2011-11-14T00:01:02 Any advice on programmatically finding a choke point (to make a wall of ants)? 2011-11-14T00:01:08 yeah, it might be a little better if we reproduce that tar file ourselves from source 2011-11-14T00:02:12 NightExcessive: increase the thickness of all the walls, until your path gets blocked 2011-11-14T00:02:13 NightExcessive: min-cut 2011-11-14T00:04:27 janzert: also, this: http://forums.aichallenge.org/viewtopic.php?f=25&t=1872 2011-11-14T00:04:42 only slightly related but I think we're well past the point of diminishing returns on adding 2011-11-14T00:04:58 adding languages? 2011-11-14T00:05:05 janzert: hmm? 2011-11-14T00:05:18 oops, yes adding languages 2011-11-14T00:09:14 of the 28 supported languages 26 are being used http://pastebin.com/ZDVsE1gB 2011-11-14T00:10:07 woot! python 2011-11-14T00:10:18 ugh.. VB is more than haskell 2011-11-14T00:10:34 ha. 2011-11-14T00:10:45 amstan: with 70 haskell bots, I'm hoping for the best! 2011-11-14T00:11:00 so.. C++11... i don't like that name 2011-11-14T00:11:09 How is OpenJDK 6 with speed in comparison to C++? (I know the JRE 7 is closing on C) 2011-11-14T00:11:18 Groovy and Scheme are the two not being used by anyone 2011-11-14T00:11:19 when i first heard about C++0x i really liked it, because i tought 0x was meant to represent the hex 2011-11-14T00:11:29 janzert: scheme starter package is barely there 2011-11-14T00:11:39 i think 2011-11-14T00:11:49 and groovy not at all of course 2011-11-14T00:11:56 yeah 2011-11-14T00:12:22 Groovy is supported because someone kept wanting it in planetwars but then never used it there either :/ 2011-11-14T00:12:34 oh well 2011-11-14T00:12:49 Groovy is a terrible language 2011-11-14T00:12:53 * bmh has opinions 2011-11-14T00:13:32 janzert: is it really a problem to support them? like it's just a few lines in the compiler and setup script 2011-11-14T00:13:50 is rexx still supported?! 2011-11-14T00:13:52 unless it's not in ubuntu 2011-11-14T00:13:53 a1k0n: lol! 2011-11-14T00:13:55 if no one uses it at least :) 2011-11-14T00:14:15 a1k0n: not by us 2011-11-14T00:14:16 oh man.. i forgot about the rexx guy 2011-11-14T00:14:39 he really wanted to use rexx, but didn't know how to make a starter package 2011-11-14T00:14:50 and this was in tron, where I/O was really easy 2011-11-14T00:14:58 the problem with supporting multiple languages is each language advocate always wants a new version or different options :/ 2011-11-14T00:15:08 ahh, I remember that 2011-11-14T00:15:20 If C# got the optimizer flag I'd definitely use it :) 2011-11-14T00:15:38 the latter problem should go away once we put the build process in the user's hands 2011-11-14T00:16:11 man, i don't wanna wait 547 minutes for a game 2011-11-14T00:16:42 whatever happened to the <(''<) anyway 2011-11-14T00:17:10 got removed since it didn't work quite right and mostly confused people 2011-11-14T00:17:23 I think McLeopold still had plans to fix it up and put it back in though 2011-11-14T00:17:40 it also confused a bunch of indexing engines and made them break the memcached stuff 2011-11-14T00:17:52 that was hilarious 2011-11-14T00:18:02 heh, that was the cause? 2011-11-14T00:18:30 janzert: the kirbys had to be htmlentitied 2011-11-14T00:18:32 hahah 2011-11-14T00:18:55 yeah, I remember fixing that at one point 2011-11-14T00:18:57 the indexing engines were parsing the links wrong, so /rankings.php/forums?page=1 appeared 2011-11-14T00:19:13 but then fixing the kirbys didn't fix it in their cache 2011-11-14T00:19:18 nice 2011-11-14T00:19:19 so they kept requesting that page 2011-11-14T00:19:49 the pagination script outputs a relative url, so forums?page=2 came into existence 2011-11-14T00:19:53 then the memcached remembered that 2011-11-14T00:19:53 yeah, I didn't realize that's how they got the bad link though 2011-11-14T00:19:57 and it showed it to the users 2011-11-14T00:20:24 while fixing that, i restarted memcached, which temporarly broke the frontpage game display 2011-11-14T00:20:41 while looking for a manual update on that i crashed the mysql server, and then the workers crashed 2011-11-14T00:20:53 all from the one kirby presented to the wrong bing bot! 2011-11-14T00:20:58 error cascade :) 2011-11-14T00:22:52 janzert: we should probably write something about the wall tile vs cpu time 2011-11-14T00:22:53 Are there debugger commands in the main tools package? (from git) 2011-11-14T00:22:54 and why 2011-11-14T00:23:05 NightExcessive: what? 2011-11-14T00:23:17 http://forums.aichallenge.org/viewtopic.php?f=25&t=1633 2011-11-14T00:23:41 NightExcessive: i don't think so 2011-11-14T00:23:58 Darn. I have to browse the forks to figure out which ones have the best debugging commands then. :< 2011-11-14T00:24:58 NightExcessive: this one's definitelly not it: https://github.com/j-h-a/aichallenge/ 2011-11-14T00:25:07 it's pretty much a perfect copy 2011-11-14T00:25:15 let me browse the branches 2011-11-14T00:25:22 in network 2011-11-14T00:25:47 *** Kommander has quit IRC () 2011-11-14T00:25:51 jab_bott seems to be the one who has done the most work in that area 2011-11-14T00:26:06 from my skimming of the thread as it has developed 2011-11-14T00:26:26 oh.. it is j-h-a then 2011-11-14T00:26:36 just the vis_overlay branch 2011-11-14T00:26:49 https://github.com/j-h-a/aichallenge/tree/vis_overlay 2011-11-14T00:27:30 *** TheLinker has quit IRC (Quit: Bye) 2011-11-14T00:28:08 tc pants .com: http://tcpants.com/ 2011-11-14T00:28:10 lol 2011-11-14T00:28:31 I always see "pants" in there too 2011-11-14T00:29:51 is figuring out what ant should move to which square actually NP-hard? (a1k0n) 2011-11-14T00:30:39 woah! 4 workers up now 2011-11-14T00:31:38 well.. i should really go to bed 2011-11-14T00:31:56 this contest got to the point where just reading forum posts and doing small changes is addictive 2011-11-14T00:34:20 janzert: using csc for myisampacking the pw stuff is a go 2011-11-14T00:34:49 bmh: I'm not sure it's even in NP 2011-11-14T00:35:08 remember NP means if someone dropped a solution out of the sky you could verify it was correct in polynomial time 2011-11-14T00:35:20 point 2011-11-14T00:35:21 so if I gave you a bunch of ant moves, could you even tell me if they wre correct? 2011-11-14T00:36:07 How about this formulation: Assign a score to each point on the board. Find the arrangement of ants that maximizes the total score. 2011-11-14T00:36:26 amstan: great 2011-11-14T00:36:26 isn't figuring out where to move the whole game? 2011-11-14T00:36:32 there's not much else you can do 2011-11-14T00:36:35 besides timing out 2011-11-14T00:36:51 amstan: I'm thinking of the case where you have several ants jostling for a square 2011-11-14T00:36:58 bmh: if you can make that score assignment cheap enough there are loads of ways to use that to write a bot 2011-11-14T00:37:11 that's actually what my current bot does 2011-11-14T00:37:21 mine too 2011-11-14T00:37:21 it's basically a giant cosntratin optimizatin solver 2011-11-14T00:37:44 Antimony: where's your profile? I like guessing how the bots work from watching the replays 2011-11-14T00:37:49 Parasprites 2011-11-14T00:37:58 *** datachomper has quit IRC (Quit: Leaving.) 2011-11-14T00:38:02 tell me what you think 2011-11-14T00:39:30 your feeding and exploration looks solid but you have no combat 2011-11-14T00:39:57 yep 2011-11-14T00:39:59 hm, mazes a weakness 2011-11-14T00:40:17 well exploration isn't too great either 2011-11-14T00:40:58 you don't clump up or waste many moves in open spaces that I saw 2011-11-14T00:43:03 yeah, I have anticlumping. Scores are decreased for being near other ants, and distances increase faster when propogating thorugh space near an ant 2011-11-14T00:43:27 *** Ashoka has joined #aichallenge 2011-11-14T00:44:55 one other thing I still need to do is keep track of stored food so I can tell ants to move off the mound 2011-11-14T00:45:00 but combat is a higher priority 2011-11-14T00:45:36 why don't you always keep the mound open? 2011-11-14T00:45:51 http://aichallenge.org/visualizer.php?game=91058&user=3845 2011-11-14T00:46:00 curious how both bots have very similar issues escaping spawn 2011-11-14T00:47:27 *** Palmik has joined #aichallenge 2011-11-14T00:48:00 so this weekend I had hoped to come up with the idea to take my bot to the next level 2011-11-14T00:48:15 buut despite thinking interesting thoughts I didn't really do that 2011-11-14T00:48:30 BenJackson: symmetry detector? :) 2011-11-14T00:48:32 I had same plan. lol 2011-11-14T00:48:44 I did figure out a lot about symmetry and write some interesting code 2011-11-14T00:49:19 I can find hills without seeing them 2011-11-14T00:49:42 how do you find it useful? I always suspect raiding far-away hill is a risky strategy 2011-11-14T00:49:44 however in my tests that wasn't necessarily a good thing: the rest of my bot tuning then tended to attack hills it couldn't really reach 2011-11-14T00:49:48 what's the magic? Translational symmetry seems easy 2011-11-14T00:50:16 bmh: seemed easier until I looked into it :) 2011-11-14T00:50:48 analyst74: the replay that got me interested in it was the same map as linked above 2011-11-14T00:51:12 The reason I don't always keep the mound clear is that it's sometimes beneficial to not have it clear 2011-11-14T00:51:16 it's so far (topologically) between the two enemy hills that there's often a big lull while both bots eke out that last bit of exploration 2011-11-14T00:51:17 like in a defence situation 2011-11-14T00:51:37 so it could be a big advantage on that map to know your enemy's location sooner 2011-11-14T00:51:50 or possibly in any 1-v-1 2011-11-14T00:52:27 that's most effective against people with no hill-defending logic 2011-11-14T00:53:00 I think the reason my ants got stuck in that maze game 2011-11-14T00:53:03 if you can determine symmetry, you can race to the bottlenecks 2011-11-14T00:53:15 is because I attempted goal occlusion by increasing distancse when propgated past an ant 2011-11-14T00:53:24 I haven't really done anything about bottlenecks 2011-11-14T00:53:30 so the distances seen near the mound are essentially noise 2011-11-14T00:53:48 because you have to go past so many ants to get to the nearest food 2011-11-14T00:56:57 after that long march, if you don't win, you'd lose soon after 2011-11-14T00:57:02 I wonder if I should try geting rid of the goal occlusion thing 2011-11-14T00:57:30 you'd probably end up with something like my first bot 2011-11-14T00:57:36 which has very snake-like behavior 2011-11-14T00:57:41 sending too many ants after each food 2011-11-14T00:58:02 i haven't had much time to work on my bot :( 2011-11-14T00:58:18 i'm still halfway through my pathfinding code 2011-11-14T00:58:49 even if I don't do any better than I am now I still learned some interesting things 2011-11-14T00:59:47 when does the winter competition start? 2011-11-14T00:59:50 very true, and if you can think of other non-obvious ways to use that knowledge, you'd be sitting on gold. :) 2011-11-14T01:00:07 isn't it winter already?? 2011-11-14T01:00:08 cyphase: someone was suggesting there'd probably be a 3-6 month gap 2011-11-14T01:00:25 yearly would be best... 2011-11-14T01:00:43 I'm sorry I missed tron 2011-11-14T01:00:46 I didn't hear about it 2011-11-14T01:01:00 I heard about planetwars and my immediate reaction was that I had no idea how I'd play that as a human 2011-11-14T01:01:13 i miight have heard about this before, but i didnt look into it enough to remember it 2011-11-14T01:01:29 there's still over a month left 2011-11-14T01:01:31 ug 2011-11-14T01:01:34 yea 2011-11-14T01:01:44 I hate how calling erase on an invalid iterator causes an infinite loop 2011-11-14T01:01:51 I've had that hpapen to me twice now 2011-11-14T01:02:02 there's an idiom for that 2011-11-14T01:02:13 eg myset.erase(it++) 2011-11-14T01:02:18 I think I'm just going to try to solve every possible position on a cluster and then have my bot look up the optimal behavior ;) 2011-11-14T01:02:31 it's somewhat awkward because you have to structure the loop so it's if X erase+increment else increment 2011-11-14T01:02:43 bmh, good luck with that 2011-11-14T01:02:57 yeah, lots of people want to do reinforcement learning or similar here 2011-11-14T01:03:00 cyphase: the universe will live long enough for that to work, right? 2011-11-14T01:03:02 the state space is enormous 2011-11-14T01:03:25 I thought at first i could do combat based on pattern matching some local area 2011-11-14T01:03:31 I was makign a copy of the iteraotr and then incrementing 2011-11-14T01:03:33 bmh, let's ask Deep Thought 2011-11-14T01:03:41 but I accidently called erase on the wrong iterator 2011-11-14T01:03:51 cyphase: how do you reverse entropy? 2011-11-14T01:03:53 even with rotational symmetry I wound up with something like 2^80 map/bot configurations inside active combat radius 2011-11-14T01:04:45 ouch! 2011-11-14T01:05:31 bmh, cat /dev/zero > /dev/universe/entropy 2011-11-14T01:05:40 so how do you detect symmetries? The best I could think of is to hash each visible NxN block, but I haven't tried yet 2011-11-14T01:05:51 and look for duplicate hash values 2011-11-14T01:05:54 BenJackson: are you ben.com? 2011-11-14T01:05:58 yes 2011-11-14T01:06:11 awesome, 3-letter domain 2011-11-14T01:06:15 (your cats are cute) 2011-11-14T01:06:20 Antimony: the actual game engine does it brute force 2011-11-14T01:06:36 bmh: I wisely quit photographing them after they grew into monsters 2011-11-14T01:06:49 the orange one is 14lbs and the orange&white one is 12.5 as of last week 2011-11-14T01:06:49 *** retybok_ has joined #aichallenge 2011-11-14T01:07:31 BenJackson, you got in early :P 2011-11-14T01:07:37 ben.com makes me think of Majora's Mask 2011-11-14T01:07:59 yep, when I registered ben.com you did it by ftp'ing a text file, filling in areas of the file and emailing it in 2011-11-14T01:08:08 they added new domains on tues and thurs at 6pm 2011-11-14T01:08:08 lol 2011-11-14T01:08:13 my shortest domain is 'ch4n.org' 2011-11-14T01:11:21 maze_04p_01 is so annoying for testing 2011-11-14T01:11:28 player 1 always loses 2011-11-14T01:11:39 which is usually the slot I put my testing bot in 2011-11-14T01:11:52 *** twymer has quit IRC (Ping timeout: 244 seconds) 2011-11-14T01:12:12 I like it for testing because you can tweak the difficulty slightly 2011-11-14T01:12:19 how? 2011-11-14T01:12:34 eg you put your "old" bot across from an easy opponent and another copy across from yoru "new" bot 2011-11-14T01:12:42 the one with the easy opponent gets a head start 2011-11-14T01:13:06 what I don't get is why the bot I put as the first player always loses 2011-11-14T01:13:12 hm, gvies me an idea, I should add an option to delay starting or run at half speed or something so I can artificially increase the difficulty 2011-11-14T01:13:47 the pattern I see on that map is that one of each pair of neighbors wins and the winners of those fight each other 2011-11-14T01:13:54 if everyone is closely balanced 2011-11-14T01:13:57 *** datachomper has joined #aichallenge 2011-11-14T01:14:05 *** dr- has quit IRC (Read error: Connection reset by peer) 2011-11-14T01:14:08 but the ones going right at the start havea uge advantage for some reason 2011-11-14T01:14:15 maybe it has to do with teh order you consider moves in 2011-11-14T01:14:37 or ordering of nodes when doing a search or something 2011-11-14T01:15:07 *** datachomper has quit IRC (Client Quit) 2011-11-14T01:15:29 *** retybok_ has quit IRC (Ping timeout: 244 seconds) 2011-11-14T01:16:42 most people do search N-E-S-W due to the default game logic 2011-11-14T01:17:21 I've tried mixing it up in my bots 2011-11-14T01:18:09 one of the methods I tried leads to oscilating ants 2011-11-14T01:18:24 that was in the previous version of my bot, the one that made squares all the tiem 2011-11-14T01:18:50 I haven't seen any bots travelling in 2x2 blocks lately 2011-11-14T01:20:22 BenJackson: 2x2 warbands, 'eh? 2011-11-14T01:20:34 my old bot just made a 5x5 square around the mound 2011-11-14T01:20:39 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-11-14T01:20:55 it definitely shows the power of CSP, but it was a terrible strategy 2011-11-14T01:21:01 let me see if I can dig up a replay 2011-11-14T01:21:04 my current bot chases food and bumps into itself and dies 2011-11-14T01:21:28 *** ikaros has joined #aichallenge 2011-11-14T01:22:18 wow your site is way cooler than mine 2011-11-14T01:23:33 Here's my SquareBot http://aichallenge.org/visualizer.php?game=48944&user=3845 2011-11-14T01:23:41 heh, you were even doing PIC programming 2011-11-14T01:24:35 someone ought to make their bot play Conway's game of life :) 2011-11-14T01:24:45 Antimony: that's hilarious 2011-11-14T01:24:49 heh, send out gliders from the base 2011-11-14T01:25:01 i wonder if that's even achievable 2011-11-14T01:25:49 one of the ants has to move diagonally 2011-11-14T01:27:43 *** Harpyon has joined #aichallenge 2011-11-14T01:32:01 *** Antimony has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-11-14T01:32:38 someone actually sent me comments on some ancient PIC mul/div macro I posted a lnog time ago 2011-11-14T01:32:42 apparently it has an obscure bug 2011-11-14T01:33:22 *** orange8 has quit IRC (Quit: Page closed) 2011-11-14T01:34:06 *** Fandekasp has quit IRC (Read error: Operation timed out) 2011-11-14T01:34:21 *** Fandekasp has joined #aichallenge 2011-11-14T01:41:10 *** orange002 has joined #aichallenge 2011-11-14T01:41:20 *** Harpyon has quit IRC (Quit: Computer has gone to sleep.) 2011-11-14T01:43:06 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 252 seconds) 2011-11-14T01:52:20 *** imp5imp5 has joined #aichallenge 2011-11-14T01:53:34 *** Conorach has quit IRC (Ping timeout: 260 seconds) 2011-11-14T01:54:53 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-14T01:56:07 *** pcsin has joined #aichallenge 2011-11-14T01:56:48 *** Odyssey has joined #aichallenge 2011-11-14T01:57:02 *** pcsin has quit IRC (Client Quit) 2011-11-14T01:58:28 *finally* my bot does something 2011-11-14T01:59:54 *** ChrisH has joined #aichallenge 2011-11-14T02:01:02 bmh: http://dl.dropbox.com/u/3166522/screenshots/2011/11/Screenshot-2011-11-14_02.00.45.png <- All your fault! :) 2011-11-14T02:01:28 NightExcessive: sorry -- you wrote a visualizer or something? 2011-11-14T02:01:31 or your ants are being dumb? 2011-11-14T02:01:40 I used the visualizer someone wrote 2011-11-14T02:01:49 To visualize my very simply collaborative diffusion. 2011-11-14T02:01:54 simple, even 2011-11-14T02:03:11 *** yoden has quit IRC (Quit: Leaving.) 2011-11-14T02:07:19 how do you make your bots not cluster up? 2011-11-14T02:08:18 *** Odyssey has quit IRC (Quit: Page closed) 2011-11-14T02:11:45 *** Flort has joined #aichallenge 2011-11-14T02:15:01 *** Mooloo has quit IRC (Ping timeout: 240 seconds) 2011-11-14T02:16:11 bmh: All of my bots give off a very local, strong "negative scent"... so once they get close to food the food gets cancelled out by it. 2011-11-14T02:18:10 Anyway, I'm off. G'night all. 2011-11-14T02:18:13 *** NightExcessive has quit IRC (Quit: Page closed) 2011-11-14T02:19:11 *** Garf has joined #aichallenge 2011-11-14T02:19:15 *** ChrisH has quit IRC (Quit: Page closed) 2011-11-14T02:21:51 *** delt0r_ has quit IRC (Ping timeout: 248 seconds) 2011-11-14T02:27:45 *** kilae has joined #aichallenge 2011-11-14T02:30:00 *** ikaros has joined #aichallenge 2011-11-14T02:35:06 *** delt0r_ has joined #aichallenge 2011-11-14T02:41:27 *** Cyb has joined #aichallenge 2011-11-14T02:50:09 *** bmh has quit IRC (Quit: bmh) 2011-11-14T02:56:14 *** Cyb has quit IRC (Ping timeout: 244 seconds) 2011-11-14T02:58:46 *** pairofdice has joined #aichallenge 2011-11-14T03:06:01 *** knyppeldynan has joined #aichallenge 2011-11-14T03:18:20 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-11-14T03:20:56 @later amstan did you uninstall apparmor? Anyway it seems it was the apparmor profile restricting the file access. apparmor was seemingly uninstalled but still enforcing the profiles. Fixing the profile and rebooting fixed the problem. I also reinstalled apparmor (restarted again for good measure as well) and it is still working 2011-11-14T03:20:57 janzert: User error, it's not my fault. 2011-11-14T03:21:04 *** HaraKiri has joined #aichallenge 2011-11-14T03:24:43 *** mj41 has joined #aichallenge 2011-11-14T03:25:08 @later tell amstan did you uninstall apparmor? Anyway it seems it was the apparmor profile restricting the file access. apparmor was seemingly uninstalled but still enforcing the profiles. Fixing the profile and rebooting fixed the problem. I also reinstalled apparmor (restarted again for good measure as well) and it is still working 2011-11-14T03:25:08 janzert: OK 2011-11-14T03:25:44 *** Blkt has joined #aichallenge 2011-11-14T03:26:13 *** sigh has joined #aichallenge 2011-11-14T03:29:14 *** ronnyk_ has joined #aichallenge 2011-11-14T03:33:06 *** netantho has quit IRC (Remote host closed the connection) 2011-11-14T03:34:07 *** djr_ has quit IRC (Quit: Leaving) 2011-11-14T03:34:22 *** orange002 has quit IRC (Ping timeout: 265 seconds) 2011-11-14T03:40:54 good morning everyone 2011-11-14T03:43:03 *** mviel has joined #aichallenge 2011-11-14T03:56:14 Blkt : GM! 2011-11-14T03:56:47 janzert: u there ? 2011-11-14T04:01:28 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-14T04:02:47 *** Alexer has quit IRC (Ping timeout: 276 seconds) 2011-11-14T04:04:03 *** Alexer has joined #aichallenge 2011-11-14T04:13:35 *** Alexer has quit IRC (Ping timeout: 256 seconds) 2011-11-14T04:13:42 *** Alexer has joined #aichallenge 2011-11-14T04:14:34 *** marijnfs has quit IRC (Quit: leaving) 2011-11-14T04:15:39 *** Morendil has joined #aichallenge 2011-11-14T04:17:47 *** bqf has quit IRC (Changing host) 2011-11-14T04:17:47 *** bqf has joined #aichallenge 2011-11-14T04:17:54 *** marijnfs has joined #aichallenge 2011-11-14T04:18:08 *** Alexer has quit IRC (Ping timeout: 258 seconds) 2011-11-14T04:22:56 *** trcjr has quit IRC (Quit: Lost terminal) 2011-11-14T04:30:24 *** Alexer has joined #aichallenge 2011-11-14T04:31:02 *** Akranis has joined #aichallenge 2011-11-14T04:31:24 *** imp5imp5 has quit IRC (Ping timeout: 265 seconds) 2011-11-14T04:32:25 *** digger3 has joined #aichallenge 2011-11-14T04:46:16 *** Cyndre has quit IRC (Read error: Connection reset by peer) 2011-11-14T04:46:29 *** Cyndre has joined #aichallenge 2011-11-14T04:52:53 *** hkraal has quit IRC (Remote host closed the connection) 2011-11-14T04:55:21 *** p_l has quit IRC (Read error: Operation timed out) 2011-11-14T04:56:13 *** jasox has joined #aichallenge 2011-11-14T04:56:54 *** netantho has joined #aichallenge 2011-11-14T04:57:07 *** Ashoka has quit IRC (Quit: Page closed) 2011-11-14T04:57:39 *** djstrong has quit IRC (Ping timeout: 244 seconds) 2011-11-14T05:00:24 *** BenJackson has quit IRC (Ping timeout: 265 seconds) 2011-11-14T05:05:52 *** replore_ has quit IRC (Remote host closed the connection) 2011-11-14T05:12:25 *** grwip has joined #aichallenge 2011-11-14T05:13:27 *** Morendil has quit IRC (Ping timeout: 265 seconds) 2011-11-14T05:17:48 *** ronnyk_ has quit IRC (Ping timeout: 265 seconds) 2011-11-14T05:24:02 *** liberforce has joined #aichallenge 2011-11-14T05:25:10 hi all 2011-11-14T05:25:33 janzert: hi, are you one of the contest organizers ? 2011-11-14T05:25:50 *** NotABug has joined #aichallenge 2011-11-14T05:27:10 it's about my timeout problem 2011-11-14T05:33:46 anyone knows where is the time measuring code ? I can't find it in the game engine code... 2011-11-14T05:34:21 tried looking in a different starter kit? 2011-11-14T05:35:31 hum... not sure of what you mean 2011-11-14T05:35:56 I was looking in the playgame.py, ants.py and game.py 2011-11-14T05:36:21 *** Zaphus has joined #aichallenge 2011-11-14T05:36:30 i'm looking at how i'ts measured in the game engine, not in the client 2011-11-14T05:36:43 not in the bots 2011-11-14T05:39:12 it's in engine.py 2011-11-14T05:40:31 *** arscan has quit IRC (Ping timeout: 265 seconds) 2011-11-14T05:41:00 *** Zaphus has quit IRC (Ping timeout: 265 seconds) 2011-11-14T05:42:04 oh, that's in the worker... 2011-11-14T05:42:25 I was looking in the "ants" directory. 2011-11-14T05:42:29 Thanks Minthos 2011-11-14T05:43:59 *** p_l has joined #aichallenge 2011-11-14T05:44:08 *** erufu has joined #aichallenge 2011-11-14T05:44:33 *** Conorach has joined #aichallenge 2011-11-14T06:01:02 *** ikaros has joined #aichallenge 2011-11-14T06:01:45 *** Varan has joined #aichallenge 2011-11-14T06:02:07 *** marijnfs has quit IRC (Quit: leaving) 2011-11-14T06:02:36 Is there a reason why some many less games are run now then before the weekend? 2011-11-14T06:04:11 Server prices rose 2011-11-14T06:04:16 hmm oke 2011-11-14T06:04:24 sucky 2011-11-14T06:09:04 *** rexou has joined #aichallenge 2011-11-14T06:09:07 hi guys 2011-11-14T06:09:33 i'm having an issue while compiling my code 2011-11-14T06:09:58 it contains c++ox features using but it's not detected as c++11 2011-11-14T06:10:07 is anyone else getting random spikes of a few 100ms for some turns using java or scala? 2011-11-14T06:10:19 (that is locally though) 2011-11-14T06:10:37 *** coramoor has joined #aichallenge 2011-11-14T06:11:00 Is it possible to see how long turns take on the server? 2011-11-14T06:13:24 *** coramoor has quit IRC (Client Quit) 2011-11-14T06:13:53 Can someone using c++ox features in his code tell me if there's something special to do, brb 2011-11-14T06:16:13 rexou: use MyBot.cc for c++ and MyBot.cpp for c++11 2011-11-14T06:16:32 at least worker/compiler.py looks like that 2011-11-14T06:16:36 oh :) 2011-11-14T06:16:42 you can use whatever extension you want for the other files 2011-11-14T06:16:45 thank you 2011-11-14T06:17:01 (from the set .{c,cpp,cc} that is) 2011-11-14T06:17:09 yeah i got used to put .cpp for both but i saw .cc on the website 2011-11-14T06:17:20 just a convention related story :p 2011-11-14T06:17:34 anyway, thanks, will be helpful 2011-11-14T06:20:35 hum... bots should get an "end" message when they timeout 2011-11-14T06:20:54 currently the game engine just kills the timed out bots 2011-11-14T06:28:51 *** coramoor has joined #aichallenge 2011-11-14T06:32:21 heh... the big 200ms spikes only happen when my visualizer app is running in parallel 2011-11-14T06:32:48 nevertheless I now have code that runs a GC if there is spare time at the end of the turn just in case 2011-11-14T06:33:40 running GC is a very had idea in general 2011-11-14T06:34:10 it stops the entire VM 2011-11-14T06:34:48 Varan: depends on VM, and sometimes on its settings 2011-11-14T06:34:56 yes 2011-11-14T06:35:04 (and language) 2011-11-14T06:35:04 Varan: I only do it when I have spare time... the VM will decide to do a GC at some point anyway 2011-11-14T06:35:25 jix: Or you could write code that never allocates. 2011-11-14T06:35:33 or allocates less 2011-11-14T06:35:39 Zao: that must be fun in scala 2011-11-14T06:35:49 *** erufu has quit IRC (Ping timeout: 256 seconds) 2011-11-14T06:35:50 It's pretty much what one has to do if you ever target the .NET Compact Framework on the 360. 2011-11-14T06:35:58 (not that it's overly related to ants, but still) 2011-11-14T06:36:23 Zao: and it's frankly ages old thing Lisp programmers learnt before Java was called Oak ;-) 2011-11-14T06:37:51 well I'm not going to change code to allocate less unless I identified it as a bottleneck 2011-11-14T06:38:33 right now I don't have any bottlenecks (nor any good AI) I just had random spikes and wanted to figure out what causes them 2011-11-14T06:38:39 Spikes in isochronous code is suboptimal :D 2011-11-14T06:38:52 Well, soft-realtime, I guess. 2011-11-14T06:39:13 I thought it might be the GC but well it only was the system scheduler deciding to give my visualizer more time 2011-11-14T06:39:14 jix: attach a profiler? 2011-11-14T06:39:49 seriously, experiment design *and* software measurement should be compulsory CS classes 2011-11-14T06:40:07 p_l: I wouldn't know how to find the cause of these random spikes with visualvm (which is the only java profile I ever used) 2011-11-14T06:40:18 *profiler 2011-11-14T06:40:30 jix: you need to brush up your profiler-fu then 2011-11-14T06:40:49 p_l: well it was easy enough to identify the cause... I don't run my visualizer app at the same time, I don't get any spikes 2011-11-14T06:40:58 lol 2011-11-14T06:41:54 so I guess visualvm wouldn't even be able to show this, as it only measures the time the code actually runs 2011-11-14T06:42:20 jix: if the visualizer runs in separate process, then the spike isn't in your code 2011-11-14T06:42:23 you could view the stacktrace at the time of the spike 2011-11-14T06:42:33 p_l: yeah it is a seperate process 2011-11-14T06:42:35 and well... there's always OProfile and such 2011-11-14T06:42:42 *** jasox has quit IRC (Quit: Leaving) 2011-11-14T06:42:59 Varan: it's not that the spike is a concrete point in time... 2011-11-14T06:43:19 ah oke 2011-11-14T06:43:49 you can maybe also view gc behavior with visual vm 2011-11-14T06:43:50 Varan: I mean I measure the time a chunk of code takes.. and only after that I know there is a spike 2011-11-14T06:44:14 *** goffrie has quit IRC (Read error: Operation timed out) 2011-11-14T06:44:21 does anyone know if it is possible to view the turn time on the server? 2011-11-14T06:44:33 you cannot view any logging right? 2011-11-14T06:44:54 jix: yeah I get it now 2011-11-14T06:46:34 *** goffrie has joined #aichallenge 2011-11-14T06:47:05 *** Parsley_ has joined #aichallenge 2011-11-14T06:49:56 *** ALplus has quit IRC (Ping timeout: 258 seconds) 2011-11-14T06:56:11 *** jasox has joined #aichallenge 2011-11-14T06:58:30 *** b0rder has joined #aichallenge 2011-11-14T06:59:09 *** gfhhjk has joined #aichallenge 2011-11-14T06:59:23 sfsdf 2011-11-14T06:59:57 *** gfhhjk has quit IRC (Client Quit) 2011-11-14T07:03:01 *** Akranis has quit IRC (Ping timeout: 256 seconds) 2011-11-14T07:07:17 tcp server down? 2011-11-14T07:09:10 *** GeorgeSebastian has joined #aichallenge 2011-11-14T07:09:16 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-11-14T07:10:57 *** amstan has joined #aichallenge 2011-11-14T07:10:57 *** ChanServ sets mode: +o amstan 2011-11-14T07:11:01 *** mjwhitt has joined #aichallenge 2011-11-14T07:15:04 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-11-14T07:15:28 *** g0llum has joined #aichallenge 2011-11-14T07:15:30 *** liberforce has quit IRC (Quit: Leaving.) 2011-11-14T07:15:43 *** TheLinker has joined #aichallenge 2011-11-14T07:16:53 *** hkraal has joined #aichallenge 2011-11-14T07:19:01 *** liberforce has joined #aichallenge 2011-11-14T07:23:14 *** coramoor has quit IRC (Quit: Page closed) 2011-11-14T07:27:32 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-14T07:39:11 does anyone know which version of g++ is used to compile our code ? 2011-11-14T07:40:33 ++11 IIRC 2011-11-14T07:40:47 Oor -compliant, whatever 2011-11-14T07:47:16 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 252 seconds) 2011-11-14T07:49:19 *** Jak_o_Shadows has quit IRC (Remote host closed the connection) 2011-11-14T07:50:32 damn it, "auto" keyword is implemented since gcc 4.6 2011-11-14T07:52:46 *** jasox has quit IRC (Remote host closed the connection) 2011-11-14T08:00:28 *** amstan has joined #aichallenge 2011-11-14T08:00:32 *** ChanServ sets mode: +o amstan 2011-11-14T08:17:55 rexou: It's on 4.5 2011-11-14T08:18:01 (the builder, that is) 2011-11-14T08:18:59 Oh 2011-11-14T08:20:17 contestbot: later tell mleise any way around this? XMLHttpRequest cannot load http://aichallenge.org/game/14313. Origin http://forums.aichallenge.org is not allowed by Access-Control-Allow-Origin 2011-11-14T08:20:17 amstan: I think that worked... 2011-11-14T08:20:30 *** bergmark has quit IRC (Ping timeout: 260 seconds) 2011-11-14T08:23:58 Is fluxid tcp server down for everyone or just me? What are people using in the meantime? 2011-11-14T08:26:48 *** GeorgeSebastian is now known as sm0key 2011-11-14T08:27:28 *** sm0key is now known as georgesebastian 2011-11-14T08:27:46 *** georgesebastian is now known as Calvin 2011-11-14T08:33:07 *** Calvin is now known as Eleanora 2011-11-14T08:34:31 retybok: yeah, it's down for me too 2011-11-14T08:35:04 *** Parsley_ has quit IRC (Quit: Page closed) 2011-11-14T08:35:31 *** GeorgeSebastian has joined #aichallenge 2011-11-14T08:36:08 *** Eleanora is now known as georgesebastian_ 2011-11-14T08:37:04 *** ikaros has joined #aichallenge 2011-11-14T08:37:17 *** GeorgeSebastian is now known as Eleanora 2011-11-14T08:37:54 *** Eleanora has left #aichallenge 2011-11-14T08:38:04 *** georgesebastian_ is now known as georgesebastian 2011-11-14T08:39:00 *** noob has joined #aichallenge 2011-11-14T08:39:38 amstan janzert j3camero: Are you guys trying to get Google to give you computational resources? There's an internal mailing list post from someone trying to get in contact with your Google contact; it says that you are confused about who at Google you should be speaking to 2011-11-14T08:39:48 I guess I'll just have to wait :-/ 2011-11-14T08:40:06 jbroman: i think that was initiated by bmh 2011-11-14T08:40:16 jbroman: he thinks we should be getting more resources 2011-11-14T08:40:21 bmh is another googler 2011-11-14T08:40:26 is this an annual event? 2011-11-14T08:40:33 Yeah, bmh would match this person's initials. 2011-11-14T08:41:02 noob: ideally we would have it every term, but development time takes a while, so we usually skip a term or 2 2011-11-14T08:41:21 Does j3camero not already know his contact? 2011-11-14T08:41:41 i haven't heard from j3camero is a while 2011-11-14T08:42:20 I suspect that using Google's resources for this is impractical, so is it that you have insufficient dollars? 2011-11-14T08:42:51 dollars = google's resources 2011-11-14T08:42:59 jbroman: i don't know, i haven't requested this personally 2011-11-14T08:43:00 pairofdice: By which I meant computational resources. 2011-11-14T08:43:26 OK. Just odd to see an email saying "We don't know who we were talking to earlier; does anyone here know?" 2011-11-14T08:43:27 jbroman: bmh is for getting us more money, but he also thinks we should be using google's comps(as opposed to amazon) 2011-11-14T08:44:38 Hmm. bmh may have a better idea how viable is than I do. But Google doesn't have a straightforward EC2 analogue, so I would think it would be tricky. 2011-11-14T08:45:05 Well, anyways, thanks. Just was curious what was going on. 2011-11-14T08:45:27 *** praveen has joined #aichallenge 2011-11-14T08:45:31 well.. he's kinda right, we're sponsored by google, but giving the money to amazon 2011-11-14T08:45:48 so the next one will be summer or fall 2012? 2011-11-14T08:45:54 noob: hopefully 2011-11-14T08:45:57 Yeah, but it really depends on how many SWE hours would be required to make it possible to run on Google infrastructure. 2011-11-14T08:46:08 yeah 2011-11-14T08:46:40 *** u_ has joined #aichallenge 2011-11-14T08:47:36 amstan:ty 2011-11-14T08:47:39 *** u__ has joined #aichallenge 2011-11-14T08:47:54 *** praveen has quit IRC (Client Quit) 2011-11-14T08:50:46 *** noob has quit IRC (Quit: Page closed) 2011-11-14T08:50:52 jbroman: can you help me with something? 2011-11-14T08:51:03 i'm trying to use corn-syrup to myisampack the planetwars database 2011-11-14T08:51:04 *** u_ has quit IRC (Ping timeout: 240 seconds) 2011-11-14T08:51:04 *** u__ is now known as u_ 2011-11-14T08:51:07 i have it in /scratch 2011-11-14T08:51:15 but myisampack doesn't seem to be installed 2011-11-14T08:52:41 g2g, i'll be back in like 10 minutes 2011-11-14T08:55:47 *** NotABug has quit IRC (Ping timeout: 265 seconds) 2011-11-14T08:56:17 *** amstan has quit IRC (Read error: Connection reset by peer) 2011-11-14T08:56:27 *** amstan has joined #aichallenge 2011-11-14T08:56:27 *** ChanServ sets mode: +o amstan 2011-11-14T09:03:48 sent an email to syscom 2011-11-14T09:04:17 *** bergmark has joined #aichallenge 2011-11-14T09:05:42 *** mleise has joined #aichallenge 2011-11-14T09:09:16 14:23:58 < retybok> Is fluxid tcp server down for everyone or just me? What are people using in the meantime? ← duh, 3 hour downtime, sorry, it's up again 2011-11-14T09:09:47 i'll need to finally fix it so it could work for the next month without problems.. 2011-11-14T09:10:05 *** Antimony has joined #aichallenge 2011-11-14T09:10:08 Fluxid: np, really, thanks for providing such a great service 2011-11-14T09:10:47 *** aarossig has joined #aichallenge 2011-11-14T09:11:44 amstan: I don't think there is a way around cross-site scripting prevention. Maybe via secure HTTP, but I don't know. 2011-11-14T09:12:07 what about an iframe? 2011-11-14T09:13:12 I don't know. It could work, if it doesn't allow the embedded frame to access the sourrounding website 2011-11-14T09:14:11 Fluxid: How about a script that just resets the thing every now and then? Say once every 24 hours... considering that we won't need it working much longer than a month? 2011-11-14T09:14:43 *** UncleVasya has joined #aichallenge 2011-11-14T09:15:52 delt0r: maybe i'll just drop database on nov 18th, exactly month before finals 2011-11-14T09:16:17 the old tcp server was reset a few times for planet wars 2011-11-14T09:16:45 I was thinking of doing a TCP server... but then this one was done... seems to still have some issues? 2011-11-14T09:16:59 Hi. Are there official sample bots on the server? It's interesting to see how many contestants overpower greedybot. 2011-11-14T09:17:00 its it process not getting shutdown properly? 2011-11-14T09:19:45 *** sigh has quit IRC (Remote host closed the connection) 2011-11-14T09:21:17 *** b0rder has quit IRC (Ping timeout: 276 seconds) 2011-11-14T09:32:27 amstan: Installed. Let me know when done, as it looks like it's in the mysql-server package and I'd like to remove that. 2011-11-14T09:32:38 jbroman: ok, thanks 2011-11-14T09:33:35 jbroman: it's going 2011-11-14T09:33:48 *** xathis has joined #aichallenge 2011-11-14T09:34:27 yeah, this is going 10 times faster than jeff's server 2011-11-14T09:34:29 lol 2011-11-14T09:35:40 amstan: Hi. Can you please upload greedy_bot to the server like official hunter)_bot during pw so we can figure out the average strengthof contestants? 2011-11-14T09:36:02 UncleVasya: you have my permission to upload it yourself 2011-11-14T09:36:27 i'm not exactly sure where to find it and all that, so go ahead 2011-11-14T09:36:28 thanks. How to name it? 2011-11-14T09:36:32 greedy_bot 2011-11-14T09:36:40 put in the bio that it's a sample bot or something 2011-11-14T09:37:02 yep, and Waterloo as organization and Canada as country. 2011-11-14T09:37:12 sure 2011-11-14T09:37:37 inb4 greedy bot is in the top 200 2011-11-14T09:40:37 Fluxid: are you serious ? 2011-11-14T09:41:21 is greedy bot really in the top 200 ? 2011-11-14T09:41:55 liberforce: that was a joke ;) 2011-11-14T09:42:13 because UncleVasya didn't yet upload it, it is not... 2011-11-14T09:43:12 Fluxid: during a ew seconds I thought I was a moron and couldn't even achieve to beat a sample bot :p 2011-11-14T09:44:07 :D 2011-11-14T09:44:37 but this will be a good way to check how many contestants that do someting are there 2011-11-14T09:44:58 Fluxid: On the Planet Wars one stock bot ended up ~1600th from total 4700 bots. So we can say there were only ~1600 real enemies. 2011-11-14T09:45:18 :) 2011-11-14T09:45:36 But 310th from 4700 sounds better so real knowing is only for internal use :D 2011-11-14T09:48:34 I can beat GreedyBot \o/ 2011-11-14T09:50:21 *** NightExcessive has joined #aichallenge 2011-11-14T09:51:51 Why is the class Ilk called Ilk in the java starter package? 2011-11-14T09:52:03 seems like a really crappy name 2011-11-14T09:52:50 umm 2011-11-14T09:52:57 0.01781% for myisampack 2011-11-14T09:53:15 is that of total, or only 0.01781% was actually lost? 2011-11-14T09:54:36 *** Harpyon has joined #aichallenge 2011-11-14T09:54:57 can't register new account because I have no more e-mails. 2011-11-14T09:55:30 UncleVasya: ok, just give me the zip 2011-11-14T09:56:20 UncleVasya: but i g2g 2011-11-14T09:56:29 *** twymer has joined #aichallenge 2011-11-14T09:56:30 amstan@whateverdomain the contest's at 2011-11-14T09:57:07 ok, but I'll ask my friend first so you don't need to care. 2011-11-14T09:58:47 *** Zulfiraq has joined #aichallenge 2011-11-14T10:01:03 *** amstan has quit IRC (Ping timeout: 248 seconds) 2011-11-14T10:04:04 pairofdice: You can beat greedy_bot so here is the next stage for you: http://www.filefactory.com/file/c1c34e3/n/UncleVasya_22_0_Tabitha.7z 2011-11-14T10:04:20 *** Zulfiraq has quit IRC (Quit: Page closed) 2011-11-14T10:04:39 *** Zulfiqar has joined #aichallenge 2011-11-14T10:08:47 *** twymer has quit IRC (Quit: leaving) 2011-11-14T10:08:59 *** kilae has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-11-14T10:13:48 *** valentin has joined #aichallenge 2011-11-14T10:16:25 *** twymer has joined #aichallenge 2011-11-14T10:17:38 meh, the python greedybot times out on my computer 2011-11-14T10:18:07 Yeah, it does that sometimes 2011-11-14T10:18:20 maybe because of my 300ms interval 2011-11-14T10:18:24 UncleVasya: oh.. there is no source code included :-( 2011-11-14T10:18:51 You actually got somethind downloaded from that horrible place? 2011-11-14T10:18:55 something* 2011-11-14T10:19:05 after it lured me into a casino site, yes 2011-11-14T10:19:15 *** delt0r_ has quit IRC (Read error: Operation timed out) 2011-11-14T10:19:19 ahah, I was too close to pick UncleVasya_22_0_src.zip but noticed id right in time :) 2011-11-14T10:19:25 "the download is at the bottom of this page"-link -> click -> *doh* 2011-11-14T10:19:48 pairofdice: What is a good place to upload 2011-11-14T10:19:53 ? 2011-11-14T10:19:55 *cough* dropbox 2011-11-14T10:20:10 by bot uses .5 cpu load while greedybot uses 90% 2011-11-14T10:20:15 :p 2011-11-14T10:20:44 Dropbox is pretty handy anyway 2011-11-14T10:20:54 So think what you are doing wrong :D 2011-11-14T10:21:54 *** Akranis has joined #aichallenge 2011-11-14T10:21:58 hmm memory usage, let me check it 2011-11-14T10:22:08 avdg, yeah. Which is why I've dug into Eckler's Thinking in c++ 2011-11-14T10:22:17 Python that is, not GreedyBot 2011-11-14T10:22:21 my bot uses more memory though 2011-11-14T10:22:41 Eckel, sorry 2011-11-14T10:22:57 UncleVasya: so you code in Ocaml, is it any good ? :) 2011-11-14T10:23:02 @seen mcstar 2011-11-14T10:23:02 UncleVasya: mcstar was last seen in #aichallenge 1 day, 3 hours, 48 minutes, and 16 seconds ago: funny, metal-fluorine fire 2011-11-14T10:24:39 It does it's job but it is a throw-away code. You don't wanna see it without aspirinus :) 2011-11-14T10:24:46 *its 2011-11-14T10:25:09 UncleVasya: What is add_my_attack() supposed to do in detail? 2011-11-14T10:26:36 meh, my bot is growing faster, but only if greedybot doesn't attack me 2011-11-14T10:26:42 It is a part of old scripted battle resolution. Yesterday I rewrote it to the alphbeta. 2011-11-14T10:28:06 I did this way: 2011-11-14T10:28:55 1. When my ant want to move he checks if that square is under (possible) attack of enemy. 2011-11-14T10:29:13 *by enemy 2011-11-14T10:29:42 *** analyst74 has quit IRC (Ping timeout: 260 seconds) 2011-11-14T10:29:46 *** amstan has joined #aichallenge 2011-11-14T10:29:46 *** ChanServ sets mode: +o amstan 2011-11-14T10:30:14 janzert: still 48GB 2011-11-14T10:30:33 2. He checks if it still safe (square is under attack but that enemy is under my attack)/ 2011-11-14T10:30:59 jbroman: i'm done, thanks 2011-11-14T10:31:51 3. If safe then ant moves. Else he waits but I count him as a attacker of that enemy. 2011-11-14T10:32:31 After all other ants moved that my ant checks again if it's safe iand moves if it is. 2011-11-14T10:32:39 So it is something like: 2011-11-14T10:33:09 amstan: Does each worker currently run multiple games simultaneously? 2011-11-14T10:33:21 NightExcessive: runs one worker per core 2011-11-14T10:33:27 since we have 2 cores, it'll run 2 at the same time 2011-11-14T10:33:32 Ah 2011-11-14T10:33:42 Hey! I wanna to move there but I'm afraid do it alone. If someone else will attack that enemy than I will too. 2011-11-14T10:33:55 if it's an 8 player game, it'll run 4 pairs sequencially 2011-11-14T10:34:08 so the bots are running on a single core at once? or sequentially? 2011-11-14T10:34:29 they don't run on the same core at once 2011-11-14T10:34:41 *** delt0r_ has joined #aichallenge 2011-11-14T10:34:56 mleise: But alphabeta s better, faster to code, easier to understand so scripted battle resolution was a huge mistake. 2011-11-14T10:34:56 contestbot: later tell janzert the db will barely fit in /media/archive 2011-11-14T10:34:56 amstan: Aye, aye, sir 2011-11-14T10:35:26 UncleVasya: I see 2011-11-14T10:37:24 *** bmh has joined #aichallenge 2011-11-14T10:37:27 *** bmh has joined #aichallenge 2011-11-14T10:38:25 hey, are the workers still melting? 2011-11-14T10:38:39 Are you interested in OCaml? Or you asked to scare me a little? :) 2011-11-14T10:38:57 Not melting, just less workers 2011-11-14T10:39:38 *** valentin has quit IRC (Quit: Leaving) 2011-11-14T10:39:39 speaking of purist languages, I've got a haskell bot to submit. Does main need to be in MyBot.hs? Or are the build scripts smarter than that? 2011-11-14T10:40:03 I believe main has to be in the MyBot.ext file. 2011-11-14T10:40:04 Whatever the starter bot extension is 2011-11-14T10:40:40 *** mcstar has joined #aichallenge 2011-11-14T10:41:17 *** Antimony has quit IRC (Ping timeout: 245 seconds) 2011-11-14T10:41:27 o 2011-11-14T10:41:28 ok 2011-11-14T10:42:05 oh boy.. copying 50GB of data from servers to servers is satisfying 2011-11-14T10:42:07 It's nice to finally have something that eradicates dumb bots 2011-11-14T10:42:13 *** praveen1 has joined #aichallenge 2011-11-14T10:42:59 Yes, it is :) 2011-11-14T10:43:15 UncleVasya: To scare you a little. Does OCaml use a garbage collector or can it be used without? 2011-11-14T10:43:19 bmh: Is collaborative diffusion working out for you? I'm still struggling to get my diffusion maps right. 2011-11-14T10:43:44 it is. I finally wrote a reasonable cell scoring function 2011-11-14T10:44:06 I can't make mine run fast enough in python 2011-11-14T10:44:16 pairofdice: diffusion or your bot? 2011-11-14T10:44:25 diffusion 2011-11-14T10:44:41 pairofdice: pull as much code as possible out of the bot so you can run metrics on it 2011-11-14T10:44:57 at least that's what wound up working for me speeding up my python, had a couple dumb mistakes that were taking enormous amounts of time 2011-11-14T10:44:59 It was a struggle making it fast enough in Haskell. On my laptop I can do >40 diffusion cycles without timing out 2011-11-14T10:45:04 bmh: your bot name? I have been waiting to see a bot implemented using colloborative diffusion 2011-11-14T10:45:28 praveen1: skynet. Haven't submitted a real bot yet, pursuant to worker woes 2011-11-14T10:46:06 mleise: The only thing I know from this side of OCaml is that I can compile to byte-code (need ocamlrun to execute) and to native executable (faster, doesn't need anything). 2011-11-14T10:46:25 bmh: are you using tcp server also for games? 2011-11-14T10:46:31 generating 5^13 configurations in 4.2 secs, new record 2011-11-14T10:46:31 praveen1: nope, just got a working bot 2011-11-14T10:46:40 bmh: ok 2011-11-14T10:46:45 Not having any problems with diffusion in C++ here, but I'm running on 4.2Ghz cores :x 2011-11-14T10:46:49 UncleVasya: but do you have to free aloocated memory yourself, or does the language do that? 2011-11-14T10:47:04 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-11-14T10:48:23 ok, I understood you. No, I didn't saw this kind of operators in the Smiley's starterpack. 2011-11-14T10:49:00 There is 'new' to make instances of classes but no something like 'free'. 2011-11-14T10:49:24 is there any reason to leave gcc 4.5 and not update to version 4.6 ? cause 'auto' keyword is implemented in g++ 4.6 2011-11-14T10:49:31 *** stupq has joined #aichallenge 2011-11-14T10:49:58 So I think the removed automatically when no links th them. 2011-11-14T10:50:07 rexou: and in 4.5 2011-11-14T10:50:14 and in all of c and c++ in fact 2011-11-14T10:50:18 *** Anilm3 has joined #aichallenge 2011-11-14T10:51:11 Hi 2011-11-14T10:51:58 hi 2011-11-14T10:52:09 i used goto again today 2011-11-14T10:52:16 and im proud of it 2011-11-14T10:52:33 heheh 2011-11-14T10:53:13 mcstar: using goto is not as bad as they usually teach 2011-11-14T10:53:15 Is this a way to kid people who will read your code after contest ends? 2011-11-14T10:53:32 rexou: 4.6 has a lot of amusing breakages for legacy code, in general. 2011-11-14T10:53:37 no, its a way for breaking double loops from the inside 2011-11-14T10:53:43 in fact the Linux Kernel uses goto and recommends using it in some cases 2011-11-14T10:53:54 D has more uses of goto 2011-11-14T10:53:58 mcstar: Not quite the 'auto' intended though. 2011-11-14T10:54:10 Heck, I don't think C++03 assigns any semantics to it. 2011-11-14T10:54:12 Zao: i realize he meant the c++11 auto 2011-11-14T10:54:15 *** Antimony has joined #aichallenge 2011-11-14T10:54:16 goto in switch-statements for example 2011-11-14T10:54:20 *** bmh has quit IRC (Quit: bmh) 2011-11-14T10:54:35 goto is a good way of getting out of deep nested logic. It has its uses, but should only be used when it is clearer than doing anything else. 2011-11-14T10:55:06 well, after prototypingmy thing, i "functinalized" it, and voila, goto disappeared 2011-11-14T10:55:09 Usually problem solving code cares little about good practices 2011-11-14T10:55:10 but it was a handy way... 2011-11-14T10:55:31 mcstar: Ah, it denotes automatic storage duration, akin to 'register'. 2011-11-14T10:55:51 Zao: everything that is not declared otherwise, is auto 2011-11-14T10:56:14 thats what i know about it, thats why people never see it 2011-11-14T10:56:22 its like signed int 2011-11-14T10:57:16 so anyway, people didnt notice my benchmark 2011-11-14T10:57:22 5^13 is quite many 2011-11-14T10:58:03 and its ~6 lines of code 2011-11-14T10:58:13 so im gonna replace my tree search with this new one 2011-11-14T10:58:38 which one? 2011-11-14T10:59:02 the one i have now, a lengthier and slower recursive move generator 2011-11-14T10:59:17 how crazy 2011-11-14T10:59:40 what is? 2011-11-14T10:59:47 for (int i=0; i < 10; ++i) 2011-11-14T10:59:49 { 2011-11-14T10:59:49 bool do_break = false ; 2011-11-14T10:59:49 for (int j = 0; j < 10; ++j) 2011-11-14T10:59:49 { 2011-11-14T10:59:49 if (j > 5) then { 2011-11-14T10:59:50 do_break = true ; 2011-11-14T10:59:50 break ; 2011-11-14T10:59:51 } 2011-11-14T10:59:51 2011-11-14T10:59:52 if (do_break) then 2011-11-14T10:59:52 break ; 2011-11-14T10:59:53 } 2011-11-14T10:59:53 2011-11-14T10:59:54 You don't like to see this condition and to compute it every single iteration? Or you have other reasons? 2011-11-14T10:59:57 ahhhhhhh 2011-11-14T11:00:18 lol 2011-11-14T11:00:24 whats the prupose if this UncleVasya ? 2011-11-14T11:00:24 btw, it's not a past, I wrote it by myself right here, in KVIRC :P 2011-11-14T11:00:35 bb 2011-11-14T11:00:37 thats why you miss a } 2011-11-14T11:00:41 *** Anilm3 has quit IRC (Quit: Page closed) 2011-11-14T11:00:45 Zao, mcstar : ok ty 2011-11-14T11:01:02 rexou: what you cant use is the range based for 2011-11-14T11:01:11 but im sure you will manage with for_each 2011-11-14T11:01:48 you don't get your own hills in the input list if you can't see them, correct? 2011-11-14T11:01:58 l1: foreach (i; 0..10) 2011-11-14T11:01:58 { 2011-11-14T11:01:58 foreach (j; 0..10) 2011-11-14T11:01:58 { 2011-11-14T11:01:58 if (j > 5) break l1; 2011-11-14T11:01:59 } 2011-11-14T11:01:59 } 2011-11-14T11:02:00 Yeah, i think i can handle it :p 2011-11-14T11:02:01 i think yes it is 2011-11-14T11:02:02 yes 2011-11-14T11:02:17 mleise: d? 2011-11-14T11:02:21 but is there any reason to leave version 4.5 / 2011-11-14T11:02:22 :) 2011-11-14T11:02:45 rexou: i think this is an installation issue 2011-11-14T11:02:50 not a version issue 2011-11-14T11:02:57 mcstar: yes, it is 2011-11-14T11:03:03 if 4.5 comes with natty or whatnot it will be used 2011-11-14T11:03:11 mleise: which one? 2011-11-14T11:03:31 hehe 2011-11-14T11:03:39 if natty comes with 4.5 2011-11-14T11:03:47 maybe its both correct 2011-11-14T11:03:57 ok i get it now 2011-11-14T11:04:08 k, im shutting dowm my language processor for maintenance 2011-11-14T11:04:16 rexou: You could always compile most of your bot yourself into a library, and link it into a shell? 2011-11-14T11:04:38 *** bergmark has quit IRC (Ping timeout: 276 seconds) 2011-11-14T11:05:17 mleise: installation or version? it wasnt clear to me what you meant by "yes it is" 2011-11-14T11:06:52 Zao: yeah that seems to be the best way, thank you 2011-11-14T11:07:25 *** jasox has joined #aichallenge 2011-11-14T11:07:36 Zao: what if libstdc++ differes? 2011-11-14T11:07:39 -e 2011-11-14T11:07:59 im just asking, probably it wont matter 2011-11-14T11:10:38 mcstar: we are qite close to each other now in rankings: 5878 (you) vs 5896 (me) . This is the table page where really cool guys spend their time. 2011-11-14T11:10:43 *quite 2011-11-14T11:10:58 Bot of us had 1 game. 2011-11-14T11:11:04 *both 2011-11-14T11:11:10 *** NoxiaZ^ has joined #aichallenge 2011-11-14T11:11:10 usa* 2011-11-14T11:11:24 *** JorgeB has joined #aichallenge 2011-11-14T11:11:24 haha, i still beat you 2011-11-14T11:11:30 :D 2011-11-14T11:11:37 How is that possible 2011-11-14T11:11:44 what? 2011-11-14T11:11:57 mcstar: oh i was referring to your guess, which language i wrote the code in 2011-11-14T11:12:11 mleise: i thought you answered that with a smiley 2011-11-14T11:12:40 yes, but then I thought that a smiley could just mean that i am happy 2011-11-14T11:13:03 UncleVasya: hopefully the clouds will realize that this contest is very important and will decide that it needs to run the contest on all of its resources 2011-11-14T11:13:49 Let's pray for money instead 2011-11-14T11:14:47 i will sacrifice a goat tomorrow 2011-11-14T11:15:23 you need to shoot silver ( peroxide ) into the cloud 2011-11-14T11:15:37 that only brings rain 2011-11-14T11:15:42 i dont want rain 2011-11-14T11:15:51 silver coin rain? 2011-11-14T11:18:56 *** bergmark has joined #aichallenge 2011-11-14T11:20:39 Zao: you said that it was possible to compile my code into a lib before uploading it but we can't compile with .a cause it's server side 2011-11-14T11:21:21 the only solutions is to upload directly my objects files 2011-11-14T11:21:46 im sure you can upload a so 2011-11-14T11:22:00 i havent done any a, but why couldnt you? 2011-11-14T11:22:26 because even if you have a Makefile, it doesn't matter 2011-11-14T11:22:37 https://github.com/aichallenge/aichallenge/blob/d1ab4125163adfaab690af3b6257156365d23ebc/worker/compiler.py 2011-11-14T11:22:55 here is the script compiling our code 2011-11-14T11:23:22 rexou: i think if you upload a dll, ld will find it 2011-11-14T11:23:30 yes 2011-11-14T11:23:40 so thats ok, isnt it? 2011-11-14T11:23:42 Wow. My bot works surprisingly well without collision resolution 2011-11-14T11:24:00 you dont have to link-in a lib 2011-11-14T11:24:07 yes it's ok i thought i could use .a files but i will use dlopen and others stuffs :) 2011-11-14T11:24:25 yet another review of .so files 2011-11-14T11:25:00 s/stuffs/stuff/ 2011-11-14T11:25:52 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-14T11:29:58 *** Darrow has quit IRC (Read error: Connection reset by peer) 2011-11-14T11:35:08 *** mj41 has quit IRC (Ping timeout: 258 seconds) 2011-11-14T11:37:34 *** mj41 has joined #aichallenge 2011-11-14T11:39:14 *** bergmark has quit IRC (Ping timeout: 260 seconds) 2011-11-14T11:44:12 *** ProperChaos has joined #aichallenge 2011-11-14T11:44:18 Wow. Never realized that using double is so much slower than using 32-bit integers. 2011-11-14T11:44:45 NightExcessive: Yes... the curse of using the FPU 2011-11-14T11:45:19 Though on a 64-bit system it is probably better than on a 32-bit system 2011-11-14T11:45:46 Yeah. I was timing out trying to test on my netbook (32-bit) and not on my computer. So more than likely. 2011-11-14T11:46:22 amstan: once the database is on /media/archive that is pretty much all that needs to go on isn't it? 2011-11-14T11:47:44 *** Harpyon has quit IRC (Quit: Computer has gone to sleep.) 2011-11-14T11:49:19 *** Antimony has quit IRC (Ping timeout: 258 seconds) 2011-11-14T11:49:28 NightExcessive: Part of the problem is that floating-point values are stored as mantissa and exponent and the numbers are 'normalized'. That means they have exactly one binary digit in front of the dot (1.xxxxxx...). The actual 'magnitude' is stored in the exponent. 2011-11-14T11:50:14 Eeep. I like my 32-bit integers that are stored much more simply in RAM :) 2011-11-14T11:50:22 A simple addition requires the the cpu get the numbers to a common exponent etc... 2011-11-14T11:51:03 NightExcessive: Yep. And they are still the fastest even on 64-bit machines. 2011-11-14T11:51:05 *** bergmark has joined #aichallenge 2011-11-14T11:57:04 UncleVasya: How comes that OCaml suffers from mostly the same problems as D? 2011-11-14T11:59:32 *** Antimony has joined #aichallenge 2011-11-14T12:00:32 mleise: what problem? 2011-11-14T12:00:46 user error 2011-11-14T12:00:52 @error 2011-11-14T12:00:53 mcstar: I'm sorry Dave, err mcstar; I cannot 'error'. 2011-11-14T12:00:56 @error 2011-11-14T12:00:57 mcstar: An error occured while trying to show the previous error. 2011-11-14T12:01:00 @error 2011-11-14T12:01:01 mcstar: User error -- Replace user. 2011-11-14T12:01:05 thats it 2011-11-14T12:01:36 *** ikaros has joined #aichallenge 2011-11-14T12:02:45 UncleVasya: the toolchain lacks, the library that comes with it is missing some stuff, the existence of a garbage collector 2011-11-14T12:03:15 the type system is seen as too strict 2011-11-14T12:03:56 I like I don't need to care about freeing objects :p 2011-11-14T12:04:15 I love having direct control over when my objects are freed. 2011-11-14T12:04:31 UncleVasya: but embedded and realtime (game) developers seem to care much 2011-11-14T12:05:27 ocaml gc is actually really good 2011-11-14T12:07:42 *** delt0r_ has quit IRC (Ping timeout: 260 seconds) 2011-11-14T12:08:33 *** kilae has joined #aichallenge 2011-11-14T12:08:45 Woops. I accidentally made some cool patterns trying to diffuse. http://dl.dropbox.com/u/3166522/screenshots/2011/11/Screenshot-2011-11-14_12.08.21.png 2011-11-14T12:09:17 *** Cyndre has quit IRC (Read error: Connection reset by peer) 2011-11-14T12:09:32 *** Cyndre has joined #aichallenge 2011-11-14T12:09:35 ffff why is everybody using a diffusion-based strategy all of a sudden? 2011-11-14T12:09:43 it was MY IDEA! 2011-11-14T12:09:48 lol 2011-11-14T12:09:51 not really 2011-11-14T12:09:56 anyway, it sucks 2011-11-14T12:10:01 http://www.cs.colorado.edu/~ralex/papers/PDF/OOPSLA06antiobjects.pdf <- That paper's idea. 2011-11-14T12:10:03 what my man, you're saying diffusion isn't my invention!? 2011-11-14T12:10:05 *** peyton has quit IRC (Quit: peyton) 2011-11-14T12:10:06 i abandoned it totally 2011-11-14T12:10:13 but dont tell anyone 2011-11-14T12:10:16 yeah so did i 2011-11-14T12:10:17 that's why i chose to invent a super complicated path finding algorithm that even I haven't implmented to date 2011-11-14T12:10:17 ditto 2011-11-14T12:10:20 *** Antimony has quit IRC (Ping timeout: 260 seconds) 2011-11-14T12:10:31 path finding eh 2011-11-14T12:11:03 noone will ever *want* to steal that idea 2011-11-14T12:11:06 lericson: i tried potentials first 2011-11-14T12:11:23 but these attempts were made months before 2011-11-14T12:11:38 i think the first postential based bot was made by bhasker 2011-11-14T12:11:39 potentials = which move n/s/e/w is the most beneficial? 2011-11-14T12:12:05 well, i summed a c/(r^2+b) potential for every nbor 2011-11-14T12:12:12 then chose the minimal 2011-11-14T12:12:22 and it spread my ants pretty damn well 2011-11-14T12:12:31 i had a hexagonal lattice most of the time 2011-11-14T12:13:02 but even simpler methods are better 2011-11-14T12:13:12 so if it is, why bother with diffusion 2011-11-14T12:14:51 because it is the most generic solution :-) 2011-11-14T12:15:07 generic how? 2011-11-14T12:15:12 you mean common? 2011-11-14T12:15:28 most reusable? i don't know, something about my programmer brain liked it 2011-11-14T12:15:57 *** amstan has joined #aichallenge 2011-11-14T12:15:57 *** ChanServ sets mode: +o amstan 2011-11-14T12:16:02 it is certainly simple, but you need smething more sophisticated than that for battle 2011-11-14T12:16:22 anyway, let everyone figure it out for themselfes 2011-11-14T12:16:39 I'm modelling desires for each ant, then I test each ant's desire (outputs a movement vector) and round it to n/s/e/w 2011-11-14T12:16:41 keep up the work guys, even if you gonna throw it away eventually 2011-11-14T12:17:08 bbl 2011-11-14T12:18:31 *** Varan has quit IRC (Quit: Leaving) 2011-11-14T12:19:45 *** Antimony has joined #aichallenge 2011-11-14T12:21:04 *** delt0r_ has joined #aichallenge 2011-11-14T12:21:18 *** Blkt has quit IRC (Quit: ERC Version 5.3 (IRC client for Emacs)) 2011-11-14T12:22:35 *** bmh has joined #aichallenge 2011-11-14T12:25:49 *** liberforce has left #aichallenge 2011-11-14T12:25:58 *** jasox has quit IRC (Quit: Leaving) 2011-11-14T12:26:32 *** aarossig has quit IRC (Quit: leaving) 2011-11-14T12:26:38 *** Antimony has quit IRC (Ping timeout: 244 seconds) 2011-11-14T12:28:46 *** aarossig has joined #aichallenge 2011-11-14T12:29:17 *** kilae_ has joined #aichallenge 2011-11-14T12:29:25 *** Antimony has joined #aichallenge 2011-11-14T12:29:39 aarossig: ping 2011-11-14T12:29:59 aarossig: ping pong 2011-11-14T12:31:30 *** kilae has quit IRC (Ping timeout: 260 seconds) 2011-11-14T12:32:02 *** georgesebastian has quit IRC (Read error: Connection reset by peer) 2011-11-14T12:33:17 *** stupq has quit IRC (Ping timeout: 265 seconds) 2011-11-14T12:37:45 *** mj41 has quit IRC (Ping timeout: 260 seconds) 2011-11-14T12:39:09 *** aarossig has quit IRC (Quit: leaving) 2011-11-14T12:42:35 *** test has joined #aichallenge 2011-11-14T12:45:14 *** peyton has joined #aichallenge 2011-11-14T12:46:25 *** Antimony has quit IRC (Ping timeout: 260 seconds) 2011-11-14T12:46:40 *** NoxiaZ^ has quit IRC (Ping timeout: 240 seconds) 2011-11-14T12:47:34 *** NightExcessive has quit IRC (Quit: Page closed) 2011-11-14T12:48:50 *** BlueMotion has joined #aichallenge 2011-11-14T12:51:46 *** bmh has quit IRC (Quit: Page closed) 2011-11-14T12:53:19 *** amstan has quit IRC (Ping timeout: 248 seconds) 2011-11-14T12:56:11 *** BlueMotion has left #aichallenge 2011-11-14T13:02:24 *** UncleVasya has quit IRC () 2011-11-14T13:07:17 *** foRei has joined #aichallenge 2011-11-14T13:09:45 *** treeform has joined #aichallenge 2011-11-14T13:10:37 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-14T13:11:05 *** g0llum has joined #aichallenge 2011-11-14T13:13:24 *** LouisMartin has quit IRC (Quit: Page closed) 2011-11-14T13:13:34 @seen j3camero 2011-11-14T13:13:34 Zannick: j3camero was last seen in #aichallenge 3 weeks, 1 day, 21 hours, 59 minutes, and 58 seconds ago: If you are finding that some of the testing tools suck, please consider making better ones. We generally release whatever tools at the beginning, but the contestants always end up making better ones. 2011-11-14T13:13:55 *** LouisMartin has joined #aichallenge 2011-11-14T13:14:03 *** Cyb has joined #aichallenge 2011-11-14T13:14:17 *** retybok_ has joined #aichallenge 2011-11-14T13:14:42 Amstan, http://forums.aichallenge.org/viewtopic.php?f=25&t=1872 ? :D 2011-11-14T13:14:44 *** peyton has quit IRC (Quit: peyton) 2011-11-14T13:14:57 close enough 2011-11-14T13:14:58 :D 2011-11-14T13:15:12 haha 2011-11-14T13:15:34 when I voted the votes were tied 2011-11-14T13:16:01 *** Antimony has joined #aichallenge 2011-11-14T13:16:32 *** knyppeldynan has quit IRC (Read error: Operation timed out) 2011-11-14T13:16:56 can't imagine why someone in C# would really vote no. Optimization of a managed language != optimization of a native one. 2011-11-14T13:17:14 ..in terms of potential problems that is. 2011-11-14T13:17:28 LouisMartin: i didnt vote 2011-11-14T13:18:04 as for the compilation, it's so fast on my side that I can't count anything in more than 1 second (optimization or not) and by bot is relatively "fat" 2011-11-14T13:18:30 mcstar: I own you one mcstar, thanks ;) 2011-11-14T13:18:40 *** retybok_ has quit IRC (Ping timeout: 240 seconds) 2011-11-14T13:19:56 *** knyppeldynan has joined #aichallenge 2011-11-14T13:20:07 "As some languages have to be precompiled, they already have some handicap to interpreted languages." hahaha 2011-11-14T13:22:34 as for people arguing that processing speed doesn't matter in this contest, I'm afraid they think sweet algo runs in another dimension or something. "It's all about algo, not processing power!" Sure. Let's drop the turn time of those arguing this to 50ms and give them a chance to prove their point. 2011-11-14T13:23:43 Zannick: Yea. That's one of my favorite. I like to think he's trolling. :) 2011-11-14T13:24:05 sure 2011-11-14T13:24:53 if a player starts with multiple hills is it guaranteed to spawn a single ant for each one in the first round? 2011-11-14T13:26:40 Yeah, this is one challenge where speed does matter 2011-11-14T13:27:09 But still, you can do quite a lot 2011-11-14T13:27:37 *** JorgeB_ has joined #aichallenge 2011-11-14T13:28:55 matters as much as program size: the longer the better 2011-11-14T13:29:24 and ladies say you need to write thick code too 2011-11-14T13:29:28 *** JorgeB has quit IRC (Ping timeout: 256 seconds) 2011-11-14T13:29:28 *** JorgeB_ is now known as JorgeB 2011-11-14T13:30:43 *** Larose has joined #aichallenge 2011-11-14T13:32:37 :p 2011-11-14T13:33:34 *** retybok_ has joined #aichallenge 2011-11-14T13:35:28 but it has to be clean code 2011-11-14T13:35:37 anyone has any idea how is that stl vectors can grow as needed, yet you can access its data pointer continuously? 2011-11-14T13:35:55 yes, they move as needed 2011-11-14T13:36:02 hm 2011-11-14T13:36:11 what is the typical allocation size? 2011-11-14T13:36:17 im guessing it is overallocated 2011-11-14T13:36:19 i hope it is 2011-11-14T13:37:21 most things like that (arraylist, etc) will double in capacity when they need to grow larger 2011-11-14T13:37:31 *** Larose has quit IRC (Quit: Leaving.) 2011-11-14T13:37:33 than their current capacity 2011-11-14T13:38:13 ok, but how big it is when i create such vector/array 2011-11-14T13:38:25 is<->it 2011-11-14T13:38:42 i see no mention here about that 2011-11-14T13:38:53 maybe starts with 1 element and doubles from that? 2011-11-14T13:39:18 anyway, so theres a hidden copying involved when this grows 2011-11-14T13:39:41 mcstar: IIRC you give it an initial capacity when creating it 2011-11-14T13:39:50 i dont 2011-11-14T13:39:52 never did 2011-11-14T13:39:54 been a while since I used C++ 2011-11-14T13:39:55 you can't access the data pointer over push_back and friends 2011-11-14T13:40:10 yeah I had a funny issue with that yesterday. 2011-11-14T13:40:11 err, should have been "you can give..." 2011-11-14T13:40:11 ? 2011-11-14T13:40:15 anything that potentially invokes reallocation, basically 2011-11-14T13:40:30 there is a .data() member function that gives you direc access 2011-11-14T13:40:43 I thought I would be smart: adding a Ant value type to a vector and returning the pointer to that point. 2011-11-14T13:40:51 well, that turned out to fail horribly when you started adding more ants :] 2011-11-14T13:41:01 yes, and you can't use that pointer over any operation that adds, for example 2011-11-14T13:41:09 Conorach: you saved the iterator? 2011-11-14T13:41:14 or a direct pointer? 2011-11-14T13:41:31 mcstar it was the address that the iterator was pointing at. 2011-11-14T13:41:40 Garf: i see, it wasnt clear for me what was your point 2011-11-14T13:41:43 I figured that was a smart way of taking the pointer to a value ;) 2011-11-14T13:41:51 turns out it wasn't ;) 2011-11-14T13:41:54 i have a lis<_ant*> 2011-11-14T13:42:05 t 2011-11-14T13:42:15 *** Larose has joined #aichallenge 2011-11-14T13:42:22 just allocate dynamically your ants 2011-11-14T13:42:24 or whatever 2011-11-14T13:42:47 Yeah I changed to use Ant * instead of Ant values ;) 2011-11-14T13:43:01 but dont forget to free them 2011-11-14T13:43:03 ^^ 2011-11-14T13:43:21 they get freed when the program ends :P 2011-11-14T13:43:29 i thought i might use smart pointers for that, but i managed wo it 2011-11-14T13:43:31 although there might be hundreds of dead ants in the memory ;) 2011-11-14T13:43:33 how? 2011-11-14T13:43:56 well, the program is running in it's own allocated space, when the application ends, it will release all memory. 2011-11-14T13:43:56 you have to iterate through the vector/list to delete each ant 2011-11-14T13:44:02 pfff 2011-11-14T13:44:04 :p 2011-11-14T13:44:20 it's not like I'm holding any resources or anything right? :/ 2011-11-14T13:44:38 I agree it should be cleaned up properly. 2011-11-14T13:44:41 Conorach: in that sense c++ IS garbage collected 2011-11-14T13:44:46 Tough young ants, sinewy bodies. Tiny ... toes. 2011-11-14T13:44:52 yeah thanks to the OS. 2011-11-14T13:45:32 pathplanning stuff is more important to clean. 2011-11-14T13:45:56 that will really explode on you on a bigger map/more ants :] 2011-11-14T13:47:08 ? 2011-11-14T13:47:15 i just reuse the board 2011-11-14T13:47:33 i dont really allocate anything when the bot runs 2011-11-14T13:47:53 only trivial auto vars, dynamically only ants 2011-11-14T13:48:18 yeah :( my version is horribly inefficient in many ways. 2011-11-14T13:48:27 probably the reason I'm not getting very far still :P 2011-11-14T13:48:45 even though I did spend a considerable amount of time on it in the past week :S 2011-11-14T13:48:59 =] 2011-11-14T13:49:10 I'm letting my ants coast, enjoy the ride 2011-11-14T13:49:12 maybe I just need some time away from this project :P 2011-11-14T13:49:45 like for just slightly more than the minimum required sleep :S 2011-11-14T13:49:51 *** Hexren has joined #aichallenge 2011-11-14T13:49:56 *** Hexren has left #aichallenge 2011-11-14T13:50:22 Conorach: try a shower 2011-11-14T13:50:30 ^_^ 2011-11-14T13:50:32 not that i smell anythin... 2011-11-14T13:51:51 Heh, I just took out the trash, that's enough fresh stuff for today :p 2011-11-14T13:54:06 Fresh, yes 2011-11-14T13:54:33 *** bugnuts2 has quit IRC (Quit: Leaving) 2011-11-14T13:55:36 *** praveen1 has quit IRC (Quit: Page closed) 2011-11-14T13:58:45 *** retybok_ has quit IRC (Ping timeout: 260 seconds) 2011-11-14T13:59:30 *** Cyb has quit IRC (Ping timeout: 256 seconds) 2011-11-14T14:00:31 *** NightExcessive has joined #aichallenge 2011-11-14T14:03:18 I have to use more tricks on my memory usage since I'm doing it in js, but otherwise I have no real garbage in my code 2011-11-14T14:03:23 hey guys is using some sort of minimax algorithm for combat any good? 2011-11-14T14:03:38 *** dvladim has joined #aichallenge 2011-11-14T14:03:47 oh, yeah 2011-11-14T14:03:57 *** choas has joined #aichallenge 2011-11-14T14:04:05 but i wouldnt call trees without branches trees 2011-11-14T14:04:16 thats the problem 2011-11-14T14:04:32 Shrubbery 2011-11-14T14:04:39 *** Cyb has joined #aichallenge 2011-11-14T14:11:05 *** Antimony has quit IRC (Ping timeout: 260 seconds) 2011-11-14T14:11:58 *** Coltragon has joined #aichallenge 2011-11-14T14:12:16 Hey people, I just started and I was wondering. How can you detect enemy Ants? 2011-11-14T14:12:44 Coltragon: By seeing them :) 2011-11-14T14:13:01 How far is the seeing range of your ant? 2011-11-14T14:13:40 It's passed in through the configuration 2011-11-14T14:13:40 *** Cyb has quit IRC (Ping timeout: 256 seconds) 2011-11-14T14:13:49 It's the square root of viewradius2 2011-11-14T14:18:18 *** Cyb has joined #aichallenge 2011-11-14T14:21:12 *** Antimony has joined #aichallenge 2011-11-14T14:21:24 *** oniondip has joined #aichallenge 2011-11-14T14:22:32 *** jcdny has joined #aichallenge 2011-11-14T14:24:39 *** okan_ has joined #aichallenge 2011-11-14T14:28:16 *** mleise has quit IRC (Ping timeout: 240 seconds) 2011-11-14T14:28:30 *** Redgis has joined #aichallenge 2011-11-14T14:29:44 oleg ocharenko, your bot climbed pretty high last time 2011-11-14T14:31:20 *** Hexren1 has joined #aichallenge 2011-11-14T14:31:42 *** okan_ has quit IRC (Quit: Page closed) 2011-11-14T14:32:04 "Sorry could not find any user with that name." 2011-11-14T14:33:13 http://aichallenge.org/profile.php?user=53 2011-11-14T14:33:24 this is unclevasya 2011-11-14T14:33:26 i think 2011-11-14T14:33:34 but he isnt here dan 2011-11-14T14:33:37 damn 2011-11-14T14:35:54 *** epicmonkey has joined #aichallenge 2011-11-14T14:35:57 *** ckennelly has quit IRC (Remote host closed the connection) 2011-11-14T14:38:03 *** retybok_ has joined #aichallenge 2011-11-14T14:38:35 uh oh 2011-11-14T14:39:02 *** Hexren1 has left #aichallenge 2011-11-14T14:41:15 *** ALplus has joined #aichallenge 2011-11-14T14:48:42 *** ALplus has quit IRC (Read error: Connection reset by peer) 2011-11-14T14:50:44 *** ALplus has joined #aichallenge 2011-11-14T14:58:38 *** Antimony has quit IRC (Ping timeout: 276 seconds) 2011-11-14T15:01:42 *** anky has joined #aichallenge 2011-11-14T15:02:43 my bot crashes when i test locally on my computer but it works on the same map on the server , any idea what could be wrong? 2011-11-14T15:03:24 no clue ,but if it crashes locally you could quite simply debug it 2011-11-14T15:03:30 the other way around would be way harder 2011-11-14T15:03:30 *** bergmark has quit IRC (Read error: Operation timed out) 2011-11-14T15:04:55 i knda tried but its stupid the ant doesnt even leaves the hill when testing locally , even though i have check in place for that 2011-11-14T15:06:02 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-14T15:06:15 anky: print print print 2011-11-14T15:06:22 *** g0llum has joined #aichallenge 2011-11-14T15:06:27 what language are you using? 2011-11-14T15:06:44 python 2011-11-14T15:06:50 oh 2011-11-14T15:07:02 cant help with that 2011-11-14T15:07:23 which language are u using? 2011-11-14T15:07:30 c++11 2011-11-14T15:09:40 anky: ah, a general remark, use the --engine_seed parameter when you try to dbug 2011-11-14T15:10:04 so that you fix the engine random seed, thereby making your errors more reproducible 2011-11-14T15:10:29 thanks i will give it a try 2011-11-14T15:10:31 *** NoxiaZ^ has joined #aichallenge 2011-11-14T15:10:37 i dont think there is amore convenient way to know the engine seed of a replay other than extracting from the html directly 2011-11-14T15:10:52 probably i will submit a change or whatnot 2011-11-14T15:11:25 *** Harpyon has joined #aichallenge 2011-11-14T15:11:52 *** Max___ has joined #aichallenge 2011-11-14T15:11:57 Hi 2011-11-14T15:12:41 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-11-14T15:13:23 I have a question for you: does the order of the move order matter? ie: AAA_ => _AAA, do I need to start by the right one and so on? Or it doesn't matter? 2011-11-14T15:13:54 where do I report bugs with the starter packages? The C# bot uses Millisecond TimeSpan.Milliseconds instead of TimeSpan.TotalMilliseconds 2011-11-14T15:13:59 I guess that only the ones that _end up_ in the same space die 2011-11-14T15:14:03 so it is all up to you :) 2011-11-14T15:14:37 I had ants that move to eachother's location in one turn :S 2011-11-14T15:14:39 that went fine 2011-11-14T15:14:41 that what I guessed in a first place but I have a strange collision now, I am trying to understand what happened :D 2011-11-14T15:17:30 I have a group of 13 ants and next turn, one died... but only one, I have a group of 12 ants now 2011-11-14T15:17:43 so I don't understand how only one died... 2011-11-14T15:17:58 Max___, 2 ants, can't be in the same loc, so additianolly check ,if you don't send 2 ants to the same spot 2011-11-14T15:18:32 I checked => only one ant is going to that position, and it died 2011-11-14T15:18:46 only one victim 2011-11-14T15:19:31 *** bergmark has joined #aichallenge 2011-11-14T15:19:32 Max___: replay? 2011-11-14T15:19:40 Max___: enemy ants nearby? 2011-11-14T15:20:38 how can I give you the replay? 2011-11-14T15:21:03 I have this file : 0.stream 2011-11-14T15:21:24 Max___: http://paste.aichallenge.org/ 2011-11-14T15:24:19 *** Antimony has joined #aichallenge 2011-11-14T15:24:55 hum 2011-11-14T15:25:01 http://paste.aichallenge.org/iKr9h/ but it does not work 2011-11-14T15:27:04 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-14T15:27:25 Max___: seems you couldnt copy it in one piece 2011-11-14T15:27:38 make sure you copy/paste the whole thing 2011-11-14T15:27:55 wait 2011-11-14T15:28:01 this doesnt seem to be a replay file 2011-11-14T15:28:19 that's my 0.stream file 2011-11-14T15:28:44 thats just the bot's io 2011-11-14T15:28:54 you need the json file that the visualizer uses 2011-11-14T15:29:16 0.replay 2011-11-14T15:29:20 in the game_logs dire 2011-11-14T15:29:37 I ran : ./play_one_game.sh -So "../cpp/Release/aichallenge" "../cppold3/MyBot" | java -jar visualizer.jar 2011-11-14T15:29:52 Max___: is there a 0.replay file under game_logs? 2011-11-14T15:30:07 not from today 2011-11-14T15:30:28 well, its in the html too 2011-11-14T15:30:45 visualizer.loadReplayData('{"status" 2011-11-14T15:30:54 the replay starts after ' 2011-11-14T15:31:21 I don't use that html file either 2011-11-14T15:31:32 gawd 2011-11-14T15:31:34 -rw-r--r-- 1 max max 274 oct. 28 19:51 0.replay -rw-r--r-- 1 max max 2010310 nov. 14 20:44 0.stream -rw-r--r-- 1 max max 998 oct. 28 19:51 replay.0.html 2011-11-14T15:31:35 what do you use? 2011-11-14T15:31:55 the line command I gave you with java -jar visualizer.jar 2011-11-14T15:32:10 blame delt0r 2011-11-14T15:32:45 anyway, rerun the scene somhow, and give me the replay 2011-11-14T15:32:47 or not 2011-11-14T15:32:50 i dont know 2011-11-14T15:33:23 if that method doesnt preserve the replay file that sucks, tell delt0r to save it nevertheless 2011-11-14T15:36:15 I can rerun the match with cat game_logs/0.stream | java -jar visualizer.jar, I'm trying to understand what to send to the replay paste 2011-11-14T15:36:31 *** aibot has joined #aichallenge 2011-11-14T15:36:40 i told you 2011-11-14T15:37:54 how can you recreate the whole game from 0.stream anyway? 2011-11-14T15:38:01 thats only one bot's perspective 2011-11-14T15:38:23 guess you aren't a solipsist ;p 2011-11-14T15:38:51 with "cat game_logs/0.stream | java -jar visualizer.jar " I have the whole game... 2011-11-14T15:39:11 *** Zaphus has joined #aichallenge 2011-11-14T15:39:49 digger3: i have no idea what that is 2011-11-14T15:39:59 *** Fandekasp has quit IRC (Read error: Operation timed out) 2011-11-14T15:40:10 Max___: is it delt0r's visualizer? 2011-11-14T15:40:12 wget http://paste.aichallenge.org/iKr9h/raw/ -O - | java -jar visualizer.jar is doing the trick too 2011-11-14T15:40:30 lol, if you uploaded it to the paste 2011-11-14T15:40:36 what does that prove? 2011-11-14T15:40:40 that you can use wget? 2011-11-14T15:41:12 just that I can share the replay anyway :D 2011-11-14T15:41:39 mcstar: reading the first paragraph is probably sufficient: http://en.wikipedia.org/wiki/Solipsism , damn, don't they teach philosophy these days? 2011-11-14T15:42:16 *** ProperChaos has quit IRC (Ping timeout: 265 seconds) 2011-11-14T15:42:23 digger3: i dont know how you figured it out(well i did), but you are spot on 2011-11-14T15:42:30 was nice talking to my projectionand realizing this though 2011-11-14T15:42:50 mcstar: hah, you catch on quick :) 2011-11-14T15:43:23 Max___: i will take a look at that stream file 2011-11-14T15:43:35 *** Accoun has quit IRC () 2011-11-14T15:44:04 (my issue is at turn 234 btw) 2011-11-14T15:44:14 (234-235) 2011-11-14T15:45:04 *** Antimony has quit IRC (Ping timeout: 258 seconds) 2011-11-14T15:45:37 Max___: ok, youre right, it save everything 2011-11-14T15:45:46 s 2011-11-14T15:46:40 (and I am player 1) 2011-11-14T15:46:51 Max___: now tell me how did you compile visualizer.jar 2011-11-14T15:46:59 i always wanted to try it 2011-11-14T15:47:08 (not really, but i will now) 2011-11-14T15:47:22 what ? I just downloaded from there : http://aichallenge.org/tools.tar.bz2 2011-11-14T15:48:00 aha 2011-11-14T15:48:15 i use git, and its not compiled in the github repository 2011-11-14T15:48:23 so ive got to do that first 2011-11-14T15:48:45 What does the following message mean? INFO: ignored o 22 10 e # move blocked 2011-11-14T15:48:54 does it mean I tried to walk on water? 2011-11-14T15:49:23 Only jesus_bot is allowed to do that. 2011-11-14T15:49:52 there is a build.xml 2011-11-14T15:50:01 anyone knows what i need? ants? maven? 2011-11-14T15:50:05 -s 2011-11-14T15:51:02 where do I report bugs with the starter packages? The C# bot uses Millisecond TimeSpan.Milliseconds instead of TimeSpan.TotalMilliseconds 2011-11-14T15:51:28 Yexo: you use tickets on github 2011-11-14T15:51:38 *** Fandekasp has joined #aichallenge 2011-11-14T15:51:39 or make a change and do a pull request 2011-11-14T15:51:51 ok, thanks 2011-11-14T15:53:26 retybok_: you'll also get that if you try to step into food that just spawned next to you 2011-11-14T15:54:40 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-11-14T15:54:58 *** willem has joined #aichallenge 2011-11-14T15:56:42 *** aibot has quit IRC (Quit: Page closed) 2011-11-14T15:57:06 Max___: i dont understand 2011-11-14T15:57:21 thestinger: thanks, that's my problem 2011-11-14T15:57:23 you have 61 ants, after that turn you have 59 2011-11-14T15:57:29 and 2 clearly stepped on each other 2011-11-14T15:57:46 hum 2011-11-14T15:57:48 I've always considered that ignoring the fact that you don't need to actually step on food to grab it was unimportant 2011-11-14T15:57:55 but I kept getting this message :) 2011-11-14T15:58:12 retybok_: i think you can safely ignore it 2011-11-14T15:58:23 doesnt happen often 2011-11-14T15:58:36 do we know which python packages are installed on the server? 2011-11-14T15:58:36 mcstar: I'm glad to know that I can fix it if I need too 2011-11-14T15:59:00 Max___: btw, that java visulizer sucks 2011-11-14T15:59:01 Also knowing what it is means that I can stop worrying about it :) 2011-11-14T15:59:04 *** xathis_ has joined #aichallenge 2011-11-14T15:59:04 why do you use it at all? 2011-11-14T15:59:25 Max___: it runs js inside java 2011-11-14T15:59:30 mcstar, i'm sorry, i've counted the dead one lol 2011-11-14T15:59:58 you are forgiven, im not clint eastwood 2011-11-14T16:00:20 Ah darn. How should I make ants block scents from diffusing? (Collaborative diffusion) 2011-11-14T16:00:32 My thought was extremely simple "ray" traces... but that doesn't seem like the best idea. 2011-11-14T16:00:41 ? 2011-11-14T16:00:52 that goes against the method's principle isnt it? 2011-11-14T16:01:13 the whole point of this diffusion thing is that you dont want to deal with complexity 2011-11-14T16:01:23 and dont want to understand whats going on 2011-11-14T16:01:24 Well, I can't think of how else to block the scents from going through ants. 2011-11-14T16:01:34 so you say: what the hell, lets just diffuse ant farts on the map 2011-11-14T16:01:44 Exactly. 2011-11-14T16:01:52 *** xathis has quit IRC (Ping timeout: 240 seconds) 2011-11-14T16:01:55 But the ants sniffing those farts first should make them completely dissipate. 2011-11-14T16:02:06 And not travel any further. 2011-11-14T16:02:23 NightExcessive: it depends how much you iterate each turn 2011-11-14T16:02:29 diffusion has a speed limit 2011-11-14T16:02:44 *** Accoun has joined #aichallenge 2011-11-14T16:02:52 so it cant reach beyond steps greater than the number of iterations 2011-11-14T16:04:53 oh, ive just seen now 2011-11-14T16:05:03 "going through ants" what do you mean? 2011-11-14T16:05:29 *** bergmark has quit IRC (Ping timeout: 252 seconds) 2011-11-14T16:06:55 *** NightExcessive has quit IRC (Ping timeout: 265 seconds) 2011-11-14T16:07:42 *** Antimony has joined #aichallenge 2011-11-14T16:07:48 *** NightExcessive has joined #aichallenge 2011-11-14T16:08:10 Oh you web chat... disconnecting me. 2011-11-14T16:08:44 Setting the value to 0 if the square has an ant? Wouldn't that work? 2011-11-14T16:09:18 *** bergmark has joined #aichallenge 2011-11-14T16:09:41 I still need a way to make sure I don't diffuse beyond the ant. 2011-11-14T16:11:54 weird 2011-11-14T16:11:54 *** Zulfiqar has quit IRC (Quit: Page closed) 2011-11-14T16:11:59 my colors got messed up 2011-11-14T16:12:57 im gonna restart my wm 2011-11-14T16:12:59 *** mcstar has quit IRC (Quit: WeeChat 0.3.6) 2011-11-14T16:13:28 *** mcstar has joined #aichallenge 2011-11-14T16:15:10 *** yahim91 has joined #aichallenge 2011-11-14T16:16:30 *** dvladim has quit IRC (Ping timeout: 258 seconds) 2011-11-14T16:16:38 *** mj41 has joined #aichallenge 2011-11-14T16:17:27 could anyone tell me how to debug my java project? 2011-11-14T16:18:31 *** amstan has joined #aichallenge 2011-11-14T16:18:31 *** ChanServ sets mode: +o amstan 2011-11-14T16:19:58 ive heard java is especially hard to debug on average 2011-11-14T16:21:14 there must be a way to do it though 2011-11-14T16:21:47 but i have no idea 2011-11-14T16:22:12 get bot input from replay, run your bot in a debugger with that input? 2011-11-14T16:22:17 *** amstan_ has joined #aichallenge 2011-11-14T16:22:17 *** ChanServ sets mode: +o amstan_ 2011-11-14T16:22:59 *** BartV_ has joined #aichallenge 2011-11-14T16:23:27 Hi 2011-11-14T16:24:07 I'm trying to compile the c++ starter package 2011-11-14T16:24:10 replay from game logs? 2011-11-14T16:24:26 but for some reason Im not getting an exe 2011-11-14T16:24:38 BartV_: what error messages do you get? 2011-11-14T16:24:55 Java isn't hard to debug if you're used to c or c++ 2011-11-14T16:25:08 none tbh I dont really know how to do it 2011-11-14T16:25:22 what are you using to compile it? 2011-11-14T16:25:23 I've tried eclipse and wxdevcpp on windows 2011-11-14T16:25:29 g++ works 2011-11-14T16:25:31 personally, I use Code::BLocks 2011-11-14T16:25:38 as my IDE 2011-11-14T16:25:52 I'm using MSVS 2010... worst decision ever. 2011-11-14T16:26:14 BartV_: "g++ *.cc -o yourbot.exe" works for me 2011-11-14T16:26:28 I made some simple c++ programs before, but they all had a main.cpp 2011-11-14T16:26:36 Yeah, the lines in the readme worked for me just fine 2011-11-14T16:26:47 Antimony: it was a joke on the average java programmer, since it is hard for him to debug and there are many average java programmers.... 2011-11-14T16:27:15 *** ltriant has joined #aichallenge 2011-11-14T16:27:31 I'd use the lines in the readme if I weren't on Windows :) 2011-11-14T16:27:32 I have no patience for IDEs -_- 2011-11-14T16:27:45 They work ion windows just fine 2011-11-14T16:28:10 it says g++ is not recognized 2011-11-14T16:28:24 NightExcessive it's not that bad right? I'm making my bot in C++ and it's working fine for me. 2011-11-14T16:28:48 kdevelop here 2011-11-14T16:28:50 Conorach: The most difficult thing for me has been making it cross-platform compatible... since the workers run Ubuntu. 2011-11-14T16:28:50 oh, I think i got MingW installed 2011-11-14T16:29:02 NightExcessive yeah it took me about 15 tries to get that right :P 2011-11-14T16:29:12 *** Coltragon has quit IRC (Ping timeout: 245 seconds) 2011-11-14T16:29:13 VS is just too lenient about all these kind of rules :P 2011-11-14T16:29:13 Hehe. 2011-11-14T16:29:14 BartV, add MingW g++ to your path 2011-11-14T16:29:38 Conorach: I've been uploading it to my home server and testing from there... still really annoying. 2011-11-14T16:30:30 *** Antimony has quit IRC (Ping timeout: 260 seconds) 2011-11-14T16:31:03 I see. well, as soon as I got the uploaded version working it's been working fine. Just the one thing that keeps bugging me is the damn window console for when I want to paste some lines of input :P why couldn't they just make it to accept ctrl+v :/ 2011-11-14T16:31:21 hint:Right mouse 2011-11-14T16:31:39 *** pedrosorio has joined #aichallenge 2011-11-14T16:31:48 :o 2011-11-14T16:32:00 Shift+insert also works 2011-11-14T16:32:25 pairofdice: I've added C:\Program Files (x86)\Dev-Cpp\mingw32 , but still it's not working 2011-11-14T16:32:49 *** Surya has joined #aichallenge 2011-11-14T16:33:18 Hi there :) 2011-11-14T16:33:23 Heya. 2011-11-14T16:33:29 I have just C:\MinGW\bin 2011-11-14T16:33:45 I really love how this type of activity can make so many people want to try programming (or some language) for the first time in their life :P 2011-11-14T16:34:14 So, you probably need the \bin there because that's where the executable is ;p 2011-11-14T16:34:18 Hehe. It made me really dive into C++ instead of just poking it with a stick occasionally. 2011-11-14T16:35:44 Same, pretty soon I'm going to rewrite my bot in c++ 2011-11-14T16:35:54 I have a weird issue: the IssueOrder(antLoc, direction); sometimes doesn't work. I mean it was working fine and now, sometimes the order doesn't make the ant moves(while it's possible) Any idea? (even the obvious ones ><). I haven't modified the way of writing System.Console.Out.WriteLine("o {0} {1} {2}", loc.Row, loc.Col, direction.ToChar()); 2011-11-14T16:36:07 well, the motivation of wanting to perform in such an awesome event will really help you overcome these obstacles that you will find in C++ :P 2011-11-14T16:36:28 still not recognized :( 2011-11-14T16:36:29 I mean, I work with that stuff every day, and I still happen upon the weiiiirdest compiler errors sometimes :/ 2011-11-14T16:37:25 *** bergmark has quit IRC (Ping timeout: 240 seconds) 2011-11-14T16:37:29 *** asdasdmnvcb has joined #aichallenge 2011-11-14T16:37:34 *** Anilm3 has joined #aichallenge 2011-11-14T16:37:52 *** Max___ has quit IRC (Ping timeout: 265 seconds) 2011-11-14T16:38:15 Must be really frustrating to work as tech help 2011-11-14T16:38:23 *** Palmik has quit IRC (Remote host closed the connection) 2011-11-14T16:38:35 Surya: Is it actually printing the line? You could check by starting it and feeding it input manually. 2011-11-14T16:39:05 *** Conorach_ has joined #aichallenge 2011-11-14T16:39:20 *** NoxiaZ^ has quit IRC () 2011-11-14T16:40:38 Thank you NightExcessive. I've added a log trace in the IssueOrder method which logs all the moves (even those which don't work). I don't know how i can make sure it writes the interpreted line. 2011-11-14T16:41:01 im download code::blocks + mingw now, see if that works 2011-11-14T16:41:04 Surya: Start it yourself (without running the Python engine) and input code from any game. 2011-11-14T16:41:36 Surya: http://aichallenge.org/visualizer.php?game=94301 -> Click the "BOT IN" thing, type one of the players' names and enter the range 1-1 2011-11-14T16:41:52 *** Conorach has quit IRC (Ping timeout: 244 seconds) 2011-11-14T16:43:15 Done. I have the list of actions and it seems allright 2011-11-14T16:43:35 Odd. It may be a bug with the engine then. 2011-11-14T16:44:09 *** kilae_ has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-11-14T16:44:21 *** anky has quit IRC (Quit: Page closed) 2011-11-14T16:44:55 I downloaded the tools pack, but it has the same behavior. I'm going to investigate and post something on the forum if I can't find the problem. Thank you for your attention :) 2011-11-14T16:45:31 Surya: No problem :) 2011-11-14T16:45:42 Surya: I might be experiencing the same problem 2011-11-14T16:45:50 How can I debug the engine? I mean I launch the game with the script .cmd 2011-11-14T16:46:01 but I assumed the engine was very well-tested, so I disabled the assert on that in my own bot 2011-11-14T16:46:47 Surya: You'd need to know Python. I'm not sure of much of anything about the engine. 2011-11-14T16:46:59 great, code blocks is working 2011-11-14T16:47:09 thx for the help pairofdice 2011-11-14T16:47:24 ok, seems complicated. I'm going to try with an old version of an IA. And see what's going on. Thx again. 2011-11-14T16:47:53 *** lavalamp has joined #aichallenge 2011-11-14T16:50:05 In this game, how was I able to win when teinemaa has 87 ants waiting in the hill? http://aichallenge.org/visualizer.php?game=94295&user=2238 2011-11-14T16:50:31 *** bergmark has joined #aichallenge 2011-11-14T16:51:52 *** mceier has joined #aichallenge 2011-11-14T16:52:04 lavalamp: if he moves away from his hill 2011-11-14T16:52:13 razing hills is done before spawning new ants 2011-11-14T16:52:17 his ants moved away from the hill... 2011-11-14T16:52:40 That has to be my luckiest win... 2011-11-14T16:54:39 contestbot: seen janzert 2011-11-14T16:54:39 amstan_: janzert was last seen in #aichallenge 3 hours, 14 minutes, and 27 seconds ago: err, should have been "you can give..." 2011-11-14T16:55:52 well 2011-11-14T16:55:59 the archive server now has 800MB free 2011-11-14T16:56:11 that should be enough for the pw website and phpmyadmin 2011-11-14T16:57:24 *** strupq has joined #aichallenge 2011-11-14T17:00:58 *** asdasdmnvcb has quit IRC (Quit: Page closed) 2011-11-14T17:03:16 *** choas has quit IRC (Read error: Operation timed out) 2011-11-14T17:04:35 *** pairofdice has quit IRC (Quit: Just keep breathing) 2011-11-14T17:04:57 *** Antimony has joined #aichallenge 2011-11-14T17:07:20 *** amstan_ has quit IRC (Ping timeout: 276 seconds) 2011-11-14T17:12:57 *** twymer has quit IRC (Ping timeout: 245 seconds) 2011-11-14T17:14:13 *** arscan has joined #aichallenge 2011-11-14T17:20:34 YES. I got one-pass collaborative diffusion working. >:] 2011-11-14T17:22:49 *** willem has quit IRC (Quit: leaving) 2011-11-14T17:24:15 are you using that for food, or attacking hills? 2011-11-14T17:26:05 *** jab_bott has joined #aichallenge 2011-11-14T17:26:21 strupq: Everything 2011-11-14T17:26:25 *** mcstar has left #aichallenge ("WeeChat 0.3.6") 2011-11-14T17:27:34 amstan: I added some toggle-buttons to the visualizer in my fork: http://forums.aichallenge.org/viewtopic.php?f=25&t=1633&start=50 2011-11-14T17:28:01 *** BartV_ has quit IRC (Quit: Page closed) 2011-11-14T17:28:03 amstan: so now you can toggle the ai-state visualizations on/off for each player. 2011-11-14T17:28:43 jab_bott: how did you extend the protocol for this? 2011-11-14T17:29:10 *** strupq has quit IRC (Quit: Page closed) 2011-11-14T17:29:11 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-14T17:31:36 amstan: it's described in the forum thread - but basically I modified ants.py so that it filters bot output into 'orders' and 'overlays' arrays, then it parses the orders as normal and parses the overlay commands (starting with v for visualizer instead of o for order) and pass the visualizer commands out in the replay data - then I modified the visualizer code to parse those commands and display the overlays on top of the map 2011-11-14T17:32:56 amstan: do you wnt me to send a pull request so you can see the changes? 2011-11-14T17:33:23 amstan: you can probably see them easily enough in github, just I don't know how :D 2011-11-14T17:33:30 yeah, don't worry about it 2011-11-14T17:33:45 in ants.py, can those changes be easily disabled? 2011-11-14T17:33:59 or easily enabled actually.. they should be disabled by default 2011-11-14T17:34:09 NightExcessive: any replay to see you cd? 2011-11-14T17:34:12 your* 2011-11-14T17:34:24 jab_bott: i just don't know if we should be merging this in the main code 2011-11-14T17:34:53 Anilm3: It's currently broken because I'm toying with the coefficient. I'll get a replay data on the next working one. 2011-11-14T17:35:06 *** ikaros has joined #aichallenge 2011-11-14T17:35:19 amstan: yeah - I think probably best not for now, but seems that people are using it and like it quite a bit so it's worth having a think about what would need to be done to include it 2011-11-14T17:35:46 jab_bott: i can surely give your version a plug in the faq or something 2011-11-14T17:36:05 look through the site, you might see a good place to link your code 2011-11-14T17:36:12 maybe in the upcoming debug section 2011-11-14T17:37:04 aichallenge: Thijs Marinussen epsilon * r2f9b0bb / ants/dist/starter_bots/csharp/GameState.cs : Fix: use TimeSpan.TotalMilliseconds instead of TimeSpan.Milliseconds to measure the time in C# starter bot - http://git.io/VWwg6g 2011-11-14T17:37:05 aichallenge: Janzert epsilon * r97b0c98 / ants/dist/starter_bots/csharp/GameState.cs : 2011-11-14T17:37:06 aichallenge: Merge pull request #368 from Yexo/epsilon 2011-11-14T17:37:06 aichallenge: Fix: Use TotalMilliseconds instead of Milliseconds to measure turn time in C# starter - http://git.io/ENR_vA 2011-11-14T17:37:26 woot! janzert's here 2011-11-14T17:37:35 janzert: i want to talk to you 2011-11-14T17:37:36 sort of just for a moment :) 2011-11-14T17:37:55 yes? 2011-11-14T17:38:13 *** mikewintermute has joined #aichallenge 2011-11-14T17:38:22 wow, that was fast :) 2011-11-14T17:38:27 *** twymer has joined #aichallenge 2011-11-14T17:38:41 :) 2011-11-14T17:38:49 *** hgh has joined #aichallenge 2011-11-14T17:39:11 Anilm3: My biggest problem has been with multiple goals. The point between two (or more) close goals is stronger than the goals themselves- so the ant gets stuck. 2011-11-14T17:39:28 amstan: I think there's more work to make it work in the realtime visualizer too before it's ready to go into the main code... in terms of disabling by default, all it does is allow bot output starting with 'v' and sends it to replay - it would be very easy to disable it by emptying the overlays array or just ignoring it so it keeps the replay clean - but really bots should not output those commands on the server anyway - so you 2011-11-14T17:39:34 *** foRei has quit IRC (Quit: Bye) 2011-11-14T17:39:37 *** hgh has left #aichallenge 2011-11-14T17:40:00 *** bergmark has quit IRC (Read error: Operation timed out) 2011-11-14T17:40:17 NightExcessive: On my paper approach I saw that problem too 2011-11-14T17:40:21 going through the tutorial and the Java code seems to be missing a visible() method. anyone run into this too? 2011-11-14T17:40:53 Anilm3: I've yet to find a good way to fix it. My temporary fix was making the ants diffuse away a negative value. 2011-11-14T17:41:06 Anilm3: have your goal field/heatmap or whatever it is you're using stop when it gets to a stronger field, instead of continuing in an additive fashion 2011-11-14T17:41:31 jab_bott: I'll try that 2011-11-14T17:41:45 jab_bott: that's what I thought 2011-11-14T17:41:55 Anilm3: that way the 'valley' between goals will always have higher-value squares adjacent to it 2011-11-14T17:42:35 jab_bott: are you using cd? 2011-11-14T17:43:24 mikewintermute: i think there is a method called isVisible() 2011-11-14T17:43:51 Anilm3: what's cd? 2011-11-14T17:43:59 collaborative diffusion 2011-11-14T17:44:23 mikewintermute: I may be able to port the C++ visibility check to Java a bit later if you need. 2011-11-14T17:47:13 Anilm3: oh - no I don't thinks so - I'm just making stuff up mostly.. I'm doing ai-class.com so learning some techniques from there, haven't heard of collaborative diffusion though.. sounds like some kind of gaussian blur on a heat-map, which seems like a good idea... I'm currently using something similar to make my ants explore 2011-11-14T17:47:50 jab_bott: what ai-class technique? i haven't seen much in there except chapter 2 that's relevant to ants 2011-11-14T17:47:58 don't have an isVisible either 2011-11-14T17:48:09 amstan: breadth-first and A* search 2011-11-14T17:48:21 yeah, but that's easy 2011-11-14T17:48:37 amstan: heh - yeah but I'm only learning AI for the first time! :D 2011-11-14T17:48:39 *** t2027 has joined #aichallenge 2011-11-14T17:48:40 NightExcessive: cool - thanks. 2011-11-14T17:48:44 mikewintermute: In C++, Bot.cc line 75-107 2011-11-14T17:48:45 jab_bott: same here 2011-11-14T17:48:48 *** underflow has joined #aichallenge 2011-11-14T17:48:53 hi 2011-11-14T17:48:55 looking… 2011-11-14T17:49:01 underflow: hi 2011-11-14T17:50:31 amstan: some of the value iteration stuff looks interesting - I think my food maps which are generated from a breadth first search actually behave a bit like a policy.. tbh, I'm really just making stuff up, struggling with how to make my ants explore without getting stuck atm. I've only got food gathering and basic exploration sorted so far... attack and defence are a way off yet 2011-11-14T17:51:21 NightExcessive: just downloaded C++ starter package. Bot.cc only has 61 lines. 2011-11-14T17:51:39 mikewintermute: Oh, sorry. State.cc. Same lines. 2011-11-14T17:51:57 mikewintermute: remember that you can edit the starter packages, feel free to add the isVisible 2011-11-14T17:52:25 jab_bott: you can do exploring the same way you do food. A breadth first search starting from all unknown (or invisible) tiles 2011-11-14T17:53:11 Anilm3: Check out my exploration visualization here: http://forums.aichallenge.org/viewtopic.php?f=25&t=1633&start=50 - it's based on additive fields around the ants and it doesn't work very well :/ 2011-11-14T17:53:24 how can I debug a python script with aichallenge ? 2011-11-14T17:53:34 (print variables etc...) 2011-11-14T17:53:39 aichallenge is not a debugger 2011-11-14T17:53:50 Yexo: wouldn't that be waaay too slow? there's lots more of those than ants or foods 2011-11-14T17:53:51 you can write to std error and supply --log_error 2011-11-14T17:54:00 that doesn't matter 2011-11-14T17:54:10 there are maximum 200x200 =40k tiles 2011-11-14T17:54:20 you only need to visit each to those once in a BFS 2011-11-14T17:54:33 underflow: you use the little bot icon to, to get the input text you can feed it into your bot and see what it does 2011-11-14T17:54:47 *** Larose has quit IRC (Quit: Leaving.) 2011-11-14T17:56:50 Yexo: bit if you need to do that from all/each unexplored tile - even just the ones on the edge of your expolored range - that's a lot 2011-11-14T17:57:23 one BFS can have multiple source tiles 2011-11-14T17:57:35 you don't need to start a separate BFS for every unknown tile 2011-11-14T17:57:49 Yexo: ahhh, interesting!! 2011-11-14T17:58:01 Yexo: nice point :) 2011-11-14T17:58:06 that is an interesting point 2011-11-14T17:58:26 jab_bott: thx, interesting 2011-11-14T17:58:45 Yexo: but surely if you start on all of them, they're already 'explored' (by the BFS) and so won't be visited again - do you need to pick a selection and search from those? 2011-11-14T17:59:33 no, the bfs doesn't mark tiles as "explorered", it just stores the distances from the closest unknown tile 2011-11-14T17:59:48 ie all unknown tiles get value 1, all tiles at the edge of your visibility get value 1 etc. 2011-11-14T18:00:03 your ants look around them and pick the lowest value 2011-11-14T18:01:04 *** Surya has quit IRC (Quit: Page closed) 2011-11-14T18:01:20 are there any methods that returns enemy ants location in starters? ? i tried with "for row, col in Ants.enemy_ants(self):" but unfortunally error was shown 2011-11-14T18:01:45 IIRC it's ants.enemyAnts() 2011-11-14T18:02:01 Yexo: not sure exactly what you mean, but you've certainly given me some ideas to go and try - I think if I make a list of previously-unexplored but now-visible-and-land tiles that should give me the edge of my explored field and I can start from there 2011-11-14T18:02:35 t2027: don't pass it 'self' as an argument - that's implicitly given as the first argument for class methods 2011-11-14T18:02:53 thestinger: Only if you use a semicolon 2011-11-14T18:03:07 what? no, you never need semicolons in python 2011-11-14T18:03:10 Colon, rather 2011-11-14T18:03:17 I thought he was talking about Lua :x 2011-11-14T18:03:35 @thestinger i tried to use an object of Ants class for exa mple ants, but then some errors come up abour wrong types ;o 2011-11-14T18:03:36 t2027: I'm sorry Dave, err t2027; I cannot 'thestinger'. 2011-11-14T18:03:49 treeform i don't see my print in this file 2011-11-14T18:04:28 t2027: if a class method is defined as 'def method(self, arg):', you use it by passing it 1 argument - the method you're using doesn't take any arguments 2011-11-14T18:04:38 you probably want self.enemy_ants() 2011-11-14T18:05:25 the terminology is confusing here - a classmethod is defined as '@classmethod\n def method(cls, args)' 2011-11-14T18:05:40 yeah, used the wrong term 2011-11-14T18:05:52 should have said 'method' 2011-11-14T18:06:24 but yes, if it's defined as def method(self, ...) in the class, then you call it by self.method(...) 2011-11-14T18:06:53 amstan: If you see mleise can you ask him to help me out in fixing the update/invalidation bug when I toggle the state on/off for the AI state visualizations? http://forums.aichallenge.org/viewtopic.php?f=25&t=1633&start=50#p11458 2011-11-14T18:07:04 *** Accoun has quit IRC (Ping timeout: 240 seconds) 2011-11-14T18:07:06 although Class.method(self, ...) could technically work if you do it right, i don't really recommend it 2011-11-14T18:07:08 @Zannick can i use enemy_ants to search in it for ant_loc? like for ant_loc in self.enemy_ants? 2011-11-14T18:07:09 t2027: An error occured while trying to show the previous error. 2011-11-14T18:07:57 t2027: so, if self.enemy_ants() returns a list, then yes. 2011-11-14T18:08:04 but don't forget the parentheses 2011-11-14T18:08:37 *** Jak_o_Shadows has joined #aichallenge 2011-11-14T18:08:40 @contestbot what do you mean ? its the starter pack and there was created an object of Ants, and they was using it to call method from Ants, but when i tried to do it iit just called some problems with types, i just surrendered and started to call these methods like (static? ) without creating objects 2011-11-14T18:08:41 t2027: Run as fast as you can and don't look back. 2011-11-14T18:08:49 t2027: it's a bot 2011-11-14T18:09:02 t2027: responds to stuff with @ in front 2011-11-14T18:09:05 t2027: you're confusing it by addressing people with @ before their names 2011-11-14T18:09:22 ops. 2011-11-14T18:10:03 jab_bott: use @later tell mleise 2011-11-14T18:10:23 amstan: oh nifty - is that an IRC thing? 2011-11-14T18:10:23 *** Antimony has quit IRC (Ping timeout: 276 seconds) 2011-11-14T18:10:32 jab_bott: contestbot thing 2011-11-14T18:10:37 cool 2011-11-14T18:10:38 he can also do this: 2011-11-14T18:10:40 @rankings 2011-11-14T18:10:41 amstan: Top 10 players: xathis(90.7), GreenTea(86.2), delineate(85.7), MomoBot(85.3), teapotahedron(84.3), GarySWest(83.9), ThisIsNotABug(83.7), Migi32(82.5), peskoj(82.1), Fluxid(81.5) 2011-11-14T18:10:47 he can do lots of things :) 2011-11-14T18:10:49 @source 2011-11-14T18:10:49 thestinger: My source is at http://supybot.com/ 2011-11-14T18:10:53 duh 2011-11-14T18:10:53 *** Conorach_ has quit IRC (Ping timeout: 252 seconds) 2011-11-14T18:11:00 grepping through the source is fun 2011-11-14T18:11:04 thestinger: it's actually gribber 2011-11-14T18:11:10 just now i came back from drinking and got hilited 2011-11-14T18:11:13 ;( 2011-11-14T18:11:44 Fluxid: to fix that, you'll have to drop out of the top ten 2011-11-14T18:11:58 HOW CAN I DO THAT 2011-11-14T18:12:10 @later tell mleise Hiya - can you help me fix a bug: http://forums.aichallenge.org/viewtopic.php?f=25&t=1633&start=50#p11458 - How do I ensure the map gets invalidated/redrawn when playback is stopped (I could pass true to director.draw() but fog toggle doesn't do this, so I'm thinking there's a better way) 2011-11-14T18:12:10 jab_bott: I come to serve. 2011-11-14T18:12:11 myabe i'll reuploqad 2011-11-14T18:12:25 but meh, i'm too drunk to do that 2011-11-14T18:12:29 ;) 2011-11-14T18:12:40 a1k0n: I think you can change the name of your bot now :) 2011-11-14T18:13:17 *** underflow has quit IRC (Ping timeout: 252 seconds) 2011-11-14T18:13:25 what's 'hilited'? 2011-11-14T18:13:33 oh highlighted? 2011-11-14T18:13:42 highlighted, highlit, hilit, etc. 2011-11-14T18:13:49 ahh ok 2011-11-14T18:14:18 I was thinking it's something to do with ant-hills - hillited! :D 2011-11-14T18:14:45 "ranking jab_bott 2011-11-14T18:14:50 jab_bott: afair in irssi the command is hilite 2011-11-14T18:14:53 @ranking jab_bott 2011-11-14T18:14:54 jab_bott: I have no idea what you mean. 2011-11-14T18:15:04 jab_bott: hmm 2011-11-14T18:15:09 jab_bott: good idea 2011-11-14T18:15:14 hilight, actually 2011-11-14T18:15:37 jab_bott: not sure though.. unless there's a json that does user searching 2011-11-14T18:15:44 *** Accoun has joined #aichallenge 2011-11-14T18:15:51 jab_bott: for now it only does this: 2011-11-14T18:15:53 @help rankings 2011-11-14T18:15:53 amstan: (rankings [how many]) -- Get the top $(how many) players. 2011-11-14T18:16:00 *** HaraKiri has quit IRC () 2011-11-14T18:16:15 i hope its limited =) 2011-11-14T18:16:55 oooh, boy, i hope so too 2011-11-14T18:17:07 amstan: rankings [all] | grep -A 5 -B 5 playername 2011-11-14T18:17:07 ? 2011-11-14T18:17:12 hehe 2011-11-14T18:17:37 it makes you do @more if there's a lot of text 2011-11-14T18:17:41 jab_bott: sure... but i don't think there 2011-11-14T18:17:51 there's a way to show all users from the rankings easily 2011-11-14T18:18:01 the bot can only do what the site can 2011-11-14T18:18:11 Ah 2011-11-14T18:18:17 2rankings 2000 2011-11-14T18:18:20 @rankings 2000 2011-11-14T18:18:20 Fluxid: Error: Cannot do that many players. 2011-11-14T18:18:25 @rankings 200 2011-11-14T18:18:25 Fluxid: Error: Cannot do that many players. 2011-11-14T18:18:30 fluxid was brave enought to test it :) 2011-11-14T18:18:30 @rankings 100 2011-11-14T18:18:30 Fluxid: Error: Cannot do that many players. 2011-11-14T18:18:38 @rankings 50 2011-11-14T18:18:38 Fluxid: Error: Cannot do that many players. 2011-11-14T18:18:44 one more chance 2011-11-14T18:18:45 or drunk enough :) 2011-11-14T18:18:48 @rankings 25 2011-11-14T18:18:49 Fluxid: Top 25 players: xathis(90.7), GreenTea(86.2), delineate(85.7), MomoBot(85.3), teapotahedron(84.3), GarySWest(83.9), ThisIsNotABug(83.7), Migi32(82.5), peskoj(82.1), Fluxid(81.5), ifdef(81.5), Komaki(81.5), RVeerdonk(81.3), gcapell(80.9), vittim(80.7), FLashM(80.6), xRaider(80.5), Ender(80.4), pguillory(80.3), olex.s(79.9), xoposhiy(79.8), cumbuz(79.5), BenJackson(79.4), lorill(79.4), (1 more message) 2011-11-14T18:18:52 @more 2011-11-14T18:18:52 Fluxid: bugnuts(79.3) 2011-11-14T18:18:56 lol 2011-11-14T18:19:08 i think it's like 40 2011-11-14T18:19:10 the limit 2011-11-14T18:19:21 maybe 2011-11-14T18:19:34 too drunk to count! 2011-11-14T18:19:52 too drunk to binary search 2011-11-14T18:19:53 at least you can divide 2011-11-14T18:20:04 ;) 2011-11-14T18:20:14 just don't take the server offline 2011-11-14T18:20:30 won't do more command 2011-11-14T18:20:32 s 2011-11-14T18:20:39 cu, gnite 2011-11-14T18:20:41 retybok_: heh. yeah, after i fixed some major bugs it seems to actually work 2011-11-14T18:21:04 i'm still a bit disappointed. 2011-11-14T18:22:07 so is google working on a virtual currency? http://www.reddit.com/r/Bitcoin/comments/mbvn5/accesscoincom_offline_for_good_looks_like_it_plea/c2zp0le 2011-11-14T18:22:22 that would be cool 2011-11-14T18:23:24 Fluxid: i'm becoming more and more in favor of a total wipe 2011-11-14T18:23:47 the rankings are basically meaningless unless you filter out anyone last seen more than a day ago or so 2011-11-14T18:24:15 it's too bad it's not using a whole-history ranking update 2011-11-14T18:24:17 a1k0n: maybe tomorrow i'll do thid 2011-11-14T18:24:22 this* 2011-11-14T18:24:30 might be fraught with peril while drunk 2011-11-14T18:24:43 actually, today, as it is 00:24 here 2011-11-14T18:26:52 amstan: got to laugh at this comment: "So, a worthless, manipulated currency. Gotcha." - so just the same as traditional fiat currencies then 2011-11-14T18:27:04 hahaha 2011-11-14T18:27:44 jab_bott: well.. it's just that it would be nice to have properly backed bitcoin 2011-11-14T18:28:10 amstan: isn't the main benefit of bitcoin that it's not 'properly backed'? 2011-11-14T18:28:24 by backed i mean supported 2011-11-14T18:28:29 and used 2011-11-14T18:28:37 having a large company back it is not the intent at all of a distributed currency 2011-11-14T18:28:54 amstan: bitcoin is used... for currency speculation :P 2011-11-14T18:29:14 as in.. get bitcoins for adsense, offer bitcoins via google checkout and a bunch of others 2011-11-14T18:29:35 I think most of the people 'using' it are just mining and selling them to people who think the value will go up :\ 2011-11-14T18:29:44 a1k0n: at least your bot's behavior is pretty original compared to others 2011-11-14T18:29:45 bitcoins for ads would be...interesting 2011-11-14T18:29:45 *** oliv3 has left #aichallenge 2011-11-14T18:30:07 thestinger: not necessarly, you can use it just fine to transfer money to other people, at like 5% losses 2011-11-14T18:30:09 amstan: The probably I see with Google getting involved is their love for money (as any company has) combined with their simply massive computing power. 2011-11-14T18:30:17 s/probably/problem/ 2011-11-14T18:30:56 yeah - they'd just devise one they can track and control 2011-11-14T18:31:44 Well if they wanted to they could mine Bitcoins at insane rates. 2011-11-14T18:31:54 *** yoden has joined #aichallenge 2011-11-14T18:31:58 WebCL seems like it will be a great way to mine bitcoins - just put some code on your site and all your visitors are mining for you 2011-11-14T18:32:05 *** Sir_Ragnarok_ has quit IRC (Read error: Connection reset by peer) 2011-11-14T18:32:28 you know what we should do.. 2011-11-14T18:32:36 have you seen this: http://www.eudebtwriteoff.com/ - seems like a pretty good idea to me 2011-11-14T18:32:48 design a contest that looks like an ordinary game, but in reality is a way to mine for bitcoins 2011-11-14T18:32:56 it could pay for itself 2011-11-14T18:33:14 *** mceier has quit IRC (Quit: leaving) 2011-11-14T18:33:41 a1k0n: is it true you got into google thanks to ai contest? 2011-11-14T18:33:49 *** Sir_Ragnarok has joined #aichallenge 2011-11-14T18:33:58 he was at yahoo before, so you can only assume it had to do with it 2011-11-14T18:34:22 problem is, "The Economy" only works by unsustainable exponential growth, and sensible ideas like eudebtwriteoff.com are against the grain on this - the banks who skim off millions on all those interest payments floating around and being converted to different currencies would never allow it 2011-11-14T18:34:30 thestinger: yeah, but bitcoins lose sense 2011-11-14T18:34:33 *** capa has joined #aichallenge 2011-11-14T18:40:00 *** retybok_ has quit IRC (Ping timeout: 260 seconds) 2011-11-14T18:40:28 jab_bott: Your fix of ignoring CB on tiles with a higher value didn't work. http://dl.dropbox.com/u/3166522/screenshots/2011/11/Screenshot-2011-11-14_18.39.32.png 2011-11-14T18:41:17 jab_bott: you know where your visualizer would be useful? 2011-11-14T18:41:21 well.. your patches 2011-11-14T18:41:24 paste.aichallenge.org 2011-11-14T18:41:36 that way NightExcessive could have pasted his thing there 2011-11-14T18:41:49 Hehe. 2011-11-14T18:42:07 NightExcessive: what the hell is this 2011-11-14T18:42:39 Fluxid: Collaborative diffusion... as done by someone with no AI experience. 2011-11-14T18:42:57 how did you render this 2011-11-14T18:43:03 *** capa has quit IRC (Remote host closed the connection) 2011-11-14T18:43:08 jab_bott's visualizer patches. 2011-11-14T18:43:14 *** capa has joined #aichallenge 2011-11-14T18:43:18 * nickjohnson has found CD works really well for finding food, not so well for finding enemy hives :/ 2011-11-14T18:43:23 O-o 2011-11-14T18:43:49 Fluxid: it lets you give 'v' commands to the visualizer to draw stuff (instead of just 'o' for orders) 2011-11-14T18:43:55 amstan: um yeah - good plan! 2011-11-14T18:44:22 thestinger: link to fork or something? 2011-11-14T18:45:02 dude 2011-11-14T18:45:15 Fluxid: http://forums.aichallenge.org/viewtopic.php?f=25&t=1633&start=50#p11458 2011-11-14T18:45:23 There's the link to the latest commit 2011-11-14T18:45:32 Fluxid: well, basically true, yes. someone referred me 2011-11-14T18:45:44 sorry - I mean that's the forum thread - that post has the link to the latest commit and the earlier ones explain how to use it 2011-11-14T18:47:22 a1k0n: cool, though i'm satissfied whith my current job ;D 2011-11-14T18:47:27 jab_bott: cool! 2011-11-14T18:47:59 *** Accoun has quit IRC (Ping timeout: 248 seconds) 2011-11-14T18:48:07 i saw sir_macelon and lardop reuploaded their bots... 2011-11-14T18:48:16 hardcore is coming 2011-11-14T18:48:39 anyway, i said before i'm going to sleep, so i;m going really to sleep rigth now 2011-11-14T18:48:42 cu 2011-11-14T18:48:46 night 2011-11-14T18:49:05 im gonna roll a new bot once i can put this baby down 2011-11-14T18:49:24 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-14T18:50:52 *** mj41 has quit IRC (Ping timeout: 245 seconds) 2011-11-14T18:52:59 *** yahim91 has quit IRC (Quit: Page closed) 2011-11-14T18:55:02 *** datachomper has joined #aichallenge 2011-11-14T18:55:25 *** xathis has joined #aichallenge 2011-11-14T18:55:35 *** Harpyon has quit IRC (Quit: Computer has gone to sleep.) 2011-11-14T18:57:52 *** xathis_ has quit IRC (Ping timeout: 240 seconds) 2011-11-14T18:57:55 *** Antimony has joined #aichallenge 2011-11-14T19:00:41 *** Accoun has joined #aichallenge 2011-11-14T19:02:48 *** Redgis has quit IRC (Quit: ... mains libres) 2011-11-14T19:05:36 * nickjohnson imagines a1k0n carrying a baby while programming 2011-11-14T19:06:06 *** Cyb has quit IRC (Ping timeout: 252 seconds) 2011-11-14T19:09:10 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-11-14T19:09:52 *** Accoun has quit IRC (Ping timeout: 240 seconds) 2011-11-14T19:10:41 *** amstan_ has joined #aichallenge 2011-11-14T19:10:41 *** ChanServ sets mode: +o amstan_ 2011-11-14T19:12:25 *** grom358 has joined #aichallenge 2011-11-14T19:13:01 falling behind in my bot.. lot of bots are getting quite good 2011-11-14T19:13:48 *** iwantwin has joined #aichallenge 2011-11-14T19:14:42 can anyone help me with editing my make.cmd, so i can use it to compile java files that are in need of a package? 2011-11-14T19:14:42 *** lavalamp has quit IRC (Quit: Page closed) 2011-11-14T19:15:02 iwantwin: are you using the starter kit from git? 2011-11-14T19:15:05 iwantwin: don't use make.cmd, get yourself make, the real make 2011-11-14T19:15:09 iwantwin: and use the makefile 2011-11-14T19:15:15 *** amstan_ is now known as amstan 2011-11-14T19:15:17 i am using the starter kit yeah 2011-11-14T19:15:24 make.cmd works fine from git. I been using it on my windows box 2011-11-14T19:15:24 how can i use the makefile then? 2011-11-14T19:15:34 you need to get make 2011-11-14T19:15:56 i don't get what is your issue with make.cmd?? 2011-11-14T19:16:03 what is it not doing? 2011-11-14T19:16:31 *** datachomper has quit IRC (Quit: Leaving.) 2011-11-14T19:17:17 well, grom.. when i call to other classes in myBot.java, for example when i call Tile, it gives error: cannot find symbol.. but it only gives this error when i give every java class a package definition 2011-11-14T19:17:48 when putting it in default package (or, no package definition) make.cmd works perfectly fine 2011-11-14T19:18:03 why do u want to put it in package? 2011-11-14T19:18:26 if i don't my netbeans IDE gives errors.. 2011-11-14T19:18:35 Which is crappy 2011-11-14T19:19:18 yeah.. you should be able to use default package.. well if you using netbeans.. Why not just use Ant 2011-11-14T19:19:41 IIRC its built into netbeans 2011-11-14T19:20:06 huh im not keepnig up now.. 2011-11-14T19:20:50 forget using make.cmd or the Makefile.. Get netbeans to build for you 2011-11-14T19:21:14 *** mikewintermute has quit IRC (Quit: mikewintermute) 2011-11-14T19:22:00 *** Akranis has quit IRC (Ping timeout: 260 seconds) 2011-11-14T19:22:26 uhm.. i feel real stupid now:) 2011-11-14T19:23:22 are you new to netbeans? 2011-11-14T19:24:46 only thing though.. I just realised you probably need working Makefile so it can be built on the servers when uploading the code. 2011-11-14T19:25:35 *** analyst74 has joined #aichallenge 2011-11-14T19:25:37 Im not new to netbeans.. I just didnt think of it:p 2011-11-14T19:25:48 Thats why i feel so stupid^^ 2011-11-14T19:26:03 How do i use the working makefile then/ 2011-11-14T19:26:24 grom358: you don't need a makefile for the servers 2011-11-14T19:26:34 uhm wait 2011-11-14T19:26:44 they do their own thing 2011-11-14T19:27:20 i cant run a game when i compile it with netbeans? while i can when i use make.cmd? 2011-11-14T19:27:36 amstan: so you could upload without a makefile and it works out how to build it? 2011-11-14T19:27:58 iwantwin: you should be able to, if you make a executable jar 2011-11-14T19:28:30 iwantwin: have manifest that specifys the main class 2011-11-14T19:29:39 grom358: yes, it'll just put all files together 2011-11-14T19:29:52 grom358: see worker/compiler.py 2011-11-14T19:30:29 okay 2011-11-14T19:30:35 thanks 2011-11-14T19:30:43 i think ill figure something out now:) 2011-11-14T19:32:57 ah I see.. yeah so your main file has to be called MyBot.java 2011-11-14T19:33:53 it kind of looks like you can't have folders.. and therefore packages .. Could be wrong.. only skimming through this code 2011-11-14T19:35:23 would explain the problem i had=P 2011-11-14T19:36:10 im just moving it in package while programming and moving it in default package for making.. i guess 2011-11-14T19:36:19 netbeans compile sure doesnt work 2011-11-14T19:36:29 *** datachomper has joined #aichallenge 2011-11-14T19:36:30 or it might in default packageXD 2011-11-14T19:36:32 iwantwin: something seems wrong. I don't recall netbeans forcing packages 2011-11-14T19:36:41 *** Akranis has joined #aichallenge 2011-11-14T19:36:53 you should be able to use the default package in netbeans 2011-11-14T19:37:13 yeah i can, but it gives errors because Tools and other stuff is in the same default package.. 2011-11-14T19:37:29 huh? what tools? 2011-11-14T19:37:31 unless you know a way to make my netbeans ignore certain maps or something? 2011-11-14T19:37:47 *** datachomper has quit IRC (Client Quit) 2011-11-14T19:37:56 aichallenge tools, the one you have to download for samplebots and stuff 2011-11-14T19:38:25 oh.. I have a java folder where I got my bot code.. seperate from the other folders.. and just make that the source folder 2011-11-14T19:38:53 netbeans only needs to know about the java folder 2011-11-14T19:39:06 *** t2027 has quit IRC (Ping timeout: 265 seconds) 2011-11-14T19:39:34 yeah but that forces me to change the make/run.cmd's 2011-11-14T19:39:54 hmm let me try something 2011-11-14T19:40:10 no it don't 2011-11-14T19:40:19 you have folder with your *.java files and make.cmd 2011-11-14T19:41:01 so where do you put makefile, version, and stuff like that? 2011-11-14T19:41:39 I just go in the java folder and run make.cmd .. And then I got MyBot.jar which I use to run my bot against the other sample bots 2011-11-14T19:41:54 WOO. Got a near perfect round from my CB. FINALLY. :] 2011-11-14T19:42:18 NightExcessive: what do you mean by perfect round? 2011-11-14T19:42:37 The AI performed as it should (at the moment it only collects food) 2011-11-14T19:42:44 but i use make.cmd and run.cmd from the same commandline.. if i put them in seperate directories that makes my testing harder=P 2011-11-14T19:42:53 aah nice night! 2011-11-14T19:43:05 iwantwin: I have make.cmd in the folder with the *.java files 2011-11-14T19:43:16 yeah me to 2011-11-14T19:43:22 iwantwin: no changes require to make to that file 2011-11-14T19:43:28 okay 2011-11-14T19:43:58 but when i put run.cmd in the .java directory i cant acces tools:p 2011-11-14T19:44:01 http://paste.aichallenge.org/FRhLu/ <- You'll note it doesn't do anything yet except collect the food it can see... and my goal obfuscation is nowhere near good. 2011-11-14T19:44:49 iwantwin: yes you can.. ..\tools or just have run.cmd be in folder where tools is and have java -jar java/MyBot.jar 2011-11-14T19:45:19 the second option is what I'm doing.. works fine 2011-11-14T19:45:50 lets take a look night:D 2011-11-14T19:46:06 the second option will give me harder time testing it 2011-11-14T19:46:33 1st option isnt working: "python tools/sample_bots/python/GreedyBot.py" change -> "python ../tools/sample_bots/python/GreedyBot.py" 2011-11-14T19:47:01 *** u_ has left #aichallenge 2011-11-14T19:47:12 NightExcessive: what map is that? 2011-11-14T19:47:21 The tutorial map 2011-11-14T19:47:23 example1 2011-11-14T19:47:34 okay night^^ 2011-11-14T19:47:34 ok i guess i did not follow any tutorials 2011-11-14T19:47:48 it does work indeed 2011-11-14T19:47:51 Oh maps/example/tutorial1.map 2011-11-14T19:48:13 *** u_ has joined #aichallenge 2011-11-14T19:49:08 @grom will python translate ../ the right way? 2011-11-14T19:49:10 iwantwin: You have no gotten any error messages recently, so here's a random one just to let you know that we care. 2011-11-14T19:49:31 lol 2011-11-14T19:49:33 iwantwin: Yes, it will 2011-11-14T19:49:41 why winter and fall? 2011-11-14T19:50:14 i even tried hardcoding the path.. 2011-11-14T19:50:51 python: can't open file 'tools/playgame.py': [Errno 2] No such file or directory 2011-11-14T19:51:57 haha 2011-11-14T19:51:58 okay 2011-11-14T19:51:59 stupid 2011-11-14T19:52:11 run.cmd uses multiple paths to tools -_- 2011-11-14T19:52:16 i forgot one(A) 2011-11-14T19:52:17 and will there be a winter 2011? 2011-11-14T19:52:44 thanks for help grom and night:D 2011-11-14T19:53:05 and @amstan^^ 2011-11-14T19:53:18 any time 2011-11-14T19:54:00 How should I go about goal obfuscation? (Making the ants "vacuum the scents" to prevent multiple ants from engaging in one objective) 2011-11-14T19:54:40 NightExcessive: what if the ant you originally sent for food dies? 2011-11-14T19:54:58 amstan: Its vacuum will no longer exist, so another ant will grab the goal. 2011-11-14T19:55:55 *** jab_bott has left #aichallenge 2011-11-14T19:59:58 Wrote my own line algorithm... and then found out about Bresenham's line algorithm. 2011-11-14T20:00:52 the way I do it is 2011-11-14T20:01:02 when expanding a node near the ant, increase the distance by more than 1 2011-11-14T20:02:15 So essentially lower the coefficient by small amounts? 2011-11-14T20:02:26 Antimony: that's what I do atm, I increase it by 3 instead of 1 2011-11-14T20:02:59 I use the algorithm from the published article, so it's not currently directly dependent on any kind of distances :x 2011-11-14T20:03:20 what article? You arne't talking about that stupid diffusion thing are you? 2011-11-14T20:03:40 Indeed I am. http://www.cs.colorado.edu/~ralex/papers/PDF/OOPSLA06antiobjects.pdf 2011-11-14T20:04:57 by the way, the actual term for that is Influence Maps, but the authors were apparently not aware of common AI practice 2011-11-14T20:05:54 *** NightExcessive_ has joined #aichallenge 2011-11-14T20:06:02 Antimony: what do you mean increase distance by more then 1? 2011-11-14T20:06:06 by* 2011-11-14T20:06:36 <- Has no AI experience either :] 2011-11-14T20:07:33 wee http://www.slideshare.net/mobius.cn/influence-map 2011-11-14T20:08:02 *** delt0r_ has quit IRC (Ping timeout: 276 seconds) 2011-11-14T20:08:06 *** arscan has quit IRC (Ping timeout: 265 seconds) 2011-11-14T20:08:06 *** NightExcessive has quit IRC (Ping timeout: 265 seconds) 2011-11-14T20:09:33 *** Zaphus has quit IRC (Ping timeout: 265 seconds) 2011-11-14T20:09:33 *** LouisMartin has quit IRC (Ping timeout: 265 seconds) 2011-11-14T20:09:59 *** Antimony has quit IRC (Ping timeout: 276 seconds) 2011-11-14T20:10:21 thestinger: what did he and you meant by encreasing the distance? 2011-11-14T20:16:50 *** xathis has joined #aichallenge 2011-11-14T20:20:04 *** delt0r_ has joined #aichallenge 2011-11-14T20:25:06 *** treeform has quit IRC (Remote host closed the connection) 2011-11-14T20:30:51 yeah NightExcessive, same here. That why been taking part in this challenge 2011-11-14T20:33:05 got A* and BFS going 2011-11-14T20:33:20 and my ant does reasonable.. same of the times 2011-11-14T20:33:49 I recently changed to trade ant for an ant... lots like my bot doing worse 2011-11-14T20:34:49 grom: trade ant for an ant? what you mean by that? 2011-11-14T20:35:55 like.. I have code where an ant will avoid other ants.. But now if its only one ant.. I will still move towards my goal.. so if that ant doesn't move away.. Means both ants die 2011-11-14T20:38:48 well not sure if it is doing worse yet.. need more sample games.. My rank still increasing.. Yet to see if it gets higher than my previous version 2011-11-14T20:39:02 rank 296 and climbing 2011-11-14T20:39:16 with zero combat 2011-11-14T20:40:03 avoidance is a type of combat 2011-11-14T20:43:49 yeah.. I guess it kind of is. But its the weakest form 2011-11-14T20:44:08 I need to like get my other ants to group up and kill them off 2011-11-14T20:45:02 try and make it so get the food and not lose ants in the process 2011-11-14T20:45:14 yeah 2011-11-14T20:45:14 *** Antimony has joined #aichallenge 2011-11-14T20:45:15 *** grwip has quit IRC (Quit: Leaving) 2011-11-14T20:45:32 im not that experienced with ai 2011-11-14T20:45:37 and my ants tend to go single file towards enemy hills 2011-11-14T20:45:48 but ive been using a* before and have some ideas for strategies 2011-11-14T20:45:55 so I can lose 30 ants to good placement by the enemy 2011-11-14T20:45:57 i even tried to put my ideas at use at first.. 2011-11-14T20:46:24 iwantwin: I got A* implement.. but its barely used so far.. cause A* requires start and end points 2011-11-14T20:46:40 but now im first putting a lot of effort trying to make the starterpack usefull or more object oriented:D 2011-11-14T20:46:54 in either case im storing more data then before^^ 2011-11-14T20:47:01 and gameplay is still the same=D 2011-11-14T20:47:15 ehm.. we do have start and end poitns right? 2011-11-14T20:47:28 Yeah I got heaps of state that I store.. water tiles, when last seen a tile, etc etc 2011-11-14T20:47:58 what I really need though, is getting my ants to work together in combat 2011-11-14T20:48:07 lolXDF 2011-11-14T20:48:18 dont be so hard on yourself:D 2011-11-14T20:48:22 and get it working^^ 2011-11-14T20:48:22 haha 2011-11-14T20:48:26 its complicated though 2011-11-14T20:48:37 then make it simple :p 2011-11-14T20:48:42 *** Anilm3 has quit IRC (Ping timeout: 265 seconds) 2011-11-14T20:48:57 yeah.. the thing with combat.. you sort of need to predict the opponents moves 2011-11-14T20:49:27 * avdg would think about that if he is writing code to do that 2011-11-14T20:49:55 yeah thats hard grom 2011-11-14T20:50:24 * avdg was thinking at scanning behavior of enemies, but in a simple way 2011-11-14T20:51:17 but we have to start with what we know already 2011-11-14T20:51:18 well I was thinking of not really doing it.. and just brute force attack 2011-11-14T20:51:32 like.. use 4-5 ants against 2 enemy ants 2011-11-14T20:51:47 well, I am trying to avoid useless things as much as possible 2011-11-14T20:52:04 though you can really see that in my current bot 2011-11-14T20:52:10 *can' 2011-11-14T20:52:11 t 2011-11-14T20:52:18 *** Accoun has joined #aichallenge 2011-11-14T20:52:20 where did I learn typing :/ 2011-11-14T20:54:17 *** KP13 has joined #aichallenge 2011-11-14T20:54:45 meh, my bot won a few games lonely, thats rare good news :p 2011-11-14T20:55:21 Minimax/alpha-beta would probably work well for combat scenarios 2011-11-14T20:56:13 *** Kingpin13 has quit IRC (Ping timeout: 240 seconds) 2011-11-14T20:57:05 *** Rinum has joined #aichallenge 2011-11-14T20:57:46 Huh. The C# starter pack has duplicate entries for ants. 2011-11-14T20:57:53 (Friendly ants) 2011-11-14T20:58:03 $1 says xathis wins this year 2011-11-14T20:58:47 I already heard migi32 complaining that the bots are getting better 2011-11-14T20:58:59 and his bot was getting 8th 2011-11-14T20:59:25 at least those in the top tier 2011-11-14T20:59:25 the bots are definitely getting better 2011-11-14T20:59:25 lol 2011-11-14T20:59:46 maybe, check the ones who are updated 2011-11-14T21:00:23 hmm, there are more active ones atm 2011-11-14T21:00:28 so that situation changed 2011-11-14T21:03:03 i wish I knew what techinques some of the top bots where using 2011-11-14T21:03:12 yeah, I'm starting to think that I've got no shot at reaching (and maintaining a spot in) the top 100 x.x 2011-11-14T21:03:16 idk where to even start for this combat stuff 2011-11-14T21:03:22 heh, I'm more worried about getting my code written 2011-11-14T21:03:23 grom358: tell me about it x.x 2011-11-14T21:03:54 I have no problems writing my ideas into code... I'm just in need of better ideas 2011-11-14T21:04:05 *** KP13 has quit IRC (Ping timeout: 260 seconds) 2011-11-14T21:04:09 Rinum: yeah I thinking my latest version might crack top 100 on aichallenge site.. But the bots at ants.fluxid.pl seem to be more advanced 2011-11-14T21:04:23 even mine? :p 2011-11-14T21:04:27 and as newer version are uploaded gonna get smashed 2011-11-14T21:04:32 its almost always losing 2011-11-14T21:04:34 grom358: yeah, fluxid bots are waaay too good 2011-11-14T21:04:58 well, its still a good benchmark if you don't get too emotional about it 2011-11-14T21:05:34 based on my exp w/ fluxid, a skill ~37 results in skill ~63 on aichallenge... at least based on my bot 2011-11-14T21:05:56 mine does just hit 30 2011-11-14T21:06:03 so I guess if you can manage to get a skill ~45-ish, you should be top 100 material 2011-11-14T21:09:06 *** datachomper has joined #aichallenge 2011-11-14T21:11:24 i did it again.. 2011-11-14T21:11:29 i stayed up till 3 am.... 2011-11-14T21:11:39 just because im to enslaved to ants:p 2011-11-14T21:11:47 I wonder how many times I stayed up till 7 am :p 2011-11-14T21:12:01 I used to stay up working on this too 2011-11-14T21:12:07 avdg: I have done it yesterday and will do it today again 2011-11-14T21:12:07 yeah lol 2011-11-14T21:12:11 then I realized I wasn't making progress 2011-11-14T21:12:12 *** bmh has joined #aichallenge 2011-11-14T21:12:12 *** bmh has joined #aichallenge 2011-11-14T21:12:13 but its better if itake sleepxD 2011-11-14T21:12:17 but not working on this xD 2011-11-14T21:12:27 I'm making more progress the last days 2011-11-14T21:12:31 haha 2011-11-14T21:12:32 guys 2011-11-14T21:12:34 so I've been taking breaks every few days 2011-11-14T21:12:34 im going off 2011-11-14T21:12:36 cya later! 2011-11-14T21:12:42 avdg: me too! We finally got collaborative diffusion to work 2011-11-14T21:12:44 but thats because I solved a (for me) complex problem 2011-11-14T21:12:53 I don't use diffusion 2011-11-14T21:13:07 though it's a nice system 2011-11-14T21:13:09 I use diffusion 2011-11-14T21:13:17 bmh: I've almost got it working! Having problems with the values being higher between goals than at the goal. :l 2011-11-14T21:13:18 no fit in my brain's ideas atm 2011-11-14T21:13:20 I'm still trying to work combat into diffusion though 2011-11-14T21:13:38 Rinum: have you looked at pguillory? 2011-11-14T21:13:51 pedrosorio: not lately 2011-11-14T21:13:54 well, diffusion has its own problems, and I'm too lazy to learn new consequences 2011-11-14T21:13:54 (related to the $1 - xathis) 2011-11-14T21:13:57 Rinum: yeah… count how many enemy ants will be within the fight radius and how many friendlies. If there are more enemy ants, run away 2011-11-14T21:14:07 yeah Ants has been affecting my sleep too lol 2011-11-14T21:14:11 It's not particularly bright. 2011-11-14T21:14:15 I haven't had time to touch my bot in weeks :( 2011-11-14T21:14:21 bmh: right, but coordinating them into formation 2011-11-14T21:14:48 Rinum: sure. I'm going to modify my ants to be state machines and have an 'infectious' warband state that spreads to adjacent ants 2011-11-14T21:15:14 nice 2011-11-14T21:15:18 I've done 8 commits the last 24 hours 2011-11-14T21:15:27 and have 89 atm 2011-11-14T21:15:36 whoa 2011-11-14T21:16:04 but most of them are very small changes or cleanup 2011-11-14T21:16:12 we're at 98 2011-11-14T21:16:30 i have done 1:D 2011-11-14T21:16:40 but will do one soon^^ 2011-11-14T21:16:41 pedrosorio: well I'm sure xathis has a better bot in the works (or already built) 2011-11-14T21:16:46 iwantwin: hmm, I wonder if your commit is actually too big 2011-11-14T21:17:08 Rinum: i'm not sure about that... 2011-11-14T21:17:20 and if I commit, I only commit parts of a file using "git add -i" 2011-11-14T21:17:33 Rinum: everytime i touch it, it plays worse :D 2011-11-14T21:17:36 avdg: my first commit was just after i completed the tutorialXD 2011-11-14T21:17:40 xathis: =) gives us hope! 2011-11-14T21:17:41 xD 2011-11-14T21:17:43 xathis: What magic are you using? 2011-11-14T21:17:46 xathis: nice one 2011-11-14T21:18:03 avdg: but i will commit a few times tomorrow 2011-11-14T21:18:07 it's really terrible when you have an awesome thing, tweak it, but then can't get it to do any better 2011-11-14T21:18:08 avdg: with real changes 2011-11-14T21:18:28 iwantwin: don't believe in real changes, my commits rarely have real changes ;-) 2011-11-14T21:18:36 I kind of want to try and make a real bot 2011-11-14T21:18:40 xathis: btw, congrats on maintaining your solid lead thus far! 2011-11-14T21:18:45 the only thing that counts is that you are committing regularly 2011-11-14T21:18:45 avdg: every change improves the tutorial bot:O 2011-11-14T21:18:56 bmh: no magic, just a lot of bfs searches and some combat logic 2011-11-14T21:18:57 but I'm too busy trying to finish my thesis and develop an Octave bot that doesn't time out 2011-11-14T21:19:16 iwantwin: yeah, but doesn't mean your bot actually get better by every commit, just on long term 2011-11-14T21:19:18 xathis: "and some combat logic" 2011-11-14T21:19:28 avdg: true haha 2011-11-14T21:19:33 bmh: "and some combat magic" 2011-11-14T21:19:35 The tutorial bot is terrible 2011-11-14T21:19:51 grom358: wakarimashite (japanese for i know) 2011-11-14T21:20:02 grom358: it still gave me enough instructions to say I want from my bot 2011-11-14T21:20:11 *what I 2011-11-14T21:20:20 xathis: are there any particular users you are worried about? 2011-11-14T21:20:31 but ill be gonig off now for real 2011-11-14T21:20:35 * iwantwin waves goodnight 2011-11-14T21:20:44 pedrosorio: i actually found two big bugs in my combat and i have a version that can handle much more ants, but it looks like the bugs are part of the caracter of my bot, fixing them doesn't necessarily result in better games 2011-11-14T21:20:46 gn 2011-11-14T21:21:27 Rinum: a lot of the other top ones are dangerous 2011-11-14T21:21:28 xathis: exactly, bug fixing that turns out to be "feature removing" =P 2011-11-14T21:21:37 xathis: I wonder if you reached a point where you have to make more decisions between tasks 2011-11-14T21:21:48 *** iwantwin has quit IRC (Quit: Page closed) 2011-11-14T21:21:57 xathis: yet as far as I remember you've always had ~5 pts of distance from them 2011-11-14T21:23:24 pedrosorio: yeah, but if you look at the mu (hover over the skill on the rankings page) it's only about 3 2011-11-14T21:23:40 <_flag> xathis: Why'd you take your bot off tcp? 2011-11-14T21:23:49 avdg: what do you mean, which tasks? 2011-11-14T21:23:59 _flag: i'm working on it atm 2011-11-14T21:24:22 <_flag> xathis: I see, don't work too hard, it's hard enough keeping up as it is :P 2011-11-14T21:24:39 well, that your bot requires better and tougher decisions 2011-11-14T21:24:42 xathis: do you cache path information? or recalculate that sort of thing every turn. 2011-11-14T21:24:50 _flag: but you're doing good on tcp :) 2011-11-14T21:25:47 grom358: i save destination information between the turns, but calculate the path again every turn. but not all of my ants go to some destination 2011-11-14T21:26:40 <_flag> xathis: Thanks, but when trying to improve it helps to be doing bad, which is why I was wondering why you took your bot off :) 2011-11-14T21:26:51 *** bmh has quit IRC (Quit: bmh) 2011-11-14T21:27:13 hmm, I wonder how fast the bots of you guys are running every turn, mine doesn't reach 30 ms atm (but its also almost doing nothing) 2011-11-14T21:27:48 mine does nothing but move all its ants west and it takes 20 ms for one ant... 2011-11-14T21:28:00 My CD bot was taking in excess of 100ms every turn before I gave up on it. 2011-11-14T21:28:08 well, mine can do 400 in that time span 2011-11-14T21:28:26 but again, its almost doing nothing, except some information tracking 2011-11-14T21:28:30 _flag: you're right, i just didn't want my bot to run in the background while developing 2011-11-14T21:28:34 my bot is around 10-30 ms a turn 2011-11-14T21:28:39 u_: that's too long... what language are you using? 2011-11-14T21:28:41 on my macbook 2011-11-14T21:28:43 actually, it's 1/5 of a ms 2011-11-14T21:28:55 <_flag> xathis: Understandable, I have a few cores so it doesn't effect me as much 2011-11-14T21:28:56 ...i just finished it a few seconds ago :P 2011-11-14T21:29:11 xathis: I'm just running my bots to see if I made a syntax error or broke my bot :p 2011-11-14T21:30:50 avgd: yeah, before uploading to the official server, i will test on the tcp servers extensivly as well 2011-11-14T21:30:59 *** bmh has joined #aichallenge 2011-11-14T21:30:59 *** bmh has joined #aichallenge 2011-11-14T21:31:31 meh, the longer I don't run my bots, the more errors I have to fix, so I just do it as regularly as possible while developing 2011-11-14T21:31:52 thats also why I sometimes get frustrated if I see a slow bot actually :p 2011-11-14T21:31:56 but meh :-) 2011-11-14T21:32:44 the time used by my bot is highly dependent on the situation 2011-11-14T21:32:56 *** datachomper has quit IRC (Quit: Leaving.) 2011-11-14T21:33:22 *** GeorgeSebastian has joined #aichallenge 2011-11-14T21:33:26 Antimony: I don't want that to happen with my bot ;-) 2011-11-14T21:33:40 why not? 2011-11-14T21:33:52 if it doesn't scale, its borken ;-) (unless you don't need the scale) 2011-11-14T21:34:20 xathis: Is your bot state-based? 2011-11-14T21:34:28 and the last thing I want to see except a crash is a time out 2011-11-14T21:35:03 NightExcessive_: you mean if i save information in between turns? 2011-11-14T21:35:34 xathis: Well, do you give each bot a state (role?) 2011-11-14T21:35:55 currently, in the last test my bot peaked around 50ms without doing anythign fancy 2011-11-14T21:36:14 it could be a lot more once I get combat implemented 2011-11-14T21:36:14 hmm, mine idles at 0/1ms 2011-11-14T21:36:16 Pro-tip: Telling your bots to run toward your hill is generally not what you want them to do. 2011-11-14T21:36:29 NightExcessive_: ah, uhm not really, at the beginning of each turn i decide for each ant what to do 2011-11-14T21:36:43 xathis: Ah. 2011-11-14T21:36:53 combat is hard 2011-11-14T21:37:04 i can't think of a good way to do it besides pasting together hacks and heuristics :( 2011-11-14T21:37:07 interesting 2011-11-14T21:37:29 yoden: information tracking? 2011-11-14T21:37:33 yoden: a1k0n fully solved some positions :) 2011-11-14T21:37:40 most of the 50ms seems to be spent doing routine updates 2011-11-14T21:37:57 NightExcessive_: but if the ant has nothing special to do and had a destination given in a prior turn, it follows that route 2011-11-14T21:38:04 well, fully solving or trying to do some sort of alpha-beta... 2011-11-14T21:38:07 doesn't really scale :\ 2011-11-14T21:38:21 *** GeorgeSebastian has quit IRC (Read error: Connection reset by peer) 2011-11-14T21:38:27 xathis: Ahh. Do you re-calculate paths every turn? 2011-11-14T21:38:28 my plan for combat is to assume all enemies are stationary 2011-11-14T21:38:34 it's better than nothing 2011-11-14T21:38:46 but I haven't even got that implemented yet 2011-11-14T21:38:49 xathis: I'm thinking of keeping around my old diffusion grid and adding it to the new grid after penalizing the whole thing 2011-11-14T21:38:50 if your gathering is good enough maybe you can time them all out. 2011-11-14T21:39:09 *** datachomper has joined #aichallenge 2011-11-14T21:39:11 Antimony: yep, I started working on that but it's not done, keep getting distracted by random tweaks, bugs and tiny features 2011-11-14T21:39:20 NightExcessive_: yes, even if the destination stays the same, i recalculate the path via A*. I don't have any time problems, even with long distance searches 2011-11-14T21:39:27 I recalculate food dsitances every turn. It usually takes up to 10ms depending on how much searching needs to be done 2011-11-14T21:39:36 thestinger: do you test your bot regularly? 2011-11-14T21:39:45 yeah, a good a* impl is so fast on these small boards, no reason to store paths 2011-11-14T21:39:47 avdg: yes, I'm strcat :) 2011-11-14T21:40:01 avdg: I implement every new feature/change in a branch, and test it before merging 2011-11-14T21:40:05 cool, my ant actually plays well until it times out 2011-11-14T21:40:06 I find that TCP is not very useful for testing. There's not enough food and too many timeouts 2011-11-14T21:40:08 I've tested it hard enough so I can't get new bugs easily without modifying the core code 2011-11-14T21:40:13 combat isn't done, so I can't really test much yet :P 2011-11-14T21:40:22 thestinger: quit killing my bot :> 2011-11-14T21:40:22 wiat, what? 2011-11-14T21:40:33 I thought strcat had awesome combat 2011-11-14T21:40:48 I have awesome movement now, that makes my ants not totally fail at combat 2011-11-14T21:41:21 thestinger: my head can't keep track of many branches in my current development state 2011-11-14T21:41:31 its a long sprint atm 2011-11-14T21:41:45 that's what my combat code is like 2011-11-14T21:42:00 it's just a massive blob of spaghetti code atm with no testing xD 2011-11-14T21:42:02 *** replore_ has joined #aichallenge 2011-11-14T21:42:16 you should see my code 2011-11-14T21:42:42 well, I'm not really testing my code to be honest, I'm just asking myself it that piece of code gets executed and how it affects its behavior 2011-11-14T21:43:30 the more and clearer it affects the behavior, the more bugs you can find 2011-11-14T21:43:55 and in some cases, just unit test or manual debugging 2011-11-14T21:45:04 I like watching A slaughter my ants 2011-11-14T21:45:11 :p 2011-11-14T21:45:22 http://ants.fluxid.pl/replay.35507 2011-11-14T21:45:51 xathis: I also have A* but I tend not to use it.. because say there is food.. How do you know which ant to path to that food.. To find that out I use BFS.. which calculates the path. So I only need A* when I need to recalculate a path for some reason 2011-11-14T21:46:38 it'll be interesting to do some tests afterwards... 2011-11-14T21:46:39 grom358: I just grab the closest ant using Manhattan distance :) 2011-11-14T21:46:47 my a* variant allows reusing partially solved boards 2011-11-14T21:46:47 It fails me sometimes because of the mazes. 2011-11-14T21:47:00 but it might be more complicated and therefore slower than just running BFS :X 2011-11-14T21:47:04 NightExcessive: yes that is why I use BFS.. Manhattan distance fails 2011-11-14T21:47:07 NightExcessive_: how many ants do you run manhattan distance for? 2011-11-14T21:47:30 thestinger: All of them, every turn. 2011-11-14T21:47:34 running BFS from a food across the _entire_ map and storing all the distances (which can be reused for pathfinding) was faster for me than the way the tutorial does it 2011-11-14T21:47:44 grom358: for collection food i don't use A* as well, but i use it to move my ants away from the hill 2011-11-14T21:48:01 and it gets you "real" distance (you have to make assumptions about unseen areas) instead of manhattan distance 2011-11-14T21:48:55 my food code is completely different now b/c I don't send multiple ants to get a few food that are right next to each other 2011-11-14T21:49:32 my food collection works well.. I use 1 ant to collect cluster of food 2011-11-14T21:49:40 thestinger: How many times do you run a BFS per turn? 2011-11-14T21:50:11 using BFS per food tile.. capped to 30 tile depth 2011-11-14T21:50:38 well, "once" for attacking or searching for enemy hills (but I do it from each enemy hill and unseen location) 2011-11-14T21:50:44 once per food tile 2011-11-14T21:51:10 and then I use BFS a bunch more for various other heuristics/minor features 2011-11-14T21:51:40 thestinger: So you use a few of them, but with multiple seeds? 2011-11-14T21:51:53 (Multiple seeds for each search) 2011-11-14T21:52:14 NightExcessive_: well, I actually didn't implement it that way, but that's basically what I'm doing 2011-11-14T21:52:54 thestinger: Ah. 2011-11-14T21:53:55 most of my breadth-first searches are slightly altered to store various things they find along the way, so my codebase is sort of a pile of graph searches atm 2011-11-14T21:54:42 graph search is easy 2011-11-14T21:54:45 actual combat is hard :X 2011-11-14T21:55:02 yeah :) 2011-11-14T21:55:33 Yeah. Tactical planning is the hardest. 2011-11-14T21:55:40 literally all i have is a* with different starts and goals 2011-11-14T21:55:45 and various configurations :X 2011-11-14T21:56:28 Sounds about like me. Mostly because I'm working on implementing D*-Lite now. 2011-11-14T21:57:18 i wish the game rate was higher 2011-11-14T21:57:20 *** Rinum has quit IRC (Quit: Page closed) 2011-11-14T21:57:24 people on fluxid too pro :X 2011-11-14T21:59:58 *** ltriant has quit IRC (Ping timeout: 258 seconds) 2011-11-14T22:00:04 just use it :p 2011-11-14T22:00:15 the more noobs there are, the more noobs will come 2011-11-14T22:00:25 i use it a lot 2011-11-14T22:00:55 but it's mostly just me losing to A and thestinger :| 2011-11-14T22:01:26 xathis: does your combat lookahead more than one step? 2011-11-14T22:01:27 :p 2011-11-14T22:01:35 I hope you are winning against mine bot 2011-11-14T22:01:35 *** ltriant has joined #aichallenge 2011-11-14T22:02:01 really, I just wanna know if the changes I made last week are enough to get me into the top 100 on the main site 2011-11-14T22:02:14 but it takes forever for rank to stabilize :( 2011-11-14T22:02:42 no kidding 2011-11-14T22:02:54 a1k0n: no 2011-11-14T22:02:54 i get my next game with random starter bots in two hours 2011-11-14T22:03:05 xathis: really? i didn't even expect an answer to that. :) 2011-11-14T22:03:20 i think my method isn't quite right then 2011-11-14T22:05:02 a1k0n: hehe, i really would love to implement some alpha-beta, but there are so many possible move states even after 1 turn... 2011-11-14T22:05:08 *** b0rder_ has joined #aichallenge 2011-11-14T22:05:18 a1k0n: I made my bot not stupid! 2011-11-14T22:05:29 bmh: awesome! 2011-11-14T22:05:48 xathis: the problem with alpha-beta is it assumes there aren't simultaneous turns 2011-11-14T22:06:08 i implemented it as an experiment on a 2v1 scenario, 3x7 map, totally solved it with dynamic programming 2011-11-14T22:06:32 if the 1 can see the 2 ants moves first, then it can always prevent the 2 from killing it without losing any ants 2011-11-14T22:06:36 "dynamic programming" -- psha, you probably like 'dynamic scope' as well 2011-11-14T22:06:47 i do, but i did this in C++ 2011-11-14T22:07:07 a1k0n: that's nice, i'm thining about doing something like that as well, in a lot of case two of my ants can't get a single enemy 2011-11-14T22:07:07 dynamic programming as in, well, the tabular solution thing 2011-11-14T22:07:42 *** datachomper has quit IRC (Quit: Leaving.) 2011-11-14T22:07:53 you have to take the payoff matrix into account somehow 2011-11-14T22:07:55 a1k0n: but not if there is water nearby 2011-11-14T22:08:17 this was a 3x7 map surrounded by water. the 1 can be cornered. 2011-11-14T22:08:39 sure, under certain starting conditions it's screwed, but under many it isn't 2011-11-14T22:09:23 anyway yeah my ants tend to get trapped in the little 1x1 'holes' in the walls of the maze maps because they don't look more than one step ahead 2011-11-14T22:09:56 have you done it the other way around? so if you have two ants and there's one enemy? 2011-11-14T22:09:59 *** roflmao has joined #aichallenge 2011-11-14T22:10:04 my code has an option to avoid them 2011-11-14T22:10:08 well the situation is symmetric 2011-11-14T22:10:26 a1k0n: yeah that's true 2011-11-14T22:10:29 avdg: heh, that's an excellent idea 2011-11-14T22:10:36 I really like your ant pattern btw a1k0n 2011-11-14T22:10:41 it looks cool 2011-11-14T22:10:41 ant pattern? 2011-11-14T22:10:48 yeah the way the ants move 2011-11-14T22:10:49 you mean when foraging? 2011-11-14T22:10:53 or attacking? 2011-11-14T22:10:57 nah, they seem to march to battle then sparse when forage 2011-11-14T22:11:02 creates a sort of spasm effect methinks 2011-11-14T22:11:05 heh 2011-11-14T22:12:03 i think my code is buggy.. it makes way too many 1 for 1 trades of ants 2011-11-14T22:12:17 not necessarily buggy 2011-11-14T22:12:18 havent figured that out 2011-11-14T22:12:22 unless your code is supposed to avoid them 2011-11-14T22:12:26 what do you think about 1v1 trades? should usually avoid? 2011-11-14T22:12:28 yeah I'm *exactly* in the same spot 2011-11-14T22:12:31 yoden: HELL yes 2011-11-14T22:12:40 yoden: if you're doing 1v1 it might be a good idea, but in an 8p map 2011-11-14T22:12:41 well i told it to lose 20 for losing an ant and gain 1 for killing one 2011-11-14T22:12:54 lose 20 what? 2011-11-14T22:13:01 *** foo___ has joined #aichallenge 2011-11-14T22:13:09 unitless points, for comparing moves 2011-11-14T22:13:16 :O 2011-11-14T22:13:22 i feel like if you're not willing to at least allow 1v1 trades, won't an aggressive bot push you back early? 2011-11-14T22:13:50 yeah the 20:1 thing is temporary. really it should depend on how many ants you have and how many you think your enemy has 2011-11-14T22:13:50 you mean back into your base? 2011-11-14T22:14:07 ..abnd whether they're attacking your base 2011-11-14T22:14:08 so you have some sort of point cost rating system, nice. 2011-11-14T22:14:16 don't you? 2011-11-14T22:14:23 you timed out. 2011-11-14T22:14:26 bs! 2011-11-14T22:14:36 me? 2011-11-14T22:14:41 i did on fluxid 2011-11-14T22:14:46 ah 2011-11-14T22:14:52 lol yes that is annoying 2011-11-14T22:14:57 nah I don't have a point system 2011-11-14T22:15:00 i just count allies and enemies 2011-11-14T22:15:02 and make decisions based on that 2011-11-14T22:15:32 mine would be statistic based 2011-11-14T22:15:39 a1k0n: have you built a hierarchical model to learned your optimal parameterization? ;) 2011-11-14T22:15:42 i have a kind of weird system that doesn't work too shabby nor too well 2011-11-14T22:15:49 (every time I give a1k0n a stupid idea, he goes and implements it) 2011-11-14T22:15:55 bmh: no, just a flat dirichlet 2011-11-14T22:16:06 bmh: :p 2011-11-14T22:16:24 "oh, just a flat dirichlet, something most people here don't even know wtf it is, but no big deal!" 2011-11-14T22:17:30 ...are you using bayes? 2011-11-14T22:17:38 i'm using a statistical method but have no proof it converges 2011-11-14T22:17:39 bot!! stop timing out :( 2011-11-14T22:18:00 bmh: Is your diffusion intensive too? :P 2011-11-14T22:18:07 NightExcessive_: yes. 2011-11-14T22:18:43 NightExcessive_: if I was doing this in C, there would be no issue. 2011-11-14T22:18:58 bmh: Wrote mine in C# (not sure why) 2011-11-14T22:19:05 my method is really elegant, so i'm kinda bummed it doesn't beat A or xathis :) 2011-11-14T22:20:02 sounds like it 2011-11-14T22:20:11 hmm, I wonder how close I would get to diffusion, but it would be rather a more controlled version and much faster 2011-11-14T22:20:30 at least, if I implement it in a faster language 2011-11-14T22:20:38 NightExcessive_: how many times can you update your automata? 2011-11-14T22:21:05 bmh: I made mine one-pass... but to work properly I also have to go through *every* cell afterwards and run a diffuse function. 2011-11-14T22:21:25 yeah. I'm trying to run my diffuse function 40 times, but I can't do it on big maps 2011-11-14T22:21:46 what are you diffusing, the scent of food? 2011-11-14T22:21:56 run it five times and keep it between frames? 2011-11-14T22:22:14 seriously like bfs is a lot easier 2011-11-14T22:22:17 a1k0n: yeah, that's what I'm trying to convince Eric to implement. I don't want to write code tonight :) 2011-11-14T22:22:37 I don't want to use my code I've just written :p 2011-11-14T22:22:53 you are destined for managerhood 2011-11-14T22:23:06 bmh: It took me the past day to get my one-pass method semi-working :P 2011-11-14T22:23:07 65 lines in beta 2011-11-14T22:23:31 ok, my foraging is apparently pretty darn good, but my combat is horrendous 2011-11-14T22:23:37 It still gets caught on really "simple" things though. :( 2011-11-14T22:23:37 actually 64 added and 1 removed 2011-11-14T22:23:51 that was good enough for 20th place a few weeks ago 2011-11-14T22:27:07 bmh: How did you solve areas between goals having higher values than the actual goals? 2011-11-14T22:27:34 NightExcessive_: I don't. 2011-11-14T22:28:04 bmh: Then how do you keep ants from getting stuck there? 2011-11-14T22:28:25 NightExcessive_: there isn't a lot to do except prioritizing, adjusting reward values or leaving it in the current state 2011-11-14T22:28:35 NightExcessive_: I don't. 2011-11-14T22:28:57 hmm, use distance to wall as penalty? 2011-11-14T22:29:05 bmh: Hum. Well that's been a major problem for me. :[ 2011-11-14T22:29:39 (well, you don't have to calculate it for every tile, just the ones that are close enough to a wall) 2011-11-14T22:30:12 *** yoden has quit IRC (Quit: Leaving.) 2011-11-14T22:30:34 i get ants stuck between food sometimes still 2011-11-14T22:31:00 *** xathis has quit IRC () 2011-11-14T22:31:08 *** foo___ has quit IRC (Quit: Page closed) 2011-11-14T22:31:26 avdg: The problem is on maps like the tutorial map where you have multiple pieces of food. Using the formula specified in the collaborative diffusion paper, the area between the food has a higher value than the food itself... and since the ants use hill climbing, they just sit there going "yay, I made it to a goal!" 2011-11-14T22:31:38 NightExcessive_: I got that problem and it turned out the diffusion coefficient was too high. 2011-11-14T22:31:59 Raimondi: Oh? Mine is currently 0.19. 2011-11-14T22:32:04 *** JorgeB has quit IRC (Quit: Computer has gone to sleep.) 2011-11-14T22:32:06 *** hjax has joined #aichallenge 2011-11-14T22:32:28 Weird, I'm around .1-.2 2011-11-14T22:32:33 replay? 2011-11-14T22:32:52 http://paste.aichallenge.org/K1vit/ (Warning: 3MB because I didn't remove debug information) 2011-11-14T22:33:23 have you guys considered applying diffusion on top of a base bfs map 2011-11-14T22:34:30 NightExcessive_: hmm, I would assume that every reached goal would be removed, so in that case, you could fix it temp with a random move 2011-11-14T22:34:43 *** hjax has quit IRC (Remote host closed the connection) 2011-11-14T22:34:53 there might be of course something better to really fix the issue 2011-11-14T22:35:20 avdg: They simply iterate over the four adjacent squares and move to the one with the highest value. If none have a higher value, they stay on the current square. I'm thinking I just need to get better math. 2011-11-14T22:35:25 roflmao: I'm going to do just that to save recomputation costs 2011-11-14T22:35:57 I wonder if there is a case (except combat) where staying is more rewarding 2011-11-14T22:36:09 the problem with always avoiding enemy ants is.. you run away from food and let them get it 2011-11-14T22:36:10 roflmao: My implementation is based around just that. It creates a BFS map from each source and goes outward. 2011-11-14T22:36:30 oh so by diffusion you mean a simple bfs map? 2011-11-14T22:37:07 roflmao: Nonono. It uses a BFS map to get the surrounding area and diffuses over the cells from the BFS map. 2011-11-14T22:37:10 NightExcessive: but doing it like that.. how do you avoid multiple ants going to the same food? 2011-11-14T22:37:40 grom358: That would happen by goal obfuscation, which I haven't implemented yet. Basically: An ant sucks up the scent around it like a vacuum. 2011-11-14T22:37:50 grom358: in my diffusion scheme my ants diffuse ant smell which gobbles up the food smell 2011-11-14T22:38:00 Yeah. 2011-11-14T22:38:09 I planned to just have ants give off a negative diffusion. 2011-11-14T22:38:21 In my bot, I use regular A* but with goal obfuscation 2011-11-14T22:38:26 it works pretty well 2011-11-14T22:38:41 I simply use the tut code for matching up ant to food 2011-11-14T22:38:50 then use a simple bfs food map to do the actual pathfinding 2011-11-14T22:39:10 the tutorial uses manhattan distance does it not? 2011-11-14T22:39:20 I believe so 2011-11-14T22:39:21 yea 2011-11-14T22:39:22 terrible on mazes 2011-11-14T22:39:32 ACTUALLY 2011-11-14T22:39:33 hang on 2011-11-14T22:39:41 *** treeform has joined #aichallenge 2011-11-14T22:39:56 ugh. I want this game on fluxid to end. I'm on turn 366 with three ants 2011-11-14T22:39:57 yeah still using tut code 2011-11-14T22:39:59 I BFS from the food until I find an ant.. therefore I have the closest ant and the apht to get there 2011-11-14T22:40:08 yeah i need to update it to do that 2011-11-14T22:40:11 path* 2011-11-14T22:40:13 but the mech i have with manhattan distance 2011-11-14T22:40:16 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-11-14T22:40:17 always gives you the best paths 2011-11-14T22:40:24 I should really make my ants *stop* foraging if my hill gets whacked 2011-11-14T22:40:27 I don't even make any distance calls anymore. 2011-11-14T22:40:37 or 2011-11-14T22:40:39 actually 2011-11-14T22:40:43 no what am I saying stupid me 2011-11-14T22:40:59 you know what, I think I might go ahead and fix that right now 2011-11-14T22:41:07 then.. I remember which ant was closest.. If the same ant is close to bunch of different food it goes to closest one first.. ends up 1 ant collects a bunch of nearby food 2011-11-14T22:41:11 my ants do stop foraging but only if they actually see the hill get razed 2011-11-14T22:41:39 grom358: How do you manage which food is closest? Length of the path? 2011-11-14T22:42:02 NightExcessive: yes length of path.. True distance (as used BFS for path) 2011-11-14T22:42:34 grom358: Ah. 2011-11-14T22:42:50 Antimony: hey.. I should add that to my bot.. Stop collecting food if I have no hills left and just look to kill enemy hills 2011-11-14T22:43:29 *** treeform_ has joined #aichallenge 2011-11-14T22:44:33 NightExcessive: without enemy ants my bot is a pro at exploring and gathering lol.. unforunately combat seems to be extremely important 2011-11-14T22:45:15 a1k0n: here's how good I am: http://ants.fluxid.pl/replay.35607 2011-11-14T22:45:37 I should really make my bot defend the hill 2011-11-14T22:45:57 how come all your ants are going the same way? 2011-11-14T22:46:14 a1k0n: my repulsion isn't strong enough repulsion 2011-11-14T22:46:20 a1k0n: murders my token defence.. lol http://ants.fluxid.pl/replay.35607 2011-11-14T22:46:22 erm.. 2011-11-14T22:46:38 *** treeform has quit IRC (Ping timeout: 276 seconds) 2011-11-14T22:46:45 yeah, i used to get killed by that, but no longer. :) 2011-11-14T22:46:46 *** bmh is now known as bmh_away 2011-11-14T22:46:53 bmh: yeah.. real defence requires combat :( 2011-11-14T22:47:02 gotta love the square defense 2011-11-14T22:47:34 ok, gotta go play trains with my kid 2011-11-14T22:47:36 *** amstan has joined #aichallenge 2011-11-14T22:47:36 *** ChanServ sets mode: +o amstan 2011-11-14T22:47:52 a1k0n: laters 2011-11-14T22:50:45 *** Antimony_ has joined #aichallenge 2011-11-14T22:51:16 *** bmh_away is now known as bmh 2011-11-14T22:51:49 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-11-14T22:51:51 *** djr_ has joined #aichallenge 2011-11-14T22:53:35 So I've noticed that with my debugging on, my script takes about 2200% longer per turn. 2011-11-14T22:54:20 what you programming in night? 2011-11-14T22:55:11 jab_bott's visualizer fork... so printing lines to stdout. 2011-11-14T22:55:18 roflmao: C# 2011-11-14T22:55:33 *** Antimony_ has quit IRC (Ping timeout: 258 seconds) 2011-11-14T22:55:52 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-11-14T22:56:16 *** amstan has joined #aichallenge 2011-11-14T22:56:16 *** ChanServ sets mode: +o amstan 2011-11-14T23:00:40 a1k0n: The reason all my ants walk in one direction is because I don't have any scoring logic. I just ask "Where is there the most food? Where is there not a friendly ant?" in order without combining the values in a model 2011-11-14T23:01:19 ah, night 2011-11-14T23:01:28 i would help you if I knew more c# 2011-11-14T23:01:31 but, alas, i don't :( 2011-11-14T23:01:42 well, in theory I can give each of my ants a job as I can store information in each ant 2011-11-14T23:01:51 * u_ is doing that 2011-11-14T23:01:51 but that would require me to write a new bot 2011-11-14T23:01:54 they even have name 2011-11-14T23:01:56 s 2011-11-14T23:02:00 lol 2011-11-14T23:02:07 hahaha! name 2011-11-14T23:02:33 do you have an array of actually names? or are they just kinda "stupid"? :p 2011-11-14T23:02:48 well, my ants have id's 2011-11-14T23:02:50 naw they're just letters 2011-11-14T23:03:43 What's everyone's favorite Linux distro? (Fairly random question with no followup) 2011-11-14T23:03:55 SLS 2011-11-14T23:04:04 OpenBSD 2011-11-14T23:04:12 debian'ish, though I haven't use many distrubutions 2011-11-14T23:04:15 debian sid 2011-11-14T23:04:29 I hope I can as get better informed 2011-11-14T23:05:12 * avdg would wonder how many people here would prefer windows :p 2011-11-14T23:05:20 It was ubuntu.. but I hate Unity and Gnome 3 2011-11-14T23:05:28 burn the heretic! 2011-11-14T23:05:32 grom358: Same. I switched to openSUSE. 2011-11-14T23:05:32 i dont mind using either 2011-11-14T23:05:49 ubuntu 2011-11-14T23:06:04 with gnu 2011-11-14T23:06:12 err gnoam 2011-11-14T23:06:17 (i little durnk) 2011-11-14T23:06:18 @gnu ubuntu 2011-11-14T23:06:18 thestinger: GNU/ubuntu 2011-11-14T23:06:22 I'm currently on os x, but my other laptop (which died) has ubuntu 2011-11-14T23:06:28 roflmao: program drunk 2? 2011-11-14T23:06:40 i find i can program about up until drunk 2011-11-14T23:06:40 ye 2011-11-14T23:06:42 s 2011-11-14T23:06:48 LFS? anyone? 2011-11-14T23:06:50 Arch is my distro of choice atm 2011-11-14T23:07:05 I would like to try arch :-) 2011-11-14T23:07:19 I use Debian on my server, but only because it works and I'm much too lazy to change something that works fine and doesn't bother me at all 2011-11-14T23:07:30 well, I used like 5 min, but I don't know enough how to adjust it atm 2011-11-14T23:08:22 Oooo. Arch looks shiney. (I've never heard of it before now) 2011-11-14T23:08:39 arch is very primitive ;-) 2011-11-14T23:08:47 thestinger: ur not stinger from utonet are you? 2011-11-14T23:08:51 nope 2011-11-14T23:09:23 avdg: Primitive is shiney nowadays... I hate the amount of stuff Ubuntu comes packed with. 2011-11-14T23:09:29 *** NightExcessive_ is now known as NightExcessive 2011-11-14T23:09:47 well yeah, I'm not telling its a bad thing 2011-11-14T23:10:04 but I wouldn't suggest it to many people 2011-11-14T23:10:19 just because of its type of simplicity 2011-11-14T23:10:30 Ah 2011-11-14T23:10:52 I don't care what distro it's running, as long as I can have my xmonad 2011-11-14T23:11:29 xmonad? 2011-11-14T23:11:41 ah.. xmonad.. I really should try use it 2011-11-14T23:11:43 I currently only care about a gui to run a browser and few apps (but it doesn't have to be shiny, though it has to be simple) and a good cli env with good tools 2011-11-14T23:11:50 NightExcessive: it's a tiling window manager. It makes using a computer bearable. 2011-11-14T23:11:58 Arch ships the latest stable upstream releases, and generally there are no patches applied to software - that's mostly what I like 2011-11-14T23:12:01 damn u can battle through water... interesting for those maze maps 2011-11-14T23:12:23 yeah, the water only limits movements 2011-11-14T23:12:41 hrm..... that kinda changes things 2011-11-14T23:12:48 you can get the same with Debian, but bleeding edge debian isn't nearly as stable as Arch in my experience and the package management isn't as nice 2011-11-14T23:12:49 can you drown your own ants? 2011-11-14T23:12:53 nope 2011-11-14T23:12:57 they just refuse to move 2011-11-14T23:12:58 that's a shame. 2011-11-14T23:12:58 i was using a diffusiony map kinda thing 2011-11-14T23:13:03 that changes things.... a lot hrm.... 2011-11-14T23:13:04 Oh god. I couldn't stand xmonad... preference, I suppose. 2011-11-14T23:13:13 i guess i needed to revamp the battle code anyways 2011-11-14T23:13:26 NightExcessive: I think that OSX has the most awful window manager known to man (as I type on my macbook) 2011-11-14T23:13:38 i think i need to stop progrmming my ant while mentally impaired 2011-11-14T23:13:40 bmh: I agree. 2011-11-14T23:13:44 i think i might jnust me making it wrose 2011-11-14T23:13:56 I wish osx had a little bit more useful spotlight though 2011-11-14T23:13:57 roflmao: turn on spellcheck and stop drinking on monday night 2011-11-14T23:13:59 * bmh puts down the wine 2011-11-14T23:14:05 dunno if its better on lion 2011-11-14T23:14:23 My personal favorite is X11+KDE 2011-11-14T23:14:31 (and I don't want an other one installed parallel with it) 2011-11-14T23:14:33 Drinking increases your chances of getting k-lined a billion percent. 2011-11-14T23:14:50 NightExcessive: I use i3 now, dwm-style tags and layouts (awesome, xmonad, etc.) annoy me too much 2011-11-14T23:16:24 can i issue a "stay still" order by replacing NESW with ' '? 2011-11-14T23:16:34 u_: just don't issue an order 2011-11-14T23:16:45 bmh: it's convienent.. 2011-11-14T23:16:59 u_: abstract harder :) 2011-11-14T23:17:19 *** Antimony has joined #aichallenge 2011-11-14T23:17:47 Welp, I'm off for a bit. Ciao all. 2011-11-14T23:17:50 *** NightExcessive has quit IRC (Quit: Page closed) 2011-11-14T23:18:10 *** bmh has quit IRC (Quit: bmh) 2011-11-14T23:18:15 hm, i can, but it complains about "invalid actions" 2011-11-14T23:18:25 well, I have to go as well *takes laptop and moves* 2011-11-14T23:18:44 hope it won't fire me for that 2011-11-14T23:18:58 u_: so when outputting moves, if(order == '-') return; 2011-11-14T23:19:16 er ' '. i use -. heh. 2011-11-14T23:21:50 we get to see the code to top 100 bots after it finishes? 2011-11-14T23:22:08 cause if I don't get combat going.. I really want to see how its done lol 2011-11-14T23:22:18 i'm sure many will release theirs. i certainly will. 2011-11-14T23:22:45 *** pyDude has joined #aichallenge 2011-11-14T23:23:46 I'll show you mine, if you show me yours 2011-11-14T23:24:04 <_flag> Most of the top entries from past contests have all been released 2011-11-14T23:24:22 <_flag> There's no real prize, so just having your code on display is the main prize really 2011-11-14T23:25:13 Basically, most top coders are flashers, that's what _flag is saying 2011-11-14T23:25:40 *** UncleVasya has joined #aichallenge 2011-11-14T23:26:13 <_flag> Most people are flashers :) 2011-11-14T23:26:16 *** djr_ has quit IRC (Ping timeout: 240 seconds) 2011-11-14T23:27:13 "i timed out." 2011-11-14T23:27:14 BS 2011-11-14T23:30:17 mcstar: unlucky-unlucky SimbaCat, your second game was against him and my one too :) 2011-11-14T23:32:14 *** bmh has joined #aichallenge 2011-11-14T23:32:18 *** bmh has joined #aichallenge 2011-11-14T23:32:33 when does fluxid stop running the game? I'm on turn 1309 2011-11-14T23:32:42 1500 2011-11-14T23:32:46 thanks 2011-11-14T23:33:18 *** djr_ has joined #aichallenge 2011-11-14T23:34:28 *** roflmao has quit IRC (Quit: Leaving.) 2011-11-14T23:39:13 *** u_ has quit IRC (Quit: u_) 2011-11-14T23:46:17 *** Antimony has quit IRC (Ping timeout: 245 seconds) 2011-11-14T23:49:32 *** bmh has quit IRC (Quit: bmh) 2011-11-14T23:50:25 *** pedrosorio has quit IRC (Quit: Page closed) 2011-11-14T23:53:52 *** twymer has quit IRC (Ping timeout: 240 seconds) 2011-11-14T23:57:22 *** UncleVasya has quit IRC () 2011-11-14T23:57:57 *** BenJackson has joined #aichallenge 2011-11-14T23:58:45 *** bmh has joined #aichallenge