2011-09-27T00:01:26 *** amstan_ has quit IRC (Ping timeout: 252 seconds) 2011-09-27T00:05:30 aichallenge: janzert epsilon * rf0d0fd4 / ants/ants.py : Fix initial scores in case of multiple hills - http://git.io/5958Uw 2011-09-27T00:07:49 *** dvladim has joined #aichallenge 2011-09-27T00:09:46 aichallenge: janzert epsilon * rc73de43 / ants/ants.py : Whitespace clean up in ants.py - http://git.io/H1Ycaw 2011-09-27T00:15:11 *** dvladim has quit IRC (Ping timeout: 258 seconds) 2011-09-27T00:17:43 *** Doddzy39 has joined #aichallenge 2011-09-27T00:31:25 *** ibdknox has quit IRC (Remote host closed the connection) 2011-09-27T00:32:02 *** ibdknox has joined #aichallenge 2011-09-27T00:36:11 *** ibdknox has quit IRC (Ping timeout: 252 seconds) 2011-09-27T00:39:41 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 276 seconds) 2011-09-27T00:40:08 aichallenge: janzert epsilon * r9187e39 / website/nice.php : Fix map name display when no slash in name - http://git.io/NUFUkA 2011-09-27T01:00:47 if xathis wasn't crashing in 75% of his games he'd be doing *really* good :P 2011-09-27T01:01:13 better that #1? ;) 2011-09-27T01:01:25 haha i'll have to look into that 2011-09-27T01:01:35 *than 2011-09-27T01:01:58 yeah, super king of the world good 2011-09-27T01:03:58 *** Doddzy39 has quit IRC (Ping timeout: 252 seconds) 2011-09-27T01:17:35 okay, found thebug, but if i resubmit i'll have to climb up the rankings again, so i'll let my bot keep crashing for now :) 2011-09-27T01:22:15 I'd be surprised if it takes 10 games to climb the rankings right now ;) 2011-09-27T01:24:29 *** Doddzy39 has joined #aichallenge 2011-09-27T01:31:21 a few more but yeah, you're right, i'm just lazy :D 2011-09-27T01:33:03 anyone got any complaints/suggestions for the multi maze maps? 2011-09-27T01:33:46 also, would a "hills remaining" line on the visualiser maybe be useful? 2011-09-27T01:35:47 Seems there is a error with the windows tools, in ants.py - when there is more then one ant in the range of food, it tries to call remove_food with the incorrect number of arguments. 2011-09-27T01:46:29 is LaTeX ever going to work on the actual specs page? or should I just do what I can without it? 2011-09-27T01:49:35 *** smiley1983 has quit IRC (Ping timeout: 248 seconds) 2011-09-27T01:56:44 *** smiley1983 has joined #aichallenge 2011-09-27T02:21:25 *** smiley1983 has quit IRC (Ping timeout: 240 seconds) 2011-09-27T02:38:31 *** smiley1983 has joined #aichallenge 2011-09-27T02:51:53 *** ibdknox has joined #aichallenge 2011-09-27T02:55:04 *** Doddzy39 has quit IRC (Ping timeout: 252 seconds) 2011-09-27T02:56:55 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-09-27T03:04:15 *** onensora has joined #aichallenge 2011-09-27T03:20:22 *** mcstar has quit IRC (Quit: WeeChat 0.3.5) 2011-09-27T03:24:17 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-09-27T03:28:29 *** boegel has joined #aichallenge 2011-09-27T03:35:17 *** kaemo has quit IRC (Ping timeout: 258 seconds) 2011-09-27T03:40:20 *** Akranis has joined #aichallenge 2011-09-27T03:48:20 *** ChrisH has quit IRC (Quit: Page closed) 2011-09-27T03:49:40 *** aerique has joined #aichallenge 2011-09-27T03:53:24 *** xathis has quit IRC () 2011-09-27T04:01:31 *** berak has joined #aichallenge 2011-09-27T04:07:35 *** gattschardo has joined #aichallenge 2011-09-27T04:09:29 *** cichociemno has joined #aichallenge 2011-09-27T04:34:50 *** ibdknox has quit IRC (Remote host closed the connection) 2011-09-27T04:35:25 *** ibdknox has joined #aichallenge 2011-09-27T04:39:39 *** ibdknox has quit IRC (Ping timeout: 252 seconds) 2011-09-27T05:04:36 *** GreenTea has joined #aichallenge 2011-09-27T05:05:06 Hi! 2011-09-27T05:06:10 who knows does bot control from which hill ant will be spawned after taking food? I found no info about this in specification.. 2011-09-27T05:08:25 GreenTea: it spawns at the ant hill which spawned an ant longest ago 2011-09-27T05:08:28 it breaks ties randomly 2011-09-27T05:10:05 random is not good. and if my and is directly on the hill. then this hill can't spawn? 2011-09-27T05:10:12 * my ant 2011-09-27T05:11:53 *** crazymaze has joined #aichallenge 2011-09-27T05:15:54 GreenTea: yeah, you can control where it spawns by placing ants on hills if you want 2011-09-27T05:16:08 i didn't do it, but i'm not really sure how else it would be done that's reasonably easy? 2011-09-27T05:16:51 i guess an alternative would be to tell people how many ants they can spawn and let them execute spawn orders 2011-09-27T05:17:05 i'll run that by mcleopold 2011-09-27T05:17:20 then the starter bots could implement what the engine currently does 2011-09-27T05:17:56 *** crazymaze has quit IRC (Quit: Page closed) 2011-09-27T05:20:17 *** patrisk has joined #aichallenge 2011-09-27T05:20:21 yes, thank you, it would be much easy then balancing ants. because for example for defense i need to be sure that all ants are spawned from attacked hill. 2011-09-27T05:21:08 how do you like the maze maps? 2011-09-27T05:21:12 that's what i care about :P 2011-09-27T05:21:54 *** Doddzy39 has joined #aichallenge 2011-09-27T05:22:21 they are aMAZing) 2011-09-27T05:22:55 :P 2011-09-27T05:24:09 with this maps the game is now much close to how I imagine this game before it is started. 2011-09-27T05:24:55 yeah, i always wanted it to be like caves 2011-09-27T05:25:57 Are hills supposed to spawn on water? 2011-09-27T05:26:53 that was my bad 2011-09-27T05:27:06 In game #741, xkor-RU spawns with two ant hills on water, with one of them being surrounded 2011-09-27T05:27:10 i updated my map generator last night to add multiple hills per ant and forgot to make sure i was adding them to land squares 2011-09-27T05:27:24 i've since updated the generator on github but those maps are still there 2011-09-27T05:44:18 *** Kingpin13 has joined #aichallenge 2011-09-27T05:44:54 *** sigh has joined #aichallenge 2011-09-27T05:46:16 *** berak has quit IRC (Remote host closed the connection) 2011-09-27T05:46:22 *** Blkt has joined #aichallenge 2011-09-27T05:49:47 slight map error: hills in water: http://aichallengebeta.hypertriangle.com/visualizer.php?game=758 2011-09-27T05:58:43 *** GreenTea has quit IRC (Quit: Page closed) 2011-09-27T05:59:14 smiley1983: read up a tiny bit :) 2011-09-27T06:10:08 *** Doddzy39 has quit IRC (Ping timeout: 252 seconds) 2011-09-27T06:14:16 *** Blkt has quit IRC (Remote host closed the connection) 2011-09-27T06:44:44 just passing through, but my idea right now on spawning is give bots the ability to set the prioirity on hills 2011-09-27T06:47:37 multi_maze_11 disabled 2011-09-27T06:57:24 *** Blkt has joined #aichallenge 2011-09-27T06:58:28 *** Blkt has quit IRC (Remote host closed the connection) 2011-09-27T06:58:48 *** Blkt has joined #aichallenge 2011-09-27T07:01:59 antimatroid: ah, I see. I'm way behind on reading. 2011-09-27T07:14:09 *** Keth has quit IRC (Ping timeout: 260 seconds) 2011-09-27T07:18:03 *** Keth has joined #aichallenge 2011-09-27T07:19:04 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-09-27T07:26:06 *** kaemo has joined #aichallenge 2011-09-27T07:29:24 *** delt0r_ has quit IRC (Read error: Operation timed out) 2011-09-27T07:30:02 janzert: what if we just told bots how many ants they can spawn each turn (up to) then let them execute orders to spawn ants? 2011-09-27T07:34:18 *** olexs has joined #aichallenge 2011-09-27T07:43:26 *** delt0r_ has joined #aichallenge 2011-09-27T07:49:18 *** berak has joined #aichallenge 2011-09-27T07:49:26 *** kaemo has quit IRC (Ping timeout: 258 seconds) 2011-09-27T07:50:07 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 252 seconds) 2011-09-27T07:52:30 *** olexs1 has joined #aichallenge 2011-09-27T07:54:08 *** olexs has quit IRC (Ping timeout: 260 seconds) 2011-09-27T08:15:11 *** chris__0076 has joined #aichallenge 2011-09-27T08:18:03 *** Chris_0076 has quit IRC (Ping timeout: 260 seconds) 2011-09-27T08:24:34 *** chris__0076 is now known as chris_0076 2011-09-27T08:26:01 *** Akranis_ has joined #aichallenge 2011-09-27T08:26:53 *** mceier has joined #aichallenge 2011-09-27T08:29:24 *** Akranis has quit IRC (Ping timeout: 260 seconds) 2011-09-27T08:32:56 *** Akranis_ has quit IRC (Quit: Lämnar) 2011-09-27T08:44:31 *** Akranis has joined #aichallenge 2011-09-27T08:49:15 antimatroid: So it would look like that at beginning of turn you get number of ants you can place in any free hill and make order to them? 2011-09-27T08:51:33 antimatroid: cause normally if in turn N my ant capture food, then in turn N+1 I got one extra ant. So it will be requiring to putting ant and ordering them in the N+1 turn. 2011-09-27T08:58:35 *** Keth_ has joined #aichallenge 2011-09-27T08:58:35 *** Keth has quit IRC (Read error: Connection reset by peer) 2011-09-27T09:24:26 *** test_nickcname has joined #aichallenge 2011-09-27T09:24:46 test message, please ignore 2011-09-27T09:25:03 \quit 2011-09-27T09:25:06 *** test_nickcname has quit IRC (Client Quit) 2011-09-27T10:15:56 *** patrisk has quit IRC (Quit: Leaving) 2011-09-27T10:17:26 *** Palmik has joined #aichallenge 2011-09-27T10:24:11 *** McLeopold has joined #aichallenge 2011-09-27T10:24:21 janzert: you around? 2011-09-27T10:25:26 antimatroid: ? 2011-09-27T10:26:13 i am 2011-09-27T10:26:58 so, did you read the logs about xathis squatting trick? 2011-09-27T10:27:14 McLeopold: for spawning, have you considered making bots execute spawn moves themselves? so for the input of each turn tell them the maximum number of ants they are allowed to spawn? and only allow one ant to spawn at an ant hill each turn? 2011-09-27T10:27:50 antimatroid: yes, but that would change the bot input. I think it takes the game to the complexity that jeff doesn't want 2011-09-27T10:28:09 i was thinking the starter bots could implement what the engine does now 2011-09-27T10:28:17 then people who don't want to touch it could just leave it 2011-09-27T10:28:20 but yeah 2011-09-27T10:28:21 for casual coders, they don't need to care where the ants spawn at and they can just run around with ants, and the spawning will be fairly even 2011-09-27T10:28:56 right now, if you care to, you can control the spawning by just leaving an ant on top of the hill to block it 2011-09-27T10:29:19 that's messy to track and requires holding an ant next to hills though 2011-09-27T10:29:25 it's not really ideal :P 2011-09-27T10:29:41 it's 1 ant at each hill, not a big deal 2011-09-27T10:29:44 it would also solve the hill squating problem, although i don't think that's really much of a problem 2011-09-27T10:29:56 I already have a good solution for that 2011-09-27T10:30:14 if your ant dies on the hill, the hill is destroyed 2011-09-27T10:30:23 it doesn't actually make it any easier to kill hills 2011-09-27T10:30:45 spawning happens, then moves, then attack, so the hill owner would have to let the hill die 2011-09-27T10:30:53 sort of does if they kill it when there's a battle around the hill doesn't it? 2011-09-27T10:31:09 nope 2011-09-27T10:31:14 move before attack 2011-09-27T10:31:23 you still have total control 2011-09-27T10:31:39 if you want to stop the siphoning, you can do it 2011-09-27T10:32:47 hmmm okay, it sounds like a lot of messy edge cases would be added though 2011-09-27T10:33:06 they are edge cases, and I don't want to change the bot output for edge cases 2011-09-27T10:33:09 my guesses are, handling that, worrying about defending approaching enemies with one ant on a hill in case it gets taken out 2011-09-27T10:33:55 i don't really mind though 2011-09-27T10:34:02 in the strategy guide, we simple explain that you should leave 1 ant next to each hill in a multi-hill map to control 2011-09-27T10:34:25 it's really not that hard to implement 2011-09-27T10:34:54 if you have the logic to handle where you think spawning should occur, you should be able to control a gatekeeper 2011-09-27T10:35:11 *** sir_macelon has joined #aichallenge 2011-09-27T10:35:27 sir_macelon: does the no food bug still happen with the new engine? 2011-09-27T10:36:02 yes, I saw it happening recently also 2011-09-27T10:36:17 but I didn't note down the game id 2011-09-27T10:36:51 it is mainly related to the "hidden" food_visible parameter 2011-09-27T10:37:49 McLeopold: btw. have you noticed one worker down? 2011-09-27T10:38:00 no, I didn't 2011-09-27T10:38:11 ec2 spot prices have spiked recently 2011-09-27T10:38:51 did we finalise a launch date in the end? 2011-09-27T10:39:27 antimatroid: yes, read the logs 2011-09-27T10:39:40 McLeopold: so you there will be only one worker now? 2011-09-27T10:39:56 McLeopold: i did but you guys seemed to jump all over the place :P 2011-09-27T10:39:59 sir_macelon: just until launch 2011-09-27T10:40:25 i'm updating my c++ starter bot to have hill information now 2011-09-27T10:40:43 McLeopold: okay 2011-09-27T10:45:35 *** boegel has quit IRC (Ping timeout: 252 seconds) 2011-09-27T10:46:35 *** sigh_ has joined #aichallenge 2011-09-27T10:50:16 *** sigh has quit IRC (Ping timeout: 256 seconds) 2011-09-27T10:52:57 *** Corgano has joined #aichallenge 2011-09-27T10:55:39 aichallenge: McLeopold raze * red4239c / ants/ants.py : start of dead and razing - http://git.io/T3AE2g 2011-09-27T10:56:05 *** McLeopold has left #aichallenge 2011-09-27T10:59:09 aichallenge: Nick Ham epsilon * r637bc54 / (3 files): updated to c++ starter bot - http://git.io/H_v4OQ 2011-09-27T11:01:36 *** Corgano has quit IRC (Quit: Page closed) 2011-09-27T11:09:03 *** onensora has quit IRC () 2011-09-27T11:09:23 *** sigh_ has quit IRC (Remote host closed the connection) 2011-09-27T11:10:20 *** aerique has quit IRC (Quit: ...) 2011-09-27T11:12:02 *** sir_macelon has quit IRC (Quit: Page closed) 2011-09-27T11:20:41 *** Corgano has joined #aichallenge 2011-09-27T11:29:35 *** Palmik has quit IRC (Remote host closed the connection) 2011-09-27T11:40:47 *** ChrisH has joined #aichallenge 2011-09-27T11:42:58 *** Keth has joined #aichallenge 2011-09-27T11:43:01 *** Keth_ has quit IRC (Ping timeout: 240 seconds) 2011-09-27T11:45:49 *** mleise has joined #aichallenge 2011-09-27T11:48:48 *** Akranis has quit IRC (Read error: Connection reset by peer) 2011-09-27T11:49:05 *** ChrisH has quit IRC (Quit: Page closed) 2011-09-27T11:50:52 *** McLeopold has joined #aichallenge 2011-09-27T11:51:13 *** mceier has quit IRC (Read error: Operation timed out) 2011-09-27T11:55:40 *** mceier has joined #aichallenge 2011-09-27T12:05:05 *** ibdknox has joined #aichallenge 2011-09-27T12:08:50 since there are so few active players after the reset still, I've only started one worker back up 2011-09-27T12:09:12 McLeopold: here now 2011-09-27T12:09:20 hi 2011-09-27T12:09:41 *** ibdknox has quit IRC (Ping timeout: 260 seconds) 2011-09-27T12:09:48 janzert: so, I'm inclined to let the xathis siphon happen rather than implement any more bot output 2011-09-27T12:10:12 or change the rule so that an enemy ant in attack range will raze a hill 2011-09-27T12:10:23 nah, I'd rather leave it 2011-09-27T12:10:30 what's xathis siphon? 2011-09-27T12:10:36 let the siphon happen? 2011-09-27T12:10:53 What was the result of the launch meeting? 2011-09-27T12:11:28 Will emails be sent to everyone, to notify them of the start? 2011-09-27T12:11:47 I was thinking about adding something like "s row col priority", and the priority of each hill increases every turn by one and is reset by an ant occupying the hill 2011-09-27T12:12:26 that's too complicated 2011-09-27T12:12:35 so the everything would be the same as it is now if the bot doesn't issue an order but it can explicitly set the order if it wants to 2011-09-27T12:12:37 ? 2011-09-27T12:12:38 really 2011-09-27T12:12:39 ? 2011-09-27T12:12:46 bot output needs to be simple 2011-09-27T12:13:04 that's no more complicated than an ant move order 2011-09-27T12:13:06 this is really an edge case, and I don't want to change the fundamentals just to support an edge case 2011-09-27T12:13:50 ok, I'm still don't really see why multi-hills really add anything to the game 2011-09-27T12:14:11 janzert: it does, I plan to have a combinations tesselation 2011-09-27T12:14:43 it'd be nice of the visualizer linked to the map page 2011-09-27T12:14:47 of->if 2011-09-27T12:15:49 http://aichallengebeta.hypertriangle.com/visualizer.php?game=784&user=26 has a map with a hill in water, I think I already took that out of rotation this morning, but I'm not sure :} 2011-09-27T12:16:05 antimatroid already fixed it 2011-09-27T12:16:19 fixed the generator you mean? 2011-09-27T12:16:31 yeah, I just generated the current set before that yesterday 2011-09-27T12:17:14 does a low priority come first or last? 2011-09-27T12:17:39 ? 2011-09-27T12:17:43 low priority where? 2011-09-27T12:17:48 for the "s row col priority" 2011-09-27T12:18:08 oh, closer to 0 zero go later 2011-09-27T12:18:26 basically highest priority gets the spawn, but that will set it back to 0 2011-09-27T12:18:43 so then the next highest will get the next unless another order is issued 2011-09-27T12:18:48 so you'd have to issue an s order every turn if you want to turn a hill off 2011-09-27T12:18:50 *** xathis has joined #aichallenge 2011-09-27T12:19:17 plus, you can't really turn it off if you gather more than 1 food per turn 2011-09-27T12:19:20 *** ChrisH has joined #aichallenge 2011-09-27T12:19:21 well, every n-1 turns at least for n hills active 2011-09-27T12:19:40 only one spawn per turn right? 2011-09-27T12:19:44 no 2011-09-27T12:19:47 oh 2011-09-27T12:19:48 1 per turn per hill 2011-09-27T12:20:05 I thought it was one per turn only 2011-09-27T12:20:29 so, my scheme still makes more sense, to me at least 2011-09-27T12:21:00 *** GeorgeSebastian has joined #aichallenge 2011-09-27T12:21:21 it makes the game a little faster, by 2 moves per hill raze :) 2011-09-27T12:22:20 I thought we had decided multiple hills spawning was unwanted the other night 2011-09-27T12:22:29 no, we didn't 2011-09-27T12:22:41 what time, I'll read the logs :) 2011-09-27T12:22:50 heh, no idea :P 2011-09-27T12:24:01 oh 2011-09-27T12:24:13 janzert: I agree with you that multiple hills do not make the game "better" 2011-09-27T12:24:47 watching bot behavior the other night I wonder if we should tell a bot when a hill is dead 2011-09-27T12:24:59 It is a lot harder to follow what is going on in the game when watching the visualizer. 2011-09-27T12:25:10 when there are a lot of hills 2011-09-27T12:25:10 i.e. not just that a hill came into view but whether the hill is active or not 2011-09-27T12:25:21 it gets a hill line each turn, just like food, so the bots would have the same delima 2011-09-27T12:25:39 My current submission figures out when hills are dead 2011-09-27T12:25:56 because once a hill is dead it is no longer sent in the input when that square is visible. 2011-09-27T12:25:59 janzert: dead hills are not sent to the bots 2011-09-27T12:26:03 ahh 2011-09-27T12:26:04 ok 2011-09-27T12:26:13 just like eaten food 2011-09-27T12:26:20 I just saw some bots go and camp out on dead hills 2011-09-27T12:26:29 I thought it might be because they were still seeing it 2011-09-27T12:26:32 but I guess not 2011-09-27T12:26:37 what game? 2011-09-27T12:26:42 no, they are just remembering it. 2011-09-27T12:26:54 heh, I've watched a couple dozen 2011-09-27T12:26:55 My first version that was hill aware did that too. 2011-09-27T12:27:00 *** amstan_ has joined #aichallenge 2011-09-27T12:27:00 *** ChanServ sets mode: +o amstan_ 2011-09-27T12:27:25 ok 2011-09-27T12:27:30 probably what it was 2011-09-27T12:28:32 janzert: how about no spawning if an enemy is in range of the hill? 2011-09-27T12:29:06 hmm 2011-09-27T12:29:24 it makes defense just a little harder 2011-09-27T12:29:26 not sure one way or the other 2011-09-27T12:29:34 but only if you were on the line 2011-09-27T12:29:42 what about if it is your last hill? 2011-09-27T12:30:07 then you're about to get eliminated :) 2011-09-27T12:30:14 not necessarily 2011-09-27T12:30:32 sacrifice some ants to get it active again 2011-09-27T12:30:38 it could be a single enemy bot shutting the hill down that wouldn't be able to capture it by itself. 2011-09-27T12:30:45 depends on if you have much in the hive 2011-09-27T12:30:59 that's been my point in this discussion 2011-09-27T12:31:01 1 ant can always raze a hill 2011-09-27T12:31:08 I still don't get that. 2011-09-27T12:31:15 hmm, I see in our original conversation we had discussed using hill with closest friendly to set spawn 2011-09-27T12:31:24 we did 2011-09-27T12:31:35 but that sounds expensive for the engine to do 2011-09-27T12:31:35 but I didn't want to implement the search :) 2011-09-27T12:31:40 right :) 2011-09-27T12:32:06 sitting ants on top of hills to block just makes so much sense, plus it is easy to see in the visualizer 2011-09-27T12:32:20 jeff doesn't want hidden state, remember 2011-09-27T12:32:41 but the state is already there 2011-09-27T12:32:50 it's just not directly manipulable by the bot 2011-09-27T12:32:52 hidden from the visualizer 2011-09-27T12:32:58 it is 2011-09-27T12:34:22 hmm... 2011-09-27T12:34:38 McLeopold: I really don't want to be a pest. But I'm still not sure I understand your assertion that a single enemy ant can raze a hill with an ant on top and more in the hive ready to spawn. 2011-09-27T12:34:41 I'm fine with leaving it for now though 2011-09-27T12:35:24 McLeopold: Could you step me through how that is resolved? 2011-09-27T12:35:42 janzert: I think I'll implement my way. If it becomes a big issues, we can just drop multiple hills. 2011-09-27T12:35:44 heh, found it ;). You did say 1 spawn per turn not hill 2011-09-27T12:35:59 janzert: it must have been a typo :P 2011-09-27T12:36:11 (1:26:12 PM) McLeopold: okay, 1 per turn, not per hill, least recently spawned? 2011-09-27T12:36:16 :P 2011-09-27T12:36:26 oh, that was way back 2011-09-27T12:36:32 hehe 2011-09-27T12:36:53 I also then said 2011-09-27T12:37:02 okay, if 1 per turn, and stepping on a hill to activate, then how do you switch? 2011-09-27T12:37:02 (1:26:52 PM) janzert: ok, since I don't expect per turn or per hill to be much different in practice 2011-09-27T12:37:29 assuming a constance flow in the hive 2011-09-27T12:37:30 constant 2011-09-27T12:37:53 *** UncleVasya has joined #aichallenge 2011-09-27T12:38:24 *** Blkt has quit IRC (Remote host closed the connection) 2011-09-27T12:38:28 yeah, you'd have to get the spawned ant to survive a turn and step back off with the other 2011-09-27T12:38:49 ChrisH: http://ietherpad.com/Rz6lUKCSZn 2011-09-27T12:38:55 I'm convinced "most recently spawned" won't work terribly well 2011-09-27T12:39:03 I don't think so either 2011-09-27T12:39:11 if we need to change I'd rather add an explicit order for the bot to issue 2011-09-27T12:40:03 that's what I'm trying to avoid 2011-09-27T12:40:30 I like to think of the ant hills as pressure values that you can block :) 2011-09-27T12:40:41 I don't see it as that horrible of a thing, not much worse than multi-hill maps ;) 2011-09-27T12:40:41 *** dvladim has joined #aichallenge 2011-09-27T12:40:51 hehe 2011-09-27T12:41:45 I'm liking fairly significant initial food placement now 2011-09-27T12:41:58 like more initial food? 2011-09-27T12:42:10 10-20? 2011-09-27T12:42:28 We could just start with 10 in the hive 2011-09-27T12:42:29 absolute depends on the map size :) 2011-09-27T12:42:41 nah, having it spread out is better 2011-09-27T12:42:56 *** meduza has joined #aichallenge 2011-09-27T12:43:43 maybe 1 per 100 squares of land area or so 2011-09-27T12:45:57 *** |UncleVasya| <|UncleVasya|!kvirc@46.133.69.155> has joined #aichallenge 2011-09-27T12:46:23 *** UncleVasya has quit IRC (Read error: Connection reset by peer) 2011-09-27T12:49:58 not sure if this is purely a visualizer thing, but I wonder if it makes sense to not show an ants hive size after its last hill is destroyed 2011-09-27T12:50:10 since it can never spawn those anyway 2011-09-27T12:54:53 here's one where paratrechina seems to hover on and around a dead hill http://antsbeta.janzert.com/visualizer.php?game=955 2011-09-27T12:55:05 from about move 179 till almost the end of the game 2011-09-27T12:55:10 *** nux67 has joined #aichallenge 2011-09-27T12:55:12 the lower left hill 2011-09-27T12:55:33 *** Corgano has quit IRC (Quit: Page closed) 2011-09-27T12:58:51 janzert: raze happens after attack, and you must have a surviving ant 2011-09-27T12:59:10 hmm? 2011-09-27T12:59:18 janzert: so, if you can maintain ants in the hive, or always gather enough food, your hill can't be razed 2011-09-27T12:59:34 since an new ant will always spawn just before move 2011-09-27T12:59:55 if you don't move it off, an enemy can't move on to the hill 2011-09-27T13:00:05 but if the attacker has two ants adjacent neither will die and one can step on 2011-09-27T13:00:12 if it does, it collides and doesn't survive to the raze phase 2011-09-27T13:00:18 ahh 2011-09-27T13:00:20 cool 2011-09-27T13:00:30 nope, an new ant always spawns 2011-09-27T13:00:50 I had thought it would work out that way until someone convinced me otherwise :P 2011-09-27T13:01:12 so, is this good? :) 2011-09-27T13:01:15 I'm happier with it that way by the way 2011-09-27T13:01:23 yes, I think so 2011-09-27T13:01:45 if I implement my idea, it break 2011-09-27T13:01:46 s 2011-09-27T13:02:10 the stopping the hill from spawning if the ant dies idea? 2011-09-27T13:02:25 actually, its a virtual win for the attackers, given enough time 2011-09-27T13:02:28 yeah, I think probably just leave it alone for now 2011-09-27T13:02:59 to move, or not to move 2011-09-27T13:03:22 if you move off, and the opponent moves on, you lose 2011-09-27T13:03:35 if you stay and the opponent stays, you lose 2011-09-27T13:03:46 so you need to stay on the hill and hope you can survive until reinforcements arrive 2011-09-27T13:03:48 there is no nash equilibrium 2011-09-27T13:04:09 I don't understand that first scenario statement 2011-09-27T13:04:11 it's just a guessing game, but once the attacker guesses right, the game is over and he wins 2011-09-27T13:04:40 unless external reinforcements (from another hill) can save the day. 2011-09-27T13:04:40 you can't move off before dieing 2011-09-27T13:04:54 can you? 2011-09-27T13:05:03 move before attack 2011-09-27T13:05:15 spawn on hill -> move -> attack phase 2011-09-27T13:05:24 the xathis siphon was predicated on there being a small number of enemy ants squatting on the hill 2011-09-27T13:05:58 so if there are some friendly ants that can maneuver to force them out of the area or kill them, then the squatted hill could be liberated. 2011-09-27T13:06:24 currently, that is 2011-09-27T13:06:30 yes, now I've changed my mind on that possibility 2011-09-27T13:06:43 but, if the bot doesn't want to try and save the hill? 2011-09-27T13:06:46 I'm glad we agree. 2011-09-27T13:06:49 if it would rather have the ants? 2011-09-27T13:07:05 not sure what to do there. 2011-09-27T13:07:27 oh, I know 2011-09-27T13:07:30 othere than allowing the bot to control spawning as someone suggested 2011-09-27T13:07:42 how about a hill won't spawn if there is a dead ant on it 2011-09-27T13:07:54 if that needs to be fixed then I still think an explicit order is the best most straightforward solution 2011-09-27T13:08:29 *** eashoka has joined #aichallenge 2011-09-27T13:08:30 how does a dead ant get removed? 2011-09-27T13:08:34 that's basically back to your first suggestion then McLeopold 2011-09-27T13:08:41 isn't it? 2011-09-27T13:09:16 assuming dead ants disappear in one of the phases each turn 2011-09-27T13:10:27 or, if the enemy chooses not to raze the hill, then it lowers the hill's spawn rate to every other turn. 2011-09-27T13:12:49 no, it doesn't kill the hill 2011-09-27T13:12:59 janzert: dead ants only last 1 turn 2011-09-27T13:13:24 not exactly, but it gives the enemy a chance to rzae the hill after a 1 turn delay. 2011-09-27T13:13:25 yeah, lowers the spawn rate 2011-09-27T13:13:29 means keep some ants in the hive is no longer a last ditch hill defense 2011-09-27T13:14:30 lets just leave it alone for the moment 2011-09-27T13:14:42 ok 2011-09-27T13:14:42 siphon is in 2011-09-27T13:15:10 that seems like the best decision to me. 2011-09-27T13:15:16 (as if you cared) 2011-09-27T13:15:34 it's got some cool consequences 2011-09-27T13:15:38 but I'm with janzert about multiple hills being less interesting 2011-09-27T13:16:20 just wait on that, we can always not include maps, but rules changes need to be fixed now 2011-09-27T13:16:38 interesting http://pastebin.com/Pesgywzv 2011-09-27T13:16:39 I agree completely on that point. 2011-09-27T13:16:43 the hill siphone also works on 1 hill maps 2011-09-27T13:16:50 *** onensora has joined #aichallenge 2011-09-27T13:16:54 plus the last ditch effort 2011-09-27T13:17:02 no disagreement from me there McLeopold 2011-09-27T13:17:27 that's the count of games for each end condition currently 2011-09-27T13:17:45 cool, I'm glad some rank cutoffs are in there 2011-09-27T13:18:03 quite a few really 2011-09-27T13:18:44 did you look at the games? 2011-09-27T13:19:03 especially considering it wasn't working until not that long ago 2011-09-27T13:19:09 I've seen a few 2011-09-27T13:19:28 but the siphon issue is less of a problem on single hill maps because it doesn't allow an enemy to squat on a highly exposed hill to harm an otherwise well protected colony, and the multi-hill maps tend to have a lot of difficult to defend hills. 2011-09-27T13:19:32 it'd be great for the visualizer and game list to start displaying it 2011-09-27T13:20:08 janzert: just to make sure we agree -> if there is a bot with a hill that can still gain in rank, then the game continues, otherwise it ends 2011-09-27T13:20:10 and now I'll be quite about that topic. ;) 2011-09-27T13:20:25 err, quiet 2011-09-27T13:20:53 if there is only one bot that has a hill remaining then the game ends immediately 2011-09-27T13:21:05 because it can't gain in rank 2011-09-27T13:21:16 right 2011-09-27T13:21:35 okay, we agree, last night I think we argued the same thing :) 2011-09-27T13:21:36 *** meduza has quit IRC (Quit: Page closed) 2011-09-27T13:21:44 *** ibdknox has joined #aichallenge 2011-09-27T13:22:02 your scenario last night didn't make sense to me 2011-09-27T13:22:17 * McLeopold is deleting channel logs... 2011-09-27T13:22:26 :) 2011-09-27T13:22:58 it would seem possible for the ranks of the other bots to shuffle if they could kill some hills though, doesn't it? 2011-09-27T13:23:13 did you see my bug report about bots not losing points for hills razed after their own death 2011-09-27T13:23:23 yes, but if they lost their hills, we deny them the chance 2011-09-27T13:23:31 no 2011-09-27T13:23:35 ok, as long as that is in the rules. 2011-09-27T13:23:53 janzert: that is a visualizer thing 2011-09-27T13:23:56 https://github.com/aichallenge/aichallenge/issues/265 2011-09-27T13:23:58 ok 2011-09-27T13:23:59 i noticed that too janzert 2011-09-27T13:24:02 wasn't sure 2011-09-27T13:24:23 I cut off the score history to fix the graph, but I think mleise uses something else now 2011-09-27T13:24:51 they do seem to lose 1 point at the moment of elimination though 2011-09-27T13:24:56 not yet, no. But you said you added something, right? 2011-09-27T13:25:06 yes 2011-09-27T13:25:26 actually, I stop adding to the score history, the modify the last one to be what it was at the end 2011-09-27T13:25:42 so it looks like they lose all the points at the point of death 2011-09-27T13:25:42 which is something I like 2011-09-27T13:26:10 mleise: two feature requests too, I can add them in the bug tracker if you like, a link to the map page and display of the cutoff reason 2011-09-27T13:26:26 janzert: let's move that out of the visualizer? 2011-09-27T13:26:31 yeah probably 2011-09-27T13:26:58 McLeopold: hmm, then the score isn't being calculated properly 2011-09-27T13:27:09 janzert: do you have a sample game? 2011-09-27T13:28:18 * janzert looks for one 2011-09-27T13:28:29 aichallenge: McLeopold epsilon * r92fee6d / ants/ants.py : simplify starting score - http://git.io/ewGWgA 2011-09-27T13:32:04 janzert: The cutoff reason is part of the replay and I think it makes sense to see it in the visualizer. Where would I display it though? 2011-09-27T13:32:07 http://antsbeta.janzert.com/visualizer.php?game=954 2011-09-27T13:32:40 mleise: I'm going to add stuff to the page itself, so you can't wait if you want. 2011-09-27T13:33:08 hmm, maybe that game isn't an example 2011-09-27T13:33:28 mleise: I was thinking maybe between the names and scores 2011-09-27T13:33:34 but I'm not sure 2011-09-27T13:33:37 McLeopold: I *can't wait* if I want? I'll wait anyway. 2011-09-27T13:34:45 okay, I get an 'F' in typing :) 2011-09-27T13:34:55 multi_maze_23 has a water bound hill at row 85, col 97 2011-09-27T13:35:23 and other symetric locations 2011-09-27T13:36:01 ok, the games I'm seeing now seem to have the correct scores but I could swear their were at least a couple last night that were wrong :/ 2011-09-27T13:36:13 *** eashoka has quit IRC (Quit: Page closed) 2011-09-27T13:37:35 could have been 2011-09-27T13:37:44 I was fast and loose with the updates last night 2011-09-27T13:39:42 *** amstan_ has quit IRC (Ping timeout: 255 seconds) 2011-09-27T13:41:48 *** ChrisH has quit IRC (Quit: Page closed) 2011-09-27T13:44:24 *** okayzed is now known as okay| 2011-09-27T13:45:27 looks like the rank change cutoff is still taking into account ants without hills? 2011-09-27T13:46:21 http://antsbeta.janzert.com/visualizer.php?game=992 2011-09-27T13:46:55 *** chris_0076 has quit IRC (Read error: Operation timed out) 2011-09-27T13:47:44 I see 2 hills to the end 2011-09-27T13:47:47 *** Chris_0076 has joined #aichallenge 2011-09-27T13:48:09 mleise: can we have hill raze events in the timeline? 2011-09-27T13:48:29 the rank change cutoff should kick in as soon as medrimonia's hill is destroyed, but it waits until his ants are also killed 2011-09-27T13:48:35 *** Palmik has joined #aichallenge 2011-09-27T13:49:17 oh, let me check that 2011-09-27T13:50:01 *** GeorgeSebastian has quit IRC (Quit: Leaving) 2011-09-27T13:50:53 McLeopold: yes we can! 2011-09-27T13:51:33 I wish the add_maps_to_database script alphabatized the list before adding them :/ 2011-09-27T13:51:48 I'm disabling multi_maze_04 it has hills in water 2011-09-27T13:51:59 3 or 4 sides? 2011-09-27T13:52:08 all 4 sides 2011-09-27T13:52:15 http://antsbeta.janzert.com/map.php?map=multi_hill_maze/multi_maze_04.map 2011-09-27T13:52:56 row 18, col 27 2011-09-27T13:53:00 yep 2011-09-27T13:53:12 mleise: the hills look so good, even zoomed in 2011-09-27T13:53:21 can't you make the dirt and water look that good? 2011-09-27T13:53:25 :P 2011-09-27T13:53:39 the hills *are* really sweet 2011-09-27T13:54:47 I don't think I ever zoomed into the maximum before looking at the hills up close 2011-09-27T13:58:12 smiley updated the OCaml starter (good) but did it by first deleting the whole bot in one commit then adding it back in a second commit :( 2011-09-27T13:59:56 The hills look so good because everything else looks so bad :p 2011-09-27T14:00:28 janzert: that game should have ended http://pastebin.com/ww8Q9h3A 2011-09-27T14:01:21 *** amstan_ has joined #aichallenge 2011-09-27T14:01:21 *** ChanServ sets mode: +o amstan_ 2011-09-27T14:01:26 *** amstan_ has quit IRC (Remote host closed the connection) 2011-09-27T14:01:32 hmm 2011-09-27T14:02:03 unless the score was wrong, or the hill objects where wrong 2011-09-27T14:04:43 http://antsbeta.janzert.com/map.php?map=multi_hill_maze/multi_maze_23.map another map with hills in water. see row 48 col 48 2011-09-27T14:05:13 yep, thanks 2011-09-27T14:05:14 McLeopold: were you looking for such gfx detail? http://imageshack.us/photo/my-images/32/bildschirmfotoen.png/ 2011-09-27T14:06:14 yes :D 2011-09-27T14:06:17 also, games on such maps (with many players and multiple spawns per player spread everywhere) are very difficult to watch properly. 2011-09-27T14:06:30 difficult how? 2011-09-27T14:06:53 when the map is big, one can only just see the hills 2011-09-27T14:07:14 there is little indication in the moment where a hill is destroyed, so one has to scroll back and forth through the timeline 2011-09-27T14:07:48 find the moment where the score increased, and then search the map to see where it happened. ants of all 8 colors spread evenly across the screen don't help :) 2011-09-27T14:07:48 maybe a giant ping on the map that lasts a few seconds? 2011-09-27T14:07:53 I know what you want! You want a proximity indicator on the map that shows up when an enemy ant approaches a hill 2011-09-27T14:09:04 any of those two ideas would help 2011-09-27T14:09:55 for the proximity indicator, first thing that comes to mind is Air Control game for iphone/android/..., if you know what I'm talking about 2011-09-27T14:10:02 McLeopold: that pastebin shows false for me 2011-09-27T14:10:16 which means the rank can't change 2011-09-27T14:10:17 oh that's what it should show :P 2011-09-27T14:10:25 yeah, reversed the boolean 2011-09-27T14:10:42 the setup of it certainly seems to match the game too 2011-09-27T14:10:44 maybe I should have done that the other way :) 2011-09-27T14:11:02 is_rank_stabalized() 2011-09-27T14:11:09 I think I've actually stumbled on that a couple times now, it'll stick in my mind soon :P 2011-09-27T14:11:26 I'm going to change it 2011-09-27T14:11:45 *** amstan has joined #aichallenge 2011-09-27T14:11:45 *** ChanServ sets mode: +o amstan 2011-09-27T14:12:19 I think I'm going to update the worker, cause I really can't see why that game didn't end 2011-09-27T14:12:46 aichallenge: McLeopold epsilon * rb02125e / ants/ants.py : restore janzert's sanity - http://git.io/AhLxmA 2011-09-27T14:12:53 :) 2011-09-27T14:14:10 on a 4 player map, where 1 player kills 2 hills, the game should always end 2011-09-27T14:14:27 * janzert nods 2011-09-27T14:14:49 can I rename the maps? 2011-09-27T14:15:07 it'll break game lists 2011-09-27T14:15:11 or for launch, add the player count in the name 2011-09-27T14:15:29 yeah 2011-09-27T14:15:42 for the open map I went with number of players and size 2011-09-27T14:16:50 about maps... since the game now is quite similar to "proper" strategy games, one could try and replicate a few "classic" maps from starcraft, c&c and others 2011-09-27T14:17:05 the symmetric ones, there are a lot of those. would add some fun factor 2011-09-27T14:17:14 jeff wanted hunters 2011-09-27T14:17:36 hopefully the community will get involved in map making again this contest 2011-09-27T14:19:03 *** ibdknox has quit IRC (Remote host closed the connection) 2011-09-27T14:19:07 http://antsbeta.janzert.com/visualizer.php?game=855&user=26 ChrisH loses the hill, but still gets some points :) 2011-09-27T14:19:37 *** ibdknox has joined #aichallenge 2011-09-27T14:19:54 worker updated to latest 2011-09-27T14:20:19 we need to shove the game replays into mongodb so I can do some searching :) 2011-09-27T14:22:58 hey, new users show up on top of the rankings list 2011-09-27T14:22:58 as a ghost 2011-09-27T14:23:03 :) that'd be interesting 2011-09-27T14:23:08 I noticed that before 2011-09-27T14:23:43 *** ibdknox has quit IRC (Ping timeout: 248 seconds) 2011-09-27T14:24:43 http://antsbeta.janzert.com/visualizer.php?game=992 the cutoff worked? 2011-09-27T14:24:47 I need to make a map list page now 2011-09-27T14:25:08 that's the one it didn't work 2011-09-27T14:25:26 it should end at turn 190 2011-09-27T14:26:02 dang, I need reproduce the games 2011-09-27T14:26:49 janzert: can you check in the open maps? 2011-09-27T14:27:05 for a broken example? 2011-09-27T14:27:50 heh, there's some really, umm, not smart bots :P 2011-09-27T14:27:53 http://antsbeta.janzert.com/visualizer.php?game=999 2011-09-27T14:28:20 three starters 2011-09-27T14:28:36 yeah 2011-09-27T14:28:37 and a random 2011-09-27T14:28:47 http://antsbeta.janzert.com/visualizer.php?game=970 it works right here 2011-09-27T14:29:54 http://antsbeta.janzert.com/visualizer.php?game=967 is an example of it broken 2011-09-27T14:30:21 should end at turn 241 2011-09-27T14:30:39 so, about those open maps? can you commit them? 2011-09-27T14:30:52 sure, it's just the one 2011-09-27T14:31:39 *** boegel has joined #aichallenge 2011-09-27T14:31:39 I just put it in the maps root on the server, do you want it in a subdirectory? 2011-09-27T14:31:52 I don't care 2011-09-27T14:33:02 aichallenge: janzert epsilon * r16b647e / ants/maps/open_4_98.map : Add open map with no water - http://git.io/OdGIZA 2011-09-27T14:33:37 want me to look for more broken examples or are you set for now? 2011-09-27T14:33:38 *** ibdknox has joined #aichallenge 2011-09-27T14:34:01 *** okay| is now known as okayzed 2011-09-27T14:34:18 I think I'm good 2011-09-27T14:34:28 *** ibdknox has quit IRC (Remote host closed the connection) 2011-09-27T14:34:59 *** ibdknox has joined #aichallenge 2011-09-27T14:35:11 http://antsbeta.janzert.com/visualizer.php?game=955 should have ended at 231 went till 838 :( 2011-09-27T14:35:23 same situation 2011-09-27T14:37:37 http://antsbeta.janzert.com/visualizer.php?game=777 that game worked and was played before the others, so I don't think it was old code on the worker causing it 2011-09-27T14:39:28 *** ibdknox has quit IRC (Ping timeout: 260 seconds) 2011-09-27T14:44:54 *** kaemo has joined #aichallenge 2011-09-27T14:46:04 *** berak_ has joined #aichallenge 2011-09-27T14:48:47 *** berak has quit IRC (Ping timeout: 248 seconds) 2011-09-27T14:49:00 *** berak_ is now known as berak 2011-09-27T14:54:07 *** amstan has quit IRC (Ping timeout: 248 seconds) 2011-09-27T14:55:13 *** arlequ1 has joined #aichallenge 2011-09-27T14:55:15 *** amstan has joined #aichallenge 2011-09-27T14:55:40 *** ChanServ sets mode: +o amstan 2011-09-27T15:01:08 janzert: got a repeatable game where the issue happens 2011-09-27T15:01:20 great 2011-09-27T15:06:03 aichallenge: Marco Leise epsilon * r304980b / (7 files in 4 dirs): added support for "playerturns" array in storage format replays - http://git.io/dg5AIA 2011-09-27T15:13:59 janzert: I think I know the error 2011-09-27T15:14:32 I'm short circuiting when I shouldn't, maybe 2011-09-27T15:14:50 or, since I revered the logic, maybe 2011-09-27T15:15:16 well, reversing it was stupid 2011-09-27T15:15:25 *** dvladim has quit IRC (Ping timeout: 240 seconds) 2011-09-27T15:15:29 where? 2011-09-27T15:15:52 can_rank_change/is_rank_stabilized 2011-09-27T15:16:18 I don't see how reversing it the way you did can make a functional difference? 2011-09-27T15:16:48 it simply reversed what is returned and the test the return value is used in 2011-09-27T15:17:13 so I'm pretty sure if it's wrong now it was wrong before 2011-09-27T15:17:45 not with a short circuit 2011-09-27T15:17:53 I'm trying to prove the game should continue 2011-09-27T15:18:02 not prove it should stop 2011-09-27T15:18:34 I'm looking for 1 case where a bot can overtake, and can stop if I found 1 2011-09-27T15:18:44 otherwise, I need to prove everyone can't 2011-09-27T15:18:45 *** ibdknox_ has joined #aichallenge 2011-09-27T15:19:05 you're just testing for true at the return instead of false though 2011-09-27T15:19:24 *** ibdknox_ has quit IRC (Remote host closed the connection) 2011-09-27T15:19:26 before the short circuit would do the same and you'd look for the opposite boolean value 2011-09-27T15:19:58 I need to think... 2011-09-27T15:20:00 *** ibdknox_ has joined #aichallenge 2011-09-27T15:20:50 <|UncleVasya|> Is Hill sent only one time (like water) or every turn? 2011-09-27T15:20:53 btw, reading the current code I still don't see where any short circuit is wrong in the method 2011-09-27T15:21:17 My understanding is it is like food and is sent every turn 2011-09-27T15:21:26 while the hill is active that is 2011-09-27T15:22:31 I think I see a problem, although not necessarily the problem in these games though 2011-09-27T15:22:58 nevermind it's not what I thought 2011-09-27T15:23:02 <|UncleVasya|> And if it is in the fog of war but we already saw it? 2011-09-27T15:23:04 *** okayzed is now known as okay| 2011-09-27T15:23:15 |UncleVasya|: same delima as with food 2011-09-27T15:23:20 just like food you won't get info about it 2011-09-27T15:23:30 <|UncleVasya|> thanks 2011-09-27T15:24:31 *** ibdknox_ has quit IRC (Ping timeout: 252 seconds) 2011-09-27T15:24:31 McLeopold: I also have a feeling the problem isn't in is_rank_stabilized itself since your extract of that returned the correct result 2011-09-27T15:25:00 *** xathis has quit IRC () 2011-09-27T15:25:27 I think it's the has_hills function 2011-09-27T15:26:38 ooh, yeah 2011-09-27T15:27:10 it should be hill.owner == player and hill.killed_by is None right? 2011-09-27T15:28:26 hmm, I didn't realize that method existed last night when I modified remaining_players() 2011-09-27T15:28:31 janzert: i've found the problem with the erlang starter package - only two of three files get compiled 2011-09-27T15:28:56 although what I wrote should be better performing than calling has_hills on each player ;) 2011-09-27T15:29:05 could you change the TargetCompiler to ExternalCompiler in compile.sh? 2011-09-27T15:29:07 oh 2011-09-27T15:29:29 you can look at the output in my account, it's "ydinic" 2011-09-27T15:29:41 any idea why isn't TargetCompiler working though? 2011-09-27T15:29:48 no, no idea 2011-09-27T15:29:58 seems to work fine for c++, i tried that starter package 2011-09-27T15:30:00 janzert: I didn't notice the change to remaining players, I'd like to undo that 2011-09-27T15:30:31 McLeopold: why? it'll break game_over if just undone 2011-09-27T15:30:35 maybe i just gave it wrong options, although i don't see that 2011-09-27T15:30:48 and finish_game 2011-09-27T15:30:56 which are the only two places it is used 2011-09-27T15:31:29 *** delt0r_ has quit IRC (Ping timeout: 276 seconds) 2011-09-27T15:32:20 gattschardo: hmm, I'll try and look at it 2011-09-27T15:32:39 janzert: thanks 2011-09-27T15:33:12 might not get time for a several hours though :( 2011-09-27T15:33:27 janzert: it won't make any difference 2011-09-27T15:33:35 except the cutoff reason will now be correct 2011-09-27T15:33:40 it does, though 2011-09-27T15:34:02 the lone survivor is calculated wrong if you take it out and doesn't get the remaining hill scores 2011-09-27T15:34:14 the hills are already done 2011-09-27T15:34:34 the case is that hills are left but not ants 2011-09-27T15:34:34 hmm... 2011-09-27T15:35:00 i.e. a few hills without ants and a few ants without hills 2011-09-27T15:35:07 that's why I changed it 2011-09-27T15:35:33 janzert: i just looked at the target compiler -- it's just the wrong thing to use for erlang, because it passes a "-o" flag on the command line 2011-09-27T15:36:04 so just changing it to the ExternalCompiler is the correct fix 2011-09-27T15:36:07 so it could lose? 2011-09-27T15:36:49 if the game went on, but the game shouldn't continue since all the other bots alive don't have a hill 2011-09-27T15:37:19 at least it could still theoretically go down in ranking if the game played out 2011-09-27T15:38:09 gattschardo: that is just a default argument so you can overide it if wanted 2011-09-27T15:38:25 yeah, but it doesn't need any 2011-09-27T15:38:34 for Erlang just "" is appropriate then 2011-09-27T15:38:46 I can add and commit that now if you want 2011-09-27T15:38:47 okay, you can also use that 2011-09-27T15:38:55 yep that would be great 2011-09-27T15:39:04 *** Zimmux has joined #aichallenge 2011-09-27T15:39:18 janzert: the bot with both hills and ants still needs to find and destroy the hill before the rank is stabilized 2011-09-27T15:39:37 it's not a given 2011-09-27T15:39:50 you handing the win over to a bot that doesn't deserve it 2011-09-27T15:40:30 the game should only be decided to continue by ants that still have active hills, that was the compromise between not eliminating them entirely 2011-09-27T15:40:40 and was your idea ;) 2011-09-27T15:40:41 close, but wrong 2011-09-27T15:41:00 and I like it so much :P 2011-09-27T15:41:02 the ant may not be in the lead 2011-09-27T15:41:08 bot 2011-09-27T15:41:27 doesn't have to be, it should just be considered the lone survivor at that stage 2011-09-27T15:41:32 so, there are still 2 bots alive, 1 with a hill, 1 without *and* the rank is not stabilized 2011-09-27T15:42:00 the only bot with a hill can still overtake, so the game continues 2011-09-27T15:42:07 it'll cut out a lot of pointless moves 2011-09-27T15:42:22 it gives a win to a dumb bot 2011-09-27T15:42:32 you consider it lone survivor and the game ends and it gets the points for all remaining hills 2011-09-27T15:42:41 that's what the lone survivor ending is 2011-09-27T15:42:49 not in my book 2011-09-27T15:43:26 but it explicitly is already, look at finish_game (I didn't even write that code) 2011-09-27T15:43:29 :) 2011-09-27T15:43:31 *** delt0r_ has joined #aichallenge 2011-09-27T15:43:54 you rewrote remaining players :P 2011-09-27T15:44:05 *** _flag <_flag!~flag@69.165.173.172> has joined #aichallenge 2011-09-27T15:44:29 bot A razes B, B eliminates A's ants, bot C sits in a corner and wins? 2011-09-27T15:44:31 but it is the same outcome as if the game continues and it does destroy all the remaining hills 2011-09-27T15:44:48 yes, but it still needs to do that 2011-09-27T15:45:11 the bot without a hill that has ants is in the lead 2011-09-27T15:45:20 but it already showed that it can hang onto ants and hills 2011-09-27T15:45:32 I think the above scenario is going to be quite rare 2011-09-27T15:45:39 or that it did get any attention 2011-09-27T15:46:02 and the normal case is going to be that a lot of pointless turns are eliminated 2011-09-27T15:46:07 okay, then let's do it my way :) 2011-09-27T15:46:31 and it has the nice compromise position of not allowing bots without hills to prolong the game 2011-09-27T15:46:59 but not actually just eliminating them 2011-09-27T15:47:11 I really like that part 2011-09-27T15:47:15 I don't 2011-09-27T15:47:29 then let's just eliminate them :P 2011-09-27T15:47:31 If you gained the lead, you should *always* be given the opportunity to maintain it 2011-09-27T15:47:55 which means the game will go the full length 2011-09-27T15:47:59 then hang onto your hill ;) 2011-09-27T15:48:09 I just don't agree 2011-09-27T15:48:25 nope if they are eliminated when the hill is dead then we have lone survivor again 2011-09-27T15:48:38 We are saving so many turns with cutoff and max_turn per map, we should stop messing up the rules just for a few cycles 2011-09-27T15:48:52 <_flag> What timezone is the server in? And shouldn't that be localized? Perhaps something like "x hous ago" 2011-09-27T15:49:06 server is on GMT now 2011-09-27T15:49:20 amstan: ? 2011-09-27T15:50:59 I'm almost sure the "compromise" position always ends the game at the same scenarios eliminating players does 2011-09-27T15:51:33 but allows bots that lost their hills to keep moving their ants and attempting to gain points in the meantime 2011-09-27T15:51:43 I'm not worried about what normally happens, I don't want to hand a game to a bot that doesn't deserve it, it screws up the rankings with noise 2011-09-27T15:52:08 there are always going to be bots that get games that don't deserve it 2011-09-27T15:52:29 We should decrease that number, not increase it 2011-09-27T15:53:25 this isn't going to effect that part a whole lot either way though, and I'm willing to add a little noise for more samples 2011-09-27T15:53:52 *** nux67 has quit IRC (Ping timeout: 252 seconds) 2011-09-27T15:55:35 we already compromised on not allowing bots without hills to try and gain rank 2011-09-27T15:55:42 *** amstan has quit IRC (Ping timeout: 256 seconds) 2011-09-27T15:55:46 this is too far 2011-09-27T15:56:01 now your handing the win to a bot not in the lead 2011-09-27T15:56:31 I'd rather go back to any bot is given the chance to overtake anyone 2011-09-27T15:57:21 unfortunately I've gotta go and get ready for work, I probably won't be back for about 3-4 hours 2011-09-27T15:57:41 cya 2011-09-27T15:59:21 aichallenge: janzert epsilon * r51c8c47 / worker/compiler.py : Update Erlang compilation - http://git.io/ckdkaw 2011-09-27T16:00:40 *** Accoun has quit IRC () 2011-09-27T16:05:08 *** Blkt has joined #aichallenge 2011-09-27T16:05:48 *** smiley1983 has quit IRC (Ping timeout: 255 seconds) 2011-09-27T16:06:15 good evening everyone 2011-09-27T16:06:18 *** smiley1983 has joined #aichallenge 2011-09-27T16:17:47 *** Accoun has joined #aichallenge 2011-09-27T16:18:46 janzert: that didn't do the trick, it didn't compile at all... i guess it's best to just use ExternalCompiler which works fine for all the other languages 2011-09-27T16:19:14 *** boegel has quit IRC (Quit: Leaving) 2011-09-27T16:23:28 aichallenge: McLeopold epsilon * r357a091 / ants/ants.py : fix for wrong cutoff, testing rule difference - http://git.io/088dbw 2011-09-27T16:23:28 aichallenge: McLeopold epsilon * r47cd4ff / (8 files in 5 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/RqUPBw 2011-09-27T16:23:31 *** Zimmux has quit IRC (Quit: Page closed) 2011-09-27T16:35:12 *** amstan has joined #aichallenge 2011-09-27T16:35:37 *** ChanServ sets mode: +o amstan 2011-09-27T16:40:42 *** cody__ has joined #aichallenge 2011-09-27T16:42:46 *** amstan_ has joined #aichallenge 2011-09-27T16:42:46 *** ChanServ sets mode: +o amstan_ 2011-09-27T16:43:08 McLeopold: hmm? 2011-09-27T16:43:31 *** Kingpin13 has quit IRC (Ping timeout: 245 seconds) 2011-09-27T16:47:12 *** okay| is now known as okayzed 2011-09-27T16:48:18 the games would be more quicker and more decisive if all dying ants in the attack radius where to turn into food... 2011-09-27T16:52:38 it would at least stop stale mates...and long drwan battels.. 2011-09-27T16:59:20 amstan_: janzert and I disagree about a rule, so you are the deciding vote 2011-09-27T16:59:48 *** asmoo has joined #aichallenge 2011-09-27T17:00:32 amstan_: should a bot that is winning and has also lost its hills be allowed to continue to defend its lead? 2011-09-27T17:00:41 *** UncleVasya has joined #aichallenge 2011-09-27T17:02:16 *** |UncleVasya| <|UncleVasya|!kvirc@46.133.69.155> has quit IRC (Ping timeout: 244 seconds) 2011-09-27T17:02:30 *** Keth has quit IRC (Ping timeout: 255 seconds) 2011-09-27T17:05:14 aichallenge: Marco Leise epsilon * r1c0faca / (4 files in 2 dirs): added hive proximity indicators - http://git.io/qKGWAw 2011-09-27T17:05:44 *** cody__ has quit IRC (Ping timeout: 252 seconds) 2011-09-27T17:05:53 *** UncleVasya has quit IRC (Ping timeout: 244 seconds) 2011-09-27T17:06:42 *** okayzed is now known as okay| 2011-09-27T17:07:56 *** Keth has joined #aichallenge 2011-09-27T17:09:28 *** mceier has quit IRC (Quit: leaving) 2011-09-27T17:09:39 *** gattschardo has quit IRC (Quit: leaving) 2011-09-27T17:10:09 aichallenge: Marco Leise epsilon * r8c976d2 / ants/visualizer/js/Application.js : regression: undecorated replays don't work - http://git.io/UmTj1A 2011-09-27T17:10:17 *** _flag <_flag!~flag@69.165.173.172> has quit IRC (Ping timeout: 276 seconds) 2011-09-27T17:12:21 mleise: that's pretty sweet, but I think i'd fade them out to invisible unless within the view radius 2011-09-27T17:15:55 McLeopold: yes, it should 2011-09-27T17:17:31 amstan_: okay, question 2 2011-09-27T17:18:09 amstan_: should a bot that loses its hills be killed, so that the ants just stay put 2011-09-27T17:18:24 we'd make the game go faster and get more games in 2011-09-27T17:18:30 no 2011-09-27T17:18:41 \o/ 2011-09-27T17:18:48 what i want, is if a bot looses its hill, he should be able to take revenge on the other ants and win the game 2011-09-27T17:19:19 yes, but must do it knowing there will be no more spawns available 2011-09-27T17:19:29 yes 2011-09-27T17:19:44 dang, I didn't think you'd agree with me, but I'm glad you do :) 2011-09-27T17:20:25 we are still cutting the game short if the bot without hills is not in the lead, which I'm okay with 2011-09-27T17:22:28 *** Blkt has quit IRC (Remote host closed the connection) 2011-09-27T17:26:31 *** berak has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443]) 2011-09-27T17:30:31 aichallenge: Marco Leise epsilon * ra7ec1b4 / ants/visualizer/js/CanvasElement.js : proximity indicators fade out to invisible - http://git.io/h16_7Q 2011-09-27T17:31:27 *** amstan_ has quit IRC (Ping timeout: 255 seconds) 2011-09-27T17:31:27 *** amstan has quit IRC (Ping timeout: 248 seconds) 2011-09-27T17:33:35 who is that person constantly updating the server code? :) 2011-09-27T17:35:08 aichallenge: McLeopold epsilon * r2c89faf / ants/ants.py : test extended play - http://git.io/6KD7Dw 2011-09-27T17:35:08 aichallenge: McLeopold epsilon * rf1838f3 / ants/visualizer/js/CanvasElement.js : Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/bABS_w 2011-09-27T17:36:20 *** xathis has joined #aichallenge 2011-09-27T17:37:45 *** JamesMG has quit IRC (Ping timeout: 260 seconds) 2011-09-27T17:38:00 *** JamesMG has joined #aichallenge 2011-09-27T17:38:54 woah these new rings in the visualizer look really slick! 2011-09-27T17:39:15 xathis: I'm not sure if you caught this, but the siphon is going to be part of the game 2011-09-27T17:40:10 McLeopold :( 2011-09-27T17:41:29 McLeopold: so janzert convinced you that the idea to destroy a hill when an ant is killed on it isn't good? 2011-09-27T17:41:36 xathis: it turns out you can't raze a hill until the hive is emptied anyway 2011-09-27T17:42:18 why's that? 2011-09-27T17:42:30 http://ietherpad.com/Rz6lUKCSZn see this 2011-09-27T17:42:57 *** ltriant has joined #aichallenge 2011-09-27T17:43:23 so if you step on a hill and at the same turn an ant spawns on the hill than they both die? 2011-09-27T17:43:42 well, the ant spawns first 2011-09-27T17:43:42 so it is a collision 2011-09-27T17:44:01 razing happens after than, so you'd have to have a clear hill to raze 2011-09-27T17:44:35 okay 2011-09-27T17:44:42 aichallenge: Marco Leise epsilon * r2e393e3 / ants/visualizer/js/CanvasElement.js : made circles around hills scale better - http://git.io/KDiOMg 2011-09-27T17:44:54 i say this will isn't going to work well... but we'll see 2011-09-27T17:47:31 if you have two hills and your enemy gets one of them but is clever and doesn't step on it, every second ant you spawn is killed right away, so you have a huge disadvantage in the game 2011-09-27T17:48:03 xathis: once the visualizer looked like this: http://imageshack.us/photo/my-images/32/bildschirmfotoen.png/ 2011-09-27T17:48:33 mleise: holy shit :D 2011-09-27T17:48:33 *** amstan has joined #aichallenge 2011-09-27T17:48:35 *** ChanServ sets mode: +o amstan 2011-09-27T17:48:36 xathis: Thats right, you cannot stop spawning at a hill 2011-09-27T17:49:04 placing two ants around it is the best attack strategy :) 2011-09-27T17:49:23 i thought it would be clever to stop caring about one of your hills, but looks like you're pretty much lost if one of your hills is controlled by an enemy, unless he's stupid and steps on it 2011-09-27T17:50:59 http://antsbeta.janzert.com/map.php?map=multi_hill_maze/multi_maze_02.map row 20 col 86 - I'm not sure but I think it's again a hill in the water. circle makes is hard to see 2011-09-27T17:51:21 *** Keth has quit IRC (Ping timeout: 244 seconds) 2011-09-27T17:52:12 *** okay| is now known as okayzed 2011-09-27T17:53:27 olexs1: oh thats a bug then 2011-09-27T17:53:58 the fat circle on a simple map 2011-09-27T17:56:36 *** Keth has joined #aichallenge 2011-09-27T17:58:35 aichallenge: Marco Leise epsilon * rffe0ccb / ants/visualizer/js/CanvasElement.js : fixed: proximity indicator rendered on maps - http://git.io/WkNELg 2011-09-27T17:59:49 Some of the hills are definitely on water, but this may be an old map. Antimatroid has fixed hills on water. 2011-09-27T18:00:34 that's pretty evil actually. You cannot defend those hills, but an enemy can get into position around them and bleed you out 2011-09-27T18:08:04 *** asmoo has quit IRC (Ping timeout: 252 seconds) 2011-09-27T18:12:26 *** Palmik has quit IRC (Remote host closed the connection) 2011-09-27T18:16:45 *** onensora has quit IRC (Read error: Connection reset by peer) 2011-09-27T18:24:44 http://ants.zeroviz.us/x8Qqe/ a little proof of concept, every second ant is killed right after being spawned 2011-09-27T18:24:58 *** arlequ1 has quit IRC (Quit: Page closed) 2011-09-27T18:40:46 xathis: did you change the engine to do that? 2011-09-27T18:44:36 McLeopold: to do what? 2011-09-27T18:44:56 McLeopold: i haven't changed anything but i think i'm not on the latest version 2011-09-27T18:45:02 I was wondering if you where testing rule changes or not 2011-09-27T18:45:31 oh no 2011-09-27T18:47:27 what about "your hill is razed when an ant dies on it and the hive is empty" ? 2011-09-27T18:47:52 i know it might be strange to explain that to newcomers... 2011-09-27T18:56:30 xathis: how about if an ant dies on it, it is blocked until another friendly ants steps on it again 2011-09-27T18:56:55 McLeopold: sounds good 2011-09-27T18:59:52 *** denebil has joined #aichallenge 2011-09-27T19:05:31 *** McLeopold has quit IRC (Ping timeout: 256 seconds) 2011-09-27T19:17:48 *** ChrisH has joined #aichallenge 2011-09-27T19:17:48 *** mleise has quit IRC (Read error: Connection reset by peer) 2011-09-27T19:18:06 *** mleise has joined #aichallenge 2011-09-27T19:18:45 those rings are full on 2011-09-27T19:18:53 it's almost impossible to tell what's even going on :P 2011-09-27T19:19:14 :-/ 2011-09-27T19:19:48 oh now i get it 2011-09-27T19:19:54 the map i opened had lots of hills 2011-09-27T19:20:36 antimatroid: your visualizer is outdated! 2011-09-27T19:21:34 mleise: i'm watching on janzerts site? 2011-09-27T19:22:39 yes, I fixed that bug 2011-09-27T19:23:32 which bug? 2011-09-27T19:23:50 the commonwealth bug 2011-09-27T19:23:51 ehe 2011-09-27T19:23:54 These programmers at Darkworks seem to be writing good bots. I wish them luck with their zombie games 2011-09-27T19:24:11 antimatroid: The 'rings full on on maps' bug 2011-09-27T19:25:48 *** McLeopold has joined #aichallenge 2011-09-27T19:28:51 This is sick: http://aichallengebeta.hypertriangle.com/visualizer.php?game=928&decorated=false&config={%22fullscreen%22:true} 2011-09-27T19:30:04 mleise: what's with the circles? 2011-09-27T19:31:46 which ones? the dashed circles, or the transparent? 2011-09-27T19:31:55 both? 2011-09-27T19:32:21 the dashed ones make it easier to spot hills especially on low resolutions 2011-09-27T19:32:42 the others show when an enemy ant gets close to a hill, so you see where the action is going on 2011-09-27T19:32:54 they're kinda intrusive though 2011-09-27T19:33:20 you shouldn't have them if there's only one enemy ant around it 2011-09-27T19:33:29 however, if that green thing comes, sure 2011-09-27T19:34:34 one enemy ant can easily raze the hill 2011-09-27T19:37:14 I think it is the maze maps, that make the circles look intrusive, because they tend to stay when an enemy is behind water and miles away. 2011-09-27T19:39:48 hills! 2011-09-27T19:39:56 i've not been paying attention obviously 2011-09-27T19:40:19 that reminds me: i should ping all starter package creators 2011-09-27T19:40:29 McLeopold: ping, where can i find more docs? 2011-09-27T19:41:18 docs for what? 2011-09-27T19:41:28 McLeopold: for the starter package guys 2011-09-27T19:42:06 McLeopold: should i tell them to follow the python starter package again? is there a place with the exact new game rules in? 2011-09-27T19:43:00 amstan: I just added a myHills and emenyHills array to my bot and parsed them the same way as ants. 2011-09-27T19:43:42 In this game the circles don't looks so intrusive: http://aichallengebeta.hypertriangle.com/visualizer.php?game=1094 2011-09-27T19:45:31 Token WTF? OMG "go" 2011-09-27T19:45:39 amstan: https://github.com/aichallenge/aichallenge/wiki/Ants-Starter-Pack-Guide 2011-09-27T19:45:56 mleise: that's what i did for the starter bot 2011-09-27T19:46:15 and marked the squares as hills with a hillPlayer variable 2011-09-27T19:46:21 c++ starter bot* 2011-09-27T19:48:09 McLeopold: I don't know if this is a scoring problem or not, http://aichallengebeta.hypertriangle.com/visualizer.php?game=1016&user=3 2011-09-27T19:48:40 McLeopold: Notice how one of the bots has timed out during initialization and only has 1 point to start the game. 2011-09-27T19:48:58 is it an issue that players can now end with a minus score? 2011-09-27T19:49:12 i think we tried to avoid that previously? 2011-09-27T19:49:27 McLeopold: But what's also interesting is that that point never leaves the display, while other bots's scores disappear when they are eliminated. 2011-09-27T19:49:30 mleise: can you keep the circles in the main map, not the shadow? 2011-09-27T19:50:00 McLeopold: python starter package is all up to date, right? 2011-09-27T19:50:20 amstan: except for 1 bug, yes, I'm not sure if I commited the fix yet 2011-09-27T19:50:27 McLeopold: please do soonb 2011-09-27T19:50:37 McLeopold: since i'm referring it in my message 2011-09-27T19:50:42 amstan: just tell them to look at the c++ one :P 2011-09-27T19:50:45 antimatroid: ok 2011-09-27T19:50:51 McLeopold: I'm not sure I understand you. There is no shadow layer. 2011-09-27T19:50:58 specifically, state.h, state.cc and square.h 2011-09-27T19:51:23 mleise: i think he means the border? 2011-09-27T19:52:03 ah right, yeah I noticed that, but the border is just a copy of what is rendered in the center 2011-09-27T19:52:25 *** Doddzy39 has joined #aichallenge 2011-09-27T19:52:48 ChrisH: as luck would have it, he only lost 4 hills :) 2011-09-27T19:53:39 mleise: the repeated sections 2011-09-27T19:53:43 McLeopold: It is possible, but not straight forward 2011-09-27T19:54:21 McLeopold: so the score is right, but the display is inconsistent with other bot elimination situations. 2011-09-27T19:54:57 i really don't think the squatting thing is that much of an issue the more i think about it? 2011-09-27T19:54:59 McLeopold: and is the visualizer looking ahead in the game to know the final score for that player? 2011-09-27T19:55:22 aichallenge: McLeopold epsilon * rc21894b / (2 files in 2 dirs): fix for python starter bot - http://git.io/aWx7Sg 2011-09-27T19:55:22 aichallenge: McLeopold epsilon * r1302cd3 / ants/visualizer/js/CanvasElement.js : Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/MQPuKg 2011-09-27T19:55:25 you want to take out as many ant hills as fast as possible, because as the game progresses any remaining hills are going to be better guarded and more spread out, making it more costly to gain points 2011-09-27T19:55:29 ChrisH: yes, kind of 2011-09-27T19:55:58 McLeopold: all good? https://gist.github.com/062e1cb6bfea05011903 2011-09-27T19:55:59 and you aren't going to be sure how many ant hills your opponent has etc. personally I wouldn't worry about taking out a few ants and just get rid of their hill already, and if someone is taking out your ants then it's on you to send reinforcements to stop them 2011-09-27T19:56:45 amstan: reference the python starter 2011-09-27T19:56:51 McLeopold: ok 2011-09-27T19:56:59 antimatroid: I agree with you in general, but each bot knows how many hills it started with and it is trivial to assume all players get the same number of hills. 2011-09-27T19:57:17 yes, but depending on how long hte game has been going for you don't know how many they've lost 2011-09-27T19:57:22 aichallenge: Alexandru Stan epsilon * rd912799 / (5 files): 2011-09-27T19:57:22 aichallenge: Merge pull request #268 from smiley1983/epsilon 2011-09-27T19:57:22 aichallenge: OCaml starter: added hill support - http://git.io/WBROQg 2011-09-27T19:57:28 plus mention that ants and hills can co-exists, so a simple matrix is not sufficient, coffescript needs to change that 2011-09-27T19:57:59 yeah i had to think about that when i fixed my debug state output function, i have hills priority and capitalised them 2011-09-27T19:58:36 antimatroid: did you follow our new map format? 2011-09-27T19:58:47 what's the new map format? 2011-09-27T19:58:50 antimatroid: true, it would be difficult to know because a hill could be razed before you ever scout it's location and you would never get the info from the server, but once you've scouted 100% of the map you could infer it's absence. 2011-09-27T19:58:58 0=hill, a=ant, A=ant on hill 2011-09-27T19:59:30 i'll update my map generator lator 2011-09-27T19:59:38 or is that to stay the same? 2011-09-27T19:59:41 i'm off to uni soonish 2011-09-27T19:59:54 the map generator should output numbers now 2011-09-27T20:00:09 we want the other stuff so we can paste inprogress games as maps 2011-09-27T20:00:37 yeah cool, i had a moment of displeasure when I realised I wouldn't have been able to do that anymore yesterday :) 2011-09-27T20:00:48 oh man, cool! 2011-09-27T20:00:50 git.io 2011-09-27T20:00:55 my bot is set up so I can just geed state information to it through a file and check the output :) 2011-09-27T20:01:00 feed* 2011-09-27T20:01:50 antimatroid: in other words you have a stateless bot :) 2011-09-27T20:01:57 *** Corgano_ has joined #aichallenge 2011-09-27T20:02:15 depends, i don't clear everything each turn, but i'd need to provide all relevant water etc. info :P 2011-09-27T20:02:41 it'd have a go at playing stateless, put it that way :) 2011-09-27T20:02:55 mleise: depending on the shape of the visualizer, sometimes it looks like bots have double the starting hills 2011-09-27T20:03:49 McLeopold: I know, and I would have rendered the circles only in the center if that was a quick change. 2011-09-27T20:05:41 the new hill cues and proximity circles make games with lots of hills a lot more watchable. Kudos to all involved 2011-09-27T20:06:33 mleise: could you maybe put the hill circle on only when you know it will be captured? 2011-09-27T20:06:46 say, 20 turns before or something? 2011-09-27T20:06:57 I like the ping and the start 2011-09-27T20:07:46 I guess that would mess up the live games... 2011-09-27T20:09:47 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-09-27T20:10:12 McLeopold: Live games are a niche product. They could live without it. 2011-09-27T20:10:40 I just like the hints it gives on defense situations 2011-09-27T20:10:53 even if the hill is not razed 2011-09-27T20:12:26 mleise: I don't know if there are a better choice of colors, but the green and the orange are really tough for red-green color blind people like me. 2011-09-27T20:13:30 ChrisH: I see... almost no difference in brightness 2011-09-27T20:13:54 mleise: do you still have the pallette tool? 2011-09-27T20:14:00 mleise: yeah, the orange is the one I would prefer to eliminate 2011-09-27T20:14:03 we can limit it to 10 colors 2011-09-27T20:14:43 plus blue and brown 2011-09-27T20:14:48 12 colors 2011-09-27T20:14:54 and the food 2011-09-27T20:15:01 McLeopold: The palette tool turned out to perform badly. I should have used a 'real world' case instead of squares on water/sand/food backgrounds 2011-09-27T20:15:54 ChrisH: maybe you can come up with a colorized screenshot for a 10 player game? that would be helpful for me 2011-09-27T20:16:42 i could try, any suggestions how to go about it? 2011-09-27T20:16:52 I've got GIMP. 2011-09-27T20:16:52 I heard red-green blind people are good at distinguishing shades of green and khaki. 2011-09-27T20:17:06 GIMP is ok. 2011-09-27T20:17:17 it's free ;) 2011-09-27T20:17:20 there is a filter for switching colors i think 2011-09-27T20:17:57 so just capture a 10 player game in progress and do some color replacements? and paste it somewhere? 2011-09-27T20:18:51 *** Corgano_ has quit IRC (Quit: Page closed) 2011-09-27T20:19:20 ChrisH: yeah, that's what I had in mind 2011-09-27T20:19:41 ok, i'll see what I can come up with. 2011-09-27T20:20:01 anyone have any links to a 10 player game? 2011-09-27T20:21:16 the most I've ever seen is 8. 2011-09-27T20:28:03 ChrisH: try this screenshot http://img94.imageshack.us/img94/8211/10players.png 2011-09-27T20:28:24 mleise: thanks 2011-09-27T20:28:25 it is from an old replay with no hills 2011-09-27T20:30:24 *** Drac32Drac has joined #aichallenge 2011-09-27T20:30:45 *** amstan_ has joined #aichallenge 2011-09-27T20:30:50 *** ChanServ sets mode: +o amstan_ 2011-09-27T20:32:24 *** Drac32Drac has quit IRC (Client Quit) 2011-09-27T20:35:44 mleise: what about changing the orange to more of a yellow, such as efd900 2011-09-27T20:35:56 hello all 2011-09-27T20:36:30 ChrisH: possible, but if it becomes too bright, the timeline becomes barely visible 2011-09-27T20:36:30 aarossig: yo computerwiz_222 2011-09-27T20:36:55 mleise: good point, let me try something 2011-09-27T20:37:05 That's why I tried to restrict all the colors to some brightness range 2011-09-27T20:45:33 mleise: it's not easy finding another hue when there are already 9 chosen. 2011-09-27T20:45:55 ChrisH: I know ;). Guess how hard it was with 26 players 2011-09-27T20:45:57 mleise: at the least, try to make it so the orange is the last color used, so it only shows up in 10 player games 2011-09-27T20:46:42 ok I'll do that 2011-09-27T20:46:44 mleise: another idea would be to decorate some of the ants with an outline, or a little dot in the middle, or some other shape. 2011-09-27T20:47:15 I have greek letters for you 2011-09-27T20:47:53 unfortunately the circles are so on the big maps that something that intricate is just lost. 2011-09-27T20:49:05 there was an online game I played once that had an orange and lime green even harder for me to tell apart, even with much bigger icons. Luckily they put a texture on each piece that was randomly rotated each game. 2011-09-27T20:49:29 the games where the green and orage got the same rotation angle I would just skip to the next board. 2011-09-27T20:49:41 hehe 2011-09-27T20:50:26 I still support old replay files with up to 26 players, so I would definitly want to draw 26 textures for the ants 2011-09-27T20:50:45 that could end up being a lot of work 2011-09-27T20:51:06 *** olexs1 has quit IRC (Quit: Leaving.) 2011-09-27T20:51:21 green could have a leaf texture, brown is wood, grey is chrystal 2011-09-27T20:51:34 I just tried the greek letters. That's definitely an option for when I need it. 2011-09-27T20:52:12 aichallenge: McLeopold epsilon * r0fe1b1f / (website/visualizer_widget.php website/example_games/1.replay): allow embedded visualizer to load uris - http://git.io/UptAxg 2011-09-27T20:52:12 aichallenge: McLeopold epsilon * r7f6a9bf / (5 files): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/zcGTfg 2011-09-27T20:52:23 They used to show up in the score graph as well, but I lost them when I added the hive/bonus sections 2011-09-27T20:52:51 *** mleise has quit IRC (Quit: Leaving.) 2011-09-27T20:55:01 aichallenge: McLeopold epsilon * r880146f / website/visualizer_widget.php : update embedded visualizer - http://git.io/dtYrBQ 2011-09-27T20:56:55 aichallenge: McLeopold epsilon * rb387e22 / website/visualizer_widget.php : update embedded visualizer - http://git.io/ABVOjQ 2011-09-27T21:01:46 aichallenge: McLeopold epsilon * rb11dd4d / website/visualizer_widget.php : added embedded flag - http://git.io/s-MaXw 2011-09-27T21:03:04 @later tell mleise http://aichallengebeta.hypertriangle.com/problem_description.php I can't turn off the graphs, what am I doing wrong? 2011-09-27T21:03:04 McLeopold: OK 2011-09-27T21:03:07 *** McLeopold has left #aichallenge 2011-09-27T21:08:55 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-09-27T21:11:00 *** ChrisH has quit IRC (Quit: Page closed) 2011-09-27T21:37:14 *** ibdknox has joined #aichallenge 2011-09-27T21:56:17 *** okayzed is now known as okay| 2011-09-27T21:56:22 *** AlliedEnvy has quit IRC (Read error: Connection reset by peer) 2011-09-27T21:57:15 *** AlliedEnvy has joined #aichallenge 2011-09-27T22:25:38 *** ChrisH has joined #aichallenge 2011-09-27T22:26:13 *** amstan has joined #aichallenge 2011-09-27T22:26:13 *** ChanServ sets mode: +o amstan 2011-09-27T22:27:02 *** ibdknox has quit IRC (Remote host closed the connection) 2011-09-27T22:29:35 *** amstan_ has quit IRC (Ping timeout: 248 seconds) 2011-09-27T22:32:28 <_flag> Is the new game taking longer or are there less workers or something? I feel like I don't play very many games, but that could just be because the "x players are ahead of you" is missing 2011-09-27T22:36:36 janzert: I just filed an issue about the visualizer and the user profile pages not always showing the same ranking of players for a completed game. 2011-09-27T22:37:20 janzert: the issue has a link to a game that shows the problem 2011-09-27T22:41:02 *** ChrisH has quit IRC (Quit: Page closed) 2011-09-27T22:54:14 *** McLeopold has joined #aichallenge 2011-09-27T23:02:39 *** Corgano has joined #aichallenge 2011-09-27T23:03:55 *** Corgano has quit IRC (Client Quit) 2011-09-27T23:05:34 McLeopold: oh come on :P you don't think the way you asked amstan about the game end condition was just a bit the far side of biased? :/ and amstan already agreed with you about not eliminating players outright when their last hill is killed. He's the one that convinced me to go along with it. :} 2011-09-27T23:05:58 and it really was your idea ;) 2011-09-26T01:07:49 if you lose your hills, you still get to participate somewhat, but the cutoffs and other game length decisions won't depend on you 2011-09-27T23:06:22 I still think you had a great idea there 2011-09-27T23:06:45 where? 2011-09-27T23:06:51 from a game rule perspective I also still think the most natural is that when you lose all your hills (bases) you're eliminated from the game 2011-09-27T23:06:55 janzert: it might have been biased, but i was convinced about the answer since i talked to you 2011-09-27T23:07:16 janzert: check the database for the cutoff reason 2011-09-27T23:07:24 amstan: I'm not even sure you know what the question is ;) 2011-09-27T23:07:40 I've stopped cutting off the games to see if there is any difference in rank 2011-09-27T23:07:48 but I do agree that just having a bunch of ants freeze is less than ideal for watching games 2011-09-27T23:08:11 McLeopold: yeah I saw that, that is good to see 2011-09-27T23:08:34 so far, no different results 2011-09-27T23:08:45 so anyway I think your "compromise" you came up with is the perfect position in between 2011-09-27T23:09:00 to repeat it: 2011-09-26T01:07:49 if you lose your hills, you still get to participate somewhat, but the cutoffs and other game length decisions won't depend on you 2011-09-27T23:09:11 doing that is the question amstan 2011-09-27T23:09:36 okay, but you did redefine "remaining players" and when I made that statement it was different 2011-09-27T23:09:57 janzert: i'm not that familiar with the game length decisions i guess, but sure 2011-09-27T23:10:00 ? that statement doesn't refer to remaining players 2011-09-27T23:10:17 amstan: whether the game should continue or not 2011-09-27T23:10:37 it's in the word "cutoffs" and the phrase "game length decisions" 2011-09-27T23:10:38 janzert: let's put it this way, the scenario is too specific for people to have to explain in an argument if they don't like it 2011-09-27T23:11:33 aka, you're not leading, so well let you continue to try and win, even though you are the only bot with hills 2011-09-27T23:11:33 yes, there is one scenario that so far hasn't been seen to happen in practice that is seen as unfavorable when following that 2011-09-27T23:11:44 /well/we'll/ 2011-09-27T23:12:49 I think it is a nice spot in between outright elimination and keeping you fully in the game, you have a chance but you can't make the game keep going just so you continue to have a chance 2011-09-27T23:13:49 speaking practically it also has a nice side effect of eliminating lots of moves that don't have much chance of changing anything in the outcome 2011-09-27T23:14:58 so far since McLeopold added the measurement code it would be the single largest reason for ending a game and not once would it have yet effected the game outcome 2011-09-27T23:15:20 err, single largest reason except for game length limit 2011-09-27T23:15:49 we've only extended 23 games 2011-09-27T23:15:51 which is certainly effecting game outcomes but also is something that we can't do without 2011-09-27T23:16:05 oh, what timestamp are you using? 2011-09-27T23:16:18 20:25 2011-09-27T23:16:26 how long ago? 2011-09-27T23:16:59 7 hours? 2011-09-27T23:17:17 commit log says 5 hours 2011-09-27T23:17:33 I show 23 extended games anyway 2011-09-27T23:17:51 current time on server is 03:17 2011-09-27T23:18:25 I pulled the time from the time you shutdown the worker to restart it with the new code I assumed 2011-09-27T23:19:29 if I'm too long I'm including games as not extended that should count as extended then 2011-09-27T23:21:25 have you looked at the turn savings? 2011-09-27T23:22:29 not rigorously in any way 2011-09-27T23:23:43 some are quite small, some are quite large 2011-09-27T23:23:57 I haven't looked at enough to get a feel for an average or anything 2011-09-27T23:24:13 fairly small http://aichallengebeta.hypertriangle.com/visualizer.php?game=1099 2011-09-27T23:24:24 pretty large http://aichallengebeta.hypertriangle.com/visualizer.php?game=1105 2011-09-27T23:26:24 janzert: 1105 looks incorrect 2011-09-27T23:27:03 isn't the leader the only one with a hill? 2011-09-27T23:28:08 even without that, hasn't the rank been stabalized? 2011-09-27T23:28:31 hmm, it seems like 2011-09-27T23:28:41 that's got to be a bug 2011-09-27T23:28:49 it should have been cut off 2011-09-27T23:29:05 at least after fxwz141 dies 2011-09-27T23:29:25 so still would have been extended more than 50% 2011-09-27T23:29:42 actually, right at turn 144 2011-09-27T23:30:03 with ants that don't control a hill don't extend the game 2011-09-27T23:30:20 ranks don't stabilize till 574 2011-09-27T23:31:02 wait, at 144, eAshoka is already second, and can't make first, so it should be cutoff before that 2011-09-27T23:31:24 ranks aren't stabilized till fxwz is killed off 2011-09-27T23:31:32 no, he's third, so 144 is the correct cutoff 2011-09-27T23:31:39 since he could destroy a hill and tie with trumpet 2011-09-27T23:31:50 *** delt0r_ has quit IRC (Ping timeout: 260 seconds) 2011-09-27T23:31:53 and he is alive till 574 2011-09-27T23:32:18 but after he loses the hill, we don't give him the chance, since he's not the leader 2011-09-27T23:33:52 so only the leader of the board without a hill gets to extend the game? that seem really special casing it and unfair, not at all a natural rule 2011-09-27T23:34:09 janzert: oh, I removed the hill check for extended play, oops 2011-09-27T23:34:40 seems natural to me 2011-09-27T23:35:52 it's not so much the leader of the board gets to extend, it's more than someone else gets a chance to win 2011-09-27T23:36:31 handing over points to someone doesn't seem right 2011-09-27T23:36:55 so, we are only doing it in rare circumstances when we think we know who should get the points 2011-09-27T23:37:06 giving someone a win is just not right 2011-09-27T23:37:40 this isn't a game of survival anymore, it's about attacking hills 2011-09-27T23:37:45 we're not 'giving' them the win, they get to the situation by keeping there hill 2011-09-27T23:37:57 I don't want to emphasize defense 2011-09-27T23:38:13 the -1 is only there to break up the ranks a little, not to hand a win to a dumb bot 2011-09-27T23:39:20 I wish we would have stuck with your own hill having the same weight as taking an opponents 2011-09-27T23:39:37 how come? 2011-09-27T23:40:19 maybe you wouldn't feel like holding onto your own hill was so different from taking an opponents :/ 2011-09-27T23:41:04 well, if holding on to your own hill was 10x attacking, no one would ever attack 2011-09-27T23:41:16 it would be a stupid game 2011-09-27T23:41:51 if it was 1to1, it's still not enough benefit to attack 2011-09-27T23:41:51 my thought on the game is that holding on to your own hill till the end of the game is the same as taking an opponents hill 2011-09-27T23:41:54 *** dvladim has joined #aichallenge 2011-09-27T23:42:06 I totally disagree 2011-09-27T23:42:14 a game of defense is a bad game 2011-09-27T23:42:20 *** Keth has quit IRC (Read error: Connection reset by peer) 2011-09-27T23:42:34 *** Keth has joined #aichallenge 2011-09-27T23:42:36 I don't think it would lead to a game stuck in defense 2011-09-27T23:42:38 have you played galcon online? 2011-09-27T23:42:47 it would, it really would 2011-09-27T23:42:56 if one opponent is breaking the mold you just lost the tournament 2011-09-27T23:42:58 it would be a broken game and the contest wouldn't work 2011-09-27T23:43:57 I know you want shorter games. More attacking means shorter games. 2011-09-27T23:44:16 not extending means shorter games ;) 2011-09-27T23:44:20 if we encourage defense, every games goes the full length, and the cutoff adjuster skyrockets 2011-09-27T23:44:33 but that isn't my largest reason even 2011-09-27T23:44:56 *** delt0r_ has joined #aichallenge 2011-09-27T23:46:14 the problem with the no hill games was no one was attacking 2011-09-27T23:46:15 but when we can shorten games by what looks like an average of maybe 20% with a change in rankings of less than 1 in 20 that does seem like quite a straight up win 2011-09-27T23:46:55 but your looking at bots that haven't fully adjusted to the rules 2011-09-27T23:48:00 yeah, once bots will at least take one step out of there way to gain a point the gain will probably be much larger ;) 2011-09-27T23:49:05 yes, once defense actually enters the game hills are going to become actually hard to destroy 2011-09-27T23:49:22 but that isn't going to change by this 2011-09-27T23:49:38 well, it would by the points per hill 2011-09-27T23:49:43 not by the cutoff 2011-09-27T23:50:27 I don't really see how 1 or 2 points on one hill isn't going to make that much difference either 2011-09-27T23:50:35 isn't->is 2011-09-27T23:53:04 I'm going to add the hill check back in, so only when the last ant with hills isn't in the lead will the game be extended 2011-09-27T23:53:10 let's see what we get then 2011-09-27T23:53:19 ok 2011-09-27T23:53:30 can you fire up a second worker for this? 2011-09-27T23:53:35 I gotta get to bed, back to work in 7 hours 2011-09-27T23:53:48 ok 2011-09-27T23:54:16 aichallenge: McLeopold epsilon * r9206465 / ants/ants.py : add hill check back in - http://git.io/7wfVGg 2011-09-27T23:54:56 worker archive ready 2011-09-27T23:55:07 ok 2011-09-27T23:55:24 so, what should we do besides argue? :P 2011-09-27T23:55:42 I'll working on the docs :( 2011-09-27T23:56:16 I should sleep ;) 2011-09-27T23:56:27 two workers should be up in 15 minutes or so 2011-09-27T23:57:03 last game before adding hill check back in is 1205