2011-08-31T00:01:54 *** McLeopold has joined #aichallenge 2011-08-31T00:03:32 *** ArchMonkey has joined #aichallenge 2011-08-31T00:32:30 *** deeper-blue has quit IRC (Quit: deeper-blue) 2011-08-31T01:07:56 *** McLeopold has quit IRC (Ping timeout: 264 seconds) 2011-08-31T01:09:43 *** McLeopold has joined #aichallenge 2011-08-31T01:53:56 *** okay is now known as okayzed 2011-08-31T02:01:07 *** okayzed is now known as okay 2011-08-31T02:38:54 mleise: I think needing to write path finding and all the other stuff makes this game more ideal for an ai contest 2011-08-31T02:39:09 because it provides people with obvious things to improve on, and even working out paths in this isn't trivial 2011-08-31T02:39:43 although I am a little worried about how different the top bots are, and nobody wants first place to be decided by chance rather than a bot being clearly superior 2011-08-31T02:39:44 are/will be 2011-08-31T02:56:54 *** okay is now known as okayzed 2011-08-31T03:24:33 *** aerique has joined #aichallenge 2011-08-31T03:27:15 *** berak has joined #aichallenge 2011-08-31T03:30:25 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-08-31T03:50:09 *** Palmik has joined #aichallenge 2011-08-31T04:06:38 *** mceier has joined #aichallenge 2011-08-31T04:19:59 *** sigh has joined #aichallenge 2011-08-31T05:02:24 *** delt0r_ has quit IRC (Read error: Operation timed out) 2011-08-31T05:17:01 *** delt0r_ has joined #aichallenge 2011-08-31T05:25:56 *** mcstar has joined #aichallenge 2011-08-31T07:00:52 *** mleise has joined #aichallenge 2011-08-31T07:53:12 mmm 2011-08-31T07:53:45 If you are running a website for the TCP server... how would you want it to work? 2011-08-31T07:54:12 PHP/Phython (yuck--but hay its one way to do it) 2011-08-31T07:54:51 Application that writes to a database --then you do the web site anyway you want 2011-08-31T07:55:24 application that writes to a set of flat files --possibly even with html formating 2011-08-31T07:55:28 etc. etc etc 2011-08-31T07:57:00 since i won't run the site or have a server for it...what others would prefer is perhaps relevant 2011-08-31T08:09:56 *** mceier has quit IRC (Quit: leaving) 2011-08-31T08:15:40 *** mceier has joined #aichallenge 2011-08-31T08:19:22 hi delt0r ;) how far did you get ? 2011-08-31T08:21:19 not far... didn't have the time in the weekend in the end 2011-08-31T08:21:31 so really only just started looking at the problem 2011-08-31T08:21:53 just about finished a engine then i will finsih off a JAntManager thing 2011-08-31T08:22:01 then perhaps look at tcp 2011-08-31T08:22:10 you're trying to rewrite the whole thing ? game/server/web/gui front ? 2011-08-31T08:22:21 really not sure how folks would want to run the server 2011-08-31T08:22:33 sure why not... its not very complicated 2011-08-31T08:22:52 and that gets me to grips with game/rules 2011-08-31T08:23:00 a bot will be much more work 2011-08-31T08:23:45 the server would not be a gui thing... 2011-08-31T08:23:54 its for running on a server after all 2011-08-31T08:24:02 thats if i get too it... 2011-08-31T08:24:19 real life does interfere 2011-08-31T08:25:12 started playing with that again, too on the wekend. all in python. grrr 2011-08-31T08:25:24 yea not a fan myself 2011-08-31T08:25:34 but better than php or perl 2011-08-31T08:25:57 tried to interface the sandbox thing, and feed a list of that into the engine. 2011-08-31T08:25:59 still quite readable --in a python kind of way. 2011-08-31T08:26:05 *** UncleVasya has joined #aichallenge 2011-08-31T08:26:23 berak: are you looking at doing a server? 2011-08-31T08:26:31 works quite nice, but it's multithreaded now 2011-08-31T08:26:54 thats fine for a server... linux can handle 100-1000 of threads 2011-08-31T08:26:59 i had one 2 month ago, sort of. quite a failure, trying to repair that 2011-08-31T08:27:00 windows.... not so much 2011-08-31T08:27:26 right cool... well if you are going to have one going --i don't need to bother 2011-08-31T08:27:32 debugging multithreaded stuff on the console ... 2011-08-31T08:27:36 how are you structuring yours? 2011-08-31T08:28:05 berak: you get use to it.. done so much thread work and many core work now, that its all normal for me 2011-08-31T08:28:13 a tcp server, a web server and a mem-db shared 2011-08-31T08:28:32 java does in fact have some great features there. compared to some other langs 2011-08-31T08:29:03 right... so the tcp server writes to a db --then you plug the web pages into that 2011-08-31T08:29:20 hmm, yes. and a compiler. that's what i'm missing most 2011-08-31T08:29:57 both server start from the same process, so they can share memory 2011-08-31T08:30:17 this is in python? 2011-08-31T08:30:22 yes 2011-08-31T08:30:42 well i think you have the winner ... I wasn't going to use python 2011-08-31T08:30:52 folks will want to use yours rather than mine 2011-08-31T08:31:04 *** UncleVasya has quit IRC (Ping timeout: 252 seconds) 2011-08-31T08:31:09 *** onensora has joined #aichallenge 2011-08-31T08:31:12 (there is a "thats what she said" joke in there somewhere) 2011-08-31T08:31:39 so i guess i won't bother with a TCP server... 2011-08-31T08:31:47 well, it has to work properly first... my 1st effort failed miserably 2011-08-31T08:31:50 its all up to you berak --no pressure ;) 2011-08-31T08:32:20 well i was thinking of the tcp game server just writing out some html pages 2011-08-31T08:32:27 brutally simple 2011-08-31T08:33:07 yeah shure, just open another server socket there and go 2011-08-31T08:33:08 a single file of usernames/passwords 2011-08-31T08:33:18 that is it.. 2011-08-31T08:33:40 oh let something else serve the html...like Apache 2011-08-31T08:33:50 *** UncleVasya has joined #aichallenge 2011-08-31T08:34:10 But that would mean doing the ranking thing somewhere or not at all... that was going to be a PITA 2011-08-31T08:34:41 did you take a closer look at jskills ? 2011-08-31T08:35:39 yea 2011-08-31T08:35:43 it doesn't like [0,0,0,0] rankins, obviously 2011-08-31T08:35:50 but without a stack trace 2011-08-31T08:36:36 didn't get far 2011-08-31T08:36:42 ah, wait, just got one: 2011-08-31T08:36:52 not really writen the way i would do it 2011-08-31T08:37:04 its a c# port IIRC 2011-08-31T08:38:41 deltor, thank you for amazing JBotManager. It was very, VERY helpful tool. I see you have time for making a new one, for ants, so thanks again :) 2011-08-31T08:39:07 its getting close to done... assuming i get time to work on it 2011-08-31T08:39:09 http://pastebin.com/RcF2sdLK 2011-08-31T08:39:22 a week of real time is not really a full day on this suff 2011-08-31T08:39:47 berak: will have a look 2011-08-31T08:40:10 that is the error the main game web page has too 2011-08-31T08:40:31 look at the ranks it' 4 times 5 there. 2011-08-31T08:40:32 they bump it up to 1000 iterations and saw the error less often 2011-08-31T08:40:52 Is JBotManager open source? 2011-08-31T08:40:57 but really there should be a clear exit clause 2011-08-31T08:41:00 UncleVasya: yes 2011-08-31T08:41:14 The jar always contained the latest source 2011-08-31T08:41:37 note that 2 others submitted patches so it was not 100% my work 2011-08-31T08:42:03 ah, good to know, it's the same thing, the issues on the wiki are a bit vague there 2011-08-31T08:42:13 The licence if you need one was either BSD or public domain -- i don't recall which 2011-08-31T08:44:40 ah, so I don't need to search sources elsewhere, I just need to take it from jar somehow. I'm not familiar with java, but will lurk for it. 2011-08-31T08:44:57 its just a zip file UncleVasya 2011-08-31T08:45:15 ok 2011-08-31T08:45:51 berak: well from what i understand setting max iterations to something like 10000 then silently droping out of the loop will be fine 2011-08-31T08:46:30 the loop is to solve the system of equations approximately anyway 2011-08-31T08:46:39 so the error should be small 2011-08-31T08:46:54 without testing i can't be sure 2011-08-31T08:47:21 delt0r, can you recommend an easy decompiler ? 2011-08-31T08:47:31 jaad here, pretty old 2011-08-31T08:47:35 there are some good ones... 2011-08-31T08:47:45 but no --i almost never need to do that 2011-08-31T08:48:00 why do you need one? JSkills is on git 2011-08-31T08:48:09 github 2011-08-31T08:48:12 wait, where did you ge 2011-08-31T08:48:14 ok 2011-08-31T08:48:24 https://github.com/nsp/JSkills/blob/master/src/jskills/factorgraphs/ScheduleLoop.java 2011-08-31T08:48:34 the front door ;) 2011-08-31T08:50:59 delt0r: reading still, but it'd be really neat if the tcp server site was actually part of the contest webpage (ie. make tcp more official) 2011-08-31T08:51:20 my guess is that the delta is large because you can order people in a draw either way. So the max iteration in this case will be about how long you go "eni meni mo" 2011-08-31T08:51:24 hopefully that might encourage more people to test on tcp too 2011-08-31T08:52:03 antimatroid: since berak is doing it in python, this stands a far better chance of that than anything i would do 2011-08-31T08:52:28 berak: ^^ that :P 2011-08-31T08:53:02 ^^^°< 2011-08-31T08:53:08 berak: do you use your own DB schema or the one used on the beta site 2011-08-31T08:53:45 no, just dome dicts in mem, pickled from time to time 2011-08-31T08:53:52 right 2011-08-31T08:54:16 probably good enough for government work ;) 2011-08-31T08:54:34 forgot about python pickling 2011-08-31T08:55:32 but moving that to sqlite might be a better idea, just started with the chapest.. 2011-08-31T08:55:59 well just to some sql db right.. no need to make it specific... a 2011-08-31T08:56:11 and the games go straight to disk 2011-08-31T08:56:36 at any rate --something that works is better than something that doesn't --but would be perfect if it did 2011-08-31T08:57:43 * UncleVasya so happy to see some life here 2011-08-31T08:57:56 also since folks often have many names on a tcp server, it can and perhaps should be a little more lax with player management. The ranks are not really representative 2011-08-31T08:58:51 even what players you match with is going to be largely a function of who is connected --not your rank 2011-08-31T08:59:23 berak: how do you deal with 4 player games? do clients get to choose what they want to play? do you get a mix? 2011-08-31T08:59:26 player management. good topic. how would you go at that ? 2011-08-31T08:59:40 or is it always 4 player --and i just missed the memo 2011-08-31T08:59:46 berak: like the last one... 2011-08-31T08:59:56 i just choose a random map, then collect players for it 2011-08-31T09:00:10 fist time seen add them... first time with a password, they always need a password 2011-08-31T09:00:19 no ranges from 2 to 10 players 2011-08-31T09:00:25 right... simple is good.. 2011-08-31T09:01:27 but if the server is not busy a 10 player map my "block" since you never have enough connected? 2011-08-31T09:01:48 s/my/may 2011-08-31T09:01:57 you can never tell, how many more are going to come, true 2011-08-31T09:02:21 well a timeout to a smaller map may be a idea... again.. perhaps for laters 2011-08-31T09:02:46 * delt0r goes to get coffee 2011-08-31T09:04:03 * berak AHHHHHH 2011-08-31T09:04:39 thanks reminding me! once a month i forget to place the cup under the espresso machine... 2011-08-31T09:05:51 ever had to swap up a triple espresso ? 2011-08-31T09:11:39 * delt0r smells his wonderful coffee 2011-08-31T09:12:59 yeah, thanks for saving mine in the very last second ;) 2011-08-31T09:16:14 it would be nice if you could try and match up bots with a closer ranking 2011-08-31T09:16:24 antimatroid: on tcp? 2011-08-31T09:16:42 yeah, but if it's waiting too long then just start a game with the bots there 2011-08-31T09:16:53 otherwise it's hard for people like bocsi to test on tcp 2011-08-31T09:16:56 that really doesn't work well unless you start with waiting ques... then how long do you wait? not a simple problem to solve 2011-08-31T09:17:05 *** kaemo has joined #aichallenge 2011-08-31T09:17:13 one would need to pick a wait amount 2011-08-31T09:17:51 also often there are only a few people online... basically TCP will be a case of taking what you get 2011-08-31T09:17:56 hopfully, there's already bots playing, that will be available soon 2011-08-31T09:18:10 hell even the ranking are pretty arbitrary 2011-08-31T09:18:14 during the contest quite a few people leave bots running non stop on tcp 2011-08-31T09:18:22 yea i did too 2011-08-31T09:18:26 but its not many 2011-08-31T09:18:46 and often players have 2-5 different names etc 2011-08-31T09:19:06 the simple way is probably fine for the moment, although i'd try and make it not too difficult to add in some kine of matching stuff if needed later if possible 2011-08-31T09:19:31 the other thing is making sure players have password access to usernames for 24 hours? 2011-08-31T09:19:51 sorry if that's been covered, that's just something I thought of :p 2011-08-31T09:19:53 how would you match them ? group similar ranks ? 2011-08-31T09:20:34 yeah, how are you planning to rank them? 2011-08-31T09:20:50 antimatroid: have you done any integer programming? im interested in maximizing an expression with absolute value functions in it 2011-08-31T09:21:08 antimatroid: trueskill 2011-08-31T09:21:18 mcstar: I haven't, I've done simplex for linear programs but not integer, it's grosser too :P 2011-08-31T09:21:58 right, have you got any idea, on how to deal with the absolute value functions? 2011-08-31T09:22:23 plus, the variable is strictly bounded from both sides 2011-08-31T09:22:35 nope, sorry 2011-08-31T09:22:37 maybe there are constraints among them too 2011-08-31T09:22:41 ok 2011-08-31T09:22:52 ill just browse wiki then 2011-08-31T09:24:02 integer programing in the general case is NP hard IIRC 2011-08-31T09:24:21 still, it is solvable 2011-08-31T09:24:23 :) 2011-08-31T09:24:32 however a large class can be effeivly solved by relaxing the integer constraint 2011-08-31T09:24:39 *** deeper-blue has joined #aichallenge 2011-08-31T09:24:43 yeah, not going to work this time 2011-08-31T09:24:56 then finding the "closest" integer solution IIRC 2011-08-31T09:25:38 right..well the np hard thing is not as bad as it sounds. when dealing with 1millon varables or something 2011-08-31T09:25:47 maybe, if i scale up my variables quite much, it could be possible to use a linear programming solution, ill look into that 2011-08-31T09:25:49 they can still get pretty good answers 2011-08-31T09:27:05 it is related to my bot, so ultimately it will need to be very fast 2011-08-31T09:27:50 and the optimal solution condition can be relaxed too later, once ill get a grasp of the problem 2011-08-31T09:29:37 optimal is the enemy of good enough 2011-08-31T09:30:29 you cant know what is good enough without having a feel for the optimal 2011-08-31T09:31:11 true 2011-08-31T09:32:28 *** sigh has quit IRC (Quit: AndroIRC - Android IRC Client ( http://www.androirc.com )) 2011-08-31T09:32:47 Or maybe you can define 'good enough' as 'bad for me'. 2011-08-31T09:33:21 UncleVasya, and his brain-teasers 2011-08-31T09:33:58 oh, i left out the laughing face :D 2011-08-31T09:36:14 no no no 2011-08-31T09:36:21 good enough for me to win ;) 2011-08-31T09:41:11 in fact i doubt i will have a entry 2011-08-31T09:42:30 now your prev. line makes sense to me 2011-08-31T09:42:38 so why not? 2011-08-31T09:42:52 just leave the engine/tcp server alone, and concentrate on a bot 2011-08-31T09:43:44 i think break will be more than happy to write it on his/her own 2011-08-31T09:49:43 *** deeper-blue has quit IRC (Quit: deeper-blue) 2011-08-31T09:51:30 even then i doubt i will have time 2011-08-31T09:51:48 a tcp server + jantmanager is a lot less work that a bot 2011-08-31T09:52:00 and working on a bot is very addictive 2011-08-31T09:52:01 a bot is quite a bit of work IMO 2011-08-31T09:52:19 its not a good year for hobby time this year for me 2011-08-31T09:52:35 but meh... i like my job so it is not all bad 2011-08-31T09:52:42 i'm all good if the contest is running part way through november still, otherwise i have no spare time 2011-08-31T09:52:49 so you prefer to build the grill, but not to roast the steak, interesting 2011-08-31T09:53:09 not a good metaphor 2011-08-31T09:53:12 i want to see some games on my maps :( 2011-08-31T09:53:19 steak is easy once you have a grill 2011-08-31T09:53:26 delt0r: you've seen my maze maps yeah? 2011-08-31T09:53:27 delt0r: best i had, sorry :D 2011-08-31T09:53:35 antimatroid: no 2011-08-31T09:53:50 http://ants.zeroviz.us/rLPKO/ 2011-08-31T09:53:53 they look like that :) 2011-08-31T09:53:58 mcstar: should involve cars or library's of congress 2011-08-31T09:54:18 ill think about it 2011-08-31T09:54:30 lol 2011-08-31T09:54:45 antimatroid: freaky --not sure many would like it however 2011-08-31T09:55:11 most simple bots would keep to themselves since they won't find their way out 2011-08-31T09:55:18 http://ants.zeroviz.us/bLiMI/ that's a different kind of symmetry 2011-08-31T09:55:30 personally.. would be interesting 2011-08-31T09:55:44 thats a very very good map, for advanced bots 2011-08-31T09:56:09 it'll make path finding more important 2011-08-31T09:56:20 path finding is easy peasy 2011-08-31T09:56:36 http://www.reddit.com/r/Algebra/comments/j27dt/generating_tessellations_on_a_discrete_torus/ 2011-08-31T09:56:42 i think its not that easy for a bot to recognize it is in a maze 2011-08-31T09:56:44 that post explains how to make the maps 2011-08-31T09:56:45 A* and Div whatever that guys name is are pretty simple 2011-08-31T09:56:57 dijkstra 2011-08-31T09:56:57 delt0r: it depends, my bot does A* with multiple sources and targets 2011-08-31T09:57:46 with h(x) := min_* mdist(x, *) (* are food locations) for example 2011-08-31T09:57:52 again not such a hard extension. Information is still propagated over the graph. I would not expect it to be much more code 2011-08-31T09:58:05 and i will re use ants when collecting food, including ants i expect to spawn from collecting food 2011-08-31T09:58:35 using information about when they'll be available while searching 2011-08-31T09:59:05 it's not that complicated, but it's good for beginners who haven't played with path finding before and offers quite a few options for people 2011-08-31T09:59:20 the real complication comes with macro strategy and battling enemies imo 2011-08-31T09:59:27 path finding is the best place to start though 2011-08-31T09:59:31 *** sigh has joined #aichallenge 2011-08-31T09:59:31 yea-- 2011-08-31T09:59:54 keep the ground you have, or be aggressive and get more ground 2011-08-31T10:00:07 assemble hunting parties 2011-08-31T10:00:12 parties 2011-08-31T10:00:13 hopefully the best bot will be agressive 2011-08-31T10:00:14 yea 2011-08-31T10:00:20 otherwise i see a fairly boring contest :P 2011-08-31T10:00:50 if you get more of the map, then you have more food 2011-08-31T10:01:17 yeah, i expect the better bots to be defensively agressive 2011-08-31T10:01:21 the best bot will utilize deception, ambush, suicide and luck too 2011-08-31T10:01:22 you can just sit tight if survival is the goal 2011-08-31T10:01:26 i'm just not sure how that will go when they play each other 2011-08-31T10:01:44 deception and ambushing are things i would like to see 2011-08-31T10:01:59 (i want a bot that brings enemies closer together) 2011-08-31T10:02:21 you don't want to do that... you want to spread their line --surly 2011-08-31T10:02:28 surely 2011-08-31T10:02:43 i want them to battle against each other, not with me 2011-08-31T10:02:44 you can use a single ant to lead enemies away from your "swarms", or in particular path so it's surrounded by your "swarms" which would look sweet 2011-08-31T10:03:06 then swoop in from outside view 2011-08-31T10:03:07 and dont call me Surely 2011-08-31T10:03:14 :D 2011-08-31T10:05:59 pfff, people dont like humor anymore 2011-08-31T10:06:10 if a bot times out, does the orders it sent before timing out still get carried out? 2011-08-31T10:06:36 delt0r: hmmmm :P i hope not 2011-08-31T10:07:09 you mean the bot times out in between sending output? 2011-08-31T10:07:14 shure, that was one turn before 2011-08-31T10:07:16 the only reason their ants stay in the game is to try and minimise teritory advantage to neighbouring bots 2011-08-31T10:07:40 yea --its sending its orders... then the time is up. So now its send orders, but has timed out. 2011-08-31T10:07:40 although we can make 5 player games where you can remove any subset of the players and the game remains symmetric 2011-08-31T10:08:15 the engine waits for the bot's go signal 2011-08-31T10:08:21 ie it didn't get to the go 2011-08-31T10:08:28 so if it never came, the orders cannot be carried out 2011-08-31T10:08:41 yea but i currently process orders as the arrive 2011-08-31T10:08:55 you shouldnt, if you think it matters 2011-08-31T10:09:01 that is not implied in the game definition. 2011-08-31T10:09:04 i said, we can make 5 player maps .. does that not blow anyone elses mind? 2011-08-31T10:09:11 no 2011-08-31T10:09:12 in fact nothing is really implied 2011-08-31T10:09:14 sorry 2011-08-31T10:09:28 talk to Penrose, he will be even calmer 2011-08-31T10:09:35 antimatroid: I support 256 players in plantwars 2011-08-31T10:09:51 good luck running that on the servers 2011-08-31T10:09:58 i'd be interested to see a tcp server doing that 2011-08-31T10:10:02 plantwars? 2011-08-31T10:10:14 random appearance :P 2011-08-31T10:10:31 16 or so players should no be a problem 2011-08-31T10:10:33 oh, he is here, with a parabolic microphone, always lsitening 2011-08-31T10:10:48 ants can handle up to 26 2011-08-31T10:10:53 *** sigh has quit IRC (Read error: Connection reset by peer) 2011-08-31T10:11:06 why is their a limit 2011-08-31T10:11:14 map file format 2011-08-31T10:11:17 my engine can do whatever you have ram for 2011-08-31T10:11:22 or 2^31 2011-08-31T10:11:25 uses a-z for the starting positions 2011-08-31T10:11:27 a-b,A-Z? 2011-08-31T10:11:30 which ever comes first 2011-08-31T10:11:32 it could be fixed with a few changes I think 2011-08-31T10:11:33 thats 52 2011-08-31T10:11:45 just pass the map without ant info and pass ants like parameters 2011-08-31T10:11:46 stupid ascii map format 2011-08-31T10:11:51 yeah 2011-08-31T10:12:01 that what i do for my map gen... 2011-08-31T10:12:03 *** mathis has joined #aichallenge 2011-08-31T10:12:26 but at the same time I like being able to open a map file and see what the map looks like 2011-08-31T10:12:58 antimatroid: you dont really need to designate different starting positions with different letters 2011-08-31T10:13:15 well that could have been done with say.... ant start positions... spaw point if you will. Then let them get picked randomly.. or have a start ant list at the end 2011-08-31T10:13:29 yes i do, because then you can have maps with more than one starting ant per player 2011-08-31T10:13:31 s/spaw/span 2011-08-31T10:13:50 :S do you want that? 2011-08-31T10:13:50 you can with this too 2011-08-31T10:13:51 which is at least vital for debugging purposes 2011-08-31T10:14:23 internally i have a map + a list of ants and food for the intial start 2011-08-31T10:15:12 yeah my bot has a vector of locations for myAnts, enemyAnts and food 2011-08-31T10:15:29 i also need oldFood to track food information outside of view 2011-08-31T10:15:43 *** FireFly has joined #aichallenge 2011-08-31T10:16:04 well what do you know.. i do ignore commands already sent if the bot times out 2011-08-31T10:16:36 seems the correct thing to do yeah? 2011-08-31T10:17:12 i wouldn't be opposed to setting a timed out bots score to zero, although i agree with mcleopold in that we're testing peoples ability to write ai and not program 2011-08-31T10:17:28 tbh --i think the spec is not clear --but its what the current github does too 2011-08-31T10:17:53 so that seems to be trumps 2011-08-31T10:19:12 antimatroid: i see you are practicing touc-typing, hence not using tab 2011-08-31T10:19:15 h 2011-08-31T10:19:39 mcstar: what? 2011-08-31T10:19:56 you didnt use tab-completion for mcleo's name 2011-08-31T10:20:03 its just too long 2011-08-31T10:23:31 mcstar: i am really bad at ever using tab completion for names :P 2011-08-31T10:23:50 delt0r: I added a paragraph to the end of the "endbot conditions" of the specs page 2011-08-31T10:24:07 "If a bot crashes or times out on a given turn then none of the moves received from that bot will be executed for that turn, nor will it receive any points for that turn." 2011-08-31T10:25:58 hm, i really think this is implied, in the input format section however 2011-08-31T10:26:44 mcstar: clearly you haven't been around during previous contests :P 2011-08-31T10:27:01 it's worth making everything as clear as possible 2011-08-31T10:27:32 ok, i wont argue with that 2011-08-31T10:32:18 *** Accoun has quit IRC () 2011-08-31T10:32:31 *** deeper-blue has joined #aichallenge 2011-08-31T10:33:30 deeper-blue: :) 2011-08-31T10:33:49 hiho 2011-08-31T10:34:01 what happened to you last contest? 2011-08-31T10:34:11 *** onensora has quit IRC (Ping timeout: 240 seconds) 2011-08-31T10:34:13 *cough* 2011-08-31T10:34:15 sorry about messing your bot up :P 2011-08-31T10:34:18 well 2011-08-31T10:34:56 I was visiting my gf and she was severly sick so I didn't work on my bot and turns out someone increased the max number of planets that can exist 2011-08-31T10:35:02 and I had a hard coded number in my code 2011-08-31T10:35:03 :D 2011-08-31T10:35:11 we broke your bot when we added maps with more than 23 planets, yours is the last I know of that didn't get fixed :( 2011-08-31T10:35:13 sorry about that :) 2011-08-31T10:35:19 i hope your gf is better now? 2011-08-31T10:35:23 yes yes 2011-08-31T10:35:25 it's fine 2011-08-31T10:35:34 it's hilarious to look at my bot 2011-08-31T10:36:10 *** Accoun has joined #aichallenge 2011-08-31T10:36:12 all those crashes :) 2011-08-31T10:37:41 have you played around with ants much? 2011-08-31T10:39:34 2 people from sri lanka in the top 10, what is this 2011-08-31T10:39:41 where's eashoka when I need him 2011-08-31T10:40:28 hehe 2011-08-31T10:40:38 I'm working on some ideas for ants 2011-08-31T10:40:50 my ant bot so far is extremely stupid 2011-08-31T10:40:58 *** aerique has quit IRC (Quit: ...) 2011-08-31T10:41:38 fat from doing anything interesting but I'm working on the frameworks :) 2011-08-31T10:42:19 yeah, i think i have a fairly solid framework for mine 2011-08-31T10:42:33 it's a beefed up version of the c++ starter bot i wrote 2011-08-31T10:42:52 then all my bot is really doing is collecting food/enemies and exploring 2011-08-31T10:43:02 hehe, I used the c++ starter bot first but then rewrote everything in C to fit my style 2011-08-31T10:43:17 i haven't done anything for battles and don't have time again for bot writing until november 2011-08-31T10:43:35 same here 2011-08-31T10:43:37 didn't like the c++ bot style? :P 2011-08-31T10:43:46 i'm not a c++ person 2011-08-31T10:44:03 C all the way ;) 2011-08-31T10:44:08 even though it hurts :/ 2011-08-31T10:44:09 i love c++, overload ALL the operators 2011-08-31T10:45:05 I'm not sure if I should try to program in all the 'smarts' or if I should pursue a fully algorithmic/optimization path with this challenge 2011-08-31T10:45:11 *** kaemo has quit IRC (Ping timeout: 240 seconds) 2011-08-31T10:45:21 what do you mean? 2011-08-31T10:46:35 i use multi source and target A* for collecting food/enemies, bfs for exploring and would intend to use some kind of game tree for battles 2011-08-31T10:46:36 one way is to try to teach the bot everything from how to look for food and how to fight - and one approach is to just give an optimization framework all the rules and let it try to figure out the smartest move 2011-08-31T10:46:55 that falls into the first categorie 2011-08-31T10:47:02 and I think it's the most feasible one :) 2011-08-31T10:47:22 i'm a game theory fan, so i want to use that when possible :P 2011-08-31T10:47:25 I think there're too many variables to optimize 2011-08-31T10:47:40 yeah, I tried for a long long time to solve planet wars 2011-08-31T10:47:42 well at least for a full bruteforce attempt :D 2011-08-31T10:47:53 it's an interesting problem but a ridiculously hard one, I had some crazy ideas :P 2011-08-31T10:48:13 I like how different it is from the planetwars thing 2011-08-31T10:48:31 it offers more depth for aiey things to work on 2011-08-31T10:48:40 it's not as cool to watch though :\ 2011-08-31T10:48:54 i think it will become more interesting to watch! 2011-08-31T10:48:58 when bots get better 2011-08-31T10:49:01 i still think tron is my favourite, it was so simple and neat 2011-08-31T10:49:19 ants is super complex 2011-08-31T10:49:30 i wanted it to be more complex :P 2011-08-31T10:49:34 :) 2011-08-31T10:49:46 when I heard ants 2011-08-31T10:49:56