2011-09-22T00:00:56 *** Keth has joined #aichallenge 2011-09-22T00:00:56 *** Keth_ has quit IRC (Read error: Connection reset by peer) 2011-09-22T00:03:06 *** dvladim has joined #aichallenge 2011-09-22T00:06:42 *** amstan has joined #aichallenge 2011-09-22T00:06:42 *** ChanServ sets mode: +o amstan 2011-09-22T00:16:47 *** dvladim has quit IRC (Read error: Operation timed out) 2011-09-22T00:20:19 contestbot: later tell McLeopold current games, maps and uploads are copied to new server. we can use rsync to copy any new ones over as well 2011-09-22T00:20:19 janzert: Yes master! 2011-09-22T00:37:00 *** McLeopold has joined #aichallenge 2011-09-22T00:37:44 amstan: can we get an A record for beta.ai-contest.com? 2011-09-22T00:37:57 McLeopold: i don't have dns access 2011-09-22T00:38:17 can you text jeff for dns access? 2011-09-22T00:38:41 i think an email would work better, last time i din't get a reply 2011-09-22T00:38:56 what i could do is redirect aichallengebeta.hypertriangle.com 2011-09-22T00:39:17 don't do that yet :} 2011-09-22T00:39:20 no, we can all create new names 2011-09-22T00:39:34 but I think we should once everything is moved over 2011-09-22T00:39:35 I want to launch, and that means official dns records 2011-09-22T00:39:46 yes, eventually 2011-09-22T00:39:57 eventually as in within 48 hours 2011-09-22T00:40:10 you want to launch in 48 hours? 2011-09-22T00:40:15 beta, yes 2011-09-22T00:40:16 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 276 seconds) 2011-09-22T00:40:38 or at least get the server running with the right dns record 2011-09-22T00:40:55 we can do final beta without changing dns 2011-09-22T00:41:11 ai-contest.com? 2011-09-22T00:42:13 what we do need to find out from jeff is that money is available 2011-09-22T00:42:15 we had talked about a pw.ai-contest.com as well, so people can still look at the old site 2011-09-22T00:43:08 after all my finagling today I'm rather convinced the main server shouldn't be a spot instance, which is fine but raises the cost to more than I'm willing to spend on the contest :/ 2011-09-22T00:44:29 McLeopold: can you move the database...err...data over? 2011-09-22T00:44:50 I think everything else is setup 2011-09-22T00:45:25 ok 2011-09-22T00:45:47 I can shut down the current workers whenever you want to stop games being processed 2011-09-22T00:46:07 yeah, shut them down now 2011-09-22T00:46:18 and I suppose we should shut off submissions as well 2011-09-22T00:47:53 hmm, out of curiosity does turning off submissions stop new user account creation? 2011-09-22T00:49:14 workers are down 2011-09-22T00:49:35 ok, give me 5 min 2011-09-22T00:52:34 amstan: want the same username and ssh key from the current beta server setup? 2011-09-22T00:52:50 sure 2011-09-22T00:53:12 what about the pass? i still need it for root 2011-09-22T00:53:43 I'm setting up all accounts iwth a disabled password and just ssh access 2011-09-22T00:53:50 ok 2011-09-22T00:54:50 we need to get jeff back on IRC for the next 2 weeks 2011-09-22T00:54:59 then he can go awol again 2011-09-22T00:56:39 amstan: your account should be setup 2011-09-22T00:57:20 janzert: what was the hostname? 2011-09-22T00:57:23 okay, data is moved, but do we run workers on both systems? 2011-09-22T00:57:30 antsbeta.janzert.com 2011-09-22T00:57:41 it seems like we should stick with one 2011-09-22T00:57:51 I'll switch the workers to run against the new one 2011-09-22T00:57:57 janzert: i get permission denied 2011-09-22T00:58:00 so, do we point everyone there? 2011-09-22T00:58:10 hmm, username alex right? 2011-09-22T00:58:14 janzert: ya 2011-09-22T00:58:37 McLeopold: I'd rather just change the aichallengebeta.hypertriangle.com dns once we're setup there 2011-09-22T00:58:52 amstan: just sec, know the problem 2011-09-22T00:58:57 128.6 games in the last 5 minutes, wow 2011-09-22T00:59:06 McLeopold: much better compared to 20 2011-09-22T00:59:14 that can't be right 2011-09-22T00:59:21 amstan: try now 2011-09-22T00:59:29 yay 2011-09-22T00:59:46 what's with the funky hostname? 2011-09-22T00:59:50 wrong ownership of .ssh directory 2011-09-22T01:00:04 it's the ec2 internal name 2011-09-22T01:00:44 janzert: so how do i sudo? won't i need a pass for that? 2011-09-22T01:00:59 rsync'ed last games and entries over 2011-09-22T01:01:05 nope 2011-09-22T01:01:23 you shouldn't have on the other beta server either 2011-09-22T01:01:24 janzert: well, it asks for one 2011-09-22T01:01:27 at least I didn't 2011-09-22T01:01:36 ahh, oops forgot to add you to admin group 2011-09-22T01:01:51 might need to relog for it to take effect 2011-09-22T01:02:14 janzert: ok, cool 2011-09-22T01:02:22 janzert: can we get the awesome bashrc for the contest user? 2011-09-22T01:02:37 think I already got the imporant bits 2011-09-22T01:02:49 if I missed something feel free to add it 2011-09-22T01:02:50 oh yeah, good stuff 2011-09-22T01:02:53 read the problem description page carefully... 2011-09-22T01:02:58 though, coloured prompt is always nice 2011-09-22T01:03:25 McLeopold: hmm any ideas? 2011-09-22T01:03:30 mleise: you around? 2011-09-22T01:03:31 contestbot: seen CIA-24 2011-09-22T01:03:31 amstan: CIA-24 was last seen in #aichallenge 4 hours, 5 minutes, and 42 seconds ago: aichallenge: janzert epsilon * r2e7219d / worker/sandbox.py : Try harder and check better that the sandbox processes have died before leaving kill method - http://git.io/yU5AbA 2011-09-22T01:04:36 I need a pass to sudo 2011-09-22T01:04:42 McLeopold: hold on 2011-09-22T01:05:06 heh, did I forget everyone's admin group :} 2011-09-22T01:05:10 McLeopold: relogin 2011-09-22T01:06:25 hold off on the website/setup.py for a few days 2011-09-22T01:06:40 janzert: what about the sudo pass then? won't it be less secure than usual if there's no pass? 2011-09-22T01:06:48 okay, I'm signing off for the night, any last requests? 2011-09-22T01:06:49 (ignoring what i did with aichallengebeta) 2011-09-22T01:07:58 it's sort of a trade off, there's no password on the system to attack, someone has to get ahold of an ssh key instead 2011-09-22T01:08:51 *** ibdknox has joined #aichallenge 2011-09-22T01:08:54 McLeopold: workers are good to start? 2011-09-22T01:09:04 yep 2011-09-22T01:09:35 ok, anything else you can think of that I missed setting up or bringing over? 2011-09-22T01:10:09 not until it breaks :) 2011-09-22T01:10:20 so long as everything appears good tomorrow we should probably let demerzel know he can have his server back 2011-09-22T01:10:32 this server is much faster 2011-09-22T01:11:03 and as soon as I see that the workers appear to be working I think you can switch the dns over if you like amstan 2011-09-22T01:12:07 janzert: can we leave it for tomorrow? i want to go to bed soon, and i don't want to leave immediatelly after i do that 2011-09-22T01:12:29 woot, we had just passed 200000 games on demerzel's server 2011-09-22T01:16:29 *** McLeopold has quit IRC (Read error: Connection reset by peer) 2011-09-22T01:20:24 janzert: nice server, 8GB 2011-09-22T01:20:27 of rams 2011-09-22T01:22:59 yeah, ec2 has some nice specs 2011-09-22T01:50:36 *** Accoun has quit IRC (Ping timeout: 252 seconds) 2011-09-22T01:54:54 *** Accoun has joined #aichallenge 2011-09-22T02:15:11 *** chris__0076 has joined #aichallenge 2011-09-22T02:19:04 *** chris_0076 has quit IRC (Ping timeout: 276 seconds) 2011-09-22T02:20:22 *** Accoun has quit IRC (Ping timeout: 276 seconds) 2011-09-22T03:08:24 *** nux67 has quit IRC (Ping timeout: 252 seconds) 2011-09-22T03:20:34 *** ibdknox has quit IRC (Remote host closed the connection) 2011-09-22T03:24:35 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-09-22T03:32:46 *** Blkt has joined #aichallenge 2011-09-22T03:33:02 good morning everyone 2011-09-22T03:34:40 *** aerique has joined #aichallenge 2011-09-22T03:57:30 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-09-22T04:07:40 *** eashoka has joined #aichallenge 2011-09-22T04:24:32 *** AGrishchenko has joined #aichallenge 2011-09-22T04:25:10 *** antimatroid has quit IRC (Ping timeout: 276 seconds) 2011-09-22T04:45:43 *** onensora has joined #aichallenge 2011-09-22T04:47:54 *** antimatroid has joined #aichallenge 2011-09-22T05:01:58 *** berak has joined #aichallenge 2011-09-22T05:13:16 *** AGrishchenko has quit IRC (Ping timeout: 276 seconds) 2011-09-22T05:39:32 *** superflit_ has joined #aichallenge 2011-09-22T05:39:35 *** superflit has quit IRC (Ping timeout: 260 seconds) 2011-09-22T05:39:36 *** superflit_ is now known as superflit 2011-09-22T06:01:15 *** FireFly has joined #aichallenge 2011-09-22T06:25:26 *** olexs has joined #aichallenge 2011-09-22T06:37:41 *** onensora has quit IRC (Ping timeout: 260 seconds) 2011-09-22T06:55:50 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-09-22T06:57:23 <_flag> ~65 2011-09-22T06:57:46 *** sigh has joined #aichallenge 2011-09-22T07:10:03 *** olexs has quit IRC (Quit: Leaving.) 2011-09-22T07:11:31 *** fgump has joined #aichallenge 2011-09-22T07:17:32 *** fgump has quit IRC (Quit: Leaving) 2011-09-22T07:27:48 *** delt0r_ has quit IRC (Ping timeout: 260 seconds) 2011-09-22T07:35:19 *** mcstar has joined #aichallenge 2011-09-22T07:37:21 *** eashoka has quit IRC (Quit: Page closed) 2011-09-22T07:39:27 *** delt0r_ has joined #aichallenge 2011-09-22T07:49:23 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 260 seconds) 2011-09-22T08:16:18 :q 2011-09-22T08:16:22 *** mcstar has quit IRC (Quit: WeeChat 0.3.5) 2011-09-22T08:16:31 *** mcstar has joined #aichallenge 2011-09-22T08:21:57 *** Palmik has joined #aichallenge 2011-09-22T08:55:39 *** eashoka has joined #aichallenge 2011-09-22T09:04:03 *** superflit has quit IRC (Ping timeout: 260 seconds) 2011-09-22T09:32:04 *** onensora has joined #aichallenge 2011-09-22T09:38:53 *** AGrishchenko has joined #aichallenge 2011-09-22T09:46:51 *** AGrishchenko has quit IRC (Quit: Bye) 2011-09-22T10:07:46 *** xathis has joined #aichallenge 2011-09-22T10:30:14 *** JokerJo has joined #aichallenge 2011-09-22T10:37:34 *** eashoka has quit IRC (Ping timeout: 252 seconds) 2011-09-22T10:48:43 *** eashoka has joined #aichallenge 2011-09-22T10:50:40 *** aerique has quit IRC (Quit: ...) 2011-09-22T11:12:09 *** FireFly has quit IRC (Quit: FireFly) 2011-09-22T11:14:32 *** McLeopold has joined #aichallenge 2011-09-22T11:20:57 *** McLeopold1 has joined #aichallenge 2011-09-22T11:23:27 *** McLeopold has quit IRC (Ping timeout: 248 seconds) 2011-09-22T11:27:59 *** JokerJo has quit IRC (Quit: Page closed) 2011-09-22T11:29:07 *** sigh has quit IRC (Remote host closed the connection) 2011-09-22T11:31:07 *** McLeopold has joined #aichallenge 2011-09-22T11:34:07 *** McLeopold1 has quit IRC (Ping timeout: 248 seconds) 2011-09-22T11:50:22 *** McLeopold has left #aichallenge 2011-09-22T11:54:02 *** Hammerok has joined #aichallenge 2011-09-22T11:58:57 *** amstan has joined #aichallenge 2011-09-22T11:58:57 *** ChanServ sets mode: +o amstan 2011-09-22T11:58:59 i missed it by 9 2011-09-22T11:59:11 we didnt celebrate the 200000th game 2011-09-22T12:00:43 *** Keth_ has joined #aichallenge 2011-09-22T12:00:44 the winner is svenhelge from sweden 2011-09-22T12:00:44 *** Keth has quit IRC (Read error: Connection reset by peer) 2011-09-22T12:01:53 good evening everyone 2011-09-22T12:02:16 *** dvladim has joined #aichallenge 2011-09-22T12:02:22 *** Blkt has quit IRC (Remote host closed the connection) 2011-09-22T12:09:18 contestbot: later tell mleise let me know what account name you would like on the new server and if I should take the ssh key from the old beta server 2011-09-22T12:09:18 janzert: Ready to serve, my lord. 2011-09-22T12:13:02 janzert: probably he would like to use unicode characters in it 2011-09-22T12:15:52 aichallenge: James Whitehead II epsilon * r7e4c737 / ants/dist/starter_bots/go/ants.go : Update Go starter package for go.r60 - http://git.io/fM-mKA 2011-09-22T12:15:52 aichallenge: James Whitehead II epsilon * rfae4a79 / ants/dist/starter_bots/go/ants.go : player_seed is a 64-bit value - http://git.io/DVY-mQ 2011-09-22T12:15:53 aichallenge: Janzert epsilon * reaac016 / ants/dist/starter_bots/go/ants.go : 2011-09-22T12:15:53 aichallenge: Merge pull request #256 from jnwhiteh/epsilon 2011-09-22T12:15:53 aichallenge: Update Go starter package for go.r60 - http://git.io/hHPPow 2011-09-22T12:16:22 thanks for the merge 2011-09-22T12:16:25 jnwhiteh: thanks for the updates 2011-09-22T12:16:29 np 2011-09-22T12:16:53 too bad I have so much other work to do, I'd enjoy spending more time working on a bot =) 2011-09-22T12:18:23 yeah, that work thing keeps getting in the way of all the good stuff ;) 2011-09-22T12:18:55 *** amstan has quit IRC (Ping timeout: 248 seconds) 2011-09-22T12:22:28 *** McLeopold1 has joined #aichallenge 2011-09-22T12:24:30 janzert: I should not have access to the real server. 2011-09-22T12:24:33 mleise: looks like your update of the D language a few days ago broke the current D bots 2011-09-22T12:24:59 mleise: ok, this isn't qute the real server yet 2011-09-22T12:24:59 janzert: on aichallengebeta ? 2011-09-22T12:25:29 janzert: ok, then on the new server I'd like my home directory to be copied :) 2011-09-22T12:25:56 I would chose mleise as the login name, but frontier is fine 2011-09-22T12:25:57 mleise: well the workers run against antsbeta.janzert.com now but I believe it just happens to be the first time the workers completely resetup since that checkin 2011-09-22T12:26:34 I didn't see your worker_setup.py change slip in with the rest of that commit :P 2011-09-22T12:27:07 I updated the D package, yes. that's what I wanted to ask 2011-09-22T12:27:26 "how can I update a programming language package on the server" 2011-09-22T12:27:31 since it's just you and one other and we're in beta I don't think it's a problem just that you need to update your bot so it doesn't have an error 2011-09-22T12:27:32 *** mceier has joined #aichallenge 2011-09-22T12:27:45 pretty much by starting new workers 2011-09-22T12:28:26 so the package doesn't have to be installed into some secret chroot environment? ok 2011-09-22T12:28:39 *** amstan has joined #aichallenge 2011-09-22T12:28:39 *** ChanServ sets mode: +o amstan 2011-09-22T12:29:07 it gets installed in the chroot environment by worker_setup.py 2011-09-22T12:29:57 the actual method it uses is after setting up the base chroot, worker_setup.py is run inside the new chroot to install the languages 2011-09-22T12:31:12 you account named mleise should be all set 2011-09-22T12:32:05 thx 2011-09-22T12:36:14 contestbot: later tell amstan Could you contact demerzel and let him know he can have his server back 2011-09-22T12:36:14 janzert: OK 2011-09-22T12:36:29 contestbot: seen demerzel 2011-09-22T12:36:29 amstan: demerzel was last seen in #aichallenge 5 weeks, 4 days, 23 hours, 29 minutes, and 49 seconds ago: workk crazy 2011-09-22T12:36:50 janzert: that means i can also forward the domain over? 2011-09-22T12:36:54 to antsbeta.janzert.com 2011-09-22T12:36:58 I was hoping you had an email address 2011-09-22T12:37:11 yes, or just point it at the ip 2011-09-22T12:37:44 107.20.156.231 2011-09-22T12:39:29 janzert: should be up 2011-09-22T12:40:13 janzert: yep, i'm getting the key error 2011-09-22T12:40:24 wow only a 2 minute TTL 2011-09-22T12:40:32 the key error? 2011-09-22T12:40:38 janzert: yeah, i like having those on my domains since i have dynamic ip 2011-09-22T12:40:47 ahh, from ssh because the host changed 2011-09-22T12:40:49 key error because it's another server now 2011-09-22T12:40:50 yes 2011-09-22T12:41:55 janzert: k, so all good? 2011-09-22T12:41:57 can i go? 2011-09-22T12:42:26 seems like 2011-09-22T12:42:31 cool 2011-09-22T12:42:32 cya 2011-09-22T12:42:41 if you can just let demerzel know when you get a chance 2011-09-22T12:47:05 *** amstan has quit IRC (Ping timeout: 256 seconds) 2011-09-22T12:48:04 @seen j3camero 2011-09-22T12:48:04 McLeopold1: j3camero was last seen in #aichallenge 13 weeks, 2 days, 13 hours, 48 minutes, and 1 second ago: Text me at 416 464 3096. 2011-09-22T12:48:25 :/ 2011-09-22T12:49:41 so, now what? 2011-09-22T12:50:44 I think we go forward with the timing on final beta and just we'll have to move to the final server when Jeff comes around 2011-09-22T12:51:07 then we'll go live after at least a couple days on the final server 2011-09-22T12:51:53 I'm going to be gone until probably late this evening btw 2011-09-22T12:52:08 amstan should be able to prod jeff once we're fully ready? 2011-09-22T12:52:31 think he did 2011-09-22T12:52:35 ahhh :\ 2011-09-22T12:52:44 *** JokerJo has joined #aichallenge 2011-09-22T12:52:45 he did send an email recently i saw that 2011-09-22T12:59:13 McLeopold1: slightly worriesome, the last two games of GreenTea ranked 1st had Cellular2 ranked 105th, played 50 minutes apart so it shouldn't have been a double scheduling problem 2011-09-22T13:00:31 *** Cyndre has joined #aichallenge 2011-09-22T13:00:47 *** superflit has joined #aichallenge 2011-09-22T13:01:48 Hi all. Whats wrong with bots upload? Who knows? Beta is over? 2011-09-22T13:02:18 ah no we just moved to a different server 2011-09-22T13:02:30 and forgot to turn it back on is all :) 2011-09-22T13:02:45 janzert: if you find a second case of a bad matchup, then let me know 2011-09-22T13:03:03 JokerJo: should be back on now 2011-09-22T13:03:13 McLeopold1: yep, yep 2011-09-22T13:08:13 *** JokerJo has quit IRC (Quit: Page closed) 2011-09-22T13:09:12 janzert: GreenTea's games look okay 2011-09-22T13:09:25 the mu drops to 60 fairly quickly 2011-09-22T13:09:43 ok 2011-09-22T13:10:16 once he get's paired with someone at 65 or so, which is very likely, the match quality drops and we start pulling in someone that matches them both 2011-09-22T13:10:58 the deal is, even though he is the seed, the other players will also be put at the back of the line, so I have to match well against everyone in the game, not just the seed 2011-09-22T13:11:15 * janzert nods 2011-09-22T13:12:12 I suppose pretty unavoidably the people at either end of the rankings will get a worse average match quality 2011-09-22T13:12:17 did you get map turn limit automatted yet? 2011-09-22T13:12:37 running automatically, no 2011-09-22T13:12:52 it'll just be a cron entry though 2011-09-22T13:13:05 I'm still running it manually more than once a day right now ;) 2011-09-22T13:13:53 none of the maps have topped out yet so I'm still not sure the current metric will really work out 2011-09-22T13:14:15 are they just increasing? 2011-09-22T13:14:32 500 looked like a good limit... 2011-09-22T13:14:55 so far every chance they get (i.e every time a map has hit 100 games at the current limit) 2011-09-22T13:15:09 maybe it was more like, 500 if the game ends, otherwise it goes for ever, and it might have been 50% / 50% 2011-09-22T13:15:54 there's a few maps now with lower average game lengths but they haven't hit the 100 game mark yet 2011-09-22T13:16:15 how's my map distribution? 2011-09-22T13:16:18 I'll try and make a script to spit out histograms soon 2011-09-22T13:16:30 I haven't tried to fix it yet 2011-09-22T13:16:38 I haven't looked at it since earlier in the week 2011-09-22T13:16:53 I should probably put a time limit on it... 2011-09-22T13:17:01 even out the maps for the last 72 or something 2011-09-22T13:17:26 yeah, especially if we're going to add a new batch every 24 hours 2011-09-22T13:19:03 *** janzert has quit IRC (Read error: Connection reset by peer) 2011-09-22T13:19:55 are my maps being used yet? :P 2011-09-22T13:21:01 antimatroid: so, did you need to fix anything in the map generator? because you have a week to finalize... 2011-09-22T13:21:19 McLeopold1: how are you using it? 2011-09-22T13:21:35 are you still asking or for say a transationally symmetric map with n players? 2011-09-22T13:21:52 that was the one major thing that was horrible to use 2011-09-22T13:22:21 what? 2011-09-22T13:22:41 those maps i have that allow 5 player symmetric maps etc. 2011-09-22T13:22:53 it's hard to be told how many players you want for such a map 2011-09-22T13:22:55 where it is just tiled? 2011-09-22T13:22:59 yeah 2011-09-22T13:23:31 of course, we want everything we can get 2011-09-22T13:23:42 if you have it set up to ping it a few times with a few different number of players it'll eventually spit something out 2011-09-22T13:23:43 but, we need to start producing good maps with more than one starting ant 2011-09-22T13:23:48 but it had trouble with some numbers of players 2011-09-22T13:23:52 i think 5 and 7 from memory 2011-09-22T13:24:01 was it random walk? 2011-09-22T13:24:01 more than one starting ant? 2011-09-22T13:24:11 it doesn't do that, but shouldn't be that hard to make it do it 2011-09-22T13:24:43 it uses recursive backtracking to make a maze in a generating set of the symmetry of the maze 2011-09-22T13:24:49 idealy so that each starting spot is surrounded by a different set of enemy bots 2011-09-22T13:25:19 then randomly removes some more blocks to open the map up a bit and then randomly removes single water cells from the edge of the map to make the walls not straight lines 2011-09-22T13:25:26 I guess I need to review the map makers... 2011-09-22T13:25:56 what about the symmetric maps that we use now? we've still got one for that, right? 2011-09-22T13:26:43 i haven't taken map generators off github 2011-09-22T13:26:49 ok 2011-09-22T13:26:51 the maze generator is the mapgen.py one 2011-09-22T13:27:08 and is set up to let you use parameters etc. 2011-09-22T13:28:29 i think you'd be best off just working out what proportion of tiled/translational/tessellation maps you want with a minimum and maximum number of players rather than specific numbers of players 2011-09-22T13:28:47 that is tiled == translational == tessellation 2011-09-22T13:43:44 *** janzert has joined #aichallenge 2011-09-22T14:11:59 *** olexs has joined #aichallenge 2011-09-22T14:16:38 janzert: So I guess the D starter bot is working with the compiler upgrade? 2011-09-22T14:17:42 no idea really 2011-09-22T14:18:02 the only two active D bots seem to have been you and ReaperUnreal 2011-09-22T14:18:10 both of those stopped working 2011-09-22T14:21:15 *** dvladim has quit IRC (Read error: Operation timed out) 2011-09-22T14:21:32 I think your profile page should be showing you the error 2011-09-22T14:23:19 looks like two string method deprecations and a template error in core.time.TickDuration.to() 2011-09-22T14:23:37 the deprecations were expected 2011-09-22T14:23:55 it looks like the deprecations are telling you to use the same method instead? 2011-09-22T14:25:46 ahh, I see the method capitalization has changed 2011-09-22T14:26:07 e.g. splitLines instead of splitlines 2011-09-22T14:27:14 being that both of you have the same error I would guess that code is in the starterbot as well 2011-09-22T14:31:39 looks, like that error depends on compile options. I don't get it in a debug build 2011-09-22T14:33:53 ugh 2011-09-22T14:34:25 the wonderful world of the dmd compiler 2011-09-22T14:38:24 oh yes, and I really thought this upate fixed more than it broke. I even waited out the 2.054 release 2011-09-22T14:38:44 One of the following flags must go: -inline -release -noboundscheck 2011-09-22T14:38:52 wich one would you chose? 2011-09-22T14:39:39 -inline I suppose 2011-09-22T14:41:11 or let me try a downgrade 2011-09-22T14:41:26 McLeopold1: I don't understand why the food cutoff didn't kick in for this game http://antsbeta.janzert.com/visualizer.php?game=201463 2011-09-22T14:42:57 :S what is happening on that map? 2011-09-22T14:43:19 why arent the ants moving? 2011-09-22T14:43:41 basically starter bots I believe or at least close to it 2011-09-22T14:43:56 they all get stuck on the walls or in the case of rabidus the origin :} 2011-09-22T14:44:16 i thought rabidus' bot was more advanced 2011-09-22T14:45:25 janzert: are the two sites the same? aichallangebeta.hypertriangle and this one? 2011-09-22T14:45:46 yes, the dns was updated to point at the new server 2011-09-22T14:48:20 aichallenge: Marco Leise epsilon * r6523ef7 / (3 files in 3 dirs): fixed D builds, by downgrading to 2.054 - http://git.io/Uz090w 2011-09-22T14:49:23 janzert: You may trigger a new compile now. 2011-09-22T14:50:25 worker's are reloading now 2011-09-22T14:50:42 is the new server now on EC2? 2011-09-22T14:50:57 yes 2011-09-22T14:51:07 so we're now beta-testing the production environment 2011-09-22T14:51:18 pretty much yes 2011-09-22T14:51:23 ... oh does that reinstall the compiler? 2011-09-22T14:51:37 mleise: yep 2011-09-22T14:51:43 very cool 2011-09-22T14:51:47 nice. good to see progress, after a long time of waiting :) 2011-09-22T14:51:50 I'm not positive how to reactivate the submissions 2011-09-22T14:52:11 I think the status just has to be reset, but I'm not sure 2011-09-22T14:52:20 I guess by resetting some flag in the submissions list? 2011-09-22T14:52:33 I'll give it a try after the workers come back up 2011-09-22T14:53:04 ok, I'm off 2011-09-22T14:53:42 heh, just noticed you're user_id 1 2011-09-22T14:53:51 *** chris__0076 is now known as chris_0076 2011-09-22T14:54:02 the early bird catches the worm 2011-09-22T14:55:34 *** Akranis has joined #aichallenge 2011-09-22T14:55:47 that reminds me of Arthur Mitchell 2011-09-22T14:58:35 *** berak has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443]) 2011-09-22T14:59:51 user data will be reset when the challenge starts, I hope? what with all the multiple and test accounts and loads of old starter bots still in there 2011-09-22T15:00:12 yes 2011-09-22T15:00:42 we're planning on a complete reset and start from scratch for a 'final' beta, probably this weekend 2011-09-22T15:00:51 nice 2011-09-22T15:01:02 "release candidate"? :) 2011-09-22T15:02:57 i suggest to start from a negative version number to leave room for improvement 2011-09-22T15:10:26 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-09-22T15:11:11 *** TTE has joined #aichallenge 2011-09-22T15:15:15 *** Accoun has joined #aichallenge 2011-09-22T15:15:33 *** dvladim has joined #aichallenge 2011-09-22T15:18:39 *** TTE has quit IRC (Quit: Page closed) 2011-09-22T15:19:59 hey, what's up with the current matching algorithm? I'm #2 and out of my last 14 games only the most recent one included other top 10 bots. 2011-09-22T15:21:41 *** amstan has joined #aichallenge 2011-09-22T15:21:41 *** ChanServ sets mode: +o amstan 2011-09-22T15:24:19 *** xathis_ has joined #aichallenge 2011-09-22T15:24:59 *** xathis has quit IRC (Ping timeout: 258 seconds) 2011-09-22T15:25:36 *** ChrisH has joined #aichallenge 2011-09-22T15:26:57 *** xathis_ has quit IRC (Read error: Connection reset by peer) 2011-09-22T15:27:19 *** delt0r_ has quit IRC (Ping timeout: 252 seconds) 2011-09-22T15:27:39 *** xathis has joined #aichallenge 2011-09-22T15:30:38 xathis: your bot plays real nicely, good work 2011-09-22T15:31:50 mcstar: thanks :) 2011-09-22T15:37:18 *** dr0b3rts has joined #aichallenge 2011-09-22T15:41:15 *** delt0r_ has joined #aichallenge 2011-09-22T15:45:28 xathis: If you think there is an issue, we need to look at it 2011-09-22T15:46:42 *** dvladim has quit IRC (Quit: Konversation terminated!) 2011-09-22T15:46:59 McLeopold1: I know the game count has to be kept even, but the new matching is frustrating i think 2011-09-22T15:47:02 *** dvladim has joined #aichallenge 2011-09-22T15:47:05 janzert: what was the minimum rank diff when choosing an opponent for planet wars? 2011-09-22T15:47:08 *** JokerJo has joined #aichallenge 2011-09-22T15:47:22 minimum? 0 2011-09-22T15:48:02 maybe I should lower it so 80% are within 30? 2011-09-22T15:48:03 it was a straight pareto distribution excluding only the last 50 opponents 2011-09-22T15:48:43 I was thinking it should probably be lower yeah, but 30 might be a little too tight 2011-09-22T15:48:56 80% within 80 right now? 2011-09-22T15:49:08 within 50, but that is for 2 players 2011-09-22T15:49:25 *** Accoun has quit IRC () 2011-09-22T15:49:45 http://wwwcsif.cs.ucdavis.edu/~kramer/papers/self_sim.pdf 2011-09-22T15:49:51 I'm using equation 3 2011-09-22T15:49:54 ahh, yes and quite likely the second player may be the closest 2011-09-22T15:50:58 yeah, 80% in 30 may be fine 2011-09-22T15:51:04 I'm going to tighten up the range, just to see 2011-09-22T15:51:44 80% in 20 is probably even ok 2011-09-22T15:52:23 aichallenge: McLeopold epsilon * re850845 / sql/2_generate_matchup.sql : tighten pareto dist - http://git.io/jhVHXw 2011-09-22T15:52:23 aichallenge: McLeopold epsilon * r3840e2e / (7 files in 6 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/gMRxuQ 2011-09-22T15:52:53 looks like I used an alpha of 0.75 for planetwars putting 80% with 8 2011-09-22T15:53:01 whithin 2011-09-22T15:53:18 within even 2011-09-22T15:53:33 that's was probably a little too tight though 2011-09-22T15:53:33 *** dvladim has quit IRC (Ping timeout: 260 seconds) 2011-09-22T15:53:43 8? that is small 2011-09-22T15:54:48 I did my alpha at 0.7, to try and hit 1 in a 1000 uses the full range 2011-09-22T15:54:56 0.75 was about 3 in 1000 2011-09-22T15:54:56 I don't remember exactly how it was implemented so it might have been two tailed when ever in the middle of the rankings making it 16 2011-09-22T15:55:28 so, 80% within 16? 2011-09-22T15:56:50 xathis: your last match included GreenTea 2011-09-22T15:56:55 possibly :), I'm just looking at the spreadsheet I used to pick alpha and don't remember if the implementation would have made this a two tailed distribution or not 2011-09-22T15:56:57 the one before that had a rank 3 2011-09-22T15:56:58 30 2011-09-22T15:57:38 <_flag> Are the workers running? 2011-09-22T15:57:58 no, they didn't come back up. looking at it now 2011-09-22T15:58:02 McLeopold1: yeah but the last one was before you changed anything. 2011-09-22T15:58:31 McLeopold1: but the 13 games before the last one were without another top 10 bot 2011-09-22T15:58:31 and it included #1 2011-09-22T16:00:13 xathis: how frequently do you think is reasonable for you to have played a top 10? 2011-09-22T16:00:47 *** Accoun has joined #aichallenge 2011-09-22T16:00:56 remembering that we had a problem with planetwars where the top bots played each other too frequently and turned out not to be so good against the rest of the field in some cases 2011-09-22T16:01:33 with the old matching i think there was hardly any game without another top 10 bot and i liked that 2011-09-22T16:02:26 maybe 50% within 10? 2011-09-22T16:03:09 I don't think I'd want it any more frequent 2011-09-22T16:03:47 although it's something I could be easily convinced otherwise ;) 2011-09-22T16:06:07 he got alot around the 15-35 range 2011-09-22T16:06:19 McLeopold1: did you see my question about the food cutoff above? 2011-09-22T16:06:27 yeah, no idea 2011-09-22T16:06:41 they all looked like starter bots, so it might be easy to reproduce 2011-09-22T16:07:00 also to re-enable a submission that was disabled because it didn't compile for a game, do I just set the status back to 40? 2011-09-22T16:07:08 ok 2011-09-22T16:07:16 yes 2011-09-22T16:12:21 hmm, no idea why the workers didn't come up properly. they seem to be working fine when I ran the setup manually 2011-09-22T16:14:24 gota love the modern era... "it worked the other 100 times?" 2011-09-22T16:17:37 *** cody__ has joined #aichallenge 2011-09-22T16:19:18 there are currently two workers used in the beta. I hope the started challenge will use more / more powerful ones? It will obviously need a higher game rate than the beta. 2011-09-22T16:20:12 olexs: the workers are the exact machine we'd use for the official and finals 2011-09-22T16:20:19 the number of workers depends on money 2011-09-22T16:20:38 but time is money 2011-09-22T16:20:42 this is a loop 2011-09-22T16:20:45 how much do they cost to run? 2011-09-22T16:21:03 20 cents an hour? 2011-09-22T16:21:39 McLeopold1: where are those ideas standing regarding advertising? 2011-09-22T16:21:44 so ~300$ per worker for the 2 months 2011-09-22T16:22:01 about 2011-09-22T16:22:31 are you keeping the google in the name at all? 2011-09-22T16:22:35 mcstar: well, I've never done adwords before 2011-09-22T16:22:46 we'd need a volunteer to make it look nice 2011-09-22T16:23:04 that is j3camero decision 2011-09-22T16:23:08 google is pretty fast to claim illegal clicks and cut you off 2011-09-22T16:23:24 google in the name is what brought most people to planetwars, no doubt 2011-09-22T16:23:26 I didn't say click fraud, geez 2011-09-22T16:23:39 yeah, it brought me :) 2011-09-22T16:23:48 neither did i 2011-09-22T16:23:52 that is the point 2011-09-22T16:24:08 This has happened to a lot of people i know 2011-09-22T16:24:14 you can't depend on it 2011-09-22T16:24:19 hmm... 2011-09-22T16:24:21 delt0r: ads in blogs? 2011-09-22T16:24:25 or what? 2011-09-22T16:24:26 No 2011-09-22T16:24:27 j3camero: could sell his own 2011-09-22T16:24:32 forums etc 2011-09-22T16:24:43 like java-gaming.org 2011-09-22T16:24:46 300$ for a small graphic on the front page for the duration of the contest 2011-09-22T16:25:20 wouldn't a uni somewhere what to support this? I know i get more than that to run a course or two 2011-09-22T16:25:23 pay per view would mint a lot of money with all the refreshing.. 2011-09-22T16:25:24 well, when i first heard of the contest i thought, whoa, sponsored by google, that must be big/awesome 2011-09-22T16:25:44 *** detryo has joined #aichallenge 2011-09-22T16:25:47 mcstar: i think we all did 2011-09-22T16:25:57 hi all 2011-09-22T16:26:02 what exactly was the google participation, actually? :) 2011-09-22T16:26:10 i think that was it 2011-09-22T16:26:19 smell of candy 2011-09-22T16:26:29 as i understand --nothing other than permission to use the name 2011-09-22T16:26:39 yeah, that was it 2011-09-22T16:26:46 underwhelming 2011-09-22T16:27:06 McLeopold1: so what about waterloo uni? 2011-09-22T16:27:17 I've got no connections there, as amstan 2011-09-22T16:27:21 ask 2011-09-22T16:27:30 (maybe i am misremembering this, but i think i read its related to the uni) 2011-09-22T16:27:32 bit abynire euther 2011-09-22T16:27:39 not anymore either* 2011-09-22T16:27:44 it was connected to uwaterloo 2011-09-22T16:28:43 *** Palmik has quit IRC (Remote host closed the connection) 2011-09-22T16:29:20 of course my idea to pay a fee per registration was kind of a joke, i guess it would be a big deal to implement some kind of paying system 2011-09-22T16:29:30 but you could accept paypal donations 2011-09-22T16:29:38 how about putting it on the main site? 2011-09-22T16:29:47 it might be a bit late for ants, but what about kickstarter? 2011-09-22T16:29:50 a donation link wouldn't be too bad 2011-09-22T16:29:54 yes 2011-09-22T16:30:11 once you take money... you need to think of taxes and company registration 2011-09-22T16:30:17 at least in most countries 2011-09-22T16:30:29 its not-for-profit 2011-09-22T16:30:43 you need to register and qualify for that 2011-09-22T16:30:50 it is not a given 2011-09-22T16:30:50 :S 2011-09-22T16:30:57 i hate money 2011-09-22T16:31:36 It not a huge deal... but it more or less mean you need a "official" legal standing 2011-09-22T16:31:45 somewhere... it can just be a club etc 2011-09-22T16:32:10 create a stock market bot.. 2011-09-22T16:32:19 I am now in Bio--so i can't really get money for a CS comp 2011-09-22T16:32:43 if i put my paypal account up on a site and say: if you pant to donate then do, in some countries that would be illegal? 2011-09-22T16:32:52 does it depend on where the site is hosted? 2011-09-22T16:32:56 its very grey to me 2011-09-22T16:33:19 pant->want 2011-09-22T16:34:08 not illegal, just counts as income 2011-09-22T16:34:42 yep.. and then you pay income tax 2011-09-22T16:34:51 again not a biggie 2011-09-22T16:34:58 but something to consider 2011-09-22T16:35:35 also others can't claim it as a donation to a Charity or non profit or whatever 2011-09-22T16:36:47 then lets just hope jeff will be succesful, and will use his jedi skills 2011-09-22T16:51:41 cant somone just donate server time? 2011-09-22T16:51:41 *** JokerJo has quit IRC (Quit: Page closed) 2011-09-22T16:53:22 it is not just the server... but the internet connection... but yes. for a uni this should be pretty easy and very cheap since its more or less covered with overheads 2011-09-22T16:53:48 but then IT departs are toss pots 2011-09-22T16:54:10 "you can't have port 22 open. its insecure! 2011-09-22T16:56:45 how many worker serververs needed per 1000 players? 2011-09-22T16:57:35 how many workers so that the non of the 1000 players complain? 2011-09-22T16:58:11 so whats the number.. 2011-09-22T16:58:12 it's the game rate 2011-09-22T16:58:21 how often do you want a game? 2011-09-22T16:59:26 as ofen as posible.. 2011-09-22T16:59:39 well, then you need as many workers as possible 2011-09-22T16:59:55 which was my point 2011-09-22T17:00:40 technically even with few workers the final outcome just takes longer... 2011-09-22T17:01:26 but while folks are tweaking their bots and what instant feedback on how well it worked. As many as possible ends up being the answer 2011-09-22T17:03:33 current game reate is 10.1 players per minute... right 2011-09-22T17:03:40 *** Hammerok has quit IRC (Ping timeout: 252 seconds) 2011-09-22T17:04:04 btw, I'm not sure that display is very correct 2011-09-22T17:04:47 I've often waited for my next game watching that info, and it always took a lot longer than the displayed ETA. 2011-09-22T17:06:38 yeah, it is flawed 2011-09-22T17:06:42 because we don't really know 2011-09-22T17:07:22 McLeopold1: than how about, after launch, you help someone make a good tcp server? 2011-09-22T17:07:35 (im not asking you to do it, just asking about the possibility) 2011-09-22T17:07:36 geeze, I'm busy enough 2011-09-22T17:07:40 sure 2011-09-22T17:07:56 I'll host a tcp server, but I'm not going to make one 2011-09-22T17:08:25 I want to do it 2011-09-22T17:08:26 than delt0r might save the day 2011-09-22T17:08:31 lol 2011-09-22T17:08:38 I've heard a few times in this channel about there being some code for such a server already, but never any details 2011-09-22T17:08:41 but its a matter of finding the time 2011-09-22T17:09:03 and on the odd occasion... the soberness. :D 2011-09-22T17:09:39 My engine is more or less finished... that i would tie into my own server frame work 2011-09-22T17:09:56 the question is --how to interact with the http server 2011-09-22T17:10:06 via a DB 2011-09-22T17:10:12 or via something else 2011-09-22T17:10:16 im not sure than wheather we should give delt0r free beer 2011-09-22T17:10:29 oh i have free beer 2011-09-22T17:10:44 why not use the same engine the challenge uses, and just redirect the in/out streams through a tcp connection using a connection manager tool? 2011-09-22T17:10:48 I am going to be at oktoberfest on wenday too 2011-09-22T17:10:53 than lets take it away from you so you will find that elusive soberness 2011-09-22T17:10:55 wensday 2011-09-22T17:10:55 :D 2011-09-22T17:11:07 olexs: that is less fun 2011-09-22T17:11:20 delt0r: i see :) 2011-09-22T17:11:33 seriously... I wrote my own ECC stuff for my game... 2011-09-22T17:11:36 because it was fun 2011-09-22T17:11:54 but yea... still however does not change much 2011-09-22T17:12:10 its been many days... but not a lot of time. 2011-09-22T17:12:24 I just don't get to do much outside work atm 2011-09-22T17:12:25 still dont know what ecc is 2011-09-22T17:12:33 other than error correction 2011-09-22T17:12:40 ecliptic Curve Cryptography 2011-09-22T17:12:45 ah 2011-09-22T17:12:50 i remember you told me that 2011-09-22T17:13:05 did LDPC codes too 2011-09-22T17:13:15 you mean LAPD? 2011-09-22T17:13:20 haha 2011-09-22T17:14:07 that is where Bayesian learning prorogation was first applied... aka now used in true skill 2011-09-22T17:14:29 Low Density Parity Codes 2011-09-22T17:14:29 your keyboard is seriously dislexic 2011-09-22T17:14:34 that me 2011-09-22T17:14:38 that is me 2011-09-22T17:14:40 hehe 2011-09-22T17:14:55 its a german keyboard with a US layout 2011-09-22T17:15:11 made in germany? 2011-09-22T17:15:11 which is different from both a german keyboard and a US one 2011-09-22T17:15:23 probably 2011-09-22T17:15:35 in Austria ... no one makes anything for Austria 2011-09-22T17:15:40 ah, it was made during ww2, so the germans only had bad photographs of the us layout, i see 2011-09-22T17:15:51 lol 2011-09-22T17:16:30 anyway... I have a bunch of 6+hour train rides coming up next week.. i hope to do some "turn key" tcp server 2011-09-22T17:16:57 you code on the ride? 2011-09-22T17:17:23 yea 2011-09-22T17:17:29 trains here are really nice 2011-09-22T17:17:37 most of my trains are the ICE 2011-09-22T17:17:47 they have power points etc 2011-09-22T17:17:58 lol 2011-09-22T17:18:03 microsoft? 2011-09-22T17:18:07 so yea can code... or what movies or whatever the whole way 2011-09-22T17:18:09 lol 2011-09-22T17:18:20 in NZ that is what we call a power socket 2011-09-22T17:19:08 anyway there is something i have always wanted to try... with regard to the tcp server 2011-09-22T17:19:11 id call it power outlet 2011-09-22T17:19:15 is that correct? 2011-09-22T17:19:25 I understand --that is enough 2011-09-22T17:19:46 lol, i remember when we went to swtzerland 2011-09-22T17:19:54 they have a socket that no other country has 2011-09-22T17:20:04 we werent prepared 2011-09-22T17:20:23 we had us plugs, but not swiss 2011-09-22T17:20:35 the last day of conference i could charge my powerbook 2011-09-22T17:20:37 it is almost the same as the rest of Europe ... but rather than 5mm pins, they use 4mm 2011-09-22T17:20:54 they have a central 3rd pin afaik 2011-09-22T17:21:02 not needed 2011-09-22T17:21:16 well, nevertheless we couldnt plug in ours 2011-09-22T17:21:23 i dont remember exactly the shape 2011-09-22T17:21:29 Swiss plugs work in the rest of Europe 2011-09-22T17:21:34 not the other way round 2011-09-22T17:22:06 anyway, last day i discovered that the cable that run to the TV in the hotel room, had an end, that resembled an "8" 2011-09-22T17:22:26 that i could plug in the apples adapter if i pulled out the pins 2011-09-22T17:22:47 *** amstan_ has joined #aichallenge 2011-09-22T17:22:47 *** ChanServ sets mode: +o amstan_ 2011-09-22T17:23:32 yes i remember now, i tried to jam the plug in the wall-socket 2011-09-22T17:23:54 *** amstan__ has joined #aichallenge 2011-09-22T17:23:54 *** ChanServ sets mode: +o amstan__ 2011-09-22T17:24:07 anyway... the thing i was keen to try... is a "turn key app" that writes and updates a bunch of static html pages that some other webserver serves? 2011-09-22T17:24:25 whadayathink? 2011-09-22T17:24:29 *** amstan__ has quit IRC (Read error: Connection reset by peer) 2011-09-22T17:24:38 seems amstan started miosis 2011-09-22T17:24:54 *** JokerJo has joined #aichallenge 2011-09-22T17:25:43 (if youre asking me delt0r i have no idea what you talkig about) 2011-09-22T17:27:16 delt0r_: that's how my planet wars pages were generated, all static html 2011-09-22T17:28:01 *** amstan_ has quit IRC (Ping timeout: 255 seconds) 2011-09-22T17:28:21 how did you find it McLeopold1? would you do it again? 2011-09-22T17:29:46 find it? 2011-09-22T17:32:11 did it work out... would you do again --was it a good idea 2011-09-22T17:33:23 I am thinking that because html etc is white space agnostic, i can have quite big blocks of white space and update pages with random access file handles 2011-09-22T17:33:55 dose anyone know when the contest is going to be launched..... is there lots of bugs? ... lots of work leftover? 2011-09-22T17:34:48 very soon 2011-09-22T17:35:47 leftover, bugs, launch, im getting hungry 2011-09-22T17:38:04 * delt0r gets more beer 2011-09-22T17:40:29 *** ltriant has joined #aichallenge 2011-09-22T17:46:56 *** cody__ has quit IRC (Ping timeout: 252 seconds) 2011-09-22T18:01:13 delt0r_: static html was a quick a dirty solution to the system having text files for the back end data 2011-09-22T18:02:14 It feels like a poor mans caching system 2011-09-22T18:02:52 I didn't have anything to clean up old filenames, the just got orphaned by a lack of links 2011-09-22T18:04:29 I don't think treating an html file as a random access database is a good idea 2011-09-22T18:05:22 <_flag> Is it just me does the matchmaking tend to be less skill-based now? 2011-09-22T18:05:37 _flag: very much so 2011-09-22T18:05:45 <_flag> Why? 2011-09-22T18:05:53 if it is skill based, we can't keep the game rates even 2011-09-22T18:06:28 but we did just turn down the range of players available as an opponent 2011-09-22T18:07:14 we are using a pareto distribution to give us a pool of players close to the seed, so 80% of the time, your first opponent will be within 30 ranks 2011-09-22T18:07:38 that may need to be tighted up some more 2011-09-22T18:08:13 and by 30 ranks, I mean the 30 closest players in match quality, which is kinda 30 ranks, but not exactly 2011-09-22T18:09:30 <_flag> What if you try to give precedence to players who haven't played recently? The game rate won't be consistent, but maybe it will even out? 2011-09-22T18:11:51 we do, but it is the second criteria 2011-09-22T18:12:19 the first is preference to players you haven't played yet, then it is total games in last 24 hours 2011-09-22T18:12:29 *** okay|` is now known as okayzed 2011-09-22T18:12:51 so, the pool of players to choose from is limited by match quality, then the games counts are used 2011-09-22T18:13:34 <_flag> And if you limit it more top players don't play as often? 2011-09-22T18:13:57 <_flag> I think that's what I remember happening before 2011-09-22T18:14:04 well, the seed player is always whoever hasn't played in the longest time 2011-09-22T18:14:23 if I limit the pool size, then top players will player other top players more often 2011-09-22T18:15:05 <_flag> That's good right? 2011-09-22T18:15:18 mostly, we are trying to find the sweet spot 2011-09-22T18:15:32 *** JokerJo has quit IRC (Ping timeout: 252 seconds) 2011-09-22T18:16:26 *** Akranis has quit IRC (Quit: Lämnar) 2011-09-22T18:16:29 I'm checking your games 2011-09-22T18:18:01 <_flag> In the most recent one for example, the highest ranked player other than myself was 59th 2011-09-22T18:18:03 ok, you've had some crappy matchups... 2011-09-22T18:18:12 arg 2011-09-22T18:20:24 *** onensora has quit IRC (Read error: Connection reset by peer) 2011-09-22T18:26:42 aichallenge: McLeopold epsilon * rb72e121 / sql/opponent.sql : update to opponent debugging - http://git.io/xR230g 2011-09-22T18:27:51 *** mceier has quit IRC (Quit: leaving) 2011-09-22T18:29:53 *** chris_0076 has quit IRC (Ping timeout: 260 seconds) 2011-09-22T18:31:59 aichallenge: McLeopold epsilon * r341cae3 / sql/opponent.sql : update to opponent debugging - http://git.io/8QUekg 2011-09-22T18:33:44 *** chris_0076 has joined #aichallenge 2011-09-22T18:38:37 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 276 seconds) 2011-09-22T18:39:04 *** mcstar has quit IRC (Quit: WeeChat 0.3.5) 2011-09-22T18:45:54 @later tell _flag my best guess right now is trueskill is using you to decipher the true skills of the other players by giving you a good match quality with everyone cause your so awesome 2011-09-22T18:45:54 McLeopold1: Job's done. 2011-09-22T18:46:30 aichallenge: McLeopold epsilon * r2d1483b / sql/match_quality.sql : add match_quality sp - http://git.io/px4PFg 2011-09-22T18:51:21 *** chris__0076 has joined #aichallenge 2011-09-22T18:53:44 I need a math guy: antimatroid? delt0r_? 2011-09-22T18:53:52 jmcarthur: ? 2011-09-22T18:54:52 *** chris_0076 has quit IRC (Ping timeout: 276 seconds) 2011-09-22T18:58:28 *** detryo has quit IRC (Ping timeout: 255 seconds) 2011-09-22T18:59:57 *** amstan has quit IRC (Ping timeout: 256 seconds) 2011-09-22T19:11:47 *** chris___0076 has joined #aichallenge 2011-09-22T19:12:38 McLeopold1: i'm here? 2011-09-22T19:13:47 antimatroid: I need a function that returns a number in the range of [0,1) with a pareto distribution where 0 is the least common and ~1 is the most common. 2011-09-22T19:15:17 I was using b / ( rand ^ (1 / a) ) but that generates number from b and larger 2011-09-22T19:15:40 *** chris__0076 has quit IRC (Ping timeout: 276 seconds) 2011-09-22T19:16:16 McLeopold1: I'd ask Janzert? 2011-09-22T19:16:38 you can usually change the distribution of rand though 2011-09-22T19:18:14 http://docs.python.org/library/random.html that page has random.paretovariate(alpha)? 2011-09-22T19:19:35 aichallenge: McLeopold epsilon * rc1d0702 / (sql/match_quality.sql sql/opponent.sql): improvements to matchup debugging - http://git.io/SsH8aw 2011-09-22T19:22:37 in sql 2011-09-22T19:30:32 *** detryo has joined #aichallenge 2011-09-22T19:46:12 McLeopold1: isn't simply (rand ^ (1 / a) what you want? i.e if (1 / (rand ^ (1/a)) goes from one to infinity with 1 most common, then (1 / (1 / (rand ^ (1 / a)))) gives 1 to 0 and the two outside divisions just cancel out. (aside, technically to get the inclusive range with b in the original formula I think you have use b / ((1-rand) ^ (1 / a)) but it works out right for the [0, 1) range) 2011-09-22T19:46:49 doh 2011-09-22T19:47:14 yes, I think that would work 2011-09-22T19:47:25 janzert: I see more issues with the match maker 2011-09-22T19:47:44 oh :( I hope you see fixes too ;) 2011-09-22T19:47:51 The pareto is working nicely, it's just not pulling in the closest 30 for some reason 2011-09-22T19:48:03 hmm 2011-09-22T19:48:05 Some of it has to do with excluding players in a matchup 2011-09-22T19:48:19 but there are some pulled into the pool that don't make any sense 2011-09-22T19:53:54 aichallenge: McLeopold epsilon * r87172c3 / (sql/match_quality.sql sql/opponent.sql): updates to matchup sql debugging - http://git.io/OdCflQ 2011-09-22T20:07:40 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-09-22T20:13:12 *** McLeopold1 has quit IRC (Read error: Connection reset by peer) 2011-09-22T20:40:32 *** McLeopold has joined #aichallenge 2011-09-22T20:41:10 *** re_ has joined #aichallenge 2011-09-22T20:43:11 I think that p = 1 - (1 - rand) ^ (a / (a-1)) should give you the right distribution. If a = log 5 / log 4 = 1.16, then p is in [0,0.8] 20% of the time and in [0.8,1] 80% of the time. 2011-09-22T20:43:57 See http://en.wikipedia.org/wiki/Pareto_distribution#Pareto.2C_Lorenz.2C_and_Gini 2011-09-22T20:44:08 And take the inverse of L(F) 2011-09-22T20:45:33 *** ibdknox has joined #aichallenge 2011-09-22T20:47:49 *** detryo has quit IRC (Ping timeout: 255 seconds) 2011-09-22T20:48:40 *** detryo has joined #aichallenge 2011-09-22T20:53:03 *** detryo has quit IRC (Ping timeout: 248 seconds) 2011-09-22T20:56:29 *** re_ has quit IRC (Quit: Page closed) 2011-09-22T21:01:51 *** eashoka has quit IRC (Quit: Page closed) 2011-09-22T21:13:16 *** amstan has joined #aichallenge 2011-09-22T21:13:16 *** ChanServ sets mode: +o amstan 2011-09-22T21:15:00 *** olexs has quit IRC (Quit: Leaving.) 2011-09-22T22:25:26 *** ltriant has quit IRC (Read error: Connection reset by peer) 2011-09-22T22:26:20 *** ltriant has joined #aichallenge 2011-09-22T22:38:28 *** amstan has quit IRC (Ping timeout: 276 seconds) 2011-09-22T22:41:40 *** aurora has joined #aichallenge 2011-09-22T22:42:08 hey, i just downloaded the ruby starter package; how do I get started on running it? 2011-09-22T22:56:45 *** amstan has joined #aichallenge 2011-09-22T22:56:45 *** ChanServ sets mode: +o amstan 2011-09-22T23:02:00 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-09-22T23:03:11 *** McLeopold has quit IRC (Quit: Leaving.) 2011-09-22T23:21:06 *** ChrisH has quit IRC (Quit: Page closed) 2011-09-22T23:27:52 *** delt0r_ has quit IRC (Ping timeout: 276 seconds) 2011-09-22T23:36:35 *** aurora has quit IRC (Quit: Page closed) 2011-09-22T23:40:11 *** delt0r_ has joined #aichallenge 2011-09-22T23:46:40 *** dvladim has joined #aichallenge 2011-09-22T23:46:45 @later tell McLeopold p = 1 - (1 - rand) ^ (a / (a-1)) should give you the right distribution. For example, if a = log 5 / log 4 = 1.16, then p is in [0,0.8] 20% of the time and in [0.8,1) 80% of the time, and p in [0.99,1) 52.8% of the time. See http://en.wikipedia.org/wiki/Pareto_distribution#Pareto.2C_Lorenz.2C_and_Gini\ 2011-09-22T23:46:45 dr0b3rts: Error: Missing "]". You may want to quote your arguments with double quotes in order to prevent extra brackets from being evaluated as nested commands. 2011-09-22T23:47:25 @later tell McLeopold "p = 1 - (1 - rand) ^ (a / (a-1)) should give you the right distribution. For example, if a = log 5 / log 4 = 1.16, then p is in [0,0.8] 20% of the time and in [0.8,1) 80% of the time, and p in [0.99,1) 52.8% of the time. See http://en.wikipedia.org/wiki/Pareto_distribution#Pareto.2C_Lorenz.2C_and_Gini" 2011-09-22T23:47:25 dr0b3rts: Yes master! 2011-09-22T23:47:45 *** dr0b3rts has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 6.0.1/20110830202826])