2011-06-27T00:00:05 consider the case where b is reflected horizontally to a 2011-06-27T00:01:05 if there's an uneven number of columns, they can skip to every square by moving 2 cols at a time, if there are an even number of rows, they skip into columns that can't land on each other and fill the row 2011-06-27T00:01:27 you just put a restriction 2011-06-27T00:01:45 i don't see how you can have an even number of columns/rows when the maze has to be even 2011-06-27T00:01:54 uneven* 2011-06-27T00:02:20 .a.b. 2011-06-27T00:02:28 that is symmetric horizontally with 5 cols yes? 2011-06-27T00:02:47 yes, but for a maze you cannot have that 2011-06-27T00:03:07 mazes are lame :( 2011-06-27T00:03:20 or.. you can make a special case for the edges 2011-06-27T00:04:46 yes.. that would work 2011-06-27T00:05:08 you know how you have mostly walls on the right side of ########### 2011-06-27T00:05:12 http://ants.zeroviz.us/KHFUK/* 2011-06-27T00:05:23 without the * 2011-06-27T00:05:40 mostly walls where? 2011-06-27T00:05:47 on the last column 2011-06-27T00:05:56 26 Jun 8:47 Was my last game where I didn't time out. The next game at : 26 Jun 9:12 is timing out. Every C# bot is timing out. 2011-06-27T00:06:00 no more than any other column :P 2011-06-27T00:06:10 well, every second column :P 2011-06-27T00:06:29 the first column has vertical corridors as well, so it's mostly empty 2011-06-27T00:06:39 and the last one is mostly walls 2011-06-27T00:06:56 I lie, the game at 26 Jun 9:54 seemed to go alright. And then from that point on, every game has timed out. 2011-06-27T00:07:11 when the map wraps, you don't necessarly have to keep that more walls/less walls pattern 2011-06-27T00:07:17 you could just double one of them 2011-06-27T00:07:36 the problem is that it won't work for your translational symmetry types anymore 2011-06-27T00:07:45 since it won't be symmetric then 2011-06-27T00:07:50 but it will for rotational and mirror 2011-06-27T00:09:20 amstan: i think that would happen already, except my get location function has no problems passing off one side of the map 2011-06-27T00:09:34 ? 2011-06-27T00:09:46 i'm not really familiar with the mapgen 2011-06-27T00:09:51 so i don't know its perks 2011-06-27T00:10:54 *** Kingpin13 has joined #aichallenge 2011-06-27T00:11:04 if i get the location 2 west from a location on the right wall, i'm not going to get the first location on the left without severely hacking the get location function 2011-06-27T00:11:13 2 east :P 2011-06-27T00:13:15 what? 2011-06-27T00:14:04 amstan: you're saying double the fist and last column yes? 2011-06-27T00:14:10 the problem is, get location doesn't know that 2011-06-27T00:14:40 make the map first, then at the end do it 2011-06-27T00:14:58 i have generalised all the getting symmetric location stuff to writing general map generators for the symmetry types 2011-06-27T00:15:06 but actual mazes are sounding way too hard 2011-06-27T00:15:17 it would require like a separate function for each type of symmetry 2011-06-27T00:15:23 so what's wrong with just limiting mazes to even numbers only 2011-06-27T00:15:34 that doesn't fix my problem 2011-06-27T00:15:48 how did you get http://ants.zeroviz.us/KHFUK/ then? 2011-06-27T00:16:08 that's a tile map, the rotationals aren't reflected 2011-06-27T00:16:23 consider a..b 2011-06-27T00:16:33 oh, now i get it 2011-06-27T00:16:50 so.. i still don't understand how you get from a normal map to a symmetric map? 2011-06-27T00:17:06 one way wants to use uneven, but the other way wants even 2011-06-27T00:17:10 do you flip it then OR all the water? 2011-06-27T00:17:29 when i add land, i get all symmetrically equivalent locations and add land 2011-06-27T00:17:50 ic, so it is ORing all the land 2011-06-27T00:18:00 yep 2011-06-27T00:18:18 so why can't you just stop your generator when you get halfway? so you only go one pass everywhere 2011-06-27T00:18:28 let the symmetry fill the rest 2011-06-27T00:18:50 i am doing one pass everywhere 2011-06-27T00:19:00 but the map requires every second one of those to be empty 2011-06-27T00:19:08 when you do that for both players, that's the entire row/col 2011-06-27T00:19:28 #define mapsize 10x10 2011-06-27T00:19:58 #define symmetrytype mirror 2 player 2011-06-27T00:20:30 yes, but from what i understand you're generating a 10x10 unsymmetric maze first, then you're flipping it around the middle and adding it to the unflipped 2011-06-27T00:20:41 having land wherever either maps have land 2011-06-27T00:20:59 why don't you just generate a 10x5 unsymmetric maze, keep the other 10x5 water? 2011-06-27T00:21:37 http://ants.zeroviz.us/COxEJ/ :P 2011-06-27T00:22:29 what was the original maze though? 2011-06-27T00:23:04 disable the "i get all symmetrically equivalent locations and add land" part and show me the resulting map 2011-06-27T00:23:46 the problem is it's a bit of work to give you both :P 2011-06-27T00:23:56 actually, give me a minute 2011-06-27T00:26:18 http://ants.zeroviz.us/TrvNo/ 2011-06-27T00:26:18 http://ants.zeroviz.us/zbZ1o/ 2011-06-27T00:26:25 you'll need to move it if you want the ants in the same place 2011-06-27T00:26:30 it was too much effort to fix that 2011-06-27T00:27:07 do you have a text version? 2011-06-27T00:27:16 oh 2011-06-27T00:27:17 found it 2011-06-27T00:31:17 amstan: http://asciipaint.com/viewImages.php?i=672 i want to do mazes like that with dfs carves into the map 2011-06-27T00:31:21 antimatroid: http://pastebin.com/XBkM1HxU 2011-06-27T00:31:34 but it's really hard to do, and i want it to be done with like nxn blocks :\ 2011-06-27T00:33:35 antimatroid: http://pastebin.com/Gu2iPTWr 2011-06-27T00:33:37 amstan: but is there an easy way for my to do that for all types of symmetry :P 2011-06-27T00:34:01 antimatroid: not sure, you have to crop it based on how the symmetry does it 2011-06-27T00:34:11 if the symmetry already took care of a block, don't fill it 2011-06-27T00:34:36 antimatroid: http://pastebin.com/Z8LyH3rt 2011-06-27T00:35:52 i didn't even realize you were doing OR for the land before 2011-06-27T00:36:07 i guess it doesn't really matter much for the other maps, unless you get to these pattern maps 2011-06-27T00:37:13 OR? 2011-06-27T00:39:50 as in.. generating the map while symmetry was in effect 2011-06-27T00:39:57 or generating the whole map with the symmetry on 2011-06-27T00:40:18 when i did this for tron i only worried about half the map 2011-06-27T00:40:29 i will try to work out how to do that 2011-06-27T00:40:43 but seems complicated with all the symmetries to me 2011-06-27T00:40:48 i might be able to just find a basis 2011-06-27T00:40:59 i can't believe we did the tron mapgen in c, lol 2011-06-27T00:41:07 pgpaskar_: remember those times? 2011-06-27T00:41:13 i wish i could do the mapgen in c++ :( 2011-06-27T00:41:18 it would be a lot neater too :P 2011-06-27T00:42:46 amstan: i think the basis idea will work :\ 2011-06-27T00:43:42 i might just try it and see what happens 2011-06-27T00:46:03 *** okayzed is now known as okay 2011-06-27T01:01:48 *** Pyronaut has quit IRC (Ping timeout: 252 seconds) 2011-06-27T01:02:31 amstan: http://ants.zeroviz.us/oUxfS/ that seems to have worked 2011-06-27T01:02:43 one problem is now some of the maze becomes not connected 2011-06-27T01:02:50 but i can potentially flood fill those parts 2011-06-27T01:09:33 *** okay is now known as okayzed 2011-06-27T01:11:35 amstan: any idea about what to do about the now disconnected regions? :P 2011-06-27T01:12:07 generate the maze with restricted dimensions 2011-06-27T01:14:41 ? 2011-06-27T01:15:14 oh build the map on the basis set? 2011-06-27T01:15:21 i think i can do that with a little bit of work 2011-06-27T01:15:47 like with our reflected map, if i move east to the other players side, it needs to change direction? 2011-06-27T01:17:20 alright, i'm gonna go get some food then start another rewrite 2011-06-27T01:17:21 wow, nice 2011-06-27T01:17:23 that will take some effort 2011-06-27T01:29:39 *** stocha has joined #aichallenge 2011-06-27T01:30:24 janzert: Wolrld war one at version http://aichallengebeta.hypertriangle.com/visualizer.php?game=40320&user=107 (1500 turns seems too short for those slow bots ) 2011-06-27T01:32:44 @later tell zaphus: here is a game of a 256x256 map capable bot (http://aichallengebeta.hypertriangle.com/visualizer.php?game=40320&user=107) It's very sloow. But cool looking still. Not very strong though :'( 2011-06-27T01:32:44 stocha: OK 2011-06-27T01:36:31 *** stocha has quit IRC (Quit: Page closed) 2011-06-27T01:56:02 *** Apophis_ has joined #aichallenge 2011-06-27T02:00:03 *** nann has quit IRC (Ping timeout: 255 seconds) 2011-06-27T02:00:09 *** Apophis has quit IRC (Ping timeout: 264 seconds) 2011-06-27T02:00:27 *** Apophis has joined #aichallenge 2011-06-27T02:03:25 *** Palmik has joined #aichallenge 2011-06-27T02:37:29 *** foRei has quit IRC (Quit: Bye) 2011-06-27T02:54:30 *** Kingpin13 has quit IRC (Ping timeout: 255 seconds) 2011-06-27T02:56:11 *** gobang has joined #aichallenge 2011-06-27T02:56:41 *** gobang has joined #aichallenge 2011-06-27T02:57:09 2011-06-27T03:04:14 *** GeorgeSebastian has joined #aichallenge 2011-06-27T03:04:15 *** GeorgeSebastian has joined #aichallenge 2011-06-27T03:32:07 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-06-27T03:39:57 *** amstan has quit IRC (Ping timeout: 255 seconds) 2011-06-27T03:45:06 http://ants.zeroviz.us/49v4Z/ 2011-06-27T03:51:58 *** UncleVasya has joined #aichallenge 2011-06-27T04:15:45 *** Naktibalda has joined #aichallenge 2011-06-27T04:31:15 *** dr- has joined #aichallenge 2011-06-27T04:50:09 *** sir_macelon has joined #aichallenge 2011-06-27T04:50:13 hi 2011-06-27T04:52:38 šų 2011-06-27T04:52:40 seems i'm undenied at the top :) 2011-06-27T04:52:41 hi 2011-06-27T04:53:33 high flyer 2011-06-27T04:53:34 :D 2011-06-27T04:54:47 guess people are waiting for the official start of the contest 2011-06-27T04:56:24 nah 2011-06-27T04:56:31 you are just too good 2011-06-27T04:57:11 http://ants.zeroviz.us/2MGOq/ 2011-06-27T04:57:59 insane map? 2011-06-27T04:58:07 *** stocha has joined #aichallenge 2011-06-27T04:58:20 stocha: http://ants.zeroviz.us/2MGOq/ :P 2011-06-27T04:58:22 insane? 2011-06-27T04:58:34 it's sad to realize that if i do not win this thing, people won't get to see my interesting coding techniques and ideas 2011-06-27T04:58:39 quite ... interesting 2011-06-27T04:58:57 antimatroid: no we are talking about map maping :) 2011-06-27T04:59:04 antimatroid: congratulation anyway :) 2011-06-27T04:59:46 I'll try to play on it now :) 2011-06-27T04:59:46 antimatroid: now to tune this thing to have 3 width corridor (although blocking doesn't sound that practical in there :p ) 2011-06-27T04:59:48 that would be fun to play with that 2011-06-27T05:00:03 what are the dimensions ? 2011-06-27T05:00:34 i think my new bot would do quite fine on this one. And also my old reference bot probably. 2011-06-27T05:00:40 click the raw file 2011-06-27T05:00:41 i still need to get it working for all symmetry types 2011-06-27T05:00:48 there are restrictions on the dimensions for maze maps 2011-06-27T05:01:21 antimatroid: let's put an assymetric maze for show. you can always update it later to symetric one. 2011-06-27T05:01:26 where there any mazes already added to beta? 2011-06-27T05:01:31 i think people will be thrilled to try it as soon as possible :) 2011-06-27T05:01:43 sir_macelon: not yet 2011-06-27T05:01:50 besides, "maze" are probably quite fair anyway. 2011-06-27T05:01:58 especially bigs ones 2011-06-27T05:02:08 simply split the player, and it should do okay 2011-06-27T05:02:24 Although .... the "starting visible food" might break a bit there. 2011-06-27T05:02:52 What's the biggest map of the beta as for now ? 2011-06-27T05:03:14 no idea 2011-06-27T05:03:24 not above 120x120 2011-06-27T05:03:29 let's make a classical 128x128 for show also okay ? 2011-06-27T05:03:50 not really high on my list of things to do 2011-06-27T05:03:55 i'm trying to get this thing working 2011-06-27T05:04:10 antimatroid: okay you made good work. So i won't argue :) 2011-06-27T05:08:54 http://ants.zeroviz.us/BbT9O/ 2011-06-27T05:08:57 just because i can :P 2011-06-27T05:12:31 antimatroid: does the vizualizer manage it well ? 2011-06-27T05:12:45 i haven't played any games with it 2011-06-27T05:12:49 let's put it on the beat, and see how many bot times out :p 2011-06-27T05:12:54 that's a legitimate map if you take the raw file though 2011-06-27T05:13:44 antimatroid: now to use the room pattern, and make a smiley map. 2011-06-27T05:14:14 (instead of the sort of cross you have on this one) 2011-06-27T05:14:49 that map is way too big for the beta server 2011-06-27T05:15:04 it's like 307x307 2011-06-27T05:16:51 oh. make a small 256x256 one then :p 2011-06-27T05:19:16 i did a ridiculous one for you instead :P 2011-06-27T05:19:22 this thing is reasonably fast 2011-06-27T05:20:26 http://ants.zeroviz.us/vhf1b/ 2011-06-27T05:23:07 lol 2011-06-27T05:23:51 I wanna one of this on the betaserver too. For stocha's reason :) 2011-06-27T05:24:02 that's 1000x1000 :P 2011-06-27T05:24:08 and a friggin maze :P 2011-06-27T05:24:14 looking at this game, i think 1500 turn is important for weak bots 2011-06-27T05:24:15 http://aichallengebeta.hypertriangle.com/visualizer.php?game=40735&user=107 2011-06-27T05:24:59 antimatroid: i wonder if my current stocha implementation could play on this one and not time out (it slows down with map growing, and number of food seen) 2011-06-27T05:25:13 my bot might time out while path finding 2011-06-27T05:25:19 it's got a timer to check usually though 2011-06-27T05:25:29 you'd have to stretch my bot pretty far for it to timeout atm 2011-06-27T05:25:33 we ought to have big maps though. 2011-06-27T05:25:38 why? 2011-06-27T05:25:47 they just limit the amount of cool strategies people can try? 2011-06-27T05:25:52 because it's interesting to see how to handle them 2011-06-27T05:25:53 they aren't needed to get cool games 2011-06-27T05:26:13 antimatroid: so far i'm amazed at how many "fast runable" strategy i still have to try 2011-06-27T05:26:24 let's make them only 5% or even 1% 2011-06-27T05:26:32 but some people will like to handle them. 2011-06-27T05:26:35 *** onensora has joined #aichallenge 2011-06-27T05:26:39 i clearly have the means for that :) 2011-06-27T05:26:50 we will have a secondary tcp server with crazy maps 2011-06-27T05:27:04 because for some reasons, i puted this constraint on myself : my bot should have way to be the best on 256x256 maps 2011-06-27T05:27:05 we can go wild there with large maps, asymmetric maps, blockable maps, single corridor maze maps etc. 2011-06-27T05:27:17 (which happens to be the upper bound McLeopold gave me ...) 2011-06-27T05:27:33 we gave you an upper bound 2011-06-27T05:27:37 not a supremum 2011-06-27T05:27:58 antimatroid: the only ting that worries me, is the viewer. If the viewer break, then i agree that it start to be questionnable 2011-06-27T05:28:19 i'm amazed it handled even displaying 1000x1000 2011-06-27T05:28:19 if you think that small map make better games, i can understand. But don't suppress the large ones all together. 2011-06-27T05:28:49 antimatroid: you have the viewer not breaking, and then you have the ease of use / speed of display. 2011-06-27T05:29:15 On the current beta map, there is no problems i believe. But i want to find out what the upper bound is. For comfortable viewing. 2011-06-27T05:29:20 if you have 1% really large maps, then people doing cool strategies pretty much need to write an entirely separate bot just for those edge cases 2011-06-27T05:29:22 what's the point? 2011-06-27T05:29:29 why not just have them on the crazy maps server? 2011-06-27T05:29:29 then i'll probably advocate for that to be official upper bound. 2011-06-27T05:29:58 antimatroid: i want to show the performances i can achieves on those. 2011-06-27T05:30:08 why not do it on the crazy maps server? 2011-06-27T05:30:12 And beside, i feel my current strategy display is one of the coolest :p 2011-06-27T05:30:30 So currently, i would say that the only really cool strategy does work well on very big maps :p 2011-06-27T05:30:59 although it seems quite weak yet on small ones. I think i can get it quite high though. 2011-06-27T05:30:59 i think maze maps will be like having larger open maps anyway 2011-06-27T05:31:06 as paths will be longer between locations 2011-06-27T05:31:23 antimatroid: what i want to show, is that my bot can commande 7000 bots in 300 ms. 2011-06-27T05:31:23 although path finding will be quicker 2011-06-27T05:31:28 7000 ants. 2011-06-27T05:31:28 because you have less places to look 2011-06-27T05:31:32 a* will be slower 2011-06-27T05:31:33 while doing clever stuff with them. 2011-06-27T05:32:16 The point is : that some people can't make good bot on those big maps, don't mean that no people can. 2011-06-27T05:32:25 let a chance to those to express themselve. 2011-06-27T05:32:56 even if it doesn't enable them to win the contest :) 2011-06-27T05:33:02 (because 1% really is not much) 2011-06-27T05:33:02 i personally don't like having a wide range of dimensions for maps 2011-06-27T05:33:15 let's only go for the biggest then. 2011-06-27T05:33:25 that's just your preference though 2011-06-27T05:33:40 antimatroid: the problem is, you have already implemented some algorithm. So you'll choose the maps that fits the limitations of them the best. 2011-06-27T05:33:52 antimatroid: the same applies to me 2011-06-27T05:33:58 i searched for scalable algorithm 2011-06-27T05:34:06 so i want big maps because if that :) 2011-06-27T05:34:15 i haven't implemented my algorithm yet 2011-06-27T05:34:16 So i guess diversity is the solution. 2011-06-27T05:34:21 i'd done food collection and not fully 2011-06-27T05:34:25 i've* 2011-06-27T05:35:06 antimatroid: what i say, is that neither you or me will be objective when choosing the upper/lower bounds. Inconsioulsy we'll try to make you natural thinking at ease. 2011-06-27T05:35:22 our not you 2011-06-27T05:35:48 i agree that we're both biased 2011-06-27T05:35:56 i do my best to get the opinions of others etc. 2011-06-27T05:35:59 Anyway the ant game has so many varation, everyone will not be happy. 2011-06-27T05:36:16 but i'm also fairly sure most people would agree that too larger dimensions will diminish the quality of the bots possible 2011-06-27T05:36:21 but to have someone reasonably unbiased made the choices would be cool (be it polls) 2011-06-27T05:36:58 antimatroid: then you need to build an unbiased poll, to proove that. (though even if they do, it does means it's right. ) 2011-06-27T05:37:16 because the same apply for small maps. 2011-06-27T05:37:27 small maps will limite the quality of the bots possible. 2011-06-27T05:37:32 we also have resource limitations 2011-06-27T05:37:36 larger maps take longer on average 2011-06-27T05:37:44 yeah. that is a good argument. 2011-06-27T05:37:56 then make all map 10x10 2011-06-27T05:38:04 the game will run very quick then. 2011-06-27T05:38:08 trust me, i would enjoy that strategy wise :P 2011-06-27T05:38:15 we could have 1000 done in a few second. 2011-06-27T05:38:24 no, i'd still demand a second a turn 2011-06-27T05:38:34 antimatroid: i don't believe the fighting system is rich enough for very small maps 2011-06-27T05:38:35 you underestimate my ability to use time 2011-06-27T05:38:53 antimatroid: but we can make the game 100 turn or less 2011-06-27T05:38:59 and be comfortable with it. 2011-06-27T05:39:05 you want the game to reasonably play out 2011-06-27T05:39:08 so even with 1 sec a turn, it'll be quite quick 2011-06-27T05:39:38 antimatroid: let's keep this idea of building a reasonable consensual upper bound then. 2011-06-27T05:39:46 we'll work on that later. 2011-06-27T05:39:56 show us a labyrinth map on beta :) 2011-06-27T05:40:01 so people can see it close. 2011-06-27T05:40:07 i stil need to work on it 2011-06-27T05:40:10 which i might get back to 2011-06-27T05:40:14 even if it's not symetrical yet. 2011-06-27T05:40:43 While small maps are good for tactical monster the big maps are for the strategysts. I think it is good to have 3 types of maps: pure tactical, pure strategical, where tactics and strategy are near the same in profit. 2011-06-27T05:41:06 did you just refer to the larger as more strategical? 2011-06-27T05:41:18 * antimatroid 's head asplodes 2011-06-27T05:41:29 :) 2011-06-27T05:41:38 The thing nedd to be managed for big maps is that ants "countries" have a long boundaries with others. 2011-06-27T05:42:25 If I said stupid thing than neverr mind :) 2011-06-27T05:42:28 what would be really important with maze maps is the ability to track down your enemies 2011-06-27T05:42:38 which means tracking information about where enemies might be outside of view 2011-06-27T05:43:02 UncleVasya: yeah i think you are right 2011-06-27T05:43:05 it wasn't stupid, i was poking at my preference for small examples :P 2011-06-27T05:43:07 small = tactical. 2011-06-27T05:43:12 medium = average 2011-06-27T05:43:19 big = strategical 2011-06-27T05:43:22 small is like chess 2011-06-27T05:43:26 medium is like go 2011-06-27T05:43:26 tactical == strategical imo 2011-06-27T05:43:30 big is like life :) 2011-06-27T05:43:42 antimatroid: did you ever play go ? 2011-06-27T05:43:45 nope 2011-06-27T05:43:57 antimatroid: this game as both tactical and strategical. They are not the same things :) 2011-06-27T05:44:16 Chess is 95% tactical 2011-06-27T05:44:20 go is abut 50/50 2011-06-27T05:44:56 sir_macelon's bot is getting crazily efficient 2011-06-27T05:45:02 *** sigh has joined #aichallenge 2011-06-27T05:45:02 For me strategic map is a map where decision "where to fight" is more important than "how to fight". 2011-06-27T05:55:52 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-27T06:04:29 *** UncleVasya has quit IRC () 2011-06-27T06:05:18 2011-06-27T06:14:26 *** _0x47 <_0x47!~slash@p57B3FD58.dip.t-dialin.net> has joined #aichallenge 2011-06-27T06:42:29 *** mleise has joined #aichallenge 2011-06-27T06:49:32 *** stocha has joined #aichallenge 2011-06-27T06:59:35 *** sYnfo has joined #aichallenge 2011-06-27T07:04:27 *** sigh has quit IRC (Remote host closed the connection) 2011-06-27T07:09:31 *** onensora has quit IRC (Ping timeout: 244 seconds) 2011-06-27T07:10:28 *** onensora has joined #aichallenge 2011-06-27T07:18:10 mleise: http://ants.zeroviz.us/vhf1b/ :P 2011-06-27T07:22:18 antimatroid: oh my god 2011-06-27T07:22:31 i got a bit carried away :P 2011-06-27T07:23:29 I guess we can have very long games on this one :p 2011-06-27T07:23:42 i'm not sure the tools would handle the gmae 2011-06-27T07:23:45 now you can try to solve the maze 2011-06-27T07:24:22 i am working on fixing up the dimension requirements for each type of symmetry 2011-06-27T07:24:33 to make sure it doesn't wipe entire rows/cols out 2011-06-27T07:24:36 what's wrong with them? 2011-06-27T07:24:38 or the entire map in some instances 2011-06-27T07:24:43 i can do all symmetry types 2011-06-27T07:24:46 That's good! I can see the 300-like battles of spartian ant holding of thousands and thousands of their advesaries. 2011-06-27T07:24:49 just with conditions on the stuff 2011-06-27T07:25:49 like for tile maps, i need row_t and col_t to be divisible by 2, for any rotational symmetries i need even dimensions and for those that have a transpose component, then i need the ants starting locations to have either both odd or both even indexes 2011-06-27T07:25:59 it's rather messy working this shit out :P 2011-06-27T07:27:00 oh, it seems more complex than i thought 2011-06-27T07:27:20 *** UncleVasya has joined #aichallenge 2011-06-27T07:27:42 can i escape python code like an if statement onto multiple lines? 2011-06-27T07:27:53 like or then new line next condition? 2011-06-27T07:28:46 in the rediculous maze i can see some diagonal and vertical line pattern 2011-06-27T07:30:17 antimatroid: put '\' at the end of the line. 2011-06-27T07:30:32 sYnfo: cheers 2011-06-27T07:30:48 melise: that's where the symmetry has matched up 2011-06-27T07:30:53 it's not a perfect maze 2011-06-27T07:30:55 *** sigh has joined #aichallenge 2011-06-27T07:30:59 i intend to make it an even less perfect maze 2011-06-27T07:31:26 it's super fast though 2011-06-27T07:31:29 compared to my other map gen 2011-06-27T07:32:34 test 2011-06-27T07:33:40 stocha: polo 2011-06-27T07:35:41 maybe we'll nee some variation in corridor wideness. let's say it's fixed. Then 3 width 4 width and 5 width labyrynth for example 2011-06-27T07:37:29 3 width labyrinth with corridors allowed... what a corridor game 2011-06-27T07:40:54 it's suprisingly frustrating to remplace experimental ugly code/bot behavior that reach 4th place, with a far more advanced thing. (especially when the result ends up 20th :p) 2011-06-27T07:41:29 it takes about 100 games to have an idea how the bot i doing 2011-06-27T07:41:35 100 games takes hours though :'( 2011-06-27T07:43:17 seems you don't know what your bots are doing :) 2011-06-27T07:44:14 okay, these conditions are just wtf 2011-06-27T07:44:50 if i have one type of 2 player rotational symmetry, the players starting locations must be odd/odd or even/even for row/col, whereas for another one, it needs to be odd/even or eve/odd 2011-06-27T07:44:53 grrr 2011-06-27T07:45:15 sir_macelon: true. you do know what your bot is doing ? 2011-06-27T07:45:23 for me it's all a big mess :) 2011-06-27T07:45:35 the ant server is quite slow to respond http wise : / 2011-06-27T07:45:51 UncleVasya: ICQ still exists ?? 2011-06-27T07:45:58 :D 2011-06-27T07:46:38 UncleVasya has about 92 games about 15hour 2011-06-27T07:46:55 so to know a bot ranks, it take about 20 hours on the beta server. 2011-06-27T07:47:04 Is it good amount? 2011-06-27T07:47:27 UncleVasya: 100 games per second would be good :p 2011-06-27T07:47:52 i guess a very slow thinking, very slow winning very strong bot would have longer average games :p 2011-06-27T07:48:35 so, we have about 100 players right now. Let's say we get 4000 for release. And that its scale linearly. So it would be about 800 hours per 100 games. 2011-06-27T07:48:57 wow 2011-06-27T07:49:11 that's about a month, to have a vague idea how the bot is doing. 2011-06-27T07:49:17 stocha: well more or less I know, but more important is whether you know what it should be doing 2011-06-27T07:49:44 sir_macelon: so much depends on the map, when you start doing "clever" stuff : / 2011-06-27T07:50:19 sir_macelon: so i know what i want him to do. I saw sometime satifying stuff. But it really depends on so much. Map and opponent strategy for one things 2011-06-27T07:50:43 I think contest will get about 1500 contestants this time. 2011-06-27T07:50:56 I'm amazed at how powerfull avoiding fighting is right now (except for top two places) 2011-06-27T07:51:02 UncleVasya: why ? 2011-06-27T07:51:07 if i see my bot acting oddly i just write a unit test to check it and that's it, works perfectly 2011-06-27T07:51:16 UncleVasya: the game can be pretty involving in a lot of ways. 2011-06-27T07:51:38 sir_macelon: how many "features" does your bot have then ? 2011-06-27T07:51:47 do you have a unit test for each of them ? 2011-06-27T07:52:06 sir_macelon: do you have a graphic tool for hand making unit tests ? 2011-06-27T07:52:27 Oh lol, a guy at Creative Assembly talked to his boss about using D for the Total War series :p 2011-06-27T07:52:36 even desining one test for one of stocha features would be a nightmare i think. 2011-06-27T07:52:50 not to mention, i rarely keep a feature for more than a day. 2011-06-27T07:53:11 stocha: I have only one feature which my overall strategy, which applies to every situation, don't like to hardcoed stuff or any behaviour 2011-06-27T07:53:20 *which is 2011-06-27T07:53:59 sir_macelon: my testing are mostly behavior oriented right now. So i have tested let's say a few dozen of them. To try to see what i can make the bot do and look like doing. 2011-06-27T07:54:01 stocha: basically yes, currently about 20 unit tests just for battle movements 2011-06-27T07:54:12 I think in this game, there is a lot of room for "elegance" :) 2011-06-27T07:54:32 sir_macelon: wow. that need patience :) 2011-06-27T07:54:45 I create them by hand, but only with the absolutely necessary information and limit to just a small part of the map 2011-06-27T07:54:47 sir_macelon: do you think battles will be interesting in the end ? 2011-06-27T07:55:21 I am a very patient person and this is really benefitting me as you can see on the leader board 2011-06-27T07:55:22 i'm not 100% convinced by battle resolution being interesting yet. You have far more experienced in seems :) 2011-06-27T07:55:42 from previous contest I've learned that testing is the key for winning 2011-06-27T07:55:52 sir_macelon: you do very well on current maps. But there a very precious few of them. 2011-06-27T07:56:21 sir_macelon: looking at your last version, i can see that you made a great job. 2011-06-27T07:56:46 I'm more like, i test some very simple pieces of code, and see what it does. 2011-06-27T07:56:54 mainly by observing right now. 2011-06-27T07:57:04 stocha: I am waiting for more challenging opponents that should come anyway 2011-06-27T07:57:15 sir_macelon: i'm starting to work on fighting 2011-06-27T07:57:23