2011-10-30T00:00:02 *** Vesbur has joined #aichallenge 2011-10-30T00:00:56 looking for working code in https://github.com/aichallenge/aichallenge/ 2011-10-30T00:01:23 ok, thanks 2011-10-30T00:01:29 sec 2011-10-30T00:02:22 *** cowbandit has quit IRC (Quit: Page closed) 2011-10-30T00:02:42 ghjgghj: http://paste.pocoo.org/raw/500296/ here's an example of the 2d array 21 turns into a game on a small map 2011-10-30T00:03:06 -1 means water, 0 means that location is unexplored, a higher number is the distance to an explored area 2011-10-30T00:04:11 ummm, ok, so i want to move the ants to the closest 0? 2011-10-30T00:04:15 so, I iterate through each unexplored area (you could just pick out a few though if speed is an issue), and first I set that location to 0, then iterate outwards from there with a breadth-first search, increasing the value by 1 each time I move 2011-10-30T00:04:37 well, once you make a distance map like that, you can just check the value at an ant's location, and that is the distance to an unseen area 2011-10-30T00:04:57 and you get pathfinding for free - you just move to any of the 4 locations next to the ant that has a lower value 2011-10-30T00:05:29 blech, the github is so different that it's impossible to figure out 2011-10-30T00:06:46 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-30T00:06:56 ok, make sense, but how do i make that map? 2011-10-30T00:08:21 so, you should preallocate a 2d array representing the map (rows/columns), to make a distance map for one target - set the location of the target to 0, and run a BFS search - increasing the value by 1 each time you move to a new node (and storing it in the graph) 2011-10-30T00:08:58 there's a working breadth-search first used by the C++ starter bot to calculate vision, but I think the python bot uses a better way now 2011-10-30T00:09:07 breadth-first search* 2011-10-30T00:09:53 I am still not sure what breadth-first search is 2011-10-30T00:10:15 https://secure.wikimedia.org/wikipedia/en/wiki/Breadth-first_search 2011-10-30T00:10:46 you start at a node and search outwards to every location, never revisiting a node 2011-10-30T00:11:25 since you record the distance from where you started in the 2d array, you don't need to keep track of visited nodes - you just stop when a node doesn't have a lower value than the parent you came from 2011-10-30T00:11:33 dijkstra's algorithm is a famous example 2011-10-30T00:11:49 you can implement BFS in ~5 lines in python, and you can find lots of examples online 2011-10-30T00:12:02 *** avdg has quit IRC (Quit: Leaving.) 2011-10-30T00:12:04 most of them will probably use a dictionary of lists to represent a graph, because it adapts to any situation 2011-10-30T00:12:16 umm, ok, I get the basic ideas now, just have to do some reading, thanks a lot 2011-10-30T00:12:28 you'll want to code your own that uses a 2d list or array to represent the graph, and record distances in it 2011-10-30T00:13:09 ok 2011-10-30T00:14:36 <^5> Theres something I dont get, how do I actually use a search algorithm? I dont have enough time to do it per ant per turn, so how often? when? 2011-10-30T00:14:53 *** vicky has joined #aichallenge 2011-10-30T00:14:59 I run a bunch of breadth-first searches every turn 2011-10-30T00:15:17 *** vicky is now known as Guest53877 2011-10-30T00:15:58 I do one for each food (across the whole map at the moment, it's fast) 2011-10-30T00:16:50 and I also make a 2d vector for attacking hills and do a BFS for each enemy hill (sort of, they stop early) and the same for exploration 2011-10-30T00:17:16 I'm using C++ but you can do similar stuff in python, you just have to be more conservative 2011-10-30T00:17:31 I was having timeouts when I used A* for each ant 2011-10-30T00:18:11 doing a BFS gets you distances _and_ pathfinding for every location on the map to the place you started as long as you store the result somewhere 2011-10-30T00:18:39 *** aarossig has joined #aichallenge 2011-10-30T00:19:12 *** danielharan has joined #aichallenge 2011-10-30T00:19:16 plenty of opportunity to show off all that computer science theory :) 2011-10-30T00:19:37 I didn't really know any of the graph theory stuff when I started :P 2011-10-30T00:20:24 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-30T00:20:41 at least you can start! 2011-10-30T00:20:42 sheesh 2011-10-30T00:20:52 I also learned C++ for this, because I found it easier than being really careful about performance in python 2011-10-30T00:20:53 *** craklyn has joined #aichallenge 2011-10-30T00:22:02 *** aarossig_ has quit IRC (Ping timeout: 256 seconds) 2011-10-30T00:22:41 you can definitely get pathfinding and distance measurement working in python this way without timing out though 2011-10-30T00:22:59 I got the idea of using a breadth-first search to make a distance map from Fluxid, and he's using python iirc 2011-10-30T00:23:51 ghjgghj: https://github.com/aichallenge/aichallenge/blob/4cd5c2f742e55321bb3f02f8f13837ddba448f24/ants/dist/starter_bots/python/ants.py#L198 2011-10-30T00:24:35 ok, cool, thanks 2011-10-30T00:24:49 that's an example of BFS 2011-10-30T00:25:01 you can adapt that code to do what I'm doing very easily 2011-10-30T00:27:47 does it make a different weather or not I have two files like in the tutorial or combine the two files? 2011-10-30T00:28:33 nope, the server just runs MyBot.py 2011-10-30T00:28:45 *** Nakyuant has quit IRC (Quit: Page closed) 2011-10-30T00:28:50 *** TheShagg has quit IRC (Quit: Page closed) 2011-10-30T00:28:55 you can rename or get rid of ants.py 2011-10-30T00:29:03 or include whatever other files you want 2011-10-30T00:29:42 ok, thanks 2011-10-30T00:31:37 *** Jak_o_Shadows has joined #aichallenge 2011-10-30T00:32:16 *** ghjgghj has quit IRC (Quit: Page closed) 2011-10-30T00:33:43 *** Saulzar has joined #aichallenge 2011-10-30T00:34:22 *** nolo_contendere has quit IRC (Read error: Connection reset by peer) 2011-10-30T00:35:58 *** nolo_contendere has joined #aichallenge 2011-10-30T00:42:50 *** Vesbur has quit IRC (Quit: Linkinus - http://linkinus.com) 2011-10-30T00:46:50 *** craklyn has quit IRC (Quit: Page closed) 2011-10-30T00:49:16 well holy crap - my debug visualizer integration works a treat! 2011-10-30T00:49:45 nice 2011-10-30T00:52:09 let's see, I started around 19:00 and it's 04:50 - basically 10 hours work, learning python and javascript as needed :) 2011-10-30T00:52:56 I'm so impressed by how powerful these languages actually are 2011-10-30T00:54:49 gonna add a few more features and push it to the repo in case anyone else wants to try it out 2011-10-30T01:05:09 *** Naga has joined #aichallenge 2011-10-30T01:06:04 *** Fandekasp has joined #aichallenge 2011-10-30T01:06:44 *** Naga has quit IRC (Client Quit) 2011-10-30T01:11:41 *** capa has quit IRC (Ping timeout: 265 seconds) 2011-10-30T01:11:42 *** djr_ has joined #aichallenge 2011-10-30T01:12:29 *** Fandekasp has quit IRC (Read error: Operation timed out) 2011-10-30T01:13:40 *** bmh has joined #aichallenge 2011-10-30T01:14:04 *** newguy has joined #aichallenge 2011-10-30T01:14:14 hi 2011-10-30T01:15:06 *** newguy has quit IRC (Client Quit) 2011-10-30T01:19:38 *** danielharan has joined #aichallenge 2011-10-30T01:22:10 *** Fandekasp has joined #aichallenge 2011-10-30T01:28:19 *** mleise has joined #aichallenge 2011-10-30T01:31:02 *** svujic has quit IRC (Quit: leaving) 2011-10-30T01:38:13 *** Saulzar has quit IRC (Ping timeout: 258 seconds) 2011-10-30T01:44:18 *** Saulzar has joined #aichallenge 2011-10-30T01:47:16 <^5> is there a way to identify individual ants apart from their location? 2011-10-30T01:47:56 *** Nakyuant has joined #aichallenge 2011-10-30T01:48:59 nope 2011-10-30T01:49:04 not unless you write it yourself based on their location 2011-10-30T01:49:05 unless you mean different coloured ants 2011-10-30T01:50:10 <^5> damn, ok. Is there a page somewhere that has all the methods and descriptions I can use in python? I cant seem to find one 2011-10-30T01:51:17 dai-ra: you can just check ants.py: https://github.com/aichallenge/aichallenge/blob/epsilon/ants/dist/starter_bots/python/ants.py 2011-10-30T01:51:31 i think the function names and docs are pretty descriptive 2011-10-30T01:51:37 *** dana44 has joined #aichallenge 2011-10-30T01:51:47 <^5> perfect, thanks :) 2011-10-30T02:02:19 <^5> how do you guys go about saving the ants path to take? 2011-10-30T02:03:03 *** stocha has joined #aichallenge 2011-10-30T02:03:22 hi. how do one raze a hill ? i didn't find that piece of information in the spec. 2011-10-30T02:03:40 although it seems to be the goal of the game. 2011-10-30T02:03:53 <^5> I think you just have an ant on top of their hill 2011-10-30T02:04:14 shouldn't that information figure in the spec though ? 2011-10-30T02:05:02 ^5: thanks 2011-10-30T02:05:15 *** sigh has joined #aichallenge 2011-10-30T02:11:58 *** roflmao has joined #aichallenge 2011-10-30T02:12:00 hey everyone 2011-10-30T02:14:07 hey you 2011-10-30T02:16:13 *** Saulzar has quit IRC (Ping timeout: 240 seconds) 2011-10-30T02:16:27 *** stocha has quit IRC (Quit: Page closed) 2011-10-30T02:18:33 hey 2011-10-30T02:18:37 can I ask a question? 2011-10-30T02:19:24 *** puto|mokka has quit IRC (Ping timeout: 244 seconds) 2011-10-30T02:25:00 *** Knekkebjoern has quit IRC (Quit: Leaving.) 2011-10-30T02:25:23 *** wombot_ has joined #aichallenge 2011-10-30T02:26:01 *** Mooloo has joined #aichallenge 2011-10-30T02:28:32 *** bmh has quit IRC (Quit: bmh) 2011-10-30T02:29:08 *** Baus has quit IRC (Ping timeout: 245 seconds) 2011-10-30T02:33:03 *** puto|mokka has joined #aichallenge 2011-10-30T02:42:33 roflmao: Might have been answered by now had you asked it. :) 2011-10-30T02:42:33 *** roflmao has quit IRC (Ping timeout: 265 seconds) 2011-10-30T02:45:29 :p 2011-10-30T02:45:52 anybody on visual studio 2010? 2011-10-30T02:46:17 Express editon... 2011-10-30T02:47:03 *** tmandry has joined #aichallenge 2011-10-30T02:47:11 dana44: no but c++ problems? 2011-10-30T02:47:19 i use code::blocks for c++ :) 2011-10-30T02:48:15 *** wombot_ has quit IRC (Quit: ChatZilla 0.9.87 [SeaMonkey 2.4.1/20110928161145]) 2011-10-30T02:49:51 Yeah, something that uses gcc as its compiler like code::blocks is probably a better idea. A little closer to a build like the one on the server. 2011-10-30T02:59:08 mleise: hey - I got my ai-debug visualizer working 2011-10-30T02:59:09 *** jtamer has joined #aichallenge 2011-10-30T03:00:16 jab_bott: wow, that was fast. you learn quick! 2011-10-30T03:01:27 mleise: well the javascript is basically java - and the python edits to ants.py took a while to get to grips with - but it's not doing anything particularly clever so I was able to infer most of the python syntax from looking at what was already there 2011-10-30T03:02:40 mleise: also I have a games/graphics background, so the scaling/mapping of coordinates in the visualizer was second nature (I like your drawWrapped function by the way - made use of that) 2011-10-30T03:04:00 mleise: heh actually I spent quite a bit of time learning about github, amstan helped me with that :) 2011-10-30T03:04:24 *** thestinger has quit IRC (Quit: sleep) 2011-10-30T03:04:44 mleise: I forked the project and all that, so should be easy to merge if you want 2011-10-30T03:10:37 *** icefox has quit IRC (Ping timeout: 252 seconds) 2011-10-30T03:12:28 oh thanks. I'd have to test it with all use cases though. There is the Java visualizer for example and support for live streams as well as map files or older replay versions. Also a few different browser versions must be checked if new JavaScript or html canvas features were used. 2011-10-30T03:13:33 *** n4nd0 has quit IRC (Remote host closed the connection) 2011-10-30T03:13:45 *** n4nd0 has joined #aichallenge 2011-10-30T03:14:00 *** n4nd0 has joined #aichallenge 2011-10-30T03:14:14 Also McLeopold who is working on the engine would have to agree that the changes to the engine are ok, and then someone needs to update the Wiki for other to learn about that feature. Last not least the server must be updated to show the new Wiki content and the new tools download. 2011-10-30T03:14:57 *** irchs has quit IRC (Quit: irchs) 2011-10-30T03:15:58 *** dvladim has joined #aichallenge 2011-10-30T03:18:07 dana, i'm on 2010 2011-10-30T03:21:32 *** Meh has joined #aichallenge 2011-10-30T03:22:10 Hey guys, I forgot my username; now wut? :( 2011-10-30T03:22:41 any ideas? 2011-10-30T03:23:09 Meh: is there an active submission on the account? 2011-10-30T03:23:26 no 2011-10-30T03:23:36 just make another then 2011-10-30T03:23:41 *** jtamer has quit IRC (Quit: Leaving...) 2011-10-30T03:24:08 Meh: I'd bet an admin can unlock it for you if you remember your email address. 2011-10-30T03:24:19 think the organizer won't mind? It would suck to get disqualified because no one even cared to check if both accounts had submission .. 2011-10-30T03:24:43 you wont get disqualified for that, it's more than one active submission per person we don't want 2011-10-30T03:24:56 I remember the email address for sure, but how can I ask an admin? I don't even see an email address to write to .. 2011-10-30T03:24:58 but yeah amstan or someone else should be able to find the username if you know the email address you registered with 2011-10-30T03:25:09 amstan: are you about? 2011-10-30T03:25:29 nope 2011-10-30T03:25:30 3am 2011-10-30T03:25:37 watching mythbusters 2011-10-30T03:25:47 you just need to look up a username from an email adderss :P 2011-10-30T03:25:54 Well, I'm all for creating a new account anyway: Even if I can't remember the username I've used, I know for sure it wasn't one I'd like to compete under 2011-10-30T03:26:05 Meh: just do that then 2011-10-30T03:26:07 mleise: yes - all that! Actually you wouldn't want the visualizer overlays to be switched on for build systems as it's a bit insecure the way I've implemented it. I could be tightened up though 2011-10-30T03:26:08 Meh: you can probably use the user search if you know a part of your nick 2011-10-30T03:26:08 amstan: nm :) 2011-10-30T03:26:58 jab_bott: Maybe you can make a forum post about your patches? 2011-10-30T03:27:19 mleise: can't even remember what it looked like. I just remember that it sucked. Does the search engine support "nicknames that sucks"? ;) 2011-10-30T03:27:28 *** delt0r__ has quit IRC (Ping timeout: 245 seconds) 2011-10-30T03:27:47 mleise: it uses javascript function calls in the JSON and calls them with eval :/ 2011-10-30T03:27:51 uh.. technically it looks for a matching substring ignoring cases 2011-10-30T03:27:55 So, thanks everyone. I'll go and create a new account. 2011-10-30T03:28:19 jab_bott: ok eval is evil :) 2011-10-30T03:28:22 mleise: yeah - I am writing a post about it now 2011-10-30T03:28:46 mleise: technically, I was joking. 2011-10-30T03:28:54 later. 2011-10-30T03:29:02 *** Meh has quit IRC (Quit: Page closed) 2011-10-30T03:29:10 mleise: yeah it was the quickest/easiest to avoid writing another command interpreter in teh javascript - like I said though that can be tidied up, for now it's ok 2011-10-30T03:30:58 *** Meh_Again has joined #aichallenge 2011-10-30T03:31:07 uh oh: "The email whereismarty@hotmail.com is already in use. You are only allowed to have one account!" 2011-10-30T03:31:33 oops. enjoy my email, you spammers .... 2011-10-30T03:31:37 lol 2011-10-30T03:32:20 amstan, still !(not there)? 2011-10-30T03:32:43 *** dana44 has quit IRC (Quit: Page closed) 2011-10-30T03:33:05 *** Rogdham has joined #aichallenge 2011-10-30T03:33:11 *** Rogdham has left #aichallenge 2011-10-30T03:33:35 *** Jak_o_Shadows1 has joined #aichallenge 2011-10-30T03:33:36 hmmm.. anyone still there? antimatroid1? 2011-10-30T03:34:04 Meh_Again: yeah, you'll need to use a different email address :P 2011-10-30T03:34:45 but that's my real email... how will the organizer contact me when I'll win the contest? ;) 2011-10-30T03:35:54 *** Jak_o_Shadows has quit IRC (Ping timeout: 258 seconds) 2011-10-30T03:35:55 Could amstan just change my username? 2011-10-30T03:36:07 (on the existant account) 2011-10-30T03:36:28 Meh_Again: Have you tried the user search? For variations of nicks you might pick? marty comes up with two possible results. iasmarty? 2011-10-30T03:37:06 I don't think so krishna 2011-10-30T03:38:00 *** ThereWillBeBlood has quit IRC (Quit: Page closed) 2011-10-30T03:40:17 *** delt0r__ has joined #aichallenge 2011-10-30T03:42:32 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-30T03:45:41 *** wnight has quit IRC (Ping timeout: 258 seconds) 2011-10-30T03:45:56 *** wnight has joined #aichallenge 2011-10-30T03:47:55 *** epicmonkey has joined #aichallenge 2011-10-30T03:51:49 *** jtamer has joined #aichallenge 2011-10-30T03:52:40 *** irchs has joined #aichallenge 2011-10-30T03:56:07 can the forum support image uploads? 2011-10-30T03:56:30 what happened to ants.fluxid.pl? slowed down by an invasion of super slow bots? 2011-10-30T03:56:51 *** danielharan has joined #aichallenge 2011-10-30T03:57:56 *** Cybsy has joined #aichallenge 2011-10-30T03:58:38 never mind, I'm going to bed 2011-10-30T04:02:13 *** jtamer has joined #aichallenge 2011-10-30T04:02:38 how long do you guys usually have to wait between games? 2011-10-30T04:02:41 hmm. No System.Drawing support in C# ? 2011-10-30T04:04:58 *** irchs has quit IRC (Quit: irchs) 2011-10-30T04:06:43 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-30T04:07:14 *** danielharan has joined #aichallenge 2011-10-30T04:07:20 *** n4nd0 has quit IRC (Ping timeout: 260 seconds) 2011-10-30T04:09:33 *** vantreeseba has quit IRC (Ping timeout: 265 seconds) 2011-10-30T04:10:13 *** Nakyuant has quit IRC (Quit: Page closed) 2011-10-30T04:18:00 *** merk has joined #aichallenge 2011-10-30T04:18:06 hallo guys 2011-10-30T04:18:11 i have a problem can you help me? 2011-10-30T04:18:23 how i install the ants.h header in dev-cpp? 2011-10-30T04:20:35 add the directory containing it to your project's include path 2011-10-30T04:21:53 *** dvladim has quit IRC (Read error: Operation timed out) 2011-10-30T04:23:02 can you explain a little please 2011-10-30T04:24:47 actually if you have it in the same directory as your C file #include "ants.h" should do just fine 2011-10-30T04:25:03 *** Kingpin13 has quit IRC (Ping timeout: 252 seconds) 2011-10-30T04:26:26 *** Mooloo is now known as Baus 2011-10-30T04:27:03 i have the ants.h header 2011-10-30T04:27:05 in the smae 2011-10-30T04:27:08 same directory 2011-10-30T04:27:15 but when i try to compile it 2011-10-30T04:27:25 the functions from ants.h are missing 2011-10-30T04:28:02 when compiling or when linking? 2011-10-30T04:28:16 when compiling mybot code 2011-10-30T04:28:28 you're maybe missing compiling ants.c 2011-10-30T04:29:40 merk: you should be using the provided makefile 2011-10-30T04:30:05 a ok 2011-10-30T04:30:10 how i use it? 2011-10-30T04:30:15 i run it from cmd? 2011-10-30T04:30:21 run make 2011-10-30T04:30:24 make 2011-10-30T04:30:38 on a real OS like mac or liniux 2011-10-30T04:30:43 on windows i have no idea 2011-10-30T04:31:20 include ants.c in your gcc line 2011-10-30T04:31:30 gcc -o MyBot MyBot.c ants.c 2011-10-30T04:31:38 merk: for a learning programming experience i am not sure i would recommend starting with C 2011-10-30T04:32:01 *** Kingpin13 has joined #aichallenge 2011-10-30T04:32:11 When I send a link to a github repo - should it have a commit id? 2011-10-30T04:32:48 bysin: if merk has gcc then merk also has make 2011-10-30T04:32:52 delt0r__: people always say that now, 15 years ago it was taught in colleges as an introduction to programming 2011-10-30T04:33:17 bysin: there is a reason for that... i started with machine code and assembly 2011-10-30T04:33:24 * antimatroid1 has almost no idea when it comes to make files too 2011-10-30T04:33:30 because that all there was on the machines i could get access to 2011-10-30T04:33:37 that is not the same as ideal however 2011-10-30T04:33:37 i just set up a project file in code::blocks and compile from there :P 2011-10-30T04:33:41 mleise: Here are the changes I made: https://github.com/j-h-a/aichallenge/commit/642f5bb21adf5dbafe5a127715c4d47c0e36c027 2011-10-30T04:34:21 bysin: also i am not saying don't start with C, hence the "not sure i would recommend" 2011-10-30T04:34:26 *** Jak_o_Shadows has joined #aichallenge 2011-10-30T04:34:33 ah 2011-10-30T04:34:37 but i compiling hello world is a challenge..... 2011-10-30T04:34:39 if 2011-10-30T04:34:57 i remember starting with basic 15 years ago 2011-10-30T04:35:02 qbasic 2011-10-30T04:35:21 and i would totally recommend learning some C before going to C++ 2011-10-30T04:35:37 I did some qbasic on a Mac512 2011-10-30T04:35:39 IIRC 2011-10-30T04:35:45 it was some kind of basic 2011-10-30T04:35:51 lol 2011-10-30T04:36:24 *** Jak_o_Shadows1 has quit IRC (Ping timeout: 240 seconds) 2011-10-30T04:36:30 i know python and c 2011-10-30T04:36:48 c i have experience with a lot of algorithms and problems 2011-10-30T04:36:51 *** Meh_Again has quit IRC (Quit: Page closed) 2011-10-30T04:36:52 like genetic algorithms 2011-10-30T04:36:57 ofline and online algorithms 2011-10-30T04:36:59 merk: compiling C is part of knowing... it is pretty clear you don't know your C. Nothing wrong with that 2011-10-30T04:37:01 and a lot of other syff 2011-10-30T04:37:08 you are right 2011-10-30T04:37:16 i dont know the compiler part 2011-10-30T04:37:28 its kind of a really important part 2011-10-30T04:37:32 genetic algorithms are the pseudoscience of programming 2011-10-30T04:37:32 yes 2011-10-30T04:37:49 and numerical analysis in c 2011-10-30T04:37:59 also its a dogs breakfast... can't understand how the c compiling linking thing is still broken after 30 years 2011-10-30T04:38:22 delt0r__: why c before c++? 2011-10-30T04:38:29 i did use c before c++ but I find c++ way easier 2011-10-30T04:38:47 i agree, i learned C++ before C, then stuck with C 2011-10-30T04:39:00 it's easy to get started with the basics of stl then learn other stuff on a need to know basis 2011-10-30T04:39:01 antimatroid1: to learn the basic memory model that C/C++ implies that is *not* abstracted out for the coder 2011-10-30T04:39:13 *** merk has quit IRC () 2011-10-30T04:39:33 antimatroid1: to get pass by value and pass by reference through their think skulls... 2011-10-30T04:39:48 other wise base pointer etc are real hard work 2011-10-30T04:40:22 hmm fair enough, i don't find pointers/references as confusing as people make them out to be :\ 2011-10-30T04:40:43 there's heaps of other things that tripped me up way more with programming 2011-10-30T04:40:54 a lot of people find them hard. and even worse one you do OO with basepointers they get even more confused 2011-10-30T04:41:59 then move on to dynamic casts 2011-10-30T04:42:09 shesh what is so hard about that? 2011-10-30T04:42:20 RTTI is not magic 2011-10-30T04:42:24 i don't even know what you're talking about anymore :P 2011-10-30T04:42:37 there you go... ;) 2011-10-30T04:43:35 I have worked with large teems where the policy is "no dynamic casts" even though that would solve their design issues completly 2011-10-30T04:44:08 a dynamic cast is a normal cast in just about every other OO lang --but is seen as magic in C++ for some reason 2011-10-30T04:44:42 http://en.wikipedia.org/wiki/Dynamic_cast 2011-10-30T04:45:11 *** jtamer has quit IRC (Quit: Leaving...) 2011-10-30T04:45:45 The most common case is casting a base class to a specific subclass 2011-10-30T04:52:02 C++ has C in the background. Its no wonder they made a distinction for dynamic casts which actually do something at runtime. 2011-10-30T04:52:30 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-10-30T04:55:35 mleise: I did a forum post: http://aichallenge.org/forums/viewtopic.php?f=25&t=1633&start=10 :) 2011-10-30T04:58:45 *** bearoff_w has joined #aichallenge 2011-10-30T04:59:05 ah nice 2011-10-30T05:00:09 *** HaraKiri has joined #aichallenge 2011-10-30T05:00:53 *** mleise has quit IRC (Quit: Leaving.) 2011-10-30T05:03:12 *** dom7b5 has joined #aichallenge 2011-10-30T05:05:37 very awesome 2011-10-30T05:07:05 now I need some sleeeep :/ 2011-10-30T05:07:57 *** jab_bott has quit IRC (Quit: Page closed) 2011-10-30T05:11:35 Can someone help me understand the "r" command? Why would the system tell you about food, and then tell you it's removed? 2011-10-30T05:14:39 Nescio: where did you see that? 2011-10-30T05:14:42 *** mleise has joined #aichallenge 2011-10-30T05:14:54 fyi the r line was removed ages ago, there should not be any documentation for it 2011-10-30T05:15:08 *** jamesBrown has joined #aichallenge 2011-10-30T05:15:12 hello 2011-10-30T05:15:41 Ah, it was in the ParseUpdate code in the C# starter... I guess it's dead code - good to know, thanks! 2011-10-30T05:15:55 it was to cover the case when a bot saw food and it was then removed while outside the vision of that bot and given when they first came back into view of the location, however users are expected to do this themselves now 2011-10-30T05:16:02 the same goes for ant hills, enemies or your own 2011-10-30T05:16:10 how can I recompile the visualizer.jar Please ? 2011-10-30T05:16:18 *** tmandry has quit IRC (Ping timeout: 255 seconds) 2011-10-30T05:17:48 With the old rules that makes sense; I guess I joined the party late :) 2011-10-30T05:18:31 i'm working with ruby on a mac, tools package comes with a bug and the last version on github don't have a visualizer.jar ... 2011-10-30T05:18:36 can anyone help pls ? 2011-10-30T05:21:00 *** dvladim has joined #aichallenge 2011-10-30T05:23:13 i see so many bots crashing a few turns into the game 2011-10-30T05:25:47 *** djr_ has quit IRC (Read error: Connection reset by peer) 2011-10-30T05:26:11 *** djr_ has joined #aichallenge 2011-10-30T05:26:46 *** irchs has joined #aichallenge 2011-10-30T05:27:39 heh. my bot just crashed locally after 6 turns. seems doing ants * food A* searches is a bad, bad idea 2011-10-30T05:28:25 bysin: all languages, or just scripting? 2011-10-30T05:28:46 kinda all, but its learning towards C++ 2011-10-30T05:29:10 stop beeing noobs 2011-10-30T05:29:25 jamesBrown: thanks for that useful advice :) 2011-10-30T05:29:42 *** jamesBrown has quit IRC (Quit: Page closed) 2011-10-30T05:30:59 i only perform one A* per turn 2011-10-30T05:31:11 i think more then that would be too slow 2011-10-30T05:31:51 i feel this competition is going to boil down to who can pack the most computation in the 1000 ms allowed per turn 2011-10-30T05:32:12 I'm staying in Ruby... I'll be fighting on heuristics 2011-10-30T05:32:25 *** grom358 has joined #aichallenge 2011-10-30T05:32:40 thats why i'm using C 2011-10-30T05:32:56 wtf.. my bot got a timeout.. I ran it locally and it works fine on that map 2011-10-30T05:32:57 takes longer to code in C, so you have fewer algo iterations 2011-10-30T05:33:09 does the server sometimes just glitch out? 2011-10-30T05:33:29 danielharan: it doesnt necessarily take longer 2011-10-30T05:33:29 grom358: try reducing time locally by 2-3X and see if you can replicate? 2011-10-30T05:33:39 bysin: to code? certainly does forme :) 2011-10-30T05:34:34 *** wishor has joined #aichallenge 2011-10-30T05:35:09 *** Jak_o_Shadows1 has joined #aichallenge 2011-10-30T05:35:53 what is the turn time at? 2011-10-30T05:36:15 working fine at 500ms here 2011-10-30T05:38:27 *** Jak_o_Shadows has quit IRC (Ping timeout: 276 seconds) 2011-10-30T05:39:04 1000ms, or 1 sec 2011-10-30T05:39:20 *** analyst74 has quit IRC (Ping timeout: 258 seconds) 2011-10-30T05:39:27 guess i'll wait to see if it happens again 2011-10-30T05:41:07 what language? 2011-10-30T05:41:36 i've always wondered if a GC language would have problems 2011-10-30T05:42:08 as a gen 2 GC will pause the threads for an indefinite amount of time for cleanup, it might cause a timeout 2011-10-30T05:42:13 bysin: java 2011-10-30T05:42:24 you think it might be that? 2011-10-30T05:42:42 idk.. it was kind of random.. i never seen it timeout since I rewrote it 2011-10-30T05:42:53 usually it like 100ms to take its turn 2011-10-30T05:42:59 who knows what kind of parameters they're using for the jvm 2011-10-30T05:43:06 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-30T05:44:09 *** racko has joined #aichallenge 2011-10-30T05:44:56 *** bigbrowser has joined #aichallenge 2011-10-30T05:45:04 hello 2011-10-30T05:45:08 *** foRei has joined #aichallenge 2011-10-30T05:45:29 how can i compile visualizer.jar from the last git repo version ? 2011-10-30T05:48:08 bysin: >>only perform one A* per turn - it's too few, especially for C. Check other timings or A* realization 2011-10-30T05:48:44 *** bigbrowser is now known as incubatio 2011-10-30T05:49:16 *** incubatio is now known as incubaton 2011-10-30T05:49:51 oh i ment i was using other path finding algorithsm 2011-10-30T05:50:23 about java bots timeouts at first turns on main server: http://aichallenge.org/forums/viewtopic.php?f=25&t=1681 2011-10-30T05:51:23 bysin: if you're using BFS because A* is too slow for you, then you are definitely doing something wrong ;) 2011-10-30T05:51:30 ) 2011-10-30T05:51:57 haha, no not BFS, that is silly 2011-10-30T05:52:20 bysin, I want to say that 1000ms is much enough even for PHP) 2011-10-30T05:53:34 it might be for early game, but late game with 400 ants, its ~3ms per ant 2011-10-30T05:54:50 now I have near 70ms for 200 ants 2011-10-30T05:55:14 but should say that battles logic is not realized 2011-10-30T05:55:23 and some more logic ) 2011-10-30T05:55:43 yes.. bearoff my last game I got a timeout on map I have no problem with locally when I got 200 ants 2011-10-30T05:56:05 *** olexs has joined #aichallenge 2011-10-30T05:56:08 maybe you should write there too 2011-10-30T05:56:38 *** Cybsy has quit IRC (Ping timeout: 245 seconds) 2011-10-30T05:56:38 how did you do guys to manage that the visulizer works ? 2011-10-30T05:56:55 *** antiwhack has joined #aichallenge 2011-10-30T05:57:00 hiyo 2011-10-30T05:57:07 *** tmske has joined #aichallenge 2011-10-30T05:57:21 incubaton - what is the problem and what OS? 2011-10-30T05:57:22 anyone here already implement A* in their movement? 2011-10-30T05:57:31 :) 2011-10-30T05:58:00 I think each second at least 2011-10-30T05:58:04 bearoff_w: mac os tiger, i can't use visualizer 2011-10-30T05:58:20 i see other java bots have had the same problem 2011-10-30T05:58:25 antiwhack: yes 2011-10-30T05:58:31 sorry, can't help - can say that there are a lot of topics about visualiser on Mac on forum 2011-10-30T05:58:51 bearoff_w: where ? 2011-10-30T05:58:59 grom358: are you running it every turn for every ant or caching the results per ant? 2011-10-30T05:59:11 antiwhack: cache results 2011-10-30T05:59:27 bearoff_w: i checked aichallenge forum, i didn't found any result or topic about mac os ... 2011-10-30T05:59:29 I store the path in my ant class 2011-10-30T05:59:30 for examle http://aichallenge.org/forums/viewtopic.php?f=25&t=1631&p=10064&hilit=mac#p10064 2011-10-30T05:59:54 i'll check thx 2011-10-30T06:00:14 bearoff_w: how old is that project ? 2011-10-30T06:00:18 I suggest there were a few of topics 2011-10-30T06:00:22 *contest* 2011-10-30T06:00:49 what do you mean? 2011-10-30T06:01:56 i'm asking, from how much time did that project started ? 2011-10-30T06:02:28 don't now, but seems that early versions were in march-may 2011-10-30T06:02:41 wow ... 2011-10-30T06:02:52 ok 2011-10-30T06:03:04 bearoff_w: thx 2011-10-30T06:03:08 but you should search only last 10 days 2011-10-30T06:03:21 as all have been changed since may ) 2011-10-30T06:03:22 *** Recette|sleeping has joined #aichallenge 2011-10-30T06:04:08 bearoff_w: i heard somewhere that, that bug were corrected on the last commit 2011-10-30T06:04:10 and, may be you should download the latest tools package, as it is fixed some times 2011-10-30T06:04:18 aha 2011-10-30T06:04:33 bearoff_w: do you know how i can recompile visualizer.jar from the last sources ? 2011-10-30T06:04:38 *** onensora has quit IRC (Ping timeout: 252 seconds) 2011-10-30T06:05:08 antiwhack: i think there is even people who used D* lite 2011-10-30T06:05:13 bearoff_w: i did download it, i just didn't found the visualizer.jar, i guess i have to recompile it, but there is no documentation about it 2011-10-30T06:05:17 no, think it's not hard, but I can tell how to do it 2011-10-30T06:05:24 does anyone have an efficient distance formula for a map that wraps rows/cols? 2011-10-30T06:05:28 :) 2011-10-30T06:05:31 *can't tell 2011-10-30T06:05:35 bearoff_w: outchhh 2011-10-30T06:05:42 :) sorry 2011-10-30T06:05:45 ^^ 2011-10-30T06:05:56 maybe anyone does know 2011-10-30T06:06:05 somebody has to :P 2011-10-30T06:06:20 *** antiwhack has quit IRC (Quit: Page closed) 2011-10-30T06:06:29 How can I compile visualizer.jar from the last sources please ? 2011-10-30T06:06:43 the standard (x1 - x2) ^ 2 + (y1 - y2) ^ 2 must be tweaked for circular maps 2011-10-30T06:07:05 bysin: have you check you starter package code? 2011-10-30T06:07:11 is it in there? 2011-10-30T06:07:22 suppose yes 2011-10-30T06:07:26 PHP has 2011-10-30T06:07:33 oh i see it 2011-10-30T06:08:11 thank you 2011-10-30T06:09:40 *** bryku has joined #aichallenge 2011-10-30T06:10:46 *** master_ninja has joined #aichallenge 2011-10-30T06:11:16 circular map? :P 2011-10-30T06:11:24 it's the surface of a torus :P 2011-10-30T06:11:57 hey.. is there any way to show debugging information in the visualizer? 2011-10-30T06:12:19 *** postwhat0 has joined #aichallenge 2011-10-30T06:12:40 http://aichallenge.org/forums/viewtopic.php?f=25&t=1633 2011-10-30T06:13:17 ah, thanks 2011-10-30T06:14:43 quick question: are tools package for mac osx written in python 3.2.x or 2.7.x? 2011-10-30T06:16:11 *** djr_ has quit IRC (Ping timeout: 256 seconds) 2011-10-30T06:16:40 *** bryku has quit IRC (Ping timeout: 265 seconds) 2011-10-30T06:18:46 hmm ok it seems one should use python 2.7.x. tools wont 2011-10-30T06:18:54 *work in 3.2.x 2011-10-30T06:19:03 *** tmske has quit IRC (Quit: Page closed) 2011-10-30T06:20:33 python 3.x is very different from the 2.x 2011-10-30T06:20:41 so yea, stick with 2.7 2011-10-30T06:21:03 its not so hard to install several versions of python 2011-10-30T06:21:34 i have python2.7 python3.2 and python2.6 while just python points to 2.7 2011-10-30T06:22:16 I'm very noobish with programming so lot to learn ^^. but I'll stick with 2.7 2011-10-30T06:23:20 I couldnt get tutorial code working last night. got some namespace errors about global variable not defined 2011-10-30T06:23:35 gonna try start again and try to get it work 2011-10-30T06:23:48 anyone here got some experience with D programming? 2011-10-30T06:24:05 I'm running into a very stupid problem 2011-10-30T06:24:07 http://pastie.org/2782073 2011-10-30T06:24:37 I somehow can't wrap my head around how to use ranges in this contextn 2011-10-30T06:25:54 *** bearoff_w has left #aichallenge 2011-10-30T06:28:34 *** ikaros has joined #aichallenge 2011-10-30T06:39:23 *** racko has quit IRC (Ping timeout: 265 seconds) 2011-10-30T06:39:36 *** ztfw has joined #aichallenge 2011-10-30T06:45:12 *** mj41 has joined #aichallenge 2011-10-30T06:50:22 *** Saulzar has joined #aichallenge 2011-10-30T06:53:19 *** Nescio has quit IRC (Quit: Page closed) 2011-10-30T06:53:36 *** Nescio has joined #aichallenge 2011-10-30T06:56:37 *** sigh has quit IRC (Remote host closed the connection) 2011-10-30T06:57:09 *** sigh has joined #aichallenge 2011-10-30T06:57:24 oh, 26th place 2011-10-30T06:57:53 *** Fandekasp has quit IRC (Ping timeout: 244 seconds) 2011-10-30T06:59:11 hey Fluxid is the tcp server down? 2011-10-30T06:59:32 *** sigh has quit IRC (Remote host closed the connection) 2011-10-30T06:59:43 *** sigh has joined #aichallenge 2011-10-30T07:00:29 *** postwhat0 has quit IRC (Quit: Page closed) 2011-10-30T07:01:37 *** dom7b5 has quit IRC (Ping timeout: 265 seconds) 2011-10-30T07:02:08 *** racko has joined #aichallenge 2011-10-30T07:03:40 ikaros: it certainly was... it crashes from time to time :| 2011-10-30T07:04:01 but crashes and keeps accepting connections... 2011-10-30T07:08:10 *** n4nd0 has joined #aichallenge 2011-10-30T07:08:31 *** g0llum has joined #aichallenge 2011-10-30T07:10:37 ok i will try to connect again :9 2011-10-30T07:11:37 *** master_ninja has quit IRC (Read error: Connection reset by peer) 2011-10-30T07:13:15 *** master_ninja has joined #aichallenge 2011-10-30T07:14:56 *** Saulzar has quit IRC (Ping timeout: 244 seconds) 2011-10-30T07:16:04 *** overburn has joined #aichallenge 2011-10-30T07:16:12 *** Redgis has joined #aichallenge 2011-10-30T07:16:36 *** wishor has quit IRC (Ping timeout: 265 seconds) 2011-10-30T07:17:13 hello all, does anyone have an idea how to get input from the master program? i get the initial setup data, then when it says ready, i cout go and then no further input is given and i timeout 2011-10-30T07:19:09 overburn: Try doing a flush. 2011-10-30T07:19:51 overburn: Also, is there an endline characater at the end of your go? 2011-10-30T07:19:55 *** SITZ has joined #aichallenge 2011-10-30T07:20:01 krishna: no 2011-10-30T07:20:05 or 2011-10-30T07:20:05 hmm 2011-10-30T07:20:05 wait 2011-10-30T07:20:06 lol 2011-10-30T07:20:18 umm 2011-10-30T07:20:19 no 2011-10-30T07:20:45 yeah, that was it , thanks a million : D 2011-10-30T07:21:10 *** Fandekasp has joined #aichallenge 2011-10-30T07:21:10 Sure 2011-10-30T07:22:48 *** dom7b5 has joined #aichallenge 2011-10-30T07:30:30 how can we compile the visualizer from sources ? 2011-10-30T07:34:39 *** Migi32 has joined #aichallenge 2011-10-30T07:35:02 incubaton, i think all the eclipse projects are on the github repo 2011-10-30T07:35:54 incubaton: you mean i can build the visualizer with eclipse 2011-10-30T07:35:56 ok thanks 2011-10-30T07:36:51 hi. I'm having a small problem with the tools: when I try to make the visualizer generate bot input, it gives an error: "can only generate bot input for revision 3 replays" 2011-10-30T07:38:04 it's the latest playgame.py and visualizer.jar, so I don't think that's the problem 2011-10-30T07:42:39 *** Saulzar has joined #aichallenge 2011-10-30T07:43:58 *** smiley1983 has joined #aichallenge 2011-10-30T07:47:25 *** dvladim has quit IRC (Ping timeout: 240 seconds) 2011-10-30T07:53:41 *** goffrie_ has joined #aichallenge 2011-10-30T07:55:01 *** goffrie has quit IRC (Read error: Operation timed out) 2011-10-30T07:56:13 ok, I think the problem is that playgame.py outputs streaming format and the input generator wants replay format 2011-10-30T08:00:21 *** cowbandit has joined #aichallenge 2011-10-30T08:01:02 What's the best way to prevent lone ants from sneaking into your ant hill? 2011-10-30T08:03:08 *** jvhkgjvkhg has joined #aichallenge 2011-10-30T08:04:09 kill them 2011-10-30T08:04:22 *** xathis has joined #aichallenge 2011-10-30T08:06:48 burny: true... 2011-10-30T08:09:02 I guess keep some ants home.. 2011-10-30T08:10:49 how can i determine which player i am? 2011-10-30T08:11:13 *** xathis has quit IRC (Ping timeout: 265 seconds) 2011-10-30T08:14:36 *** Jak_o_Shadows1 has quit IRC (Remote host closed the connection) 2011-10-30T08:14:58 overburn: you can't 2011-10-30T08:15:00 *** irchs has quit IRC (Quit: irchs) 2011-10-30T08:15:01 thats the point 2011-10-30T08:15:08 well the bot can't 2011-10-30T08:15:12 does it matter though? 2011-10-30T08:15:24 ah no, i meant the player number i get assigned but nevermind 2011-10-30T08:15:30 you dont' know how many players there are 2011-10-30T08:15:30 it's written in the game specs page 2011-10-30T08:15:44 yea you get 1 always 2011-10-30T08:15:47 0 2011-10-30T08:15:51 sorry 2011-10-30T08:15:53 zero 2011-10-30T08:15:56 count from 0 2011-10-30T08:16:04 yeah 2011-10-30T08:16:19 but thanks :D 2011-10-30T08:18:30 *** Harpyon has joined #aichallenge 2011-10-30T08:20:56 *** svujic has joined #aichallenge 2011-10-30T08:24:52 *** avdg has joined #aichallenge 2011-10-30T08:31:32 *** Neurosys_ has joined #aichallenge 2011-10-30T08:31:57 hi 2011-10-30T08:32:10 good evening 2011-10-30T08:32:41 * antimatroid1 finds it strange that good evening is usually a greeting while good night a farewell 2011-10-30T08:33:25 I just downloaded and ran the c++ starter w/ ubuntu makefile on visual studio (with some modifications) and I was able to get it to run I was curious if anyone knew if submitting that would work ? 2011-10-30T08:33:41 I'm not even sure the ant is actually alive but its not crashing 2011-10-30T08:34:12 it just walks str8 up until it wraps around bottom and gets eaten by hunterbot 2011-10-30T08:34:28 and when i run the executable it runs.. but just a blank screen 2011-10-30T08:36:47 submitting the starter package is OK 2011-10-30T08:37:00 though you won't win :) 2011-10-30T08:37:58 *** weqwe has joined #aichallenge 2011-10-30T08:38:07 i was curious if it would be able to compile, hehe I'm gonna modify it but I was worried something was wrong with the way i did it 2011-10-30T08:38:29 *** n4nd0 has joined #aichallenge 2011-10-30T08:38:46 is it possible to use boost libraries? 2011-10-30T08:38:51 also not sure if thats the right behavior for the this starter.. or if i broke it 2011-10-30T08:39:42 i mean.. if u make a bot in c++ and then run it from cmd line (not in a game) is it supposed to just be a blank screen? 2011-10-30T08:41:37 *** pointhigh has joined #aichallenge 2011-10-30T08:42:08 does someone know if it is possible to use boost libraries for our bot? 2011-10-30T08:42:22 Neurosys_ yes 2011-10-30T08:42:39 *** SITZ has quit IRC (Read error: Connection reset by peer) 2011-10-30T08:42:43 try typeing "ready " 2011-10-30T08:43:01 typing even 2011-10-30T08:43:37 didnt do anythin 2011-10-30T08:44:16 w/o the quotes 2011-10-30T08:44:51 ya i just typed ready and hit enter :) 2011-10-30T08:45:48 strange - your bot should initialize then respond "go" 2011-10-30T08:46:50 im using this starter package... aichallenge.org/starter_packages/cpp_starter_package.zip and I had to resolve problems with not having a stdint.h (grabbed from opensource also had to grab platform.h) to make it compile 2011-10-30T08:46:59 it semes to be listening but it says nothing 2011-10-30T08:47:09 according to the spec there should at least be the line "turn 0" before "ready" 2011-10-30T08:47:59 the matchs play though.. he acts dumb but i thought that was just cuz he was basic... 2011-10-30T08:48:16 then i realized he literally just moves straight upwards 2011-10-30T08:48:17 *** smiley1983 has quit IRC (Quit: leaving) 2011-10-30T08:48:26 and thought maybe i broke something :) 2011-10-30T08:48:36 n4nd0, since boost is mostly a header-only library, maybe you can copy the headers you need? 2011-10-30T08:49:51 Neurosys_: yeah i think start behavior is to just try moving upward, and if you can't then keep trying other directions until you can 2011-10-30T08:49:57 *starter 2011-10-30T08:50:27 k I'll try and get the lil guy to obey and see if hes really alive or just stalling out 2011-10-30T08:50:40 thanx :) 2011-10-30T08:51:30 Migi32, I can try to do that, but I am afraid that it might become too messy 2011-10-30T08:52:25 n4nd0: yeah, true. 2011-10-30T08:52:26 Migi32, I mean, the idea is to include the files of boost that I am using in my zip file, but I have also to check what are the files that those that I am using are including 2011-10-30T08:52:43 is there a method that tells you if an enemy ant is close to your ant hill? 2011-10-30T08:52:46 Migi32, any better idead? 2011-10-30T08:52:50 I know what you mean. Boost uses a lot of headers :) 2011-10-30T08:53:51 Migi32, I think I will code on my own the part I have used so far and I'll remember for later not to think of boost 2011-10-30T08:54:07 cowbandit: if you can see the enemy it will be reported to you 2011-10-30T08:54:08 not really. But I'm actually not sure that we can't use the boost libraries. Maybe you should ask one of the contest organizers or something. 2011-10-30T08:54:29 or just try it yourself :) 2011-10-30T08:54:36 oi ok thx 2011-10-30T08:54:58 cowbandit: but it's up to you to figure out if it's close to a hill :) 2011-10-30T08:55:01 Migi32, it looks like there is no support according to this entry of the forum http://aichallenge.org/forums/viewtopic.php?f=25&t=1612&p=10300&hilit=boost&sid=cf853953b81e14003dc0b3a4ceb5072a#p10300 2011-10-30T08:55:34 dmcarthur: lol darn. well i just thought there was a prewritten method for it since your hill pings whenever an enemy ant gets close 2011-10-30T08:56:01 *** incubaton has quit IRC (Quit: Page closed) 2011-10-30T08:56:49 *** SITZ has joined #aichallenge 2011-10-30T08:57:05 the ping is just a visualizer thing i think 2011-10-30T08:58:06 *** pointhigh has quit IRC (Ping timeout: 265 seconds) 2011-10-30T08:58:19 ah ok, guess we gotta do it the hard way 2011-10-30T09:00:53 I got the bot console to respond by typing "turn 0" then "ready" ... it crashed saying the vector was too large lol 2011-10-30T09:04:33 *** Recette|sleeping has quit IRC (Ping timeout: 245 seconds) 2011-10-30T09:04:50 *** olexs1 has joined #aichallenge 2011-10-30T09:07:08 *** olexs has quit IRC (Ping timeout: 260 seconds) 2011-10-30T09:08:06 *** irchs has joined #aichallenge 2011-10-30T09:13:53 *** Neurosys_ has quit IRC (Quit: Page closed) 2011-10-30T09:15:44 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-10-30T09:21:23 *** kara has joined #aichallenge 2011-10-30T09:22:48 *** antiwhack has joined #aichallenge 2011-10-30T09:24:31 *** HaraKiri has quit IRC (Ping timeout: 248 seconds) 2011-10-30T09:25:08 anyone on that is java-savvy? 2011-10-30T09:32:15 antiwhack: just ask 2011-10-30T09:32:27 just asking 2011-10-30T09:32:39 antiwhack: I mean ask what you want to know 2011-10-30T09:32:39 *** goffrie_ has quit IRC (Remote host closed the connection) 2011-10-30T09:32:50 i'm looking for a bit of help to understand hashCode 2011-10-30T09:33:18 *** dvladim has joined #aichallenge 2011-10-30T09:33:44 antiwhack: what exactly? 2011-10-30T09:33:52 I'm digging through documents online to figure it out, Most of them list how to use it, not what it is used for 2011-10-30T09:34:27 I just started to learn Java, I've never dealt with hash codes, or hash tables 2011-10-30T09:34:27 it's used by containers based on hashing 2011-10-30T09:34:48 is it just me or is accoun's TCP server down at the moment? 2011-10-30T09:35:04 antiwhack: done a bit of java in my time 2011-10-30T09:35:38 any good sites I could use as reference? 2011-10-30T09:35:56 not of the top of my head 2011-10-30T09:35:59 *** merk has joined #aichallenge 2011-10-30T09:36:06 i only use the javadocs 2011-10-30T09:36:15 or specific google searches 2011-10-30T09:36:35 guys when i run the play_one_game.cmd 2011-10-30T09:36:39 from command line 2011-10-30T09:36:44 there is a java problem 2011-10-30T09:36:49 and i cant see the battle 2011-10-30T09:36:55 whats the problem? 2011-10-30T09:36:59 what's the error? 2011-10-30T09:37:04 antiwhack: you need to know your basic containers and algos... the wikipedia is pretty good for that 2011-10-30T09:37:08 *** n4nd0 has quit IRC (Quit: Leaving) 2011-10-30T09:37:22 thanks delt0r 2011-10-30T09:37:55 what should I wiki? container? 2011-10-30T09:38:02 antiwhack: then ask specific questions. Whatever you do *don't* go to ##java 2011-10-30T09:38:02 hash? 2011-10-30T09:38:08 hashmap 2011-10-30T09:38:11 http://pastie.org/2782743 2011-10-30T09:38:14 hash table 2011-10-30T09:38:17 lol, not looking for elitists 2011-10-30T09:38:18 thanks 2011-10-30T09:39:26 delt0r hallo 2011-10-30T09:39:28 merk: are you sure a game got played? or that the reply file is saved properly or in the right place 2011-10-30T09:39:43 when i run this command 2011-10-30T09:39:52 it pops up a java windows and the closes immediately 2011-10-30T09:40:11 well its getting called wrong 2011-10-30T09:40:14 why i don't know 2011-10-30T09:40:25 i suspect you don't have a valid game reply 2011-10-30T09:40:29 for whatever reason 2011-10-30T09:40:29 i call it as in the example 2011-10-30T09:40:50 *** antiwhack has quit IRC (Quit: Page closed) 2011-10-30T09:40:50 so? 2011-10-30T09:40:59 i use the example and it works 2011-10-30T09:41:12 in linux? 2011-10-30T09:41:16 something is wrong --without more details its hard to know 2011-10-30T09:41:53 i have the latest java runtime 2011-10-30T09:41:57 and the python 3.2 2011-10-30T09:42:03 also i don't let the play game thing start my browser... i just point FF to the game reply 2011-10-30T09:42:21 the viewer is in JS 2011-10-30T09:42:26 the node.js thing 2011-10-30T09:42:38 yes 2011-10-30T09:43:09 the java thing is to wrap it IIRC --so just a bowser that supports node.js aka FF or chrome is all you need 2011-10-30T09:43:33 i have chrome and firefox 2011-10-30T09:43:42 how i make it to run in my chrome? 2011-10-30T09:43:58 its just a link to the file system 2011-10-30T09:44:29 * delt0r__ plays one game 2011-10-30T09:45:44 so i have to link the java with chrome? 2011-10-30T09:45:50 No 2011-10-30T09:45:53 no java 2011-10-30T09:45:59 java nono 2011-10-30T09:46:21 sorry i can't find my ants stuff on this machine 2011-10-30T09:46:39 but either way it ends up just as something like 2011-10-30T09:47:06 file://myreplaysAndStuff/visuliser.js?data=whatever 2011-10-30T09:47:12 something like that 2011-10-30T09:47:20 then the browser only uses js 2011-10-30T09:47:37 chrome has a very fast js engine 2011-10-30T09:49:49 *** grom358 has quit IRC (Quit: Leaving) 2011-10-30T09:49:51 aichallenge: janzert epsilon * ree8953e / website/check_change_password.php : Correctly handle case of no old password being given to check_change_password - http://git.io/SpgbaA 2011-10-30T09:49:58 found it 2011-10-30T09:50:08 merk: file:///home/bob/Ants/tools/game_logs/replay.0.html 2011-10-30T09:50:16 that is the link i can use 2011-10-30T09:50:18 *** weqwe has quit IRC (Ping timeout: 265 seconds) 2011-10-30T09:50:28 note that you can't have the -S switch on. 2011-10-30T09:51:05 I just use my browser to browse my replys folder 2011-10-30T09:51:49 so i have to run the command 2011-10-30T09:51:58 and the with the browser open the replay 2011-10-30T09:52:45 that is what i am doing. It tries to open the browser in my case, perhaps i have old tools. But i set my default browser to nothing 2011-10-30T09:52:54 hate things the launch browsers... 2011-10-30T09:53:32 but my thing it doesnt 2011-10-30T09:53:33 run 2011-10-30T09:53:45 the play one game live cmd 2011-10-30T09:53:49 it doesnt rum 2011-10-30T09:53:50 What does not run... you need to be clear 2011-10-30T09:54:02 i'll play against anyone who sets up a tcpserver real quick 2011-10-30T09:54:02 look at pastie 2011-10-30T09:54:03 http://pastie.org/2782743 2011-10-30T09:54:08 well that is not what you said. I even asked if it did and if there was a game 2011-10-30T09:54:17 http://pastie.org/2782743 2011-10-30T09:54:20 no ---no more pastie 2011-10-30T09:54:25 look this is what i get 2011-10-30T09:54:27 just tell me and be clear 2011-10-30T09:54:37 it runs the code 2011-10-30T09:54:44 but i get errors in java 2011-10-30T09:54:46 I am looking at the pastie and it tells me nothng 2011-10-30T09:54:57 WHAT java for fs 2011-10-30T09:55:08 the game engine does not run in java 2011-10-30T09:55:11 does not use java 2011-10-30T09:55:15 *** onensora has joined #aichallenge 2011-10-30T09:55:23 what uses? 2011-10-30T09:55:37 play one game 2011-10-30T09:55:40 no java 2011-10-30T09:55:46 unless you put it there 2011-10-30T09:56:03 Oh wait 2011-10-30T09:56:14 perhaps the new vis does wrap thing with java... 2011-10-30T09:56:23 however this won['t be the first set of errors 2011-10-30T09:56:27 *** Belerafon has joined #aichallenge 2011-10-30T09:56:32 for example what is the engine output 2011-10-30T09:57:03 the new vidualizer uses java 2011-10-30T09:57:04 Should be a list of "turn blar blar blar" 2011-10-30T09:57:19 i dont get a list like that 2011-10-30T09:57:21 and extra info as needed 2011-10-30T09:57:32 well then there is no game to see 2011-10-30T09:57:47 as i suggested right at the start 2011-10-30T09:58:31 the error is before any java is run 2011-10-30T09:58:44 looks like a map the size 0,0 is tried to be replayed 2011-10-30T09:58:53 which is clearly wrong 2011-10-30T09:59:22 so your map bit in the play one game is probably wrong or you are exeing it from the wrong directory 2011-10-30T09:59:37 must i have a directory? 2011-10-30T10:00:29 on linux you are always in some directory ie the cwd or pwd 2011-10-30T10:00:39 most of this stuff is a bit sensitive to that 2011-10-30T10:00:45 ie where it is run from 2011-10-30T10:02:02 i am going to run it from my hard disk 2011-10-30T10:02:04 directory 2011-10-30T10:03:29 the same stuff again 2011-10-30T10:03:55 *** goffrie has joined #aichallenge 2011-10-30T10:04:01 Hi. Can anybody share some old bots for replacement of standard bots? Standard bots are stupid... 2011-10-30T10:05:45 you're supposed to improve them yourself :P 2011-10-30T10:06:00 oh you mean bots to test against? 2011-10-30T10:06:05 i would just play on tcp 2011-10-30T10:06:20 Belerafon: its a contest ;-) 2011-10-30T10:06:35 *** frosty has joined #aichallenge 2011-10-30T10:06:45 det0r i got stuck 2011-10-30T10:06:49 i dont know what to do 2011-10-30T10:09:24 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-10-30T10:10:13 *** ikaros has joined #aichallenge 2011-10-30T10:10:49 *** Rinum has joined #aichallenge 2011-10-30T10:12:49 *** frosty has quit IRC (Quit: Page closed) 2011-10-30T10:13:02 What tcp server is now avaible and popular? 2011-10-30T10:13:20 *** diskonnect has joined #aichallenge 2011-10-30T10:13:49 I've been using ants.fluxid.pl 2081 2011-10-30T10:14:01 *** merk has quit IRC () 2011-10-30T10:14:50 *** bearoff_w has joined #aichallenge 2011-10-30T10:15:09 *** mceier has joined #aichallenge 2011-10-30T10:17:51 *** cowbandit has quit IRC (Ping timeout: 265 seconds) 2011-10-30T10:17:56 The bot there named A seems really good. 2011-10-30T10:20:24 It looks like getting the combat right so that you run if you're outnumbered or attack if you have more goes a long way. 2011-10-30T10:24:37 *** Nescio has quit IRC (Ping timeout: 265 seconds) 2011-10-30T10:27:16 callahan: A is good but xathis is totally dominating 2011-10-30T10:28:03 *** dvladim has quit IRC (Quit: Konversation terminated!) 2011-10-30T10:28:28 Rinum: They haven't played against each other much. 2011-10-30T10:28:35 *** g0llum has joined #aichallenge 2011-10-30T10:29:32 *** dvladim has joined #aichallenge 2011-10-30T10:29:42 *** FCK42 has joined #aichallenge 2011-10-30T10:29:50 *** antiwhack has joined #aichallenge 2011-10-30T10:30:13 *** rajanaresh has left #aichallenge 2011-10-30T10:31:14 oh boy, A's author is 16-years old 2011-10-30T10:31:15 cool 2011-10-30T10:31:23 That is cool 2011-10-30T10:31:42 Where are you seeing that? 2011-10-30T10:31:45 "damnit, i've been beaten by a 16yo"? 2011-10-30T10:31:47 Fluxid: how do you know? 2011-10-30T10:31:55 http://aichallenge.org/profile.php?user=31 2011-10-30T10:32:10 when i was 16 i was learning vb.... 2011-10-30T10:32:14 Fluxid: how do u know that's A? 2011-10-30T10:32:35 he told me it's him ;P 2011-10-30T10:32:56 ahhh 2011-10-30T10:33:07 man I wish I wasn't such a noob at 16 x.x 2011-10-30T10:33:14 I think A is better than xathis atm 2011-10-30T10:33:43 I need some help: I didn't got the activation e-mail 2011-10-30T10:34:23 FCK42: maybe ask McLeopold 2011-10-30T10:35:00 I don't think either of them is particularly strategic though. The ants aren't going to where they will do the most good, they just do reasonably well whereever they are. 2011-10-30T10:35:37 McLeopold isn't here right now 2011-10-30T10:35:47 is it just me or are these AI challenges getting harder each year x.x 2011-10-30T10:35:58 *** bchoii has joined #aichallenge 2011-10-30T10:38:25 i think xathis has a better bot than A. xathis has his ants spreading out evenly over the board, trying to stay within X of other ants. 2011-10-30T10:38:31 best forr food harvesting 2011-10-30T10:38:47 antiwhack: but you only get points for attacking 2011-10-30T10:39:00 once you have all the food, attacking comes easy 2011-10-30T10:39:05 http://ants.fluxid.pl/player/A 2011-10-30T10:39:28 imo, your bot doesn't need to be that good at collecting food 2011-10-30T10:39:44 It does for some of the maps 2011-10-30T10:40:02 attacking hills is the only thing that gets you points 2011-10-30T10:41:06 how often do games happen on fluxid? 2011-10-30T10:42:01 should happen as soon as it has enough people 2011-10-30T10:42:01 continuously 2011-10-30T10:42:02 Rinum: i think i disagree... by rewriting food gathering in my bot it suddenly started to perform better at attacking hills - because it produced more ants in shorter period of time, attacks were more massive 2011-10-30T10:42:27 I think the fluxid maps are skewed towards particular types of ants 2011-10-30T10:42:38 lol 2011-10-30T10:42:38 lots of low resource maps 2011-10-30T10:42:45 and high hill count maps 2011-10-30T10:42:59 maps are random imo 2011-10-30T10:43:06 what are those coloured circles (not ants) sometimes emitting from hives in visualization? 2011-10-30T10:43:28 *** overburn has quit IRC (Ping timeout: 265 seconds) 2011-10-30T10:43:32 matching works other way than on official server. on official server players are chosen first, ant then map which were played the least by all players 2011-10-30T10:43:56 here map is chosen first, and when needed amount of players are connected - game is started 2011-10-30T10:44:37 Draakon: you mean when you move mouse close by ants? 2011-10-30T10:44:46 no 2011-10-30T10:45:14 i don't see other circles... 2011-10-30T10:45:33 Draakon: they highlight the hill when an enemy ant comes close 2011-10-30T10:45:43 ahaa.. I suppose thats it 2011-10-30T10:45:50 ah those circles 2011-10-30T10:46:00 jix: if the hill becomes visible to an enemy? 2011-10-30T10:46:00 don't know how to call them 2011-10-30T10:46:16 they appear when hill will be razed soon, and enemy who does it is close 2011-10-30T10:46:21 Does anybody know wheather there is another person than McLeopold who can help me? 2011-10-30T10:46:27 ahh 2011-10-30T10:46:36 *** Cybsy has joined #aichallenge 2011-10-30T10:46:38 FCK42: amstan, but he's not here, maybe janzert 2011-10-30T10:47:52 How to use tcpserver? My client print "connected to ants.fluxid.pl:2081 as Belerafon" and... where are my game? :) 2011-10-30T10:48:38 Belerafon: I think you have to wait until there's enough players? 2011-10-30T10:48:39 Belerafon: wait 2011-10-30T10:48:51 oh wait 2011-10-30T10:48:59 this fucking tcp server crashed agaion 2011-10-30T10:49:14 20 minutes ago 2011-10-30T10:49:33 or rather hung, ^C helped... 2011-10-30T10:49:43 oh, ok. Now i see a lot of stuff on terminal :) 2011-10-30T10:50:49 Hmm there is maximum turntime on "Game Settings" page (http://aichallenge.org/game_settings.php), but what about memory usage? 2011-10-30T10:51:03 1.5 GB at the least 2011-10-30T10:52:13 i think i'll replace this server with a different version, please bear with me for a while 2011-10-30T10:52:50 *** racko has quit IRC (Quit: Page closed) 2011-10-30T10:52:52 What is Mu and Sigma in statistics? 2011-10-30T10:52:57 how can I ask janzert wheather he can help me? 2011-10-30T10:53:18 Belerafon: lookup trueskill in google 2011-10-30T10:53:30 Belerafon: mu is computed skill, sigma is "uncertainty" 2011-10-30T10:53:41 Belerafon: It's a measure of how much the system knows about your skill. 2011-10-30T10:53:41 thanks 2011-10-30T10:53:44 showed skill is mu-3*sigma 2011-10-30T10:53:50 For a long-form article about this: http://www.moserware.com/2010/03/computing-your-skill.html which 2011-10-30T10:54:03 the lower the sigma, the more certain is "real" skill - mu 2011-10-30T10:54:05 discusses both Elo and TrueSkill and how the statistics apply 2011-10-30T10:54:16 Belerafon: doing that stops someone winning by luck the first few games from ending up on #1 2011-10-30T10:54:27 also the higher the sigma, the higher changes to skill are made if you fail/win 2011-10-30T10:55:20 jbroman: oh, nice article, thanks 2011-10-30T10:56:23 all: tcp server is down for few minutes 2011-10-30T11:01:58 Hi, is there documentation on how the TCP server works ? My first impression is that it pipes my bot IO to the TCP server instead of to the stdio. 2011-10-30T11:02:09 what are the test cases? I got a working bot but fails a test case apparantly 2011-10-30T11:02:33 Ionic_Groove: there is a shellscript in the tools package 2011-10-30T11:03:38 *** Baus has quit IRC (Ping timeout: 258 seconds) 2011-10-30T11:03:52 Ionic_Groove: test_bot? 2011-10-30T11:04:33 *** bmh has joined #aichallenge 2011-10-30T11:04:37 *** bmh has joined #aichallenge 2011-10-30T11:04:44 good morning 2011-10-30T11:05:11 so I replace submission_test/TestBot.py with my script and it should work? 2011-10-30T11:05:19 *** irchs has quit IRC (Quit: derp) 2011-10-30T11:07:09 *** kj has joined #aichallenge 2011-10-30T11:07:17 hi guys, i just tried to upload my first bot 2011-10-30T11:07:21 and i get 2011-10-30T11:07:24 turn 1 bot 0 invalid actions: 2011-10-30T11:07:25 o 1 18 � # invalid direction 2011-10-30T11:07:35 *** kj is now known as Guest96898 2011-10-30T11:07:47 any idea how to debug this? 2011-10-30T11:08:03 or what this means? 2011-10-30T11:08:08 TCP server is back, should work 2011-10-30T11:08:25 *** Jubjub has joined #aichallenge 2011-10-30T11:08:40 Fluxid: cool, thank you 2011-10-30T11:09:06 I guess not 2011-10-30T11:09:09 *** Baus has joined #aichallenge 2011-10-30T11:09:23 *** Sapog has joined #aichallenge 2011-10-30T11:09:45 hi 2011-10-30T11:09:46 *** humphrey1239 has quit IRC (Quit: Page closed) 2011-10-30T11:11:14 timos: you can give there only four letters 2011-10-30T11:11:28 you gave there... some kind of trash, i see a square ;) 2011-10-30T11:11:42 timos: did you test it with test_bot? 2011-10-30T11:11:49 *** smf68 has joined #aichallenge 2011-10-30T11:12:48 how do you test with test_bot 2011-10-30T11:13:22 Fluxid: yes, i tested with ./test_bot.sh and i get the same error 2011-10-30T11:13:36 but without the garbage 2011-10-30T11:14:07 *** bearoff_w has left #aichallenge 2011-10-30T11:14:09 Fluxid: just o 1 18 2011-10-30T11:14:43 *** AxVapor has joined #aichallenge 2011-10-30T11:14:44 *** AxVapor has left #aichallenge 2011-10-30T11:15:15 Fluxid: I don't know much about the output format, this is just the cpp starter package with a modified move function 2011-10-30T11:15:24 *** AxVapor has joined #aichallenge 2011-10-30T11:15:53 @spec 2011-10-30T11:15:54 Fluxid: Run as fast as you can and don't look back. 2011-10-30T11:15:59 @specification 2011-10-30T11:16:00 Fluxid: User error -- Replace user. 2011-10-30T11:16:02 bah 2011-10-30T11:16:06 Is here really nobody who can read this and help me? 2011-10-30T11:16:20 http://aichallenge.org/specification.php 2011-10-30T11:16:58 *** joakimar has joined #aichallenge 2011-10-30T11:17:04 timos: ↑ 2011-10-30T11:17:53 Fluxid: ok, i understand the spec for the moves now 2011-10-30T11:20:26 *** antiwhack has quit IRC (Quit: Page closed) 2011-10-30T11:20:32 Is there an ability to get the activation e-mail again? 2011-10-30T11:20:48 *** Guest96898 has quit IRC (Quit: Page closed) 2011-10-30T11:21:13 Fluxid: in this game http://aichallenge.org/visualizer.php?game=33565&user=757 your ants seem to thrash around your surviving hill for a while -- what's going on there? 2011-10-30T11:21:28 *** sigh has quit IRC (Remote host closed the connection) 2011-10-30T11:21:37 Fluxid: ok, I found the problem :-) thanks for the spec link 2011-10-30T11:23:35 bmh: i dispatch 4 ants to guard my hill 2011-10-30T11:23:59 Fluxid: why four? So you don't get fragged by assassins or ants working the buddy system? 2011-10-30T11:25:02 bmh: originally it was six, but this was too much. i simply didn't want to leave my hill unguarded, so one dumb enemy ant couldn't raze it alone 2011-10-30T11:25:36 you seem to be doing pretty well, so perhaps my inquiries are misplaced :) 2011-10-30T11:25:44 *** pvarga_ has joined #aichallenge 2011-10-30T11:26:10 i just make them move randomly within radius of 1 around hill... 2011-10-30T11:26:24 *** pvarga has quit IRC (Ping timeout: 240 seconds) 2011-10-30T11:26:24 *** pvarga_ is now known as pvarga 2011-10-30T11:26:25 do you decide to guard your hill when you only have 1 left or when a specific number of ants is reached 2011-10-30T11:27:23 ikaros: after turn 40, for each new ant spawned at given hill, if it is unguarded there is 50% chance it will become a guard. 2011-10-30T11:27:31 ah nice 2011-10-30T11:27:33 *** delt0r__ has quit IRC (Ping timeout: 255 seconds) 2011-10-30T11:27:37 *** x10ded has joined #aichallenge 2011-10-30T11:28:01 i'm a bit uncertain where to proceed right now 2011-10-30T11:28:21 before i spawn only two kind of explorers - one try to explore map, and one to try keep as much map visible as possible 2011-10-30T11:28:34 ikaros: watch your bot play some games and look for stupid things it does 2011-10-30T11:28:37 i wanted to implement a hill defense like yours next but it seems the top bots don't do this at all 2011-10-30T11:28:43 ikaros: then fix the stupid things :) 2011-10-30T11:28:46 yea i have ideas :) 2011-10-30T11:28:56 it's just what's most important :) 2011-10-30T11:28:57 http://ants.fluxid.pl/replay.13179 is an interesing game 2011-10-30T11:29:21 I think there was a better one. A starts covering hills to get ants to correct places. 2011-10-30T11:29:22 ikaros: I plan on determining a polygon in which enemy ants *cannot* be and ensuring that I can get ants to my hill if the line is breached 2011-10-30T11:29:30 i am doing the dumbest thing ever... working on a monolithic AI without testable intermediate stages 2011-10-30T11:29:36 callahan: who am I watching? 2011-10-30T11:29:47 bmh: A 2011-10-30T11:30:38 this isn't the most interesting game of A 2011-10-30T11:30:44 http://ants.fluxid.pl/replay.13178 for a more pronounced version 2011-10-30T11:30:46 I know 2011-10-30T11:31:03 Can somebody of you tell someone who can do it to send me the activatione-mail again? 2011-10-30T11:31:04 *** keitherz has joined #aichallenge 2011-10-30T11:31:07 I just thought it was the more interesting version of parking ants by the hive. 2011-10-30T11:31:12 interesting, yeah 2011-10-30T11:31:19 *** rioniac has quit IRC (Ping timeout: 265 seconds) 2011-10-30T11:31:35 starting with AI 2011-10-30T11:31:43 goodbye 2011-10-30T11:31:48 FCK42: sorry 2011-10-30T11:31:50 I've seen people start to park ants there. Someone parks them at the corners which seems silly because that doesn't maximize their coverage of each other. 2011-10-30T11:32:00 maybe wrong timezone 2011-10-30T11:32:03 *** FCK42 has quit IRC () 2011-10-30T11:32:46 callahan: anything that has corners is bad 2011-10-30T11:32:55 *** amstan_ has joined #aichallenge 2011-10-30T11:32:55 *** ChanServ sets mode: +o amstan_ 2011-10-30T11:33:09 I mean pgoldbr_2's behavior 2011-10-30T11:33:20 and here's amstan... just after FCK42 decided to quit 2011-10-30T11:33:21 4 ants on each hive, one per corner 2011-10-30T11:33:33 yeah 2011-10-30T11:33:59 i saw four ants doing rounds around hive 2011-10-30T11:34:10 just wondering, is there a documentation for all the list of functions and variables in the C++ starter kit? 2011-10-30T11:34:26 or four stationary anst, but not on corners but nsew of hill 2011-10-30T11:34:27 Look at the code I suppose? 2011-10-30T11:34:50 *** Rinum has quit IRC (Quit: Page closed) 2011-10-30T11:34:52 and when ant spawn, they move away to allow it to go down from the hill 2011-10-30T11:35:06 I don't think the 4 corners thing works. Seems to be quick hack to make up for the fact that your ants are terrible enough to let one wander all the way over to your hive. 2011-10-30T11:35:35 ;) 2011-10-30T11:35:42 Hello, can someone help me, please, I've got a problem 2011-10-30T11:36:35 Why does this .cmd script doesn't work 2011-10-30T11:36:38 cd "C:\Users\Olexandr\Desktop\ants" set bot="python tools\sample_bots/python\HunterBot.py" python tools/playgame.py "ants.exe" %bot% --map_file "tools\maps\maze\maze_02p_01.map" --log_dir game_logs --turns 500 --scenario --player_seed 7 --verbose -e 2011-10-30T11:39:39 *** Seebie has joined #aichallenge 2011-10-30T11:40:30 *** x10ded has quit IRC (Quit: Page closed) 2011-10-30T11:41:07 *** delt0r__ has joined #aichallenge 2011-10-30T11:41:12 *** amstan_ has quit IRC (Ping timeout: 240 seconds) 2011-10-30T11:41:27 I think the ants.fluxid.pl tcp server is stuck again 2011-10-30T11:42:45 *** Seebie has quit IRC (Client Quit) 2011-10-30T11:44:05 *** Puj has joined #aichallenge 2011-10-30T11:47:56 if its same code what i tried few months ago, it served like 10 games and then got stuck 2011-10-30T11:48:43 :( 2011-10-30T11:57:51 *** Palmik has joined #aichallenge 2011-10-30T12:07:15 *** bmh has quit IRC (Quit: bmh) 2011-10-30T12:08:44 *** epicmonkey has quit IRC (Remote host closed the connection) 2011-10-30T12:13:33 *** epicmonkey has joined #aichallenge 2011-10-30T12:14:21 henry61: IndexError: list index out of range 2011-10-30T12:14:24 awesome 2011-10-30T12:14:39 i'll try to fix it 2011-10-30T12:18:29 henry61: added guard code, restarted 2011-10-30T12:18:42 sweet 2011-10-30T12:21:12 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-10-30T12:21:15 duh, to commit/push to fork this i'd need to rebase 2011-10-30T12:21:21 aaah, i don't want to do it now 2011-10-30T12:23:25 *** capt_wilkerson has joined #aichallenge 2011-10-30T12:23:27 mmmh this is hard 2011-10-30T12:23:39 *** capt_wilkerson has left #aichallenge 2011-10-30T12:23:44 AI is harder than microcontroller programming 2011-10-30T12:24:03 the only thing that makes ai remotely hard.. is performance 2011-10-30T12:25:07 well, that sort of depends on the problem 2011-10-30T12:25:25 but most of these games would be a lot simpler with infinite computing power and memory 2011-10-30T12:25:27 *** ikaros has joined #aichallenge 2011-10-30T12:25:51 this game(like GW), would be trivial with infinite processing 2011-10-30T12:25:53 not that used to OOP 2011-10-30T12:26:15 and i stil doesn't have a path finding algorithm 2011-10-30T12:30:47 *** andy_ has joined #aichallenge 2011-10-30T12:31:13 *** andy_ is now known as Guest29668 2011-10-30T12:33:12 *** dvladim has quit IRC (Ping timeout: 244 seconds) 2011-10-30T12:33:33 *** AxVapor has quit IRC (Quit: AxVapor) 2011-10-30T12:34:58 *** bmh has joined #aichallenge 2011-10-30T12:34:58 *** bmh has joined #aichallenge 2011-10-30T12:38:59 is ants.fluxid.pl working correctly atm? 2011-10-30T12:39:34 haven't used it before, so I don't know whether it should output anything to console? 2011-10-30T12:40:27 shows only "connected to ants.fluxid.pl:2081 as draakon" 2011-10-30T12:40:32 *** npskirk has joined #aichallenge 2011-10-30T12:41:44 account help? 2011-10-30T12:41:52 *** pvarga has quit IRC (Quit: pvarga) 2011-10-30T12:44:01 Draakon: I think it's broken right now 2011-10-30T12:44:11 Oh, okai.. any other good tcp server? 2011-10-30T12:44:25 *** Chirmaya has joined #aichallenge 2011-10-30T12:44:59 not that I know of, I've been happily using this until now 2011-10-30T12:46:17 Anyone know why I would be getting a "status timeout survived" after only 274ms in a turn? (Happens on the 90-150th turn >.<) 2011-10-30T12:47:50 introductions have to be the hardest things to write 2011-10-30T12:49:21 Is it possible to delete an account? 2011-10-30T12:49:46 *** pvarga has joined #aichallenge 2011-10-30T12:51:53 *** overburn has joined #aichallenge 2011-10-30T12:52:10 umm guys, what does the "duplicate order" error mean? 2011-10-30T12:52:41 you tried to order the same ant twice in one round 2011-10-30T12:52:50 ahm , ok thanks :D 2011-10-30T12:53:28 are some of those example bots causing duplicate orders and etc. errors also? it seems to me at least 2011-10-30T12:53:29 *** dom7b5 has quit IRC (Ping timeout: 265 seconds) 2011-10-30T12:53:54 yeah I know at least greedybot makes invalid orders 2011-10-30T12:54:35 you can run fights against them with the toolkit, if you set it to strict it will kick them out if they misbehave 2011-10-30T12:54:48 *** epicmonkey has quit IRC (Remote host closed the connection) 2011-10-30T12:55:01 *** epicmonkey has joined #aichallenge 2011-10-30T12:57:59 does anyone know any tcp server that is currently working fine? 2011-10-30T12:58:15 there are some links on the forums, but haven't got any of them working 2011-10-30T12:58:25 so I don't know whether I'm doing something wrong or what 2011-10-30T12:58:29 Is it possible to delete or rename an account? 2011-10-30T12:59:05 I'd be happy to run another TCP server if someone tells me how 2011-10-30T13:00:42 Is there any way to get a more descriptive error message about a timeout from the python tool? 2011-10-30T13:01:48 Chirmaya, usually it depends on a crash in your bot, you can output those crash messages using the -E flag in the playgame.py options. You must also specify a game_logs folder to save all of the logging to. 2011-10-30T13:02:40 I have that on, but all it says is "status timeout survived" 2011-10-30T13:02:50 *** pvarga has quit IRC (Quit: pvarga) 2011-10-30T13:03:28 Unless there is a new version of the tool? 2011-10-30T13:03:28 So nothing from stderr indicates an exception, or runtime error? Are you getting other stderr output? Chirmaya? 2011-10-30T13:04:05 Yeah, I am getting it since I use that for checking my timings 2011-10-30T13:04:15 *** UncleVasya has joined #aichallenge 2011-10-30T13:05:00 Chirmaya, well then either you can do some very verbose output through stderr indicating, for example, the start and end points of your methods, or you can use an appropriate profiler. 2011-10-30T13:05:21 And trying piping in the 0.bot0.input, for example. 2011-10-30T13:06:10 I have done the verbose method, but all of my times were around 100-300 ms, so I know that it isn't timing out on my end 2011-10-30T13:06:59 maybe your computer and/or compiler, are better than on live? 2011-10-30T13:07:00 Check turntime in the playgame file, and make sure it is not set to 250ms, Chirmaya. 2011-10-30T13:07:17 Says 1000ms 2011-10-30T13:07:42 Are you developing in Python? 2011-10-30T13:07:49 Ruby 2011-10-30T13:08:42 I'm not sure then. I've been experiencing some strange timeouts in Python, but I assumed that was some clash with cygwin. Eclipse seems to work quite a bit better with the tools. 2011-10-30T13:10:23 Hmm. So maybe my Ruby believes that it times out because of a long (though not infinite loop) and sends that message to the python tool? 2011-10-30T13:11:19 Well, the Python tool keep tracks of all of the Bots and how much time they take per turn. 2011-10-30T13:11:43 It checks periodically if all the bots are finished and then continues when all of the bots are finished or the time limit runs out. 2011-10-30T13:11:58 *** npskirk has left #aichallenge 2011-10-30T13:12:17 It wouldn't be possible that an opposing bot is the one doing the timeout, is it? 2011-10-30T13:12:24 *** master_ninja has quit IRC (Quit: now my ants are ready for battle) 2011-10-30T13:16:18 Hmm, just changed the bot I am facing from HunterBot, and it seems to be working much better (over 200 turns so far) 2011-10-30T13:16:29 Maybe it just was that bot 2011-10-30T13:16:55 . 2011-10-30T13:17:20 the other bot is causing you to timeout? 2011-10-30T13:17:36 Going to try a few different maps to check 2011-10-30T13:18:10 *** racko has joined #aichallenge 2011-10-30T13:18:17 Successfully passed one map, but going to try one that usually timed out at about 50 turns 2011-10-30T13:18:48 *** thenewbie has joined #aichallenge 2011-10-30T13:19:23 *** thenewbie has left #aichallenge 2011-10-30T13:24:17 *** bmh has quit IRC (Quit: bmh) 2011-10-30T13:28:36 *** teak42 has joined #aichallenge 2011-10-30T13:31:39 *** teak42 has quit IRC (Client Quit) 2011-10-30T13:39:55 \he 2011-10-30T13:40:57 *** Sapog has quit IRC (Quit: Page closed) 2011-10-30T13:41:28 *** xathis has joined #aichallenge 2011-10-30T13:42:04 *** roflmao has joined #aichallenge 2011-10-30T13:42:09 help! 2011-10-30T13:42:38 ? 2011-10-30T13:42:42 uuh.. 2011-10-30T13:42:43 * a1k0n throws life raft 2011-10-30T13:42:48 haha 2011-10-30T13:42:56 I'm trying to implement a* in ruby, it works but the performance is terrible. 2011-10-30T13:43:01 with a manhattan heuristic estimate of 41 2011-10-30T13:43:04 hm. yep. 2011-10-30T13:43:12 it takes 3 seconds to find the path 2011-10-30T13:43:18 and I end up with 1678 nodes in the open set 2011-10-30T13:43:25 is that normal or ridiculously high? 2011-10-30T13:43:50 sounds pretty high to me. depends on the map obviously. 2011-10-30T13:43:50 pathfinding is expensive and ruby is notoriously slow 2011-10-30T13:43:58 but... what minthos said 2011-10-30T13:44:07 check that you're not revisiting the same node twice 2011-10-30T13:44:25 oh i had the forumla fucked up 2011-10-30T13:44:28 down to 89 open nodes 2011-10-30T13:44:31 =) 2011-10-30T13:44:34 heh. 2011-10-30T13:44:36 it's weird though... 2011-10-30T13:44:42 this is an independent a* algorithm 2011-10-30T13:44:55 once I integrate it into ants i get huge open_node sets again :-\ 2011-10-30T13:45:22 *** tmandry has joined #aichallenge 2011-10-30T13:45:33 *** tmandry has joined #aichallenge 2011-10-30T13:48:18 roflmao: All the paths in ants have the same cost. That is the problem. When you stand in a corner on an empty map and move to the center you can litterally take any path and A* checks them all. 2011-10-30T13:48:58 yeh :-\ 2011-10-30T13:49:12 for heuristics (manhattan) of around 20 it works well in ants 2011-10-30T13:49:24 anything above 35 distance goes crazy slow cause the set of open nodes is enormous 2011-10-30T13:49:24 so use an inadmissible heuristic 2011-10-30T13:49:34 what's that? 2011-10-30T13:49:39 20? You set the heuristics to a constant of 20? 2011-10-30T13:49:42 something that is not an underestimate. 2011-10-30T13:49:50 no 2011-10-30T13:50:05 I'm just saying that when the ant is going to calculate a path that has an estimated manhattan distance of 20 it performs well 2011-10-30T13:50:13 use 5*euclidean distance, for instance, and you'll get a path that isn't necessarily optimal but is quicker to compute. 2011-10-30T13:50:24 a1k0n: inadmissible heuristic means he has to get nodes back from the closed set, right? 2011-10-30T13:50:48 *** mj41 has quit IRC (Ping timeout: 245 seconds) 2011-10-30T13:50:48 one thing that helped (but didn't help enough) was that I weighted the heuristic 2011-10-30T13:51:22 node priority = estimated distance to food MINUS distance already travelled 2011-10-30T13:51:43 but, obviously, that ends up favoring long paths 2011-10-30T13:52:11 huh? 2011-10-30T13:52:22 You can get rid of the "open nodes with equal value" by using a tie-breaker 2011-10-30T13:52:36 roflmao: shouldn't it be (distance traveled + estimated remaining)? 2011-10-30T13:52:46 mleise: eh? not that i know of 2011-10-30T13:52:55 yes that's what it should be lavalamp but that leads to crazy huge open node sets 2011-10-30T13:52:58 *** bmh has joined #aichallenge 2011-10-30T13:52:58 *** bmh has joined #aichallenge 2011-10-30T13:53:07 but that does find the best possible route 2011-10-30T13:53:27 by simply adding + distance traveled the open set goes from 80 to 1k 2011-10-30T13:53:48 roflmao, a1k0n: http://theory.stanford.edu/~amitp/GameProgramming/Heuristics.html#breaking-ties 2011-10-30T13:53:49 but.. you've messed up the definition of path cost 2011-10-30T13:54:18 The range of the tie-breaker must remain a fraction of the cost (i.e. < 1) 2011-10-30T13:54:28 from one node to the next 2011-10-30T13:54:29 nice. 2011-10-30T13:54:32 if you switch to euclidian distance instead of manhattan you may not get the optimal path but the open set should stay much smaller 2011-10-30T13:54:48 you'll still get the optimal path. 2011-10-30T13:55:00 it's an underestimate of the distance, and is hence admissible 2011-10-30T13:55:04 well, you'll get *an* optimal path :) 2011-10-30T13:55:12 on this tiebreaker 2011-10-30T13:55:13 heh. touché 2011-10-30T13:55:33 basically I'm sorting my nodes in the open set by lowest cost 2011-10-30T13:55:33 wtf 2011-10-30T13:55:40 this server crashed again 2011-10-30T13:55:46 obv cost is determined by estimated distance to goal 2011-10-30T13:55:49 so this tie breaker thing 2011-10-30T13:55:59 not entirely sure I get it 2011-10-30T13:56:13 p <(minimum cost of taking one step)/(expected maximum path length) 2011-10-30T13:56:15 Fluxid: yours or the official worker? 2011-10-30T13:56:21 every step in ants costs the same 2011-10-30T13:57:19 so that will always be 1 / expected maximum path length? 2011-10-30T13:57:39 *** capa has joined #aichallenge 2011-10-30T13:57:55 mleise: amit is awesome! :) 2011-10-30T13:58:00 a1k0n: hiya! 2011-10-30T13:59:04 a1k0n: wht do you mean? 2011-10-30T13:59:18 bmh: yep 2011-10-30T13:59:33 Fluxid: you said a server crashed 2011-10-30T13:59:37 I've exchanged a little bit of e-mail with him on terrain generation. Very friendly guy. 2011-10-30T13:59:42 bmh: hey, sup 2011-10-30T13:59:46 Fluxid: any chance you could set a watchdog script to restart it when it crashes? 2011-10-30T14:00:24 Fluxid: Are your server startup scripts public? I'd be happy to run an additional server 2011-10-30T14:00:26 just to make sure I understand 2011-10-30T14:00:32 f_score is what is used to determine node priority 2011-10-30T14:00:33 correct? 2011-10-30T14:01:00 a1k0n: my tcp 2011-10-30T14:01:07 ah 2011-10-30T14:01:15 Minthos: this code sucks, it crashes but only one thread 2011-10-30T14:01:19 so proces still runs 2011-10-30T14:01:33 how much margin do you guys leave for the server to process your input? I mean, do you break your main loop at 490ms or like 499.9ms? 2011-10-30T14:01:43 i added another fix 2011-10-30T14:02:03 so it's up again 2011-10-30T14:02:07 sorry for the problems :( 2011-10-30T14:02:15 Migi32: I was planning on ignoring the clock and just praying that I could get everything done in time 2011-10-30T14:02:23 500ms is a *long* time 2011-10-30T14:03:05 bmh: i would recommend against that approach 2011-10-30T14:03:06 bmh: I use this: https://github.com/berak/ants-tcp 2011-10-30T14:03:34 i begin to think about timeouts if they happen :) 2011-10-30T14:03:35 bmh: yeah I was doing the same but my pathfinding preprocessing algorithm takes very variable time. Sometimes it's 50ms, sometimes 2000 (in which case I have to break the calculations up in pieces, which is kind of tricky) 2011-10-30T14:03:51 a1k0n: in that it's dangerous or that you risk wasting cycles? 2011-10-30T14:03:52 i used ashod's fork at the beginning, but it went out of date with berak changes after a while 2011-10-30T14:04:09 bmh: both! 2011-10-30T14:04:28 but maybe now it will be more stable 2011-10-30T14:04:37 lets hope so Fluxid 2011-10-30T14:04:41 wtf 2011-10-30T14:04:49 a1k0n: I'll delegate that to someone else on the team. I'm concentrating on infrastructure 2011-10-30T14:05:12 I've just set my pathfinding algo to return failure if remaining time is less than 10% 2011-10-30T14:05:16 i wonder why irssi doesn't always highlight my nick 2011-10-30T14:05:17 * bmh is destine to become a manager 2011-10-30T14:05:31 in practice that only happens when I have >100 ants so it should be fine 2011-10-30T14:05:56 Minthos, what pathfinding algorithm do you use? A*? 2011-10-30T14:06:15 nah, a custom dijkstra-based thing 2011-10-30T14:06:21 k 2011-10-30T14:06:23 g_score[neighbor] = g_score[node] + 1 2011-10-30T14:06:23 f_score[neighbor] = heuristic(neighbor, goal) + g_score[neighbor] 2011-10-30T14:06:28 but I'm going to write an A* for point to point paths 2011-10-30T14:06:33 and use both 2011-10-30T14:06:36 *** danielharan has joined #aichallenge 2011-10-30T14:06:36 if I add the + g_score my algorithm crashes, if I remove it the open set is 80 nodes 2011-10-30T14:06:45 (nodes are sorted by f_score of course) 2011-10-30T14:06:50 *** merk has joined #aichallenge 2011-10-30T14:07:11 *** Murasha has joined #aichallenge 2011-10-30T14:07:13 hallo guys 2011-10-30T14:07:16 but if you remove it your algorith is wrong 2011-10-30T14:07:29 but shouldn't the + g_score REDUCE the amount of nodes trasversed? 2011-10-30T14:07:41 it instead increases them ten fold, even with a tiebreaker on the heuristic 2011-10-30T14:07:48 ((start.row - goal.row).abs + (start.col - goal.col).abs) * (1.0 + 1/1000) 2011-10-30T14:07:56 is your priority queue taking the maximum instead of the minimum? 2011-10-30T14:08:05 nope it sorts by minimum 2011-10-30T14:08:20 (i only ask because i had that problem yesterday) 2011-10-30T14:08:21 *** Chirmaya has quit IRC (Quit: Page closed) 2011-10-30T14:08:23 ok but back to my first question. Break the main loop at 490ms, 499ms or 499.9ms? 2011-10-30T14:08:26 a1k0n: haha, I did that too :O 2011-10-30T14:09:02 Migi32: what's the variance in your inner loop iteration? 2011-10-30T14:09:18 pretty much 0. 2011-10-30T14:09:38 then 499.9ms? 2011-10-30T14:09:50 does anyone here 2011-10-30T14:09:55 codes the bot in c? 2011-10-30T14:10:05 all right, ants.fluxid.pl is up again and my bot needs opponents! 2011-10-30T14:10:08 C++ count? 2011-10-30T14:10:16 yeah, but if you submit at 499.9ms, don't you risk that the server only reads your input at 500.1ms? 2011-10-30T14:10:18 thanks Fluxid :) 2011-10-30T14:10:20 Minthos: i've connected 2011-10-30T14:10:38 np 2011-10-30T14:10:44 Migi32: yeah, potentially. i guess we don't know what the scheduling slop is going to be. 2011-10-30T14:11:30 oh shiiiit 2011-10-30T14:11:42 I think I'm going with 495 or something. I know it's kind of a big margin but that extra 1% of time probably won't do much anyway 2011-10-30T14:11:55 *** Saulzar has quit IRC (Ping timeout: 260 seconds) 2011-10-30T14:12:42 *** Bostown has joined #aichallenge 2011-10-30T14:12:47 I have duplicates in my open set rofl 2011-10-30T14:12:50 Hello 2011-10-30T14:12:58 for timing I wouldn't use a normal distribution 2011-10-30T14:13:16 is it just me or a* doesn't seem to be the best choice for pathfinding on this specific problem? 2011-10-30T14:13:29 overburn: define best 2011-10-30T14:13:36 @overburn: when mine doesn't time out it is a beast 2011-10-30T14:13:37 overburn: "this problem" has many subproblems 2011-10-30T14:13:37 roflmao: I'm sorry Dave, err roflmao; I cannot 'overburn:'. 2011-10-30T14:13:53 Migi32: do you so many stuff you need to limit time? 2011-10-30T14:14:09 yeah 2011-10-30T14:14:15 Migi32: what language>? 2011-10-30T14:14:45 ah , ok , good point. meaning for the basic pathfinding from point a to point b , without taking into account the choices of the programmer when it encounters an enemy ant or such 2011-10-30T14:15:04 C++, but it's not the language that's slow or the way I program, it's just the complexity of the algorithms I use 2011-10-30T14:15:06 I'm looking into segmenting the map and then doing pathfinding on the abstract graph produced by the segmentation 2011-10-30T14:15:44 wtf? Why so many people have problems with slow pathfinding? I launch simple BFS several times (2 for food, 3 for enemy hiils, 2 for boundary) and still have a lot of time. Bot spend nearly 160ms for the turn. I wonder if it's only thanks to the fast language (OCaml) or your A* is something weird. 2011-10-30T14:15:56 Migi32: for sufficiently large N, I can execute a linear algorithm by hand faster than you can execute an exponential algorithm on all the computers in the world :) 2011-10-30T14:16:07 My a* is probably weird 2011-10-30T14:16:24 the thing is, trying to get my bot up and running and need a really simple pathfinding algorithm to start with , 2011-10-30T14:16:26 why the fuck do i have duplicates onmy open set 2011-10-30T14:16:30 I don't use A*. I use something with a lot of preprocessing time but O(1) lookup 2011-10-30T14:16:33 can anyone recomment one that can be implemented really fast? 2011-10-30T14:17:06 overburn: Breadth-first-search 2011-10-30T14:17:21 UncleVasya: when you're trying to get a really fast A* implementation, it's easy to make mistakes. 2011-10-30T14:17:23 UncleVasya: my bots usually don't hit the time limit when I run it locally, but on the official server they do when I have around 100-150 ants on big maps 2011-10-30T14:17:36 written in D 2011-10-30T14:17:56 hmm thanks uncle 2011-10-30T14:18:29 Minthos: So my advice is to make search from destinations. :) 2011-10-30T14:18:33 Just one question,,, Im trying to start now, and get cant get it working with the tutorials, can someone help me ? 2011-10-30T14:18:56 UncleVasya: I search both directions in different parts of my ai 2011-10-30T14:18:57 don't forget that it also depends on how often you're asking for a distance calculation 2011-10-30T14:19:13 ah.. Fluxid's server is back? 2011-10-30T14:19:16 UncleVasya: but yes, I need better pathfinding, that's my next step 2011-10-30T14:19:20 yes 2011-10-30T14:19:30 I'm doing it far more often than just asking for the distance between each ant and each food/enemy hill 2011-10-30T14:19:31 http://aichallenge.org/visualizer.php?game=35057&user=4823 here's a good example 2011-10-30T14:19:36 Draakon: yes 2011-10-30T14:19:47 is it normal to have duplicate nodes in the open set? 2011-10-30T14:19:49 no right? 2011-10-30T14:19:57 at some point here my ai just breaks down from starvation and my ants stop moving 2011-10-30T14:20:16 roflmao: correct. 2011-10-30T14:20:16 @Minthos: Here's what I do. If food is visible, I use a* to hunt down that food 2011-10-30T14:20:17 roflmao: You have no gotten any error messages recently, so here's a random one just to let you know that we care. 2011-10-30T14:20:19 I installed one myself for shits and giggles while yours was down.. can someone check whether its even accessible for outside world? 2011-10-30T14:20:21 duplicates = you're doing it wrong 2011-10-30T14:20:25 http://ants.ahju.eu:2080/ 2011-10-30T14:20:47 Draakon: seems to work 2011-10-30T14:20:54 it seems to crash really easily 2011-10-30T14:21:00 at least the tcpserver.py part 2011-10-30T14:21:22 for example if I send some misformed packets 2011-10-30T14:21:25 Minthos: at turn 628 your ants suddenly move, what happens there? 2011-10-30T14:21:37 Draakon: yeah, code is not of the best quality 2011-10-30T14:21:50 that was the problem 2011-10-30T14:21:53 that was the fkin problem. WOW 2011-10-30T14:21:54 also, who is darko, he uses invalid passowrd 2011-10-30T14:21:55 Minthos: are your ants on guard duty? :D 2011-10-30T14:22:01 *** invader has joined #aichallenge 2011-10-30T14:22:07 also lol, passwords are in plaintext in db -_-' 2011-10-30T14:22:23 Fluxid: no idea, maybe the server scheduled my process more than usual 2011-10-30T14:22:31 lol 2011-10-30T14:22:38 emilk: 4 around each hill yes 2011-10-30T14:22:45 WHy is it that I am unable to go on top of the food?? 2011-10-30T14:22:58 Fluxid: you could sell them for few $ ;) 2011-10-30T14:23:03 Minthos: thats a nice touch... I'm gonna steal that ;) 2011-10-30T14:23:05 Draakon: no way 2011-10-30T14:23:11 just kidding.. 2011-10-30T14:23:12 heuristic 59, open nodes are 55 2011-10-30T14:23:14 woohooo! 2011-10-30T14:23:16 i'm not of that kind 2011-10-30T14:23:17 emilk: go right ahead, I stole it from someone else :) 2011-10-30T14:23:39 Minthos: what scerver's scheduling has to do with it? O_o 2011-10-30T14:23:48 given that everyone is using 4 defenders, I'm going to hardcode 5 ant kill squads 2011-10-30T14:23:54 Fluxid: this match was on the official server, not yours 2011-10-30T14:24:14 Minthos: i know 2011-10-30T14:24:36 i just wonder what in your bot made all of your ants move for two turns 2011-10-30T14:24:41 ;) 2011-10-30T14:25:05 I assume my bot doesn't get exclusive cpu time, so maybe a game just finished and it got more cpu time than normal 2011-10-30T14:25:16 it's the only explanation I can think of 2011-10-30T14:25:18 numpy is officially allowed and works, right? 2011-10-30T14:25:52 *** xathis has quit IRC () 2011-10-30T14:25:53 Minthos: this still doesn;t make sense 2011-10-30T14:26:09 bot are run sequentialy, one has whole core for itself 2011-10-30T14:26:17 to make game fair 2011-10-30T14:26:44 Minthos: do you limit time? 2011-10-30T14:26:59 it works! IT WORKS!! 2011-10-30T14:27:11 congrats 2011-10-30T14:27:23 yes, I abort all pathfinding when remaining time is less than 10% 2011-10-30T14:27:31 to avoid timeouts 2011-10-30T14:27:38 Fluxid: bots are not run sequentally... you are alowed to use non blocking IO... so while you wait for input from the server you could be doing something usefull 2011-10-30T14:28:14 Minthos: then i think you need to optimize hard ;) 2011-10-30T14:28:17 ie when the other bots are running... so to do the round robin thing, the rules would need to explicitly ban that 2011-10-30T14:28:30 Fluxid: yes, that's my next step 2011-10-30T14:29:03 delt0r__: ok, maybe i misunderstood. 2011-10-30T14:30:01 Hi anyone knows how to debug it? I'm programming in C. 2011-10-30T14:30:02 Minthos: what language? 2011-10-30T14:30:09 D 2011-10-30T14:30:32 never played with it 2011-10-30T14:30:42 it's nice but still a bit immature 2011-10-30T14:30:43 capa: debug what? bot? log to stderr 2011-10-30T14:30:59 *** Islacrusez has joined #aichallenge 2011-10-30T14:31:04 had some problems getting my bot to compile right on the server 2011-10-30T14:31:10 *** AD1960 has joined #aichallenge 2011-10-30T14:31:13 Minthos: Looling at your submissions history I can say that some of the future ones will be on E,F,G etc. languages :) 2011-10-30T14:31:17 *** asdasda has joined #aichallenge 2011-10-30T14:31:18 *looking 2011-10-30T14:31:19 why does my tutorial bot keep timing out? o.O 2011-10-30T14:31:21 hehe 2011-10-30T14:31:58 3 of 8 matches have resulted in timeout >< 2011-10-30T14:32:02 Fluxid: I want to know what is going on with my variables 2011-10-30T14:32:15 Islacrusez: you use too much time 2011-10-30T14:32:30 write faster code 2011-10-30T14:32:39 and check remaining time in your code 2011-10-30T14:32:39 Can someone help me setting up the game, cant get it working with the tutorials 2011-10-30T14:32:57 Bostown: which starter kit are you trying to use? 2011-10-30T14:33:03 C# 2011-10-30T14:33:05 ... 2011-10-30T14:33:22 so basically "the code you're given is inherently wrong and broken" 2011-10-30T14:33:37 well 2011-10-30T14:33:47 I try to make something... 2011-10-30T14:33:57 but When i want to test it doesnt work 2011-10-30T14:33:59 uhhh my weighted heuristic gives greater open node sets :| 2011-10-30T14:34:05 ((start.row - goal.row).abs + (start.col - goal.col).abs) * 1.001 2011-10-30T14:34:09 Islacrusez: it's a start to show you how it works, it's not meant to be used as it is 2011-10-30T14:34:11 55 open node set to 118 2011-10-30T14:34:13 i start play_one_game.cmd 2011-10-30T14:34:43 put it says python is not recognized .... 2011-10-30T14:34:55 roflmao: sometimes heuristics get it wrong. 2011-10-30T14:34:56 I added it to environment variables 2011-10-30T14:35:23 Minthos: but it doens't show how it works, since it doesn't actually work 2011-10-30T14:35:35 Islacrusez: not at all? 2011-10-30T14:35:46 Islacrusez: when does it time out? 2011-10-30T14:36:47 anyone here who writes the code in c? 2011-10-30T14:37:01 merk: I do, but not for this. 2011-10-30T14:37:12 me too 2011-10-30T14:37:18 not for this 2011-10-30T14:37:25 and i want someone to explaing me a little 2011-10-30T14:37:29 WINWINWINWINWINWIN 2011-10-30T14:37:30 the bot in c 2011-10-30T14:37:38 roflmao: "start.row - goal.row" is broken over the edges 2011-10-30T14:37:49 one timed out at 75 ants, one at 16 *shrugs* 2011-10-30T14:38:06 if I knew what was causing it I not be asking >.> 2011-10-30T14:38:22 *** Murasha has quit IRC (Ping timeout: 265 seconds) 2011-10-30T14:38:25 yeah true lavalamp 2011-10-30T14:38:31 Islacrusez: running locally? 2011-10-30T14:38:48 on the server 2011-10-30T14:38:59 merk: I'm trying to understand the code in c 2011-10-30T14:39:19 well I'm good now :) 2011-10-30T14:39:24 test it locally, add debug output that shows how much time remains each round 2011-10-30T14:39:25 nodes in open is way smaller 2011-10-30T14:39:25 Islacrusez: the example bots are only to show you what you need. We have even discussed only giving them one game and then blocking them since they load the server needlessly 2011-10-30T14:39:35 they are currently slow for big maps 2011-10-30T14:39:36 merk: maybe we can share ideas 2011-10-30T14:39:42 but are not hard to make fast. 2011-10-30T14:39:47 tut bot > sample bot 2011-10-30T14:39:52 http://www.mavetju.org/programming/songbook.php 2011-10-30T14:40:09 tut bot are == the same 2011-10-30T14:40:36 not something you should expect to use on the server, at least not past testing submission 2011-10-30T14:41:18 in fact the main page even says don't share bots till after the comp 2011-10-30T14:41:28 *** AD1960 has left #aichallenge 2011-10-30T14:41:35 since then if anything is half decent that is what everyone will submit 2011-10-30T14:41:39 Can someone help me setting up tools 2011-10-30T14:41:48 When I do play_one_game.exe 2011-10-30T14:41:55 C:\Users\****\Desktop\tools>play_one_game.cmd 'python' is not recognized as an internal or external command, operable program or batch file. 2011-10-30T14:42:12 Bostown: you need to install python 2011-10-30T14:42:12 http://ants.fluxid.pl/replay.13309 omfg, i'm so proud how my bot deals with column of enemy ant! 2011-10-30T14:42:13 Bostown: do you have python installed? 2011-10-30T14:42:17 O jabe 2011-10-30T14:42:19 I Have 2011-10-30T14:42:22 2.7 python in particular 2011-10-30T14:42:27 Yes 2011-10-30T14:42:29 in the path? 2011-10-30T14:42:30 Maybe reinstalling python will help. 2011-10-30T14:42:37 can an ant travel diagonally? 2011-10-30T14:42:38 Environment variable 2011-10-30T14:42:40 yes 2011-10-30T14:42:50 ie can you type python at the cmd line? 2011-10-30T14:42:52 Fluxid: That is really cool 2011-10-30T14:42:57 knucklehead is knucklehead 2011-10-30T14:43:05 Bostown: yesto me or to UncleVasya ? 2011-10-30T14:43:07 i cant type python 2011-10-30T14:43:09 in the cmd 2011-10-30T14:43:11 what i have to do? 2011-10-30T14:43:17 overburn: no 2011-10-30T14:43:23 ah ok thanks 2011-10-30T14:43:25 to deltor 2011-10-30T14:43:28 ants can only go e,w n or s 2011-10-30T14:43:38 You said you added it to your environment vars. So you have something like c:\python27 in your system path? 2011-10-30T14:43:56 The column is definitely not the right attack formation. 2011-10-30T14:44:05 I cant type python in cmd 2011-10-30T14:44:10 *** bchoii has quit IRC (Ping timeout: 265 seconds) 2011-10-30T14:44:14 delt0r 2011-10-30T14:44:14 so the path is a problem 2011-10-30T14:44:18 Fluxid, yeah, that was like lemmings :) 2011-10-30T14:44:19 merk 2011-10-30T14:44:22 napoleon all over again, eh bmh? xD 2011-10-30T14:44:26 how i install python in path/ 2011-10-30T14:44:28 ? 2011-10-30T14:44:32 Fluxid, later you timed out though... 2011-10-30T14:44:35 I'm almost all done with this AI 2011-10-30T14:44:36 on windows i have no idea 2011-10-30T14:44:38 Islacrusez: crossing the T, dammit! 2011-10-30T14:44:39 and it works pretty well :) 2011-10-30T14:44:40 been years 2011-10-30T14:44:45 on linux 2011-10-30T14:44:47 its easy 2011-10-30T14:44:55 well i think so 2011-10-30T14:45:16 Migi32: yeah, i'm on wifi ;P 2011-10-30T14:45:32 ah, like that :) 2011-10-30T14:45:53 merk: those last few lines where to you 2011-10-30T14:46:08 thanks 2011-10-30T14:46:20 Bostown: don't know how to fix a windows path --need a windows person to chime in 2011-10-30T14:46:27 merk: are you on linux 2011-10-30T14:46:31 does the server only give 300ms? 2011-10-30T14:46:36 for bot response time 2011-10-30T14:46:38 500ms 2011-10-30T14:46:44 delt0r__: Thanx 2011-10-30T14:46:57 Migi32: last time i checked my bot was able to run 200 ants in around 150ms 2011-10-30T14:46:59 hehe Fluxid knuckle is my experimental :) it doesn't have combat tactics yet :) that's why they all are dying :) 2011-10-30T14:47:05 :) 2011-10-30T14:47:13 but this was a nice sequence 2011-10-30T14:47:18 mine crashes at about 20 something ants lol 2011-10-30T14:47:19 *** npskirk has joined #aichallenge 2011-10-30T14:47:19 yea lol 2011-10-30T14:47:25 fluxid, what pathfinding do you use? 2011-10-30T14:47:46 *** Nescio has joined #aichallenge 2011-10-30T14:47:48 roflmao: dunno what name it has, someone called it "heightmap" 2011-10-30T14:48:00 is it faster than a*? 2011-10-30T14:48:01 but it is not A* or anything special 2011-10-30T14:48:06 bah, pathfinding. I should just make a dictionary storing every path I've ever computed 2011-10-30T14:48:09 delt0r__: alas, that doesn't help those not already proficient in programming, nor does it help me in enhancing my code >.> 2011-10-30T14:48:15 roflmao: it is fast enough 2011-10-30T14:48:22 what language do you program in? 2011-10-30T14:48:27 python 2011-10-30T14:48:30 Islacrusez: well we can't write it for you... 2011-10-30T14:48:31 @bmh: wouldn't that cause your bot to crash at turn 0? :P 2011-10-30T14:48:31 Question about languages: does the absence of a starter kit imply that the game challenge server is not prepared to run clients in that language? 2011-10-30T14:48:31 roflmao: Run as fast as you can and don't look back. 2011-10-30T14:48:33 python 3 to be exact 2011-10-30T14:48:39 i just use BFS for food and hills and it is very fast.. don't understand all that trouble 2011-10-30T14:48:41 roflmao: The idea is probably less stupid than it sounds 2011-10-30T14:48:44 roflmao: don't use @ at the beginning 2011-10-30T14:48:59 fluxid: isn't python performance similar to ruby? 2011-10-30T14:49:22 Islacrusez: start with something simple and start adding complexity slowly. Strip stuff out of the tut to see what makes it slow/fast 2011-10-30T14:49:38 you know we all have to start somewhere.... 2011-10-30T14:49:40 I mean, look -- you can only travel in four directions 2011-10-30T14:49:43 I have yet to see a timeout locally 2011-10-30T14:49:52 that's insane 2011-10-30T14:49:59 and as for whoever told me that it times out because it takes too much time, you're an asshole 2011-10-30T14:50:06 Islacrusez: a big thing is map size 2011-10-30T14:50:10 so to get from any given square to any other square, you can partition the entire world into at most four pieces 2011-10-30T14:50:14 Islacrusez: and ant count 2011-10-30T14:50:22 i guess I really need to optimize this bitch 2011-10-30T14:50:31 so if you pathfind from (x -> y), store the path and the partition 2011-10-30T14:50:34 roflmao: comparable. and depends on implementation, because there is cpython (reference, default, most popular), jython, ironpython (.net) and pypy (self-hosting rocket-science interpreter with JIT, 5x faster than cpython) 2011-10-30T14:51:09 fluxid: just wondering cause I get horrible performance on mine :), gradually improving it but makes me want to cry when i hear about people managing 200+ ants with a* 2011-10-30T14:51:10 always wondered why python and perl are not on JITs yet gerneally 2011-10-30T14:51:26 roflmao: have you profiled? 2011-10-30T14:51:42 its never where you expect it to be... the performance bottle neck that is 2011-10-30T14:51:48 delt0r: that's a very good point 2011-10-30T14:52:00 delt0r: however I'm fairly certain it's with my a* 2011-10-30T14:52:09 famous last words 2011-10-30T14:52:11 cause it crashes when the open node set is too long 2011-10-30T14:52:12 =) 2011-10-30T14:52:15 roflmao: i had some problems with performance but i was able to optimize it after profiling and then it started to run 3x faster 2011-10-30T14:52:36 will do that then :) 2011-10-30T14:52:55 I'll show all of you! Static pathfinding shall win the day 2011-10-30T14:52:58 a* with the right heuristic and memory between ants and turns should not be that slow 2011-10-30T14:53:01 bmh: if you do it 2011-10-30T14:53:11 and it works 2011-10-30T14:53:21 then I'm totally doing it since thyat's a huge performance boost to have pre-computed paths 2011-10-30T14:53:24 roflmao: finding a simple enough partition is probably the hard part 2011-10-30T14:53:26 even if it means sacrificing 30 turns or so 2011-10-30T14:54:25 fluxid: what heuristic do you use? 2011-10-30T14:54:31 i'm using manhattan, maybe that's the problem 2011-10-30T14:54:34 due to the maze nature 2011-10-30T14:54:49 what do you mean? 2011-10-30T14:54:58 ah, right, you're running a different pathfinding algorithm 2011-10-30T14:55:23 heuristic = formula for estimating cost to goal 2011-10-30T14:55:24 i run only few pathfinds a turn no matter how many ants i have, and cache the results 2011-10-30T14:56:03 *** Islacrusez has left #aichallenge 2011-10-30T14:56:04 Here, this will blow you mind: Dijkstra's algorithm is A* with a zero-cost heuristic. 2011-10-30T14:56:10 and many ants may use the same "path map" 2011-10-30T14:56:12 is there anyone here from the challenge adminstration? 2011-10-30T14:56:20 zero-cost heuristic? 2011-10-30T14:56:22 <_flag> http://aichallenge.org/visualizer.php?game=35425&user=31 :) 2011-10-30T14:56:26 wut? 2011-10-30T14:56:44 roflmao: h(x) = 0 2011-10-30T14:56:44 Fluxid, how have you handled partial path caching and lookup? 2011-10-30T14:56:55 YaY Got it all working :) 2011-10-30T14:57:03 I'm out. 2011-10-30T14:57:04 *** bmh has quit IRC (Quit: bmh) 2011-10-30T14:57:37 heh bmh 2011-10-30T14:57:40 oh you're gon 2011-10-30T14:57:52 npskirk: McLeopold is and he is away but lurking 2011-10-30T14:58:14 fluxid: ah so you limit the path finding 2011-10-30T14:58:22 Bostown: great to here 2011-10-30T14:58:22 ergo in one turn you can't have 20 pathfinding computations 2011-10-30T14:58:27 _flag: i like how you move ants, and how they are able to find path to enemy hive around enemy ants 2011-10-30T14:58:49 Fluxid: kill all ants 2011-10-30T14:59:01 *** Nismorack has joined #aichallenge 2011-10-30T14:59:09 problem solved... going around things that move won't work with A* or whatever either 2011-10-30T14:59:10 ;) 2011-10-30T14:59:30 <_flag> Fluxid: Thanks, I like that game because all or almost all the bots in it are in the top ~20 2011-10-30T14:59:47 in that game no ant appears to be using a* 2011-10-30T14:59:49 *** exezive has joined #aichallenge 2011-10-30T15:00:00 <_flag> Basically I try to spread out as much as possible and collect food and steal hills until everyone else kills each other 2011-10-30T15:00:26 _flag: you avoid combat? 2011-10-30T15:00:43 _flag: soon i'll be in top 20 ;) 2011-10-30T15:00:45 i wish ;) 2011-10-30T15:00:46 <_flag> I avoid combat unless I need to defend or until I know I can win 2011-10-30T15:01:14 <_flag> I'm sure you can, most of the top 20 are leftovers from the beta who still have hiveless strategy implemented 2011-10-30T15:01:17 You mean you avoid attacking hills until the above, but you engage if you can win in small skirmishes? _flag 2011-10-30T15:01:24 anyway, i explained my pathfinding once, i will do it again only with a screenshot: http://fluxid.pl/misc/2011-10-30-195611_1782x1112_scrot.png 2011-10-30T15:01:26 <_flag> So they're good at pathfinding and combat, but will sacrifice their hives 2011-10-30T15:01:41 *** Chirmaya has joined #aichallenge 2011-10-30T15:02:00 this is, what i call "direction map", or heightmap to edge of unexplored part of map 2011-10-30T15:02:06 the lower the number, the close it is 2011-10-30T15:02:18 Ugh, I feel so dumb. I started my timer after a function of 900ms ran >.< But woohoo, it works now! Thanks to that person for the help! 2011-10-30T15:02:23 the closer it is to unseen part of map* 2011-10-30T15:02:43 I think it was Puj, so thanks again! 2011-10-30T15:02:50 Sure man. 2011-10-30T15:03:02 <_flag> Interesting, do you ever try to just hold an area or are you always trying to explore unseen area? 2011-10-30T15:03:04 i dunno what it really is, this the first thing which came to my mind when i wondered how to find paths for many ants at once 2011-10-30T15:03:06 yea, flood fill type algos can do that fast Fluxid --even left right scan can to 2011-10-30T15:03:27 delt0r__: yup, and i use it, fo everything almost 2011-10-30T15:03:28 lol 2011-10-30T15:03:34 fuck A* 2011-10-30T15:04:05 _flag: oh, im on 20 place! 2011-10-30T15:04:05 I was going to a version that includes all ants too... with objective scores 2011-10-30T15:04:10 just few minutes go a game! 2011-10-30T15:04:12 <_flag> Fluxid: nice :) 2011-10-30T15:04:13 let's see 2011-10-30T15:04:30 then each ant just moves to the next lowest/highest squre 2011-10-30T15:04:50 but then i have too much work to get done so it just won't happen 2011-10-30T15:05:49 fuck a* 2011-10-30T15:05:51 it's not fast enough 2011-10-30T15:05:59 http://aichallenge.org/visualizer.php?game=35437&user=757 my bot was SO STUPID when attacking greentea :( 2011-10-30T15:06:37 <_flag> How likely do you guys think it is two bots timedout on the same turn: http://aichallenge.org/visualizer.php?game=35441&user=31 ? 2011-10-30T15:07:00 *** bchoii has joined #aichallenge 2011-10-30T15:07:21 _flag: holy shit, this is scary 2011-10-30T15:07:36 *** danielharan has quit IRC (Ping timeout: 240 seconds) 2011-10-30T15:07:37 630 ants and 414 waiting to spawn 2011-10-30T15:07:51 <_flag> Fluxid: Also, I'm right behind you on the leaderboard :) 2011-10-30T15:07:56 _flag: maybe server choked or something 2011-10-30T15:07:59 _flag: cool! 2011-10-30T15:08:06 next game will be ours! 2011-10-30T15:08:42 _flag: oh wait! 2011-10-30T15:08:49 *** Bostown has quit IRC (Ping timeout: 265 seconds) 2011-10-30T15:08:54 just got another game, two minutes later 2011-10-30T15:08:57 tiwh you ;) 2011-10-30T15:09:14 <_flag> Yeah that was the game I linked where I timedout at the same time as someone else 2011-10-30T15:09:17 and this is the same game you linked above, lol 2011-10-30T15:09:36 what the shit timeout with 4 ants 2011-10-30T15:09:37 <_flag> Silly map IMO 2011-10-30T15:09:50 <_flag> Or at least it needs a lower food spawn rate 2011-10-30T15:10:03 hehe Fluxid that was a knucklehead scenario there :) 2011-10-30T15:10:25 ikaros: yes :( 2011-10-30T15:10:45 but my bot is really stupid in case of close combat 2011-10-30T15:10:56 i have really small game tree 2011-10-30T15:10:59 oh well, many ants can be a challenge as well 2011-10-30T15:11:19 and ants do not communicate well when attacking 2011-10-30T15:11:22 and I have a timeout in my bot :/ 2011-10-30T15:11:44 http://aichallenge.org/visualizer.php?game=33780&user=280 2011-10-30T15:12:07 *** analyst74 has joined #aichallenge 2011-10-30T15:12:24 oh, and one more game 2011-10-30T15:12:26 if you look good, you also see bugs in my ai 2011-10-30T15:13:33 http://aichallenge.org/visualizer.php?game=35451&user=757 fascinating 2011-10-30T15:13:59 paratrechina hive literally explodes at turn 947 2011-10-30T15:14:06 Is there one of those servers open for doing many fast matches? 2011-10-30T15:14:23 :-) 2011-10-30T15:14:30 @tcp 2011-10-30T15:14:31 Fluxid: tcp could be http://ants.fluxid.pl/howto. 2011-10-30T15:14:34 it chould work 2011-10-30T15:14:52 Maybe it is just my current location, but that one says that it is refusing my requests 2011-10-30T15:14:54 yeah, it didn't crash :) 2011-10-30T15:15:06 _flag: I don't go to fight just fot the fight too. Only if enemy ant stands between my ants and their destination (and it's winning battle or, if my ants num > 60, draw battle). 2011-10-30T15:15:17 anyway, i'm wondering what made paratrechina's ants do such sudden move outwards 2011-10-30T15:15:20 "socket.error: [Errno 10061] No connection could be made because the target machine actively refused it" 2011-10-30T15:15:22 fighting range is euclidian distance right? 2011-10-30T15:15:46 Chirmaya: what username? 2011-10-30T15:16:05 Anyone using Scala or Groovy? 2011-10-30T15:16:25 *** Nismorack has left #aichallenge 2011-10-30T15:16:45 I'm trying a new username. 2011-10-30T15:17:23 *** bchoii_ has joined #aichallenge 2011-10-30T15:18:40 Ah, it seems that quotation marks around the name messed it up 2011-10-30T15:18:46 Kay, nevermind =3 2011-10-30T15:19:00 :) 2011-10-30T15:20:55 *** Gr1m has joined #aichallenge 2011-10-30T15:21:04 hello 2011-10-30T15:21:09 WHAT THE FUCK 2011-10-30T15:21:15 wtf? 2011-10-30T15:21:22 after some thought 2011-10-30T15:21:28 http://aichallenge.org/visualizer.php?game=35441&user=757 this game ending is boring 2011-10-30T15:21:36 bots implement the same strategy 2011-10-30T15:21:46 im having trouble with collision detection' 2011-10-30T15:21:47 and event GarySWest timeout'd 2011-10-30T15:21:55 ok either I'm terrible or a* is a piece of shit for this game cause of the 300ms limit 2011-10-30T15:21:57 xathis is just too scared to make a move 2011-10-30T15:22:15 and warline doesn't move at all 2011-10-30T15:22:16 boooooring 2011-10-30T15:22:41 <_flag> That's maze maps for ya 2011-10-30T15:23:00 so, _flag, would youa gree that a* is not appropriate 2011-10-30T15:23:02 because it is too slow? 2011-10-30T15:23:10 at least a non-precomputed a* 2011-10-30T15:23:13 <_flag> roflmao: I don't use it 2011-10-30T15:23:18 what do you use? :) 2011-10-30T15:23:33 <_flag> I'm not really sure what I'd call it. I kind of bfs I guess 2011-10-30T15:23:43 which is faster I assume 2011-10-30T15:23:45 guys does anyone here using windows? 2011-10-30T15:23:52 > 2011-10-30T15:23:59 on the python tut, the collision detection code doesn't work 2011-10-30T15:24:06 im using windows 2011-10-30T15:24:13 <_flag> Depends, you have to code it in such a way that it's faster. I'm pretty sure a few top bots use A*, sir_macelon being one of them 2011-10-30T15:24:19 how you run the test_bot.cmd 2011-10-30T15:24:20 *** dvladim has joined #aichallenge 2011-10-30T15:24:21 ? 2011-10-30T15:24:30 because my a* is just too fucking slow :( 2011-10-30T15:24:44 I would like to use it since I've already got it implemented but holy shit if it's taking 300ms with 4 ants 2011-10-30T15:25:12 <_flag> What language are you using? 2011-10-30T15:25:14 there's probably something wrong with it. even in ruby it shouldnt take that long i guess 2011-10-30T15:25:19 ruby, which might be the problem 2011-10-30T15:25:23 Are you computing the entire search network for A* each turn? 2011-10-30T15:25:30 anybody want to help me? 2011-10-30T15:25:33 yea for each ant 2011-10-30T15:25:38 i find the path 2011-10-30T15:25:41 <_flag> I don't think anyone will be able to make anything competitve with anything less than java in terms of speed 2011-10-30T15:25:41 if I see a food 2011-10-30T15:25:53 otherwise i use a simple algorithm 2011-10-30T15:26:00 I might switch to java. 2011-10-30T15:26:03 hm _flag garyswest has python doesnt he 2011-10-30T15:26:03 right but creating your datastructure that A* searches over... are you doing that more than once? 2011-10-30T15:26:23 not sure what you mean exactly by creating my datastructure 2011-10-30T15:26:31 do you mean if I store the paths i find or not? 2011-10-30T15:26:45 odd how java sucks irl but its good for this game 2011-10-30T15:27:10 <_flag> ikaros: He does, but I don't think it will be enough in the long run 2011-10-30T15:27:15 Gr1m: irl? 2011-10-30T15:27:21 hm i dont know 2011-10-30T15:27:33 *** smf68 has quit IRC (Read error: Connection reset by peer) 2011-10-30T15:27:36 http://pastie.org/2784189 2011-10-30T15:27:42 that's it.. 2011-10-30T15:27:43 No, I mean A* searches a graph of nodes and arcs... do you build that graph fresh every turn? 2011-10-30T15:27:46 not within the game 2011-10-30T15:28:01 anyways anybody want to help me? 2011-10-30T15:28:01 i think implementing good strategies is worth more than alot of minmax or whatever algos 2011-10-30T15:28:07 I remember in tron, Python bots slowly disappeared from leaderboard as the game goes 2011-10-30T15:28:08 because you cant go deep anyways 2011-10-30T15:28:18 Gr1m: don't ask to ask. Just ask 2011-10-30T15:28:20 I'm fairly certain I do because the ruby framework refreshes the entire graph of nodes every turn 2011-10-30T15:28:27 i did ask 2011-10-30T15:28:35 nobody listened 2011-10-30T15:28:46 I only see you asking to ask 2011-10-30T15:28:48 ok... you might look for a way to maintain the graph and just update it with information 2011-10-30T15:28:49 *** azertt has joined #aichallenge 2011-10-30T15:29:12 on the python tut, the collision detection code doesn't work 2011-10-30T15:29:24 if you are creating new objects to populate the graph each time, you are doing a bunch of object instantiation that can be expensive 2011-10-30T15:29:26 do you have an old version 2011-10-30T15:29:38 its python 2.7 2011-10-30T15:29:53 the graph being the actual map 2011-10-30T15:29:58 Gr1m: ? since some of the starter bots don't use it, but the python code as stuff for it IIRC 2011-10-30T15:30:13 Gr1m: no i mean the tools+starter bots? 2011-10-30T15:30:16 If you can profile your code, look and see if the search itself is the bottleneck or if the setup of what you search over is the bottleneck 2011-10-30T15:30:30 no, all up to date 2011-10-30T15:30:37 i downloaded today 2011-10-30T15:30:40 fairly certain it is the search 2011-10-30T15:30:47 because my non a* ai works just fine 2011-10-30T15:30:53 does anyonde have a example of collision detection in C ? 2011-10-30T15:30:57 now, the setup might add just enough overhead to push the a* over the limit, that's true 2011-10-30T15:31:02 *** danielharan has joined #aichallenge 2011-10-30T15:31:16 but I'm having - massive - performance problems, one a* search will crash it if the open_set gets to be too big 2011-10-30T15:31:45 has anyone successfully run a bot with Scala or Groovy? 2011-10-30T15:31:54 too big = 200 or so nodes 2011-10-30T15:31:54 Gr1m: some of the starter packs have different quality but IIRC the python one does have a "isWater" method somewhere 2011-10-30T15:31:59 which actually isn't that much if I'm right 2011-10-30T15:32:39 ya, it does but it collides with other ants 2011-10-30T15:32:45 not with water 2011-10-30T15:32:59 oh well yea you have to that bit on your own 2011-10-30T15:33:19 since at least for enemy ants its not possible to not collide 2011-10-30T15:33:26 i tryed to use the tut's code but it doesn't work 2011-10-30T15:33:44 http://www.pastebucket.com/966 2011-10-30T15:33:52 well you are supose to write your own. ITs not all on a silver plate... 2011-10-30T15:33:52 I assume the problem is ruby then.. 2011-10-30T15:34:02 I guess I'll redo this in java, what the hell 2011-10-30T15:34:05 thanks for the help everyone 2011-10-30T15:34:12 im still learning how to do this 2011-10-30T15:34:17 im new to ai 2011-10-30T15:34:21 and the game 2011-10-30T15:34:29 *** wombot_ has joined #aichallenge 2011-10-30T15:34:36 I don't know rofmao, I doubt large scale A* would be performant 2011-10-30T15:34:40 but I could be wrong 2011-10-30T15:34:50 well it isn't large scale that's the thing 2011-10-30T15:34:58 I'm talking about 4 ants scale 2011-10-30T15:35:09 one single a* search in a turn, no more 2011-10-30T15:35:16 searching for long distance? 2011-10-30T15:35:18 with a max estiamted heuristic of 15 2011-10-30T15:35:20 manhattan 2011-10-30T15:35:40 if estimated manhattan is over 15 the path is not calculated 2011-10-30T15:35:43 http://www.pastebucket.com/966 code doesn't work, http://aichallenge.org/ants_tutorial_step_1.php 2011-10-30T15:35:46 and even with that it STILL crashes 2011-10-30T15:35:50 help 2011-10-30T15:36:05 hmm 2011-10-30T15:36:43 *** Nescio_ has joined #aichallenge 2011-10-30T15:36:44 I haven't tried to do pathfinding that long, but that does sound like a lot of searches, how big does your open_set get? 2011-10-30T15:36:52 Grlm: what is the crash message? 2011-10-30T15:36:57 crashes at around 200 nodes on the open_set 2011-10-30T15:37:13 no crash message, the code doesn't do anything 2011-10-30T15:37:15 that doesn't sound right, 200 is nothing 2011-10-30T15:37:17 so distance of 15 = long distance? i thought that was short 2011-10-30T15:37:20 yeah exactly 2011-10-30T15:37:26 200 should be a pushover 2011-10-30T15:37:41 maybe it's with the way I'm storing the nodes or something. i do store the entire square object in the array, but could it really be that> 2011-10-30T15:37:52 tut said is was collison detection but it doesn't do htat 2011-10-30T15:37:58 *** ccc has joined #aichallenge 2011-10-30T15:38:18 *** Nescio has quit IRC (Ping timeout: 265 seconds) 2011-10-30T15:38:26 hi, am I guaranteed to see my ants before the food at each turn ? 2011-10-30T15:38:37 *** amstan has joined #aichallenge 2011-10-30T15:38:37 *** ChanServ sets mode: +o amstan 2011-10-30T15:39:01 *** Gr1m has quit IRC () 2011-10-30T15:39:14 I can't profile the code easily, too 2011-10-30T15:39:30 I guess I want to know what is the order in which I get input? is it totally random or not random 2011-10-30T15:39:33 ya, Java probably has better toolset 2011-10-30T15:39:43 ccc: the specs don't make any statements about the order of the input you get so it could change at any point 2011-10-30T15:39:47 and, as you said, 200 should be a pushover 2011-10-30T15:40:02 I just love ruby so much as a language, so pretty :P 2011-10-30T15:40:02 *** goffrie has quit IRC (Remote host closed the connection) 2011-10-30T15:40:12 Grlm: so your ant stays in one place? 2011-10-30T15:40:19 thanks jix, 2011-10-30T15:40:27 I use python for the same reason ;) 2011-10-30T15:40:41 yep 2011-10-30T15:40:49 but isn't py and ruby performance similar? 2011-10-30T15:40:52 that's the thing. :-\ 2011-10-30T15:41:06 ya, if my bot somehow got competitive enough, I might consider migrating 2011-10-30T15:41:10 on ants.fluxid.pl, why does A have much more games and yet also much larger sigma than the bots rated higher than him? 2011-10-30T15:41:13 so far it hasn't. :p 2011-10-30T15:41:19 doesn't sigma go down with number of games played? 2011-10-30T15:41:20 so why the fuck am I having such bad performance aaaaaaargh 2011-10-30T15:41:31 *** test has joined #aichallenge 2011-10-30T15:41:50 like I said, I don't do pathfinding longer than 300 node in bfs 2011-10-30T15:42:05 yea well I'm at 200 :( 2011-10-30T15:42:40 does running the live visualizer slow things down? 2011-10-30T15:42:45 I'm not familiar, is multi-agent A* suppose to be faster than bfs? 2011-10-30T15:43:01 I think so because it doesn't have as many nodes in the open set 2011-10-30T15:43:11 it's both faster and gives a more optimal path 2011-10-30T15:43:23 it's really just bfs with a heuristic for choosing nodes in the open set 2011-10-30T15:46:32 ok running the live visualizer certainly slows things down 2011-10-30T15:47:31 need a better machine. :p 2011-10-30T15:47:35 *** paulwal has joined #aichallenge 2011-10-30T15:47:59 *** ZeroBrain has quit IRC (Ping timeout: 248 seconds) 2011-10-30T15:48:15 no but I mean that without it 2011-10-30T15:48:17 hte bot doesn't time out 2011-10-30T15:48:26 as fast :P 2011-10-30T15:48:32 lol 2011-10-30T15:48:42 how expensive is your heuristic? 2011-10-30T15:48:50 it's manhattan 2011-10-30T15:48:51 very simple 2011-10-30T15:49:08 how many manhattan do you calculate per node? 2011-10-30T15:49:14 one 2011-10-30T15:49:20 Bot at: 9:68 finding food 3:3 2011-10-30T15:49:21 turn 173 bot 0 timed out 2011-10-30T15:50:08 there is a problem if your ant has to go that far to gather food. 2011-10-30T15:51:38 *** npskirk has left #aichallenge 2011-10-30T15:51:39 what's manhattan? 2011-10-30T15:51:51 it's a very simple heuristic 2011-10-30T15:52:09 overburn: when you go from point A to point B in a grid without going diagonally 2011-10-30T15:52:24 overburn: manhattan distance is the distance it takes to do that 2011-10-30T15:52:45 I'm going to exit out of my pathfinding if the open_set length gets to be bigger than 100 2011-10-30T15:52:59 ooh useful 2011-10-30T15:54:11 *** merk has quit IRC () 2011-10-30T15:54:18 are you familiar with BFS overburn? 2011-10-30T15:54:33 breadth first search 2011-10-30T15:54:38 yeah 2011-10-30T15:54:46 well a* is basically a modified version of that 2011-10-30T15:54:56 where you sort the nodes in the open_set by their cost 2011-10-30T15:55:03 you check lower cost nodes first 2011-10-30T15:55:04 yeah ,i know :) 2011-10-30T15:55:14 and you can use manhattan as the heuristic for the cost 2011-10-30T15:55:16 simple and effective 2011-10-30T15:55:27 oh , that makes sense 2011-10-30T15:55:29 thanks mate 2011-10-30T15:55:30 : D 2011-10-30T15:55:41 (start.row - goal.row).abs + (start.col - goal.col).abs 2011-10-30T15:55:44 that's it 2011-10-30T15:55:59 that solves me a shitload of issues 2011-10-30T15:56:02 thanks : D 2011-10-30T15:56:11 *** wilsone has quit IRC (Ping timeout: 265 seconds) 2011-10-30T15:56:31 you'll want to add in a tiebreaker in 2011-10-30T15:56:34 as well as g_score 2011-10-30T15:58:01 indeed 2011-10-30T15:58:05 mate, you just made my day 2011-10-30T15:58:07 heh 2011-10-30T15:58:36 welcome 2011-10-30T15:58:50 now if you have a way of doing that without having your code time out then please share :P 2011-10-30T15:59:05 wiki has a great example of a* is psuedocode 2011-10-30T15:59:07 *** wombot_ has quit IRC (Remote host closed the connection) 2011-10-30T15:59:10 aichallenge: janzert epsilon * r9867618 / website/profile_submissions.php : Add missing ) - http://git.io/QsXI1Q 2011-10-30T15:59:12 this is what I use and it's failing kind of bad cause of perf: http://pastie.org/2784189 2011-10-30T15:59:32 *** xar0l has joined #aichallenge 2011-10-30T15:59:36 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-10-30T15:59:47 I'm thinking of maybe just posting my code in the forums and hoping someone will be kind enough to help 2011-10-30T15:59:58 roflmao: you could try using bigger scoped nodes 2011-10-30T16:00:05 what do you mean? 2011-10-30T16:00:33 uhh wait, i expressed myself wrong 2011-10-30T16:01:08 i menat that you could basically divide the path in several segments and process them as you go after getting a big picture 2011-10-30T16:01:37 you mean break it up through multiple turns? 2011-10-30T16:01:42 yeah 2011-10-30T16:01:44 that's what i meant 2011-10-30T16:01:48 hey that's a good idea 2011-10-30T16:02:04 heh yeah , i know 2011-10-30T16:02:23 i've got timers in most places , and then it takes too long , it pauses and continues next turn 2011-10-30T16:02:36 too bad i haven't had the time to make the algorithms useful tho :( 2011-10-30T16:03:21 *** Belerafon has quit IRC (Remote host closed the connection) 2011-10-30T16:03:55 *** invader has quit IRC (Ping timeout: 265 seconds) 2011-10-30T16:03:59 a* should work well for you 2011-10-30T16:04:02 what lang are you using? 2011-10-30T16:04:20 *** Accoun has quit IRC () 2011-10-30T16:04:26 c++ 2011-10-30T16:05:05 a* is a bit tricky when you need to compute whole path, and things usually changes during moving on that path 2011-10-30T16:05:32 i guess you could compute the whole path taking only water into account 2011-10-30T16:05:32 and a* for only first move is very ... ummm.. heavy 2011-10-30T16:05:44 xathis has insanely high mu 2011-10-30T16:05:51 and then check the vicinity every turn and compensate for various obstacles on a waaay lower set 2011-10-30T16:06:01 well, I've found that 2011-10-30T16:06:13 *** dvladim has quit IRC (Ping timeout: 245 seconds) 2011-10-30T16:06:14 also, what's mu? 2011-10-30T16:06:26 it's very common for ants to pick on unexpected food on the way to the target food 2011-10-30T16:06:34 so i almost never modify the path based on the appearence of new food 2011-10-30T16:06:41 of course not food 2011-10-30T16:06:43 lol 2011-10-30T16:06:44 overburn: "true skill" 2011-10-30T16:06:45 i meant other ants 2011-10-30T16:06:53 oh yes 2011-10-30T16:06:57 i haven't gotten there yet :P 2011-10-30T16:06:58 sigma is uncertainty of true skill 2011-10-30T16:07:07 visible skill = mu-3*sigma 2011-10-30T16:07:10 first I need to get the fucking thing working well then I can worry about ant collision 2011-10-30T16:07:15 heh i haven't gotten to implementing almost anything yet 2011-10-30T16:07:23 however i have a great timeshifting system haha ;)) 2011-10-30T16:08:03 and ants that move randomly + useless calculations to test it 2011-10-30T16:08:24 Fluxid: Xathis has almost the same mu as A atm. But A has very high sigma, why is that? Shouldn't sigma go down with number of games played? 2011-10-30T16:08:27 anyone making their ai take into account, not only what target is best, but what target is likely to be best after that? 2011-10-30T16:09:03 nope but thanks for the idea :D 2011-10-30T16:10:43 hey Fluxid, just curious, is python performance a problem for you? 2011-10-30T16:11:19 anyone know a great uml tool 2011-10-30T16:11:29 Migi32: i'm talinkg about official server 2011-10-30T16:11:35 ah ok 2011-10-30T16:11:37 analyst74: not really 2011-10-30T16:11:37 i keep forgetting key parts of my implementation and need to log them somehow 2011-10-30T16:12:00 good to know. :) 2011-10-30T16:12:24 analyst74: do you want to know exact stats? 2011-10-30T16:12:38 exact stats on? 2011-10-30T16:12:44 but yes, always! 2011-10-30T16:12:48 how much time per turn is used 2011-10-30T16:12:57 ok, lemme see 2011-10-30T16:13:01 *** Fandekasp has quit IRC (Ping timeout: 240 seconds) 2011-10-30T16:13:10 *** diskonnect has quit IRC (Quit: KVIrc 4.0.2 Insomnia http://www.kvirc.net/) 2011-10-30T16:13:12 *** bchoii_ has quit IRC (Quit: Page closed) 2011-10-30T16:14:37 *** merk has joined #aichallenge 2011-10-30T16:15:50 fluxid: how many hours have you spent on your algorithm? 2011-10-30T16:17:08 roflmao: i've got first working version uploaded within first 24 hours 2011-10-30T16:17:20 nice 2011-10-30T16:17:32 it got to 30 place 2011-10-30T16:17:34 i've been owrking on mine for a week and it's still total utter shit. time to move to java and do the tut first :P 2011-10-30T16:17:38 now it would be 70something 2011-10-30T16:17:42 and within next week fixes and refinements 2011-10-30T16:19:13 *** Fandekasp has joined #aichallenge 2011-10-30T16:19:47 anyone here coding the bot in c? 2011-10-30T16:20:26 merk: why? 2011-10-30T16:22:20 *** puto|mokka has left #aichallenge ("Leaving") 2011-10-30T16:22:21 if you want to use C, you could just as well use C++ and write C-style code but still use some C++ functionality wherever it's useful. 2011-10-30T16:22:26 that's my opinion anyway 2011-10-30T16:23:10 analyst74: on my compy with q6600 managing near 600 ants for the time gets over 500ms 2011-10-30T16:23:26 for the first time* 2011-10-30T16:23:46 600 ants? shouldn't your opponents be dead by then? 2011-10-30T16:23:47 Migi32: like? 2011-10-30T16:23:53 nice 2011-10-30T16:23:59 Migi32: what kind of c++ stuff? 2011-10-30T16:24:03 Migi32: yeah ;D but i'm stress testing now 2011-10-30T16:24:16 790 ants - over 1s 2011-10-30T16:24:22 amstan: std::map and stuff. That stuff is quite a bit less user-friendly in C. 2011-10-30T16:24:26 but that was a spike 2011-10-30T16:24:27 *** Fandekasp has quit IRC (Ping timeout: 252 seconds) 2011-10-30T16:26:05 culd anyonw tell me how i culd send arguments to my bot? i end up withh an error telling me that the game cant stat my bot :/ 2011-10-30T16:26:12 amstan: I just don't get why you'd limit yourself to a almost-subset of another language. But you know, that's just my opinion, don't take it too seriously ;) 2011-10-30T16:26:29 good to know that decent combat logic can be achieved without too much computation! 2011-10-30T16:26:45 Migi32: there's some C things that the C++ compiler doesn't like 2011-10-30T16:27:01 like? 2011-10-30T16:27:02 like not using extern 2011-10-30T16:27:46 ah, ok then. I don't ever use extern, but fair point nonetheless :P 2011-10-30T16:27:56 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-30T16:27:57 this fails: python tools/playgame.py "MyBot 1" "python tools/sample_bots/python/GreedyBot.py" 2011-10-30T16:28:34 *** paulwal has quit IRC (Ping timeout: 265 seconds) 2011-10-30T16:28:36 *** danielharan has joined #aichallenge 2011-10-30T16:28:39 "echo 1 | MyBot" ? 2011-10-30T16:28:48 Puj: you don't want to do that 2011-10-30T16:28:54 No I don't 2011-10-30T16:29:01 isbric: linux? 2011-10-30T16:29:05 Puj: thats not passing it an argument. 2011-10-30T16:29:08 isbric: write a small shell script and use that to pass the arguments? 2011-10-30T16:29:10 amstan: nix* 2011-10-30T16:29:34 does "MyBot" work just like that? shouldn't it be "./MyBot"? 2011-10-30T16:29:51 Migi32: i have a startup script, that takes arguments that need to be passed to my bot. 2011-10-30T16:30:25 isbric: I wanted to do that as well - fought with it a bit and did the shell script too 2011-10-30T16:30:30 amstan: in my case "../MyBot" didn't work. I had to create a simlink in the /tools directory and use "MyBot" directly 2011-10-30T16:30:38 amstan: python doesnt care about the env variable when executing the bots 2011-10-30T16:30:45 guys when i am trying to compile and link my files 2011-10-30T16:30:47 with make 2011-10-30T16:30:49 Migi32: one dot, 2 means the parent folder 2011-10-30T16:30:51 a get an error 2011-10-30T16:30:59 not 'ld' directory 2011-10-30T16:31:10 amstan, in my case tools is a subdirectory of where my bot's executable is 2011-10-30T16:31:28 *** ^5 <^5!3aaaaf23@gateway/web/freenode/ip.58.170.175.35> has quit IRC (Ping timeout: 265 seconds) 2011-10-30T16:31:47 isbric: does ./MyBot 1 work in the terminal? 2011-10-30T16:32:08 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-30T16:32:17 *** kmp has joined #aichallenge 2011-10-30T16:32:37 *** danielharan has joined #aichallenge 2011-10-30T16:32:47 amstan: yes 2011-10-30T16:32:52 weird.. 2011-10-30T16:33:04 if i pass the greedybot an argument it works aswell 2011-10-30T16:33:09 isbric: what arguments do you want to pass anyway? When you upload your code, the server won't pass these arguments 2011-10-30T16:33:30 *** Fandekasp has joined #aichallenge 2011-10-30T16:33:58 are there any (other than the mybot) C samples of bots out tere? 2011-10-30T16:34:05 Migi32: i know, i need to pass it a filename to open, the filename will allways be uniq, hance the need of an argument 2011-10-30T16:34:25 I added a debug flag so I would not have to comment out print statements etc. 2011-10-30T16:34:48 Migi32: and its ontly for ofline training of my bot. 2011-10-30T16:34:59 isbric, a simple hack might be to append a timestamp to the filename? 2011-10-30T16:35:12 inside your code 2011-10-30T16:35:39 *** azertt has quit IRC (Quit: Page closed) 2011-10-30T16:35:53 then it's always unique 2011-10-30T16:36:02 guys 2011-10-30T16:36:13 merk 2011-10-30T16:36:21 migi32 2011-10-30T16:36:21 Migi32: that wount work, im making a simple IPC with files, cant be bothered to try to use the pythons semaphores. 2011-10-30T16:36:24 i get an error 2011-10-30T16:36:32 when i am trying to compiling with make 2011-10-30T16:36:41 *** dfsd has joined #aichallenge 2011-10-30T16:36:47 does anyone see any point in using trained neural nets in this challenge ? 2011-10-30T16:36:51 gcc: installation problem, cannot exec `ld': No such file or directory 2011-10-30T16:36:53 isbric, why don't you set a local environmental variable in your script and read it in your bot? 2011-10-30T16:36:59 what it is the problem? 2011-10-30T16:37:02 overburn, yes I do :) 2011-10-30T16:37:14 hello, im on the python tut and i seem to be having problems with the collision detection part 2011-10-30T16:37:14 overburn: i do, just having a load of stuggle training them. :P 2011-10-30T16:37:27 isbric: how are you training them ? 2011-10-30T16:37:32 migi32 what do you think? 2011-10-30T16:37:51 Migi32: umm care to share? :D would really want to use that, but no ideas currently 2011-10-30T16:37:55 overburn: genetic algorithums works best imo 2011-10-30T16:37:58 merk, you don't have ld installed 2011-10-30T16:38:05 what is this? 2011-10-30T16:38:16 merk, the linker part of gcc 2011-10-30T16:38:21 can anybody help me 2011-10-30T16:38:25 analyst74: in average 1.3ms for every ant per turn 2011-10-30T16:38:33 i use it in windows 2011-10-30T16:38:38 isbric: are you trying to use neural nets overall, or only for specific problems? 2011-10-30T16:38:39 1057 ants took 1370ms 2011-10-30T16:38:41 *** skunx has joined #aichallenge 2011-10-30T16:39:15 overburn: lets just say its nn heavy, i dont want to get in to much details hehe 2011-10-30T16:39:22 overburn: when an ant isn't going after food or an enemy hill or stuff, but is just somewhere in the middle of nowhere, i was thinking of letting a neural net decide the direction he should go 2011-10-30T16:39:40 anybody want to help me with my problem? 2011-10-30T16:39:45 overburn: but i use offline training with ga, and online adoptation with rl 2011-10-30T16:40:12 isbric: rl? 2011-10-30T16:40:18 reinforced learning? 2011-10-30T16:40:24 merk: could it be that you've downloaded the linux starter bot instead of the windows one? 2011-10-30T16:40:30 overburn Puj yes 2011-10-30T16:41:05 hello? 2011-10-30T16:41:11 oh, have yet to stumble on this concept 2011-10-30T16:41:13 learning time :D 2011-10-30T16:41:20 isbric, did you try using a local environmental variable to store your filename and grab it at runtime? 2011-10-30T16:41:20 isbric, what are the outputs of the network? suggested direction(s)? 2011-10-30T16:41:27 dfsd - you can ask without asking to ask mate 2011-10-30T16:41:37 i already asked 2011-10-30T16:41:46 [15:37] hello, im on the python tut and i seem to be having problems with the collision detection part 2011-10-30T16:41:50 dfsd, nope, you said you were having trouble. 2011-10-30T16:41:55 dfsd, with what? 2011-10-30T16:41:55 Puj: no, i will look in to that. thx 2011-10-30T16:42:09 dfsd what kind of problems? 2011-10-30T16:42:26 the code that they provide doesn't work 2011-10-30T16:42:39 What's the problem with the code? dfsd? 2011-10-30T16:42:42 it has no errors but it doesn't do what the tut said it should do 2011-10-30T16:42:56 Migi32: 6 outputs, 5 moves, 1 comunication node 2011-10-30T16:43:16 dfsd: are you sure? it worked well for me 2 days ago 2011-10-30T16:43:37 and how does it not work? 2011-10-30T16:43:39 *** cutoff has joined #aichallenge 2011-10-30T16:43:40 what doesn't it do? 2011-10-30T16:43:50 you're not being specific enough mate 2011-10-30T16:44:01 isbric, communication node. That's interesting :) I wonder if they'll use it and what for 2011-10-30T16:44:35 http://fluxid.pl/misc/2011-10-30-213601_1920x1200_scrot.png i like how no ant is killed and no hive is blockd for long time 2011-10-30T16:44:35 ya, i put the code in, i saved it, and it does what it always did which is run togeather. tut is here http://aichallenge.org/ants_tutorial_step_1.php 2011-10-30T16:45:00 the ant run into each other is what i mean 2011-10-30T16:45:04 ants* 2011-10-30T16:45:10 Migi32: generaly swarms need to comunicate :P so my hopes is to use it and aswer it in attack situations. 2011-10-30T16:45:36 *** tmandry has quit IRC (Ping timeout: 260 seconds) 2011-10-30T16:45:38 nice 2011-10-30T16:45:57 * avdg wishes he would be back in his coding zone :-) 2011-10-30T16:46:10 *** swinejihad has joined #aichallenge 2011-10-30T16:46:26 isbric: will you assist it with other heuristics like pathfinding or just let the NN do everything? 2011-10-30T16:46:30 haha avdg, same here 2011-10-30T16:46:32 *** mleise has quit IRC (Ping timeout: 258 seconds) 2011-10-30T16:47:02 is there something im missing 2011-10-30T16:47:16 stressful situation today with someone who had a virus on his computer, its no fun if every step it takes to fix it are interrupted 2011-10-30T16:47:21 but anyway 2011-10-30T16:47:33 *** Chirmaya_ has joined #aichallenge 2011-10-30T16:47:35 Migi32: depends on my first analysis if i get this ga working on the game, but if its to training heavy ill have to resurt to some A* like alg to pin out the main logic. 2011-10-30T16:47:53 *** ancechu has joined #aichallenge 2011-10-30T16:47:59 *** kmp has quit IRC (Quit: Page closed) 2011-10-30T16:48:16 get a new harddrive. :) 2011-10-30T16:48:19 huzzah my a* almost works 2011-10-30T16:48:22 hello? 2011-10-30T16:48:22 for now its nn only. 2011-10-30T16:48:29 Migi32: how about you? 2011-10-30T16:48:49 Quick question. The text data that we are passed per turn, does any of it include information of whether or not we have visibility of the square? Or do we calculate that ourselves? 2011-10-30T16:49:05 Chirmaya_, you have to calculate it 2011-10-30T16:49:21 *** ancechu has quit IRC (Client Quit) 2011-10-30T16:49:22 most starter kits do have that logic implemented though 2011-10-30T16:49:26 hello? 2011-10-30T16:49:46 My ruby one didn't, and my custom method isn't efficient on big maps =( 2011-10-30T16:49:48 All information passed implies visibility of the square which contains the information, beyond that, no empty tiles, that is, non-land, non-water tiles are given in the turn information. 2011-10-30T16:49:53 isbric: very interesting. It'll be nice to see if this works out or not. At the moment NN stuff is way down at the bottom of my todo list. I have some ideas, but that's all :) 2011-10-30T16:50:02 dfsd, you need to implement that logic yourself 2011-10-30T16:50:06 my java pack doesn't have visibility logic 2011-10-30T16:50:06 Not many of the starter packages have optimal visibility functions. 2011-10-30T16:50:21 Oh, Puj, haha, thanks you again! 2011-10-30T16:50:38 analyst74: how? 2011-10-30T16:50:42 I am so blind for not thinking of that. I was checking visibility to each tile from each ant, hahaha 2011-10-30T16:50:56 Migi32: how come, you doubt the use or just want to try somthing else out first? 2011-10-30T16:50:58 isbric: currently I assign scores to each of my ant's possible moves and I select the best moves each time. The NN stuff would be part of that scoring function, but probably not much more. 2011-10-30T16:51:00 *** girandroid has joined #aichallenge 2011-10-30T16:51:25 NN = neural network? 2011-10-30T16:51:29 yes 2011-10-30T16:52:15 *** ltriant has joined #aichallenge 2011-10-30T16:52:22 fuck this, none of you guys are helpfull 2011-10-30T16:52:31 Has anyone implemented any back-propogating iterative improvement algorithms for pathfinding? 2011-10-30T16:52:43 what do you want 2011-10-30T16:52:48 dfsd, there are many ways to achieve that 2011-10-30T16:53:22 the one fucking way im looking to do is is the fucking tuts way so i can fucking continue with the fucking tut 2011-10-30T16:53:36 Whoa, settle down 2011-10-30T16:53:43 *** merk has quit IRC () 2011-10-30T16:53:59 ... 2011-10-30T16:54:03 dfsd, peace and lov man. It's cool, we just need the juicy details of your problem yo 2011-10-30T16:54:46 Most of the people in here that want to help wait until something pops up like "Oh shit, I had this problem .... , now this is happening ... . I've tried this.. . and this ..." Now what? 2011-10-30T16:54:59 And everyone's like "geezomg, metoo, dothis." 2011-10-30T16:55:04 And then everyone is happy. 2011-10-30T16:55:08 *** Accoun has joined #aichallenge 2011-10-30T16:55:16 But like "omg wtf?" people don't really respond to. 2011-10-30T16:55:23 the tut 2011-10-30T16:55:26 what algorithm does it use? 2011-10-30T16:55:31 for food hunting 2011-10-30T16:55:34 bfs? 2011-10-30T16:55:46 http://aichallenge.org/ants_tutorial_step_1.php I TRIED THIS CODE AND IT DOES NOTHING 2011-10-30T16:56:00 I don't really think it had a real algorithm 2011-10-30T16:56:08 In what language.... dfsd 2011-10-30T16:56:20 *** exezive has quit IRC (Remote host closed the connection) 2011-10-30T16:56:25 python 2011-10-30T16:56:45 Okay. Have you tried using printout statments to verify what is going on dfsd? 2011-10-30T16:57:10 have you made other changes to the program? 2011-10-30T16:57:11 I assume by "doing nothing" you mean, it exhibits the same behavior as before you put in the new code? 2011-10-30T16:57:23 yes 2011-10-30T16:57:36 Okay, and you are absolutely sure you modified the correct MyBot.py? 2011-10-30T16:57:45 *** goffrie has joined #aichallenge 2011-10-30T16:57:45 yes 2011-10-30T16:58:33 Okay. So, have you tried printing out ants.unoccupied and see if it corresponds to what you think should happen? 2011-10-30T16:59:01 how do you print? 2011-10-30T16:59:22 Ah! Herin lies the key to answering most "omg wtf just happened" questions. 2011-10-30T16:59:32 sys.stderr.write("This is a test") 2011-10-30T16:59:39 how much experience do you have with programming? 2011-10-30T16:59:41 Now, make sure you import sys! 2011-10-30T17:00:13 a little c/c++ and python 2011-10-30T17:00:14 * isbric thinks this is a prank of some sort? 2011-10-30T17:00:24 * Puj likes being trolled. 2011-10-30T17:00:45 just write to an error stream or a file 2011-10-30T17:01:52 *** Bostown has joined #aichallenge 2011-10-30T17:01:55 dfsd, I would recommend to you exactly what swinejihad has. Print a lot and often. Make sure you know what the code is doing. If in doubt, code it yourself by looking at the code you want to write and run the program after you type each line, just to understand exactly. 2011-10-30T17:01:59 Hello :) 2011-10-30T17:02:05 Sup gfunk 2011-10-30T17:02:08 turn 1 bot 0 timed out 2011-10-30T17:02:08 hi 2011-10-30T17:02:22 Okay!! That's what's happening then dfsd. 2011-10-30T17:02:29 Your bot is timing out! 2011-10-30T17:02:42 That's why it is not doing the shiznittle omg bamfuzz. 2011-10-30T17:02:46 One question :) Can I test with the tool with just 1 bot ? 2011-10-30T17:02:51 It's probably because of an error. 2011-10-30T17:03:01 actually that just happend after i put the sys.stderr.write in 2011-10-30T17:03:06 Which, in python means, it's probably because of your copy paste. 2011-10-30T17:03:06 *** danielharan has quit IRC (Ping timeout: 252 seconds) 2011-10-30T17:03:12 Well, then you have not imported sys 2011-10-30T17:03:24 you might be able to find a blank bot that just doesn't do anything... 2011-10-30T17:03:26 Bostown: With few instances of the same bot or only with one player in a game? 2011-10-30T17:03:26 Bostown, multiples of the same bot yes. 2011-10-30T17:03:38 *** jkant76 has joined #aichallenge 2011-10-30T17:03:54 UncleVasya: one player in the game 2011-10-30T17:04:21 *** girandroid has quit IRC (Quit: Page closed) 2011-10-30T17:04:37 I think you may need speshul taktix for that Bostown 2011-10-30T17:04:47 I've never tried it actually, I think if you actually did manage to kinda wedge that in you would just end up with an instant-win 2011-10-30T17:04:53 the sys.stderr.write function is not working 2011-10-30T17:05:02 dfsd, have you imported sys? 2011-10-30T17:05:05 "sys" 2011-10-30T17:05:09 The smallest count of player is 2 for map but Puj's splution should be ok if we understood correctly. 2011-10-30T17:05:31 *** ikaros_ has joined #aichallenge 2011-10-30T17:05:34 kk Ill try that :) 2011-10-30T17:05:46 Thanx a bunch :) 2011-10-30T17:05:55 there 2011-10-30T17:05:55 http://pastebin.com/0EFJXA1N 2011-10-30T17:06:08 Bostown, you can mess around with the playgame.py to make the game not end when there is only one surviving player, if you just want to test food collection for instance, but I would keep a backup copy on hand. 2011-10-30T17:06:28 *** danielharan has joined #aichallenge 2011-10-30T17:07:05 *** racko has quit IRC (Quit: Page closed) 2011-10-30T17:07:07 Yes Ill try the do_nothing bots 2011-10-30T17:07:11 *** skunx has quit IRC (Ping timeout: 260 seconds) 2011-10-30T17:07:43 Bostown: Also it can be usefult trick to make a version of your bot that performs as usual but after some amount of turns (let's say it 300 or 600) it stops moving ants and just finishing every turn. 2011-10-30T17:07:48 tut bot aint too shabby 2011-10-30T17:08:08 dfsd, that works for me. 2011-10-30T17:08:18 it doesn't for me 2011-10-30T17:08:35 maybe you have an old version of python? 2011-10-30T17:08:43 Ye I am focussing on moving and collecting food :) 2011-10-30T17:08:44 python 2.7.2 and on windows and linux 2011-10-30T17:08:58 *** skunx has joined #aichallenge 2011-10-30T17:09:06 actually i tried 2.6.5 and 2.7.2 on linux 2011-10-30T17:09:38 Maybe you have a replay opened in your browser and a new game of your bot cannot create a new replay. 2011-10-30T17:09:41 what would be the best data structure to keep the map data in (and best order key/keys) so that i can assign orders to each ant in only one map pass? 2011-10-30T17:09:45 *** ikaros_ has quit IRC (Client Quit) 2011-10-30T17:09:49 umm actually, is there any way to do that? 2011-10-30T17:09:58 So you watch the same (old) replay again and again. 2011-10-30T17:10:06 *** danielha_ has joined #aichallenge 2011-10-30T17:10:18 * Puj claps for UncleVasya 2011-10-30T17:10:22 Agree! 2011-10-30T17:10:32 the txt in console doesn't change either 2011-10-30T17:10:34 *** ikaros_ has joined #aichallenge 2011-10-30T17:10:35 *** danielharan has quit IRC (Read error: Connection reset by peer) 2011-10-30T17:10:37 so it isn't taht 2011-10-30T17:10:59 'is there a way to use the eclipse debugger with this thing? 2011-10-30T17:11:05 since it is run through python and not just java 2011-10-30T17:11:18 nja....a.. 2011-10-30T17:11:24 You could.... roflmao 2011-10-30T17:11:28 overburn: Assuming sets are iterable with your language those are good 2011-10-30T17:11:43 You could rewrite the playgame.py to run your bot in code and not in a separate process. 2011-10-30T17:11:51 *** jkant76 has quit IRC (Quit: Page closed) 2011-10-30T17:11:55 That requires quite the rework, but it would be nice. 2011-10-30T17:12:00 I might in fact do that. 2011-10-30T17:12:30 and im not using replay 2011-10-30T17:12:36 if you do that 2011-10-30T17:12:48 you will send me the code puj or i will hunt you down and kill you with a shotgun :P 2011-10-30T17:12:53 lol 2011-10-30T17:13:08 cause eclipse debugger RULES 2011-10-30T17:13:17 swinejihad: i expressed myself in a crappy way 2011-10-30T17:13:20 what i mean 2011-10-30T17:13:33 I think it should be fairly simple actually.. maybe just wrapping the Bot class should do it. 2011-10-30T17:13:39 say i have the ants vector , and a food vector 2011-10-30T17:13:56 which would be the most effective way to get the closest food item for an ant ? 2011-10-30T17:14:03 *** asdasda has quit IRC (Quit: Page closed) 2011-10-30T17:14:06 overburn 2011-10-30T17:14:11 It depends 2011-10-30T17:14:12 is there any other way besides sys.stderr.write to print out information 2011-10-30T17:14:13 what language are you using again? 2011-10-30T17:14:16 c++ 2011-10-30T17:14:18 ah right 2011-10-30T17:14:19 *** skunx has quit IRC (Ping timeout: 258 seconds) 2011-10-30T17:14:27 dont' know about c++ but simplest way is to 2011-10-30T17:14:28 dfsd, logging 2011-10-30T17:14:35 loop through all possible ant-food combinations 2011-10-30T17:14:37 overburn: FYI I'm not a C++ user but I'm familiar with vectors 2011-10-30T17:14:51 add them to a list along with their cost (could be manhattan) 2011-10-30T17:14:54 and then sort the list 2011-10-30T17:14:59 logging how? 2011-10-30T17:15:03 sort by the cost of course 2011-10-30T17:15:08 dfsd, write/append to a file. 2011-10-30T17:15:19 // Find all possible combinations of ant-to-food routes and add them 2011-10-30T17:15:20 // to foodRoutes 2011-10-30T17:15:20 for(Tile foodLoc : sortedFood) { 2011-10-30T17:15:20 for(Tile antLoc : sortedAnts) { 2011-10-30T17:15:20 int distance = ants.getDistance(antLoc, foodLoc); 2011-10-30T17:15:20 Route route = new Route(antLoc, foodLoc, distance); 2011-10-30T17:15:21 foodRoutes.add(route); 2011-10-30T17:15:21 } 2011-10-30T17:15:21 overburn: You could do what I do and calculate the distance to every piece of food for the target ant 2011-10-30T17:15:22 } 2011-10-30T17:15:22 // sort foodRoutes by distance cost to choose closest ants to food 2011-10-30T17:15:23 Collections.sort(foodRoutes); 2011-10-30T17:15:25 roflmao: that's just wasteful hah 2011-10-30T17:15:34 meh dude it's not that much cpu power :P 2011-10-30T17:15:43 ovverburn: so that's a O(n) operation for one ant 2011-10-30T17:15:51 overburn: up^ 2011-10-30T17:15:58 yeah , not that much 2011-10-30T17:16:11 trying to find a solution that uses less 2011-10-30T17:16:12 *** skunx has joined #aichallenge 2011-10-30T17:16:13 dfsd, http://docs.python.org/library/logging.html 2011-10-30T17:16:17 well, think of it this way 2011-10-30T17:16:24 overburn: or you could do a breadth first search from the ant looking for food 2011-10-30T17:16:26 by spending cpu power on the sorting part 2011-10-30T17:16:28 that sort is O(m*n), no? ants * food_hills 2011-10-30T17:16:28 cyphase, to the rescue! 2011-10-30T17:16:31 you save cpu power on the pathfinding 2011-10-30T17:16:35 *** Bostown_ has joined #aichallenge 2011-10-30T17:16:37 by ensuring that you'll always use optimal paths 2011-10-30T17:16:49 * cyphase puts on his tights 2011-10-30T17:16:52 lol 2011-10-30T17:17:09 roflmao: oh, yeah , i'm totally cool with sorting 2011-10-30T17:17:32 *** simlay has joined #aichallenge 2011-10-30T17:17:37 Guys, wouldn't the getPath() function return a path of a different cost than the distance() function? I mean don't you have to find the path to know the cost? Or are you just guessing? And i fyou are guessing why not just use A*? 2011-10-30T17:17:41 *** Qua has joined #aichallenge 2011-10-30T17:17:43 I just wrote a generic find closest function for one tile and a set of tiles 2011-10-30T17:17:51 but realy, doing manhattan for every ant with a grid of 17*17 (ants sight) is not realy that bad.. 2011-10-30T17:17:59 How do you detect new friendly ants spawned? 2011-10-30T17:18:16 but on the server the grid can be even 200x200 and i can have a lot of ants 2011-10-30T17:18:25 overburn, burn! 2011-10-30T17:18:43 it amounts to wasted time iterating through every food item for every ant 2011-10-30T17:18:43 hmm 2011-10-30T17:18:59 *** tmandry has joined #aichallenge 2011-10-30T17:18:59 *** tmandry has joined #aichallenge 2011-10-30T17:19:01 *** paulwal has joined #aichallenge 2011-10-30T17:19:03 I use an implemented of a KD-tree to find items in range and nearest neighbour searches. 2011-10-30T17:19:18 Thats plenty fast even with 20k elements in the list 2011-10-30T17:19:24 wait, put the log code on MyBot.py or another file 2011-10-30T17:19:31 Qua, how do you account for pathfinding? 2011-10-30T17:19:31 ? 2011-10-30T17:19:35 hey, can we get logs from the servers at all in any way? im getting timeouts there, but not locally or on tcp games 2011-10-30T17:19:46 dfsd, don't overwrite your MyBot.py 2011-10-30T17:19:49 I don't 2011-10-30T17:19:58 Qua, okay. 2011-10-30T17:20:15 its logging to log.txt 2011-10-30T17:20:23 dfsd, good. 2011-10-30T17:20:31 but bot crashed after i put the logging code on MyBot.py 2011-10-30T17:20:46 *** Bostown has quit IRC (Ping timeout: 265 seconds) 2011-10-30T17:20:48 You are not allowed to write to the file system 2011-10-30T17:20:49 I believe 2011-10-30T17:21:01 He's talking locally Qua 2011-10-30T17:21:02 Only read 2011-10-30T17:21:07 Ahh, sorry 2011-10-30T17:21:11 np 2011-10-30T17:21:31 dfsd, do you have all your tabs correctly indented and such and this 2011-10-30T17:21:51 ya 2011-10-30T17:22:04 dfsd, do you get any error in the here or where? 2011-10-30T17:22:53 it timed out 2011-10-30T17:23:03 *** UncleVasya has quit IRC (Ping timeout: 244 seconds) 2011-10-30T17:23:10 and log.txt wasn't writen into 2011-10-30T17:23:13 What is that dependent on, dfsd, around the corner and which? 2011-10-30T17:23:42 Any takers on how you can detect newly spawned ants? 2011-10-30T17:23:44 dfsd, have you a stacktrace where the who and how? 2011-10-30T17:23:55 hwo? 2011-10-30T17:24:04 Qua, you get a message, like "a 0" 2011-10-30T17:24:25 dfsd, A stacktrace, receive thee a stacktrace, around the who abouts? 2011-10-30T17:24:31 *** Fandekasp has quit IRC (Ping timeout: 248 seconds) 2011-10-30T17:24:55 how? 2011-10-30T17:25:09 Yeah, but I get that for all my ants. Also those I had last turn. I need to identify just ants that just spawned at the hill Puj 2011-10-30T17:25:24 dfsd, Ah "where' you mean my friend of the what and why. dfsd, try the stderr output file in the fly and by. 2011-10-30T17:25:46 aichallenge: janzert epsilon * rd8419ca / website/profile_submissions.php : Move header to top of profile_submissions so it doesn't overwrite the user_id variable - http://git.io/GCKUOQ 2011-10-30T17:25:57 Qua, that is a course many have taken. You may have to account for the orders you issued last turn? 2011-10-30T17:26:32 Damn, hoped there would be an easier way 2011-10-30T17:26:33 *** skunx has quit IRC (Quit: leaving) 2011-10-30T17:26:40 Puj, http://pastebin.com/0EFJXA1N does this code work for you or does the ants just collide 2011-10-30T17:26:58 and does it print right 2011-10-30T17:28:14 dfsd, you are not using Eclipse, and you are not using the -E flag 2011-10-30T17:28:26 Your code spits out 2011-10-30T17:28:32 Traceback (most recent call last): File "../MyBot.py", line 62, in Ants.run(MyBot()) NameError: name 'Ants' is not defined turn 0 bot 0 crashed 2011-10-30T17:29:08 Or you are checking the wrong log file. 2011-10-30T17:29:26 Something of the what and sniff is up. 2011-10-30T17:30:39 python tools/playgame.py "python MyBot.py" "python tools/sample_bots/python/HunterBot.py" --map_file tools/maps/example/tutorial1.map --log_dir game_logs --turns 60 --scenario --food none --player_seed 7 --verbose -e 2011-10-30T17:30:43 thats the sh code 2011-10-30T17:30:44 Snabblebun! 2011-10-30T17:31:07 and the code should work because its the tut's code 2011-10-30T17:32:06 eh not necessarily 2011-10-30T17:32:20 the php code doesn't get the input properly for example 2011-10-30T17:32:54 hello 2011-10-30T17:33:23 *** Chirmaya has quit IRC (Quit: Page closed) 2011-10-30T17:33:25 most of it is contributed by volunteers and they may not be fully compatible with all the versions of the language they were created for 2011-10-30T17:33:55 wouldn't that mean the php code is faulty? 2011-10-30T17:34:17 i get to talk about accoun in my management assignment 2011-10-30T17:34:39 not necessarily 2011-10-30T17:34:51 it's just a starter bot 2011-10-30T17:34:57 ants roaming randomly 2011-10-30T17:35:02 but 2011-10-30T17:35:13 still for me it didn't get the input properly 2011-10-30T17:35:22 @Puj. After 30-some hours of work, my AI is finally working on all maps without messing up thanks to your help! Danke danke! 2011-10-30T17:35:23 Chirmaya_: User error -- Replace user. 2011-10-30T17:35:28 meaning the coordinates ant player ownership 2011-10-30T17:35:33 the rest was ok 2011-10-30T17:35:56 how is it getting the input improperly? 2011-10-30T17:36:04 *** plasto has joined #aichallenge 2011-10-30T17:36:06 not necessarily, other people have said that the code works 2011-10-30T17:36:24 idk why it doesn't for me 2011-10-30T17:37:00 well, for example instead of say "a 26 18 0" it only gets "a " 2011-10-30T17:37:14 may be dependent on php version tho 2011-10-30T17:37:21 the like to change it a lot 2011-10-30T17:37:44 *they 2011-10-30T17:37:46 useless, im using java 2011-10-30T17:37:49 python* 2011-10-30T17:38:02 irrelivent 2011-10-30T17:39:35 overburn, that sounds like a bug in the php code. 2011-10-30T17:40:02 ya 2011-10-30T17:40:09 anyone have an example of avoid collisions in C ? 2011-10-30T17:40:18 playgame.py should be writing to stdin the same for every language 2011-10-30T17:40:20 and python 2011-10-30T17:41:51 paulwal: probably, tho that's the standard starter output 2011-10-30T17:41:57 *output for the input 2011-10-30T17:42:20 did you save? 2011-10-30T17:42:34 erm, never mind 2011-10-30T17:42:36 * paulwal downloading php starter bot 2011-10-30T17:43:18 i'm really curois if it was buggy writing to file on my part or bug in the starter bot 2011-10-30T17:43:23 do tell if you find something 2011-10-30T17:43:34 k 2011-10-30T17:43:40 how did you write to file? 2011-10-30T17:44:17 in python? 2011-10-30T17:44:25 I just use stderr 2011-10-30T17:44:34 fwrite($file_ptr, $value); 2011-10-30T17:44:39 sys.stderr.write("foo bar baz") 2011-10-30T17:44:43 *** Qua has quit IRC (Quit: Page closed) 2011-10-30T17:45:08 on sys.stderr.write, where does it write to? 2011-10-30T17:45:22 capa: can't you check for the state of the tile where you are trying to move your ant and if the status is another ant present there, don't move there? 2011-10-30T17:45:59 capa: to avoid collisions in C code i'm using hcreate() and hsearch() hash functions, look at their man pages, they are really simple 2011-10-30T17:46:12 anyone know you to write to stderr in php? 2011-10-30T17:46:16 know how* 2011-10-30T17:46:33 stderr is a different stream from stdout (that is the default) 2011-10-30T17:46:35 *** cutoff has quit IRC (Ping timeout: 260 seconds) 2011-10-30T17:47:07 so you sys.stderr.open? 2011-10-30T17:47:12 ccc it works on windows ? 2011-10-30T17:47:13 no 2011-10-30T17:47:23 paulwal: quick google search yelds "file_put_contents("php://filter/write=string.rot13/resource=example.txt","Hello World");" 2011-10-30T17:47:24 it's pretty much always open. 2011-10-30T17:47:28 don't close it. 2011-10-30T17:47:37 sorry 2011-10-30T17:47:38 where does it write to? 2011-10-30T17:47:39 wrong paste 2011-10-30T17:47:42 *** Palmik has quit IRC (Remote host closed the connection) 2011-10-30T17:47:44 capa: those are from gnu libc, don't have a clue about windows 2011-10-30T17:47:50 *** Bostown_ has quit IRC (Ping timeout: 265 seconds) 2011-10-30T17:48:02 ok yeah , no idea about the stderr 2011-10-30T17:48:08 if you don't redirect it, just the console 2011-10-30T17:48:17 dfsd: then on linux you could do somthing like this ./(play bot sctipt) 2> debug.txt 2011-10-30T17:48:17 overburn, found it 2011-10-30T17:48:18 yeah 2011-10-30T17:48:22 how is it? 2011-10-30T17:48:23 oh 2011-10-30T17:48:35 yeah, isbric's on it. 2011-10-30T17:48:43 janzert: are you planning to run the turn limit update? 2011-10-30T17:48:49 i tried but it didn't write to console 2011-10-30T17:48:51 ccc: i found search.h here :-) 2011-10-30T17:49:24 you can just redirect it to a different file like isbric said. 2011-10-30T17:49:27 McLeopold: not without changing it 2011-10-30T17:49:28 isbric: i have console open so i dont need to save it in file 2011-10-30T17:49:30 *** freenity has joined #aichallenge 2011-10-30T17:50:33 dfsd: ofc, but if you feel theres to much debug infor mixed with the stdout stream you know where to go 2011-10-30T17:50:41 -r 2011-10-30T17:51:34 *** plasto has quit IRC (Quit: Page closed) 2011-10-30T17:52:22 wtf? it works now 2011-10-30T17:52:40 overburn, try putting this in the 3rd line of the while loop in the function 'run': fprintf(STDERR, "%s\n", $current_line); 2011-10-30T17:53:25 all i did was take out -senerio 2011-10-30T17:53:29 that should print in your console the input exactly as it is received 2011-10-30T17:54:21 and sys.stderr.write still doesn't work 2011-10-30T17:55:06 *** pvarga has joined #aichallenge 2011-10-30T17:55:51 paulawal: heh no longer have it . c++ is the way for me here heh 2011-10-30T17:56:26 janzert: what needs to be changed? 2011-10-30T17:56:37 weird. locally: "turn 4 bot 0 timed out" -- but I can't even get something out of stderr at the top of AI.run 2011-10-30T17:56:51 it's like it times out before it gets to my code 2011-10-30T17:56:56 did you just call me paula? 2011-10-30T17:57:33 lulz 2011-10-30T17:57:35 my bad 2011-10-30T17:57:38 sorry typo 2011-10-30T17:58:32 daniel, add -E (or is it -e) parameter to playgame.py 2011-10-30T17:58:33 McLeopold: it was never updated for the expanded reporting of game end reason, also after watching the maps the current criteria is it using to lengthen or shorten the map limit isn't any good 2011-10-30T18:00:18 is there teams forming or is it just every man for him self? :P 2011-10-30T18:00:23 *** girandroid has joined #aichallenge 2011-10-30T18:00:40 * isbric might be looking for a teamate 2011-10-30T18:00:41 janzert: what needs to change? 2011-10-30T18:01:05 *** bearoff has joined #aichallenge 2011-10-30T18:01:19 * overburn might be looking for a teammate once he gets a rather functional bot up and running 2011-10-30T18:01:22 *** dfsd has quit IRC (Ping timeout: 265 seconds) 2011-10-30T18:02:07 no hopes tehn huh? ;) 2011-10-30T18:02:26 * paulwal didn't know there were teams 2011-10-30T18:02:46 McLeopold: besides the above? 2011-10-30T18:03:57 I think before I can take a guess at a good criteria I need to make a tool so I can view some histograms of game lengths per map and ending type 2011-10-30T18:04:29 is it legal for your bot to dynamically load a .so? For instance, Tcl loading a sqlite3.so for a memory-only sql database 2011-10-30T18:05:13 janzert: if you can use some type of js visualization, like d3, that would be nice to put on the site 2011-10-30T18:06:01 paulwal: thats realy intresting, im hoping for the readonly rule, but if it fails ill have to compile my db. 2011-10-30T18:06:03 crossed my mind although resulting load worried me a bit too 2011-10-30T18:06:34 if we produce json and can cache it, it shouldn't be too bad 2011-10-30T18:06:46 isbric, what language are you using? 2011-10-30T18:06:46 *** dfsd has joined #aichallenge 2011-10-30T18:06:58 paulwal: c++ 2011-10-30T18:07:01 *** tmandry has quit IRC (Ping timeout: 240 seconds) 2011-10-30T18:07:21 analyst74: it's already there :( 2011-10-30T18:07:27 paulwal: I would think in memory database structures make more sense, and would be faster 2011-10-30T18:07:35 *data structures* 2011-10-30T18:07:49 and, yes, you can try that 2011-10-30T18:08:13 does all the python tut work? 2011-10-30T18:08:28 dfsd: it should, I wrote the bot before I wrote the tutorial 2011-10-30T18:08:32 The test script spits out one frame of a text view, is there a way to use this instead of the java visualizer? 2011-10-30T18:08:46 well Step 1 is not working for me 2011-10-30T18:09:25 McLeopold: awesome. you're probably right, though sqlite can be quite fast and convenient. I figured i'd experiment 2011-10-30T18:10:00 the python collision detection code doesn't do anything when i add it to MyBot.py 2011-10-30T18:10:03 dfsd: pebkac 2011-10-30T18:10:19 dfsd: pastebin your code 2011-10-30T18:11:27 Is there a text-mode visualizer? 2011-10-30T18:11:30 http://pastebin.com/NTaCSDZg 2011-10-30T18:12:00 dfsd: you still have some default move stuff down below 2011-10-30T18:12:57 dfsd: get rid of the old default move code 2011-10-30T18:13:06 oh, the whole do_turn funtion is supposed to be replaced 2011-10-30T18:13:29 yeah, except you can keep the timing if statement if you want 2011-10-30T18:13:43 *** capa has quit IRC (Quit: Page closed) 2011-10-30T18:13:47 you are using ants.issue_order down below, which is skipping the collision check 2011-10-30T18:14:11 everything has to go though the new do_move_direction function, or the collision check doesn't work 2011-10-30T18:15:37 ah, it makes much more sense now, thnx 2011-10-30T18:15:44 *** nolo_contendere has quit IRC (Read error: Connection reset by peer) 2011-10-30T18:16:02 *** nolo_contendere has joined #aichallenge 2011-10-30T18:18:16 *** djr_ has joined #aichallenge 2011-10-30T18:18:34 * fingers crossed 2011-10-30T18:18:54 Fluxid: Does your server work with ruby files? My ants aren't moving on it =( 2011-10-30T18:20:00 *** djr_ has quit IRC (Read error: Connection reset by peer) 2011-10-30T18:20:27 Chirmaya_: it works in whatever language your bot is written in - because you run your bot locally 2011-10-30T18:20:50 and tcpclient.py just proxies its input/output to manager on the server 2011-10-30T18:21:33 McLeopold: when you get a chance could you comment on https://github.com/aichallenge/aichallenge/issues/344 2011-10-30T18:21:45 amstan: ^^ as well 2011-10-30T18:21:50 *** icefox has joined #aichallenge 2011-10-30T18:22:13 looks good 2011-10-30T18:22:35 That is strange. My ants are working on all the test maps locally, but they aren't moving at all in the games on the server. Do you know what I might be doing incorrectly? 2011-10-30T18:23:37 McLeopold, amstan: thanks, figured it'd be good to have something with a record for changing the rules 2011-10-30T18:24:02 d'oh. "turn 4 bot 0 timed out" means bot 0 timed out during turn 3... 2011-10-30T18:24:07 Chirmaya_: It's not saying timeout or crashed for your bot? 2011-10-30T18:24:18 *** Harpyon has quit IRC (Quit: Textual IRC Client: http://www.textualapp.com/) 2011-10-30T18:24:25 *** djr_ has joined #aichallenge 2011-10-30T18:24:32 analyst74: ^^ :) 2011-10-30T18:24:55 *** girandroid has quit IRC (Quit: Page closed) 2011-10-30T18:25:07 yes? 2011-10-30T18:25:30 It seems to timeout about 80 some turns in, but I don't get a single timeout here =/ 2011-10-30T18:25:45 *** test has quit IRC (Quit: Page closed) 2011-10-30T18:26:11 *** cameronz has quit IRC (Quit: Ex-Chat) 2011-10-30T18:26:23 *** cameronz has joined #aichallenge 2011-10-30T18:26:38 Some of the games: http://ants.fluxid.pl/player/Rarity 2011-10-30T18:26:42 chirmaya, possibly invalid orders also?(eg. movign same ant more than once) 2011-10-30T18:26:52 Not a single one of those 2011-10-30T18:26:59 *** avdg has quit IRC (Quit: Leaving.) 2011-10-30T18:27:01 Plus it passes all test cases on the offical server 2011-10-30T18:27:23 Chirmaya_: which script do you use? 2011-10-30T18:27:27 for python or python3? 2011-10-30T18:27:31 python3 2011-10-30T18:27:39 use one for python2 2011-10-30T18:27:49 and i'll remove it from server 2011-10-30T18:27:53 Akay, I'll give it a try ^^ 2011-10-30T18:27:54 *** icefox has quit IRC (Quit: icefox) 2011-10-30T18:28:00 because it doesn't fucking work 2011-10-30T18:29:46 *** avdg has joined #aichallenge 2011-10-30T18:29:50 *** tmandry has joined #aichallenge 2011-10-30T18:29:50 *** tmandry has joined #aichallenge 2011-10-30T18:30:20 *** wombot_ has joined #aichallenge 2011-10-30T18:30:22 *** swinejihad has quit IRC (Ping timeout: 265 seconds) 2011-10-30T18:30:41 *** overburn has quit IRC (Quit: Page closed) 2011-10-30T18:31:15 done 2011-10-30T18:31:35 McLeopold: wouldn't it make more sense for game_settings.php to live in the sidebar? probably directly under game specification. Before giving a link to it anywhere though we should probably add a mention at the top that 'turns' is over ridden by the map limit. 2011-10-30T18:31:49 *** choas has joined #aichallenge 2011-10-30T18:32:02 janzert: I agree 2011-10-30T18:32:06 *** bearoff has left #aichallenge 2011-10-30T18:32:54 ok, I'll do it 2011-10-30T18:33:10 ah, so it's not wrong, it's just unclear 2011-10-30T18:34:00 *** AxVapor has joined #aichallenge 2011-10-30T18:34:36 McLeopold: any objection to taking the rankings link out of the footer as well? doesn't seem to belong there either 2011-10-30T18:34:44 sure 2011-10-30T18:36:21 *** ikaros_ has quit IRC (Quit: Ex-Chat) 2011-10-30T18:37:32 *** danielharan has joined #aichallenge 2011-10-30T18:37:36 *** Sir_Ragnarok_ has joined #aichallenge 2011-10-30T18:37:51 hi 2011-10-30T18:39:31 *** danielha_ has quit IRC (Ping timeout: 244 seconds) 2011-10-30T18:42:18 *** Mo has joined #aichallenge 2011-10-30T18:42:19 *** bluegaspode has joined #aichallenge 2011-10-30T18:42:44 *** Mo is now known as Guest70422 2011-10-30T18:42:47 with your do_turn loops 2011-10-30T18:43:16 is it better to loop though each ant and decide an actions? 2011-10-30T18:43:33 aichallenge: janzert epsilon * r0240040 / (3 files): Add game settings link and extra explanatory text - http://git.io/tifoug 2011-10-30T18:43:54 *** avdg has quit IRC (Quit: Leaving.) 2011-10-30T18:44:29 umm.. is it me or is aichallenge.org down? 2011-10-30T18:44:50 is there a api list? 2011-10-30T18:45:12 Draakon: not just you 2011-10-30T18:45:24 its down for me too 2011-10-30T18:45:30 McLeopold: mysqld seems to be pegged on cpu usage 2011-10-30T18:45:32 *panic* 2011-10-30T18:45:36 *** avdg has joined #aichallenge 2011-10-30T18:48:34 tempted to back out my last change but it doesn't even touch the database 2011-10-30T18:49:28 hmm.. 2011-10-30T18:50:34 *** AxVapor has quit IRC (Quit: AxVapor) 2011-10-30T18:51:53 *** grom358 has joined #aichallenge 2011-10-30T18:51:58 janzert: we may have deadlocks 2011-10-30T18:51:59 *** Guest70422 is now known as asdsfafas 2011-10-30T18:52:05 morning 2011-10-30T18:52:05 *** Guest29668 has quit IRC (Quit: Page closed) 2011-10-30T18:52:16 ok, where? 2011-10-30T18:52:26 *** Puj has quit IRC (Quit: ~ Trillian Astra - www.trillian.im ~) 2011-10-30T18:52:27 *** xathis has joined #aichallenge 2011-10-30T18:52:48 I wonder... should I be looping through each ant one by one to decide actions... hmmm 2011-10-30T18:53:07 janzert: I'm just going cycle mysqld, sound ok? 2011-10-30T18:53:34 ok, wish we could figure out where it is stuck but... 2011-10-30T18:54:00 asdsfafas: that is what I do.. but it has its problems 2011-10-30T18:54:20 janzert: okay, I'll reasearch for about 10 minutes first 2011-10-30T18:54:33 grom358: like what kinds? 2011-10-30T18:54:57 *** choas has quit IRC (Ping timeout: 258 seconds) 2011-10-30T18:55:10 janzert: I need mysql admin access :( 2011-10-30T18:55:10 ahh, found 'show processlist' :) 2011-10-30T18:55:28 you've got root? :) 2011-10-30T18:55:40 *** kara has quit IRC () 2011-10-30T18:55:52 grom358: well I can see food gathering posing a problem... since distances b/w ants and food is calculated for each ant rather than just once (if you do it that way) 2011-10-30T18:55:56 *** lunz has joined #aichallenge 2011-10-30T18:56:44 asdsfafas: like moving ants together todo things like attack 2011-10-30T18:57:31 *** bmh has joined #aichallenge 2011-10-30T18:57:36 hi 2011-10-30T18:57:36 *** Raimondi has joined #aichallenge 2011-10-30T18:57:49 asdsfafas: so say the enemy has two ants in front of you.. you only doing one ant at a time.. so how do you know you can move. You only want to move into attack if another ant is also moving with you 2011-10-30T18:57:52 is there a way to enable print statements in the code to be seen when running in the terminal? 2011-10-30T18:57:53 McLeopold: I could take down apache for a bit to stop new requests piling on 2011-10-30T18:58:04 roflmao: I thought some more about static pathfinding. I think I can compute all possible paths efficiently. 2011-10-30T18:58:20 grom358: yeah that could be tough 2011-10-30T18:58:26 asdsfafas: no.. log to a file is the best option.. stdout is for bot commands 2011-10-30T18:58:36 grom358: thanks 2011-10-30T18:58:37 *** cutoff has joined #aichallenge 2011-10-30T18:59:03 ah i see, webpage is down? 2011-10-30T18:59:03 *** gcflymoto has joined #aichallenge 2011-10-30T18:59:13 yes 2011-10-30T18:59:16 ok 2011-10-30T18:59:35 wanted to whine about it but saw the discussion ;) 2011-10-30T18:59:41 :) 2011-10-30T18:59:44 asdsfafas: so my idea for workaround.. was to have ant packs that move together... but haven't figured out how to code moving them as group when dealing with obstacles and narrow areas etc 2011-10-30T19:00:16 and fit in with the path I calculate with A* 2011-10-30T19:00:50 *** frogman has joined #aichallenge 2011-10-30T19:01:22 janzert: how are the workers? 2011-10-30T19:01:45 *** yascha has joined #aichallenge 2011-10-30T19:02:08 grom358: let them deviate from the path! 2011-10-30T19:02:20 I'm approaching it as a flow problem. 2011-10-30T19:02:29 *** lunz has quit IRC (Quit: Page closed) 2011-10-30T19:03:13 janzert: okay, kicking mysqld 2011-10-30T19:03:18 bmh: that is easier said then done.. if I let it deviate.. how do I make sure it still follows it? 2011-10-30T19:03:28 like the A* is giving me back a list of tiles 2011-10-30T19:03:43 worker I'm on was waiting on the server to accept last result 2011-10-30T19:03:59 so say I move away from that list of tiles.. I have to path find back to it or something 2011-10-30T19:04:26 *** frogman has quit IRC (Client Quit) 2011-10-30T19:04:41 think about it this way -- imagine you have a graph that abstracts over the grid 2011-10-30T19:04:48 whats the best way to debug a bot? 2011-10-30T19:04:58 dfsd: snarky answer: with your brain 2011-10-30T19:05:20 you have edge that connect the nodes of the abstract graph and these have weights that correspond to the number of ants abreast you can push though 2011-10-30T19:05:50 pretty hard to see whats going on in the background with only my brain 2011-10-30T19:05:55 It looked like there was one matchup creation in progress and a few checks for submissions 2011-10-30T19:06:03 If you have five ants, find a flow on the abstract graph and kill off the edges that hit capacity 2011-10-30T19:06:30 I did notice the other day matchup creation seemed to be getting a bit long again 2011-10-30T19:06:43 at least it was 10 seconds sometimes 2011-10-30T19:07:47 *** mceier has quit IRC (Quit: leaving) 2011-10-30T19:07:47 15 seconds just now 2011-10-30T19:08:05 for 10 player game 2011-10-30T19:08:25 grom358: I used A* initially but I'm beginning to think herd pathfinding doesn't work well with it 2011-10-30T19:08:58 *** rharmsen has joined #aichallenge 2011-10-30T19:09:06 grom358: so I'm going to try a sort of... move in the direction of least resistance type of algorithm 2011-10-30T19:09:24 grom358: with a center of mass using A* 2011-10-30T19:09:31 *** paulwal has quit IRC (Ping timeout: 265 seconds) 2011-10-30T19:09:31 well. i was thinking of have ants that follow another ant.. so one ant is doing the pathfind 2011-10-30T19:10:25 and the rest just follow... but the leader ant will have to stop whenever it sees enemy ants ahead and let the ants get back into formation 2011-10-30T19:10:47 or just find the path and store it for any ant in the start area that wants to get to the destination area 2011-10-30T19:10:53 *** dana44 has joined #aichallenge 2011-10-30T19:11:06 grom358: yup, I'm thinking of the same thing except my "leader ant" is a ghost ant at the center of mass 2011-10-30T19:11:38 cypahse: how does that let them travel in a group though? that just lets you cache A* results 2011-10-30T19:11:55 asdsfafas: ghost ant, I like that :) 2011-10-30T19:11:59 grom358, you could still do what you said with that 2011-10-30T19:12:27 cyphase: interesting idea... pre-calculate the major paths (or cache them) 2011-10-30T19:12:31 *** Chirmaya_ has quit IRC (Quit: Page closed) 2011-10-30T19:12:48 yea, it' not really *pre*-calculating 2011-10-30T19:12:53 it's* 2011-10-30T19:13:23 basically as first step.. I want my ants to cluster up if there enemy ants.. atm my ants are suicidal 2011-10-30T19:13:39 lol 2011-10-30T19:14:03 "hey look, there's food behind enemy lines! let me go get it!" 2011-10-30T19:14:12 grom358: bring piece to the ants, in hope in may help (just kidding) 2011-10-30T19:14:43 brb... gonna go watch some real ant attack videos x.x do they travel in a straight-ish line until they reach their destination? 2011-10-30T19:14:57 learn from nature :P 2011-10-30T19:15:19 cyphase: oh no, we push them back :-) 2011-10-30T19:15:24 http://www.youtube.com/watch?v=HL3sHuK3iGE 2011-10-30T19:15:44 according to that they do what I'm wanting to implement =D 2011-10-30T19:16:02 we would need more food for that 2011-10-30T19:16:03 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-10-30T19:16:35 i also need defenders big time 2011-10-30T19:16:45 actually that should be my first step 2011-10-30T19:17:08 how to defend is the question 2011-10-30T19:18:50 maybe we need to simulate this bug: http://www.youtube.com/watch?v=mA37cb10WMU 2011-10-30T19:19:16 For the main server, has anyone suggested the idea of giving slight preference to 1) newer submissions and 2) submissions in the upper x% (where x could be as big as the top 50%, etc.). That way, 80% of the server resources aren't tied up just playing starter packages against each other over and over again... 2011-10-30T19:19:52 henry61: those are already implemented 2011-10-30T19:20:01 interesting game.. http://aichallenge.org/visualizer.php?game=35451 2011-10-30T19:20:30 avdg: LMAO, perhaps so 2011-10-30T19:20:30 ah, okay. that's great, thanks 2011-10-30T19:20:59 i wonder what percentage of the bots are unedited starter bots? 2011-10-30T19:21:24 amstan: and I guess they would just revolve around the ant hill 2011-10-30T19:21:30 avdg: * 2011-10-30T19:21:38 ? 2011-10-30T19:22:11 that video looks like an evil hurricane 2011-10-30T19:22:20 :p 2011-10-30T19:22:53 *** AxVapor has joined #aichallenge 2011-10-30T19:23:31 huh -- has anyone experimented with having ants lay down pheromone trails? 2011-10-30T19:23:49 lol 2011-10-30T19:24:15 *** Jak_o_Shadows has joined #aichallenge 2011-10-30T19:24:15 cyphase: really! 2011-10-30T19:24:31 I haven't tried pheremones yet. Probably will eventually 2011-10-30T19:24:48 But the game food doesn't work like real world food 2011-10-30T19:24:52 "Test Error: compiled, but failed test cases": turn 1 bot 0 timed out 2011-10-30T19:24:59 mh.. what kind of test this is? 2011-10-30T19:25:03 So you don't really need to create persistent paths to it. 2011-10-30T19:25:13 on local everything seem to work fine 2011-10-30T19:25:22 tools/submission_teste 2011-10-30T19:25:34 it works fine :\ 2011-10-30T19:25:36 callahan: I wish it did work like real world food (food dropped in bunches and having to carry it back to the hill) 2011-10-30T19:25:51 supply lines :D 2011-10-30T19:26:48 *** cutoff has quit IRC (Ping timeout: 240 seconds) 2011-10-30T19:27:33 *** randomdude has joined #aichallenge 2011-10-30T19:28:05 *** delt0r__ has quit IRC (Ping timeout: 244 seconds) 2011-10-30T19:28:41 lol, i said supply lines, then cutoff left the channel 2011-10-30T19:28:41 *** Jak_o_Shadows1 has joined #aichallenge 2011-10-30T19:30:43 *** Jak_o_Shadows has quit IRC (Ping timeout: 252 seconds) 2011-10-30T19:31:31 bleh, this would be the ultimate bot: http://www.youtube.com/watch?v=IBTjQMtbViU 2011-10-30T19:31:52 hmm.. sure that the tests are working fine on the server atm? I mean it was down for some time. 2011-10-30T19:31:57 *** kgk has joined #aichallenge 2011-10-30T19:32:47 I'm unable to reproduce this error on my local machine 2011-10-30T19:35:22 *** freenity has quit IRC (Quit: Leaving) 2011-10-30T19:35:55 *** tmandry has quit IRC (Ping timeout: 260 seconds) 2011-10-30T19:36:35 *** kgk has quit IRC (Ping timeout: 265 seconds) 2011-10-30T19:36:50 *** wombot_ has quit IRC (Remote host closed the connection) 2011-10-30T19:37:32 *** Migi32 has quit IRC (Quit: bye) 2011-10-30T19:39:58 *** bluegaspode has quit IRC (Ping timeout: 265 seconds) 2011-10-30T19:40:23 there is no timeout even if I'll set turntime to 100 for the test_bot script 2011-10-30T19:40:34 *** delt0r__ has joined #aichallenge 2011-10-30T19:42:11 *** srjek has joined #aichallenge 2011-10-30T19:42:11 *** lastaid_ has joined #aichallenge 2011-10-30T19:42:26 hmm.. I don't know what should I test or investicate next 2011-10-30T19:42:29 Hey, i have a problem, i have compiled my bot, but i dont know how test it 2011-10-30T19:42:56 there's test_bot.cmd 2011-10-30T19:43:02 i wrote it in java, 2011-10-30T19:43:20 jes, but do i give test_bot.cmd any kind of argument=? 2011-10-30T19:43:40 "java -jar ../MyBot.jar" i think 2011-10-30T19:44:07 at least I did that way 2011-10-30T19:44:22 ok, did that now i wait 2011-10-30T19:44:40 but how can i see the visualization? 2011-10-30T19:44:50 also nothing seems to happen 2011-10-30T19:45:36 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-10-30T19:46:36 i seriously doubt that what i just did is doing anything 2011-10-30T19:47:11 code says it reads system input ... 2011-10-30T19:47:39 anyone? 2011-10-30T19:47:53 lastaid: did u look at the tutorial? 2011-10-30T19:48:23 lastaid: it has examples in Java and how to ran a game against hunter bot 2011-10-30T19:48:23 yes i did, but i seems that it ignores my changes. 2011-10-30T19:48:26 make works fine 2011-10-30T19:48:49 so.. your bot doesn't move? 2011-10-30T19:49:28 *** wombot_ has joined #aichallenge 2011-10-30T19:49:46 it doesnt update my bot according to the code 2011-10-30T19:49:50 just uses the stock code 2011-10-30T19:49:52 *** dfsd has quit IRC (Quit: Page closed) 2011-10-30T19:50:01 i setup the directory just as it is in the tutorial 2011-10-30T19:50:14 *** dinosoep has joined #aichallenge 2011-10-30T19:50:46 I get compilation errors on submitted files which don't give compilation errors on my own computer 2011-10-30T19:51:04 not failed test cases errors? 2011-10-30T19:51:29 no, everything worked fine. no assert's failed, all unitests ran,... 2011-10-30T19:51:29 I get strange failed test case error and am unable reproduce it on local machine 2011-10-30T19:51:36 hello i'm a friend of lastaid...i had a similar competition already like this and it was necessary to register your own bot there in an INI file 2011-10-30T19:51:45 is such stuff necessary here also? 2011-10-30T19:51:53 *** tmandry has joined #aichallenge 2011-10-30T19:51:53 *** tmandry has joined #aichallenge 2011-10-30T19:51:53 no 2011-10-30T19:52:16 and Draakon, what language are you using? 2011-10-30T19:52:29 *** JDawg has joined #aichallenge 2011-10-30T19:52:52 lastaid: no.. you just point the scripts to your jar file 2011-10-30T19:52:59 hrm, so if my java bot tries to print to the terminal using System.out.Printf(), it doesn't show up in the terminal 2011-10-30T19:53:01 *** olexs1 has quit IRC (Quit: Leaving.) 2011-10-30T19:53:06 anyone got any clever ideas for debugging? 2011-10-30T19:53:13 dinosoep: Java 2011-10-30T19:53:14 JDawg: yes.. use a logger 2011-10-30T19:53:22 JDawg: write out to a file 2011-10-30T19:53:34 well duh 2011-10-30T19:53:34 i am using python 2.7, might this be a problem?? 2011-10-30T19:53:34 I always write all my output to a file :) 2011-10-30T19:53:35 System.out.printf is for writing bot commands 2011-10-30T19:53:41 oh, lol 2011-10-30T19:53:46 that could be a problem 2011-10-30T19:53:51 :D 2011-10-30T19:53:52 lastaid: nope using 2.7 here too 2011-10-30T19:53:58 damnit ... 2011-10-30T19:54:11 *** srjek has left #aichallenge 2011-10-30T19:54:16 stdin/stdout is for bot input and output 2011-10-30T19:54:21 and java's kit won't have any obvoious errors 2011-10-30T19:54:26 jes 2011-10-30T19:54:36 aichallenge: Paul Walton epsilon * rac6a9f2 / (2 files): Edited a couple of comments (+6 more commits...) - http://git.io/H1sdqw 2011-10-30T19:54:37 *** nolo_contendere has quit IRC (Read error: Connection reset by peer) 2011-10-30T19:54:37 and grom, if you write to stdout it'll print that output too 2011-10-30T19:55:00 mh... i have the java_starter_pack and in this folder i also have the tools folder 2011-10-30T19:55:14 stdout from your bot feeds into the game.. the engine reads what gets put onto stdout 2011-10-30T19:55:36 so.. you can't use it for printing out other stuff 2011-10-30T19:55:38 rats, it looks like System.err doesn't print ot the terminal either 2011-10-30T19:55:49 JDawg: just use a logger 2011-10-30T19:56:02 JDawg: java.util.logging .. built into Java for you to use 2011-10-30T19:56:13 then I'm abusing it grom :p it prints my output with #error next to it. Doesn't give any problems though 2011-10-30T19:56:14 ah, great, thank you much 2011-10-30T19:56:25 was hoping not to write java file IO from scratch 2011-10-30T19:56:42 *** dinosoep has quit IRC (Quit: Page closed) 2011-10-30T19:57:17 dineosoep: mmm.. well I wouldn't do it. You basically sending a bunch of invalid commands to the engine 2011-10-30T19:59:48 wow, this java logging thing is beastly 2011-10-30T20:00:05 if you guys want to test your bot, what exactly do you do? 2011-10-30T20:00:24 lastaid: I start by playing against the hunter bot 2011-10-30T20:00:30 on the tutorial map 2011-10-30T20:00:32 *** bob__ has joined #aichallenge 2011-10-30T20:00:37 then move to a maze map 2011-10-30T20:01:13 JDawg: http://pastebin.com/K5QiPUBK 2011-10-30T20:01:18 *** bmh has quit IRC (Quit: bmh) 2011-10-30T20:01:28 JDawg: sets up logger that writes to a file called log.txt 2011-10-30T20:01:32 "Test Error: compiled, but failed test cases": turn 1 bot 0 timed out 2011-10-30T20:01:35 still no ideas? 2011-10-30T20:01:41 wow thanks 2011-10-30T20:02:22 JDawg: then you just do logger.info("my debug message"); to use it 2011-10-30T20:02:27 i got it, had to use another srypt to use the java bot instead of the python but 2011-10-30T20:02:28 bot 2011-10-30T20:02:42 *** Cybsy has quit IRC (Ping timeout: 244 seconds) 2011-10-30T20:02:54 lastaid: you didn't select the Java tab in the tutorial did u? 2011-10-30T20:03:27 *** rharmsen has quit IRC (Quit: Page closed) 2011-10-30T20:03:49 *** ztfw has quit IRC (Remote host closed the connection) 2011-10-30T20:06:27 *** xathis has quit IRC () 2011-10-30T20:08:30 C++ may be helpp to code in, but being able to do depth 90 breadth first searches in 12ms is awesome 2011-10-30T20:09:10 bob: Java pretty fast too, :P 2011-10-30T20:10:17 unfortunately, I still can't think of a good way to handle fighting 2011-10-30T20:10:33 bob__: same 2011-10-30T20:10:50 my idea at the moment.. cluster up ants and hope for the best 2011-10-30T20:11:14 should be lot better than my current suicidal ants 2011-10-30T20:11:28 mine just go for their goal while completely ignoring enemies 2011-10-30T20:11:36 bob: sames 2011-10-30T20:12:22 so any form of defence means bad news for my bot 2011-10-30T20:12:48 got me into above 400 rank so far though 2011-10-30T20:12:57 yeah 2011-10-30T20:13:08 mine just does BFS and it's already rank 116 2011-10-30T20:13:18 you don't need fighting ability till you get near the top 2011-10-30T20:13:34 116.. how? do you have defender ants? 2011-10-30T20:13:47 nope 2011-10-30T20:13:50 your bot kind of sounds like at same level as mine 2011-10-30T20:14:13 on a single hill map, it usually has ants spawning near continuously from the hill 2011-10-30T20:14:19 on multihill maps it doesn't do so well 2011-10-30T20:14:30 so defence kind of happens on accident 2011-10-30T20:14:31 ah.. yeah i been in a few multihill maps 2011-10-30T20:14:52 yeah.. do you priority food over hills? 2011-10-30T20:15:19 my ants just go for the closest food, unexplored space, or enemy hill 2011-10-30T20:15:35 bob: yeap.. same as mine 2011-10-30T20:15:35 while trying to avoid colliding with each other 2011-10-30T20:15:43 just wondering how u got to 116 is all 2011-10-30T20:15:52 how many games have you played? 2011-10-30T20:16:02 with current one.. 5 2011-10-30T20:16:07 it will go up a lot then 2011-10-30T20:16:11 the games are taking way too long 2011-10-30T20:16:13 *** bchoii has quit IRC (Ping timeout: 265 seconds) 2011-10-30T20:16:14 mine's at 20 right now 2011-10-30T20:16:31 there's a big uncertianty penalty, so just playign more games will make you shoot up the rankings 2011-10-30T20:19:46 I wonder if I should make mine go for non visible spaces 2011-10-30T20:19:56 because right now, once the map is explored and no food is visible, they just sit there 2011-10-30T20:20:01 *** replore_ has joined #aichallenge 2011-10-30T20:20:15 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-30T20:20:47 bob: yeah how can they just sit there.. if map is explored then there is enemy hill to go for 2011-10-30T20:20:48 *** danielharan has joined #aichallenge 2011-10-30T20:20:56 I don't know 2011-10-30T20:20:58 it's strange 2011-10-30T20:21:20 I did code up a fallback maze walk routine.. but it never used cause it never gets into that state 2011-10-30T20:21:25 I guess the search could be timing out 2011-10-30T20:21:29 but it doesn't seem likely 2011-10-30T20:21:44 even with 38 searchs and 90 depth, it only takes 12ms on my laptop 2011-10-30T20:22:17 yeah.. my bot taking less then 100ms for it turn.. even with 200 ants 2011-10-30T20:22:46 though for some weird reason one of my games my bot timeout on turn 1. but I seen on forums a post about java bots doing that sometimes 2011-10-30T20:22:56 duh, i promised myself that after newest version i won't work on my bot anymore... 2011-10-30T20:23:04 but i know already how to make it better 2011-10-30T20:23:09 someone suggested its was Java warmup or GC 2011-10-30T20:23:13 how? 2011-10-30T20:23:14 *any*more? 2011-10-30T20:23:20 well, I'm excited about this new Java bot 2011-10-30T20:23:22 I got heaps todo yet 2011-10-30T20:24:03 *** bmh has joined #aichallenge 2011-10-30T20:24:16 but i need to create first a scenariobott - it would gather minimal amount of food and place ants in given places, so i could test my bot how he handles those situations 2011-10-30T20:24:17 atm I said the closest ant to food (if its walkable - using BFS).. but just thought of something before. I should group up food into patches.. and one ant can collect all in that patch 2011-10-30T20:24:42 sent* 2011-10-30T20:24:49 anyone exactly knows how the fight system works? 2011-10-30T20:25:06 liek i saw 2 ants take down single ants multiple times 2011-10-30T20:25:07 lastaid_: http://aichallenge.org/specification_battle.php 2011-10-30T20:25:17 thank you 2011-10-30T20:26:02 lastaid_: lso, lookup do_attack_focus function in aichallenge/ants/ants.py in repo 2011-10-30T20:26:18 *** bmh has quit IRC (Client Quit) 2011-10-30T20:28:07 *** danielharan has quit IRC (Remote host closed the connection) 2011-10-30T20:28:42 *** danielharan has joined #aichallenge 2011-10-30T20:28:44 *** foRei has quit IRC (Read error: Connection reset by peer) 2011-10-30T20:33:11 i am currently thinking to implement flocks 2011-10-30T20:33:28 but i am quite new to this, so it should be quite challenging 2011-10-30T20:33:42 grom358: thanks for the help on the logger, was able to solve my problem pretty quickly with it 2011-10-30T20:36:11 *** chipsbob has joined #aichallenge 2011-10-30T20:37:26 *** chipsbob has left #aichallenge 2011-10-30T20:42:00 *** tmandry has quit IRC (Ping timeout: 240 seconds) 2011-10-30T20:47:25 *** Redgis has quit IRC (Ping timeout: 240 seconds) 2011-10-30T20:48:02 *** Jak_o_Shadows has joined #aichallenge 2011-10-30T20:50:05 *** onensora has quit IRC () 2011-10-30T20:50:57 *** Jak_o_Shadows1 has quit IRC (Ping timeout: 255 seconds) 2011-10-30T20:52:21 *** yascha has quit IRC (Quit: Page closed) 2011-10-30T20:55:13 *** venom has joined #aichallenge 2011-10-30T20:56:21 *** wombot_ has quit IRC (Ping timeout: 255 seconds) 2011-10-30T20:57:06 JDawg: no problem. I try to help where I can 2011-10-30T20:57:29 this challenge is hard enough then having to learn unrelated stuff like setting up a logger 2011-10-30T20:59:38 *** Relax has quit IRC (Remote host closed the connection) 2011-10-30T21:00:07 Hey, where can I find info about the TCP bot? 2011-10-30T21:01:09 No TCP Bot, but TCP contest engine 2011-10-30T21:01:11 @TCP 2011-10-30T21:01:12 venom: TCP could be http://ants.fluxid.pl/howto. 2011-10-30T21:01:43 Seems like what I was looking for. 2011-10-30T21:05:00 *** cutoff has joined #aichallenge 2011-10-30T21:05:02 *** JDawg has quit IRC (Ping timeout: 265 seconds) 2011-10-30T21:05:14 *** wombot_ has joined #aichallenge 2011-10-30T21:07:11 this year's challenge is definitely harder than the other ones 2011-10-30T21:11:41 *** pvarga_ has joined #aichallenge 2011-10-30T21:13:19 *** bmh has joined #aichallenge 2011-10-30T21:15:12 *** pvarga has quit IRC (Ping timeout: 240 seconds) 2011-10-30T21:15:12 *** pvarga_ is now known as pvarga 2011-10-30T21:16:46 more complicated, at least 2011-10-30T21:17:10 *** bmh has quit IRC (Client Quit) 2011-10-30T21:20:00 *** pvarga has quit IRC (Ping timeout: 240 seconds) 2011-10-30T21:20:31 jmcarthur: yup 2011-10-30T21:23:53 *** gcflymoto has quit IRC (Ping timeout: 265 seconds) 2011-10-30T21:28:43 *** willrandship has joined #aichallenge 2011-10-30T21:30:11 question 2011-10-30T21:30:43 when the rules mention bad sportsmanship, that doesn't refer to the actual bpt strategies, right? 2011-10-30T21:30:57 bot* 2011-10-30T21:32:14 *** willrandship has left #aichallenge 2011-10-30T21:36:16 I'm trying to debug my python bot with print statements... except that won't work because print statements won't do anything... what can I do? 2011-10-30T21:38:16 actually nvm 2011-10-30T21:38:23 I'll just write to a file 2011-10-30T21:40:43 *** TinBane has joined #aichallenge 2011-10-30T21:41:51 Hi, anyone else getting or solve the error "unknown action"? My script runs fine locally, but seems to fail after a random number of turns when testing online. 2011-10-30T21:44:37 TinBane: the message should provide enough details 2011-10-30T21:44:40 *** bob__ has quit IRC (Ping timeout: 265 seconds) 2011-10-30T21:44:48 and if you don't know the game spec, read it 2011-10-30T21:44:54 its good to know it 2011-10-30T21:45:01 http://aichallenge.org/specification.php 2011-10-30T21:45:19 It only gives me: turn 1 bot 0 invalid actions: 11 # unknown action 2011-10-30T21:45:53 so, basically I've got to run it locally, and see if I can catch it out giving an invalid command, even though it runs fine on the viewer locally? 2011-10-30T21:46:03 *** Jak_o_Shadows1 has joined #aichallenge 2011-10-30T21:46:12 did you also check the output file of the bot? 2011-10-30T21:46:31 I think its option -O in case it isn't there 2011-10-30T21:46:36 *** xoposhiy has quit IRC (Ping timeout: 265 seconds) 2011-10-30T21:47:08 locally, or on the server? 2011-10-30T21:47:18 local 2011-10-30T21:47:27 Cheers, I'll check that. 2011-10-30T21:47:28 you can't retrieve debug information from the server 2011-10-30T21:47:38 That's what I thought :( 2011-10-30T21:47:42 unless you can rely on the very little information 2011-10-30T21:49:33 *** Jak_o_Shadows has quit IRC (Ping timeout: 276 seconds) 2011-10-30T21:56:16 *** AxVapor has quit IRC (Quit: AxVapor) 2011-10-30T21:56:26 *** AxVapor has joined #aichallenge 2011-10-30T21:56:32 *** AxVapor has left #aichallenge 2011-10-30T21:56:51 *** AxVapor has joined #aichallenge 2011-10-30T21:56:58 i wish aichallenge didnt use -O3 on gcc 4.5 2011-10-30T21:57:12 it generates invalid assembly in certain cases due to an optimization bug 2011-10-30T21:58:39 *** AxVapor has quit IRC (Client Quit) 2011-10-30T22:01:36 do AI built in compiled languages have an advantage in speed in this game? 2011-10-30T22:01:44 yes 2011-10-30T22:01:46 obviously 2011-10-30T22:01:47 does "unknown action" mean a command that means nothing, or could it be an illegal move? Because it's outputting absolutely standard orders, as per the spec :/ 2011-10-30T22:01:54 hmpf 2011-10-30T22:02:04 Yes, you can do a better search, because you have more timelimit. 2011-10-30T22:02:21 I figured some form of equilization would be used 2011-10-30T22:02:27 how? 2011-10-30T22:02:40 That would be pretty contentious. 2011-10-30T22:04:45 that sucks 2011-10-30T22:05:17 Well how would you weight it to be fair? It's impossible. 2011-10-30T22:05:26 Odds are, they will split the results by language type. 2011-10-30T22:05:32 Top python implementation, etc. 2011-10-30T22:06:03 You can't factor a direct comparison, because the speed to execute depends on the mix of commands used. 2011-10-30T22:10:29 *** bmh has joined #aichallenge 2011-10-30T22:12:24 *** Da_Blitz has joined #aichallenge 2011-10-30T22:15:26 well.. last contest.. the winner was in lisp I think 2011-10-30T22:16:45 Ionic_Groove: A Python bot has been number one for a while. It's in the top three now. Java isn't considered a particularly fast language and it holds the top two spots at the moment. 2011-10-30T22:16:47 Ionic_Groove: language choice is a factor in this contest. it's up to you to decide what tradeoffs to make 2011-10-30T22:18:24 *** djr_ has quit IRC (Read error: Connection reset by peer) 2011-10-30T22:18:58 *** djr_ has joined #aichallenge 2011-10-30T22:20:58 *** tmandry has joined #aichallenge 2011-10-30T22:20:58 *** tmandry has joined #aichallenge 2011-10-30T22:24:22 processing speed really isn't much of a factor in this contest 2011-10-30T22:25:05 bysin: i thought if in our makefile we don't use -O3 then that should be good enough, no? 2011-10-30T22:25:17 they dont honor the makefile 2011-10-30T22:25:30 so no c99 either 2011-10-30T22:29:08 *** lastaid_ has quit IRC (Ping timeout: 265 seconds) 2011-10-30T22:30:36 * avdg wonders how js would perform 2011-10-30T22:30:59 there are still no good bots in js though 2011-10-30T22:31:16 there are several close ones though 2011-10-30T22:31:33 that's cause js sucks :P 2011-10-30T22:31:46 avdg: someone should write a bot in node ;) 2011-10-30T22:31:50 what doesn't ;-) 2011-10-30T22:31:57 I am doing in js 2011-10-30T22:32:09 I just need to write some tools atm 2011-10-30T22:32:14 for debugging purposes 2011-10-30T22:34:57 for fixing this game: http://aichallenge.org/visualizer.php?game=33780 2011-10-30T22:34:58 *** pvarga has joined #aichallenge 2011-10-30T22:36:59 *** joakimar has quit IRC (Quit: Page closed) 2011-10-30T22:38:31 *** grom358 has quit IRC (Read error: Connection reset by peer) 2011-10-30T22:38:55 *** grom358 has joined #aichallenge 2011-10-30T22:41:19 anyone knows the specs of the contest's game server? 1 second per turn doesn't mean nothing when planning your algo, strategies & tactics 2011-10-30T22:42:06 venom: make it flexible. Do something, ask how long it took and plan accordingly 2011-10-30T22:42:07 the specs of the server wouldn't give a good idea, but at least your could speculate a little 2011-10-30T22:42:19 In the beginning of the universe you have a lot of time to kill 2011-10-30T22:42:27 they are quite powerful afaik 2011-10-30T22:42:30 just burn your first few turns figuring out how many horses you have 2011-10-30T22:43:23 bmh: of course, but some strategies and their algo just aren't good when you don't get enough processing power/time 2011-10-30T22:43:42 venom: implement a bunch and pick which one to use based on how fast the box is :) 2011-10-30T22:44:05 and think about things you (could) have 2011-10-30T22:44:06 bmh: yeah, we're pretty stuck with that, I know. 2011-10-30T22:45:56 avdg: by the stats it gives on the profile page, I feel it's not that powerful. My system could run as much games as that server and mine is only a Athlon II X4 630 with 4gb ram 2011-10-30T22:46:07 "only" 2011-10-30T22:46:50 Anyway, I wish I game logs would give us some info about how much times each turn took. Wait... are they? meh. 2011-10-30T22:47:32 venom: yes, but are you running 8 players at the same time? 2011-10-30T22:48:34 bmh: you're right. It's pretty good and I come off as a non-appreciative asshole. :) 640kb should be enough for everyone! and here I am crying of my 4gb and awesome cpu. 2011-10-30T22:49:09 venom: I think the box under my desk at work has 16gb, I've never really noticed. 2011-10-30T22:49:09 amstan: nope, but I factored the number of players that we're playing per minutes. 2011-10-30T22:49:40 did you also take into account that the players use the full time? 2011-10-30T22:49:48 I can't recall many problems where I've actually been constrained by the specifications of my machine 2011-10-30T22:51:27 bmh: what an awesome technological era we're into. My first computer had a 40mb harddrive and about 2 megs of RAM and your machine has 8192 time more ram than my teenage years rig. wow. 2011-10-30T22:51:40 venom: Was that 1992? 2011-10-30T22:51:57 amstan: what do you mean by 'use the full time'? 2011-10-30T22:52:11 I remember having a 486 with 40mb harddrive, 2mb ram, and a 2400 bps modem! 2011-10-30T22:52:29 I only remember 2gb drives :/ 2011-10-30T22:52:36 they can't just be the starter bots, they have to use 0.5s each turn if you wanna test it like that. 2011-10-30T22:52:48 in 99 I had a system with 1gig of ram, and it pwned so hard 2011-10-30T22:53:01 bmh: yeah, near enough. *close the light, raise the volume* Second Reality .. awwwwwww 2011-10-30T22:53:24 bmh: about the same rig as mine, except I had a 386. 2011-10-30T22:53:58 For the first time in 13 years my internet has gotten faster! 2011-10-30T22:53:59 avdg: Well, your :/ should be -> :D lol 2011-10-30T22:54:19 burny: that's a serious amount of ram for that long ago 2011-10-30T22:54:41 amstan: I get it. Btw, is the server spec a secret? 2011-10-30T22:54:51 venom: no 2011-10-30T22:55:03 venom: i'm just a little lazy to get it right now 2011-10-30T22:55:03 burny: 1gb in 99 was epic 2011-10-30T22:55:13 i know it's dual core and has 12GB ram 2011-10-30T22:55:14 ya.. it was 2011-10-30T22:55:29 I could copy cd's to ram drive 2011-10-30T22:56:35 amstan: good enough. but seeing the number of registration each days and the time it already take to have a game played, I feel like it could become to slow by mid-november. Do you feel too? 2011-10-30T22:56:41 some games wouldn't install though 2011-10-30T22:57:31 *to = too 2011-10-30T22:57:31 cause 2gig < 16meg requirements, in 32bit signed 2011-10-30T22:57:43 venom: i think we'll get more comps near the end 2011-10-30T22:57:46 burny: that's beautiful. 2011-10-30T22:57:52 burny: lol 2011-10-30T22:57:53 contestbot: seen janzert 2011-10-30T22:57:53 amstan: janzert was last seen in #aichallenge 3 hours, 49 minutes, and 47 seconds ago: for 10 player game 2011-10-30T22:58:05 the machine was for work though, not personal 2011-10-30T22:58:08 yes? 2011-10-30T22:58:21 amstan: okay 2011-10-30T22:58:24 I ran into a problem with a satellite phone. The drivers were written for Windows XP and I was using a computer with Windows 7 -- "Sorry, this application requires Windows XP or later" 2011-10-30T22:59:26 wow.. I haven't seen that error since 95 or 98 versions 2011-10-30T22:59:32 "later" as "whatever, later! *runs*" 2011-10-30T23:00:02 it eventually did work in compatibility mode, but for a couple of minutes my stomach was on the floor 2011-10-30T23:00:27 *** venom is now known as venom_away 2011-10-30T23:00:51 i had this one program, which wouldn't run on 98 and 95 claiming it only worked in win3.1 2011-10-30T23:00:55 but then it worked on xp 2011-10-30T23:02:15 We should surpass planetwars number of total users tomorrow 2011-10-30T23:02:21 woot! 2011-10-30T23:02:24 sweet. 2011-10-30T23:02:31 hopefully I can get up a bot that isn't the started bot by tomorrow... 2011-10-30T23:03:32 I got some good leads 2011-10-30T23:03:38 bmh: are you still doing the map predictions? 2011-10-30T23:03:40 imo what works best is a scavenger/hunter combo 2011-10-30T23:03:50 hunter = a* — quick food grabbing 2011-10-30T23:04:01 no food, go to a kickass explorer mode to find food 2011-10-30T23:04:03 amstan: I would like to, I need to get other stuff working first 2011-10-30T23:21:47 this thing is like crack 2011-10-30T23:21:49 can't put it down 2011-10-30T23:22:06 it drives me crazy cause it's hard to get a good optimized bot but i can't put it down 2011-10-30T23:22:10 love hate relationship all the way 2011-10-30T23:25:47 roflmao: I agree... I should be doing my homework but nope... just spent the entire day on this x.x 2011-10-30T23:26:23 I finally got my alternative pathfinder working! (it's not A* mwahahahaha and I had to build it from scratch ) 2011-10-30T23:28:45 * avdg thinks a* is fitting less in his idea, the more he thinks about it 2011-10-30T23:28:59 though its still a good one to implement 2011-10-30T23:29:25 a* fits perfectly in my "idea" 2011-10-30T23:29:35 it just doesn't fit perfectly for the speed I require 2011-10-30T23:29:52 btu that was the old ruby code, now I'm trying in java and doing things a lot more intellegently so hopefully this time around I'll fit it in :) 2011-10-30T23:30:00 yeah, I was already thinking about the pf problem from day 1 2011-10-30T23:30:25 and I'm still stuck with how I should implement my idea 2011-10-30T23:30:28 thing is a* is actually more optimized than other simpler pathfinding algorithms like bfs 2011-10-30T23:30:45 i think a correctly implemented a* should be able to solve the whole board in one turn pretty easily 2011-10-30T23:30:45 :X 2011-10-30T23:30:47 for point to point pathfinding yes 2011-10-30T23:30:51 yeah A* was slowing mine down so I created my own 2011-10-30T23:31:04 so far no timeouts at all even with 100s of ants =D 2011-10-30T23:31:11 using my method for each ant woohoo 2011-10-30T23:31:13 but I am looking for situations where I can rely less on point to point navigation 2011-10-30T23:31:19 please share your method asdfafas 2011-10-30T23:31:19 and on python lol 2011-10-30T23:31:28 NICE dude cause I would like to use ruby 2011-10-30T23:31:35 instead of ugly verbose java 2011-10-30T23:31:45 I'll probably do a re-write in C++ 2011-10-30T23:31:46 if you want to keep it sekkrit i understand :P 2011-10-30T23:32:11 roflmao: will hack it anyway ;-) (just kidding :-)) 2011-10-30T23:32:12 well... part of it is turning the map into a graph 2011-10-30T23:32:21 you'll have to figure out the rest ;) 2011-10-30T23:32:33 s/:// 2011-10-30T23:32:43 my brain just hurts thinking about map into graph 2011-10-30T23:33:14 yeah, that reward would be very great for me 2011-10-30T23:33:15 a graph? isn't that what A* and others do? (it's been awhile for me) 2011-10-30T23:33:19 just remember... the maps we play on is a torus... use that to your advantage too 2011-10-30T23:33:26 torus? 2011-10-30T23:33:36 tmandry: yup, but I made a few changes 2011-10-30T23:33:36 roflmao: it wraps 2011-10-30T23:33:41 actually quite a few changes 2011-10-30T23:33:47 ah yes, that's key and I haven't implemented that 2011-10-30T23:33:49 its like a donut 2011-10-30T23:34:16 asdsfafas: I take it you sacrificed better paths for more speed? 2011-10-30T23:34:17 but a very strange one 2011-10-30T23:34:33 roflmao: exactly... if you're at (0,0) and you want to get to (29,0) you can do it in 1 step (map height = 30) 2011-10-30T23:34:41 tmandry: yessir 2011-10-30T23:34:54 yep 2011-10-30T23:35:04 i just haven't thought it through yet 2011-10-30T23:35:06 tmandry: although the paths it finds aren't the worst either 2011-10-30T23:35:18 on how to implement it through the heuristic and pathfinding 2011-10-30T23:35:18 tmandry: just not the absolute shortest 2011-10-30T23:35:22 I will, eventually, of course 2011-10-30T23:35:24 aint too hard 2011-10-30T23:35:39 tmandry: since map conditions change anyway (enemies moving n whatnot) there's no point using A* 2011-10-30T23:36:08 random question: 2011-10-30T23:36:13 asdsfafas: fair enough. sounds like you've analyzed the problem pretty well 2011-10-30T23:36:18 asdsfafas: you can generate targets ;-) 2011-10-30T23:36:22 overestimating the heuristic cost in a standard a* to take off emphasis from the G 2011-10-30T23:36:25 * tmandry just started looking at AI challenge.. no code yet :) 2011-10-30T23:36:35 does that mean LOWERING the heuristic cost artificially or INCREASING it 2011-10-30T23:37:43 logically I think it is lowering it artificially cause that'll lead to lower costs for the H which pushes G nodes back on the open set a bit 2011-10-30T23:37:53 if that makes any sense at all 2011-10-30T23:38:01 but "overestimating" implies increasing the cost of h artificially 2011-10-30T23:39:04 *** dvladim has joined #aichallenge 2011-10-30T23:39:05 really I just want to use a* to snatch very closeby foods 2011-10-30T23:39:19 and then use something else for the rest due to changing conditions 2011-10-30T23:39:37 a* would only be used for foods that are 5-6 squares away from ant 2011-10-30T23:39:59 * avdg is trying to think outside the box 2011-10-30T23:40:18 one thing I have discovered though is because food pops up everywhere 2011-10-30T23:40:35 *** bmh has quit IRC (Changing host) 2011-10-30T23:40:35 *** bmh has joined #aichallenge 2011-10-30T23:40:36 ON AVERAGE if I allow a* to craft a path for an ant to a food that is rather far away 2011-10-30T23:40:45 that ant ends up picking up 4-5 foods automatically along the way 2011-10-30T23:40:50 pathfindin would be so much easier if the damn ants had eyes 2011-10-30T23:41:22 :) the best way to pf is not having to pf at all 2011-10-30T23:41:31 bmh: yeah these ants are nothing like REAL ants 2011-10-30T23:41:33 you don't pathfind? 2011-10-30T23:41:39 well, minimize 2011-10-30T23:42:03 I was thinking in only pathfinding for nearby foods 2011-10-30T23:42:09 and if the path goes over 5 steps abort 2011-10-30T23:42:20 cause sometimes a nearby food is actually behind a huge wall that you need to go around 2011-10-30T23:42:22 and can be deceptive 2011-10-30T23:42:26 and thats where you put a* for, right... 2011-10-30T23:42:31 correct 2011-10-30T23:42:40 a* is good for long distances 2011-10-30T23:42:46 what would you recommend for short? 2011-10-30T23:42:46 with relative cheep cost 2011-10-30T23:42:57 5-6 square distance 2011-10-30T23:42:58 * bmh grumbles. 2011-10-30T23:43:02 static. pathfinding. 2011-10-30T23:43:15 a* is good for known mazes, that don't change 2011-10-30T23:43:44 well, that condition only applies when a* is searching 2011-10-30T23:43:51 Frozen yogurt is more fun than pathfinding. I'm going to do that. 2011-10-30T23:43:55 bmh: I just don't see how you are going to reasonably compute all possible paths on a 127x127 grid when you only have 300ms per turn 2011-10-30T23:44:17 you don't need to compute them all 2011-10-30T23:44:25 if there is only 1 wall between you and the food.. 2011-10-30T23:44:31 just follow the wall 2011-10-30T23:44:32 *** ccc has quit IRC (Ping timeout: 265 seconds) 2011-10-30T23:44:34 only 2 possible paths 2011-10-30T23:44:53 then you either know whichone is shorter.. or guess 2011-10-30T23:44:55 *** bmh has quit IRC (Quit: bmh) 2011-10-30T23:44:56 assuming ants only pursue food that are in their line of sight 2011-10-30T23:44:59 "{ 2011-10-30T23:45:01 :P 2011-10-30T23:45:08 if it's far away.. 2011-10-30T23:45:17 then either naother ant is closer.. or you can't see the food 2011-10-30T23:45:28 yep I just made that brilliant discovery today :) 2011-10-30T23:45:29 if you can't see it(cause you died).. then probably an enemy already got it anyway 2011-10-30T23:45:32 and it's going to change my game completely 2011-10-30T23:45:47 oh, the DUHs you miss when you start out 2011-10-30T23:45:57 *** mrbanana has joined #aichallenge 2011-10-30T23:46:11 that's actually very interesting burny, hugging the wall 2011-10-30T23:46:25 so you could implement a sort of combined explorer and lefty bot 2011-10-30T23:46:28 huging the wall fails if there are 2 walls in the way ofcourse 2011-10-30T23:46:45 not if you implement it correctly 2011-10-30T23:46:46 but if there's 2 walls in the way, but it's in sight range.. is a semi rare occurance 2011-10-30T23:46:53 it just means the wall is 2 squares thick 2011-10-30T23:47:05 2squares thick = 1wall 2011-10-30T23:47:18 oh i see what you mean 2011-10-30T23:47:22 aichallenge: McLeopold epsilon * r37f6eb3 / (3 files): speed up matchup creation - http://git.io/QHuHOQ 2011-10-30T23:47:22 aichallenge: McLeopold epsilon * r2667a05 / (17 files in 5 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/nWSIOA 2011-10-30T23:47:23 wall - space - wall - food 2011-10-30T23:47:59 even if it's 2 away, and you assume it's 1away.. there's roughly a 30% chance that you'll see the proper path before you get to it 2011-10-30T23:48:36 (this is getting interesting :-)) 2011-10-30T23:48:43 you are a genius burny 2011-10-30T23:48:48 fuck a* 2011-10-30T23:48:53 I'm going with some of your infinite wisdom 2011-10-30T23:48:59 np 2011-10-30T23:49:04 plus a cool scavenger algorithm that effectively spreads ants 2011-10-30T23:49:18 a* is still a good fallback algorithm ;-) 2011-10-30T23:49:30 yea but once you manage like 150 ants 2011-10-30T23:49:36 you can't afford the expense imo 2011-10-30T23:49:41 true 2011-10-30T23:49:42 unless you use it VERY sparingly 2011-10-30T23:49:42 a* is going to be good for the final attack.. if you don't just save ants paths 2011-10-30T23:49:58 but the trick with a* is to optimize is for your purpose 2011-10-30T23:50:20 and before using a*, you could try to use a few checks to reduce your lookup 2011-10-30T23:50:36 *** bmh has joined #aichallenge 2011-10-30T23:50:36 *** bmh has joined #aichallenge 2011-10-30T23:51:00 anyone using java here that knows how to output to terminal? 2011-10-30T23:51:07 ala python stderr or ruby warn 2011-10-30T23:51:12 *** cutoff has quit IRC (Ping timeout: 240 seconds) 2011-10-30T23:51:13 I'm gonna start my bot in a few weeks.. 2011-10-30T23:51:22 std.out.write or something? 2011-10-30T23:51:25 my goal is going to be to get top 100, with the least ammount of code possible 2011-10-30T23:51:29 s/out/err/ 2011-10-30T23:51:53 *** bmh has quit IRC (Client Quit) 2011-10-30T23:52:15 s/std/system/ 2011-10-30T23:52:18 ? 2011-10-30T23:52:25 that's not java 2011-10-30T23:52:26 top 100 would be awesome! 2011-10-30T23:52:29 it's probably System.out.something 2011-10-30T23:52:32 roflmao: System.err.println() 2011-10-30T23:52:51 logging to syserr feels dirty :> 2011-10-30T23:52:51 meh, I almost never programmed in java 2011-10-30T23:53:01 its the best way ;-) 2011-10-30T23:53:11 if you want to do it quickly 2011-10-30T23:53:12 certainly was the easiest to set up 2011-10-30T23:53:13 haha 2011-10-30T23:53:29 oh.. if anyone else is having problems solving something.. and wants to know how to effectively replace the solution with 'nothing': just ask :) 2011-10-30T23:54:11 yoden: it is very dirty :P 2011-10-30T23:54:12 burny: do you have tnt? :-) 2011-10-30T23:54:20 what's tnt? 2011-10-30T23:54:26 the bomb? 2011-10-30T23:54:26 boooom! 2011-10-30T23:54:35 nope, but I got nitro 2011-10-30T23:54:43 this stupid tutorial bot doesn't chase food like it says :P 2011-10-30T23:54:57 mine doesn't do it as well 2011-10-30T23:54:59 *** djr_ has quit IRC (Quit: Leaving) 2011-10-30T23:55:52 it just bounces back and forth between the same 2 squares 2011-10-30T23:56:30 roflmao: use it with a collision protection 2011-10-30T23:56:53 it may actually win the game in some maps :-) 2011-10-30T23:57:03 it already has collision protection :) 2011-10-30T23:57:18 the problem is that the algorithm they use for going to squares is borked 2011-10-30T23:57:20 at least, if they get enough food 2011-10-30T23:57:29 essentially it sort of has an internal manhattan heuristic 2011-10-30T23:57:40 so lets say food is 3 squares above me but there is wall seperating 2011-10-30T23:57:47 so the only movement actually INCREASES the heuristic 2011-10-30T23:58:03 so then next turn it will go right back to the previous spot since that spot decreases the heuristic 2011-10-30T23:58:13 creating an endless loop of failure (this is the tut bot not my code) 2011-10-30T23:58:22 you need something that is always too optimistic ;-) 2011-10-30T23:58:29 at least with a* 2011-10-30T23:58:51 moving randomly, avoiding collisions, is surprisingly not 100% terrible 2011-10-30T23:58:58 it's the tortoise approach xD 2011-10-30T23:59:02 lol yoden 2011-10-30T23:59:13 one thing that works surprisingly well for exploring is a bouncer 2011-10-30T23:59:19 move into one direction until you hit into wall 2011-10-30T23:59:23 change direction 2011-10-30T23:59:35 obviously you can get into a square of death 2011-10-30T23:59:39 my bot is actually a bad bouncer 2011-10-30T23:59:47 no pathfinding whatsoever?? 2011-10-30T23:59:52 no 2011-10-30T23:59:54 dude 2011-10-30T23:59:55 thats my next layer 2011-10-30T23:59:58 do you actually win games? :P