2011-03-10T00:00:00 at some point 2011-03-10T00:00:02 huh? 2011-03-10T00:00:03 yeah... 2011-03-10T00:00:17 why not just make gameLoop, update, and finishTurn all the same thing? 2011-03-10T00:00:39 because it is good practice to make function do a single thing 2011-03-10T00:00:54 it would be easier for a beginner to understand the different parts 2011-03-10T00:00:57 well, mine has all that stuff separate, but it's still just one function call 2011-03-10T00:01:55 I'm thinking that if we can make all the starters more consistent, it will be easier to help people with questions about them 2011-03-10T00:01:59 like... you have a few monad transformers that have different parts of the functionality and they all implement the same interface. if you compose them then you get the complete functionality, but you can also stick your own components in to do additional or different things 2011-03-10T00:02:21 haskell is going to differ the most, I think 2011-03-10T00:02:41 i don't even think this naming scheme quite works for the library i'm making 2011-03-10T00:02:49 which is why if we can agree on a good spec for the starter bots, then we've covered most of the languages 2011-03-10T00:02:50 i could just try to match as close as i can 2011-03-10T00:02:56 hmm 2011-03-10T00:03:42 I didn't even mention separating a library from the bot logic, I should have 2011-03-10T00:07:26 well, the library i'm doing has different components for getting and parsing input, for updating seen water, for handling output, for handling static information about the game, and for managing turn-by-turn information and aborting the game loop cleanly while providing meaningful debugging information 2011-03-10T00:07:46 they are intended to be composed with each other, and they all contribute to a few common interfaces 2011-03-10T00:08:21 and i don't think it's really the kind of thing most starter writers will want to do for most languages 2011-03-10T00:09:50 ok 2011-03-10T00:10:18 jmcarthur: " are we wanting the starters to track water at all?" well yeah, you only get told about water once :P 2011-03-10T00:10:58 let me put in some info on the wiki about what should be tracked in the game state 2011-03-10T00:11:09 i am just tracking water 2011-03-10T00:11:09 antimatroid: i mean as opposed to making the developers handle that themselves 2011-03-10T00:11:34 i think we should at least track water in the starter bots 2011-03-10T00:11:54 my water tracking is optional/extensible 2011-03-10T00:12:22 so if somebody comes along and "knows better" they don't even suffer any performance penalty or boilerplate 2011-03-10T00:13:43 i expect anyone who cares about performance to write their own bot from scratch 2011-03-10T00:14:09 McLeopold: "Depending on the order of the ants, you may get different results with the same configuration of ants." <-- does that imply that we should be avoiding sending ants to collide with each other in the starters too? 2011-03-10T00:14:19 i think the objective of the starter bots should be to be as helpful as possible for beginners who may not be able to do their own rewrite 2011-03-10T00:14:20 antimatroid: i consider that a waste 2011-03-10T00:14:21 antimatroid: jmcarthur wants to be able to swap out pieces. 2011-03-10T00:14:38 jmcarthur: you simply can't release something that will work the way everyone wants 2011-03-10T00:14:41 antimatroid: That way it's not al or nothing = huge skill gap. 2011-03-10T00:14:48 antimatroid: you aren't used to good languages. shut up :) 2011-03-10T00:14:52 :P 2011-03-10T00:16:06 antimatroid: "i think the objective of the starter bots should be to be as helpful as possible for beginners who may not be able to do their own rewrite" <-- that is also one of my objectives 2011-03-10T00:16:38 jmcarthur: yes, avoid sending ants to the same square, but do it naively. 2011-03-10T00:16:51 but i'd like to do it without alienating the majority of developers who know what they are doing 2011-03-10T00:17:00 that is my objective, with a side note of being reasonably efficient 2011-03-10T00:17:31 i think people could re use most of my code if they wanted, but i imagine a lot of people will just write their own for c++ 2011-03-10T00:17:48 McLeopold: how naively? 2011-03-10T00:17:56 it should also showcase good design and best practices for the language 2011-03-10T00:18:09 McLeopold: we are avoiding collisions now? 2011-03-10T00:18:45 jmcarthur: 1 at a time, don't move on an ant, whether it has been given an order or not. 2011-03-10T00:18:55 i was considering making a component to somewhat intelligently avoid collisions, but that could be overstepping what a starter should be 2011-03-10T00:18:58 so, that means updating the state in the issue_order function 2011-03-10T00:19:30 McLeopold: so... the starter should be tracking the new positions of ants as it issues orders? doesn't sound very naive to me 2011-03-10T00:19:50 well, not starter-level naive 2011-03-10T00:20:23 Then I'll make it optional 2011-03-10T00:20:24 *** chemuduguntar has quit IRC (Ping timeout: 276 seconds) 2011-03-10T00:20:35 i was thinking "naive" might mean "don't move an ant to a position where another ant is already, regardless of whether it will be moving 2011-03-10T00:20:51 so if you have AA. and try to move both of them to the right, only the right one will move 2011-03-10T00:21:08 unless the right one moves first 2011-03-10T00:21:13 and that would be regardless of ordering 2011-03-10T00:21:32 maybe your language is too powerful :) 2011-03-10T00:21:42 no this is simpler 2011-03-10T00:21:46 *more naive 2011-03-10T00:21:58 you just don't update some internal game state just because an order was given 2011-03-10T00:22:03 but then, 2 ants could go to the same spot 2011-03-10T00:22:11 yeah. naive 2011-03-10T00:22:16 it's avoiding some collisions, not all 2011-03-10T00:22:35 okay, let's say the first item in the stategy guide will be avoiding collision 2011-03-10T00:22:51 that's what i thought the original plan was :) 2011-03-10T00:22:54 and we can specify that the starter should not include the logic 2011-03-10T00:23:15 my collision avoidance is as follows: don't move into a square containing an ant, once i've moved an ant, implement that move 2011-03-10T00:23:15 I've been losing track, which is why I started the wiki page 2011-03-10T00:24:03 what kind of collision avoidance should the strategy guide have? 2011-03-10T00:24:07 what does no logic? 2011-03-10T00:24:08 decent? 2011-03-10T00:24:09 mean? 2011-03-10T00:24:26 sigh: it means just don't look for ant-ant collisions 2011-03-10T00:24:30 only look for ant-water collisions 2011-03-10T00:24:40 oh ok 2011-03-10T00:24:53 it's a flexible definition of "no" 2011-03-10T00:25:19 i'm looking for an occurrence of "no logic" 2011-03-10T00:25:23 not seeing it 2011-03-10T00:25:32 "not include the logic" 2011-03-10T00:25:50 "not include the logic [for avoiding ant-ant collisions]" 2011-03-10T00:25:58 ah, my sucky reading comprehension 2011-03-10T00:27:24 "Optionally attempt to remember ants and food that was last seen." <-- sounds way beyond starter-level to me, too 2011-03-10T00:27:55 and sounds like something that could be a critical part of a bot... like, most people wouldn't want to stick to a default one anyway 2011-03-10T00:28:11 IMO, it's stuff like that that leads people to rewrite unnecessarily 2011-03-10T00:29:00 yeah, I remember planet wars had dody data structures which forced anyone serious to rewrite 2011-03-10T00:29:00 so, I should take it out? 2011-03-10T00:29:08 *dodgy 2011-03-10T00:29:33 just put stuff like that in a strategy guide? 2011-03-10T00:29:37 yeah, I lot of beginners felt obligated to keep the fleet class 2011-03-10T00:29:43 McLeopold: i would take it out unless you can put a very specific algorithm in instead of a vague description 2011-03-10T00:29:56 even then... maybe that is strategy guide material 2011-03-10T00:31:06 yeah i'd say it's strategy guide worthy 2011-03-10T00:31:16 also, there is no obviously optimal way to reason about unseen food and ants, right? 2011-03-10T00:31:19 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-03-10T00:31:41 no, I don't think so 2011-03-10T00:31:47 "determine difference between unseen area and land" <-- does that mean "distinguish between land and unseen squares which could actually be water"? 2011-03-10T00:32:14 yes, but it's gone now :) 2011-03-10T00:32:14 i was debating whether to add that 2011-03-10T00:32:16 ah 2011-03-10T00:32:32 remembering that might be a candidate for starter I think 2011-03-10T00:32:35 becuase that never changes 2011-03-10T00:32:46 yeah 2011-03-10T00:32:54 well 2011-03-10T00:32:56 not necessarily 2011-03-10T00:33:06 if maps are known to be symmetrical then a bot could try to guess 2011-03-10T00:33:20 how about a visible helper function that returns true/false for a square? 2011-03-10T00:33:23 heck, even if it's only known that *some* maps are symmetrical, a bot might try to guess 2011-03-10T00:33:24 symetreical maps are not part of the spec, right? 2011-03-10T00:33:30 no 2011-03-10T00:33:30 they don't need to be 2011-03-10T00:33:31 I can't spell 2011-03-10T00:33:40 it could just be some fraction of maps are symmetrical 2011-03-10T00:33:49 *happen to be 2011-03-10T00:33:51 then they shouldn't be assumed in the starter pack, or strategy guide 2011-03-10T00:34:01 i wasn't suggesting that assumption 2011-03-10T00:34:03 when were they assumed? 2011-03-10T00:34:11 i was just saying that it's not something that necessarily stays static 2011-03-10T00:34:14 i think we should play around with both and see what works both 2011-03-10T00:34:19 if both work well, we can do a combination 2011-03-10T00:34:29 works best* 2011-03-10T00:34:35 I was just inferring things I shouldn't again when jmcarthur said "if maps are known to be symmetrical then a bot could try to guess" 2011-03-10T00:34:42 sigh: it was in response to "because that never changes" 2011-03-10T00:35:11 i see 2011-03-10T00:35:35 like, it's true that whether or not a square has been seen never changes 2011-03-10T00:36:05 but that doesn't mean that you will always want this ternary state for each square of the bot's internal map 2011-03-10T00:36:13 sure 2011-03-10T00:36:20 what? 2011-03-10T00:36:21 you may want to eventually put some probabilities in there or something instead 2011-03-10T00:36:27 when you see a square, doesn't that value cahnge? 2011-03-10T00:36:30 the first time 2011-03-10T00:37:18 like if the bot hasn't seen enough of the map to determine whether or not it's symmetrical, it's very unsure of things, but once it has seen enough to be 90% sure that it's symmetrical, it can go ahead and guess most of the rest of the water positions with decent accuracy 2011-03-10T00:37:36 antimatroid: this is about whether or not the *code* for that will change throughout the contest 2011-03-10T00:37:49 jmcarthur: it's a bit of work to find out if it's symmetrical 2011-03-10T00:38:01 you need to find rtranslate and ctranslate from pattern matching? 2011-03-10T00:38:02 antimatroid: not that much. just see if anything appears to be repeating 2011-03-10T00:38:15 sure why not? 2011-03-10T00:38:24 it may be a little involved, but probably worth it 2011-03-10T00:38:31 jmcarthur: oh, I was talking about the knowledge about the squares during a game... didn't even realize you were talking about the *code* during the contest 2011-03-10T00:38:37 yes, i'm just pointing out it isn't entirely trivial :) 2011-03-10T00:38:45 sigh: ah, that's why we were talking over each other's heads then 2011-03-10T00:38:58 antimatroid: doesn't mean it won't happen ;) 2011-03-10T00:39:15 i suspect one could maybe just hash nxn square blocks and find number of appearances? 2011-03-10T00:39:20 you could probably do it well like that 2011-03-10T00:39:29 i know it will happen, i intend on doing it myself 2011-03-10T00:39:36 my point was essentially, once you see a water square then you know for sure that it will be there for the rest of the game, and it is obvious how to handle that 2011-03-10T00:39:37 hashing would be problematic if there is just one square in there you've missed 2011-03-10T00:39:42 unlike ants and food 2011-03-10T00:39:46 Okay, more changes, refresh and read 2011-03-10T00:39:53 sigh: true 2011-03-10T00:40:31 jmcarthur: true, i don't think it'll be very easy at all :P 2011-03-10T00:40:59 are you talking about detecting symettries? 2011-03-10T00:41:03 that's one of the things i like about this game, there are LOTS of obvious things to do that will make the bot smarter 2011-03-10T00:41:04 antimatroid: it's another thing to put in the strategy guide 2011-03-10T00:41:11 and require playing around with different things 2011-03-10T00:41:20 McLeopold: ah, you meant for "game loop" to actually mean a real loop and not a function call of some sort 2011-03-10T00:41:27 i misinterpreted 2011-03-10T00:41:52 yes 2011-03-10T00:42:12 should I clarify? 2011-03-10T00:42:23 well, it's listed under "Core Functions" 2011-03-10T00:42:47 yeah, function in the broader sense of the word 2011-03-10T00:42:53 "Sub Routine"? 2011-03-10T00:43:17 "Core Functionality"? 2011-03-10T00:43:47 I watched a bit of Stargate Universe on Monday when they used that term in the show. I laughed and told my friend they were programming in BASIC. 2011-03-10T00:44:50 i'm avoiding writing a game loop as part of the library. the starter will have one, but with the intent to be redesigned as needed 2011-03-10T00:45:34 I put a game loop in the java and python bots. 2011-03-10T00:45:56 They don't have to use it, but it is good enough that even serious programmers could just stick with it for the contest. 2011-03-10T00:46:03 well, there will be one in the starter. it's just not "core functionality" in the sense that it's basically just a one-liner in the main fail 2011-03-10T00:46:07 s/fail/file/ 2011-03-10T00:46:34 oh yeah, I was going to talk about a library 2011-03-10T00:46:36 forever $ 2011-03-10T00:48:07 is haskell OOP? 2011-03-10T00:48:35 haskell... can be used to make OOP code. however, it's rarely the best design 2011-03-10T00:48:57 in any case, most OOP programmers wouldn't recognize it as the kind of OOP they know and love 2011-03-10T00:49:30 at least, not the opinionated ones 2011-03-10T00:49:58 okay, more changes, refresh and read. 2011-03-10T00:50:10 antimatroid: I'm going to hack apart your game spec 2011-03-10T00:50:22 go for it 2011-03-10T00:50:37 you can implement message passing or F-coalgebras in haskell, which are two common definitions for OOP, but most people don't even know what an F-coalgebra is 2011-03-10T00:50:41 I'm just going to separate it into sub pages 2011-03-10T00:50:59 I'm one of them. :) 2011-03-10T00:51:46 heck, i only know enough about it to make a simple object and know what it is, but i don't know any deep theory about it or anything 2011-03-10T00:52:17 i like it more than message passing though. more amenable to static type checking and purity :) 2011-03-10T00:52:30 *** delt0r_ has quit IRC (Read error: Operation timed out) 2011-03-10T00:53:05 okay, i approve of the gist of the starter guide 2011-03-10T00:53:06 *** delt0r_ has joined #aichallenge 2011-03-10T00:53:37 i think it's phrased sufficiently that the haskell starter needn't be considered an exceptional case 2011-03-10T00:53:41 okay 2011-03-10T00:53:43 good 2011-03-10T00:54:18 ooh, forgot the timing mechanism 2011-03-10T00:54:36 ah yeah. i still need to add a component for that to my new lib 2011-03-10T00:54:53 i'm basically just providing a way to get the amount of time remaining for the current turn 2011-03-10T00:55:16 amstan: if you look at the github wiki, my Ants Starter Pack Guide is red, I can't figure out how to make it blue? 2011-03-10T00:55:17 i debated adding some sort of interruptive timer, but i think that's unnecessary 2011-03-10T00:55:42 I think just a true/false function to determine if you should still loop through ants is good enough 2011-03-10T00:55:59 It give an example for the language 2011-03-10T00:56:06 err... how does the function know whether to go through the loop though? 2011-03-10T00:56:24 oh, you mean to demonstrate the timer in the bot? 2011-03-10T00:56:27 *in the starter? 2011-03-10T00:56:40 yeah, all language have very different "get_time" functions 2011-03-10T00:56:55 it could be a part of the starter interface though 2011-03-10T00:57:12 I'll add it to the optional list of functions 2011-03-10T00:57:14 like just have a timeRemaining function 2011-03-10T00:57:17 yeah, optional is good 2011-03-10T00:57:36 what kind of time are we measuing? 2011-03-10T00:57:41 measuring? 2011-03-10T00:57:42 wall clock 2011-03-10T00:57:51 milliseconds 2011-03-10T00:57:57 so it matches the bot input 2011-03-10T00:58:00 hehe 2011-03-10T00:58:12 wall clock should be more reliable this time around if we make the bots execute sequentially and pause them between turns 2011-03-10T00:58:20 hopefully 2011-03-10T00:58:22 s/reliable/reliable and fair/ 2011-03-10T00:58:31 why are they executed sequentially? 2011-03-10T00:58:38 more bots than cores 2011-03-10T00:58:42 in a game 2011-03-10T00:58:48 i see 2011-03-10T00:59:17 just read some of the recent conversation; why doesn't the engine treat all ant moves as simultaneous? 2011-03-10T00:59:29 janzert: the *engine* should 2011-03-10T00:59:43 janzert: this is about the bots view of issuing orders 2011-03-10T00:59:48 ahh, must have misread something then 2011-03-10T01:00:02 s/this is/that was/ 2011-03-10T01:00:07 well, any order or err, orders should be fine 2011-03-10T01:00:29 i.e. lead to the same result 2011-03-10T01:00:51 janzert: the proposal was to have some internal representation of the map in the bot such that ants are moved immediately when orders are issued, which is a naive way to look for collisions 2011-03-10T01:01:20 that way the bot can avoid actually issuing orders resulting in collisions 2011-03-10T01:01:20 ahh, ok so something easy but not quite completely correct 2011-03-10T01:01:25 yeah 2011-03-10T01:01:54 but when you phrase it that way it makes me think we should definitely not be encouraging beginners to run with it like that 2011-03-10T01:02:22 we should encourage them to notice the mistake and fix it 2011-03-10T01:02:36 McLeopold: yes, if we were to include that in the first place 2011-03-10T01:02:47 janzert: anyway, the final decision i think was that collision avoidance should be part of the strategy guide and not the starter itself 2011-03-10T01:03:05 ok 2011-03-10T01:03:29 I was mostly just afraid that was the way the engine was actually handling it 2011-03-10T01:03:46 that would be pretty bad 2011-03-10T01:03:53 * jmcarthur hopes that's not the way it works :) 2011-03-10T01:05:12 McLeopold: so ant-water collisions actually result in the engine ignoring the move altogether, right? what about ant-ant collisions? do they still kill the ants, or do they prevent one or both from moving? 2011-03-10T01:05:32 ant-ant collisions kill the ants 2011-03-10T01:05:48 ant-water is an invalid order and is ignored 2011-03-10T01:06:15 McLeopold: i was thinking... given the former property of the engine, i could just send four orders for each square on the map, regardless of the actual game state, and satisfy the game logic requirements of the starter guide :P 2011-03-10T01:06:32 hehe 2011-03-10T01:06:36 s/former property/ant-water behavior/ 2011-03-10T01:07:07 you also need the 'ignore all invalid orders' property 2011-03-10T01:07:20 janzert: that's in there already. i had checked the source when i first thought of it 2011-03-10T01:07:27 hmm 2011-03-10T01:07:32 i don't know about all invalid orders 2011-03-10T01:07:38 but redundant orders are ignored, i know 2011-03-10T01:07:47 i think simply invalid ones are too though 2011-03-10T01:07:59 jmcarthur: yep, your right 2011-03-10T01:08:53 since this is multiplayer, we need to try and keep all bots participating, so there will be no bots removed from play do to bad output 2011-03-10T01:09:12 unless the bad output is infinite... 2011-03-10T01:09:22 in which case they will time out 2011-03-10T01:09:30 so just count the moves and it it's infinity they lose ;) 2011-03-10T01:09:46 *if 2011-03-10T01:09:53 yes, but what magnitude of infinity? 2011-03-10T01:10:08 what if they issue an uncountably infinite number of moves? 2011-03-10T01:10:17 we're just operating on the extended reals here. nothing fancy! 2011-03-10T01:10:26 *extended naturals 2011-03-10T01:10:29 sigh: I'll just map them to the set of integers 2011-03-10T01:10:41 McLeopold: good idea 2011-03-10T01:11:17 negative integers, to rub it in 2011-03-10T01:11:28 that's their score :P 2011-03-10T01:11:44 hmm... - 1 point for each bad order? 2011-03-10T01:12:16 they should get imaginary points for bad orders 2011-03-10T01:12:19 bad orders are obviously bugs. i wouldn't mind if it nullified *all* their points, personally 2011-03-10T01:12:32 it would still mean they play to the end 2011-03-10T01:12:38 so that other bots get decent scores 2011-03-10T01:12:44 yes, and keep the individual game balance 2011-03-10T01:12:51 ahhh... the dead but they don't even know it way of dealing with things 2011-03-10T01:13:28 -1 point per bad move sounds reasonable to me, really 2011-03-10T01:13:44 we could split their bot's input into a parallel universe game than never ended. 2011-03-10T01:13:44 but maybe restrict all final scores to non-negative numbers 2011-03-10T01:13:52 I would rather just give them 0 points 2011-03-10T01:14:07 makes no difference to me at all 2011-03-10T01:14:22 buggy bots are buggy and they deserve whatever crap scores they get 2011-03-10T01:14:22 why should a buggy bot get points? 2011-03-10T01:14:28 maybe a too many bad orders threshold 2011-03-10T01:14:56 nah, threshold just complicats things 2011-03-10T01:15:03 if we do this, i don't think ant-water collisions should be considered invalid such that they make you lose points 2011-03-10T01:15:13 it should just kill the ant 2011-03-10T01:15:19 I agree 2011-03-10T01:15:26 i mean logically ants should just drown 2011-03-10T01:15:43 it makes sense 2011-03-10T01:16:01 what if we go back to walls? 2011-03-10T01:16:30 but orders for ants that don't exist or for ants that have already been given orders should be "invalid" including whatever penalty that implies 2011-03-10T01:16:36 then we have the suicide to prevent enemy points issue 2011-03-10T01:16:41 yes 2011-03-10T01:16:56 not sure that's so terrible, myself 2011-03-10T01:17:03 I think it's good 2011-03-10T01:17:19 it's not very ant-like i guess... 2011-03-10T01:17:23 but neither is anything else about the game 2011-03-10T01:17:37 that is the nature of points, not ants 2011-03-10T01:17:42 or walls/water 2011-03-10T01:18:06 well, it's not ant-like to commit suicide in the name of points because ants don't care about points, especially not enough to commit suicide over them :P 2011-03-10T01:18:36 if you say that, then it's not ant-like to care about points at all 2011-03-10T01:18:42 i agree 2011-03-10T01:18:50 it might be ant-like to prefer a death by water than to be tortured by the enemy 2011-03-10T01:18:53 but suicides are even less ant-like 2011-03-10T01:18:56 if you want a justification :P 2011-03-10T01:18:57 heh 2011-03-10T01:19:50 having invalid orders reflect score changes that are not reflected in play behavior has a strong potential for weird ratings effects 2011-03-10T01:20:07 i agree 2011-03-10T01:20:17 i think it should be all or nothing 2011-03-10T01:20:37 they either get 0 score, or their score is not affected 2011-03-10T01:20:59 e.g. a bot that play fairly strongly but also issues lots invalid orders is going to continually play bots that it easily wipes out 2011-03-10T01:22:05 janzert: whoever comes in "second" will get the benefit 2011-03-10T01:22:26 meaning the strong bot could choose the ordering it wants 2011-03-10T01:22:42 well, a strong bot could do that anyway 2011-03-10T01:24:30 that's always going to be the case, yeah 2011-03-10T01:24:50 it's not a strong bot if its scores are miserable 2011-03-10T01:24:51 there just really is no ideal way to deal with badly behaving bots, TANSTAAFL is in full effect here 2011-03-10T01:25:32 is cooperation between bots explicitly discouraged? 2011-03-10T01:25:35 the ability to wipe out opposing teams of ants is insufficient to be considered a strong bot, is all i mean 2011-03-10T01:25:38 sigh: no 2011-03-10T01:25:45 interesting 2011-03-10T01:25:49 sigh: no, but i doubt it will happen much or at all 2011-03-10T01:25:57 I'm planning it? 2011-03-10T01:26:04 really? 2011-03-10T01:26:10 s/?/>:)/g 2011-03-10T01:26:12 how does that even work? 2011-03-10T01:26:24 it would be fairly easy to develop a protocol to detect *friendly* bots 2011-03-10T01:26:30 does your plan require working it out with other developers in advance? 2011-03-10T01:26:33 It will be an opt-in public feature. 2011-03-10T01:26:36 ah yeah with a protocol 2011-03-10T01:26:44 well in that case it's interesting 2011-03-10T01:27:00 For *python* programmers only >:) 2011-03-10T01:27:04 haha 2011-03-10T01:27:05 sigh: you mean to detect bots that are privy to your handshake at least 2011-03-10T01:27:14 jmcarthur: yes 2011-03-10T01:27:27 communication is also exploitable >:) 2011-03-10T01:27:40 yes, but fog of war give some cover 2011-03-10T01:27:59 I already have a protocol worked out. 2011-03-10T01:28:02 time to authentication schemes :P 2011-03-10T01:28:03 haha 2011-03-10T01:29:10 The real problem is what to communicate and how to get as much information per turn without wasting too many ants. 2011-03-10T01:29:22 *** BtbN has quit IRC (Remote host closed the connection) 2011-03-10T01:29:37 but if you have a friendly you are not fighting, then you probably have some left over anyway 2011-03-10T01:29:42 true, but you can afford to waste quite a few ants if you are working together 2011-03-10T01:29:46 yeah 2011-03-10T01:30:04 you could dedicate a bunch to communication 2011-03-10T01:30:15 maybe I'll write the protocol in malbloge... 2011-03-10T01:30:16 but hmm... only 2 bits/ant/turn? 2011-03-10T01:30:26 is that right? 2011-03-10T01:30:59 maybe less? you can't have them wondering too far 2011-03-10T01:31:08 yeah, I meant as a maximum 2011-03-10T01:32:07 move in one of 4 directions or stand still, 2bits + 1 maximum I believe 2011-03-10T01:32:18 we're talking pretty low bandwidth 2011-03-10T01:32:56 it's tougher than that due to other aspects of the game, like sight range, combat range, food, number of ants in area, etc. 2011-03-10T01:33:09 just knowing they will hold a truce is powerful enough 2011-03-10T01:33:21 i think we'd be lucky to get an average of 1 bit per visible ant 2011-03-10T01:34:22 janzert: I'm going to totally merge the branch sometime tomorrow. 2011-03-10T01:34:30 ish 2011-03-10T01:34:35 great :) 2011-03-10T01:34:37 ish 2011-03-10T01:34:41 :P 2011-03-10T01:35:04 and I'm totally ignoring all sandbox/jail stuff, so if you aren't going to do it, we need another recruite 2011-03-10T01:35:24 amstan: can you organize another all hands meeting? 2011-03-10T01:36:16 just keep what we had last time for now and we'll see if someone comes along 2011-03-10T01:36:37 okay, it will work, but it won't be sequential 2011-03-10T01:36:49 *** sigh has left #aichallenge ("Leaving") 2011-03-10T01:36:58 oh, yeah that part will be a problem 2011-03-10T01:37:20 *** McLeopold has quit IRC (Read error: Connection reset by peer) 2011-03-10T01:37:52 what exactly needs doing sandbox-wise? are there open tickets about it? 2011-03-10T01:38:14 the whole thing needs a good going over 2011-03-10T01:38:37 the part McLeopold was mentioning is that it needs an interface to suspend and restart bots 2011-03-10T01:39:04 which is completely nonexistent at the moment 2011-03-10T01:39:42 are we still on the old sadbox.py thing? 2011-03-10T01:40:50 yes 2011-03-10T01:41:17 actually I don't think even that is being used at the moment 2011-03-10T01:42:37 as I was about to say before McLeopold got disconnected, first thing I suppose is just get something end to end up and running 2011-03-10T02:20:16 aichallenge: janzert epsilon * r1d19525 / (3 files): More server setup script work - http://bit.ly/ewW8HH 2011-03-10T02:22:23 *** Palmik has joined #aichallenge 2011-03-10T02:42:51 *** amstan has quit IRC (Remote host closed the connection) 2011-03-10T02:46:37 *** sigh has joined #aichallenge 2011-03-10T02:48:09 *** delt0r___ has joined #aichallenge 2011-03-10T02:49:22 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-03-10T03:10:36 *** delt0r_ has joined #aichallenge 2011-03-10T03:12:13 *** delt0r___ has quit IRC (Ping timeout: 255 seconds) 2011-03-10T03:36:09 *** aerique has joined #aichallenge 2011-03-10T04:02:09 *** AlliedEnvy has quit IRC (Read error: No route to host) 2011-03-10T04:13:08 antimatroid: have you done anything else besides feeding your bot data on standard input to make sure it works? 2011-03-10T04:16:09 *** Kingpin13 has joined #aichallenge 2011-03-10T04:18:18 *** AlliedEnvy has joined #aichallenge 2011-03-10T04:28:05 *** boegel has joined #aichallenge 2011-03-10T04:33:41 *** Frontier has joined #aichallenge 2011-03-10T04:38:42 sigh: I got your notice, I don't know if there was a reason behind it. We can change that, it is probably more straight forward to parse if one doesn't have to extract the number and subtract one. Although with the right RegExp it is a one-liner. 2011-03-10T04:41:38 Frontier: having to parse the key instead of the value just seem like a kludge to me 2011-03-10T04:42:33 also, having a static set of keys is neater 2011-03-10T04:43:50 Is McLeopold around? Ok.... I think it was his idea :p 2011-03-10T04:44:14 McLeopold agreed with me :) 2011-03-10T04:44:25 if you look at the chat logs around the time I sent the message 2011-03-10T04:45:16 Could you change the Wiki accordingly? 2011-03-10T04:45:55 I haven't edited the wiki before, but I'll give it a shot 2011-03-10T04:48:42 the ai-contest is the one big education event :-). you learn how to use git, the wiki, a dozen programming languages and international project management 2011-03-10T04:49:42 yeah :) 2011-03-10T04:50:02 this is the first time I've used git to collaborate with anyone 2011-03-10T04:50:22 ok, changed 2011-03-10T05:08:06 *** mega1 has joined #aichallenge 2011-03-10T05:15:41 once a bot has received the end turn data it can just die, aight? the game engine doesn't expect a 'go' or anything 2011-03-10T05:24:58 I wonder if the engine even lets the bot read the end game state or just say, "so, match is over you can read the results at home" 2011-03-10T05:26:30 Frontier: right, good point 2011-03-10T05:48:14 *** phreeza has joined #aichallenge 2011-03-10T05:48:39 @later tell amstan I've issued a pull request for the first version of the Common Lisp example bot. 2011-03-10T05:48:40 aerique: OK 2011-03-10T05:49:17