2011-06-16T00:09:46 *** janzert has joined #aichallenge 2011-06-16T00:10:50 oh, wow ICFP programming contest starts in less than 24 hours 2011-06-16T00:14:18 interesting they're putting a hard cpu limit on entries this year, don't remember that before 2011-06-16T00:16:44 *** Chris_0076 has joined #aichallenge 2011-06-16T00:17:28 *** Chris_0076 has left #aichallenge 2011-06-16T00:54:16 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-06-16T01:04:29 *** antimatroid has quit IRC (Quit: Leaving.) 2011-06-16T01:04:35 *** antimatroid has joined #aichallenge 2011-06-16T01:18:28 *** antimatroid has quit IRC (Quit: Leaving.) 2011-06-16T01:18:42 *** antimatroid has joined #aichallenge 2011-06-16T01:20:14 *** berak has joined #aichallenge 2011-06-16T01:30:47 *** mcstar has joined #aichallenge 2011-06-16T01:32:43 *** Kingpin13 has joined #aichallenge 2011-06-16T01:45:59 *** UncleVasya has joined #aichallenge 2011-06-16T01:46:05 *** McLeopold has joined #aichallenge 2011-06-16T01:47:00 antimatroid: did you understand my idea for a new attack method? 2011-06-16T02:00:17 amstan , can you describe in detail how you see the work of a button for making submissions in testing app? 2011-06-16T02:01:08 where is that new attack method introduced? 2011-06-16T02:01:09 *** Zannick has quit IRC (Ping timeout: 240 seconds) 2011-06-16T02:02:14 *** Zannick has joined #aichallenge 2011-06-16T02:02:31 rabidus: I haven't really explained it at all, I was waiting for antimatroid to show up 2011-06-16T02:02:57 so you were talking privately 2011-06-16T02:03:13 Basically, if any ant only attacked, or was within attack range, of one other ant, then its full damage would go to that ant and kill it 2011-06-16T02:03:13 McLeopold: the outcome of your --html update and advices is that you can launch replays for every game during and after the multimap testing. Also my app inserts good bot names in the replay and you can see them in the visualizer. 2011-06-16T02:04:30 UncleVasya: I attempted to retrieve bot names from command lines passed, but it is a bit broken. If your app has manages bots, then having a name outside of the command line is a good thing. So, nice work. 2011-06-16T02:05:11 UncleVasya: also, if there are bugs or features you need in playgame, please open up issues on github for them and I will fix/code stuff 2011-06-16T02:05:24 rabidus: back to attacking... 2011-06-16T02:06:10 For defense, the goal would be to make sure your ant has a buddy to take half the damage from attacking ants. 2011-06-16T02:06:10 i was just wondering... next task for me was good attack planning, but i think i'll move that task a bit 2011-06-16T02:06:36 McLeopold: Again, it's my crazy way to work: When html-file exists I just deal with it as with a text file and adding text info about bots. :) 2011-06-16T02:06:50 This can be extended to maintaining a higher ant count than the enemy, even in localized situations 2011-06-16T02:07:15 Which allows for some trivial blocking if the enemy doesn't have freedom of movement, as in corridors. 2011-06-16T02:08:05 McLeopold: can you show static blocking of a 4 width corridor with attackradius2=9? 2011-06-16T02:08:07 http://dl.dropbox.com/u/9140666/ants-r2-9-corridors.png 2011-06-16T02:08:18 i dont get it. what's the difference between now and that ? 2011-06-16T02:08:19 UncleVasya: yeah, interjecting playernames into the replay file, or into the html replay makes sense. 2011-06-16T02:09:12 rabidus: There is not two methods of attack. I'm just disecting the damage method in general. We have not made changes nor implemented a new attack method. 2011-06-16T02:09:20 the best I see is 2 for 1 2011-06-16T02:10:40 janzert: I ran a test of a column of ants, of different widths, heading in the same direction against a sea of enemy ants. For every radius there is a shape that will block. The larger the radius, the larger the shape. 2011-06-16T02:11:13 So, there can always be blocking, but the shapes may be too large to create. 2011-06-16T02:11:36 Of course there will be variations depending on where water is and such. 2011-06-16T02:11:40 I really don't believe that is true for a radius that has both 'nibs' and 'convex shoulders' (as defined a few days ago ;) ) 2011-06-16T02:11:54 I also think attackradius2=9 may be the only radius that fulfills that 2011-06-16T02:12:01 I thought it wouldn't be either, but it is. 2011-06-16T02:12:27 I'd really like to see that static shape that will bock the 4 width corridor then 2011-06-16T02:12:49 The ants were all moving north. Perhaps I could come up with some side attacks or something, but it looks like there will always be a static shape this is solid. 2011-06-16T02:12:59 for radius 9? 2011-06-16T02:13:11 thank god my nick isn't radius 2011-06-16T02:13:12 radius squared = 9 yes 2011-06-16T02:13:24 hold on... 2011-06-16T02:13:34 I believe it is in use on the beta server now 2011-06-16T02:13:56 the beta server won't give us any info, there are no good blocking bots 2011-06-16T02:14:08 did you see the image I posted above? 2011-06-16T02:14:27 yeah 2011-06-16T02:14:41 that is what I did my check work from to see that 1, 2 and 3 width can be blocked but I don't think 4 width 2011-06-16T02:14:58 I'm running the test to prove you wrong... 2011-06-16T02:15:04 good :) 2011-06-16T02:16:56 *** Ice_Harley has joined #aichallenge 2011-06-16T02:16:57 how about giving the opportunity to ants to combine their strength? 2011-06-16T02:17:19 for example 2-3 ants could youn up on one square, for a single attack 2011-06-16T02:17:33 if their startegy is good, they could break any block 2011-06-16T02:17:53 but the formation couldnt be maintained for more than a couple of turns 2011-06-16T02:18:50 youn->team up? 2011-06-16T02:19:22 oh no, i meant join up probably, that makes sense 2011-06-16T02:19:47 I wouldn't mind a single order including an attack square for a single ant 2011-06-16T02:20:19 janzert: I've messed up my code, it will be just a minute... 2011-06-16T02:20:36 you mean, that only that ant could attack, others cannot, with more than 1 damage to distribute? 2011-06-16T02:21:19 no, meaning 1 ant can choose not to spread its 1 damage across all enemies in range, but rather choose which enemy to take out 2011-06-16T02:21:29 oh 2011-06-16T02:21:35 *** Palmik has joined #aichallenge 2011-06-16T02:21:37 yeah, thats not bad either 2011-06-16T02:22:46 i said what i said, because there would have to be some balance maintained between attacking force and the number and spread of ants 2011-06-16T02:23:29 this would certainly make the game strategy wise more interesting 2011-06-16T02:23:43 also its porbably too late for this now isnt it? 2011-06-16T02:24:10 not really, especially if the current method can lead to static blocking 2011-06-16T02:24:20 of any width corridor that is 2011-06-16T02:24:41 meaning you think the rules can change that much? 2011-06-16T02:24:46 I also like the idea of ants doling out more than 1 damage, say 1.5 2011-06-16T02:24:59 mcstar: yes, if necessary 2011-06-16T02:25:16 if static blocking is always possible the rules are basically broken 2011-06-16T02:27:42 McLeopold: I saw a new type of error during watching the games on server: 2011-06-16T02:27:58 *** iShurf has quit IRC (Ping timeout: 252 seconds) 2011-06-16T02:28:15 Creating canvas... 2011-06-16T02:28:18 status 0 2011-06-16T02:28:21 janzert: http://imgur.com/BjbzJ okay, here's the shape 2011-06-16T02:28:36 &debug=true helped 2011-06-16T02:28:55 UncleVasya: open a github issue. Also let Frontier know 2011-06-16T02:30:44 McLeopold: ahh, I see the important part is the ability to place ants not just at the end of the corridor but one space further as well 2011-06-16T02:31:22 .˙˙˙˙˙. these forms can block 2011-06-16T02:31:35 yeah, so for any radius, there is a shape 2011-06-16T02:32:03 for set of random waters in there, you could probably modify the shape 2011-06-16T02:32:38 i recently realized that you changed the radii, i was wondering why the f i dont understand a discussion, turned out i was thinking in ar2<=6 2011-06-16T02:32:38 we probably want the shape to be large enough that it is improbable or impractical to maintain 2011-06-16T02:32:40 http://pastebin.com/sMkJYMbp 2011-06-16T02:32:57 i.e. there has to be space for those top and bottom b ants 2011-06-16T02:33:34 in any case I'd agree the current method is broken enough that we need to do something else 2011-06-16T02:33:46 :( 2011-06-16T02:33:55 so if it is just a solid corridor, you can't block, or in other words, if both bots have the same degrees of freedom, then you can't block 2011-06-16T02:34:30 it's only when one bot has more freedom of movement that blocking is possible 2011-06-16T02:35:17 and what about attack formations? i dont want to talk shit, but i think 4 ants can always kill 1 ant 2011-06-16T02:35:37 active formations are fine with me :) 2011-06-16T02:35:57 I'd like to encourage them 2011-06-16T02:35:58 mcstar: breaking away from 1:1 kill ratios is the goal 2011-06-16T02:36:27 ok 2011-06-16T02:36:51 janzert: we could try giving each ant more than 1 damage or less than 1 hp? 2011-06-16T02:37:04 comulative damage? 2011-06-16T02:37:23 we are trying to avoid maintaining ant hp state 2011-06-16T02:38:04 well, it certainly makes the engine simpler 2011-06-16T02:38:27 adding extra damage might be the simplest change with a possibility of working out 2011-06-16T02:39:06 let me try 1.5... 2011-06-16T02:39:09 do you think starter packages should be able to handle them? 2011-06-16T02:39:11 standing still = attack power 0.9 and moving = attack power 1 2011-06-16T02:39:15 sure 2011-06-16T02:39:26 hey, that's a good idea 2011-06-16T02:39:43 ants have more attacking power if they moved during the turn? 2011-06-16T02:39:47 yes 2011-06-16T02:39:53 like irl 2011-06-16T02:40:03 what's that? 2011-06-16T02:40:09 you are moving towards your target, you have more power in your arm 2011-06-16T02:40:09 :) 2011-06-16T02:40:09 what if an ant cant move bc of water of his own ants, should it suffer from this restriction? 2011-06-16T02:40:10 hmm, that might work as well 2011-06-16T02:40:39 it does add more complexity that simply making damage > 1 though 2011-06-16T02:40:40 of->or 2011-06-16T02:40:53 but it avoids maintaining state 2011-06-16T02:41:29 and irl = in real life 2011-06-16T02:41:34 if you ment that 2011-06-16T02:41:51 yeah, that's what I ment 2011-06-16T02:41:52 now back to work (real work) -> 2011-06-16T02:41:53 meant 2011-06-16T02:41:57 McLeopold: I couldn't reproduce the bug. After page reloading visualizer works even without &debug=true. 2011-06-16T02:42:16 UncleVasya: you have a healing touch 2011-06-16T02:43:54 janzert: damage=1.5 still has a defense shape, but it is larger, requiring 2 ants for each incoming for a 4 width corridor 2011-06-16T02:44:48 not sure how it gets larger? 2011-06-16T02:45:13 mcstar: symptoms are hiding. Illness lies to us, like everybody else. :) 2011-06-16T02:45:41 UncleVasya: you arent an optimist are you :D 2011-06-16T02:46:34 Just watched default-TV-show too many times. 2011-06-16T02:48:47 McLeopold: maybe you should make the beta server more secure, so that people may upload their most current bots, i can imagine somebody might have very good solutions, only afraid that others might get his code... 2011-06-16T02:50:17 is there a securiti issues hanging around? 2011-06-16T02:50:21 mcstar: then we would have to remove several of the people that have accounts to help out with the coding that don't want to lose the opportunity of competing and things would develop slower 2011-06-16T02:50:24 *securyty 2011-06-16T02:50:32 http://imgur.com/6Fy2j damage 1.5 2011-06-16T02:50:50 ok, im understand 2011-06-16T02:50:51 *** smellyhippy has quit IRC (Quit: Take your stinking clause off me you damn dirty predicate!) 2011-06-16T02:50:53 -m 2011-06-16T02:51:30 imgur doesn't want to show me the image :( 2011-06-16T02:51:40 it works for me 2011-06-16T02:51:45 me too 2011-06-16T02:52:05 weird, the site loads but not the image itself 2011-06-16T02:52:38 McLeopold: what is the attackradius on the pic? 2011-06-16T02:52:43 9 2011-06-16T02:52:58 alright 2011-06-16T02:53:36 *** Kingpin13 has quit IRC (Quit: this used to be the life, but I don't need another one) 2011-06-16T02:54:12 finally got it 2011-06-16T02:54:32 *** znutar has left #aichallenge 2011-06-16T02:55:41 *** amstan has quit IRC (Ping timeout: 276 seconds) 2011-06-16T02:55:51 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-16T02:56:17 *** Kingpin13 has joined #aichallenge 2011-06-16T02:56:22 ok, so how hard would it be for you test 1.5 damage moving and 1 when standing still? 2011-06-16T02:56:41 *** UncleVasya has joined #aichallenge 2011-06-16T02:56:45 or basically for the test give the attackers 1.5 the defenders 1 2011-06-16T02:57:11 easy 2011-06-16T02:57:19 actually, I don't think it works 2011-06-16T02:57:43 just take your current shape and step it north and south on alternating turns 2011-06-16T02:57:59 I suppose it's not technically static, but close enough? 2011-06-16T02:58:19 or is there a way to get through it then? 2011-06-16T02:58:29 ugh 2011-06-16T02:58:54 the fact that the shape needs to be larger may be good enough 2011-06-16T02:59:26 yeah, possibly 2011-06-16T02:59:27 the shape is the same with or without the moving ant code 2011-06-16T02:59:56 oops, right 2011-06-16T03:00:17 since the defenders only need .5 damage 2011-06-16T03:00:26 damage = 2 works quite nicely 2011-06-16T03:00:59 the real trick is the defenders just need to make sure each enemy ant is spreading their damage to enough ants 2011-06-16T03:01:09 *** aerique has joined #aichallenge 2011-06-16T03:01:22 at which point were almost to 1 for 1 in most any realistic scenario though right? 2011-06-16T03:01:35 for damage=2? not really 2011-06-16T03:01:37 with this reasoning, damage can be a large number even 2011-06-16T03:02:33 with damage close to 2, but not quite there, it is really nice 2011-06-16T03:02:35 once damage is large enough the best tactic seems spread out individuals :( 2011-06-16T03:02:44 1 ant can't take out 2 2011-06-16T03:03:08 but the 1.5 damage blocking shape still holds 2011-06-16T03:03:19 but in a fight, if you can manage to get 2 ants close to an enemy, you can probably take them out, even with extra enemy taking some of the extra damage 2011-06-16T03:03:51 yeah, but that is for only moving 1 direction 2011-06-16T03:04:43 the fact that some of the damage is 90 degrees to the attacks may give enough room 2011-06-16T03:04:49 without "diagonal" moves it doesn't matter how you move into it when coming out of a corridor does it? 2011-06-16T03:05:49 with damage = 1.9, maybe I should reduce the attack radius again? 2011-06-16T03:06:51 ooh, that's quite nice 2011-06-16T03:07:38 2 width is blockable, but not 3 2011-06-16T03:07:43 doesn't a diagonal (or straight if a radius without nibs) line of ants just block then? 2011-06-16T03:08:32 no 2011-06-16T03:08:43 not with attackradius=4 and damage=1.9 2011-06-16T03:09:18 damage > 1 is away from what we want 2011-06-16T03:09:20 damage=1.5 does block width 3 2011-06-16T03:09:28 antimatroid: no it's not 2011-06-16T03:09:30 we wanted balance so that "bunching" up for fights was good 2011-06-16T03:10:08 except not so good that you can block with a static shape :( 2011-06-16T03:11:46 what about that 'standing still' vs 'moving' ? 2011-06-16T03:12:03 *** aumfer has quit IRC (Ping timeout: 255 seconds) 2011-06-16T03:12:31 rabidus: that doesn't make a difference to the attackers 2011-06-16T03:12:37 only in their loses 2011-06-16T03:12:53 but corridors are beaten 2011-06-16T03:13:32 *** Zaphus has joined #aichallenge 2011-06-16T03:13:49 radius=4, damage=1.9 is nice 2011-06-16T03:14:07 damage=1.875 is the sweet spot for breaking a 3 width corridor 2011-06-16T03:15:07 Evening all, talking about damage multipliers? 2011-06-16T03:15:08 you can get 3 ants in, the left and right can't attack the same set and get spread to 3 enemies, the center spreads to 5 enemies, so you need 1/(1/3+1/5) damage to break it 2011-06-16T03:15:47 antimatroid: can you make that latex? :) 2011-06-16T03:16:29 http://imgur.com/9AfZH 2011-06-16T03:17:03 *** antimatroid has quit IRC (Quit: Leaving.) 2011-06-16T03:17:17 *** antimatroid has joined #aichallenge 2011-06-16T03:17:20 Zaphus: hi. 2011-06-16T03:17:26 McLeopold: /frac{1}{\frac{1}{3} + \frac{1}{5}} is that what you want? 2011-06-16T03:17:27 Is this better? 1 / (⅓ + ⅕) 2011-06-16T03:17:33 *** smellyhippy has joined #aichallenge 2011-06-16T03:17:38 ergh, \ at the start, not / 2011-06-16T03:17:41 mleise: yes, and how? 2011-06-16T03:18:26 sorry, i haven't followed the discussion. do you mean how to unicode? 2011-06-16T03:18:39 yep 2011-06-16T03:18:55 type charmap and search for 'fifth' 2011-06-16T03:19:28 So with a 1.85, two ants can kill three (mutually assured destruction anyway) 2011-06-16T03:20:25 actaully, that might actually make fights more interesting 2011-06-16T03:20:32 I'm a bit worried that it removes too much motivation to cluster though 2011-06-16T03:20:45 but you'd still want to cluster a bit 2011-06-16T03:21:00 1 can't kill 2, 2 can kill 3, 2 can't kill 4 2011-06-16T03:21:19 3 can kill 5, but now we need to consider position 2011-06-16T03:21:24 what about 1.4? 2011-06-16T03:21:38 you need 3 ants in a battle to kill more than 1:1 2011-06-16T03:21:39 is blockable 2011-06-16T03:21:42 Seems like you found a possible solution for the blocking situations. :) 2011-06-16T03:21:43 hm okay 2011-06-16T03:21:56 even 1.5 was blockable 2011-06-16T03:22:07 McLeopold: your comment the other day about combining iterative closest with damage, hell no :P 2011-06-16T03:22:22 i'd rather move to power, which would be more understandable :P 2011-06-16T03:22:23 although the space needed to block gets larger 2011-06-16T03:22:26 antimatroid: I coded it and I agree :D 2011-06-16T03:22:29 Interesting idea. Of course you now get standoff instead I imagine 2011-06-16T03:23:58 maybe damage increases for an ant with the number of enemies, whatever makes it so a local 2 on 1 works 2011-06-16T03:24:35 i don't like the idea of damage amounts differing depending on what the ant dos 2011-06-16T03:24:40 that's complicating the rules a lot 2011-06-16T03:24:51 i'd rather we moved to power method if you want to complicate stuff :P 2011-06-16T03:25:08 The occupied method blocked as well 2011-06-16T03:25:09 What is the power method ? 2011-06-16T03:25:48 Zaphus: enemies(A) := the set of enemys within the attack radius of A, A lives if |enemies(A)| < |enemies(B)| for all B \in enemies(A) 2011-06-16T03:26:34 So ants around you but out of range of attack 2011-06-16T03:26:59 In other words, if all attackers get hit by more ants than your own ant 2011-06-16T03:27:30 you can only kill ants more occupied with enemies than yourself 2011-06-16T03:28:13 Confusing in words, is there a pic somewhere ? 2011-06-16T03:28:58 its because 3 people are talking :p 2011-06-16T03:29:17 image.google.com 2011-06-16T03:29:29 is there a moon-clipse? today? 2011-06-16T03:30:26 oh, it only shown on the defualt google search site 2011-06-16T03:30:35 i mis-spelled google 2011-06-16T03:30:50 So an ant could kill one enemy but be killed by one on the other side 2011-06-16T03:31:38 it's the coolest rule (imo) to date, but it was deemed too complicated 2011-06-16T03:33:25 It is interesting, but agree it is complex 2011-06-16T03:33:38 oh, it wsa yesterday 2011-06-16T03:33:53 ive never seen a lunar eclipse 2011-06-16T03:37:53 *** okay has joined #aichallenge 2011-06-16T03:43:16 There's another lunar eclipse on Dec 10th I think 2011-06-16T03:45:54 yep, but thats months away, ill be dying until then 2011-06-16T03:46:19 (not really) 2011-06-16T03:46:26 Lol 2011-06-16T03:47:27 So, what's happening with ants, any sign of a launch date or anything? 2011-06-16T03:51:27 Zaphus: I've released v0.3 on the forums. Statistic now is shown for all bots that played all games. 2011-06-16T03:51:48 Cool! 2011-06-16T03:53:40 Zaphus: Still eating processor. I plan to fix it in v0.4. 2011-06-16T03:54:15 Zaphus: Here it is: http://ai-contest.com/forums/viewtopic.php?f=25&t=1401#p9057 2011-06-16T03:55:57 btw, I dissapointed a bit: v0.2 downloaded 32 times while v0.3 only 6 times. 2011-06-16T03:56:14 Will grab it when I get home (this is on my phone) 2011-06-16T03:56:57 That's because there's no movement on the contest, people get bored waiting... 2011-06-16T03:58:36 *** qacek has quit IRC (Quit: WeeChat 0.3.5) 2011-06-16T04:11:11 UncleVasya: It works in Wine and looks funny ! 2011-06-16T04:12:46 wow! I thought it cannot work under Wine. Or you just launched it and didn't tried to press Start button? 2011-06-16T04:13:16 uh yes ... I just pressed start 2011-06-16T04:13:54 let me install python... 2011-06-16T04:15:09 winetricks python .... is taking some time 2011-06-16T04:22:31 UncleVasya: It says "You must specify a map file." 2011-06-16T04:23:11 I think it's because of '\' instead of '/'. 2011-06-16T04:24:45 hmm I'm not quite sure of it. It can launch the Python program which uses '\' in the path 2011-06-16T04:26:58 mleise: somewhere i used TrailingPathDelimiter, somewhere I was too lazy and hardcoded '\'. So you can get a string with both of them. 2011-06-16T04:27:58 mleise: or maybe you are using an old environment. 2011-06-16T04:28:13 tools.zip was updated few days ago. 2011-06-16T04:29:57 hmm if I run the command with bare wine python.exe ... it works 2011-06-16T04:31:27 maybe it is the extra space after the map name 2011-06-16T04:32:00 UncleVasya: It looks like --map_file "H:\...\symmetric_10,map " 2011-06-16T04:35:41 mleise: in the play_one_game.cmd: --map_file %~dp0maps\symmetric_maps\symmetric_10.map %* 2011-06-16T04:35:41 in the play_one_game.sh: --map_file maps/symmetric_maps/symmetric_10.map "$@" 2011-06-16T04:35:41 2011-06-16T04:36:07 Maybe problem is here, I don't know what that symbol means. 2011-06-16T04:36:25 %* and "$@" 2011-06-16T04:36:57 it means: add any command line parameters here 2011-06-16T04:37:00 *** Zaphus has quit IRC (Quit: Bye) 2011-06-16T04:37:18 do you use the .cmd file in your tool? 2011-06-16T04:37:27 yes 2011-06-16T04:38:03 my tool build test.cmd file like play_one_game.cmd 2011-06-16T04:38:17 andthen launches it 2011-06-16T04:38:48 So I can delete %*? 2011-06-16T04:39:12 yes, but may i ask why you dont launch the playgame.py directly? 2011-06-16T04:39:45 one less step in between may be one less source of bugs :) 2011-06-16T04:39:57 mleise: you can find test.cmd in the specified directory and look at it. 2011-06-16T04:40:45 mleise: I am a bad programmer. I can't launch playgame.py directly :) 2011-06-16T04:44:09 UncleVasya: WinExec('python.exe ...', SW_NORMAL); 2011-06-16T04:46:51 mleise: heh, I've tried WinExec('playgame.py' + options, SW_HIDE). I'll try your solution, thanks. 2011-06-16T04:47:51 there is also ShellExecute which is a bit more complicated, but you can tell it to 'open' a file. So you could just put a .txt or .py in there and it would launch the associated program 2011-06-16T04:49:11 UncleVasya: And if you want to make your window resizable look at the anchors of the controls. They define how controls resize with the window! 2011-06-16T04:50:33 * mleise is happy that there is someone using Delphi that I can actually tech something. 2011-06-16T04:51:19 *teach 2011-06-16T04:52:25 *** Zaphus has joined #aichallenge 2011-06-16T04:52:50 UncleVasya: Can you give me that link again ? or is it just the one in the ai-contest forum ? 2011-06-16T04:53:18 mleise: Hah, I thought about something like "when the size of form is changed we are going to manually resize table to the old percentage of the form occupy". 2011-06-16T04:53:22 Zaphus: the same. 2011-06-16T04:53:53 mleise: I didn't know what Anchors properties means :) 2011-06-16T04:54:15 I thought so... it took me a long time to realize they are helpful 2011-06-16T04:54:30 There is a second option: autosize = true 2011-06-16T04:55:01 mleise: I still can't launch playgame directly. 2011-06-16T04:55:10 But you have to use a few fixed size Panels to create a layout with that 2011-06-16T04:55:35 UncleVasya: Good work ! now can you make it edit my code and make my bot smarter ? :-) 2011-06-16T04:55:53 UncleVasya: I cannot run Delphi on Wine :/ 2011-06-16T04:55:55 Let's suppose I have a string that contains all the text I write into the cmd-file. 2011-06-16T04:56:00 UncleVasya: speaking of bots, did you work out how to make your bot not require ocaml to be installed ? 2011-06-16T04:56:23 Zaphus: I've downloaded tools but didn't tried it yet. 2011-06-16T04:56:50 UncleVasya: You have these %dp0 in the cmd file. These have to be replaced of course 2011-06-16T04:57:16 Best is you change the working directory to where playgame.py is and just remove all %dp0 2011-06-16T04:57:41 mleise: If I do Write(test_file, test_string) and then launch test.cmd via WinExec - it works. 2011-06-16T04:57:58 "SetCurrentDir " 2011-06-16T04:59:30 UncleVasya: it works because cmd.exe knows how to replace %dp0 2011-06-16T04:59:54 You cannot use these when you launch a program the 'normal' way 2011-06-16T05:00:50 mleise: 1. I have SetCurrentDir in my code. 2. dp0 was only an example of play_one_game.cmd text and my test.cmd doesn't consist it. 2011-06-16T05:01:02 *contains 2011-06-16T05:01:33 hmm where do you save the .cmd again? 2011-06-16T05:01:38 In my test.cmd all paths are have full format 2011-06-16T05:01:59 Near playgame.py . 2011-06-16T05:02:12 And is there a chance you add a space after the map file name? 2011-06-16T05:03:00 Ahhh now I found it! 2011-06-16T05:03:25 Yes very bad!!! very VERY bad :p 2011-06-16T05:03:34 --map_file "H:\Entwicklung\Python\aichallenge\ants\maps\symmetric_maps\symmetric_10.map %*" 2011-06-16T05:04:11 But you remove that you said. So I'm waiting for the patch :) 2011-06-16T05:10:35 mleise: I've uploaded version without "%*" but forgot to give a link. http://www.filefactory.com/file/ccd6dfe/n/AntsTesting_0.3_Linux.7z 2011-06-16T05:11:06 Zaphus: If I'll try to make your bot smarter it become stupid :) 2011-06-16T05:15:00 mleise: WinExec(PAnsiChar('python.exe ' + test_string), SW_SHOW) ; gives me only a black screen flash. 2011-06-16T05:15:07 *** boegel has joined #aichallenge 2011-06-16T05:19:16 hmm at least pythin is starting then ! 2011-06-16T05:20:52 *** sir_macelon has joined #aichallenge 2011-06-16T05:21:03 hi 2011-06-16T05:22:08 hi 2011-06-16T05:22:13 hi 2011-06-16T05:22:32 flashmob :) 2011-06-16T05:22:48 been reading the damage/blocking discussion 2011-06-16T05:23:26 i don't like the ideas with changing the battle resolution 2011-06-16T05:23:56 i would rather see maps properlu adjusted so that such situations shouldn't be decisive 2011-06-16T05:24:35 or on the other hand make it as a part of the game 2011-06-16T05:25:13 like in go you have a live formations that cannot be killed 2011-06-16T05:26:11 you were also talking about corridors without considering the other end of it, so it is blockable only from one side 2011-06-16T05:26:37 unless you have blocked a peninsula 2011-06-16T05:27:23 blocking would just another cool tactic 2011-06-16T05:28:39 you have to be aware that you "sacrifice" ants so they will be standing still, but they could have been more useful if they would be moving towards food or enemy positions 2011-06-16T05:29:56 moreover if we would have symmetric maps, then every player would have an equal opportunity to set his defenses 2011-06-16T05:30:46 i don't get it why you don't like blocking so much? 2011-06-16T05:31:09 *** olexs has joined #aichallenge 2011-06-16T05:31:24 UncleVasya: It seems like wine starts the Linux Python interpreter. That's why the file names don't work 2011-06-16T05:32:36 any adjustments to the damage resolution would lead to bunching up won't be a good tactic, because a bunch of ants can be killed with fewer enemy ants 2011-06-16T05:32:48 that's just not right 2011-06-16T05:35:49 UncleVasya: You set the working directory to where playgame.py is, right? Then it may be fixable on wine if you use only relative paths and '/' instead of '\'. Windows understands that since Windows 2000. 2011-06-16T05:40:47 *** sigh has quit IRC (Ping timeout: 276 seconds) 2011-06-16T05:41:26 mleise: Is test.cmd works for you with changes you are talking about? 2011-06-16T05:42:30 sir_macelon: I agree, I'm yet to see proof that blocking is really bad, people just say it's a bit boring... so what ? whatever happens the top bots are going to be similar anyway and be a bit boring - doing standoffs waiting for someone to make a wrong move 2011-06-16T05:43:05 I cannot run test.cmd. Don't know how in wine. But I see Linux path names "/home/..." when I change playgame.py to print the working directory 2011-06-16T05:43:35 and if you're the bot with the lower score (which you cannot tell..) then you'll lose a standoff or blocking battle anyway 2011-06-16T05:43:40 so relative paths with '/' *should* work. But I understand if you don't want to make that change 2011-06-16T05:44:13 maybe you can launch test.cmd without Wine? Just like usual binary file. 2011-06-16T05:45:17 Zaphus: what 2011-06-16T05:45:18 hmm ok let me turn this into a shell script quickly 2011-06-16T05:45:25 's boring in blocking 2011-06-16T05:45:44 Zaphus: what's boring in blocking? 2011-06-16T05:45:58 it's just another tactic 2011-06-16T05:46:14 I don't think it's boring, but thats the only 'argument' I've heard against it - nobody in the Beta has demonstrated it being a problem so far 2011-06-16T05:46:23 it shouldn't be all about attacking 2011-06-16T05:46:43 but there should be also some defense possiblities for balance 2011-06-16T05:46:47 mleise: If it's the only thing that doesn'tlet my tool work under Linux, I'll fix it. Linux users are dominating here in humber. 2011-06-16T05:47:22 Look at Zaphus_BlockTest - it doesnt even block properly, just does a half-hearted attempt - but watching other bots work out ways to cut it back or block it in so it can't get any more food is good 2011-06-16T05:47:30 UncleVasya: yes it works! I replaced all absolute paths with relative ones and replaced '\' with '/' 2011-06-16T05:47:40 I don't mean I'll abandon Windows users, I mean I'll try to make app more portable. 2011-06-16T05:48:04 it depends on what you are blocking actually 2011-06-16T05:48:19 you may be blocking yourself 2011-06-16T05:48:24 in 2011-06-16T05:48:44 UncleVasya: If Delphi had a native Linux version I would still use it for GUI applications! 2011-06-16T05:50:02 I don't know if it is good or bad that wine launches the linux python. It's good because it is already installed and it is bad because absolute file names don't work 2011-06-16T05:51:16 mleise: If I just replace '\' with '/' but left absolute paths - will it work? 2011-06-16T05:51:33 no the absolute paths are the important part. 2011-06-16T05:52:21 playgame.py is launched with windows file names like "C:\..." but cannot find the files because on linux they start with "/..." 2011-06-16T05:53:12 ok 2011-06-16T05:55:28 BTW, here check out some nicely advanced battles: http://aichallengebeta.hypertriangle.com/visualizer.php?game=25256 2011-06-16T05:55:29 :) 2011-06-16T05:55:45 http://aichallengebeta.hypertriangle.com/visualizer.php?game=25056 2011-06-16T05:56:15 *** Naktibalda has joined #aichallenge 2011-06-16T06:00:45 *** olexs has quit IRC (Quit: Leaving.) 2011-06-16T06:03:17 sir_macelon: yes, I have been admiring your "chase down and kill" behaviour that your bot has, looks cool :-) 2011-06-16T06:15:33 *** stocha has joined #aichallenge 2011-06-16T06:15:58 hi, i'd like to learn the verilog language. Do you think it would be feasible to support/submit a verilog bot ? 2011-06-16T06:17:56 create a starter pack for it :) 2011-06-16T06:20:26 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-16T06:21:02