark yellow in that is brutal 2011-12-14T02:15:18 yep, but no need to distort the open period rankings when we don't need to 2011-12-14T02:15:28 <_flag> This contest the top bots were in actual fact last weeks top bots 2011-12-14T02:16:49 <_flag> janzert: I'm just saying that the main site is basically useless as a method of testing anything, and it shouldn't be that I have to rely on tcp to get decent games 2011-12-14T02:17:21 the main site has always been a problem for testing if you're any good, cause the game rate is too slow 2011-12-14T02:18:31 pairofdice: hey, you're in that game too 2011-12-14T02:19:00 *** kapoc has joined #aichallenge 2011-12-14T02:19:03 *** McLeopold has quit IRC (Ping timeout: 245 seconds) 2011-12-14T02:19:03 *** raemde has quit IRC (Ping timeout: 245 seconds) 2011-12-14T02:19:32 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-12-14T02:20:38 so i'm 11th 2011-12-14T02:20:57 Fluxid: I kept 11th warm for a while 2011-12-14T02:21:01 i guess it's just pure luck and i'll fall down 2011-12-14T02:21:02 got a good view of 10th 2011-12-14T02:21:25 I really have no idea if this new sub is going to actually do better 2011-12-14T02:22:02 i'm higher in rankings than thestinger so i guess it's impossible for me to stay so high, he's much better than me... 2011-12-14T02:23:14 there are definitely areas in the top 20 that aren't "settling" 2011-12-14T02:23:21 maybe they will in the finals 2011-12-14T02:23:42 hey, xathis has mu = 100.79 2011-12-14T02:23:54 i thought you can't go over 100 2011-12-14T02:24:14 as long as he doesn't go over 9000 we'll be fine. 2011-12-14T02:24:23 that just means on average he beats everybody AND some people who aren't even participating 2011-12-14T02:26:48 Fluxid: there is no hard limit on mu, just more unlikely the farther away from 50 2011-12-14T02:26:50 which is actually kind of true; i'm not participating, but have been rooting for everyone _but_ xathis (just because he got his lead very early on). so by being at #1 again, he is "beating" me. 2011-12-14T02:27:09 at least 99% of bots should be between 0-100 2011-12-14T02:27:15 there's a game difficulty factor in the algorithm 2011-12-14T02:27:24 if that's misset, ratings can go wider or narrowed easily 2011-12-14T02:27:38 janzert: so what you are saying is... xathis is the 1%? (i'm so sorry.) 2011-12-14T02:27:44 :) 2011-12-14T02:27:46 I'm guessing it can compress again with more games 2011-12-14T02:27:55 probably what trueskill will try to do over time is push everyone below xathis down (so he can have the right amount of skill points over #2) 2011-12-14T02:28:04 but there has to be a lot of game play to make that happen 2011-12-14T02:32:25 *** ccarbonix has quit IRC (Quit: Page closed) 2011-12-14T02:32:56 BenJackson: based on that game your BFS seems not to stop at an ant for unexplored areas. that snake behavior is an emergent behavior of that technique. 2011-12-14T02:34:07 I only have a few pathing maps floating around 2011-12-14T02:34:20 when I have a lot of map visibility there's not much variation for each ant 2011-12-14T02:34:31 I go to food individually 2011-12-14T02:34:39 and in this latest version to enemy ants 2011-12-14T02:34:57 but when I have a huge excess of ants like that they do get aimless 2011-12-14T02:35:11 if I were starting over I'd rebudget my time 2011-12-14T02:35:18 cheap combat, expensive ant utilization 2011-12-14T02:35:39 I just didn't forsee how after combat defined the top tier it would be redefined by strategy again 2011-12-14T02:36:09 my ant doesn't move when it doesn't have any unexplored areas to go to 2011-12-14T02:36:09 *** kapoc has quit IRC (Ping timeout: 252 seconds) 2011-12-14T02:36:38 easiest "solution" to that is to redefine "unexplored" as "not explored recently" 2011-12-14T02:36:56 basically i do combat, food, defense, attack enemy hill and explore in order 2011-12-14T02:37:28 i only do "not explored recently" for non-visible tiles 2011-12-14T02:37:33 I did all those things simultaneously 2011-12-14T02:37:35 then split combat 2011-12-14T02:37:44 then split food 2011-12-14T02:37:51 then sort of split "maintain vision" 2011-12-14T02:37:57 finally split "defense" 2011-12-14T02:38:56 i have after combat and food, 10% of my ants on defense and 50% going for enemy hill attack. the remaining is for exploration 2011-12-14T02:39:48 i'm currently trying a1k0n's idea, wish i have tried this last week, seems not enough time to make this work well now 2011-12-14T02:40:02 ooh I got a double game 2011-12-14T02:40:39 10 in a row to 231st 2011-12-14T02:40:50 I seem to remember skill 74 used to be a lot higher than 231st! 2011-12-14T02:42:02 turns out I was exactly 231st last time at 72.7 2011-12-14T02:42:05 *** matthewd has quit IRC (Remote host closed the connection) 2011-12-14T02:42:43 the skill to get into the top100 increased gradually as more bots gets better 2011-12-14T02:43:01 *** matthewd has joined #aichallenge 2011-12-14T02:45:04 Fluxid: my bot just has a high sigma atm 2011-12-14T02:45:51 if you look at the sigma of the other bots above you, you can sort of guess that's you will actually go up more if mu is somewhat accurate 2011-12-14T02:45:58 that you will* 2011-12-14T02:46:15 McLeopold makes new top-mu queries from time to time 2011-12-14T02:51:55 *** replore_ has quit IRC (Read error: Connection reset by peer) 2011-12-14T02:52:25 *** replore__ has joined #aichallenge 2011-12-14T02:56:27 as is traditional I played #8 on my climb 2011-12-14T02:56:43 amstan: sure would be interesting to have the seed marked on the game page 2011-12-14T02:59:23 *** monjo has joined #aichallenge 2011-12-14T02:59:59 I'm on some kind of multi-game wave 2011-12-14T03:02:00 thestinger: you got me in my 12th game 2011-12-14T03:03:37 *** monjo has quit IRC (Client Quit) 2011-12-14T03:03:50 BenJackson: silly map though :P 2011-12-14T03:05:20 wow game rate 53 2011-12-14T03:06:33 thestinger: I suppose ideally we would have walled each other off and each taken half the map 2011-12-14T03:06:36 low food for that 2011-12-14T03:06:51 *** praveen has joined #aichallenge 2011-12-14T03:07:28 *** kaemo has quit IRC (Ping timeout: 268 seconds) 2011-12-14T03:10:39 after some turns, I just either attack/defend the enemy and this results in nearly no exploration. how to balance this? 2011-12-14T03:11:11 no exploration and also it doesnt gather food 2011-12-14T03:17:00 imperfect information is a pita 2011-12-14T03:17:10 i wish i knew enemy ant counts :P 2011-12-14T03:18:36 *** aerique has joined #aichallenge 2011-12-14T03:20:52 *** arscan has quit IRC (Ping timeout: 258 seconds) 2011-12-14T03:20:55 antimatroid1: it's easy - just assume your bot is perfect, i.e. all others can have AT MOST as many ants as you. it's a reasonable assumption if you're aiming for #1. (no, it's not.) 2011-12-14T03:21:10 *** Kurnevsky has quit IRC (Quit: Instantbird 1.0) 2011-12-14T03:22:46 bqf: that's a terrible assumption :P 2011-12-14T03:22:54 *** mviel has joined #aichallenge 2011-12-14T03:23:05 what one wants to know is when they are significantly ahead 2011-12-14T03:23:12 so you can just start maximising enemy deaths 2011-12-14T03:24:05 antimatroid1: i hope that part in parantheses was indication enough that i was kidding. 2011-12-14T03:24:07 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-12-14T03:24:15 yes i know :P 2011-12-14T03:24:35 i'm also not aiming for #1, but anywhere below that would be awesome 2011-12-14T03:24:47 slowly getting less awesome the higher the number 2011-12-14T03:27:04 *** treeform has quit IRC (Remote host closed the connection) 2011-12-14T03:27:27 i would say that combining ant counting and map dominance would actually lead to an acceptable heuristic. most "strong" bots don't seem to be easily boxed in, so if you can see 70% of the map you can probably make some rather well-founded assumptions (that could still be completely off, of course - but you'd have _something_ to go on) 2011-12-14T03:29:33 sure, but what about n>2 player games? :P 2011-12-14T03:30:05 antimatroid1: I can pretty much guarantee that if you're not #1 you'll be somewhere below that :P 2011-12-14T03:30:16 janzert: pretty much? 2011-12-14T03:30:36 yep, chance of the universe ending and such 2011-12-14T03:30:39 i look forward to watching my bots performance in the finals 2011-12-14T03:30:45 i just hope it actually runs on the contest servers :P 2011-12-14T03:30:52 hehe 2011-12-14T03:31:27 janzert: don't be too sure, if xathis can have mu > 100 maybe he'll be #0 2011-12-14T03:31:31 janzert: did you see my comment before about live leaderboard for the finals? 2011-12-14T03:31:46 no I didn't 2011-12-14T03:31:54 just wanted to confirm that is the case? 2011-12-14T03:31:59 there's a post on the forums asking about it 2011-12-14T03:32:40 we don't have a way to shut it off currently and I don't think it would be particularly easy to change either 2011-12-14T03:32:51 i don't think we'd want to anyway 2011-12-14T03:32:52 *** Blkt has joined #aichallenge 2011-12-14T03:32:59 it's much more exciting having the leaderboard be live during finals 2011-12-14T03:33:07 hmm, I wouldn't mind hiding the top 100 positions 2011-12-14T03:33:16 for the last part 2011-12-14T03:33:17 so, maybe I'm missing something, but is there a contest calendar anywhere that says when entries close, last submissions are due, finals are run, etc? 2011-12-14T03:33:24 hmmmm 2011-12-14T03:33:34 kevlar: bottom of the front page 2011-12-14T03:33:42 personally i would just leave it open but i see why one might not 2011-12-14T03:33:48 but all that is set is time for open competition to close and finals to start 2011-12-14T03:34:07 antimatroid1: yeah, I'm not real strong on it either way 2011-12-14T03:34:11 gasp! couldn't possibly be under the "contest schedule" heading *feels silly* 2011-12-14T03:34:32 probably why I haven't bothered trying to change it :) 2011-12-14T03:34:41 kevlar: :) 2011-12-14T03:34:44 BenJackson: any preferences on the leaderboard being publicly available during the final contest? 2011-12-14T03:35:10 if there's a "fog of war" at the top of the leaderboard, we should still be able to watch their games :) 2011-12-14T03:35:13 I think both ways will cause great angst 2011-12-14T03:35:14 janzert: what might be cool is hiding the rankings but showing the games (probably not feasible for this, but in the future?) 2011-12-14T03:35:19 but I'd rather see it myself 2011-12-14T03:35:22 or at least the last few games, so you can't figure out the rankings yourself or something 2011-12-14T03:35:27 kevlar beat me to it 2011-12-14T03:35:32 antimatroid1: that's what my thought was to do 2011-12-14T03:35:35 from the forums there seems to have been a bit of a dustup over how the last finals went down 2011-12-14T03:35:49 hide all display of skill and ranking for top 100 bots basically 2011-12-14T03:35:50 BenJackson: link? and what is a dustup? 2011-12-14T03:35:55 for the final 24 hours or something 2011-12-14T03:35:59 but I think if people can see live how positions bounce around and how their own rank is partly determined by the arbitrary stopping point... 2011-12-14T03:36:24 antimatroid1: slang for a scuffle 2011-12-14T03:36:25 a fight 2011-12-14T03:36:33 for planetwars the last 12-18 hours ranking display was frozen is all 2011-12-14T03:36:39 so, what strategies have people disclosed about how they plan for combat? 2011-12-14T03:36:48 kevlar: check the forum 2011-12-14T03:36:51 there are two big threads 2011-12-14T03:36:58 game trees! 2011-12-14T03:37:00 BenJackson: I don't remember anything like that in particular 2011-12-14T03:37:03 janzert: maybe you can rely on memcached bugs for that ;-) 2011-12-14T03:37:08 kevlar: you are quite late, but still fun to play around :) 2011-12-14T03:37:08 :) 2011-12-14T03:37:15 yeah, I'm *way* late 2011-12-14T03:37:31 kevlar: someone made top 100 starting with like a week to go last year 2011-12-14T03:37:34 but I managed to get to 1600 on my second day, which I consider pretty good :) 2011-12-14T03:37:35 burny 2011-12-14T03:37:52 oh, 1488 now. 2011-12-14T03:38:15 and the one online is actually a really silly algorithm, it shouldn't be doing that well... 2011-12-14T03:38:26 (but that's kinda how things go in AI) 2011-12-14T03:39:20 *** raveous has joined #aichallenge 2011-12-14T03:41:59 that's exactly how things word in ai :P 2011-12-14T03:42:17 i'm saving my bot to submit and have it come in and surprise people :P 2011-12-14T03:42:28 (hopefully surprise people anyway) 2011-12-14T03:44:35 aichallenge: janzert epsilon * rd50e32d / manager/matchup_buffer.py : Increase matchup buffer size - http://git.io/B7QCDA 2011-12-14T03:45:00 I'm kinda playing around. 2011-12-14T03:45:22 I started off trying to code up something similar to the diffusion thing that is discussed on the forums 2011-12-14T03:45:25 antimatroid1: sounds like a recipe to mostly surprise yourself ;) 2011-12-14T03:45:31 but I was apparently making it waaaay more difficult than it had to be 2011-12-14T03:45:44 janzert: it hasn't been ready anyway 2011-12-14T03:45:45 so I may try that out just for pokes and giggles. 2011-12-14T03:53:46 thestinger: i changed to calculating the enemies moves for each set of my moves 2011-12-14T03:57:24 *** mj41 has joined #aichallenge 2011-12-14T04:08:18 the turntime on my local testing seems to be 1 second. may i know where can i change the settings for turn time to 500ms? i can't seem to find where it is being set 2011-12-14T04:09:14 AntDroid: you can set it with an option to playgame.py, pretty sure it's --turntime 2011-12-14T04:11:37 *** epicmonkey has joined #aichallenge 2011-12-14T04:12:01 janzert: okay. thanks. i thought it needs to be set in one of those*.py files 2011-12-14T04:18:47 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-12-14T04:19:57 *** ikaros has joined #aichallenge 2011-12-14T04:24:09 good morning everyone 2011-12-14T04:24:13 *** kire has joined #aichallenge 2011-12-14T04:27:26 *** Gnome has quit IRC (Quit: Gnome) 2011-12-14T04:31:47 Blkt: morning Blkt 2011-12-14T04:32:45 well that looks silly 2011-12-14T04:34:28 *** raveous has quit IRC (Ping timeout: 258 seconds) 2011-12-14T04:35:42 hi aerique 2011-12-14T04:41:18 *** amstan has quit IRC (Ping timeout: 255 seconds) 2011-12-14T04:41:21 Blkt: how are you doing in the contest? 2011-12-14T05:02:43 *** replore__ has quit IRC (Remote host closed the connection) 2011-12-14T05:02:57 *** antimatroid1 has quit IRC (Read error: No route to host) 2011-12-14T05:03:06 *** antimatroid has joined #aichallenge 2011-12-14T05:11:17 *** GeorgeSebastian has quit IRC (Read error: Connection reset by peer) 2011-12-14T05:25:10 quiet channel 2011-12-14T05:25:25 are people programming furiously or sleeping? 2011-12-14T05:28:52 *** GeorgeSebastian has joined #aichallenge 2011-12-14T05:29:07 *** sigh has joined #aichallenge 2011-12-14T05:36:57 *** AntDroid has quit IRC (Ping timeout: 258 seconds) 2011-12-14T05:37:32 *** g0llum has joined #aichallenge 2011-12-14T05:37:41 *** GeorgeSebastian has quit IRC (Ping timeout: 252 seconds) 2011-12-14T05:45:10 *** Darhuuk has quit IRC (Ping timeout: 252 seconds) 2011-12-14T05:46:41 *** Anilm3 has joined #aichallenge 2011-12-14T05:46:52 @rankings 2011-12-14T05:46:53 Anilm3: Top 10 players: xathis(94.4), FlagCapper(93.8), pguillory(92.8), a1k0n(91.3), fourmidable(90.5), GreenTea(89.4), delineate(89.0), ThisIsNotABug(88.7), jacob_strauss(87.8), cumbuz(86.9) 2011-12-14T05:48:02 *** Darhuuk has joined #aichallenge 2011-12-14T05:50:09 *** Garf has quit IRC (Read error: Connection reset by peer) 2011-12-14T05:50:33 *** Garf has joined #aichallenge 2011-12-14T05:51:34 *** Garf has quit IRC (Read error: Connection reset by peer) 2011-12-14T05:51:58 *** Garf has joined #aichallenge 2011-12-14T05:53:15 *** Garf has joined #aichallenge 2011-12-14T05:54:14 *** nplus has quit IRC (Ping timeout: 248 seconds) 2011-12-14T05:54:18 *** GarfTop has joined #aichallenge 2011-12-14T05:55:06 *** b0rder has quit IRC (Quit: 离开) 2011-12-14T05:57:55 *** Garf has quit IRC (Ping timeout: 240 seconds) 2011-12-14T05:59:01 *** mviel_ has joined #aichallenge 2011-12-14T05:59:11 *** nplus has joined #aichallenge 2011-12-14T06:01:46 *** pairofdice has joined #aichallenge 2011-12-14T06:02:27 *** mviel has quit IRC (Ping timeout: 240 seconds) 2011-12-14T06:02:49 *** dmj111 has quit IRC (Quit: ERC Version 5.3 (IRC client for Emacs)) 2011-12-14T06:17:12 *** liuw has quit IRC (Read error: Connection reset by peer) 2011-12-14T06:17:40 *** nha has joined #aichallenge 2011-12-14T06:19:23 yay, I cracked the top 20 :) 2011-12-14T06:20:03 now I can die happy 2011-12-14T06:21:27 congrats 2011-12-14T06:29:27 *** aerique has quit IRC (Ping timeout: 240 seconds) 2011-12-14T06:29:32 *** aerique has joined #aichallenge 2011-12-14T06:29:40 *** dmj111 has joined #aichallenge 2011-12-14T06:40:46 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-12-14T06:41:06 dmj111: hello :) 2011-12-14T06:41:10 how's the bot coming along? 2011-12-14T06:47:31 *** ikaros_ has joined #aichallenge 2011-12-14T06:57:35 http://aichallenge.org/visualizer.php?game=229500&user=757 i'm stupid 2011-12-14T07:04:57 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-12-14T07:07:40 *** Conorach has joined #aichallenge 2011-12-14T07:11:18 *** SimonT has joined #aichallenge 2011-12-14T07:11:47 *** nplus has quit IRC (Ping timeout: 240 seconds) 2011-12-14T07:11:53 *** SimonT is now known as nplus 2011-12-14T07:12:05 *** amstan__ has joined #aichallenge 2011-12-14T07:12:05 *** ChanServ sets mode: +o amstan__ 2011-12-14T07:17:53 http://fluxid.pl/private/2011-12-14-131646_722x1593_scrot.png uh. sometimes i get rankings like that 2011-12-14T07:18:11 so xathis is not visible and suddelny lordap is 9th etc 2011-12-14T07:18:25 ah, it shows organization Other only 2011-12-14T07:18:55 but i clicked current rankings 2011-12-14T07:19:30 on first page i see only Other organization, but on second page normal rankings... 2011-12-14T07:19:58 something is not good with page, caching maybe>? 2011-12-14T07:20:35 *** SMJ has joined #aichallenge 2011-12-14T07:22:02 *** mviel__ has joined #aichallenge 2011-12-14T07:23:13 *** amstan__ has quit IRC (Ping timeout: 276 seconds) 2011-12-14T07:24:07 hey antimatroid! 2011-12-14T07:24:17 sup? 2011-12-14T07:24:28 not much 2011-12-14T07:24:50 as to your question... I have been trying hard to _not_ work on it. 2011-12-14T07:25:03 trying to maximize my fun per unit time 2011-12-14T07:25:17 hehe fair enough 2011-12-14T07:25:24 i'm still trying to get a bot i can submit 2011-12-14T07:25:28 *** mviel_ has quit IRC (Ping timeout: 245 seconds) 2011-12-14T07:25:30 good luck! 2011-12-14T07:25:36 cheers 2011-12-14T07:25:52 i worked on mine some more since my last submission, but am not doing any more on it. 2011-12-14T07:26:13 *** dmj111 has quit IRC (Quit: ERC Version 5.3 (IRC client for Emacs)) 2011-12-14T07:30:21 And similar case, normal rankings in organization rankings http://fluxid.pl/private/2011-12-14-132932_716x813_scrot.png 2011-12-14T07:30:26 McLeopold1: ↑ 2011-12-14T07:34:32 *** Anilm3 has joined #aichallenge 2011-12-14T07:34:54 hey guys 2011-12-14T07:34:58 @rankings 2011-12-14T07:34:59 Anilm3: Top 10 players: xathis(94.5), FlagCapper(94.0), pguillory(92.9), a1k0n(91.0), fourmidable(90.8), delineate(88.8), ThisIsNotABug(88.6), GreenTea(88.2), jacob_strauss(87.9), strcat(87.3) 2011-12-14T07:35:46 anilm3: what user account are you using? 2011-12-14T07:37:08 it's a secret 2011-12-14T07:38:11 GreenTea beat xathis in 1v1 2011-12-14T07:38:37 yeah 2011-12-14T07:38:43 Anilm3: that's no fun 2011-12-14T07:38:46 and _flag beat pguillory 2011-12-14T07:39:25 sorry I meant greentea 2011-12-14T07:39:37 <_flag> Anilm3: Watch the game, I didn't 2011-12-14T07:39:56 _flag: I meant greentea 2011-12-14T07:40:07 oh he timed out 2011-12-14T07:40:13 <_flag> 1v1s are tricky anyway because they're played counter intuitive to a regular game 2011-12-14T07:40:30 are you really 16 years old? 2011-12-14T07:40:40 is flag 16? 2011-12-14T07:40:44 yeah 2011-12-14T07:40:45 <_flag> In a 1v1 agression is good and regular games it's bad (generalization) 2011-12-14T07:40:46 <_flag> yes 2011-12-14T07:40:54 are you in uni? 2011-12-14T07:40:58 <_flag> no 2011-12-14T07:41:05 he's a prodigy 2011-12-14T07:41:08 * antimatroid is now rooting for flag 2011-12-14T07:41:15 one of my friends started uni at 15 2011-12-14T07:41:19 antimatroid: if he wins we all look bad xD 2011-12-14T07:41:36 <_flag> antimatroid: Home schooled? 2011-12-14T07:41:49 _flag: nope, just trying to work out whether it was 15 or 16 2011-12-14T07:41:54 one of my friends went straight from middle school to uni 2011-12-14T07:42:02 _flag: it's impressive though, how much time do you spend a day on this? 2011-12-14T07:42:05 skipped hischool 2011-12-14T07:42:09 *** u_ has joined #aichallenge 2011-12-14T07:42:21 <_flag> I was under the impression that most people who start uni at a young age are homeschooled, because at least where I am nobody skips anything 2011-12-14T07:42:46 <_flag> Anilm3: A few hours a day, depends on the day 2011-12-14T07:42:51 *** u__ has joined #aichallenge 2011-12-14T07:42:52 _flag: where I live no one skips anything either 2011-12-14T07:42:57 i've known of a few people in australia to skip ahead a few grades 2011-12-14T07:43:02 it's very strange 2011-12-14T07:43:03 i wish i had 2011-12-14T07:43:06 school sucked! 2011-12-14T07:43:08 haha 2011-12-14T07:43:10 uni was/is awesome 2011-12-14T07:43:15 tell me about it 2011-12-14T07:43:16 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-12-14T07:43:37 you scared the guy 2011-12-14T07:43:43 :p 2011-12-14T07:44:28 antimatroid: my name doesn't really matter since I abandoned my bot on december the 3rd 2011-12-14T07:46:16 *** u_ has quit IRC (Ping timeout: 240 seconds) 2011-12-14T07:46:16 *** u__ is now known as u_ 2011-12-14T07:46:53 *** goffrie has quit IRC (Ping timeout: 244 seconds) 2011-12-14T07:47:13 *** goffrie has joined #aichallenge 2011-12-14T07:58:32 *** Areks has quit IRC (Ping timeout: 268 seconds) 2011-12-14T08:08:23 *** Areks has joined #aichallenge 2011-12-14T08:12:30 *** UncleVasya has joined #aichallenge 2011-12-14T08:16:10 *** sigh has quit IRC (Remote host closed the connection) 2011-12-14T08:24:00 *** mviel_ has joined #aichallenge 2011-12-14T08:26:51 *** liberforce has joined #aichallenge 2011-12-14T08:27:27 *** mviel__ has quit IRC (Ping timeout: 240 seconds) 2011-12-14T08:28:22 *** mviel__ has joined #aichallenge 2011-12-14T08:28:47 *** mviel_ has quit IRC (Ping timeout: 240 seconds) 2011-12-14T08:29:10 http://aichallenge.org/visualizer.php?game=230372&user=757 aaaaawesome game :( 2011-12-14T08:38:40 Fluxid: very cool 2011-12-14T08:40:07 *** mviel_ has joined #aichallenge 2011-12-14T08:43:27 *** mviel__ has quit IRC (Ping timeout: 240 seconds) 2011-12-14T08:43:31 *** mj41_nb has joined #aichallenge 2011-12-14T08:55:55 thestinger: would sets of sets be horribly slow? 2011-12-14T08:56:08 well, unordered_sets of unoredered_sets 2011-12-14T09:09:34 a1k0n: fourth, that's great I didn't know you had made any changes 2011-12-14T09:10:19 *** Antimony has joined #aichallenge 2011-12-14T09:10:26 Hi antimatroid 2011-12-14T09:10:29 Antimony: 2011-12-14T09:10:34 hey 2011-12-14T09:10:46 i'm thinking about battles :) 2011-12-14T09:10:49 tab is a nightmare 2011-12-14T09:10:55 haha 2011-12-14T09:11:19 Anilm3: he hasn't; his bot is a week old. 2011-12-14T09:11:27 antimatroid: that depends on what you're trying to do with them ;) 2011-12-14T09:11:46 hi 2011-12-14T09:12:02 is there anyway to automatically download a bunch of replay input files? 2011-12-14T09:13:37 bqf: be smarter 2011-12-14T09:14:16 antimatroid: E=mc² 2011-12-14T09:14:19 http://pastebin.com/NZExUe8c 2011-12-14T09:14:25 there's a battle scenario for you guys 2011-12-14T09:15:05 you can't move the top left 'a' down 2011-12-14T09:15:20 *** TheLinker has joined #aichallenge 2011-12-14T09:17:25 *** besh has joined #aichallenge 2011-12-14T09:18:22 is tcpants.com website down ? 2011-12-14T09:18:50 my bot is playing games there but can't seem to view the results. 2011-12-14T09:22:42 i think i have a neat idea for picking combat moves 2011-12-14T09:23:01 max(min(possibleFriends)-possibleEnemies) :) 2011-12-14T09:37:12 *** Areks has quit IRC (Ping timeout: 268 seconds) 2011-12-14T09:42:05 *** bergmark has joined #aichallenge 2011-12-14T09:42:11 *** mikewintermute has joined #aichallenge 2011-12-14T09:47:59 *** Kurnevsky has joined #aichallenge 2011-12-14T09:53:08 *** kire has quit IRC (Quit: Leaving) 2011-12-14T09:57:39 besh: i cant access tcpants also 2011-12-14T09:59:38 tc pants 2011-12-14T09:59:51 now, what does t and c stand for... 2011-12-14T10:02:45 It is a mystery 2011-12-14T10:03:20 TL_: yes. I played many games in the night but i can't see the results to deciide if it improved or not ... 2011-12-14T10:03:36 thestinger: http://aichallenge.org/visualizer.php?game=230994&user=757 bawww :( 2011-12-14T10:07:20 *** tcoppi has quit IRC (Ping timeout: 252 seconds) 2011-12-14T10:11:03 can i deactivate a recently submitted version and start and old one ? 2011-12-14T10:12:11 i want to submit a new version and also have the option to deactivate it if it performs worse 2011-12-14T10:13:42 besh: don't think so 2011-12-14T10:14:52 You have to reupload the old version if you want to switch back 2011-12-14T10:15:37 *** avdg has joined #aichallenge 2011-12-14T10:15:52 besh: all rankings are reset at the very end 2011-12-14T10:16:01 if that's what you're worried about 2011-12-14T10:17:11 I have to wait until tcpants website is up to see if my new version is improved before submitting. 2011-12-14T10:21:00 *** praveen has quit IRC (Ping timeout: 258 seconds) 2011-12-14T10:26:04 there's always fluxid :P 2011-12-14T10:27:18 *** GeorgeSebastian has joined #aichallenge 2011-12-14T10:30:59 fluxid too slow :) 2011-12-14T10:31:58 even when i send 500ms turntime? 2011-12-14T10:32:48 Maybe it is the engines I am paired with that is making it slow. 2011-12-14T10:33:01 i can do 3x faster at tcpanst i think 2011-12-14T10:35:31 *** amstan has joined #aichallenge 2011-12-14T10:35:31 *** ChanServ sets mode: +o amstan 2011-12-14T10:36:41 *** Harpyon has joined #aichallenge 2011-12-14T10:39:34 Fluxid: how do you feel after being slaughtered by xathis? 2011-12-14T10:40:34 can't complain 2011-12-14T10:40:45 he's a god after all 2011-12-14T10:41:09 i mean, his bot is ant-god 2011-12-14T10:41:51 Fluxid: how do you feel after being slaughtered by xathis? 2011-12-14T10:41:55 wups 2011-12-14T10:41:56 sorry 2011-12-14T10:42:01 did you see the fight between him and flag, and some other guys? 2011-12-14T10:42:11 http://aichallenge.org/visualizer.php?game=227621&user=31 2011-12-14T10:42:22 flag was very superior imo 2011-12-14T10:47:12 that movie is amazing 2011-12-14T10:47:54 http://pastebin.com/NZExUe8c 2011-12-14T10:47:58 well i now handle that correctly 2011-12-14T10:48:31 top left stays still, top right moves down and bottom right goes left 2011-12-14T10:48:54 anyone have anything to add to that example? 2011-12-14T10:52:46 talk strategies with me people! 2011-12-14T10:54:30 why doesn't top left go down ? 2011-12-14T10:56:09 ok maybe because b can move west 2011-12-14T10:56:18 antimatroid: Are you trying to find a way to kill for sure A without giving him a chance to kill any your ant? 2011-12-14T10:56:45 UncleVasya: what's A? 2011-12-14T10:56:58 *a 2011-12-14T10:57:06 i want to kill b 2011-12-14T10:57:24 well, i want to get as close to b as possible without him being able to kill one of my ants 2011-12-14T10:57:24 yep, messed with letters. 2011-12-14T10:57:49 i've just added and extra set of moves into my game tree to handle it 2011-12-14T10:58:09 it finds "battle slack" and tries to minimise the positive battle slack 2011-12-14T10:58:29 Maybe you'll find more interesting to find a possibility for _b_ to not give away his life for free? 2011-12-14T10:59:10 wow, have I made a Shakespeare-like sentence now? 2011-12-14T10:59:53 I mean if there is a situation for _b_ where he can do 2 things: 2011-12-14T11:00:08 1. go for 1-to-1 exchange 2011-12-14T11:00:25 2. step back without exchange 2011-12-14T11:00:34 when picking out offensive moves for one of my ants, i for each move location, inside the attack radius i check for move locations of enemies and the maximum friends they could be battling, then minimise that and the battle slack is like min(possibleFriendEnemies) - possibleEnemies 2011-12-14T11:00:38 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-14T11:01:01 nah, see i already use a game tree for picking combat moves but the sample of moves i take is stupid in situations like that 2011-12-14T11:01:07 i'm hoping this extra set of moves might help 2011-12-14T11:01:08 And choise 2 is leading to position where there is no more hope for _b_ to go for 1-to-1. 2011-12-14T11:01:57 So on the very beginning _b_ must go into 1-to1 exchange, althought steping away is higher in score. 2011-12-14T11:05:14 *** sigh has joined #aichallenge 2011-12-14T11:07:41 *** denysonique__ has joined #aichallenge 2011-12-14T11:09:47 *** aerique has quit IRC (Quit: ...) 2011-12-14T11:09:50 *** UncleVasya has quit IRC (Read error: Connection reset by peer) 2011-12-14T11:10:49 *** UncleVasya has joined #aichallenge 2011-12-14T11:14:02 Hi UncleVasya 2011-12-14T11:14:14 hi 2011-12-14T11:14:46 *** Antimony has quit IRC (Ping timeout: 248 seconds) 2011-12-14T11:22:28 *** nha has quit IRC (Ping timeout: 240 seconds) 2011-12-14T11:25:42 *** foRei has joined #aichallenge 2011-12-14T11:25:50 *** GeorgeSebastian has quit IRC (Quit: popped out) 2011-12-14T11:26:12 *** dvladim has joined #aichallenge 2011-12-14T11:26:20 *** GeorgeSebastian has joined #aichallenge 2011-12-14T11:26:26 *** Antimony has joined #aichallenge 2011-12-14T11:31:16 *** Antimony has quit IRC (Client Quit) 2011-12-14T11:36:39 aichallenge: McLeopold epsilon * rf2d078e / website/ranking.php : fix for showing other category for main rankings page sometimes - http://git.io/7kNgew 2011-12-14T11:41:44 *** iglo has joined #aichallenge 2011-12-14T11:42:28 *** kire has joined #aichallenge 2011-12-14T11:42:57 Anilm3: i haven't touched it! 2011-12-14T11:45:10 that's great then! 2011-12-14T11:45:13 @rankings 2011-12-14T11:45:14 Anilm3: Top 10 players: xathis(95.6), FlagCapper(94.1), pguillory(93.0), fourmidable(91.1), a1k0n(90.9), delineate(88.9), GreenTea(88.8), ThisIsNotABug(88.6), protocolocon(88.1), jacob_strauss(88.0) 2011-12-14T11:45:17 haha 2011-12-14T11:45:18 http://aichallenge.org/visualizer.php?game=225024&user=13020 2011-12-14T11:45:23 that's really well done 2011-12-14T11:45:34 xathis is getting further 2011-12-14T11:46:26 what the hell??? 2011-12-14T11:47:33 that was a surprise haha 2011-12-14T11:47:42 *** tobym has quit IRC (Remote host closed the connection) 2011-12-14T11:48:16 dunno who chm is but somebody loves him 2011-12-14T11:49:28 it does that on every fight 2011-12-14T11:49:40 actually if it didn't do that stupid thing it would win every battle 2011-12-14T11:50:24 or it could still have a little style... it could roll in the text, and then immediately smash all enemy anthills 2011-12-14T11:50:36 that too 2011-12-14T11:51:08 I found him 2011-12-14T11:51:10 http://www.linkedin.com/groups/Chm-file-not-opening-on-2572009.S.69608429 2011-12-14T11:51:42 oh, compiled html sucks? that's the statement? 2011-12-14T11:52:10 I think chm is awesome, after you uncompile it to actual html 2011-12-14T11:52:11 I don't know, but he's Dave-O and he's talking about chm lol 2011-12-14T11:52:16 decompile 2011-12-14T11:52:25 *** sigh has quit IRC (Remote host closed the connection) 2011-12-14T11:54:36 a1k0n: don't give up on your bot! 2011-12-14T11:56:07 i don't know if i'll have any more time for it 2011-12-14T12:00:49 who am i kidding, i obviously have enough time to dink around on the aichallenge forums 2011-12-14T12:01:36 *** mj41 has quit IRC (Ping timeout: 244 seconds) 2011-12-14T12:05:54 *** GeorgeSebastian has quit IRC (Ping timeout: 255 seconds) 2011-12-14T12:06:09 *** SMJ has quit IRC (Quit: Konversation terminated!) 2011-12-14T12:06:19 *** SMJ has joined #aichallenge 2011-12-14T12:12:56 *** Blkt has quit IRC (Quit: ERC Version 5.3 (IRC client for Emacs)) 2011-12-14T12:13:51 *** LoneRanger has joined #aichallenge 2011-12-14T12:14:11 *** amstan has joined #aichallenge 2011-12-14T12:14:11 *** ChanServ sets mode: +o amstan 2011-12-14T12:17:14 *** TheLinker has quit IRC (Remote host closed the connection) 2011-12-14T12:17:37 *** GeorgeSebastian has joined #aichallenge 2011-12-14T12:19:00 *** foRei has quit IRC (Read error: Connection reset by peer) 2011-12-14T12:19:05 *** LoneRanger has quit IRC (Quit: Who was that masked man?) 2011-12-14T12:20:28 *** foRei has joined #aichallenge 2011-12-14T12:21:06 *** treeform has joined #aichallenge 2011-12-14T12:24:12 *** dr- has joined #aichallenge 2011-12-14T12:28:35 *** mikewintermute has quit IRC (Quit: mikewintermute) 2011-12-14T12:29:24 *** GeorgeSebastian has quit IRC (Quit: popped out) 2011-12-14T12:31:46 *** pguillory has joined #aichallenge 2011-12-14T12:34:47 http://paste.aichallenge.org/Osl9t/ 2011-12-14T12:36:49 Hi first McLeopold 2011-12-14T12:37:14 I don't know why but it freezes on Creating canvas... 2011-12-14T12:37:46 *** amstan__ has joined #aichallenge 2011-12-14T12:37:46 *** ChanServ sets mode: +o amstan__ 2011-12-14T12:37:57 Takes a while but it gets there 2011-12-14T12:38:11 yes now I see it 2011-12-14T12:38:58 there is *a bunch* of visualization stuff in that replay 2011-12-14T12:39:16 I was attempting to spread my ants out evenly 2011-12-14T12:39:34 it worked out better than I thought it would, but still bunches around at the hill 2011-12-14T12:40:15 *** praveen_ has joined #aichallenge 2011-12-14T12:40:55 *** dici has joined #aichallenge 2011-12-14T12:45:58 *** amstan__ has quit IRC (Ping timeout: 248 seconds) 2011-12-14T12:50:41 But what does it all mean! 2011-12-14T12:51:07 @rainbow it's so beautiful 2011-12-14T12:51:07 a1k0n: it's so beautiful 2011-12-14T12:52:27 McLeopold1: this isn't crashing my browser, right? 2011-12-14T12:52:30 I think seeing all the secret viz stuff once the contest finishes is going to be pretty fun. 2011-12-14T12:52:53 i don't have any secret viz stuff :( 2011-12-14T12:53:30 jcdny: those visuals have to be built into the bot, you can't just switch on some flag in your bot and see it (unless you implemented it, but i repeat myself) 2011-12-14T12:54:08 yeah but I imagine people will talk about what they have actually done and put it up on paste 2011-12-14T13:00:40 *** SMJ has quit IRC (Quit: Konversation terminated!) 2011-12-14T13:00:50 *** SMJ has joined #aichallenge 2011-12-14T13:09:03 *** flowenol has joined #aichallenge 2011-12-14T13:09:10 aichallenge: janzert epsilon * r0c454cb / manager/matchup_buffer.py : Expand matchup buffer logging and size - http://git.io/SKtukA 2011-12-14T13:09:10 aichallenge: janzert epsilon * rd123307 / website/ranking.php : Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/f8queQ 2011-12-14T13:09:18 *** ikaros_ has quit IRC (Quit: Ex-Chat) 2011-12-14T13:12:19 *** kire has quit IRC (Quit: Leaving) 2011-12-14T13:16:01 BenJackson: no, it is 14MB of visualization data 2011-12-14T13:16:26 kudos to django for handling my large pastes without failing 2011-12-14T13:16:59 *** JorgeB has joined #aichallenge 2011-12-14T13:17:20 McLeopold1: ah, i see you dispatch ants to hill 2011-12-14T13:17:28 i dispatch ants only for food 2011-12-14T13:18:01 I do, but on a graduated scale 2011-12-14T13:18:23 and they will be distracted by food 2011-12-14T13:21:46 *** liberforce has left #aichallenge 2011-12-14T13:30:22 *** Hexren has joined #aichallenge 2011-12-14T13:30:27 *** Hexren has left #aichallenge 2011-12-14T13:32:51 *** McLeopold1 has left #aichallenge 2011-12-14T13:37:26 *** Gnome has joined #aichallenge 2011-12-14T13:41:51 *** dvladim has quit IRC (Ping timeout: 252 seconds) 2011-12-14T13:42:34 *** grwip has joined #aichallenge 2011-12-14T13:46:16 *** xathis has joined #aichallenge 2011-12-14T13:47:02 *** besh has quit IRC (Quit: Page closed) 2011-12-14T13:53:31 *** Dlayne has joined #aichallenge 2011-12-14T13:58:34 *** TheLinker has joined #aichallenge 2011-12-14T14:10:46 Dlayne: no fun here right now 2011-12-14T14:11:43 thanx for info 2011-12-14T14:12:06 http://aichallenge.org/profile.php?user=2255 nice, made it up to a new high :) 2011-12-14T14:12:18 i'm falling asleep anyway 2011-12-14T14:12:18 I guess today will be my last reupload 2011-12-14T14:13:21 *** Redgis has joined #aichallenge 2011-12-14T14:14:15 thestinger: I'm still frustrated with obvious bad behaviors with no obvious fix 2011-12-14T14:14:27 I have lots of those :) 2011-12-14T14:14:43 and I never really merged in my improved combat... it still just does 2 sets of moves for the enemy 2011-12-14T14:14:45 this version really had very little tuning 2011-12-14T14:14:57 generating a bunch of moves just makes my bot too cautious 2011-12-14T14:15:02 it wasn't really getting meaningful games on fluxid where I lost and could see why 2011-12-14T14:15:08 although I think fluxid is just losing games 2011-12-14T14:15:12 so I just duct tapped together the memetix combat, and the payoff matrix style 2011-12-14T14:15:15 that's what my bot does atm 2011-12-14T14:15:17 because I see games go by on my console that never appear 2011-12-14T14:15:25 like I'll end up with 1 ant being chased around 2011-12-14T14:15:31 yeah 2011-12-14T14:15:33 but when I check fluxid I've won 5 in a row 2011-12-14T14:15:33 tcp has weird bugs 2011-12-14T14:15:36 for the hour around that 2011-12-14T14:15:43 I play a lot of games 2011-12-14T14:15:48 but it seems like half of them don't show up 2011-12-14T14:15:58 fluxid: where are our gaaaames 2011-12-14T14:16:00 like, I'll get some big win - and it never shows up 2011-12-14T14:16:11 yeah, it's harder for me to spot those 2011-12-14T14:16:13 or some interesting loss where I hold off for a long time and then die - nope, vanishes 2011-12-14T14:16:20 easier to see when it goes by for several minutes moving only 1 ant 2011-12-14T14:16:34 * Fluxid hears screams 2011-12-14T14:16:37 what where why 2011-12-14T14:16:37 lol 2011-12-14T14:16:45 BenJackson: ? 2011-12-14T14:16:48 we think the tcp server eats our games :P 2011-12-14T14:16:52 wha 2011-12-14T14:16:52 Fluxid: see recent sb 2011-12-14T14:16:53 anyway, too late to worry about that 2011-12-14T14:17:19 BenJackson: lemme check something 2011-12-14T14:17:21 Fluxid: 1) game rate "appears" to be way down on fluxid 2011-12-14T14:17:30 2) thestinger and I have both noticed games played that never show up 2011-12-14T14:17:53 BenJackson: try to load your profile 2011-12-14T14:18:15 IndexError: list index out of range 2011-12-14T14:18:17 ok, fuck 2011-12-14T14:18:22 hey, how about i drop db? 2011-12-14T14:18:34 drop table? 2011-12-14T14:18:38 whole db 2011-12-14T14:18:42 what db? 2011-12-14T14:18:45 sqlite 2011-12-14T14:18:46 it's fucked up 2011-12-14T14:18:47 in sqlite that's a file 2011-12-14T14:18:59 I'm pretty sure db<->file in sqlite 2011-12-14T14:19:02 by dropping db i meant removing the file, right 2011-12-14T14:19:12 because it's screwed up 2011-12-14T14:19:19 you can sqlite into it 2011-12-14T14:19:23 it lost consistency somehow 2011-12-14T14:19:40 and do a transaction to empty the tables 2011-12-14T14:19:43 *** Paradoxial has joined #aichallenge 2011-12-14T14:19:43 if you're worried about restarting 2011-12-14T14:19:52 sqlite does concurrency right 2011-12-14T14:19:58 no, I don't think so 2011-12-14T14:20:03 only one process can open the db 2011-12-14T14:20:13 and it's just a library the program uses, not a server type thing 2011-12-14T14:20:15 you can definitely have multiples 2011-12-14T14:20:21 you can have multiple readers 2011-12-14T14:20:28 multiple writers 2011-12-14T14:20:38 < thestinger> only one process can open the db 2011-12-14T14:20:40 well... 2011-12-14T14:20:40 with transactions 2011-12-14T14:20:46 oh 2011-12-14T14:20:54 http://www.sqlite.org/lockingv3.html 2011-12-14T14:21:07 ah, so it does have locking 2011-12-14T14:21:09 well that page is just "we improved over v2" but close enough 2011-12-14T14:21:20 we have two processes, where one has shitloads of threads 2011-12-14T14:21:23 what's more you can connect to multiple disjoint sets of dbs 2011-12-14T14:21:32 buuuuut 2011-12-14T14:21:39 sqlite live.db test.db and then do soem transaction to move data from live to test or whatever 2011-12-14T14:21:44 i have no idea what should i do to regain consistency 2011-12-14T14:22:08 Fluxid: this is the first time (when you prompted) that I couldn't load my profile 2011-12-14T14:22:11 because as i can see almost every profile i check is broken 2011-12-14T14:22:13 so I'd say it's getting worse 2011-12-14T14:22:21 yeah 2011-12-14T14:22:57 so you could just sqlite to the db and in one transaction "drop from table" all the tables 2011-12-14T14:23:04 err delte from 2011-12-14T14:24:25 @seen Smiley1983 2011-12-14T14:24:25 UncleVasya: Smiley1983 was last seen in #aichallenge 5 weeks, 3 days, 4 hours, 29 minutes, and 10 seconds ago: zorun: if I enter the commands into the ocaml toplevel to input_line from a fifo pipe (like in the example in my paste) then it can read from the pipe; the same functions seem to fail under ocamlrun, so I don't know what's going on, but thankfully the compiled versions all (1 more message) 2011-12-14T14:24:37 @seen solifugid 2011-12-14T14:24:37 UncleVasya: I have not seen solifugid. 2011-12-14T14:24:51 it's disturbing to see handles like "foo1983" 2011-12-14T14:24:58 because at first my mind makes that a random number 2011-12-14T14:25:05 then it realizes it's probably a birth year 2011-12-14T14:25:07 then I'm horrified 2011-12-14T14:25:16 esp when it's like "sexy1992" or something 2011-12-14T14:26:15 BenJackson: Horried by 28 year olds? 2011-12-14T14:26:34 horrified even 2011-12-14T14:27:10 well it's the 1992+ that get me 2011-12-14T14:27:21 I was born in 1992 :3 2011-12-14T14:27:22 since those could basically be my kids (I don't have kids) 2011-12-14T14:27:23 awww yeah 2011-12-14T14:28:11 fffff 2011-12-14T14:28:54 thestinger: then I have been using IRC longer than you've been alive 2011-12-14T14:29:05 (certainly not continuously, but first in 1989 iirc) 2011-12-14T14:29:14 So you nee find some sexy1992 to make kids :D 2011-12-14T14:29:20 hahaha 2011-12-14T14:29:46 new dating service is just 'select email from yahoo where username like '%1992' 2011-12-14T14:30:27 okay 2011-12-14T14:30:37 BenJackson: thestinger: maybe now 2011-12-14T14:30:44 i mean 2011-12-14T14:30:49 I really want a function in social networks like facebook 2011-12-14T14:30:50 i dropped games, but not player 2011-12-14T14:31:01 ah, nice 2011-12-14T14:31:05 looks like my client is still running 2011-12-14T14:31:08 to put music tegs in the search form 2011-12-14T14:31:11 I don't know if you can tell 2011-12-14T14:31:36 (I'm at work) 2011-12-14T14:32:20 mined didn't get disconnected so you're probably right 2011-12-14T14:32:21 mine* 2011-12-14T14:33:16 @seen mcstar 2011-12-14T14:33:16 Anilm3: mcstar was last seen in #aichallenge 2 weeks, 2 days, 20 hours, 27 minutes, and 20 seconds ago: i dont agree 2011-12-14T14:33:59 he's going to upload some super secret bot on the last day imo :P 2011-12-14T14:34:43 Anilm3: but he's still alive - he reuploads often. 2011-12-14T14:34:50 *** Conorach has quit IRC (Read error: Connection reset by peer) 2011-12-14T14:41:29 Fluxid: thanks! 2011-12-14T14:42:09 I wonder if the "184 games" and my rank are based on all games (including missing) or if the lost replays are also not in stats 2011-12-14T14:49:03 nah, there's column "ngames" in player table 2011-12-14T14:49:20 it just increments by one every finished game 2011-12-14T14:49:50 update players set ngames=ngames+1,lastseen=?,skill=?,mu=?,sigma=? where name=? 2011-12-14T14:50:50 lol, hopefully I fixed _newdefend 2011-12-14T14:51:18 just added a 'std::ceil()' to the thing that gets the ratio of defenders : attackers 2011-12-14T14:55:50 *** besh has joined #aichallenge 2011-12-14T14:58:08 *** UncleVasya has quit IRC (Read error: Connection reset by peer) 2011-12-14T15:00:38 oh, the many ways, you can be off by one.. 2011-12-14T15:02:44 BenJackson: http://ants.fluxid.pl/replay.33084 I regret doing ctrl-C on this game... 2011-12-14T15:02:53 *** Antimony has joined #aichallenge 2011-12-14T15:03:27 maybe I could have held it at 10 - 9 2011-12-14T15:07:19 nice, I'm on the front page getting murdered my greentea - right after he murdered me on tcp :) 2011-12-14T15:08:03 thestinger: hard to tell how that was going to come out otherwise 2011-12-14T15:09:17 next aichallenge game in 6 minutes 2011-12-14T15:09:20 let's hope I don't f this one up 2011-12-14T15:10:09 *** Paradoxial has quit IRC (Read error: Connection reset by peer) 2011-12-14T15:10:11 *** Harpyon has quit IRC (Quit: Textual IRC Client: http://www.textualapp.com/) 2011-12-14T15:10:26 *** Paradoxial has joined #aichallenge 2011-12-14T15:11:45 well, it was somewhat entertaining to figure out how to beat antivirus - but the solution was pretty sad 2011-12-14T15:12:00 they have all kinds of fancy unpackers to figure out how you've encoded your evil code and unpack/scan it 2011-12-14T15:12:20 the trick was to use argv[1] to decode it 2011-12-14T15:12:47 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-12-14T15:14:58 *** Accoun has quit IRC () 2011-12-14T15:20:04 is anyone going to be running their TCP server after the competition is over? 2011-12-14T15:21:29 oh good grief 2011-12-14T15:21:38 I guess I just found a new way to lose a 3 player game 2011-12-14T15:21:48 Fluxid and sivsoft killed each other 2011-12-14T15:21:56 thus 2-2-1 rank stabilized 2011-12-14T15:22:04 because being dead they weren't counted for gaining more score 2011-12-14T15:24:00 uh 2011-12-14T15:24:12 strange game 2011-12-14T15:24:22 *** Antimony has quit IRC (Ping timeout: 248 seconds) 2011-12-14T15:24:27 it's strange that it got stabilized but there were no way for you to eran points... 2011-12-14T15:24:43 it's true, there's no way for me to win and I had the only hill 2011-12-14T15:24:50 one of you could have killed ME and won outright 2011-12-14T15:24:58 unlikely though 2011-12-14T15:25:39 ;) 2011-12-14T15:26:06 you'd kill us all and wander around seeking hills ;) 2011-12-14T15:26:28 *** kire has joined #aichallenge 2011-12-14T15:26:58 I think I actually found him first 2011-12-14T15:27:04 which drew him away for you to kill him 2011-12-14T15:27:22 I can't tell why you didn't successfully defend against him 2011-12-14T15:27:26 if you HAD then I could have won :) 2011-12-14T15:29:15 now I'm trying to decide if that's a viable strategy 2011-12-14T15:29:43 *** praveen_ has quit IRC (Quit: Page closed) 2011-12-14T15:29:54 *** LoneRanger has joined #aichallenge 2011-12-14T15:29:59 in a 3 player game if you take player "B" and player B is attacking you some time later and you let B take your hill 2011-12-14T15:30:11 you might have a better chance of a tie for 1st 2011-12-14T15:30:14 *** Accoun has joined #aichallenge 2011-12-14T15:30:24 because the guy you didn't take could still beat you at that point 2011-12-14T15:31:31 I'm not sure if truskill cares how many tie for 1st 2011-12-14T15:31:49 BenJackson: shepherding enemy ants so that the right one takes your hill sounds pretty awesome ;) 2011-12-14T15:32:10 not helping them, necessarily, just not stopping them 2011-12-14T15:33:22 I guess not knowing if he took C's hill is an issue 2011-12-14T15:33:29 but if you can see C's hill you can ensure a tie for 1st by giving up yours 2011-12-14T15:33:30 i didn't have enough ants to defend. few ants got drawn away from hill to left but then more his ants got from north... 2011-12-14T15:34:33 until now I thought 3 player maps were actually "fair" because I had not expected two to trade 2011-12-14T15:34:53 I figured even if the 2nd strongest player got a fast start he still comes in second after it's down to 1v1 2011-12-14T15:36:22 BenJackson: I've noticed that before 2011-12-14T15:36:32 it really makes a lot of sense to rush on lots of maps 2011-12-14T15:37:00 because you don't win by building up a huge force and surviving until the end - you win by grabbing a few hills early on :) 2011-12-14T15:37:54 oh, another game 2011-12-14T15:38:05 third.. 2011-12-14T15:38:23 earned less than 0.01 of skill, lol 2011-12-14T15:38:40 +0.0017 2011-12-14T15:39:34 *** SMJ has quit IRC (Quit: Konversation terminated!) 2011-12-14T15:41:19 *** choas has joined #aichallenge 2011-12-14T15:44:00 Fluxid: nice 2011-12-14T15:44:15 my bot just got up to 13th :) 2011-12-14T15:47:20 good :) 2011-12-14T15:48:12 nvm, magically shifted into 12th 2011-12-14T16:00:56 thestinger: the strange thing about that is that xathis is all about territory control and winning the long game 2011-12-14T16:04:44 *** Kurnevsky has left #aichallenge 2011-12-14T16:04:52 yeah, so my only hope is to get lucky with rushing :) 2011-12-14T16:05:17 if the RNG is on my side, I'll win 2011-12-14T16:07:41 *** ltriant has joined #aichallenge 2011-12-14T16:09:39 early on I saw a bot that moves 2x2 blocks of ants around 2011-12-14T16:09:44 which was pretty effective pre-combat 2011-12-14T16:09:54 but no one now seems to intentionally launch sorties like that 2011-12-14T16:10:06 I tried making my ants look for food in pair 2011-12-14T16:10:09 pairs* 2011-12-14T16:10:27 that worked, but now it just wastes too many ants and gets destroyed anyway 2011-12-14T16:11:17 *** Murashka has joined #aichallenge 2011-12-14T16:11:23 I think my bot's problem now is that I built a strategy and then added combat so it could execute the strategy without getting killed 2011-12-14T16:11:32 yeah 2011-12-14T16:11:36 then I realized that with everyone having combat the strategy was wrong 2011-12-14T16:11:41 my bot doesn't try to kill enemy enemies 2011-12-14T16:11:42 so I've changed my strategy 2011-12-14T16:11:51 but I think now my combat is "wrong" for that 2011-12-14T16:12:05 I would need to rewrite everything to fix it, but it's not going to happen 2011-12-14T16:12:10 yeah 2011-12-14T16:12:29 I tried just adding on hunting of enemies, but it sucked 2011-12-14T16:12:43 my combat relies on the fact that my ants are willing to pathfind around enemies to go to another goal 2011-12-14T16:13:33 *** Nihil688 has joined #aichallenge 2011-12-14T16:13:39 my V11 is basically hunting enemies 2011-12-14T16:13:41 BenJackson: Did the 2x2 blocks work? I was thinking of that earlier when brainstorming. 2011-12-14T16:13:53 ronchaine: works great against "blind" bots 2011-12-14T16:13:58 ronchaine: I think it would hurt more than it would help now 2011-12-14T16:14:06 against combat bots they'll just kill your block of 4 (assuming your block is blind) 2011-12-14T16:14:06 hey is there any way to save the replays? 2011-12-14T16:14:13 it's just a nice target with 4 vulnerable corners to a bot with combat 2011-12-14T16:14:32 Nihil688: the support files are in the "tools" tree 2011-12-14T16:14:41 I was thinking that it would be kinda useful since they could reorganize themselves to 1x4 or 4x1 wall in one move. 2011-12-14T16:14:41 using that you can watch replay.html anywhere 2011-12-14T16:15:02 *** yoden has quit IRC (Read error: Connection reset by peer) 2011-12-14T16:15:33 ronchaine: it might be a strategy that gives you a boost in the top 100, but surely not in the top 20 or top 10 2011-12-14T16:15:41 I have the tools 2011-12-14T16:15:54 and I have seen the section share maps and replays 2011-12-14T16:16:20 a replay is a big json string 2011-12-14T16:16:29 it can either be embedded in replay.html or loaded from another url 2011-12-14T16:16:54 depends on where you viewed the replay 2011-12-14T16:16:56 visualizer.loadReplayData 2011-12-14T16:17:11 Nihil688 game_logs\0.replay 2011-12-14T16:17:22 visualizer.visualize_locally.launch(sys.argv[1], True, 'replay.html') 2011-12-14T16:17:23 that code is taken from that 2011-12-14T16:17:35 that line will make replay.html embedding sys.argv[1] 2011-12-14T16:18:14 Nihil688: sorry to be vague, there are just many sources of replays 2011-12-14T16:18:18 not sure which you want to save 2011-12-14T16:18:21 or view 2011-12-14T16:19:08 *** yoden has joined #aichallenge 2011-12-14T16:19:18 when I watch a replay and click on the "bot in" button 2011-12-14T16:19:29 I suppose this is where I can find the replay data 2011-12-14T16:19:36 that's different data 2011-12-14T16:19:46 no it's stream data from server to bot 2011-12-14T16:19:52 the replay is a json string with the full map, all the ants and what they did in the entire game 2011-12-14T16:20:09 *** Palmik has quit IRC (Remote host closed the connection) 2011-12-14T16:20:14 that "bot in" button is the list of lines like 'a 100 10 0' that would have been sent to one of the contestants during the game 2011-12-14T16:20:21 which you can use to re-run your bot 2011-12-14T16:20:28 for debug or performance or whatever 2011-12-14T16:20:49 http://ants.fluxid.pl/replay.33215 nice, I think I finally fixed it 2011-12-14T16:21:01 stupid off-by-one errors.... 2011-12-14T16:21:03 right, so where can I find that json string to feed it to my replay.html 2011-12-14T16:21:32 I want to view a game I played on the server, locally 2011-12-14T16:21:45 thestinger: I suggest you come up with some function of proximity to your hills and increase your willingness to kill/sacrifice as they get closer 2011-12-14T16:21:57 I thikn you should have killed some of those nearby ants no matter waht the loss 2011-12-14T16:22:05 instead of just holding them a few squares away 2011-12-14T16:22:17 Nihil688: the aichallenge server? 2011-12-14T16:22:37 let's say this: http://aichallenge.org/visualizer.php?game=233683&user=11095 2011-12-14T16:22:38 it's http://aichallenge.org/game/NNNNNN 2011-12-14T16:22:41 BenJackson: yeah, I have a score branch where I do all kinds of crazy stuff with the score 2011-12-14T16:22:47 it's http://aichallenge.org/game/233683 2011-12-14T16:22:49 I want to get this locally 2011-12-14T16:22:51 _newdefend just assigns defenders to enemies near my hill 2011-12-14T16:22:58 with ones closer to the hill getting priority 2011-12-14T16:23:06 oh, yeah 2011-12-14T16:23:11 that's what V11 of mine does 2011-12-14T16:23:19 interesting, so I just use the game number 2011-12-14T16:23:35 thank's for that 2011-12-14T16:23:44 I do it based on a ratio of defenders : attackers, and I was using int division 2011-12-14T16:23:53 so if there were more enemies, it would assign 0 to each :P 2011-12-14T16:23:59 ah 2011-12-14T16:24:13 I think what I do is assign 2 to each until I run out of ants to assign 2011-12-14T16:24:35 I find the ratio, the set it back to 3 if it's higher 2011-12-14T16:24:43 and then assign that many to each (starting with ones closest to the hill) 2011-12-14T16:25:22 what time are the submissions on Sunday? 2011-12-14T16:25:24 so ceil((double)defenders / (double)attackers) is a lot different than the buggy way I did it before 2011-12-14T16:26:13 Nihil688 topic 2011-12-14T16:26:32 I've slowly changed from gradient map stuff to hardcoded strategies like that 2011-12-14T16:26:49 just couldn't get the right behavior to emerge 2011-12-14T16:27:03 topic? 2011-12-14T16:27:47 Nihil688 yes. Submission deadline: December 18 23:59 EST 2011-12-14T16:28:14 *** SMJ has joined #aichallenge 2011-12-14T16:28:18 ah it was up there 2011-12-14T16:28:19 :$ 2011-12-14T16:28:49 right I need to fix my attackers' strategy and I'm good to go :P 2011-12-14T16:29:25 getting higher than #1500 will be tough :) 2011-12-14T16:30:09 or maybe it won't since I don't have any attackers atm :P 2011-12-14T16:32:37 is there any particular base algorithm everyone uses? 2011-12-14T16:34:32 minimax 2011-12-14T16:34:50 Everyone do not use minmax 2011-12-14T16:34:56 Not even close 2011-12-14T16:35:28 and it's variations 2011-12-14T16:35:29 *** iglo has quit IRC (Ping timeout: 252 seconds) 2011-12-14T16:36:44 really? 2011-12-14T16:36:53 There are a couple good threads in the strategy section on a couple of approaches that are being used. 2011-12-14T16:36:58 just because its turn based? :P 2011-12-14T16:37:12 I don't personally see how min-max applies 2011-12-14T16:37:14 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-12-14T16:37:32 maybe to assess the situation 2011-12-14T16:37:36 my best move if I know your move is different from my best move if I *don't* know your move 2011-12-14T16:37:40 *** g0llum has joined #aichallenge 2011-12-14T16:37:48 I actually didn't find much literature on simultaneous move games 2011-12-14T16:37:57 (like rock-paper-scissors) 2011-12-14T16:38:03 because humans aren't really built to play them 2011-12-14T16:38:07 rock-dynamite-scissors 2011-12-14T16:38:14 minimax = 5 ^ numants, not uesful, until you reduce that somehow 2011-12-14T16:39:20 *** McLeopold has joined #aichallenge 2011-12-14T16:39:51 yep it needs a lot of heuristics 2011-12-14T16:40:15 "A simple approach to to combat" and "Implementing combat with random sampling" are both good reads if you are looking for places to start. 2011-12-14T16:40:34 and it's not gives exact decision 2011-12-14T16:40:34 *** yoden1 has joined #aichallenge 2011-12-14T16:40:36 I just use masks 2011-12-14T16:42:11 they even took care, that a combat won't fit into a 64bitboard 2011-12-14T16:43:09 so I take people are quite high in the ranks here? :) 2011-12-14T16:43:21 sometimes ;) 2011-12-14T16:43:34 I don't have much time since I work all day :( 2011-12-14T16:43:43 *** yoden has quit IRC (Ping timeout: 252 seconds) 2011-12-14T16:43:46 and the game industry makes you work a lot of hours ;) 2011-12-14T16:43:47 it helps if you shirk your responsibilities 2011-12-14T16:43:53 some of the top 10 visit the channel 2011-12-14T16:44:26 pairofdice: and some of us are on the channel a lot and are sometimes in the top 10 :) 2011-12-14T16:44:34 #27 for now 2011-12-14T16:44:46 nice! 2011-12-14T16:44:51 *** Vaenom has quit IRC (Quit: Page closed) 2011-12-14T16:45:03 my bot is really simple, no fancy algorithms or code 2011-12-14T16:45:06 I was about 12th before my last resub 2011-12-14T16:45:14 *** LouisMartin has joined #aichallenge 2011-12-14T16:45:18 and I hope rise up :) 2011-12-14T16:45:27 I just did semi-automated testing using the score info you get when a game is over 2011-12-14T16:45:35 and figured out which ideas were good/bad :) 2011-12-14T16:45:53 I've considered extracting a limited set of parameters and tuning that way 2011-12-14T16:46:15 my code is a mess at the moment 2011-12-14T16:46:23 it works fast but design wise... beh 2011-12-14T16:46:23 :P 2011-12-14T16:46:28 I'm not sure if my code is beautiful or ugly 2011-12-14T16:46:31 but it works, mostly 2011-12-14T16:47:03 BenJackson: I'm checking that game you posted. We should have let that continue. 2011-12-14T16:47:08 there are some ugly parts where I've been too lazy to write proper higher-order functions 2011-12-14T16:47:27 so I have a few implementations of some stuff 2011-12-14T16:47:33 meh it just takes time 2011-12-14T16:47:38 Yep. And this is a problem. You invent a new method. Realize it .... and it's getting worse. 2011-12-14T16:47:52 McLeopold: no, both my enemies were "dead" 2011-12-14T16:47:58 and there were no hills for me 2011-12-14T16:48:02 to take 2011-12-14T16:48:14 BenJackson: but one of the enemies could have killed _you_ 2011-12-14T16:48:24 yes but players without hills don't count for that 2011-12-14T16:48:29 they should :P 2011-12-14T16:48:38 kamikaze ants 2011-12-14T16:48:41 I'm just explaining why it's not a bug as the spec is written 2011-12-14T16:48:43 BenJackson: at one point, we included the scoring leader, but I need to check the history 2011-12-14T16:49:01 right i ll go back and work a bit at it 2011-12-14T16:49:08 thanks everyone 2011-12-14T16:49:12 my real disappointment with that game is that I thought 3-way games would favor the best player 2011-12-14T16:49:20 I haven't really read the game spec xD 2011-12-14T16:49:25 but once there's a trade it doesn't necessarily 2011-12-14T16:49:35 just the map generator part, and the battle algorithm 2011-12-14T16:49:38 *** Nihil688 has quit IRC (Quit: Page closed) 2011-12-14T16:49:52 BenJackson: maybe a special bonus for lone survivor is in order 2011-12-14T16:49:56 I read the symmetry code very carefully when I was working on that 2011-12-14T16:50:58 I'm curious about how you dealt with that without using up too much time 2011-12-14T16:51:08 McLeopold: if it was still in the design phase I'd suggest that trading somehow cancel out 2011-12-14T16:51:35 since two ants each with their own hills is 1-1, but if they trade it's 2-2, but did they really do anything worthwhile? 2011-12-14T16:51:47 well, maybe losing your hill should have been -2 2011-12-14T16:52:00 that would change multi-hill mazes dramatically 2011-12-14T16:52:00 it would be a different game though 2011-12-14T16:52:13 it would make them favor the long-term guy 2011-12-14T16:52:30 yeah, xathis 2011-12-14T16:52:35 it would make my bot useless 2011-12-14T16:52:39 *** LouisMartin has quit IRC (Quit: Page closed) 2011-12-14T16:53:40 it would be neat if capturing a hill gave you control if it, and the ranking was determined by who survived the longest 2011-12-14T16:53:46 all our bots useless except xathis :) 2011-12-14T16:53:52 BenJackson: I'm not trying to make it so the ant with the hill left wins, I need to make him lose more :) 2011-12-14T16:54:03 I think I will announce my symmetry idea on Sunday morning to cause maximum chaos :) 2011-12-14T16:54:05 and you could come "back to life" by recapturing a hill 2011-12-14T16:54:18 BenJackson: is it really better than dmj111? 2011-12-14T16:54:21 BenJackson: lol, as everyone tries to implement it and reuploads buggy versions that crash? xD 2011-12-14T16:54:44 McLeopold: well.. it's not the same 2011-12-14T16:54:59 it gives you only a subset of what his does 2011-12-14T16:55:02 are you detecting food spawning in 2 places on the same turn? :) 2011-12-14T16:55:04 but it's the subset you care about 2011-12-14T16:55:07 and is practially free 2011-12-14T16:55:10 McLeopold: yes! 2011-12-14T16:55:19 detecting food spawns seems very useful 2011-12-14T16:55:27 that was my method, but I haven't finished it yet 2011-12-14T16:55:54 with that and like 10 lines of code you can ID potential enemy hills *if* the map symmetry is north 2011-12-14T16:56:00 BenJackson: so McLeopold figured out your sneaky trick? xD 2011-12-14T16:56:04 what I didn't bother with was finding the relative map symmetries around each food 2011-12-14T16:56:13 but that's pretty easy 2011-12-14T16:56:28 the problem was I'd be able to find hills on the other side of the map in like 20 turns 2011-12-14T16:56:53 McLeopold: I actually think that's a bug in the game engine 2011-12-14T16:56:57 it "leaks" symmetry information 2011-12-14T16:57:27 by design 2011-12-14T16:57:38 anyway guys like dmj111 could use that to eliminate like 90% of their cases the first time they saw 2 food 2011-12-14T16:58:03 because 2 simultaneous food are almost always two congruent points on the map 2011-12-14T16:58:41 so if you saw the food you could look at your whole list of symmetries and reject any that don't admit your new food 2011-12-14T16:58:56 (with the caveat that there are ways for food to appear other ways, but rarely) 2011-12-14T16:58:58 they are congruent 2011-12-14T16:59:12 McLeopold: not if one is delayed 2011-12-14T16:59:37 hmm, and if we spawn on multiple congruent spots, I guess 2011-12-14T16:59:49 but for my method a "rogue" food would just fail to have any local rotational symmetry with any other and I'd ignore it 2011-12-14T17:00:03 for the "eliminate matrices" version you'd have to know if you could truly eliminate them 2011-12-14T17:00:25 anyway it's very quick to implement as long as you skip the rotation part 2011-12-14T17:00:32 which at the time I worked on it worked for >90% of the maps 2011-12-14T17:00:37 so I just tested with them 2011-12-14T17:01:37 so just random walk? 2011-12-14T17:02:39 if you run ants.py as a program it spits out the symmetry info 2011-12-14T17:02:41 you probably know that 2011-12-14T17:02:54 so I ran it on all the maps and only tested on the aim=0 ones 2011-12-14T17:03:01 *** tobym has joined #aichallenge 2011-12-14T17:03:11 I've seen a fwe games where it would help 2011-12-14T17:03:25 late in the game where you have the dominant ants but you can't break through to know where their hill is 2011-12-14T17:03:42 So I think it's the best idea to try and read the map file 2011-12-14T17:04:33 exec(ps -ef | grep "--map") 2011-12-14T17:05:01 that is speicifically grounds for disqualification in the rules 2011-12-14T17:05:02 except you are in a special unionfs chrooted env, so no :P 2011-12-14T17:05:10 booo! 2011-12-14T17:05:41 McLeopold: how many maps will there be in the finals? 2011-12-14T17:05:48 dont' know 2011-12-14T17:05:49 it seems odd to me to have more maps than each bot will play games 2011-12-14T17:05:54 (which is where the contest site is now) 2011-12-14T17:06:13 maybe one map per game? 2011-12-14T17:06:25 generate a new map for every matchup? 2011-12-14T17:06:28 *** keith_ has joined #aichallenge 2011-12-14T17:06:29 ow! 2011-12-14T17:06:33 that's interesting 2011-12-14T17:06:43 I like the idea of new maps for finals 2011-12-14T17:06:55 but I *also* like the idea of picking the known-good maps for finals 2011-12-14T17:06:57 I don't like how some of the maps are more unfair based on which opponents are next to you 2011-12-14T17:07:02 the ones with the best ant-distinguishing power 2011-12-14T17:07:17 like one of the original mazes had 2 players next to each other, and then another 2 across the map 2011-12-14T17:07:26 that's still in the rotation 2011-12-14T17:07:31 anyone else having problems seeing their games on fluxid? I can see the global list of all recent games, including mine -- but when i go to my player page, it is empty 2011-12-14T17:07:38 people are talking about the game where xathis lost 04p_02 2011-12-14T17:07:54 keith_: he tried to reset it earlier when the db got corrupted 2011-12-14T17:08:09 kk, thanks ben 2011-12-14T17:08:13 BenJackson: lol, I think I know where this started 2011-12-14T17:08:30 around a week or two ago he manually fixed a broken thing 2011-12-14T17:08:35 fluxid is still missing a ton of games if you just look at the replay #s on the front page 2011-12-14T17:08:45 the page for 'strcat' wouldn't load due to a broken game 2011-12-14T17:08:51 so he deleted it 2011-12-14T17:09:02 oh, not a ton, maybe just a few 2011-12-14T17:09:17 you're correct, it is missing lots o' games. bummer 2011-12-14T17:09:18 I think the server is just buggy 2011-12-14T17:09:19 BenJackson: I only see a couple missing 2011-12-14T17:09:28 past 30 min ago 2011-12-14T17:09:33 anything else coul,d still be running 2011-12-14T17:09:51 how low does my sigma need to go before the matchmaker starts pulling me in to non-seeded games? 2011-12-14T17:11:14 BenJackson: https://github.com/aichallenge/aichallenge/blob/epsilon/ants/mapgen/cell_maze.py#L759 2011-12-14T17:13:52 basically the same loop is what I used 2011-12-14T17:14:40 it'd be fun to repeat ants with teams 2011-12-14T17:14:45 everyone pair off 2011-12-14T17:14:46 for multi-hill cell mazes, I use that to know what player should be in what position 2011-12-14T17:15:32 at first I thought I was rereading the food symmetries code in ants.py 2011-12-14T17:15:36 I assume a similar author :) 2011-12-14T17:16:38 yep :) 2011-12-14T17:16:51 probably a copy paste :) 2011-12-14T17:18:27 *** keith_ has quit IRC (Ping timeout: 258 seconds) 2011-12-14T17:18:34 *** Harpyon has joined #aichallenge 2011-12-14T17:18:35 hmmm can I fix my combat before the bruins start playing? 2011-12-14T17:19:26 McLeopold: so i wanna put my mapgen to use 2011-12-14T17:19:40 McLeopold: it doesn't produce perfect maps, but i could check them manually for correctness 2011-12-14T17:20:00 amstan: ok 2011-12-14T17:20:13 You don't need to inherit from map.py anymore 2011-12-14T17:20:24 I just run the output of any program through it to check 2011-12-14T17:20:28 hmm? 2011-12-14T17:20:32 if it fails, I don't create a file 2011-12-14T17:20:39 i don't use map.py 2011-12-14T17:20:47 I thought you did? 2011-12-14T17:20:51 are a lot of people just not submitting current bots? I am rank 145 with very basic combat 2011-12-14T17:20:59 commit the code and we will add it to the list for finals. 2011-12-14T17:21:00 I will never attack 2011-12-14T17:21:07 *** pairofdice has quit IRC (Ping timeout: 240 seconds) 2011-12-14T17:21:18 McLeopold: it's comitted, in my branch 2011-12-14T17:21:23 i could put it in a folder instead 2011-12-14T17:21:35 are there filename collisions? 2011-12-14T17:21:50 i have like 4 files, i don't really want to mix them 2011-12-14T17:22:16 okay, folder it is 2011-12-14T17:22:29 can you merge to epsilon, and then I will generate a bunch for official 2011-12-14T17:22:44 yes 2011-12-14T17:23:25 *** cuda has joined #aichallenge 2011-12-14T17:23:25 *** cuda is now known as pairofdice 2011-12-14T17:26:24 aichallenge: Alexandru Stan amstan_mapgen * r93c9096 / (12 files in 2 dirs): moved stuff to a folder - http://git.io/bXCtQg 2011-12-14T17:27:46 hi there :) is there any thread / information available about the next challenge? 2011-12-14T17:28:00 i haven't found all too much information until now 2011-12-14T17:28:05 *** pairofdice has quit IRC (Ping timeout: 252 seconds) 2011-12-14T17:28:56 aichallenge: Alexandru Stan epsilon * r93c9096 / (12 files in 2 dirs): moved stuff to a folder - http://git.io/bXCtQg 2011-12-14T17:28:56 aichallenge: Alexandru Stan epsilon * r0080bd9 / (6 files): Merge branch 'amstan_mapgen' into epsilon - http://git.io/DHZEdQ 2011-12-14T17:28:58 aichallenge: Alexandru Stan epsilon * r5225ae4 / (8 files in 5 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/kpijlQ 2011-12-14T17:31:15 McLeopold: for translational maps there's a need to fiddle with the settings 2011-12-14T17:32:01 McLeopold: since they kinda look like the maps that people on the forums are complaining about, where the players are one in a line 2011-12-14T17:32:24 strong engines please connect to tcpants :) bj & strcat ? 2011-12-14T17:34:44 amstan: what's wrong with players in a line? 2011-12-14T17:34:52 hmm flagcapper just resubmitted. It is crunch time everywhere .. 2011-12-14T17:35:02 it's based on luck then, on how the players order is 2011-12-14T17:35:21 if you're next to xathis you're screwed, if you're opposite to xathis it's slightly better 2011-12-14T17:35:33 I disagree 2011-12-14T17:35:52 if you have some defense and can get an ant out to the open field, you can keep you spawn rate up 2011-12-14T17:36:04 it's part of the game 2011-12-14T17:36:07 it doesn't matter where you ants are, you hill will still be there 2011-12-14T17:36:23 *** foRei has quit IRC (Read error: Connection reset by peer) 2011-12-14T17:36:23 it's clear that a stronger bot will prefer attacking the neighbours first, then the hills farther 2011-12-14T17:36:32 so will theirs 2011-12-14T17:36:47 high food gathering can compensate 2011-12-14T17:37:22 yes, but using the same food gathering technique can get your farther if you're far away from the crazy bot 2011-12-14T17:37:41 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-12-14T17:38:26 *** yoden-cloud has joined #aichallenge 2011-12-14T17:40:38 amstan: about the cutoff rules, didn't we agree that the leaders always need to defend even if they don't have hills? 2011-12-14T17:41:14 yes? 2011-12-14T17:41:34 I remembering we outvoted janzert on that awhile back, while testing extended play 2011-12-14T17:41:49 yes, but i don't see how it's relevant 2011-12-14T17:41:56 it is a little, but not a lot 2011-12-14T17:42:48 well, for ties, we'd let the game go longer 2011-12-14T17:43:12 it's not about ties though 2011-12-14T17:43:16 so, if you are in the lead, we check to see if your rank can go down as well 2011-12-14T17:43:23 i'm talking when you have maps like this: 2011-12-14T17:43:28 0 1 2 3 2011-12-14T17:43:31 all in the same line 2011-12-14T17:43:38 amstan: totally different issue :) 2011-12-14T17:43:38 *** dici has quit IRC (Read error: Connection reset by peer) 2011-12-14T17:43:47 let's say 0 is a good bot and can kill everyone easily 2011-12-14T17:43:53 1 and 3 will have a much harder time than 2 2011-12-14T17:44:01 BenJackson posted in the forums about a game that should have gone longer, I agree with him 2011-12-14T17:44:03 even if 1 2 and 3 were the same 2011-12-14T17:44:15 you just need to have enough games 2011-12-14T17:44:21 and that sort of positional bias should average out 2011-12-14T17:44:44 yoden-cloud: yes, but it's not a good idea to induce such randomness in the games 2011-12-14T17:44:46 amstan: can we go away from starting position discussions to cutoff rules now :) 2011-12-14T17:44:50 yoden-cloud: it just makes rankings converge slower 2011-12-14T17:45:13 yeah, and the rankings already converge so slow 2011-12-14T17:45:18 McLeopold: that's what i meant by problems with my mapgen 2011-12-14T17:45:26 yoden-cloud: therefore it's not a good idea to make them even slower 2011-12-14T17:45:26 I on a different subject now 2011-12-14T17:45:33 McLeopold: k 2011-12-14T17:45:38 the cutoff rules 2011-12-14T17:48:16 *** Murashka has quit IRC (Quit: Page closed) 2011-12-14T17:49:47 *** bergmark has quit IRC (Ping timeout: 240 seconds) 2011-12-14T17:50:54 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-14T17:51:47 *** ikaros has joined #aichallenge 2011-12-14T17:52:39 I'd really be pretty strongly opposed to rule changes at this stage 2011-12-14T17:52:48 in any direction 2011-12-14T17:53:18 *** choas has quit IRC (Ping timeout: 255 seconds) 2011-12-14T18:02:05 *** SMJ has quit IRC (Read error: Connection reset by peer) 2011-12-14T18:03:11 amstan: relative starting positions makes games "not symmetric" regardless of whether players start in a row or not 2011-12-14T18:03:49 *** SMJ has joined #aichallenge 2011-12-14T18:05:34 ack and we're introducing new map types 4 days from the contest end? 2011-12-14T18:06:09 McLeopold: where's this post about a game that shouldn't have ended? 2011-12-14T18:06:12 i can't find it anywhere 2011-12-14T18:06:29 http://forums.aichallenge.org/viewtopic.php?f=27&t=2123 2011-12-14T18:06:43 http://aichallenge.org/visualizer.php?game=234535&user=757 luuuuuuuuuuck 2011-12-14T18:06:51 *** kire has quit IRC (Remote host closed the connection) 2011-12-14T18:08:43 does a bot count as participating when they have no hills? 2011-12-14T18:08:59 not for cutoff by rank stabilized 2011-12-14T18:09:02 i thought we quite often cut bots off when they don't have hills left, even if they could still increase in ranking? 2011-12-14T18:09:18 antimatroid: it isn't yes or no, just sometimes 2011-12-14T18:09:25 ie. in a 2 player game the moment a hill is razed the game is over right? 2011-12-14T18:09:30 I wanted it to be straight up no but was over ruled 2011-12-14T18:09:49 you'll have to consult with McLeopold and amstan to figure out what exactly sometimes means ;) 2011-12-14T18:09:53 i'm with janzert, it's highly unideal to change something now, but personally i think the rank stabilised thing was pretty messed up for this 2011-12-14T18:11:59 McLeopold: whate are the rules for a bot to count as "still participating"? 2011-12-14T18:12:01 personally I think there should have been a way for white to win :) 2011-12-14T18:12:11 :) 2011-12-14T18:12:37 BenJackson; sure but we know about that "bug" with the game 2011-12-14T18:13:34 in the future i think the best thing to do would be to only cut a game short of max turns if no bot can change in ranking 2011-12-14T18:14:04 *** Antimony has joined #aichallenge 2011-12-14T18:14:27 I'd rather bots that lose all their hills be straight up knocked out of the game at that point, i.e. their controller can no longer communicate with the ants 2011-12-14T18:15:11 I don't really think we should do either or anything else of course 2011-12-14T18:16:09 janzert: i'm not sure i agree, take a 2 player example, what if the "losing player" was 1 turn from also razing the enemy hive? 2011-12-14T18:16:29 i was told months ago the game should make intuitive sense for real life, in real life an ant isn't going to just stop because it'd hill has been razed 2011-12-14T18:16:37 its* 2011-12-14T18:16:41 should of put a little thought into defense 2011-12-14T18:16:49 in real life an ant controls itself 2011-12-14T18:17:05 here we have a hive mind of some sort 2011-12-14T18:17:09 sure, and it can be here, it just has telepathic information sharing :P 2011-12-14T18:17:24 oh god, I just opened the fridge at work 2011-12-14T18:17:27 saw some diet coke cans on the shelf 2011-12-14T18:17:34 and analyzed them as an ant formation 2011-12-14T18:17:39 instinctively 2011-12-14T18:17:39 haha 2011-12-14T18:18:58 hehe, so does that stage come before or after the dreams? 2011-12-14T18:19:20 no dreams about ants yet 2011-12-14T18:21:38 *** Dlayne has left #aichallenge 2011-12-14T18:21:48 *** bluegaspode has joined #aichallenge 2011-12-14T18:25:23 *** Dan___ has joined #aichallenge 2011-12-14T18:26:04 hey everybody... any good techniques for debugging a timeout that only happens once in a while? 2011-12-14T18:26:05 haha 2011-12-14T18:26:21 use the "bot in" button and feed it in again 2011-12-14T18:26:29 then use a profiler or timing code to see where it is slow 2011-12-14T18:27:01 is there a way to make the fog of war dark instead of light in the online viewer? 2011-12-14T18:28:56 you might be able to change the opacity number in the JS if you're using chrome or Firefox 2011-12-14T18:29:00 thanks Ben... i'll try that out 2011-12-14T18:29:26 finally got a shot at the big boys and timed out in my last game 2011-12-14T18:29:28 sigh 2011-12-14T18:29:50 *** SMJ has quit IRC (Quit: Konversation terminated!) 2011-12-14T18:33:09 antimatroid: has a hill and hasn't crashed or timedout 2011-12-14T18:34:17 *** SMJ has joined #aichallenge 2011-12-14T18:34:25 if it is participating, the game continues if it can gain in rank 2011-12-14T18:41:21 McLeopold: so the game doesn't actually end early according to that spec? 2011-12-14T18:41:35 oh wait 2011-12-14T18:42:40 i guess they could increase in ranking if an enemy razed their hill 2011-12-14T18:43:17 no they couldn't 2011-12-14T18:43:23 how do you handle deciding when to make trades in combat? 2011-12-14T18:43:36 Antimony: only trade if you kill more than you lose 2011-12-14T18:43:46 otherwise i minimise friend deaths 2011-12-14T18:44:18 but it seems like 1v1 trades aren't always good 2011-12-14T18:44:24 like if they have a lot more ants than you 2011-12-14T18:44:47 you kill the same as you lose then, don't trade, minimise friend deaths 2011-12-14T18:44:52 McLeopold... they got my hill 100 turns later 2011-12-14T18:44:52 http://aichallenge.org/visualizer.php?game=234363&user=4165 2011-12-14T18:45:10 not sure what caused my timeout 2011-12-14T18:48:07 McLeopold: so why is that game cut short? 2011-12-14T18:48:24 the only player with a hill left can' get any points 2011-12-14T18:48:28 can't 2011-12-14T18:48:42 the other 2 leaders could beat each other, but that is it 2011-12-14T18:49:12 so, don't bother fighting ants if they don't have a hill, just urn 2011-12-14T18:49:13 yes, but the remaining participating player can't change in ranking right? 2011-12-14T18:49:14 run 2011-12-14T18:49:28 no, there are no hills to raze, no points to be had 2011-12-14T18:49:37 so the game hasn't been cut short? 2011-12-14T18:49:43 even if he exterminated everyone and had enough growth to cover all land 2011-12-14T18:49:51 that's my point 2011-12-14T18:49:58 people said it was cut short? but cut short of what? 2011-12-14T18:50:18 well, the 2 tied for the lead could try and raze his hill, so you might see a different outcome 2011-12-14T18:50:35 instead of 1,1,3 it could be 1,2,3 2011-12-14T18:50:39 yes, but our definition of rank stabilised here is the same as anywhere else 2011-12-14T18:50:50 1,2,2 -> 1,2,3* 2011-12-14T18:50:59 1,2,2 -> 0,2,3* 2011-12-14T18:51:36 while i don't agree with it, we are cutting plenty of other games short when a non-participating player could increase in ranking 2011-12-14T18:52:03 right, I would only want to extend it if it is the *leader* who could lose in rank 2011-12-14T18:52:17 why? 2011-12-14T18:52:23 if fact, I thought I had that code in there 2011-12-14T18:52:24 all rankings are equally important to trueskill no? 2011-12-14T18:52:47 yes, I would extend all games until all ranks are finalized up or down if I could 2011-12-14T18:52:50 why does it matter if it's the leader or second last place who could lose in ranking? 2011-12-14T18:53:05 i think we should at least be consistent either way 2011-12-14T18:53:12 for the same reason nobody cares about last place 2011-12-14T18:53:22 they do though, or they should 2011-12-14T18:53:23 *** tobym has quit IRC (Remote host closed the connection) 2011-12-14T18:53:27 *** Paradoxial has quit IRC (Ping timeout: 240 seconds) 2011-12-14T18:53:36 we are inconsistant already 2011-12-14T18:53:41 we only check for a gain in rank 2011-12-14T18:53:54 not a loss in rank 2011-12-14T18:54:02 what's inconsistent about that? 2011-12-14T18:54:38 *** Apophis has quit IRC (Read error: Connection reset by peer) 2011-12-14T18:55:39 I can't think of anything :) 2011-12-14T18:56:37 *** bluegaspode has quit IRC (Quit: Page closed) 2011-12-14T18:57:19 @ranking 2011-12-14T18:57:20 Anilm3: Run as fast as you can and don't look back. 2011-12-14T18:57:21 @rankings 2011-12-14T18:57:23 McLeopold: in the future i think we need to be more careful when cutting games short of the specified max turns 2011-12-14T18:57:24 Anilm3: Top 10 players: xathis(95.3), pguillory(93.0), fourmidable(91.2), a1k0n(91.0), delineate(88.9), protocolocon(88.6), ThisIsNotABug(88.2), bucash(88.0), jacob_strauss(87.4), bix0r4ever(87.3) 2011-12-14T18:57:40 what happened to flag? 2011-12-14T18:57:40 _flag resubmitted!!! 2011-12-14T18:57:42 resubmit i guess 2011-12-14T18:57:52 oh god, that may be a game changer 2011-12-14T18:58:09 _flag: how good is your new bot? :P 2011-12-14T19:00:00 protocolocon is doing well 2011-12-14T19:00:06 i imagine that's his final entry 2011-12-14T19:04:52 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-14T19:05:03 *** Apophis has joined #aichallenge 2011-12-14T19:05:05 ugh 2011-12-14T19:05:24 so... at some point, my bot started randomly timing out during setup 2011-12-14T19:05:45 and I cannot for the life of me figure out why 2011-12-14T19:06:24 McLeopold: what do you think? http://forums.aichallenge.org/viewtopic.php?f=21&t=2122 2011-12-14T19:07:11 i hope my computer is worse than the servers 2011-12-14T19:07:59 amstan: what would the grid be? and how is the physics going to work? 2011-12-14T19:08:10 if we're going for something like that I think we should talk to delt0r 2011-12-14T19:08:20 floating point? 2011-12-14T19:08:24 the game he had in the works sounded pretty awesome 2011-12-14T19:08:26 physics based on time steps 2011-12-14T19:08:58 amstan: I like it 2011-12-14T19:09:12 floating point will be an issue 2011-12-14T19:09:13 another possibility would be a game where you go around shooting your enemies 2011-12-14T19:09:18 and bullets have a given trajectory 2011-12-14T19:09:21 McLeopold: we could go with a very small grid then 2011-12-14T19:09:32 if we're talking about xpilot (or asteroids in general) the precision isn't that big a deal 2011-12-14T19:09:32 32 bit to represent the whole map 2011-12-14T19:09:35 is there a terminal velocity in space, besides light speed? 2011-12-14T19:09:38 or we could allow people to program their own missile scripts, like a missile can't go backwards or stay sitll? 2011-12-14T19:09:50 McLeopold: what about neutrinos? :P 2011-12-14T19:10:00 yes! we could get acceleration dependent on speed, so you can't get to speed of light 2011-12-14T19:10:15 right, I just wonder what the formula is 2011-12-14T19:10:23 what happens to an asteroid when you shoot it? 2011-12-14T19:10:37 splits in half 2011-12-14T19:10:40 antimatroid: that's an issue 2011-12-14T19:10:49 do we want to make this stochastic or what? 2011-12-14T19:10:51 I would say all objects are perfect circles 2011-12-14T19:11:05 colisions split in half at the tagent 2011-12-14T19:11:08 did any of you play that old dos game where you controlled a ship on a grid and had to shoot down enemy ships who were trying to shoot you down etc.? 2011-12-14T19:11:08 splits in half, but then where do the parts go? 2011-12-14T19:11:12 something like that could work well 2011-12-14T19:11:22 use momentum equations on points 2011-12-14T19:11:29 *** JorgeB has quit IRC (Ping timeout: 252 seconds) 2011-12-14T19:11:38 McLeopold: is there any way to predict where the smaller parts will go? 2011-12-14T19:11:39 amstan: personally i'm very against making games stochastic 2011-12-14T19:11:41 maybe add perpendicular momentum to the 2 new objects 2011-12-14T19:11:51 antimatroid: food is already stochastic 2011-12-14T19:12:08 BenJackson: sure, but i mean like stochastic resolution etc. 2011-12-14T19:12:20 ie. the random spawn tiebreak for ants I despise 2011-12-14T19:12:26 I prefer deterministic 2011-12-14T19:12:43 how about sub particle velocity=old particle velocity+velocity tangential to the bullet 2011-12-14T19:12:48 I think asteroids would be awesome BUT I think it would restrict the participants a lot 2011-12-14T19:12:59 eg tron's winnin strategy was heavy on graph theory and etc 2011-12-14T19:13:07 but anyone could play it 2011-12-14T19:13:14 ants is a little harder 2011-12-14T19:13:14 amstan: yes 2011-12-14T19:13:20 but still anyone can gather food and wander around 2011-12-14T19:13:22 I rekon a grid based game where your ships can shoot torpedos that your bot can also control would be cool 2011-12-14T19:13:27 the point of the game would be make your opponent crash with an asteroid 2011-12-14T19:13:49 in a space sim even knowing where you want to go still requires significant math to figure out how to thrust there efficiently 2011-12-14T19:13:52 the controls would be "left, right, straight" and "burn noburn" 2011-12-14T19:13:55 and "fire" 2011-12-14T19:14:01 BenJackson: yes 2011-12-14T19:14:07 "self-destruct" 2011-12-14T19:14:10 McLeopold: burn? 2011-12-14T19:14:12 thestinger: why? 2011-12-14T19:14:13 (disclaimer: I have written parts of the xpilot ai) 2011-12-14T19:14:24 amstan: dunno :P 2011-12-14T19:14:31 thestinger: exit(1) 2011-12-14T19:14:34 amstan: like, fire up the engines and go forward 2011-12-14T19:14:45 amstan: but then you don't take out nearby enemies :( 2011-12-14T19:15:07 McLeopold: what's straight for then? 2011-12-14T19:15:12 don't turn 2011-12-14T19:15:31 is there fuel? does burning fuel reduce your weight? 2011-12-14T19:15:35 does running out leave you stranded? 2011-12-14T19:15:36 unless we say turning and burning can't happen at the same time 2011-12-14T19:15:37 can you refuel? 2011-12-14T19:15:49 aww you ahve to be able to "turn and burn"! 2011-12-14T19:15:52 maybe just have a left thruster and a right thruster and you can turn them on/off 2011-12-14T19:16:05 would we want to make this physics accurate? have rotational momentum and so on? 2011-12-14T19:16:10 no 2011-12-14T19:16:15 gravity? 2011-12-14T19:16:22 unidirectional or towards suns? 2011-12-14T19:16:22 black holes! 2011-12-14T19:16:23 gravity would be nice too 2011-12-14T19:16:24 no 2011-12-14T19:16:35 hohman transfer orbits 2011-12-14T19:16:42 shooting directly at another player? 2011-12-14T19:16:48 BenJackson: yes, lol 2011-12-14T19:16:57 sling shot torpedos 2011-12-14T19:16:59 should totally be able to shoot down opponents 2011-12-14T19:17:07 well, maybe you should have a recharge rate for shooting 2011-12-14T19:17:08 i like the idea of bot controlled torpedos 2011-12-14T19:17:13 with different rules about how they can move 2011-12-14T19:17:39 it would be nice if there was a way to make particle calculations impossible to do 2011-12-14T19:17:41 like 2011-12-14T19:17:46 have asteroids in the game 2011-12-14T19:18:00 and make it so it's impossible to predict the positions of all of them 2011-12-14T19:18:06 what would obstacles be? 2011-12-14T19:18:08 you can get the game engine more cores, and code in C 2011-12-14T19:18:16 just open maps with asteroids? 2011-12-14T19:18:27 antimatroid: i'm not sure, but that would be cool 2011-12-14T19:18:32 btw one key part of xpilot is how slow the shots are 2011-12-14T19:18:39 ships can easily fly faster than shots 2011-12-14T19:18:42 trying to predict where asteroids would be without actually simulating everything manually 2011-12-14T19:18:54 so imparting your own momentum to shots is important 2011-12-14T19:19:02 BenJackson: are shot velocities relative to ship velocities? 2011-12-14T19:19:06 yes 2011-12-14T19:19:08 nice 2011-12-14T19:19:13 xpilot is a lot of fun 2011-12-14T19:19:28 k, that's it 2011-12-14T19:19:29 aptgetting it 2011-12-14T19:19:36 BenJackson: have a server or something? 2011-12-14T19:19:37 and already has a full engine with arbitrary maps and replays and etc 2011-12-14T19:19:43 xpilot-ng? 2011-12-14T19:19:50 there's a package for that I'm installing atm 2011-12-14T19:19:55 amstan: what's the plan post ants? 2011-12-14T19:19:57 yeah, I thik that's U's version 2011-12-14T19:20:07 U did a rewrite in C++ at some point, but that's after my time 2011-12-14T19:20:08 antimatroid: get another game going, implement it with django 2011-12-14T19:20:10 maybe different types of ammunition might be nice 2011-12-14T19:20:11 are we going to bite the bullet and do zeta? or just patch another game together with the existing framework? 2011-12-14T19:20:27 thestinger: it's all client-server already 2011-12-14T19:20:28 antimatroid: the patching part was supposed to take 1 month 2011-12-14T19:20:36 so you can already write a client bot if you want 2011-12-14T19:20:40 people probably have 2011-12-14T19:20:45 antimatroid: unfortunatelly it 8 2011-12-14T19:21:04 definitely a lot of similar stuff you can do, it could be with planes/subs rather than in space 2011-12-14T19:21:12 BenJackson: do you have a server? 2011-12-14T19:21:28 heat seeking torpedos... 2011-12-14T19:21:41 turn of engines and float to avoid 2011-12-14T19:21:45 or lead to enemy 2011-12-14T19:21:59 it would be nice if this could be real time 2011-12-14T19:22:04 where it doesn't matter if you timeout 2011-12-14T19:22:12 * antimatroid wants tcp servers for past games :( 2011-12-14T19:22:14 you just keep going 2011-12-14T19:22:14 :p 2011-12-14T19:22:15 it would be really neat to have some kind of mechanic like how sub/plane battles happen - where shooting reveals your location 2011-12-14T19:22:23 you can get semi-real time 2011-12-14T19:22:26 *** chris__0076 is now known as chris_0076 2011-12-14T19:22:28 amstan: no, I don't run a server 2011-12-14T19:22:30 I did years ago 2011-12-14T19:22:37 allow different actions to take different amounts of time 2011-12-14T19:22:40 BenJackson: do you know a good one we can all join? 2011-12-14T19:22:50 no, I haven't played in years 2011-12-14T19:22:53 it's easy to create a local server 2011-12-14T19:23:07 keep in mind I originally played this on a 486 with no graphics accel 2011-12-14T19:23:16 udp 0 0 *:xpilot *:* 2011-12-14T19:23:17 lol 2011-12-14T19:23:22 the port is actually assigned to it 2011-12-14T19:23:27 it crashes on me 2011-12-14T19:23:29 yep, it's ollld 2011-12-14T19:23:40 crashes? 2011-12-14T19:23:54 oh hey I thikn someone made a sever menu in the client 2011-12-14T19:23:56 I forgot about that 2011-12-14T19:24:03 I was the one who put the latency measurement in 2011-12-14T19:24:03 works for me 2011-12-14T19:24:10 I also thought a tank game might be fun, where you shoot projectiles, like scorched earth 2011-12-14T19:24:17 or atari TANK 2011-12-14T19:24:32 that was actually what I was thinking :) 2011-12-14T19:24:43 nice, shooting propels you backwards 2011-12-14T19:24:44 they had a shoot to reveal mode in that, didn't they? 2011-12-14T19:24:57 but you didn't know where you where either 2011-12-14T19:25:10 thestinger: or forwards 2011-12-14T19:25:14 wow wtf, the controls are like 20 pages long 2011-12-14T19:25:14 thestinger: with enough multishot it can overcome your thrust 2011-12-14T19:25:18 without boosted thrusters 2011-12-14T19:25:26 thestinger: there are TONS of powerups 2011-12-14T19:25:36 but most people play one particular map which is very simple 2011-12-14T19:25:38 blood's music 2011-12-14T19:26:01 early on walls were super deadly (in the classic style) but these days the walls are a bit softer 2011-12-14T19:26:43 amstan: how are you running it? 2011-12-14T19:26:59 amstan: the only problem is a game of asteroids doesn't have a natural time span 2011-12-14T19:27:08 I'm using an xpilot-ng package and there's an xpilot-ng-server and xpilot-ng-x11 as the client 2011-12-14T19:27:27 there's also xpilot-ng-sdl which works too 2011-12-14T19:27:31 unless we limit fuel and ammunition 2011-12-14T19:27:35 it has a server menu thing to join existing ones 2011-12-14T19:27:37 McLeopold: sure it does, take out all the enemy ships 2011-12-14T19:27:42 don't let people spawn more 2011-12-14T19:28:07 antimatroid: you can have a situation where 2 players don't make progress 2011-12-14T19:28:11 it was impossible in tron 2011-12-14T19:28:12 *** Antimony has joined #aichallenge 2011-12-14T19:28:19 nice, it has a GUI map editor 2011-12-14T19:28:22 xD 2011-12-14T19:28:22 the game had a natual ending, no matter what happened 2011-12-14T19:28:25 *** replore has joined #aichallenge 2011-12-14T19:28:46 *** keith_ has joined #aichallenge 2011-12-14T19:28:48 BenJackson: how long ago did you play this lol? 2011-12-14T19:28:52 yeah, i don't think you'll ever get that with games where the state isn't permanently altered from each move 2011-12-14T19:29:05 ie. tron, dots and boxes, etc. 2011-12-14T19:29:36 right, so limiting fuel and ammunition helps with that 2011-12-14T19:29:38 McLeopold: we could have the asteriods randomly split 2011-12-14T19:29:52 and they could split to the point where it's hard to avoid them 2011-12-14T19:29:57 *** Dan___ has quit IRC (Quit: Page closed) 2011-12-14T19:29:57 amstan: I think there would be enough collisions to go around without that 2011-12-14T19:29:59 amstan: banach-tarski split? 2011-12-14T19:30:06 ? 2011-12-14T19:30:16 both new asteroids are the same size :P 2011-12-14T19:30:23 so eventually they fill the map 2011-12-14T19:30:56 some kind of multiplayer bomberman could be funny too 2011-12-14T19:31:04 amstan: http://en.wikipedia.org/wiki/Banach%E2%80%93Tarski_paradox 2011-12-14T19:31:58 antimatroid: i don't get it 2011-12-14T19:32:09 ikaros: but that's pretty much minimax 2011-12-14T19:32:25 amstan: so, in classic asteroids, they split 3 times, we can split more than that 2011-12-14T19:32:32 it is possibly to take a 3-dimensional sphere and turn it into 2 identical copies 2011-12-14T19:32:36 not physically possibly, but possible 2011-12-14T19:32:39 Can you write data to the disk and read it later? 2011-12-14T19:32:40 possible* 2011-12-14T19:32:58 ronchaine: later the same game, yes, but not between games 2011-12-14T19:33:19 ronchaine: but we'd rather you didn't, and if you exceed 10MB, it's bad 2011-12-14T19:33:31 ronchaine: you might as well just use ram 2011-12-14T19:33:33 amstan, doesnt need to be a classic approach.. could include buying options or whatever to increase complexity.. but i agree it probably would end in minmax 2011-12-14T19:34:03 thing is asteroid stuff will be 70% physics 2011-12-14T19:34:42 i think it's relevant though, you have lots of cases where physics and ai are intermingled 2011-12-14T19:34:49 fyi i will try and minimax everything you throw at me 2011-12-14T19:35:02 yea its relevant 2011-12-14T19:35:13 antimatroid: so if i accelerate using a float force, go minimax that 2011-12-14T19:35:18 ikaros: barely, the physics should be simple to learn 2011-12-14T19:35:27 then it's about what you can do with it, which is going to be maths 2011-12-14T19:35:28 McLeopold, amstan: ok, thanks 2011-12-14T19:35:45 well 2011-12-14T19:35:52 amstan: i'd heuristically pick our moves for me and the opponents and run minimax on that 2011-12-14T19:36:00 like i did for planet wars and like i am for combat in ants 2011-12-14T19:36:17 and i don't think physics would be that much, you can definatelly finish everything physics related for the bot 2011-12-14T19:36:21 as for ai, you'll never know 2011-12-14T19:37:13 *** grwip has quit IRC (Remote host closed the connection) 2011-12-14T19:37:14 it wouldn't be too hard 2011-12-14T19:37:26 starter bots come with some equations 2011-12-14T19:37:48 i would love something which requires strategic planning alot more.. like sim games.. but this would probably be to complex and i cant propose an actual idea 2011-12-14T19:37:48 though i would still like some aspect of it that can't be perfectly simulated with physics, so bots would have to use other methods 2011-12-14T19:37:53 i want to do a grid based game where the squares have a randomly assigned height and you can only walk between locations when like abs(loc1.height-loc2.height) < b 2011-12-14T19:38:20 sim games? 2011-12-14T19:38:31 simulation 2011-12-14T19:38:35 antimatroid: i want to make a minecraft like game, where you can change the map around you 2011-12-14T19:38:41 antimatroid: isn't that just a maze? 2011-12-14T19:38:44 amstan: yes please :) 2011-12-14T19:38:45 thats a nice idea amstan 2011-12-14T19:38:47 unless you can always go down? 2011-12-14T19:39:01 McLeopold: sure, but you don't have any explicit walls 2011-12-14T19:39:14 and it might be cool to allow a bot to always walk down 2011-12-14T19:39:23 so, one way doors? 2011-12-14T19:39:27 in fact that's awesome 2011-12-14T19:39:35 cause then you can have sections of the map what are like walled off 2011-12-14T19:39:41 and then they need to build their way back out etc. 2011-12-14T19:40:13 but that's still minimaxable 2011-12-14T19:40:24 depends on the size of hte grids 2011-12-14T19:40:26 or a* 2011-12-14T19:40:40 you can't minimax entire battle partitions in ants 2011-12-14T19:40:44 it's just not going to happen 2011-12-14T19:41:00 if you have one agent, doesn't really matter what the size is, you can only do so many actions per turn 2011-12-14T19:41:20 lemmings? 2011-12-14T19:41:28 =) 2011-12-14T19:41:43 worms! 2011-12-14T19:41:53 the built-in AI in xpilot-ng is killing me :) 2011-12-14T19:42:27 thestinger: yeah, for the asteroids game, any kind of starter package would kill all humans 2011-12-14T19:42:51 since they can easily predict/calculate what the state will be after a delta t 2011-12-14T19:42:59 and the controls are pretty crap :P 2011-12-14T19:43:02 a and s to steer 2011-12-14T19:43:17 i could only get the mouse to work for me 2011-12-14T19:43:17 its late.. g'night 2011-12-14T19:43:21 thestinger: wanna play? 2011-12-14T19:43:22 ikaros: bye 2011-12-14T19:43:23 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-14T19:43:35 amstan: sure, but I'm having trouble getting it to work over the network 2011-12-14T19:43:41 maybe it works 2011-12-14T19:44:12 try the .ch server 2011-12-14T19:44:15 second one i think 2011-12-14T19:44:38 in that now 2011-12-14T19:47:11 *** keith_ has quit IRC (Ping timeout: 258 seconds) 2011-12-14T19:47:28 *** McLeopold has left #aichallenge 2011-12-14T19:47:31 *** TheLinker has quit IRC (Read error: Connection reset by peer) 2011-12-14T19:54:32 bleh 2011-12-14T19:54:39 my turns usually take ~30 ms 2011-12-14T19:55:00 but the first one takes about 150, since I guess java is busy JIT'ing things 2011-12-14T19:55:10 and i guess the servers are that much slower than my laptop :( 2011-12-14T20:04:41 thestinger: ggs 2011-12-14T20:04:43 *** treeform has quit IRC (Remote host closed the connection) 2011-12-14T20:05:14 *** treeform has joined #aichallenge 2011-12-14T20:07:19 *** Harpyon has quit IRC (Quit: Textual IRC Client: http://www.textualapp.com/) 2011-12-14T20:13:17 *** LoneRanger has quit IRC (Remote host closed the connection) 2011-12-14T20:22:24 Hum, can I put a file into disk that I can read 2011-12-14T20:22:29 Nvm. 2011-12-14T20:22:46 Tried to push up arrow and enter on tty. 2011-12-14T20:22:48 Failed. 2011-12-14T20:28:21 *** dmj111 has joined #aichallenge 2011-12-14T20:36:15 *** chris__0076 has joined #aichallenge 2011-12-14T20:37:11 *** thestinger has quit IRC (Quit: WeeChat 0.3.6) 2011-12-14T20:37:34 *** thestinger has joined #aichallenge 2011-12-14T20:37:38 nice, it froze X 2011-12-14T20:37:39 *** chris_0076 has quit IRC (Ping timeout: 252 seconds) 2011-12-14T20:37:57 thestinger: oh.. that's why you were acting like you were adrift 2011-12-14T20:38:24 anyway, I suck at that game 2011-12-14T20:38:26 xD 2011-12-14T20:41:12 BenJackson: doshoes1990@live.com <- just got spam from there :P 2011-12-14T20:50:52 *** Sir_Ragnarok has quit IRC (Ping timeout: 252 seconds) 2011-12-14T20:51:04 http://ants.fluxid.pl/replay.33689 yay for working defense :D 2011-12-14T20:51:08 will probably merge it in a bit 2011-12-14T20:52:58 *** alc has joined #aichallenge 2011-12-14T20:56:35 *** Jak_o_Shadows has joined #aichallenge 2011-12-14T21:06:37 haha i'm in 3rd 2011-12-14T21:07:35 *** Sir_Ragnarok has joined #aichallenge 2011-12-14T21:09:07 anyone want to sell their copy of AI: A Modern Approach 3rd Ed? 2011-12-14T21:09:17 I'm waiting for the market to crash post-ai-class 2011-12-14T21:11:22 *** Redgis has quit IRC (Ping timeout: 252 seconds) 2011-12-14T21:13:24 *** Paradoxial has joined #aichallenge 2011-12-14T21:19:05 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-12-14T21:20:20 *** yoden-cloud has quit IRC (Ping timeout: 258 seconds) 2011-12-14T21:25:41 *** flowenol has quit IRC (Ping timeout: 252 seconds) 2011-12-14T21:25:52 haha. i have that book somewhere. i might have given it to a friend already tho 2011-12-14T21:33:28 *** treeform has quit IRC (Remote host closed the connection) 2011-12-14T21:37:41 *** Fandekasp has quit IRC (Ping timeout: 244 seconds) 2011-12-14T21:39:20 *** codetiger has joined #aichallenge 2011-12-14T21:39:42 Fluxid: http://ants.fluxid.pl/player/strcat_newdefend does this load for you? 2011-12-14T21:39:47 I think the screw up happened again 2011-12-14T21:40:58 when X crashed, all my bots got disconnected - I think that screwed up the page 2011-12-14T21:43:57 the worst thing my bot can do is survive a multihill maze 2011-12-14T21:44:18 wondering why it hates close neighbours 2011-12-14T21:50:26 *** amriedle has joined #aichallenge 2011-12-14T21:50:54 *** dom7b5 has quit IRC (Quit: Ex-Chat) 2011-12-14T21:54:27 *** Fandekasp has joined #aichallenge 2011-12-14T21:55:34 *** replore has quit IRC (Read error: Connection reset by peer) 2011-12-14T21:55:42 *** 16SAADD24 <16SAADD24!~replore@203.152.213.161.static.zoot.jp> has joined #aichallenge 2011-12-14T22:01:25 *** Antimony has joined #aichallenge 2011-12-14T22:04:44 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-12-14T22:13:59 *** besh has quit IRC (Quit: Page closed) 2011-12-14T22:24:06 *** Cyndre has joined #aichallenge 2011-12-14T22:25:57 *** amriedle has quit IRC (Quit: Lost terminal) 2011-12-14T22:27:42 *** codetiger has quit IRC (Remote host closed the connection) 2011-12-14T22:30:06 *** Cyndre_ has joined #aichallenge 2011-12-14T22:31:14 *** Cyndre has quit IRC (Quit: Leaving) 2011-12-14T22:38:55 *** pguillory has quit IRC (Ping timeout: 258 seconds) 2011-12-14T22:42:00 *** TheLinker has joined #aichallenge 2011-12-14T22:44:55 *** treeform has joined #aichallenge 2011-12-14T22:52:33 *** Dan__ has joined #aichallenge 2011-12-14T22:52:54 Hey everybody... I'm having a weird timeout issue 2011-12-14T22:52:59 *** Dan__ is now known as Guest15667 2011-12-14T22:53:10 where I keep randomly timing out 2011-12-14T22:53:24 at various parts of my code... 2011-12-14T22:53:27 in Battle Res 2011-12-14T22:53:32 issuing orders, etc... 2011-12-14T22:53:51 it seems like 300-400ms of nothing happens in between 2 logging lines of code 2011-12-14T22:53:57 no loops, etc... 2011-12-14T22:54:21 anybody have any ideas how to debug? 2011-12-14T22:54:23 what king of logging? 2011-12-14T22:54:25 what language? 2011-12-14T22:55:05 Java... FileWriter to a .txt file 2011-12-14T22:55:27 that could be it 2011-12-14T22:55:58 did you try removing that? or switching it with an if statment? 2011-12-14T22:56:05 *** arscan has joined #aichallenge 2011-12-14T22:56:22 yeah I've turned off all logging and I still crash 2011-12-14T22:56:42 what happens during the crash then? 2011-12-14T22:56:50 where does it crash? 2011-12-14T22:57:01 have you checked memory usage/GC? 2011-12-14T22:57:12 i had a nasty memory leak which caused similar problems :( 2011-12-14T22:57:14 *** yoden1 is now known as yoden 2011-12-14T22:57:16 sorry... doesn't crash... times out 2011-12-14T22:57:23 it times out in random places 2011-12-14T22:57:24 Guest15667: yes, but where? 2011-12-14T22:57:36 well, you have to figure out where 2011-12-14T22:57:52 that's the thing, it times out in a different place every time 2011-12-14T22:58:04 lemme try making sure all logging is turned off 2011-12-14T22:58:06 everywhere 2011-12-14T22:58:13 because that would make sense... 2011-12-14T22:58:17 well 2011-12-14T22:58:20 not really 2011-12-14T22:58:28 *** dvladim has joined #aichallenge 2011-12-14T22:58:31 i was under the impression that your timeout was happening during the logging 2011-12-14T22:58:43 if that's not the case logging might actually help you solve your problem 2011-12-14T22:59:01 assuming you're certain it's not a significant time user 2011-12-14T22:59:18 the last logging I get is: 2011-12-14T22:59:19 [250] moveParticipantAndUpdateBattles 11: Friendly at ((12,42)) | X 2011-12-14T22:59:27 that means there are 250 ms left in the turn 2011-12-14T22:59:31 and that's the last thing I logged 2011-12-14T22:59:41 but I'll get a random assortment of those last lines 2011-12-14T22:59:53 at various places in my program 2011-12-14T23:00:01 with various amounts of time left 2011-12-14T23:00:31 well, i'm not sure, get those debugs in more places 2011-12-14T23:00:49 also.. as yoden said, GC might kick in at unexpected times 2011-12-14T23:01:07 you could try running that manually(if possible) and see how long it takes. 2011-12-14T23:01:10 interesting... 2011-12-14T23:01:23 amstan: the site is getting slower 2011-12-14T23:01:26 try running it at the start of every turn if you can, not sure if Java lets you do that 2011-12-14T23:01:29 ok i'll check that out... 2011-12-14T23:01:29 bigger db? more traffic? 2011-12-14T23:01:44 anybody have any GC tips? 2011-12-14T23:01:47 that are doing Java? 2011-12-14T23:01:50 eg I hit reload on my profile page before I said that 2011-12-14T23:01:51 it's still loading 2011-12-14T23:01:55 BenJackson: mysql is high cpuing 2011-12-14T23:02:17 janzert: help 2011-12-14T23:02:28 * BenJackson laughs 2011-12-14T23:02:39 BenJackson: lol 2011-12-14T23:02:51 i could restart it, but that means i have to restart all workers 2011-12-14T23:02:59 which would be a pain to do manually 2011-12-14T23:03:06 site is still fast for me 2011-12-14T23:03:31 oh, the rankings do lot pretty slowly 2011-12-14T23:03:40 load* 2011-12-14T23:03:57 it might be just temporary 2011-12-14T23:03:59 http://aichallenge.org/organization_profile.php?org=2693 what? 2011-12-14T23:04:34 thestinger: oh man! css issue! what is that? 2011-12-14T23:04:34 *** avdg has quit IRC (Quit: Leaving.) 2011-12-14T23:04:38 i don't even 2011-12-14T23:04:53 janzert just made a NULL vs 0 change didn't he? 2011-12-14T23:04:59 for org 2011-12-14T23:05:14 anyway I can I join [undefined]? I'd be winning 2011-12-14T23:05:22 lol 2011-12-14T23:05:30 I'm not winning "Other" or "C++" 2011-12-14T23:05:36 or probably USA but I didn't actually check 2011-12-14T23:05:38 switch to C++11 2011-12-14T23:05:42 http://aichallenge.org/language_profile.php?language=C%2B%2B11 :) 2011-12-14T23:05:47 thestinger: I actually considered that 2011-12-14T23:05:56 actually no, don't 2011-12-14T23:06:00 I want to be the top of that xD 2011-12-14T23:06:09 I haven't gotten back to top 10 since my resubmit 2011-12-14T23:06:24 it took a while for my bot to climb back up 2011-12-14T23:06:38 aichallenge: Alexandru Stan epsilon * rc0aa09c / website/css/layout.css : fixed min-height for the new sidebar height - http://git.io/PfPcOw 2011-12-14T23:06:44 it's kinda specialized at killing the mid to high level bots, and sucks against the top ones 2011-12-14T23:06:51 not that I intended to do that 2011-12-14T23:07:21 anyway fluxid's server is broken again too 2011-12-14T23:07:22 pguillor and a1k0n almost certainly keep me from winning USA. actually "C++" might be doable but I'd have to up my game 2011-12-14T23:07:25 http://ants.fluxid.pl/player/strcat_newdefend 2011-12-14T23:08:01 due to fluxid's misbehavior I don't really know how my current bot does against top bots other than my own previous submission 2011-12-14T23:08:20 well, now I have no idea if my new defense code works xD 2011-12-14T23:08:27 oh... I can check on the search page 2011-12-14T23:08:55 new bot beats old bot very reliably and slaughters old old bot 2011-12-14T23:09:07 nice, _newdefend climbed up from 50 mu with the broken code to 57.4 2011-12-14T23:09:38 http://ants.fluxid.pl/search?name=strcat too many branches lol. 2011-12-14T23:09:46 *** xathis has quit IRC () 2011-12-14T23:09:52 so I modified my code now so that often on smallish maps my bot can guarantee itself it knows where every enemy is 2011-12-14T23:10:05 this isn't such a boon because it doesn't react that smartly anyway :) 2011-12-14T23:11:11 what's driving me nuts at the moment is trying to understand what "in between" is for two points in a grid separated by a diagonal 2011-12-14T23:11:20 yeah, site seems to be grinding to a halt or something 2011-12-14T23:11:21 amstan: yes? 2011-12-14T23:11:26 because of manhattan distance potentially every point in the entire rectangle is "on the way" 2011-12-14T23:11:34 janzert: mysql is doing more cpu than usual 2011-12-14T23:11:56 because it's constantly generating matchups now 2011-12-14T23:12:02 at least that's my guess 2011-12-14T23:12:06 *** treeform has quit IRC (Remote host closed the connection) 2011-12-14T23:12:18 and throwing them away? 2011-12-14T23:12:22 shouldn't it stop when it fills a buffer or something? 2011-12-14T23:12:26 it's sorta just barely keeping up with the workers 2011-12-14T23:12:27 seems rather wasteful 2011-12-14T23:12:30 it's not getting ahead at all 2011-12-14T23:12:56 I could cut the number of workers down ;) 2011-12-14T23:13:14 can this be distributed? 2011-12-14T23:13:15 wow 24 workers 2011-12-14T23:13:35 some of the servers have 0% fail 2011-12-14T23:13:36 huh 2011-12-14T23:14:02 that's just for the last half hour like all the other stats listed there so varies wildly 2011-12-14T23:14:04 http://pastebin.com/XAt66mK6 2011-12-14T23:14:16 output from the matchup generator script 2011-12-14T23:14:25 as you can see it isn't really getting ahead 2011-12-14T23:14:32 matchup maker doing to many expensive queries? 2011-12-14T23:14:33 janzert: i don't get it though, before we had like 5gpm(at the beggining and beta) 2011-12-14T23:14:52 yes? 2011-12-14T23:14:52 janzert: isn't that just half the number of matchups? 2011-12-14T23:14:55 amstan: near 11 now 2011-12-14T23:15:00 right 2011-12-14T23:15:23 so why does it take more than 2 times the cpu? too many players? 2011-12-14T23:15:34 you should 'explain' the queries the matchup maker is doing to see if you need an index somewhere 2011-12-14T23:15:40 or if perhaps an index is missing for some reason 2011-12-14T23:15:42 because it's going constantly instead of just half the time? 2011-12-14T23:16:02 it uses the same cpu while generating the matchup I'm sure 2011-12-14T23:16:40 probably more with 8000 users than in the beta 2011-12-14T23:16:49 and it has gotten longer to generate matchups as the contest has progressed, I haven't looked yet to see if it is because of more users or more game history 2011-12-14T23:17:21 do you use complete game history (for things you try to balance per player) or only current version? 2011-12-14T23:17:49 I actually think it just uses past 24 hours for any game history in the matchup 2011-12-14T23:17:53 but I may be remembering wrong 2011-12-14T23:18:08 huh, before totaalnet that would have been like 3 games 2011-12-14T23:18:51 closer to 8 2011-12-14T23:19:11 some days :) 2011-12-14T23:20:08 janzert: what was the reason we couldn't use clock time again? 2011-12-14T23:20:11 sorry 2011-12-14T23:20:12 cpu time* 2011-12-14T23:20:25 nobody implemented it 2011-12-14T23:20:35 oh 2011-12-14T23:20:50 if you want it as a secondary check I'll implement it 2011-12-14T23:20:56 assuming python can get to it 2011-12-14T23:21:00 next contest at this point 2011-12-14T23:21:06 you can't get it directly 2011-12-14T23:21:13 you have to work through the /proc info 2011-12-14T23:21:19 for what we need 2011-12-14T23:21:29 and convert from jiffies to ms 2011-12-14T23:21:41 i still get surprised, the code is now open source, why don't more people hack at it? 2011-12-14T23:21:54 what's wrong with resource.getrusage? 2011-12-14T23:22:36 amstan: laziness 2011-12-14T23:22:57 i should be more inviting when people complain about this on the forums 2011-12-14T23:23:10 it has worked in the past actually 2011-12-14T23:23:22 we got a few website features with it 2011-12-14T23:23:24 yeah, I don't think people know help is wanted 2011-12-14T23:23:35 yeah 2011-12-14T23:23:41 BenJackson: hmm, looks like it doesn't give child information for children still alive 2011-12-14T23:23:51 although thank god I didn't hear about this sooner or I'd have been hacking on ant ai all summer 2011-12-14T23:24:15 amstan: if you do something like making a wishlist I bet people will start implementing stuff 2011-12-14T23:24:18 thestinger: where would you expect to find such info? 2011-12-14T23:24:18 janzert: ?? it should 2011-12-14T23:24:39 http://linux.die.net/man/2/getrusage doesn't seem to be what that page is saying 2011-12-14T23:24:58 "usage statistics for all children of the calling process that have terminated and been waited for" 2011-12-14T23:24:59 amstan: fwiw I thought "aicontest" was dead when I couldn't find the planetwars archives 2011-12-14T23:25:08 this was some time before ants started 2011-12-14T23:25:19 BenJackson: hmm, that's bad, lol 2011-12-14T23:25:27 ? they didn't change until ants launched 2011-12-14T23:25:59 ai-contest.com was exactly as the planetwars contest had left it until then 2011-12-14T23:26:01 yeah, the only thing that did was the forums 2011-12-14T23:26:18 *** dvladim has quit IRC (Read error: Operation timed out) 2011-12-14T23:26:27 i guess the website should be more dynamic and more informative regarding these things 2011-12-14T23:26:41 I never heard about tron at all 2011-12-14T23:26:47 so I thought pw was a one-off 2011-12-14T23:26:49 amstan: a link the github repo in the sidebar would be good 2011-12-14T23:26:51 when I had trouble finding it 2011-12-14T23:27:18 thestinger: any idea where? 2011-12-14T23:27:31 but anyway hopefully by next contest cpu time can be the primary measurement with wall clock a secondary 2011-12-14T23:27:45 cpu time can't be the only one btw 2011-12-14T23:28:04 amstan: maybe under Past Contests you could put an About section 2011-12-14T23:28:05 because sleep 10000 trolls? 2011-12-14T23:28:34 yeah, or other things that would have basically the same effect 2011-12-14T23:29:06 do you get IO waits for loading stuff from ram into cache? 2011-12-14T23:29:38 I think that time will still fall under cpu time for you 2011-12-14T23:30:02 yeah, because if not, that would be a significant time for bots that use a lot of ram 2011-12-14T23:30:08 pretty positive actually, but should confirm it at some point 2011-12-14T23:31:20 oh, didn't notice this page http://aichallenge.org/credits.php 2011-12-14T23:31:31 for some reason I've never noticed the 3 links at the bottom... 2011-12-14T23:32:26 there are several pages I know exist but I don't know where they're linked from 2011-12-14T23:32:30 like server stats, the map list 2011-12-14T23:32:34 @amstan / @yoden - thanks for the help... it was garbage collection. I created an object in my battle simulator that I wasn't nulling out 2011-12-14T23:32:35 Guest15667: You have no gotten any error messages recently, so here's a random one just to let you know that we care. 2011-12-14T23:32:39 BenJackson: yeah, scroll to the very bottom 2011-12-14T23:32:42 there are 3 links xD 2011-12-14T23:32:45 *** dom7b5 has joined #aichallenge 2011-12-14T23:32:59 oh, I never noticed the footer 2011-12-14T23:32:59 thestinger, janzert: what about this? http://hypertriangle.com/~alex/aichallenge/website/ renamed help to community and added github 2011-12-14T23:33:29 amstan / yoden - not sure if you got that last message I sent... thanks though... it was GC 2011-12-14T23:33:31 *** corey_ has joined #aichallenge 2011-12-14T23:33:43 *** corey_ is now known as Guest40126 2011-12-14T23:33:43 amstan: yeah, I think that's good 2011-12-14T23:33:48 Guest15667: we did, cool 2011-12-14T23:33:51 I don't think many people realize everything is open-source 2011-12-14T23:33:56 Guest15667: stuff with @ is interpreted by the bot 2011-12-14T23:34:03 cool, well thank you 2011-12-14T23:34:05 gotcha 2011-12-14T23:34:08 maybe move the FAQ link to the Ants or Getting Started header 2011-12-14T23:34:13 probably Getting Started 2011-12-14T23:34:42 wish I had this fix for this game: http://aichallenge.org/visualizer.php?game=234363&user=4165 2011-12-14T23:34:45 womp womp womp 2011-12-14T23:34:57 janzert: done 2011-12-14T23:35:15 Guest15667: resubmit? 2011-12-14T23:35:23 Guest15667: also.. change your name with /nick somethinghere 2011-12-14T23:35:48 looks good to me 2011-12-14T23:36:00 *** Guest15667 is now known as goldcaddy77 2011-12-14T23:36:05 Hello World 2011-12-14T23:36:11 cool, thanks again amstan 2011-12-14T23:36:14 amstan: is there a link back to the home page somewhere? 2011-12-14T23:36:22 I haven't noticed one 2011-12-14T23:36:23 thestinger: from github? yes 2011-12-14T23:36:34 no, I mean from random pages on the site 2011-12-14T23:36:35 *** Guest40126 has left #aichallenge 2011-12-14T23:36:42 thestinger: top of the menu 2011-12-14T23:36:52 oh, I'm silly xD 2011-12-14T23:36:52 *** dom7b5 has joined #aichallenge 2011-12-14T23:38:04 *** Antimony has quit IRC (Ping timeout: 244 seconds) 2011-12-14T23:38:42 aichallenge: Alexandru Stan epsilon * r269450d / (README.md website/menu.php): highlighted the github page from the main website - http://git.io/lF6Ueg 2011-12-14T23:40:35 goldcaddy77: more info on it: http://forums.aichallenge.org/viewtopic.php?f=25&t=1763 2011-12-14T23:41:39 *** u_ has quit IRC (Quit: u_) 2011-12-14T23:45:23 *** SMJ has quit IRC (Quit: Konversation terminated!) 2011-12-14T23:53:23 wow there must be a big gulf near 32nd 2011-12-14T23:59:32 aichallenge.org/visualizer.php?game=226671&user=13020&turn=500 2011-12-14T23:59:38 wtf I didn't watch that all the way last time 2011-12-14T23:59:58 I was wondering why the aversion to hills