2010-12-13T00:04:54 *** Rubicon-|-Cross has joined #aichallenge 2010-12-13T00:11:31 *** Appleman1234 has quit IRC (Ping timeout: 240 seconds) 2010-12-13T00:16:56 *** davidjliu has quit IRC (Quit: Page closed) 2010-12-13T00:22:25 *** Utkarsh has quit IRC (Ping timeout: 276 seconds) 2010-12-13T00:26:50 *** Utkarsh has joined #aichallenge 2010-12-13T00:28:19 *** fmeyer has joined #aichallenge 2010-12-13T00:35:07 i'm not sure the rotation of the board is necessary 2010-12-13T00:35:13 (i'm back) 2010-12-13T00:36:29 I meant rotation as a symettry 2010-12-13T00:38:50 yeah okay 2010-12-13T00:38:57 so i can remove the rotation at the end? 2010-12-13T00:39:07 i'm going to write that function to scan linear lines between homes 2010-12-13T00:39:09 sure 2010-12-13T00:39:19 what for? 2010-12-13T00:39:32 well, diagonal, vertical and horizontal 2010-12-13T00:39:37 to see how often it aligns 2010-12-13T00:39:38 ah yup 2010-12-13T00:39:46 we only really care if it aligns all the time 2010-12-13T00:39:58 obviously aligning at random times is fine ;P 2010-12-13T00:40:20 but yeah, that function is good... saves having to think through the math 2010-12-13T00:41:03 wait, do i want to rotate the board randomly up to 3 times? 2010-12-13T00:41:09 o_O 2010-12-13T00:41:29 bah, i'm confused, i'll play around :) 2010-12-13T00:41:56 rotations lie in a group so once should be enough :P 2010-12-13T00:42:14 *** delt0r has quit IRC (Ping timeout: 240 seconds) 2010-12-13T00:42:36 *** delt0r has joined #aichallenge 2010-12-13T00:46:32 argh, I was going to point you at the good tron map generator http://www.ai-contest.com/forums/viewtopic.php?f=10&t=71 2010-12-13T00:46:45 but apparently the file has been removed from the server 2010-12-13T00:47:36 also it looks like none of them were actually used in the final contest so I can't point you at examples either :( 2010-12-13T00:48:07 *** Appleman1234 has joined #aichallenge 2010-12-13T00:48:33 janzert: what was special about the tron generator? 2010-12-13T00:49:24 yeah? 2010-12-13T00:49:31 it didn't have wrapped walls either 2010-12-13T00:49:48 one thing is in addition to the "random" wall maps like anti's current generator makes it also made maze like maps and maps with long straight walls randomly laid out 2010-12-13T00:49:59 antimatroid1: right but it should be easy to adapt that 2010-12-13T00:50:06 ah 2010-12-13T00:50:25 that's easy to do 2010-12-13T00:50:33 I was going to suggest looking at a maze generator, but I think this random approach gives a more natural feels 2010-12-13T00:50:38 antimatroid1: I think maybe your current requirement that *every* non-wall cell be reachable may be too strong 2010-12-13T00:50:39 *feel 2010-12-13T00:50:52 janzert: what do you mean? 2010-12-13T00:51:09 sure, it should all be easy to redo but looking at the other one might have been useful for you 2010-12-13T00:51:10 in fact, you can run into every wall if you want 2010-12-13T00:51:42 it should be fine to generate maps where sections are unreachable so long as the two bases are still connected 2010-12-13T00:52:00 but then it's a waste of space? 2010-12-13T00:52:08 might want to make sure only a certain percentage is unreachable of course 2010-12-13T00:52:13 janzert: thats what he originally had :P 2010-12-13T00:52:33 *** Appleman1234 has quit IRC (Ping timeout: 240 seconds) 2010-12-13T00:52:38 give me a sec 2010-12-13T00:52:47 I think it's fine if say up to 10-15% is unreachable 2010-12-13T00:53:36 doesn't look like I even have any of the good maps saved locally 2010-12-13T00:53:52 from tron that is 2010-12-13T00:54:04 *** Appleman1234 has joined #aichallenge 2010-12-13T00:54:07 that's an easy change anyway.... change line 89 to test: squaresVisited == int(dim*dim*min_reachable_factor) 2010-12-13T00:54:43 it's not the simple change i thought, cbf atm 2010-12-13T00:55:01 what I said is not easy 2010-12-13T00:55:02 ? 2010-12-13T00:55:08 hang on, just saw that 2010-12-13T00:55:23 change == to > 2010-12-13T00:55:41 err... >= 2010-12-13T00:55:50 oh, yeah, i also need to check home bases are connected 2010-12-13T00:56:06 that's also easy 2010-12-13T00:56:15 make sure home2 is painted err 2010-12-13T00:56:17 okay, hold up 2010-12-13T00:57:30 squaresVisited >= int(dim*dim*min_reachable_percentage/100) and nMap[home2[0] ][home2[1] ] 2010-12-13T00:57:46 that should work 2010-12-13T00:59:24 * sigh codes by randomly typing characters, then slowly mutating the string until it converges to a solution 2010-12-13T00:59:49 oh, it does, i need to paint the closed off regions after adding walls now though 2010-12-13T00:59:57 cause resources get put in closed off regions 2010-12-13T01:00:40 ants see through walls right? 2010-12-13T01:01:00 i.e. visions is strictly limited by distance 2010-12-13T01:01:09 *** delt0r has quit IRC (Read error: Operation timed out) 2010-12-13T01:01:21 haha... I had an idea... put nice resources in the closed off areas and allow people to dig 2010-12-13T01:01:24 http://codepad.org/9BKCYLIz 2010-12-13T01:01:39 janzert: i wanted to have them not see through walls, delt0r thinks it's better to 2010-12-13T01:01:42 what are your thoughts? 2010-12-13T01:02:02 it's not hard to paint out each players visible region with walls blocking sight and returning what they can see 2010-12-13T01:02:06 if ants can see through I would go ahead and put resources in the closed off region so ants need to get smart enough to know when the resource is unreachable 2010-12-13T01:02:32 hmm, not really sure which would be more interesting 2010-12-13T01:02:34 yeah, actually that could make it cool, i'll leave it like that 2010-12-13T01:03:04 see through is probably a lot easier to deal with all around 2010-12-13T01:03:45 antimatroid1: your comments are out of date again :P 2010-12-13T01:03:53 sigh: i'll fix them later 2010-12-13T01:03:58 *** delt0r has joined #aichallenge 2010-12-13T01:03:58 :) 2010-12-13T01:04:06 what do you think should happen when opponent ants try to move into the same square? 2010-12-13T01:04:12 delt0r and i don't agree on that one 2010-12-13T01:04:31 he wants the engine to resolve those moves and have neither ant move, I say let them fight each other to death there and then 2010-12-13T01:04:36 I think it should segfault and crash the server 2010-12-13T01:05:19 I think a lot of that really should be resolved with playtesting :) 2010-12-13T01:05:57 that's the plan, but i think if people have to consider resolving moves that may not go through, it becomes a lot harder 2010-12-13T01:07:25 I rather like fight to the death, but it might have unforeseen problems. It's really hard to get a good feel for the dynamics without seeing games played :) 2010-12-13T01:07:46 i don't want it to be the only method of fighting, otherwise i don't think many fights will occur 2010-12-13T01:08:21 yeah, I don't really like bullets but there needs to be some method of ranged damage I think 2010-12-13T01:08:28 maybe call it acid spit :P 2010-12-13T01:08:34 it would be good if the game dynamics went along the lines of start: collect resources and build up grains at home base, middle: go explore the map for them and scope things out end: epic battle while trying to charge each others home base 2010-12-13T01:10:12 acid spit is cooler than bullets, another option is you can "hit" another ant that can be reached in x moves 2010-12-13T01:10:22 which gets processed before moves 2010-12-13T01:10:39 with acid spit 2010-12-13T01:10:45 or maybe ranged attack shouldn't just travel in straight lines but be more like a mortar lob that can go to any destination within a certain range and cause a small radius of damage 2010-12-13T01:11:18 i like that 2010-12-13T01:11:24 you give it the distance it travels 2010-12-13T01:11:29 and it has a set blast radius 2010-12-13T01:11:37 but I'll let you guys work it all out, I really should stay out of this one :/ 2010-12-13T01:11:39 and distance is capped 2010-12-13T01:12:09 Wibble 2010-12-13T01:12:15 and you get hp of 5-distance to centre of blast hp taken off 2010-12-13T01:12:53 wibble? 2010-12-13T01:13:32 sigh: you don't watch blackadder? 2010-12-13T01:13:43 I though you where from down under 2010-12-13T01:14:10 I haven't seen all of it :P 2010-12-13T01:14:46 well the last episode of blackadder goes forth has the wibble reference... 2010-12-13T01:14:53 I let you you tube 2010-12-13T01:14:55 it 2010-12-13T01:15:30 antimatroid1: it sounds like you really just want a RTS rather than "army ants" 2010-12-13T01:15:49 isn't this meant to be an rts like game? 2010-12-13T01:16:03 i really don't like the cancelling of moves 2010-12-13T01:16:10 that makes it way harder to work out what moves you can make 2010-12-13T01:18:28 delt0r_: I'm not overly set on "firing" stuff, just "attack" ant in next square is fine, but i really think oppoents should be able to land in the same square and one (or both) dies off 2010-12-13T01:19:19 stacking always results in the same tatics in all RTS games that have it. Stack when ever you can --always 2010-12-13T01:20:01 while no stacking permit more interesting formations and less "logic" (ie 3 way + n way stacks from both temes) 2010-12-13T01:20:03 teams 2010-12-13T01:20:47 sigh: i've actually kind of warmed up to krudd since he got the boot, you? 2010-12-13T01:21:09 delt0r_: I expect it to be fairly easy to avoid stacking 2010-12-13T01:21:23 although if large groups come across each other, than can charge into battle which will be over faster 2010-12-13T01:21:25 antimatroid1: well he's saying stuff that makes sense now 2010-12-13T01:21:33 and not running away from the media 2010-12-13T01:21:35 sigh: even before this last week 2010-12-13T01:22:02 honestly haven't been paying attention to him in the meantime 2010-12-13T01:27:33 does anything need to be passed at the very start of the game other than map dimensions, turns in a game and maxt time per turn? 2010-12-13T01:27:57 *** Azrathud has quit IRC (Quit: Leaving) 2010-12-13T01:29:18 vision radius would be useful 2010-12-13T01:29:32 yep 2010-12-13T01:30:23 antimatroid1: But if you permit stacking you can't stop it... so i stack *all* my ants when ever i can 2010-12-13T01:30:42 It will always be the better strat... we have seen this many times in other RTS 2010-12-13T01:30:57 delt0r_: i'm saying make two friend ants stacked be an invalid move and loss of game 2010-12-13T01:31:13 if two opponent ants land on the same square, only one (if any) can be there after they battle 2010-12-13T01:31:23 well now you have ---stacking kinda with extra rules and logic ... for what? 2010-12-13T01:31:28 what does it add? 2010-12-13T01:31:40 it means you don't have to worry about your moves not actually being implemented 2010-12-13T01:31:49 ?? 2010-12-13T01:32:08 if two oppoent ants try to move into the same square, you want that to not count for either of them? 2010-12-13T01:32:21 but then your ant isn't goint to be where it was expected if it's still alive 2010-12-13T01:32:23 yea 2010-12-13T01:32:29 so 2010-12-13T01:32:32 which is also more work to handle when working out possible moves 2010-12-13T01:32:40 it's easier if you let them land there and fight to death 2010-12-13T01:32:56 if there are *any* invalid moves... then this is something you need *anyway* 2010-12-13T01:33:18 no, you just never move one of your ants to where another one is 2010-12-13T01:33:18 and if two freindlys can't stack then that situation happens anyway 2010-12-13T01:33:20 that's easy 2010-12-13T01:33:41 and then you may move an ant to any empty square, and provided it's stil alive next turn, it will be there 2010-12-13T01:33:49 really? 2010-12-13T01:34:02 what's complicated? 2010-12-13T01:34:11 you can only ever have two opponent ants land in the same square 2010-12-13T01:34:14 (0,0)-> 0,1 <-(0,2) 2010-12-13T01:34:43 ops these two ants did not communicate that they both planed to move to the same square 2010-12-13T01:34:47 now what happens 2010-12-13T01:34:54 same team or not? 2010-12-13T01:34:55 these are on the same team 2010-12-13T01:34:57 they lose 2010-12-13T01:35:04 ?? WTF 2010-12-13T01:35:05 if all ants are the one process, they all share the same information 2010-12-13T01:35:32 it shouldn't be hard to make sure you never move where another one of your ants is going to be 2010-12-13T01:35:40 no--they only see local info... ant A doesn't know anthing about ant B without some serious pattern matching 2010-12-13T01:35:40 *** sigh has quit IRC (Ping timeout: 245 seconds) 2010-12-13T01:35:55 delt0r_: how? 2010-12-13T01:36:01 each ant only "sees" a friendly... it doesn't know which "ant: that is 2010-12-13T01:36:37 if i am using one process with all information going to there, all my ants know the same thing 2010-12-13T01:36:47 you cannot stop people from doing that 2010-12-13T01:36:50 Ant A sees: Ant (0, 2) team=1 2010-12-13T01:37:06 it needs to know its hp and what it can see 2010-12-13T01:37:10 ant B sees: Ant (0,-2) team=1 2010-12-13T01:37:18 only its HP 2010-12-13T01:37:22 not other ants. 2010-12-13T01:37:30 if ants start from the home base 2010-12-13T01:37:34 they all know where they are 2010-12-13T01:37:35 I don't even give id;s --but that could be added 2010-12-13T01:37:57 good luck with you starter packs... 2010-12-13T01:38:10 it's easy :\ 2010-12-13T01:38:23 prediction... bottom 2/3 make illegal moves and kill 1/2 there own ants 2010-12-13T01:38:28 yea right 2010-12-13T01:39:01 not really, make a vector of ants, go through and make a random valid move for each one (random bot) and return the moves 2010-12-13T01:39:03 delt0r_: I think you're imagining the information being given to the ants in a completely different way that antimatroid1 is doing it 2010-12-13T01:39:36 none of the information is being given from an ants point of view for his proposal 2010-12-13T01:39:37 the idea was "agents" --ie close to the ant per process 2010-12-13T01:39:53 but in one process for implentation reasons 2010-12-13T01:40:08 so you want to pass each ant relative information? 2010-12-13T01:40:14 *** tapwater has quit IRC (Quit: tapwater) 2010-12-13T01:40:23 bbs 2010-12-13T01:40:26 however its still very abitary to have "freindies" move onto the same square as illegal but not foes 2010-12-13T01:40:44 what if 2 guys try to attack the same ant... 2010-12-13T01:40:56 either let stacking happen --or don't 2010-12-13T01:41:05 but a mix is worse that either 2010-12-13T01:41:18 antimatroid1: I do only pass relative information 2010-12-13T01:41:20 you could instead just always have just the strongest ant survive no matter if they are same team or not :) 2010-12-13T01:41:59 *** fmeyer has quit IRC (Ping timeout: 265 seconds) 2010-12-13T01:42:13 still seems very lame... "locked" combat is very artifical... i can't retreat or break off the attack... it just doesn't seem to lead to anything interesting 2010-12-13T01:43:07 yeah, there should be a separate /normal/ way of attacking 2010-12-13T01:45:25 anyway I have a first cut ready to show people... 2010-12-13T01:45:45 its sloppy code... but at least we can "see" how it works... 2010-12-13T01:48:42 delt0r: if two ants attack the same enemy, it loses that many hp, if it goes below 0, the ant dies 2010-12-13T01:49:31 you can't do relative information with the home base idea 2010-12-13T01:49:46 i originally wanted to do that but threw it out when i liked home base 2010-12-13T01:50:39 but since they are on the same sqaure--you now have 2 friendlies on the same square 2010-12-13T01:50:55 its just seems very inconsitant... 2010-12-13T01:51:07 no, i'm saying don't let two friends go on the same square 2010-12-13T01:51:22 *** virdo has quit IRC (Ping timeout: 260 seconds) 2010-12-13T01:51:35 you could do that with ides 2010-12-13T01:51:37 antimatroid1: relative info is easy... "collision" stuff is also easy 2010-12-13T01:51:38 ids* 2010-12-13T01:51:39 and relative 2010-12-13T01:52:28 how are you having new ants made atm? and what's the decision rule for where they spawn? 2010-12-13T01:52:30 but that will kill friendlys.. seriously.. if I can share space with the enemny why not a friend... its not logical 2010-12-13T01:52:51 just spawn everything randomly... good for basic debug 2010-12-13T01:53:12 and you get an idea without too many tedious details 2010-12-13T01:53:58 when do you get a new ant? 2010-12-13T01:56:03 A new ant is "born" in the same place as the parent. Its an "egg" till the parent leaves the square. Then it spaws.. 2010-12-13T01:56:16 its a hack, but can be easily changed 2010-12-13T01:56:24 that works 2010-12-13T01:56:25 permits easy swaming in random maps 2010-12-13T01:56:36 so you can see dynamics better 2010-12-13T01:56:47 its not bad... very "alife" like 2010-12-13T01:57:15 you can see stratagies that could work well... and see what does not 2010-12-13T01:57:28 my current bot ignores walls 2010-12-13T01:57:31 i really like the home base idea though 2010-12-13T01:57:36 yea 2010-12-13T01:57:38 easy to add 2010-12-13T01:57:56 don't thinks it changes the basic dynamics... 2010-12-13T01:57:57 not really, if ants spawn there, they all know the same thing 2010-12-13T01:58:11 just adding some comments to the code... then will post 2010-12-13T01:58:15 or do you want ants to not know where their own home is? 2010-12-13T01:58:28 no.. thats fine 2010-12-13T01:58:39 but tracking state is not as easy as it sounds... 2010-12-13T01:58:59 currently ant "ids" are only valid for a round... 2010-12-13T01:59:10 so you can't easily know if its the same ant 2010-12-13T01:59:18 may fix that before release 2010-12-13T01:59:21 pattern match the walls, easy 2010-12-13T01:59:48 well not really... we don't want the starter packs to be a lot of code --just to get started really 2010-12-13T02:00:01 yeah 2010-12-13T02:00:03 that way more langs get added faster 2010-12-13T02:00:15 A* seems like it would be needed 2010-12-13T02:00:24 with walls that is 2010-12-13T02:00:37 i don't think one will get away with just one tree 2010-12-13T02:01:10 well if an ant can only see say 10x10... then you only need to do the path finding on that 2010-12-13T02:01:26 hell you only really need the "next move" part 2010-12-13T02:01:26 that's easy, paint the grid 2010-12-13T02:01:32 yea 2010-12-13T02:05:37 i don't see how you can have relative information be anything other than a tedious annoyance with extra overhead with home bases 2010-12-13T02:05:48 ?? 2010-12-13T02:06:00 you just don't give coords. 2010-12-13T02:06:19 well, each bot is just going to work out where they are relative to home base, then translate everything to the same coordinates 2010-12-13T02:06:37 unless 2010-12-13T02:06:40 you rotate the maps too 2010-12-13T02:07:11 well it stops even being "agent" relealted if you give absolute data 2010-12-13T02:07:42 yeah, my point is i think it's impossible to do relative and home bases 2010-12-13T02:12:56 I don't 2010-12-13T02:13:09 just don't give coords... whats the problem? 2010-12-13T02:15:04 *** mega1 has joined #aichallenge 2010-12-13T02:15:04 how do they know where home is without coords? 2010-12-13T02:15:12 where is mcleopold these days 2010-12-13T02:15:28 it would be cool to get him to add revised revisions of an engine to a tcp 2010-12-13T02:15:30 thats the point... thats what the "agent" code need to do 2010-12-13T02:26:15 okay, i'm with you, i say have the home hole, but ants get spawned randomly and given a relative coord of home base and spawned ants need to be at least a certain distance from the enemies home base 2010-12-13T02:26:25 i can make the map generator for that quite easily 2010-12-13T02:26:55 just wait to you see all this in action. 2010-12-13T02:27:08 you may not like it anyway... 2010-12-13T02:27:14 or others may not 2010-12-13T02:27:18 we will see 2010-12-13T02:27:20 if the ants are given relative coords to the home base that simply becomes the origin for the sides global coord system 2010-12-13T02:27:26 or folks can hack the code 2010-12-13T02:27:54 janzert: yes.. If you can see the base --you see the base... otherwise you know nothing... 2010-12-13T02:28:14 janzert: yes, but you don't necessarily know the path there 2010-12-13T02:28:22 so your ants at least have to find each other 2010-12-13T02:28:26 note that in my current implentation... you can't really "know" that you moved or not wihout checking the terrain 2010-12-13T02:28:59 yeah, home base completely throws out relative coords 2010-12-13T02:29:17 janzert: one option is to make it harder to match by rotating bots relative view 2010-12-13T02:29:50 and inverting etc. then reverting those when receiving its move 2010-12-13T02:30:13 why bother? it seems like it only makes an artificial barrier for newbies 2010-12-13T02:30:42 you could be right janzert 2010-12-13T02:30:54 giving coords does make it easier 2010-12-13T02:31:00 and we want easier 2010-12-13T02:31:19 but i remained unconvinced ants is the best idea... 2010-12-13T02:32:21 so agents are incompatible with home bases 2010-12-13T02:32:40 *** sigh has joined #aichallenge 2010-12-13T02:32:55 sigh: what do you prefer, individual agent ants or home bases 2010-12-13T02:33:00 But can't think of anything else right now 2010-12-13T02:33:21 independent agents doesn't seem feasible 2010-12-13T02:33:26 it is 2010-12-13T02:33:32 with single process 2010-12-13T02:33:42 each ant gets relative information each turn 2010-12-13T02:33:49 with them centred in a grid 2010-12-13T02:34:10 if we wanted to do an optimization/puzzle type contest instead of a head to head game sometime I did think one that might be interesting the other night, basically something based off of manufactoria i.e. basically 2d compiler writing 2010-12-13T02:34:13 http://jayisgames.com/games/manufactoria/ 2010-12-13T02:34:38 janzert: actual games are what makes it so interesting :P 2010-12-13T02:35:41 yeah, I don't expect people to jump on it but I do think it could be fun 2010-12-13T02:36:01 single process? so ants can share data? 2010-12-13T02:36:13 yes, but don't know which ones which 2010-12-13T02:36:32 janzert: flash fails for me on that one 2010-12-13T02:36:47 *** sigh has quit IRC (Remote host closed the connection) 2010-12-13T02:37:12 if you give them a relative location of home, then they can just tranlate to that as a common square 2010-12-13T02:37:20 sigh: its really not pratical to have 10-40 odd process per game... and when you want 100 games a min in the server 2010-12-13T02:37:31 practical 2010-12-13T02:37:54 *** sigh has joined #aichallenge 2010-12-13T02:37:57 antimatroid1: what do you mean you don't know which one's which? 2010-12-13T02:38:04 (6:36:46 PM) antimatroid1: if you give them a relative location of home, then they can just tranlate to that as a common square 2010-12-13T02:38:04 (6:36:54 PM) delt0r_: sigh: its really not pratical to have 10-40 odd process per game... and when you want 100 games a min in the server 2010-12-13T02:38:04 (6:37:05 PM) delt0r_: practical 2010-12-13T02:38:10 *** Palmik has joined #aichallenge 2010-12-13T02:38:45 sigh: you will just be passed information for "you have an ant, here's what it can see in it's field of vision" 2010-12-13T02:39:12 thats all i send... 2010-12-13T02:39:15 if they are randomly spawned, then they don't know where they are relative to anyone else 2010-12-13T02:39:47 what if we had 2 home bases per team? 2010-12-13T02:39:53 ah, ok... so you debating giving position relative to ant vs relative to home base? 2010-12-13T02:39:55 and each ant was randomly paired to one of the home bases 2010-12-13T02:40:15 *** Utkarsh has quit IRC (Ping timeout: 260 seconds) 2010-12-13T02:40:21 sigh: you can't have relative positions, single process and home bases, it doesn't work 2010-12-13T02:41:35 ok, I'm confused... I'll go read the forum to get familiar with your terminology 2010-12-13T02:41:42 You guys discussing new challenge? :) 2010-12-13T02:41:44 this isn't in the forums 2010-12-13T02:41:51 Palmik: not necessarily 2010-12-13T02:41:56 it's an option for the new challenge 2010-12-13T02:42:09 I see :) 2010-12-13T02:42:39 antimatroid1: ok, can you tell me clearly the difference bw individual agent ants and home bases 2010-12-13T02:42:44 pretend I don't know anything 2010-12-13T02:43:24 I have a feeling it doesn't matter if you have home bases or not, each ant will just build an individual map and when it comes within sight of another friendly merge maps 2010-12-13T02:43:41 janzert: i agree 2010-12-13T02:43:50 hence why it might just be easier to drop agents 2010-12-13T02:43:54 yay, I'm not needed 2010-12-13T02:44:01 sigh: hang on 2010-12-13T02:44:02 ants don't really know which one is which if they are fed relative information about their position, so they are sort of like individual agents, but when 2010-12-13T02:44:27 they are given a home base, they can always fill information in on the same map 2010-12-13T02:44:37 before they don't know how far away they are from each other 2010-12-13T02:45:34 ok, so the purpose of the home base is that it gives the ants a common origin? 2010-12-13T02:46:26 *** Utkarsh has joined #aichallenge 2010-12-13T02:46:26 hmmm, it's sort of something you need to protct now, it's not anything near as important as it was 2010-12-13T02:46:45 i prefer my idea, but i'm explaining delt0r's 2010-12-13T02:46:47 which works 2010-12-13T02:46:59 *** iNtERrUpT has joined #aichallenge 2010-12-13T02:47:09 ok, so you are deciding the objective of the game? 2010-12-13T02:47:23 all rules do 2010-12-13T02:47:29 the game is constantly changing at this point 2010-12-13T02:47:40 it's a matter of finding what will result in the most interesting competition 2010-12-13T02:48:20 it needs to be easy enough that people can pick up the rules fairly quickly 2010-12-13T02:48:21 meh... I'm still confused about what you are asking 2010-12-13T02:48:48 I don't know if you are discussing what ants see, or how they communicate, or how they win 2010-12-13T02:48:58 I'll just lurk a bit 2010-12-13T02:49:53 i like having the ants spawn from home and having a collective intelligence 2010-12-13T02:50:18 the other option requires ants to build their own information and match it up with other friend ants when they see them 2010-12-13T02:50:47 the second option just doesn't work with home bases, so it's a forced choice 2010-12-13T02:51:46 with the first option, you may have two ants just outside each others field of vision, and they'd know, with the second option, they wouldn't be able to distinguish between that and being complete opposite sides of the board 2010-12-13T02:52:33 so, what I said before: the purpose of the home base is that it gives the ants a common origin? 2010-12-13T02:53:07 it has more purposes, with the first option, ants are spawned there and you have to take harvest back there, with the second option you just have to make sure an enemy ant doesn't get on it 2010-12-13T02:53:24 ok, i see 2010-12-13T02:54:07 i see a lot of pattern matching being done with the second option instead of implementing cool strategies 2010-12-13T02:54:20 yeah 2010-12-13T02:54:45 also, the first 30 moves of the game will just be bumbling around 2010-12-13T02:54:56 and once you see a friend ant, you will stay together 2010-12-13T02:55:28 both can be tried though 2010-12-13T02:55:37 we need mcleopold for a tcp server once an engine is made 2010-12-13T02:55:40 *** Zannick has quit IRC (Ping timeout: 250 seconds) 2010-12-13T02:55:40 *** perror has joined #aichallenge 2010-12-13T02:55:46 :) 2010-12-13T02:56:24 it seem like individual ants would just result in a tedious initial coordination phase 2010-12-13T02:56:39 antimatroid1: we are a long way from having an engine... i like amstan idea to use JSON 2010-12-13T02:57:04 json for what? everything? 2010-12-13T02:57:14 except then you need json libraries for all languages 2010-12-13T02:57:16 i don't like that :\ 2010-12-13T02:57:23 not every language has json in the stdlib like python :P 2010-12-13T02:57:37 so there need to be a more "formal" spec at some point.. delegated to the organizes who can make executive decisions 2010-12-13T02:57:42 it's good being able to write a bot without any additional libraries 2010-12-13T02:58:04 I think json for the game output for visualiser is a good idea though 2010-12-13T02:58:13 its better to be able to change the format a little without breaking backward compatablity 2010-12-13T02:58:31 anyway s-expresions or JSON is easy to parse without a lib too 2010-12-13T02:58:58 The idea is the server is going to be better set up to support libarays 2010-12-13T02:59:29 i still don't see any reason to move away from standar input/output for communictation with the engine 2010-12-13T02:59:34 standard* 2010-12-13T02:59:47 if you pass game parameters at the start and hardcode that into starter packages 2010-12-13T02:59:52 they can be adjusted without people bitching 2010-12-13T03:00:18 you still do that 2010-12-13T03:00:27 json is orthogonal to stdin/out 2010-12-13T03:00:29 just with more formally formatted strings 2010-12-13T03:00:55 i just throw all my input into a string stream 2010-12-13T03:01:07 and then read it into whatever variables i like 2010-12-13T03:01:09 even if its not JSON (a good idea really) more structure would be good 2010-12-13T03:01:19 my format is as hacky as hell 2010-12-13T03:01:20 as long as i can continue doing that, i'm happy 2010-12-13T03:02:16 I like the idea of JSON but I don't think it will be nice for the other languages 2010-12-13T03:02:48 key value pairs with unknown keys specified as being ignored is probably sufficient for our purposes and a whole lot easier to parse than json 2010-12-13T03:03:08 agree ^^ 2010-12-13T03:03:33 *** Zannick has joined #aichallenge 2010-12-13T03:03:33 i had a look .. Every lang i have heard of has a JSON lib for it 2010-12-13T03:03:44 including haskell 2010-12-13T03:03:58 scheme, lisp 2010-12-13T03:04:06 ok --so not ook! 2010-12-13T03:04:25 but anyone using ook should be able to write their own 2010-12-13T03:05:00 *** Mathnerd314 has quit IRC (Quit: ChatZilla 0.9.86-rdmsoft [XULRunner 1.9.2.12/20101026210630]) 2010-12-13T03:05:16 but it does mean an extra library everyone has to get working (including us on the servers and we better pick the library everyone likes for that language :}) 2010-12-13T03:06:06 janzert: I understood that the server this time round will support general user libs better.. But yes this was my objection too... 2010-12-13T03:06:21 not all langs are late binding like java... 2010-12-13T03:07:01 and C/C++ compiler options are not always straightforward... 2010-12-13T03:07:11