2011-06-01T00:03:36 *** yasith has joined #aichallenge 2011-06-01T00:10:12 aichallenge: McLeopold epsilon * r0c5c8c6 / (3 files): update to server stats - http://bit.ly/m0VPl8 2011-06-01T00:12:25 aichallenge: McLeopold epsilon * ra40f772 / website/server_stats.php : update to server stats - http://bit.ly/kexyKW 2011-06-01T00:32:11 *** computerwiz_222 has quit IRC (Remote host closed the connection) 2011-06-01T00:33:56 *** amstan has joined #aichallenge 2011-06-01T00:33:56 *** ChanServ sets mode: +o amstan 2011-06-01T00:40:16 amstan: we have worker stats now 2011-06-01T00:40:31 ooh, and 3 servers 2011-06-01T00:40:58 janzert: what was our games per minute during the finals last time? 2011-06-01T00:41:05 McLeopold: like 120 2011-06-01T00:41:48 something like that, I think it peaked around 180 or so 2011-06-01T00:41:55 don't really remember exactly 2011-06-01T00:42:09 I don't think we'll get anywhere near that this time 2011-06-01T00:42:18 I don't either 2011-06-01T00:43:07 of course we only need around 1/3 or 1/4 that to get the same number of players per minute 2011-06-01T00:43:10 I think next I need to get some average game length stats 2011-06-01T00:43:10 oh man, we have a ton of beta users 2011-06-01T00:43:24 those are just starter users created by hjax 2011-06-01T00:43:53 well.. no 2011-06-01T00:43:58 only 19 in the last 24 hours 2011-06-01T00:44:07 we still have 39 before then 2011-06-01T00:44:13 janzert: I could calc player-games per minute, or turns per minute 2011-06-01T00:45:01 games per minute and an average number of players in those games is probably sufficient 2011-06-01T00:45:25 *** UncleVasya has joined #aichallenge 2011-06-01T00:45:39 the square dance makes me lol http://aichallengebeta.hypertriangle.com/visualizer.php?game=9227 2011-06-01T00:45:58 i would like an average turn duration(ms) and average game duration(turns) 2011-06-01T00:46:05 k 2011-06-01T00:46:27 the square dance is pretty effective 2011-06-01T00:48:07 McLeopold: what's the argument against using symmetric food during all the games? 2011-06-01T00:49:18 the method doesn't work for all map types does it? 2011-06-01T00:49:50 it should.. 2011-06-01T00:50:51 ok, i'm going to bed 2011-06-01T00:51:19 oh, maybe something different is being used now. The method I had heard before I thought only worked with the offset tessellation type 2011-06-01T00:51:22 night 2011-06-01T00:52:36 yes, it should work for everything 2011-06-01T00:52:39 and random is not good either 2011-06-01T01:01:14 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-01T01:22:41 *** aeter has joined #aichallenge 2011-06-01T01:28:53 aichallenge: McLeopold epsilon * r3419893 / website/ranking.php : fixed rank change arrow - http://bit.ly/m0qysf 2011-06-01T01:33:07 amstan: from what I've seen, high food amount, wide attack and spawn radii and asymmetric maps make for a very low skill game 2011-06-01T01:34:10 I think we could add a bit more food, but not much. The small radii for attack and spawn are working well. 2011-06-01T01:34:39 We could do asymmetric, but we don't really have the time or manpower to analyze it to do it right. Maybe next time. 2011-06-01T01:34:56 Food spawning is symmetric and very fair right now. 2011-06-01T01:35:06 *** McLeopold has left #aichallenge 2011-06-01T01:35:11 McLeopold: wait! 2011-06-01T01:35:32 antimatroid: ping 2011-06-01T01:40:36 *** Ice_Harley has joined #aichallenge 2011-06-01T01:42:31 *** Palmik has joined #aichallenge 2011-06-01T01:47:30 aichallenge: janzert epsilon * rd8e59ca / (3 files in 2 dirs): Add worker start script and install it with worker setup - http://bit.ly/iNuUSG 2011-06-01T02:13:22 *** needsch has joined #aichallenge 2011-06-01T02:25:57 *** smiley1983 has joined #aichallenge 2011-06-01T02:28:39 *** yasith has quit IRC (Ping timeout: 240 seconds) 2011-06-01T02:52:48 *** UncleVasya has joined #aichallenge 2011-06-01T02:54:03 smiley1983: Hi. My problem is fixed now. 2011-06-01T02:55:58 I had to change 2011-06-01T02:56:00 method set_tile_busy (row col) to the method set_tile_busy (row, col) 2011-06-01T02:56:43 I am happy now :) 2011-06-01T03:03:25 I'm glad you got it working 2011-06-01T03:03:30 *** needsch has quit IRC (Quit: Leaving.) 2011-06-01T03:09:19 *** aeter has left #aichallenge 2011-06-01T03:09:52 UncleVasya: sorry I was slow responding yesterday, I was having connection problems. 2011-06-01T03:13:02 *** mcstar has joined #aichallenge 2011-06-01T03:15:13 *** berak has joined #aichallenge 2011-06-01T03:26:56 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-06-01T03:28:04 *** qacek has quit IRC (Quit: WeeChat 0.3.5) 2011-06-01T03:28:05 *** aerique has joined #aichallenge 2011-06-01T03:49:06 *** needsch has joined #aichallenge 2011-06-01T03:51:21 *** amstan has quit IRC (Remote host closed the connection) 2011-06-01T03:55:47 *** needsch has quit IRC (Ping timeout: 276 seconds) 2011-06-01T03:58:09 *** Naktibalda has joined #aichallenge 2011-06-01T03:59:15 *** olexs has joined #aichallenge 2011-06-01T04:01:38 *** janzert has quit IRC (Ping timeout: 276 seconds) 2011-06-01T04:07:51 *** UncleVasya has quit IRC (Quit: jmIrc-m v0.34a by Archangel (http://jmirc-m.net.ru/en/)) 2011-06-01T04:09:13 Frontier: pong 2011-06-01T04:09:22 McLeopold: how would you ever get enough data to test the fairness of maps? 2011-06-01T04:09:26 i just don't see that as possible 2011-06-01T04:17:04 *** janzert has joined #aichallenge 2011-06-01T04:29:03 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-06-01T04:42:09 *** yasith has joined #aichallenge 2011-06-01T04:42:31 *** yasith has quit IRC (Read error: Connection reset by peer) 2011-06-01T05:00:23 *** yasith has joined #aichallenge 2011-06-01T05:01:51 *** yasith has joined #aichallenge 2011-06-01T05:03:09 *** yasith has joined #aichallenge 2011-06-01T05:29:00 *** olexs has quit IRC (Quit: Leaving.) 2011-06-01T05:31:22 *** sigh has joined #aichallenge 2011-06-01T05:39:52 *** xQuasar has joined #aichallenge 2011-06-01T06:02:32 *** xQuasar has quit IRC (Ping timeout: 276 seconds) 2011-06-01T06:06:55 *** xQuasar has joined #aichallenge 2011-06-01T06:20:31 *** Quasar has joined #aichallenge 2011-06-01T06:20:57 *** Quasar is now known as Guest329 2011-06-01T06:21:33 *** hjax has joined #aichallenge 2011-06-01T06:22:21 *** hjax has quit IRC (Client Quit) 2011-06-01T06:23:38 *** xQuasar has quit IRC (Ping timeout: 244 seconds) 2011-06-01T06:35:51 *** Guest329 is now known as xQuasar 2011-06-01T06:46:53 *** delt0r_ has joined #aichallenge 2011-06-01T06:48:28 *** delt0r___ has quit IRC (Ping timeout: 258 seconds) 2011-06-01T06:57:27 *** UncleVasya has joined #aichallenge 2011-06-01T07:18:47 *** FireFly has joined #aichallenge 2011-06-01T07:23:15 *** yasith has joined #aichallenge 2011-06-01T07:38:22 *** mceier has joined #aichallenge 2011-06-01T07:49:53 *** xQuasar has quit IRC (Ping timeout: 252 seconds) 2011-06-01T07:50:33 *** xQuasar has joined #aichallenge 2011-06-01T07:53:19 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86.1 [Firefox 4.0.1/20110413222027]) 2011-06-01T08:09:06 *** mceier has quit IRC (Quit: leaving) 2011-06-01T08:30:59 *** xQuasar has quit IRC () 2011-06-01T08:43:14 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-01T09:12:21 *** Naktibalda has joined #aichallenge 2011-06-01T09:15:41 *** yasith has quit IRC (Ping timeout: 244 seconds) 2011-06-01T09:22:50 *** olexs has joined #aichallenge 2011-06-01T09:29:28 *** Eruonen has joined #aichallenge 2011-06-01T09:32:30 *** yasith has joined #aichallenge 2011-06-01T09:39:33 *** sigh has quit IRC (Remote host closed the connection) 2011-06-01T09:49:57 *** needsch has joined #aichallenge 2011-06-01T10:14:46 I looked at the website's "starting your own bot" page today and noticed it was still the one from PlanetWars. Here is a stub of something which could be used for an equivalent page for Ants: http://pastebin.com/0PZVgRMa 2011-06-01T10:15:00 *** Kingpin13 has joined #aichallenge 2011-06-01T10:27:48 *** BR has joined #aichallenge 2011-06-01T10:28:06 *** BR is now known as skylord5816 2011-06-01T10:28:09 *** skylord5816 has joined #aichallenge 2011-06-01T10:28:44 where do I get the visualizer? 2011-06-01T10:28:56 or the examples 2011-06-01T10:29:27 or anything besides the starter pack, for that matter 2011-06-01T10:30:06 try here: https://github.com/aichallenge/aichallenge/tree/d8e59ca61a75496ba072dc8f8ec10d135196ae61 2011-06-01T10:30:15 thanks 2011-06-01T10:30:54 @learn html_visualizer as http://aichallengebeta.hypertriangle.com/visualizer/html_visualizer.zip 2011-06-01T10:30:54 Frontier: As you wish. 2011-06-01T10:31:22 @learn java_visualizer as http://aichallengebeta.hypertriangle.com/visualizer/visualizer.jar 2011-06-01T10:31:22 Frontier: Ready to serve. 2011-06-01T10:35:49 *** yasith has quit IRC (Read error: Connection reset by peer) 2011-06-01T10:35:56 Frontier, what is the difference? 2011-06-01T10:36:12 between java and html 2011-06-01T10:37:02 the html visualizer is faster and offers a copy&paste solution for replay text, the java version on the other hand can be used in scripts by calling "java -jar visualizer.jar replayfile" 2011-06-01T10:37:33 html is faster than java? :) 2011-06-01T10:38:25 html with css3 is said to be turing complete, but the language is actually JavaScript. And JavaScript in a browser runs faster than through a Java library. That's why :) 2011-06-01T10:38:37 but then the html visualizer ist quite good performance-wise, it seems, it works flawlessly on my android even 2011-06-01T10:39:13 you can also call the html one in scripts using "python visualizer/visualize_locally.py < 0.replay" 2011-06-01T10:39:39 Chrome has a very good implementation of JavaScript. They even compile parts of it into machine code. 2011-06-01T10:40:03 *** yasith has joined #aichallenge 2011-06-01T10:40:06 the play_one_game script that was in repo earlier uses the html visualizer, i still use that to dev my bot 2011-06-01T10:40:07 as does Opera 2011-06-01T10:40:41 skylord5816: Yeah, I thought of that when I pressed Enter :p. 2011-06-01T10:41:12 But it seems right now Opera is the slowest for the visualizer. ...maybe IE is slower 2011-06-01T10:42:09 hm, wonder why 2011-06-01T10:42:22 I tried to find some docs on the internet that would help me optimize parts of the visualizer, but I couldn't find anything. 2011-06-01T10:42:22 canvas implementation? 2011-06-01T10:42:24 so, what in there do I download? 2011-06-01T10:43:17 Don't use Chrome's HW accelerated canvas 2d, btw. It is far from production quality 2011-06-01T10:44:38 Frontier: it's also not in stable yet, is it? 2011-06-01T10:44:55 skylord5816: you mean in the git repository? there are sample bots, the engine code etc... just clone the repository :) 2011-06-01T10:45:45 Frontier: ah, ok 2011-06-01T10:45:55 olexs: No it isn't and there is a big fat warning about it when you use it. That didn't stop a 14 year old from posting a screenshot and saying "The visualizer has a bug" 2011-06-01T10:46:14 hey!!! I'm 15!!! ;) 2011-06-01T10:46:19 lol 2011-06-01T10:46:41 :p After a few questions I found out that he was using Chrome 12 with experimental features active. 2011-06-01T10:47:25 Then amstan came around with the same problem claiming he didn't activate HW acceleration. I didn't believe him, he checked again and found out that it was active. 2011-06-01T10:48:31 what do I do if I have an idea for improvement of say, the website? issue in github? 2011-06-01T10:49:15 Yeah that's probably best. What kind of idea is it? 2011-06-01T10:49:34 for example, display version number in submission list on user page 2011-06-01T10:50:37 I should be able to implement that as well, just no idea how github works to be honest :) 2011-06-01T10:50:40 Ok, that's a good one. It makes no sense to have it only in the general overview, but not on the user's page 2011-06-01T10:51:18 visualizer works good for my Opera. Maybe i should check it speed in Chrome 2011-06-01T10:51:23 off to school, see you later 2011-06-01T10:51:25 *** skylord5816 has left #aichallenge 2011-06-01T10:51:36 olexs: You would have to sign up, then become an official member of our team, or fork the repository and issue a pull request :) 2011-06-01T10:52:14 Ice_Harley: Do that. The speed gain in Chrome and Firefox 4 is astonishing :) 2011-06-01T10:52:14 Frontier: signed up I am, and I have a repo clone as well. fork would be within github and I'd have to clone that separately, right? 2011-06-01T10:52:35 right 2011-06-01T10:53:20 ok, I'll see what I can do. 2011-06-01T10:53:41 Frontier: i don't see any difference in speed. Opera 11.11 2011-06-01T10:54:06 try adding &debug=true to the url to see the speed in the title bar 2011-06-01T10:54:45 and &config={"cpu":1} to get maximum speed 2011-06-01T10:55:17 ok 2011-06-01T10:55:36 You are on Windows? 2011-06-01T10:56:13 I noticed that Opera is quite a bit slower for me on Linux, so that may be it. 2011-06-01T10:56:26 how can I label the issue in github ("website")? or can only team members do that? 2011-06-01T10:56:42 olexs: no idea 2011-06-01T10:57:09 the gui for that is not intuitive though, maybe try a bit more :p 2011-06-01T10:57:38 there is a "Labels" are on the right, but no active controls there, so I assume I can't do anything there :) 2011-06-01T10:57:44 *area 2011-06-01T11:01:27 *** yasith_ has joined #aichallenge 2011-06-01T11:02:50 *** yasith has quit IRC (Ping timeout: 276 seconds) 2011-06-01T11:08:08 Frontier: you are right. 45-50 FPS on Chrome and 35 FPS on Opera. It's on Windows 2011-06-01T11:09:32 I just gave up on an MIT algorithm for nearest neighbor search. I think it is meant for large data sets anyway, not for a few 100 ants :p 2011-06-01T11:10:43 forget - it's very good now 2011-06-01T11:22:24 *** aerique has quit IRC (Quit: ...) 2011-06-01T11:26:16 *** Ice_Harley has quit IRC (Ping timeout: 252 seconds) 2011-06-01T11:26:52 *** UncleVasya has joined #aichallenge 2011-06-01T11:29:34 *** olexs has quit IRC (Quit: Leaving.) 2011-06-01T11:30:48 Frontier: what was that? 2011-06-01T11:32:02 mcstar: you mean the algorithm? 2011-06-01T11:32:07 yep 2011-06-01T11:32:14 link? 2011-06-01T11:32:40 in a se 2011-06-01T11:32:41 c 2011-06-01T11:33:10 UncleVasya: I've made another update to the starter. It's tiny and you don't need to get it. It just takes out the "time remaining" and "turn counts disagree" debugging output. However, it also adds some suggestions for incremental changes, which I've pasted here: http://pastebin.com/0PZVgRMa 2011-06-01T11:33:35 mcstar: http://people.csail.mit.edu/ruhl/papers/2002-stoc.html 2011-06-01T11:33:42 ty 2011-06-01T11:34:10 smiley1983: ok 2011-06-01T11:37:18 *** McLeopold has joined #aichallenge 2011-06-01T11:39:08 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-06-01T11:42:31 Frontier: judging from the first page its an interesting paper 2011-06-01T11:44:51 mcstar: that's what I thought. It can handle different metrics, and scales well in every aspect, including dynamic updates. 2011-06-01T11:46:35 But then I came across the passage where they want to add 24*c³ random sample points out of the set to a list. An if I understood that right that would be 192 points to go through. 2011-06-01T11:47:37 I'll focus on pathfinding a bit now. 2011-06-01T11:59:28 later ill give it a full read 2011-06-01T12:03:23 *** amstan has joined #aichallenge 2011-06-01T12:03:23 *** ChanServ sets mode: +o amstan 2011-06-01T12:03:31 hello 2011-06-01T12:07:30 amstan: 2011-06-01T12:07:38 I'm doing stuff to beta right now 2011-06-01T12:07:39 hi 2011-06-01T12:07:46 It will be back up in a minute or two 2011-06-01T12:09:25 aichallenge: janzert epsilon * r8658338 / (3 files in 2 dirs): More fixes for worker setup - http://bit.ly/inufyj 2011-06-01T12:10:58 janzert: It looks like we don't install clojure? 2011-06-01T12:11:50 not sure offhand, I did notice the language entry for it in compiler.py is certainly incomplete 2011-06-01T12:12:06 yeah, it's causing the worker to hang, so I'm fixing that 2011-06-01T12:20:16 looks like it's another language that we need to do a manual install in order to get a recent version :( 2011-06-01T12:20:35 fortunately it looks like that involves downloading and unpacking a zip to a known location 2011-06-01T12:21:19 are you already working on setting that up or do you want me to do it? 2011-06-01T12:22:50 no, you do it, I'm making sure the worker can survive the lack of a language 2011-06-01T12:23:03 ok 2011-06-01T12:23:58 beta just got really slow... 2011-06-01T12:26:11 *** berak_ has joined #aichallenge 2011-06-01T12:29:09 *** berak has quit IRC (Ping timeout: 250 seconds) 2011-06-01T12:29:22 *** berak_ is now known as berak 2011-06-01T12:35:59 amstan/janzert: I've created a screen session under the contest user. type "screen -x contest/worker" to access it. It should be running "~/aichallenge/worker/worker.py -t -n 0" unless one of us is doing diagnostics. 2011-06-01T12:36:21 I'm going to set one up for the manager as well. contest/manager 2011-06-01T12:36:29 McLeopold: just use another tab 2011-06-01T12:36:34 in the same screen 2011-06-01T12:36:42 what are the commands for that? 2011-06-01T12:36:52 ctrl+a ctrl+c makes a new one 2011-06-01T12:37:05 ctrl+a ctrl+a changes to the previous tab 2011-06-01T12:37:30 2 sessions will be easier for me, I can access the correct one by name 2011-06-01T12:37:33 ctrl+a ctrl+$i where $i e {0-9} changes to the $ith tab 2011-06-01T12:37:55 unless there is a big downside? 2011-06-01T12:38:26 you have to constantly switch screens sessions? 2011-06-01T12:38:31 instead of ctrl+a ctrl+a? 2011-06-01T12:39:00 okay, it's in one screen 2011-06-01T12:39:13 but, if I switch, do you see the switch? 2011-06-01T12:40:21 yeah, try it 2011-06-01T12:40:35 i think there's a thing at the bottom that tells you which tab you're in for a few seconds 2011-06-01T12:40:46 but it's pretty obvious when you switch tabs 2011-06-01T12:41:00 okay 2011-06-01T12:43:39 *** _0x47 <_0x47!~slash@p5B12E0E0.dip.t-dialin.net> has quit IRC (Ping timeout: 240 seconds) 2011-06-01T12:44:09 *** bhasker has quit IRC (Quit: bhasker) 2011-06-01T12:44:38 *** FireFly has quit IRC (Quit: swatted to death) 2011-06-01T12:48:11 *** smiley1983 has quit IRC (Quit: leaving) 2011-06-01T12:54:51 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-06-01T12:58:56 *** _0x47 <_0x47!~slash@p57B3E55D.dip.t-dialin.net> has joined #aichallenge 2011-06-01T12:59:18 *** aeter has joined #aichallenge 2011-06-01T13:01:59 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86.1 [Firefox 4.0.1/20110413222027]) 2011-06-01T13:04:07 *** Meatkat has joined #aichallenge 2011-06-01T13:05:15 anyone have an opinion where the clojure.jar should live? I'm thinking something like /usr/local/lib/clojure 2011-06-01T13:05:43 (this should be great bikeshed fodder since it can really go anywhere ;) ) 2011-06-01T13:11:38 *** olexs has joined #aichallenge 2011-06-01T13:15:12 *** Meatkat has quit IRC (Ping timeout: 240 seconds) 2011-06-01T13:16:27 *** Meatkat has joined #aichallenge 2011-06-01T13:16:47 *** bhasker has joined #aichallenge 2011-06-01T13:22:41 janzert: ubuntu puts it in /user/share/java 2011-06-01T13:24:31 hmm 2011-06-01T13:25:07 I was worried about putting it somewhere that is on the regular classpath that it might cause unintentional interference with another JVM entry 2011-06-01T13:25:46 well, I'd just go with the ubuntu way. then move the symlink over to the latest jar 2011-06-01T13:25:47 I suppose everything is very well namespaced so that isn't much of a concern though 2011-06-01T13:26:38 or is /usr/share/java not on the default classpath? 2011-06-01T13:28:45 *** yasith_ has quit IRC (Remote host closed the connection) 2011-06-01T13:30:25 yeah, looks like /usr/share/java isn't on the classpath 2011-06-01T13:31:18 so no concern at all with that location 2011-06-01T13:31:37 *** yasith has joined #aichallenge 2011-06-01T13:36:29 aichallenge: McLeopold epsilon * r21aea58 / worker/worker.py : update to error reporting on functional tests - http://bit.ly/lhyEIT 2011-06-01T13:43:00 *** olexs1 has joined #aichallenge 2011-06-01T13:46:11 *** olexs has quit IRC (Ping timeout: 276 seconds) 2011-06-01T13:48:28 McLeopold: any idea what the effective food rate range was before you got food spawning rate fixed? 2011-06-01T13:50:55 I'm just thinking I'd like to see a higher end on the range than there is currently, maybe even try doubling it to 1-12 2011-06-01T13:51:40 *** tomh has joined #aichallenge 2011-06-01T14:01:03 morning 2011-06-01T14:02:56 question what parameters are being used for food spawning on the beta server 2011-06-01T14:03:03 just want to make sure my local testing is similar 2011-06-01T14:06:20 anyone 2011-06-01T14:08:48 bhasker: pretty sure the ones in use are the ones in here https://github.com/aichallenge/aichallenge/blob/epsilon/setup/server_info.php.template 2011-06-01T14:09:56 ah k 2011-06-01T14:09:59 let me do a gitpull 2011-06-01T14:10:24 *** Naktibalda has joined #aichallenge 2011-06-01T14:11:00 aichallenge: janzert epsilon * r389dd7e / setup/worker_setup.py : Add clojure install to worker setup - http://bit.ly/ihAw9M 2011-06-01T14:11:35 btw i realized in the new design page the banner on the page is really big 2011-06-01T14:11:45 the banner + user profile takes off almost 50-60% of the page 2011-06-01T14:11:56 maybe we can reduce the height of the banner 2011-06-01T14:12:16 and make the user profile take up less vertical space? 2011-06-01T14:12:19 talk to amstan, he's been doing the design grunt work ;) 2011-06-01T14:12:46 or file a github issue I suppose 2011-06-01T14:13:36 http://aichallengebeta.hypertriangle.com/visualizer.php?game=10180 2011-06-01T14:13:41 lol my new bot just sat for 30 turns 2011-06-01T14:13:45 till a food spawned nearby 2011-06-01T14:13:49 then won the game:-\ 2011-06-01T14:14:06 heh 2011-06-01T14:14:16 suboptimal exploration code? :) 2011-06-01T14:14:22 yea bug probably 2011-06-01T14:14:30 have to see why it didn't do crap for 30 turns 2011-06-01T14:14:31 lol 2011-06-01T14:14:35 it should have made some random moves 2011-06-01T14:15:11 i keep track of squares i haven't seen yet and go to those if no food is nearby. seems to work rather well 2011-06-01T14:15:33 yea well thats what i did looks like its buggy 2011-06-01T14:17:14 hm... now I have modified the submissions table code in my repo fork, but can't test it, since I don't have the website running myself... should I just commit it and issue a pull request, since there is actually nothing that could break?.. 2011-06-01T14:17:44 *** mcstar has quit IRC (Quit: WeeChat 0.3.5) 2011-06-01T14:23:18 janzert: food spawning was 1 starting food per 100 land squares and 1 food per player per turn after that. 2011-06-01T14:23:54 The start food has been reduced to 1 per 200-300 land squares 2011-06-01T14:24:16 ok, and I've kind of changed my mind a bit from what I said above. the current top end rate is plenty for maps with a low number of players 2011-06-01T14:25:02 I think maybe the spawn rate should be propotional to the number of players in the game instead of absolute 2011-06-01T14:25:09 The food spawning is between 3 per turn and 1 every 20 turns, not dependant on players 2011-06-01T14:25:48 I could see it being tied to land area, but not players. 2011-06-01T14:25:49 for a 4 player map 3 per turn is plenty but for a 9 player seems a bit low for the top end 2011-06-01T14:26:01 land area should be tied to players somewhat 2011-06-01T14:26:20 that would probably work out ok as well 2011-06-01T14:26:44 I really like having games with very little food 2011-06-01T14:27:01 It means bots have to be good at both types of maps 2011-06-01T14:27:12 I like having some with very little food, but I want some with a pretty good amount too 2011-06-01T14:27:30 very little food seems to emphasize exploration 2011-06-01T14:28:53 i.e. I'm fine with the low end of spawn rate everywhere right now, and the high end of the rate on low player games, I just want a higher end rate for the larger player games 2011-06-01T14:29:50 1 per player per turn is too high, imo 2011-06-01T14:29:56 what should be our max? 2011-06-01T14:30:48 I was thinking of a range between .25 and .8 per player per turn earlier 2011-06-01T14:32:11 although that's a somewhat higher low end than current for large player maps, so maybe even .2 or so for the low end 2011-06-01T14:33:38 *** InsaneMalkavian has quit IRC (Ping timeout: 252 seconds) 2011-06-01T14:34:02 actually I guess the current low end is .11 for a 9 player map? 2011-06-01T14:34:18 lower I think 2011-06-01T14:34:26 maybe go with 1/8th, .125 2011-06-01T14:34:51 0.05 / #players right now 2011-06-01T14:34:58 isn't between 1 and 6 food per turn currently 2011-06-01T14:35:06 0.00555... 2011-06-01T14:35:29 food_rate / food_turn 2011-06-01T14:35:43 so 6/2 down to 1/10 2011-06-01T14:35:57 ahh, I didn't understand food_turn 2011-06-01T14:36:21 # players changes throughout the game 2011-06-01T14:36:37 let's define it as food per turn per land area 2011-06-01T14:36:55 I want to base it on total number of players not players left alive 2011-06-01T14:37:24 assumedly the alive players has taken over the area the dead players were in :} 2011-06-01T14:37:38 yes 2011-06-01T14:37:56 you want the rate to change after an elimination? 2011-06-01T14:38:08 no 2011-06-01T14:38:12 good :) 2011-06-01T14:38:27 land area will be higher for higher # of players 2011-06-01T14:38:43 but I'm thinking about it in terms of how often a player gets a chance at a new ant assuming everyone is alive still 2011-06-01T14:38:50 not in terms of land area 2011-06-01T14:39:02 so I can't really give you a rate I'd like in terms of land area :) 2011-06-01T14:39:24 how much land can 1 ant see? 2011-06-01T14:40:05 don't even know what the vision radius is atm :/ 2011-06-01T14:40:53 can someone from the team check my pull request? 2011-06-01T14:42:09 55 2011-06-01T14:42:22 1 ant sees 177 squares 2011-06-01T14:42:43 olexs1: I don't know the language, sry 2011-06-01T14:43:18 McLeopold: https://github.com/aichallenge/aichallenge/pull/140 ? 2011-06-01T14:43:23 it's a website change 2011-06-01T14:43:25 janzert: should we say an ant standing still should see food at a rate of 1 per 5-15 turns? 2011-06-01T14:45:02 I don't know :) I want to say a bot given perfect conditions and everyone else in the game should be able to get a new ant from every 5-8 turns up to a little less than every turn 2011-06-01T14:45:58 i.e. from somewhere around .125-.2 up to .8 or .9 food per player per turn :/ 2011-06-01T14:46:38 I'm not sure how to translate that feel into per ant area or land area 2011-06-01T14:47:25 *** delt0r___ has joined #aichallenge 2011-06-01T14:48:25 lol my new destination code is rather funny 2011-06-01T14:48:26 http://aichallengebeta.hypertriangle.com/visualizer.php?game=10220 2011-06-01T14:48:27 *** delt0r_ has quit IRC (Read error: Operation timed out) 2011-06-01T14:48:47 actually the low end could probably be lower than that, but less than 1 every 15 turns or so seems like a very slow and barren game 2011-06-01T14:49:58 I guess that would translate into a much lower rate than an ant standing still seeing a food within 15 turns? 2011-06-01T14:51:04 bhasker: that's a bit of a weird behaviour right at the start :) almost saw the food to the left and then decided to go right back 2011-06-01T14:51:16 yea lol 2011-06-01T14:51:18 but it still won! 2011-06-01T14:51:34 my code remembers its last destination and unless it gets better orders 2011-06-01T14:51:37 will keep giong 2011-06-01T14:51:42 *** Mathnerd314 has joined #aichallenge 2011-06-01T14:51:44 in this case it never saw any food 2011-06-01T14:51:54 heh 2011-06-01T14:51:54 and it always tracks to what it thinks is the closest edge locaiton 2011-06-01T14:51:59 so it just kept going back and forth 2011-06-01T14:52:06 will have to add some randomness in there 2011-06-01T14:52:21 I should improve spreading out in the beginning, my ants will mostly keep going all the same way instead of covering maximum area 2011-06-01T14:52:33 yea well i jsut added some code to do that 2011-06-01T14:52:44 so that they explore in all directions equally 2011-06-01T14:53:35 but this new code at least got rid of oscillation on the same two squares 2011-06-01T14:53:46 otherwise a lot of my ants just keep oscillating between two positions 2011-06-01T14:54:18 I had that when navigating to closest invisible square. got rid of it by changing invisible to unexplored 2011-06-01T14:54:41 hmm 2011-06-01T14:55:08 all squares are unexplored to start with, once a square is visible, it's marked as explored for the rest of the game 2011-06-01T14:56:04 later in the game this might result in suboptimal exploration, but at that point my bot is in a different mode altogether anyway 2011-06-01T14:59:14 janzert: a game with very low food turns into a game where the bot with the best formations should win 2011-06-01T15:00:15 hmm, I think it ends up more about luck of early exploration 2011-06-01T15:01:51 and with less food it seems like there are less chances for the randomness to even out 2011-06-01T15:02:00 1 food per 1-8 turns per land_area/#players 2011-06-01T15:02:25 with more food, the game results were more random 2011-06-01T15:03:57 so 1 food per x turns per land area per player right? 2011-06-01T15:04:32 thinking... 2011-06-01T15:05:49 at 2000 turns, the high end would give us 18000 ants for a 9 player game 2011-06-01T15:05:59 1 per turn per player 2011-06-01T15:06:21 doesn't the size of map effect it with that formulation though? 2011-06-01T15:06:30 no 2011-06-01T15:06:38 *** amstan has joined #aichallenge 2011-06-01T15:06:38 *** ChanServ sets mode: +o amstan 2011-06-01T15:06:50 so what does the land area term effect? 2011-06-01T15:07:14 ignore it :) 2011-06-01T15:07:18 ok 2011-06-01T15:07:29 you wanted a bot with perfect conditions to get 1 ant per turn at the high end 2011-06-01T15:07:33 so it's really 1 food per x turns per player 2011-06-01T15:07:37 yea 2011-06-01T15:08:37 I'm thinking make the range of x between 1-10 now but yeah that looks good to me 2011-06-01T15:08:49 I think that is just way to high 2011-06-01T15:08:55 although 1 seems just a tad high it's probably fine occasionally 2011-06-01T15:09:33 I really want the range to be fractional :P 2011-06-01T15:09:40 it is :) 2011-06-01T15:09:59 I mean 1 food per 1.2 turns per player 2011-06-01T15:10:10 it is :D 2011-06-01T15:10:47 6 food per 5 turns, food_rate/food_turn 2011-06-01T15:11:29 also, the lower food rates will be choosen more often right now 2011-06-01T15:11:38 ahh, then I want a continuous range of 1.5 to 10 :) 2011-06-01T15:12:49 with 9 players, that's 6 food per turn 2011-06-01T15:13:07 on the high end yeah 2011-06-01T15:13:28 and just under 1 food per turn on the low end 2011-06-01T15:13:30 I'll try that out, but I think it is too high 2011-06-01T15:13:56 do you mind the lower end being chosen more? 2011-06-01T15:16:16 I wouldn't mind seeing less of either extreme 2011-06-01T15:16:30 hmm... 2011-06-01T15:17:15 maybe a normal distribution over a range of 1 and 15? 2011-06-01T15:18:10 so, you don't have to do as good in low and high end games? I would think we want even distrubution 2011-06-01T15:18:12 or rather normal distribution over a 99% range of 1 and 15 2011-06-01T15:18:49 you do have to do as good in extreme conditions 2011-06-01T15:18:58 err, don't have to do as good 2011-06-01T15:19:22 but there will be less games with low and high end 2011-06-01T15:19:27 but I could see going the other way too 2011-06-01T15:20:02 right, that's why you don't need to do as well in the extreme's 2011-06-01T15:20:31 I think we should encourage doing well in both extremes 2011-06-01T15:20:34 I don't really have a strong feel one way or the other on distribution 2011-06-01T15:23:34 I kinda think the extreme games may be the least interesting to watch though 2011-06-01T15:33:07 is there a published clojure bot, that I could test with? 2011-06-01T15:35:04 *** Palmik has quit IRC (Remote host closed the connection) 2011-06-01T15:35:30 olex1, oh no, don't do that in a boolean fashion!, imagine being stuck in pocket, you can't go back then 2011-06-01T15:36:00 aichallenge: janzert epsilon * r0ae85f0 / worker/compiler.py : Fill out Clojure language specification - http://bit.ly/mEwaXl 2011-06-01T15:36:18 Clojure support is basically untested but should work now :} 2011-06-01T15:36:26 berak: good point. another reason for me to rewrite what is now about 10 lines of initial exploration code :) 2011-06-01T15:38:23 fixed another bug with exploration hopefully my bot does a bit better and doesn't sit doing nothing 2011-06-01T15:38:25 for a few turns 2011-06-01T15:38:30 though i still see some ants do it 2011-06-01T15:39:16 bhasker: do you test locally? early game behaviour should be pretty easy to debug 2011-06-01T15:39:34 yea i do 2011-06-01T15:40:04 i think early game behaviour is fine, its later that ants get a little confused :) 2011-06-01T15:40:21 ouch latest submission flunked compilation 2011-06-01T15:40:21 hmm 2011-06-01T15:44:32 berak: actually, after looking at my code again, the way I implemented it ants wont get stuck... it's looking further out that its view range for unexplored squares, so after exploring the whole pocket, it'll get out again and go somewhere else 2011-06-01T15:46:49 *** olexs1 has quit IRC (Quit: Leaving.) 2011-06-01T15:47:15 *** olexs has joined #aichallenge 2011-06-01T15:50:23 olexs1, just wanna say, if a game can take like 2000 turns, it's not wise to say: 'been there, seen that', situation might change.. 2011-06-01T15:51:05 berak: that is why I switch to a completely different logic at some point in the game, the "unexplored" system is only for the early phase 2011-06-01T15:51:39 my bot defines what i call a frontline 2011-06-01T15:52:05 of enemy ant squares/dead ants squares and squares which have either unexplored adjacent squares or squares that haven't been seen in some time 2011-06-01T15:52:11 then when an ant has nothing better to do 2011-06-01T15:52:16 it just routes to one of the frontline squares 2011-06-01T15:52:48 frontiers? :D 2011-06-01T15:52:49 in later game, I have a grid of ants keeping fog of war away, and so-called "interest points", where most of the other ants go, as well as some micro-management when encountering enemies 2011-06-01T15:53:06 frontiers indeed:) 2011-06-01T15:53:40 olexs: what kind of grid do you use? 2011-06-01T15:53:59 Frontier: hexagonal, to use minimal amount of ants for maximum view area 2011-06-01T15:54:36 hex grid, cool! 2011-06-01T15:54:59 GreenTea had a similar idea it seems, only his is rectangular and much tighter than mine 2011-06-01T15:55:00 olexs: me too - in theory, implementation has to follow now 2011-06-01T15:55:30 well to be fair a lot of people realized that a grid might be useful 2011-06-01T15:55:30 random shuffle causing weird behaviour 2011-06-01T15:55:31 http://aichallengebeta.hypertriangle.com/visualizer.php?game=10314 2011-06-01T15:55:40 Frontier: I use the viewradius var at game initialisation to calculate grid distance, and build it from there 2011-06-01T15:55:40 bots routing across each other lol 2011-06-01T15:55:42 ants* 2011-06-01T15:56:24 my micro needs massive improvement though, I lose way more ants in battles than necessary... 2011-06-01T15:57:02 with the current maps we can do evil things in micro *hint*hint* 2011-06-01T15:58:04 Frontier, micro ? 2011-06-01T15:58:22 my bot does a bit of micro:) 2011-06-01T15:58:29 it does some smart battles if possible 2011-06-01T15:58:41 at this point my micro-management is pretty much "am I alone ? ruuuuun : this is sparta!" 2011-06-01T15:58:42 but it doesn't do the evil I have in mind. 2011-06-01T16:00:26 however, I have an idea how I can improve it massively... just too lazy to actually sit down and code it properly 2011-06-01T16:01:32 Maybe a stupid question; can the locations of dead ants, given to the starter package, be the same as a location of an alive ant? 2011-06-01T16:01:50 no 2011-06-01T16:02:06 ok, thanks - wasn't sure. 2011-06-01T16:02:06 you can't walk on food 2011-06-01T16:02:29 any ants at a dead ant location would have been considered to "collide" with it and be dead also 2011-06-01T16:03:02 what do you guys think about the starting food amounts? 2011-06-01T16:03:36 *** Accoun has quit IRC () 2011-06-01T16:06:17 bhasker/olexs/Frontier/berak: starting food amount? good? bad? 2011-06-01T16:06:30 seems fine to me 2011-06-01T16:06:49 McLeopold, seems you just need longer games this way ( since they all have to walk some time to find food ) 100 turns not enough 2011-06-01T16:06:50 makes bots a bit more interesting than a food gatherer 2011-06-01T16:07:31 and battles count now 2011-06-01T16:07:34 bhasker +1 2011-06-01T16:07:42 stupid battle logic will kill the bot very fast 2011-06-01T16:07:53 a bit low for my taste. my bot loses some games just because he doesn't happen to come across a food piece in first few turns, and others do, by pure random 2011-06-01T16:08:03 that should not happen olex 2011-06-01T16:08:10 spawning is symmetric 2011-06-01T16:08:17 unless there was no food nearby 2011-06-01T16:08:17 okay 2011-06-01T16:08:23 in which case yes there is some luck 2011-06-01T16:08:31 I'm going to try and raise it a bit, so that a bot will on average see 1 food to start with 2011-06-01T16:08:43 I observed that yesterday and earlier today, don't know if you changed something since then 2011-06-01T16:09:46 McLeopold: that would be good I think. The bots need a little kick start to get the game to the interesting phase 2011-06-01T16:09:55 *** yasith has quit IRC (Ping timeout: 260 seconds) 2011-06-01T16:10:16 I want it low enough that a stupid bot doesn't do a cascading growth spurt 2011-06-01T16:10:24 then again sir_macelon seems to work well with the current amount 2011-06-01T16:10:44 Frontier: whatever gets you the win :p 2011-06-01T16:12:22 I need to decrible my bot, then I will see if I like the food amount. 2011-06-01T16:12:54 lol http://aichallengebeta.hypertriangle.com/visualizer.php?game=10332 2011-06-01T16:13:01 my bot sits in one place for 100 turns 2011-06-01T16:13:05 wtf is it doing 2011-06-01T16:13:12 and then wins :-\ 2011-06-01T16:13:15 and still wins, wtf :) 2011-06-01T16:13:41 somethign is seriously broken 2011-06-01T16:13:46 need to spend some time debugging 2011-06-01T16:13:51 it didn't start until it saw food 2011-06-01T16:13:56 yea 2011-06-01T16:13:57 weird 2011-06-01T16:14:03 my explorationc ode is broken 2011-06-01T16:14:05 :-\ 2011-06-01T16:14:15 seems so 2011-06-01T16:14:19 ode to the explorationc 2011-06-01T16:14:36 do we have a replay script like the one for planetwars to take a game and replay it to a bot 2011-06-01T16:15:02 at least my battle logic seems to be decent 2011-06-01T16:15:05 bhasker: we would need a replay to stream converter 2011-06-01T16:15:10 bot isn't doing stupid moves 2011-06-01T16:15:18 i thought the engine logs both streams and replays 2011-06-01T16:15:25 bhasker: care to write one in python? 2011-06-01T16:15:27 it does 2011-06-01T16:15:45 let me see what i can come up with 2011-06-01T16:15:46 but I'm not saving the stream on the server 2011-06-01T16:15:56 ah 2011-06-01T16:16:03 simpler to just store the stream zipped? 2011-06-01T16:16:05 bhasker: turn 150-152, are you sure? :) 2011-06-01T16:16:05 it's 20x larger 2011-06-01T16:16:09 rathere than back convert 2011-06-01T16:16:21 let me see olexs 2011-06-01T16:16:29 that was pretty much a suicide move 2011-06-01T16:16:37 bhasker: but then the worker needs to transmit it back to the main server 2011-06-01T16:16:57 hmm which location 2011-06-01T16:17:11 25:1 2011-06-01T16:17:16 oh the ant that didn't move 2011-06-01T16:17:20 I suppose I could store it on the worker, then replay it if requested? 2011-06-01T16:17:23 * berak still wonders, why this go-bot, doing straight random is doing so fine 2011-06-01T16:17:27 *** Accoun has joined #aichallenge 2011-06-01T16:17:31 not a suicide 2011-06-01T16:17:34 i took one of his ants with me 2011-06-01T16:17:37 i didn't lose it 2011-06-01T16:17:41 even trade 2011-06-01T16:17:42 okay :) 2011-06-01T16:17:48 i allow that 2011-06-01T16:17:51 explicitly 2011-06-01T16:17:52 random does surprisingly well 2011-06-01T16:20:00 we need a rage-bot for this ;) 2011-06-01T16:20:31 McLeopold: I need to show you something, and you tell me what you think. 2011-06-01T16:20:38 ok 2011-06-01T16:21:00 berak: HunterBot from python samples? :) 2011-06-01T16:21:50 i think, it's not the same idea 2011-06-01T16:22:21 it always goes for closest food or enemy 2011-06-01T16:22:31 it does not show that kind of hate ;) 2011-06-01T16:25:18 since it does not distinguish between food and enemy 2011-06-01T16:25:28 someone should upload it though, just to see how it does :) 2011-06-01T16:32:34 won't work here, in pw, you could win attacking in turn 1, (being daring), but here, you'll never see an opponent before ~ turn 50, and you're lucky finding food/spawning more ants before that 2011-06-01T16:32:41 just done that... let's see whatever happens. 2011-06-01T16:32:57 ;) 2011-06-01T16:33:38 it was actually fairly good among the test bots, at least until food got dialed back 2011-06-01T16:33:53 argh, forgot to rename the executable 2011-06-01T16:34:07 most fun for me, in this beta-phase, is that's almost anything possible 2011-06-01T16:34:53 ok, now it's in the contest, and I'm afk for a while 2011-06-01T16:36:04 * berak 's f5 key is the most shining one ... 2011-06-01T16:42:04 *** GreenTea has joined #aichallenge 2011-06-01T16:42:10 Hi 2011-06-01T16:42:17 some bug in visulizer 2011-06-01T16:42:29 incorrect length? 2011-06-01T16:42:43 yes.. 2011-06-01T16:42:55 heppen when by bot is timeout.. 2011-06-01T16:42:58 we know, you can add &debug=true to the query string 2011-06-01T16:43:14 we'll get to it soon :) 2011-06-01T16:43:43 ok) 2011-06-01T16:44:28 McLeopold: can you give more detail ? 2011-06-01T16:45:00 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86.1 [Firefox 4.0.1/20110518052719]) 2011-06-01T16:45:22 http://aichallengebeta.hypertriangle.com/visualizer.php?game=10366 2011-06-01T16:45:26 http://aichallengebeta.hypertriangle.com/visualizer.php?game=10366&debug=true 2011-06-01T16:45:36 ty 2011-06-01T16:46:49 *** hjax has joined #aichallenge 2011-06-01T16:47:05 lol someone uploaded the hunter bot 2011-06-01T16:47:09 smart move :) 2011-06-01T16:47:29 ^_^ 2011-06-01T16:48:06 im cornered between the hunterbot and the go starter 2011-06-01T16:48:39 now you are worse :) 2011-06-01T16:48:45 yep :/ 2011-06-01T16:49:02 the bots scores are sooo close 2011-06-01T16:49:25 less than 1, i can beat it after i fix my anti sucide 2011-06-01T16:49:53 btw want to tell me why my anti suicide doesnt work? 2011-06-01T16:50:05 sure 2011-06-01T16:50:50 im trying to keep a list of all moves, and use index to check if a move is in a list 2011-06-01T16:51:11 but my recent debug code showed that i messed something up 2011-06-01T16:52:47 http://aichallengebeta.hypertriangle.com/visualizer.php?game=10379&debug=true 2011-06-01T16:53:04 why after my timebout - Frontier got more then 3000 scores? 2011-06-01T16:53:21 *** |UncleVasya| <|UncleVasya|!kvirc@178-133-18-25.dialup.umc.net.ua> has joined #aichallenge 2011-06-01T16:53:22 * berak suicide / alan vega, they're still alive ! 2011-06-01T16:54:48 :D 2011-06-01T16:55:03 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 246 seconds) 2011-06-01T16:55:19 GreenTea: he got the food bonus, which is the food spawn rate times the number of turns 2011-06-01T16:55:48 %) 2011-06-01T16:55:50 turns is at 2000 2011-06-01T16:56:09 so (2000-267)*~1.5 2011-06-01T16:56:44 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-01T16:57:15 This is funny none the less, because GreenTea is so shocked 2011-06-01T16:57:29 lol 3k points 2011-06-01T17:01:20 Gr 2011-06-01T17:01:50 <|UncleVasya|> Hi. Today I've made a simple gui for testing ants. It's like a very-very basic version of Galcon Testing Arena. I wanna share it with you in hope that someone will find it useful. Where I can easy upload 350kb? 2011-06-01T17:02:44 <|UncleVasya|> It's Windows-only (works on my XP and I hope Vista in 7 will be ok too). 2011-06-01T17:02:46 GreenTea: i liked that idea of leaving some ants on a raster, to improve the vision, did you give up on this ? 2011-06-01T17:04:46 McLeopold 2011-06-01T17:05:13 i need to give my bot something to do if there is no targets in sight :/ http://aichallengebeta.hypertriangle.com/visualizer.php?game=10388 2011-06-01T17:09:02 |UncleVasya| thank you, I don't need it 2011-06-01T17:09:03 *** UncleVasya has joined #aichallenge 2011-06-01T17:09:41 berak, what you mean, improve vision? 2011-06-01T17:10:35 *** Meatkat has quit IRC (Ping timeout: 260 seconds) 2011-06-01T17:11:54 *** |UncleVasya| <|UncleVasya|!kvirc@178-133-18-25.dialup.umc.net.ua> has quit IRC (Ping timeout: 260 seconds) 2011-06-01T17:12:11 seems, you left some ants as a sitting duck, to keep that area within sight ( if there's food spawning there ) 2011-06-01T17:15:13 in regula distance 2011-06-01T17:16:30 this is good taktick when all food on map was eaten 2011-06-01T17:17:51 I thought that food spawning randomly.. 2011-06-01T17:17:58 yeah, like now, when new food only happening rarely ;) 2011-06-01T17:19:26 *** Meatkat has joined #aichallenge 2011-06-01T17:22:24 *** aeter has left #aichallenge 2011-06-01T17:22:40 *** hjax has quit IRC (Ping timeout: 252 seconds) 2011-06-01T17:24:01 who was solifugid, again, smiley ? 2011-06-01T17:24:16 GreenTea: If you run same map with same bots few times you'll see same things every time (if bots don't use random). So it seems like food spawning is calculating by some determined alhorytm 2011-06-01T17:25:07 :) 2011-06-01T17:25:24 this algorithm is random generator with constant seed.. 2011-06-01T17:26:56 *** Meatkat has quit IRC (Ping timeout: 240 seconds) 2011-06-01T17:27:49 *** FireFly has joined #aichallenge 2011-06-01T17:27:57 GreenTea: I've uploaded testing utility: http://www.filedropper.com/antstesting It's a podvig with my connection. 2011-06-01T17:28:45 *** Meatkat has joined #aichallenge 2011-06-01T17:31:12 UncleVasya: use a better pastebin, it's showing anything, but your code here 2011-06-01T17:32:30 There is no code. I uploaded 7z-archive with exe-file inside. 2011-06-01T17:32:39 it looks like someone trying to lure me into downloading a virus, lol 2011-06-01T17:33:13 My antennas give me bad signals. 2011-06-01T17:33:33 paranoid :) 2011-06-01T17:34:15 People here are good enough to not do things like that. 2011-06-01T17:34:34 better give code, you got nothing to loose, in this stage 2011-06-01T17:34:38 I have no AV software- was replaced with paranoia. Paranoia doesn't have to run as a background task slowing down the pc 2011-06-01T17:35:24 If you are awared I can pack project's src and upload it instead of exe. I've made it in Delphi. 2011-06-01T17:38:07 *** GreenTea has quit IRC (Quit: Page closed) 2011-06-01T17:38:54 *** GreenTea has joined #aichallenge 2011-06-01T17:39:43 src: http://www.filedropper.com/antstestingsrc 2011-06-01T17:53:11 Screenshots: http://www.imageupload.org/?d=4E0C1D731 2011-06-01T17:54:26 I think I had to start with screenshots, not with exe :) 2011-06-01T17:57:36 seems you got a taste for ugly pastebins, though 2011-06-01T17:58:02 *** Meatkat has quit IRC (Ping timeout: 250 seconds) 2011-06-01T17:59:04 berak: I dan't use any of them before. I used first ones I found in Google now. 2011-06-01T17:59:19 *hadn't 2011-06-01T18:00:26 much easy is to write script which lauch test.. 2011-06-01T18:00:54 then start it every time by one click.. 2011-06-01T18:01:52 ok 2011-06-01T18:09:23 *** ltriant has joined #aichallenge 2011-06-01T18:11:49 *** Meatkat has joined #aichallenge 2011-06-01T18:11:54 *** Frontier has quit IRC (Ping timeout: 250 seconds) 2011-06-01T18:12:25 *** UncleVasya has quit IRC (Read error: Connection reset by peer) 2011-06-01T18:16:24 *** berak has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443]) 2011-06-01T18:19:43 McLeopold: Could we move "Serial" out of game options and make it a worker option easily? 2011-06-01T18:20:07 maybe it's not worth it though 2011-06-01T18:26:34 *** sigh has joined #aichallenge 2011-06-01T18:38:15 *** sigh has quit IRC (Remote host closed the connection) 2011-06-01T18:45:29 *** Redgis has joined #aichallenge 2011-06-01T18:45:54 *** FireFly has quit IRC (Quit: swatted to death) 2011-06-01T18:59:25 so question, when a bot times out why does the replay get borked? 2011-06-01T19:00:56 answer that and you can probably fix the problem :} I believe a further constraint on when the problem appears is only when the starting ant for that player is still alive 2011-06-01T19:01:36 at least I think that is what Frontier tracked it down to 2011-06-01T19:01:47 hmm k 2011-06-01T19:03:26 and are the workers down? 2011-06-01T19:05:10 looks like possibly 2011-06-01T19:05:31 ah k 2011-06-01T19:10:54 What up! 2011-06-01T19:10:59 Is amstan around? 2011-06-01T19:11:04 yo 2011-06-01T19:11:16 Hey. I am at a Reddit coding night in Toronto. 2011-06-01T19:11:30 cool 2011-06-01T19:11:35 have you seen the beta site? 2011-06-01T19:11:42 I have yes. 2011-06-01T19:11:45 Played with it a bit. 2011-06-01T19:12:03 I noticed theres a bit of a leaderboard going. Are those mostly real people? 2011-06-01T19:12:16 j3camero: yes, people from here 2011-06-01T19:12:20 Sick. 2011-06-01T19:12:28 Then we must be pretty close to ready to go. 2011-06-01T19:12:54 yeah, so what's the plan? 2011-06-01T19:13:08 biggest thing missing for launch at this point is probably email sending 2011-06-01T19:13:32 http://aichallengebeta.hypertriangle.com/visualizer.php?game=10541 omg 2011-06-01T19:13:44 *** needsch has quit IRC (Read error: Connection reset by peer) 2011-06-01T19:15:07 Yeah last time I tried to send email through Gmail's servers (using @ai-contest.com addresses). They blacklisted us after only a few minutes after launch. 2011-06-01T19:15:08 GreenTea: lol, soo close 2011-06-01T19:15:26 gmail has a limit of 500 per day 2011-06-01T19:15:34 Oh wow. 2011-06-01T19:15:43 But makes sense if you only want humans using it. 2011-06-01T19:15:48 I think the amazon mail service is probably the way to go 2011-06-01T19:15:53 Who sends more than 500 emails a day pff. 2011-06-01T19:15:56 who's doing the hosting? you just buy access to a relay 2011-06-01T19:15:58 Okay. 2011-06-01T19:16:43 Last time we were just sending mail right from my dedicated server's mail server. Surprisingly, hotmail and gmail seemed to accept our emails as valid and non-spam. Could we do that again? It's so free. 2011-06-01T19:17:19 I brought this up before, but do we really need email other than for registering? 2011-06-01T19:17:36 Not at this point I don't think. 2011-06-01T19:17:49 McLeopold: not really, not anymore, since we have the errors being displayed on the user page 2011-06-01T19:17:50 Just lost password. I don't even know if we have that. 2011-06-01T19:18:01 no way to change passwords.. 2011-06-01T19:18:21 and i'm really not going to bother, anyone who loses their password must submit a patch to add the recovery feature, lol 2011-06-01T19:18:24 if we are only sending registration/password reset emails I don't think we'll hit amazons free limit of 2000 free emails 2011-06-01T19:18:34 janzert: /day? 2011-06-01T19:18:42 yes 2011-06-01T19:18:46 oh, decent 2011-06-01T19:18:51 Oh wait yeah. The compile status emails. We may be able to do without those, especially since people's entries are being compiled by the workers. 2011-06-01T19:18:53 i'd go with that 2011-06-01T19:19:06 http://aws.amazon.com/ses/pricing/ 2011-06-01T19:20:13 McLeopold has the results nicely integrated into the profile page now anyway 2011-06-01T19:20:16 We will have to pay 0.10 for every extra thousand. Geez guys that could bankrupt us :-) 2011-06-01T19:20:22 an email wouldn't really add much of anything 2011-06-01T19:20:52 So how do people find out if their code doesnt compile? 2011-06-01T19:20:58 refresh the page 2011-06-01T19:21:01 look at their profile page 2011-06-01T19:21:06 Ah nice. 2011-06-01T19:21:14 Well why don't I just try it. 2011-06-01T19:21:16 j3camero: example: http://aichallengebeta.hypertriangle.com/profile.php?user=35 2011-06-01T19:21:18 How can I create an account? 2011-06-01T19:21:20 at the bottom 2011-06-01T19:21:24 j3camero: top left 2011-06-01T19:22:03 Ohhh wow. 2011-06-01T19:22:06 That's nice. 2011-06-01T19:22:49 *** GreenTea has quit IRC (Quit: Page closed) 2011-06-01T19:23:07 What if the compiler output is really long? Does it get truncated to the first few lines? 2011-06-01T19:23:24 first 1024 chars 2011-06-01T19:23:52 j3camero: people are supposed to test their bots locally anyway, the functional test is included in the tools download 2011-06-01T19:23:56 Oh solid. Did you check with a few languages' compilers to make sure that will usually be enough? 2011-06-01T19:24:15 nobody tested that that i know 2011-06-01T19:24:18 no, but we've got most language in the system with no complaints 2011-06-01T19:24:28 Okay. 2011-06-01T19:25:05 In practice, most people will happily use the site as their compiler. Hopefully people will be able to get somewhat clear error messages from the site. 2011-06-01T19:25:21 McLeopold: maybe we should do tail instead? 2011-06-01T19:25:30 i know python gives the most relevant error at the end 2011-06-01T19:25:40 maybe 2011-06-01T19:25:50 amstan: but for C/++ you want the head :( 2011-06-01T19:26:00 I was thinking about that before 2011-06-01T19:26:01 but I want to encourage people not to fill up stderr, so I went with head 2011-06-01T19:26:04 first 512 and last 512? 2011-06-01T19:26:11 Maybe a head/tail thing. 1024 from the head, 1024 from the tail. Unless it's less than 2048. 2011-06-01T19:27:08 what is there now is probably fine for the moment though 2011-06-01T19:27:17 For sure. 2011-06-01T19:27:22 McLeopold: what do you think of changing the serial game option to an integer on how many bots to run in parallel? 2011-06-01T19:27:46 support is already in engine 2011-06-01T19:27:50 sure 2011-06-01T19:28:05 but, will in make a difference? 2011-06-01T19:28:23 I want to run as many bots in parallel as there are cores on the worker 2011-06-01T19:28:38 I've actually got it fleshed out in the engine right now 2011-06-01T19:28:42 How sensitive are the workers to the Ubuntu version? Do they need a specific version of Ubuntu? 2011-06-01T19:28:48 j3camero: yes, natty 2011-06-01T19:28:48 just not at all tested :/ 2011-06-01T19:28:54 Okay. 2011-06-01T19:29:30 there's quite a few things that had to be hardcoded and that change between ubuntu versions 2011-06-01T19:30:04 We could try upping the games/minute by using boxes owned by us and other IRC people as workers in addition to the EC2 workers. good idea/bad idea? 2011-06-01T19:30:16 bad idea 2011-06-01T19:30:21 ;) 2011-06-01T19:30:24 Haha okay. 2011-06-01T19:30:31 Why? Different specs would cause problems? 2011-06-01T19:30:54 yeah, I think people will really complain about having differences in different workers 2011-06-01T19:31:11 all these starter bots are annoying :-\ 2011-06-01T19:31:34 there was a fair amount of complaining about having games on the main server and workers while that was happening last time 2011-06-01T19:35:30 http://aichallengebeta.hypertriangle.com/visualizer.php?game=10561 2011-06-01T19:35:44 interesting game frontier 2011-06-01T19:35:57 though my bot's exploration logic is oor 2011-06-01T19:35:58 poor* 2011-06-01T19:36:27 how do you do against the square dancer? 2011-06-01T19:36:35 who is that 2011-06-01T19:37:32 he was 5th a while ago 2011-06-01T19:40:55 McLeopold: is anyone actively working on the engine at this point? 2011-06-01T19:41:18 I'm working on ants.py, not engine.py 2011-06-01T19:42:26 ok, I've got serial mode changed to allow a set number of bots at a time 2011-06-01T19:42:51 janzert: I thought it was already there? 2011-06-01T19:42:57 not quite 2011-06-01T19:43:23 the basis for it was, but it could only handle everyone at once or one at a time 2011-06-01T19:43:37 oh, yeah, I see the TODO's now 2011-06-01T19:44:07 did you suffle the bot order ;) 2011-06-01T19:44:12 So amstan. Do you feel like being a director of a corporation? :-P 2011-06-01T19:44:23 aichallenge: janzert epsilon * recb0178 / worker/engine.py : Allow a restricted number of bots at once in serial mode - http://bit.ly/lIUwnC 2011-06-01T19:44:33 shuffle I did :) 2011-06-01T19:45:36 the next thing that is bugging me is that all the bots are kept in one list for the whole time and it constantly has to be checked if a bot is alive or not 2011-06-01T19:46:16 but I'll probably just leave that for now 2011-06-01T19:46:47 I think we did that so the numbers don't get off 2011-06-01T19:47:18 yeah, I'd just give them an explicit player number 2011-06-01T19:47:59 also that way the numbers don't have to passed around separately 2011-06-01T19:48:37 but I'll not touch it right now 2011-06-01T19:53:56 *** Kingpin13 has quit IRC (Quit: time to go) 2011-06-01T19:56:31 I wonder if people would prefer .5GB of memory and 1.3 times more processing power or 1GB of memory and less processing power 2011-06-01T19:57:44 oops, that's actually not a choice :} 2011-06-01T20:00:08 it's basically 1GB of memory per bot and 1.3x times processing power for each turn but a quarter the throughput (in games per unit of time) (2 cores instead of 8) 2011-06-01T20:00:25 1gb memory per bot? that's a lot 2011-06-01T20:00:36 Are the bots memory-limited in some hard way? 2011-06-01T20:00:40 yes 2011-06-01T20:01:01 Oh nice. Using the same shady sandbox that I cobbled together last contest? 2011-06-01T20:01:03 currently the sandbox is set to just over a 1GB limit 2011-06-01T20:01:15 not quite 2011-06-01T20:01:16 That should be plenty haha. 2011-06-01T20:01:24 built off of it though 2011-06-01T20:01:25 were passing language specific params as well 2011-06-01T20:01:45 it chroots the sandboxed process 2011-06-01T20:01:54 Oh nice. More condoms. 2011-06-01T20:02:40 And the language-specific treatment should really help. The sandbox tends to irritate users of some languages more than others. 2011-06-01T20:02:43 I'm not completely sure java will start with a smaller hard limit even with giving the lower memory options 2011-06-01T20:03:00 guess I should test that 2011-06-01T20:03:13 janzert: so far, I think you're right because java is still crashing on turn 1 2011-06-01T20:03:25 We need to test on a real server 2011-06-01T20:04:37 java is such a pain, too bad it's so popular :P 2011-06-01T20:04:46 Okay, so I need to do a ton of paperwork before we can get mega google money (or Facebook or Amazon or whatever) 2011-06-01T20:05:24 So I don't know what to do about that. Ideally we would be able to light up a few servers right now to get things rolling. 2011-06-01T20:05:46 do the paperwork? 2011-06-01T20:05:56 Yep, working on that now! 2011-06-01T20:06:23 What would it cost per month to light up a web server and two workers at this time? 2011-06-01T20:07:16 I just need a ballpark idea what each worker costs per month basically. 2011-06-01T20:07:23 All-in. 2011-06-01T20:08:19 aichallenge: McLeopold epsilon * rc13ff98 / (4 files in 3 dirs): update to ensure ants have starting food, upped food amounts - http://bit.ly/iOPYNa 2011-06-01T20:08:20 aichallenge: McLeopold epsilon * r6c07018 / (setup/worker_setup.py worker/compiler.py worker/engine.py): Merge branch 'epsilon' of https://github.com/aichallenge/aichallenge into epsilon - http://bit.ly/kRaofg 2011-06-01T20:09:42 webserver at most $0.34 per hour and worker at most $0.25 per hour on ec2 for both 2011-06-01T20:10:13 so ~245 for server and 180 per worker 2011-06-01T20:10:23 My current plan is to basically start funding the contest out of pocket (given my stable employment and low expenses and such) and then have the corporation pay me back once it's set up. That will be good because it will make me actually do stuff. If I don't, I won't get paid back :-) 2011-06-01T20:10:33 that's maximum amount basically 2011-06-01T20:10:37 Okay that's reasonable. Thanks janzert. 2011-06-01T20:11:10 I haven't really decided what instance type is best bang for the buck for workers still 2011-06-01T20:11:15 So 600 per month for the server and two workers, to get started. 2011-06-01T20:11:40 yeah, that would do it for sure 2011-06-01T20:12:01 I think ultimately we will be able to raise whatever cash we need. So we don't have to be stingy. The concern is purely from an engineering point of view, which instance type gives us the best games/minute/dollar. 2011-06-01T20:12:27 also I do worry about the instance stability for the main server though 2011-06-01T20:12:42 We don't have to use EC2 if it's a problem. 2011-06-01T20:12:57 The web server is the only thing that really has to stay up. 2011-06-01T20:13:11 yeah, I'm thinking a more traditional server for the webserver might be better 2011-06-01T20:13:16 say linode or something 2011-06-01T20:13:38 Yep. What does that mean for the cost of sending data in and out of ec2? 2011-06-01T20:13:48 I assume that's pretty light. 2011-06-01T20:14:06 Game assignments go out, results come back in. Here I am assuming that the database is with the web server. 2011-06-01T20:14:30 I don't think it should be huge, it's basically $0.10 per GB in and $0.15 per GB out 2011-06-01T20:14:43 Okay that should be practically nothing. 2011-06-01T20:15:04 yeah, the instance cost should completely out weigh that I'm pretty sure 2011-06-01T20:15:36 Cool cool. 2011-06-01T20:16:17 *** tomh has quit IRC (Quit: tomh) 2011-06-01T20:17:10 In the past I have done pretty extensive stress testing before launching. I think we should light up an instance on real servers, then use scripts to create a few thousand accounts and submit thousands of bots that all behave differently. We can try a few scenarios to make sure that the whole house of cards doesnt blow over when we launch. 2011-06-01T20:30:09 Okay I am out for the night. Catch y'all later. 2011-06-01T20:38:39 *** computerwiz_222 has joined #aichallenge 2011-06-01T20:41:25 *** Eruonen has quit IRC () 2011-06-01T20:51:40 *** hjax has joined #aichallenge 2011-06-01T20:53:46 *** bhasker has quit IRC (Read error: Operation timed out) 2011-06-01T21:17:09 *** hjax has quit IRC (Quit: Rooms • iPhone IRC Client • http://www.roomsapp.mobi) 2011-06-01T21:41:50 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-06-01T21:54:51 *** olexs has quit IRC (Quit: Leaving.) 2011-06-01T21:56:09 *** Redgis has quit IRC (Ping timeout: 255 seconds) 2011-06-01T22:38:28 *** qacek has joined #aichallenge 2011-06-01T22:42:08 http://aichallengebeta.hypertriangle.com/visualizer.php?game=10770 2011-06-01T22:43:02 *** computerwiz_222 has quit IRC (Remote host closed the connection) 2011-06-01T22:47:26 *** delt0r_ has joined #aichallenge 2011-06-01T22:49:06 *** delt0r___ has quit IRC (Ping timeout: 258 seconds) 2011-06-01T22:57:19 *** amstan has joined #aichallenge 2011-06-01T22:57:19 *** ChanServ sets mode: +o amstan 2011-06-01T22:59:27 what's new? 2011-06-01T23:07:50 amstan: Having trouble with golang 2011-06-01T23:08:04 how so? 2011-06-01T23:08:31 I'm on i386 and have to use 8g and 8l. I'm still trying to get it to work 2011-06-01T23:08:46 yeah 2011-06-01T23:08:57 and the worker is hardcoded on 64 bit in a few places 2011-06-01T23:09:00 i wouldn't bother 2011-06-01T23:09:15 i gave up on my worker 2011-06-01T23:09:23 i should delete that branch 2011-06-01T23:10:24 It's just settings in compiler.py I think 2011-06-01T23:11:21 are you looking to make the code on the repo work? or just for your comp? 2011-06-01T23:11:58 just for me 2011-06-01T23:12:31 i think grep -Re 6\w . should do it 2011-06-01T23:13:04 hmm, it doesn't.. 2011-06-01T23:13:38 at line 222 and 223 of compiler.py change the all the 6s to 8s 2011-06-01T23:13:54 s/6/8/gc 2011-06-01T23:14:17 I got that, but now I'm getting 'Output file /home/contest/compiled/tmp???/bot/MyBot was not created 2011-06-01T23:14:36 what's with the ??? 2011-06-01T23:14:40 ? 2011-06-01T23:15:15 McLeopold: hmm, not sure. pretty sure that is all I changed when moving it to 64bit 2011-06-01T23:15:51 it actually still needs the nukeglobs expanded to delete both *.8 and *.6 2011-06-01T23:16:30 McLeopold: this is under linux right? 2011-06-01T23:16:41 well obviously from the path 2011-06-01T23:17:27 yeah, ubuntu 2011-06-01T23:18:54 *** Apophis_ has quit IRC (Read error: Connection reset by peer) 2011-06-01T23:19:18 aichallenge: janzert epsilon * r0605525 / worker/compiler.py : Expand oft forgotten golang nukeglobs to get both 32 and 64 bit object files - http://bit.ly/mrtG07 2011-06-01T23:19:18 aichallenge: janzert epsilon * r26c1f3f / (4 files in 3 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://bit.ly/mJ091f 2011-06-01T23:19:21 *** Apophis_ has joined #aichallenge 2011-06-01T23:20:56 *** Frontier has joined #aichallenge 2011-06-01T23:32:58 janzert: ./main.go:46: too many arguments in call to os.Open 2011-06-01T23:33:37 O.o 2011-06-01T23:33:43 looks like a code issue? 2011-06-01T23:33:55 arg 2011-06-01T23:33:58 yeah, in the compiler itself though? 2011-06-01T23:34:05 the go compiler that is 2011-06-01T23:34:19 or is main.go a bot file? 2011-06-01T23:35:06 it's a bot file 2011-06-01T23:35:15 gtg 2011-06-01T23:35:25 ahh 2011-06-01T23:49:43 contestbot: later tell hjax how's it going? 2011-06-01T23:49:43 amstan: OK 2011-06-01T23:57:38 so how do I create maps for a server? 2011-06-01T23:57:59 is there a script to create n maps at once? 2011-06-01T23:59:09 see symmetric_mapgen.py, but i don't think it creates more