2012-11-03T00:41:59 *** thestinger has quit IRC (Quit: WeeChat 0.3.9) 2012-11-03T00:59:06 *** antimatroidl has joined #aichallenge 2012-11-03T01:26:45 *** thestinger has joined #aichallenge 2012-11-03T03:07:56 *** thestinger has quit IRC (Quit: WeeChat 0.3.9) 2012-11-03T03:36:54 *** smiley1983 has quit IRC (Ping timeout: 252 seconds) 2012-11-03T03:37:46 *** smiley1983 has joined #aichallenge 2012-11-03T04:10:33 *** antimatroidl has quit IRC (Ping timeout: 240 seconds) 2012-11-03T04:18:46 *** amstan has quit IRC (Quit: Konversation terminated!) 2012-11-03T04:28:43 *** pairofdice has joined #aichallenge 2012-11-03T04:45:43 *** dici has joined #aichallenge 2012-11-03T05:00:59 *** antimatroidl has joined #aichallenge 2012-11-03T05:36:17 Ants beginners: 106.187.94.97:2085 website: http://106.187.94.97:2084/ 2012-11-03T05:36:34 Ants intermediate: 106.187.94.97:2083 website: http://106.187.94.97:2082/ 2012-11-03T05:36:49 Ants Free for All: 106.187.94.97:2081 website: http://106.187.94.97:2080/ 2012-11-03T05:37:06 TRON: 106.187.94.97:2087 website: http://106.187.94.97:2086/ 2012-11-03T05:44:28 *** smiley1983 has quit IRC (Ping timeout: 256 seconds) 2012-11-03T05:45:05 *** smiley1983 has joined #aichallenge 2012-11-03T05:56:38 *** smiley1983 has quit IRC (Ping timeout: 252 seconds) 2012-11-03T05:57:27 *** antimatroidl has quit IRC (Ping timeout: 252 seconds) 2012-11-03T05:57:28 *** smiley1983 has joined #aichallenge 2012-11-03T06:03:26 *** antimatroidl has joined #aichallenge 2012-11-03T06:27:33 antimatroidl: http://106.187.94.97:2086/replay.27 2012-11-03T06:37:09 smiley1983: too fast 2012-11-03T06:37:31 also not square 2012-11-03T06:37:39 looking good though 2012-11-03T06:42:13 *** smiley1983 has quit IRC (Ping timeout: 240 seconds) 2012-11-03T06:43:42 *** smiley1983 has joined #aichallenge 2012-11-03T06:44:00 internet connection has died about five times in the last half hour, glad I'm not trying to run the server off my home connection 2012-11-03T06:45:34 The aspect ratio definitely needs fixing, I've treated visualizer things as the lowest priority 2012-11-03T06:51:11 *** ramn[mba] has joined #aichallenge 2012-11-03T06:51:39 ramn[mba]: TCP server is up: http://forums.aichallenge.org/viewtopic.php?f=21&t=3641 2012-11-03T06:52:14 smiley1983: wicked stuff, well done! 2012-11-03T06:52:23 Cheers :) 2012-11-03T06:52:25 now I have to get my ant into shape.. 2012-11-03T06:53:07 hosting it yourself? 2012-11-03T06:53:31 I'm trying to keep some bots running against the server at all times, but my home connection dies frequently, and they don't restart automatically 2012-11-03T06:53:38 Hosting it on a linode server 2012-11-03T06:53:44 ok 2012-11-03T06:54:36 ramn[mba]: screw ants, tron 2012-11-03T06:54:47 The server currently has plenty of spare CPU and a modest amount of ram, so I'll probably get it to run one bot for each server, and then they won't die all the time 2012-11-03T06:55:07 ah nice 2012-11-03T06:55:14 antimatroidl: you said the replays were too fast, are they stupidly fast, or about 1/2 second per move? 2012-11-03T06:55:16 antimatroidl: you prefer tron? 2012-11-03T06:55:41 smiley1983: what's the stated and actual time limits? 2012-11-03T06:56:02 have you written up what input parameters the tron bots get? 2012-11-03T06:56:05 well io specs? 2012-11-03T06:56:18 I think it's currently 3000ms, 500ms 2012-11-03T06:56:27 i could be tempted to write a tron bot tonight 2012-11-03T06:56:34 see how good i can get the bot in one night 2012-11-03T06:56:42 yes, even serving it up from the webserver: http://106.187.94.97:2086/trondoc 2012-11-03T06:57:31 smiley1983: why do you give the number of columns before the number of rows? 2012-11-03T06:57:46 er... legacy? 2012-11-03T06:57:51 Will change it :) 2012-11-03T06:57:52 please tell me you've got proper indexing (like a matrix) 2012-11-03T06:58:19 oh god :P 2012-11-03T06:58:39 Not sure what you mean; internally, the map is accessed by grid[row][col] 2012-11-03T06:58:45 (0,0) should be top left 2012-11-03T06:58:57 with (1,0) being second row first column 2012-11-03T06:59:12 that's how input looks on a text document 2012-11-03T06:59:23 It isn't? That was certainly my intention... 2012-11-03T06:59:44 look at the example map at the bottom of the link above 2012-11-03T06:59:51 it has the bots at 2,2 and 7,7 2012-11-03T07:00:12 those % signs are not bots 2012-11-03T07:00:20 bots are not marked on the map 2012-11-03T07:00:30 oh, okay 2012-11-03T07:00:41 so, uhhh, the map files don't specify starting location? 2012-11-03T07:00:45 those marks are walls (water from ants) 2012-11-03T07:00:46 the engine does that? 2012-11-03T07:00:59 i like being able to set up very specific test examples 2012-11-03T07:01:02 especially for tron 2012-11-03T07:01:10 yes, they do; "a 2 2 s 0" is a starting position 2012-11-03T07:02:29 I realize when I said bots were not marked on the map, I should have been more specific: they are specified in the map file, but not marked on the grid 2012-11-03T07:04:38 *** antimatroidl1 has joined #aichallenge 2012-11-03T07:05:13 smiley1983: I am slightly confused :P 2012-11-03T07:05:17 *** antimatroidl has quit IRC (Ping timeout: 252 seconds) 2012-11-03T07:05:22 have you got sample io for a game? 2012-11-03T07:05:41 grabbing one to paste... 2012-11-03T07:06:03 also how do you tell me what move people just made? the current location of heads or the moves issued last turn? 2012-11-03T07:06:21 i'd prefer to just be told what moves were issued last turn rather than having to look up which direction people moved 2012-11-03T07:07:18 Currently it tells you the location of heads 2012-11-03T07:07:30 oh hmmm, are you starting heads with a direction? 2012-11-03T07:07:45 old tron let you move up to 4 directions on the first turn 2012-11-03T07:07:59 http://pastebin.com/4XasU8zf 2012-11-03T07:08:30 heads are assigned a direction to begin with, but you never have to move in that direction; it's just for bots who only say "go" 2012-11-03T07:09:52 my suggestions: (feel free to ignore), on the 0th turn heads are just given a location "a r c p" and on each following turn they are tolds moves "m r c d" though don't use m obviously 2012-11-03T07:10:14 when you say "move up to 4 directions", you're referring to the four options? Or you actually took four steps in a turn? 2012-11-03T07:10:16 and i never understood why we use "ready" to indicate the end of input on the 0th turn 2012-11-03T07:10:19 why not just go 2012-11-03T07:10:29 the four options 2012-11-03T07:10:44 every turn after that there's at most 3 cells not blocks 2012-11-03T07:10:48 blocked* 2012-11-03T07:10:52 I'm not sure if that's even part of the game format - I think "ready" and "go" are handled by the engine 2012-11-03T07:10:59 ah okay 2012-11-03T07:11:37 i shall start converting my ants starter package anyway 2012-11-03T07:11:45 then i can write myself a game tree 2012-11-03T07:11:53 Here's the bot output for that game: http://pastebin.com/pMKvVbTu 2012-11-03T07:12:33 bah i still don't have c++11 aha 2012-11-03T07:12:43 starter package can be old school anyway 2012-11-03T07:13:17 actually, here 2012-11-03T07:13:27 Since you're told the direction each head is facing in on any given turn, I'm not sure what the benefit is of telling them the orders rather than the head locations; the bot can just apply a reversed version of the same logic 2012-11-03T07:13:38 0th turn has "a r c p" then future just spits out the input "o r c d" 2012-11-03T07:13:45 don't even need to know which direction the head is facing 2012-11-03T07:14:00 yeah head direction is useless 2012-11-03T07:14:12 you can just regurgitate all bots orders to each player 2012-11-03T07:14:21 bots'* 2012-11-03T07:15:12 personally I'm still a fan of people being able to easily train a bot 2012-11-03T07:15:18 ie. give them the outcome of the game at the end 2012-11-03T07:15:22 the orders contain the same data; head direction, move direction, it's the same thing; the question is, though, do you tell the player where the bot was last turn and which direction it moved in, or do you tell it where it is now and which direction it moved in 2012-11-03T07:15:47 i'd prefer the old location 2012-11-03T07:15:53 personally 2012-11-03T07:16:09 my current state when reading the input will still have them in the old positions 2012-11-03T07:16:21 Why? Also, it would have to recreate orders which were not sent; agents move whether or not they receive a new order 2012-11-03T07:16:31 what? 2012-11-03T07:16:49 Sorry, the "why" was typed before seeing your previous line 2012-11-03T07:17:05 think of it like this, if I receive "o 2 2 n", then for my current state object I go to 2,2 and move the head north 2012-11-03T07:17:10 updating wall info etc. 2012-11-03T07:17:29 or i get 2, 2, find the opposite of north and that location, then move the head north and update info 2012-11-03T07:17:34 i think my way is easier personally 2012-11-03T07:18:06 your way you'd end up just swapping what each direction meant 2012-11-03T07:18:15 Yes; I hadn't envisioned a situation where you would want to record the agent in the map; to me, the grid map is for recording trails, and I would keep agents in a separate data structure 2012-11-03T07:18:42 i recycle my game tree for tron too 2012-11-03T07:19:16 though I see your point, that if you record data between turns for a specific agent, you'll have to find it from its previous location 2012-11-03T07:19:38 i guess your way is easier if you don't remember state information between turns 2012-11-03T07:19:50 which is probably better for beginners 2012-11-03T07:20:00 you may be right 2012-11-03T07:20:42 I'll have a think about it, but I'm working tomorrow and not likely to change it myself tonight :) 2012-11-03T07:20:54 *** mceier_ has joined #aichallenge 2012-11-03T07:21:07 *** mceier has quit IRC (Disconnected by services) 2012-11-03T07:21:14 *** mceier_ is now known as mceier 2012-11-03T07:21:35 working on a sunday? :| 2012-11-03T07:22:30 Every week, better money than other days for the same amount of time; the fact that it's Sunday just means the buses run at different times 2012-11-03T07:22:44 ha, fair call 2012-11-03T07:22:58 i shall get to work on this starter bot anyway 2012-11-03T07:23:08 yes, starters are needed :) 2012-11-03T07:23:15 i'll do c++ 2012-11-03T07:23:16 as always 2012-11-03T07:23:29 should only take an hour or two hopefully 2012-11-03T07:23:37 what's your default strategy? 2012-11-03T07:23:42 wall hugger? 2012-11-03T07:24:17 random non-suicidal bot wouldn't be a bad default strategy 2012-11-03T07:24:21 easy for people to improve on 2012-11-03T07:35:40 antimatroidl1: default strategy is the Ants starter :) 2012-11-03T07:35:46 n, e, s, w bot 2012-11-03T07:36:45 not mandated for others, of course, I just left that part of the Python Ants starter untouched when converting it 2012-11-03T07:46:07 i definitely want to input changed later :P 2012-11-03T07:46:30 i assume you're currently resetting the head info each turn before reading it? 2012-11-03T07:46:41 or are you reading the info and deleting old info based on that? 2012-11-03T07:47:02 smiley1983: what is returned when there's a collision? 2012-11-03T07:47:23 and presumably the collision cell becomes a wall? 2012-11-03T07:47:47 thinking of more than 1snake v 1snake here 2012-11-03T07:48:55 Head on collisions are detected (or should be...) and both agents die and become a wall 2012-11-03T07:49:03 bots are told about dead agents 2012-11-03T07:49:15 including owner and where they died 2012-11-03T07:50:02 "both agents die..." -> "all agents die..." 2012-11-03T07:50:21 eg? 2012-11-03T07:50:41 how are you told when a snake walks into a wall too? 2012-11-03T07:50:59 and what happens if a snake doesn't issue a move? does it die? 2012-11-03T07:51:16 i would treat that like an individual fail from that snake and it dies where it is 2012-11-03T07:51:56 Currently bots are not required to issue orders; they can just say "go", and the agents will proceed in their starting direction until they crash 2012-11-03T07:52:37 Head on collisions and agents walking into walls are signalled in the same way - you are told about the dead agents 2012-11-03T07:54:48 It would be easy to require orders to be given; it's easier to verify that your I/O channels are correctly set up in a new starter package if all you have to do is say "go", and the binary can then "participate" 2012-11-03T07:56:08 smiley1983: but how am i told about dead agents? 2012-11-03T07:56:24 ie. am i just given multiple heads in the same location and expected to work it out myself? 2012-11-03T08:01:26 *** smiley1983 has quit IRC (Ping timeout: 260 seconds) 2012-11-03T08:03:10 *** smiley1983 has joined #aichallenge 2012-11-03T08:03:41 antimatroidl1: You get a line like "d 5 7 0" meaning there is a dead agent at row 5, col 7 belonging to player 0 2012-11-03T08:04:10 though now I realize that you should also be told the heading, so you can work out where it was last turn 2012-11-03T08:06:41 okay, i do like that the engine handles that for you 2012-11-03T08:07:02 my way would require the bots do collision detection 2012-11-03T08:09:05 smiley1983: let me know if you change the io at all :) 2012-11-03T08:09:15 i'm assuming d is remaining as is for tonight 2012-11-03T08:09:23 i think i'm almost done 2012-11-03T08:09:32 yes, I just did: d r c o -> d r c h o (dead row col heading owner) 2012-11-03T08:09:44 however, that's not on the server yet 2012-11-03T08:09:54 it's just been pushed to github 2012-11-03T08:11:31 it just compiled, so i'm going to feed it some sample output 2012-11-03T08:22:27 smiley1983: python tcpclient.py li414-97 2085 "python MyBot.py" username password [num_rounds] is that right? 2012-11-03T08:22:37 shoudl i just use the ip and 2086? 2012-11-03T08:27:33 *** smiley1983 has quit IRC (Ping timeout: 240 seconds) 2012-11-03T08:28:46 *** smiley1983 has joined #aichallenge 2012-11-03T08:28:55 port 2087 and it should work 2012-11-03T08:29:13 accursed internet connection 2012-11-03T08:29:41 i'm connected but failing to start :\ 2012-11-03T08:30:27 have you set it to loop a few times? Often the first connection attempt in a sequence will fail 2012-11-03T08:31:09 If it actually says "bot failed to start" or something like that, then it is failing to find your bot; might lack permissions, or want a complete path or something 2012-11-03T08:31:23 I'll paste the scripts I'm using to connect 2012-11-03T08:33:31 http://pastebin.com/EzqaE94A 2012-11-03T08:33:34 it seems to be playing 2012-11-03T08:34:08 Hooray! http://106.187.94.97:2086/replay.275 2012-11-03T08:34:38 which one am i? 2012-11-03T08:34:40 did i crash? 2012-11-03T08:35:10 In that replay, yes 2012-11-03T08:35:13 Green 2012-11-03T08:35:25 hmmm, there's a bug 2012-11-03T08:35:30 with me that is 2012-11-03T08:35:32 well i assume me 2012-11-03T08:35:54 k, stopped it for now 2012-11-03T08:36:20 You know you can test locally with the contents of the tron_dist/tron directory? 2012-11-03T08:36:26 play_one_game.sh and all that 2012-11-03T08:36:50 i'll worry about that later :P 2012-11-03T08:38:00 ah i see my problem 2012-11-03T08:38:09 you're given starting locations and walls in turn 0 now too 2012-11-03T08:38:29 Walls are only given in turn 0 2012-11-03T08:38:48 Now you mention it, perhaps they should be withheld until turn 1 2012-11-03T08:39:17 starting locations, too 2012-11-03T08:40:37 yes 2012-11-03T08:40:38 definitely 2012-11-03T08:40:53 otherwise people will use up that time for their first move 2012-11-03T08:43:38 an idea to consider: 2012-11-03T08:43:55 well not sure how to implement this actually 2012-11-03T08:44:23 but if you have many friendly snakes in the same blocked off region the first thing you'll do is kill off all but one of them 2012-11-03T08:44:38 choosing which to kill will be a challenge in itself 2012-11-03T08:48:35 Yes; also, currently there is a score penalty for losing snakes while opponents are still alive, so there may be circumstances where you cannot win, but you can elevate your rank above someone else by keeping two snakes alive for longer in a confined space 2012-11-03T08:52:37 Copy of engine which withholds walls and agents 'til turn 1 now available on github, will have it up on the tcp server in a minute 2012-11-03T08:56:44 smiley1983: i dunno if i like difficult scoring like that 2012-11-03T08:56:57 i prefer less fiddling around with crap like that :P 2012-11-03T08:57:04 is the dead thing updated on tcp now too? 2012-11-03T08:57:35 Yes 2012-11-03T08:58:16 The scoring isn't too hard to calculate; you just use the dead agent lines to add one point to every person who is not the owner, and subtract one from the owner 2012-11-03T08:58:51 It has the result that if you survive longest, you always win, and there is a means of differentiating between subsequent ranks in n-player games 2012-11-03T08:59:39 Though I realize I have to add the correction "add one point to evrey person who is not the owner and still has living agent(s)" 2012-11-03T09:00:25 TCP server now also withholds walls and starting locations until turn 1 2012-11-03T09:03:17 smiley1983: hmmm, i now wonder if water could be given first turn 2012-11-03T09:03:23 well 0th turn 2012-11-03T09:03:38 would allow for more setting up of the grid during turn 0 2012-11-03T09:03:51 then turn 1 you get the starting positions and actually start working out what move to make 2012-11-03T09:04:09 I thought about that, and the only argument I can find against is that in a contest, you would have to replace all maps at launch of contest 2012-11-03T09:04:23 or people will write code to recognize the maps and extrapolate 2012-11-03T09:05:27 you always want to do that anyway 2012-11-03T09:05:58 yes. If it's to be changed, it should be now. So; send walls on turn 0? 2012-11-03T09:06:08 yeah i think that's a good idea 2012-11-03T09:07:35 smiley1983: you could always add scores as an input parameter 2012-11-03T09:08:13 send scores to bots each turn? 2012-11-03T09:08:40 hmm i guess there's not much to infer each turn aye 2012-11-03T09:08:44 nm 2012-11-03T09:08:56 The server should now be sending walls on turn 0 again 2012-11-03T09:09:24 players could start on score n where n is the number of snakes they have? 2012-11-03T09:09:30 then update the scores as snakes die 2012-11-03T09:10:00 does that work? 2012-11-03T09:10:17 er... something is going wrong, now it sends the walls on turn 2. :/ 2012-11-03T09:10:21 lol 2012-11-03T09:10:40 Yes, that scoring system is correct 2012-11-03T09:11:08 i think the scoring is a little confusing when it requires that correction 2012-11-03T09:11:56 oh wait 2012-11-03T09:12:09 if you get eliminated you're equal last with current scoring righ? 2012-11-03T09:12:13 dunno if i like that 2012-11-03T09:13:03 so if one person remains at the end everyone else comes last? 2012-11-03T09:13:12 that's pretty crappy scoring if i'm interpresting that right :P 2012-11-03T09:15:30 What about your score increases by the turn number at which east snake dies? 2012-11-03T09:15:43 so the earlier each snake dies the lower the score that snake gets 2012-11-03T09:15:56 i think that would eliminate the need to suicide snakes too 2012-11-03T09:16:24 and if anything i think more snakes would be better for trying to fill in space 2012-11-03T09:16:48 your score would essentially be your wall space then too 2012-11-03T09:16:55 smiley1983: what do you think? 2012-11-03T09:17:15 not sure what to score when snakes collide with that 2012-11-03T09:17:34 i'd suggest 1 point per player per colliding square 2012-11-03T09:17:57 *** smiley1983 has quit IRC (Ping timeout: 264 seconds) 2012-11-03T09:18:44 *** smiley1983 has joined #aichallenge 2012-11-03T09:19:06 smiley1983: how much of what i just wrote did you get? 2012-11-03T09:19:37 going back through the log to find out; reposting stuff which didn't make it through: 2012-11-03T09:19:47 People should end up ranked in the order of elimination 2012-11-03T09:19:56 Three people, one agent each; A is eliminated first, 2012-11-03T09:20:00 dropping to 0 points and raising B and C to 2 each. Then B 2012-11-03T09:20:04 is eliminated, dropping to 1 point and raising C to 3 2012-11-03T09:20:40 so you get one point for every enemy snake that dies? 2012-11-03T09:21:07 and you lose one point if enemy snakes are alive 2012-11-03T09:21:21 i wonder if there's any bad situations with that 2012-11-03T09:21:40 i kind of like my idea of scoring people based on how many squares they turned into walls 2012-11-03T09:21:50 Assigning one point per square occupied could work; though it would look awkward if the winning player dies early, and the one who came second has plenty of space to run around in 2012-11-03T09:22:12 how would they be winning? 2012-11-03T09:22:24 last remaining player gets all free squares? 2012-11-03T09:22:29 or all accessible free squares? 2012-11-03T09:22:33 by filling in more space (having kept two agents alive while the second rank had only one) 2012-11-03T09:22:42 shared between remaining players if the turn limit is reached 2012-11-03T09:23:05 yeah okay 2012-11-03T09:23:27 yeah i am with the original one now provided no bad examples come up 2012-11-03T09:23:28 You don't lose a point if enemy snakes are alive; you always lose a point for one of your snakes dying; you gain a point if you are alive when an enemy snake dies 2012-11-03T09:24:08 *** Scooper has joined #aichallenge 2012-11-03T09:24:37 do we really want more than one snake per player? 2012-11-03T09:24:54 writing ai for that will be messy 2012-11-03T09:25:22 i guess it doesn't hurt to allow it 2012-11-03T09:25:52 indeed; if experimentation shows that it's bad, we can just get the map generator to make simple 1v1 maps 2012-11-03T09:26:20 so what are your thoughts on giving scores to players each turn? 2012-11-03T09:26:31 rather than having each person calculate them individually? 2012-11-03T09:26:54 i'd still want dead ant info though 2012-11-03T09:27:03 dead snake* 2012-11-03T09:27:54 rebooting server 2012-11-03T09:28:18 Giving scores would be fine; not too much bloat for the data stream. 2012-11-03T09:29:04 I probably won't add it tonight, since it doesn't make a difference for the core functionality of the bot 2012-11-03T09:29:35 yep 2012-11-03T09:29:39 i think my starter bot is done 2012-11-03T09:29:47 In fact, I'd be surprised if it would change any bot's decisions at all 2012-11-03T09:30:31 Right, server is back up, and is serving wall information on turn 0 2012-11-03T09:30:32 k, i'ma start on a proper bot 2012-11-03T09:31:00 If you want to paste the code somewhere I'll add it to the repo 2012-11-03T09:31:09 I mean for the starter 2012-11-03T09:31:23 let's wait a day or two just to see what else changes in the meantime 2012-11-03T09:31:29 then i'll zip it up 2012-11-03T09:31:44 will let me find bugs too aha 2012-11-03T09:37:30 smiley1983: trying the tcp client 2012-11-03T09:37:36 http 400 2012-11-03T09:39:49 it seems it complains about the password parameter 2012-11-03T09:40:01 Error code explanation: 400 = Bad request syntax or unsupported method. 2012-11-03T09:44:33 ramn[mba]: can you paste the script you're running somewhere? 2012-11-03T09:44:43 sure 2012-11-03T09:45:11 I don't have a script I'm just following http://106.187.94.97:2084/howto 2012-11-03T09:45:25 this command> 2012-11-03T09:45:26 python tcpclient.py li414-97 2081 "python MyBot.py" username password [num_rounds] 2012-11-03T09:46:34 python tcpclient.py 106.187.94.97 2084 'scala target/scala-2.9.0-1/HymenopteranOverlord.jar' user1 passwd1 2012-11-03T09:46:52 2084 should be 2085 2012-11-03T09:46:57 ah 2012-11-03T09:47:07 ah now i see 2012-11-03T09:47:08 Sorry, I need to modify the howto pages it seems 2012-11-03T09:47:11 the howto is wrong 2012-11-03T09:49:25 smiley1983: working now, yay 2012-11-03T09:49:39 *** loglog has quit IRC (Remote host closed the connection) 2012-11-03T09:49:53 *** loglog has joined #aichallenge 2012-11-03T09:49:56 Hooray :) 2012-11-03T09:49:59 whats the timeout 2012-11-03T09:50:39 ah it got printed out nvm 2012-11-03T09:50:45 500 2012-11-03T09:50:52 is that ms? 2012-11-03T09:51:10 so that is shorter than the actual tournament? 2012-11-03T09:52:23 I think my bot died horribly 2012-11-03T09:53:13 not updated on teh games board? 2012-11-03T09:54:43 ah I connected to the wrong one after all 2012-11-03T09:56:50 I'll look into setting up a heroku instance running a bot continuously 2012-11-03T09:57:06 so ppl don't have to worry keeping their bot online all teh time 2012-11-03T09:57:12 yes, 500ms 2012-11-03T09:57:19 The contest used 500ms in the end 2012-11-03T09:57:27 ah, why so low? 2012-11-03T09:57:31 the docs probably still say 1000ms 2012-11-03T09:58:11 hm... games were pretty slow, and bots were managing to do a lot with 500ms, so I think it was mostly for higher game turnover 2012-11-03T09:58:28 not really an issue over TCP 2012-11-03T09:58:41 About to restart the server, games will be disconnected... 2012-11-03T10:00:44 ah 2012-11-03T10:02:36 Server running games again 2012-11-03T10:03:38 i'm building myself a specialised starter 2012-11-03T10:12:51 *** g0llum has joined #aichallenge 2012-11-03T10:14:21 rebooting server again (sorry) 2012-11-03T10:16:22 running games again 2012-11-03T10:28:40 smiley1983: does cols come after rows now? 2012-11-03T10:29:09 if water is given during turn 0 you need to set up the grid before the wall info comes 2012-11-03T10:31:33 No, cols are still being issued first; I'll try to get that fixed in the next 10 min or so 2012-11-03T10:32:02 cool 2012-11-03T10:32:11 sorry to keep you up :P 2012-11-03T10:33:12 I still have time (begin work at 2pm) 2012-11-03T10:33:23 easy 2012-11-03T10:33:47 i'm reconsidering storing neighbour info 2012-11-03T10:34:06 as i need a new copy of the grid for each extendable node in the tree 2012-11-03T10:34:18 Any other sequence changes you think should be made, let me know now :) 2012-11-03T10:34:35 i think i'm good after that at the moment 2012-11-03T10:35:45 *** Eibwen has quit IRC (Read error: Connection reset by peer) 2012-11-03T10:36:16 *** Eibwen has joined #aichallenge 2012-11-03T10:44:38 Rebooting server, when it's back up it should have rows before cols 2012-11-03T10:46:38 online again 2012-11-03T10:48:45 confirmed, rows issued before cols 2012-11-03T10:49:36 The previous Python starter contained an error, and failed when the order was reversed. :) Anyone starting development should make sure they have the updated version from github. 2012-11-03T10:52:01 ramn[mba]: previously I accidentally connected the "solifugid_final" bot to the beginner server; the game you got there was not representative of a beginner opponent. Just thought I should let you know. 2012-11-03T10:52:20 ok :) 2012-11-03T11:02:29 *** UncleVasya has joined #aichallenge 2012-11-03T11:04:58 UncleVasya: I now have a TCP server running Ants and Tron 2012-11-03T11:05:57 smiley1983: It's so great for the community, thank you! 2012-11-03T11:06:13 I hope so, we'll see how much it gets used 2012-11-03T11:06:30 UncleVasya: have you started a tron bot yet? 2012-11-03T11:06:53 I've got one bot running on the server itself for each TCP server, so there should always be an opponent there 2012-11-03T11:07:05 No, I haven't even made an ocaml starter 2012-11-03T11:07:07 shame on me 2012-11-03T11:07:19 I made the Python one first 2012-11-03T11:07:52 actually I made an ocaml one which just said "go", but realized I had to get Python working so other people could run play_one_game.sh without installing ocaml 2012-11-03T11:08:34 antimatroidl1: oh, I don't even know when I'll start it. When I'll have a possibility to work on something new I think I'll try to do some stuff for the challenge first, and only then bots. 2012-11-03T11:12:16 smiley1983: do you know, does ramn or somebody already works on adapting PW? If not I'll try to do something today-tomorrow. 2012-11-03T11:13:09 *** smiley1983 has quit IRC (Ping timeout: 264 seconds) 2012-11-03T11:13:52 *** smiley1983 has joined #aichallenge 2012-11-03T11:14:47 I'm not working on pw! 2012-11-03T11:15:28 ramn[mba]: ok 2012-11-03T11:16:05 UncleVasya: I don't think anybody is even thinking about it except you and I, and I won't really think about doing something new until I'm a bit happier with Tron (visualizer improvements, testing and such) 2012-11-03T11:17:01 I have a long way to make my ant bot behave! 2012-11-03T11:17:09 just discovered the visualization api 2012-11-03T11:17:13 seems neat 2012-11-03T11:17:40 such a thing should really be in the game from start, much easier/more educational to work then 2012-11-03T11:18:08 do you mean you discovered the plain visualizer, or the upgraded one which lets you draw colored shapes on the replay? 2012-11-03T11:19:47 Having that from the start would certainly have made things easier! 2012-11-03T11:19:54 hm 2012-11-03T11:19:59 what is the plain visualizer 2012-11-03T11:20:18 ramn[mba]: heheh, I remember this long minutes when I was looking at the log.txt with drawed map via symbols :D Like B for border, E for enemy. 2012-11-03T11:20:18 This: http://106.187.94.97:2080/replay.75 2012-11-03T11:20:24 *these 2012-11-03T11:21:01 ah yes the replay I have discovered before, 2012-11-03T11:21:03 it is very nice! 2012-11-03T11:21:21 it's on the frontpage of aichallenge.org! really sells in the game 2012-11-03T11:21:37 was that what you meant by visualization api? Or something else? 2012-11-03T11:21:40 smiley1983: So, are you inviting my final version? :D 2012-11-03T11:21:49 something else 2012-11-03T11:22:01 there is special commands you can emit, 2012-11-03T11:22:02 On the Free for All server, yes :) 2012-11-03T11:22:07 then run a java client 2012-11-03T11:22:08 right? 2012-11-03T11:22:38 https://github.com/j-h-a/aichallenge/blob/vis_overlay/VIS_OVERLAY.md 2012-11-03T11:22:48 ramn[mba]: if it's the one I'm thinking of, it's Javascript like the one on the website - 2012-11-03T11:22:52 yes, that's the one! 2012-11-03T11:23:01