2011-09-16T00:08:34 *** ChrisH has joined #aichallenge 2011-09-16T00:10:42 *** ChrisH has quit IRC (Client Quit) 2011-09-16T00:11:22 *** ChrisH has joined #aichallenge 2011-09-16T00:18:54 *** dvladim has quit IRC (Quit: Konversation terminated!) 2011-09-16T00:19:19 *** dvladim has joined #aichallenge 2011-09-16T00:19:38 delt0r: did you get a JSkills patch? 2011-09-16T00:21:43 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 260 seconds) 2011-09-16T00:25:13 *** dvladim has quit IRC (Ping timeout: 260 seconds) 2011-09-16T00:46:20 *** ChrisH has quit IRC (Quit: Page closed) 2011-09-16T00:48:37 aichallenge: McLeopold epsilon * r6807472 / website/api_game_result.php : fix missing status - http://git.io/cJ7_eA 2011-09-16T00:48:38 aichallenge: McLeopold epsilon * r452a08b / sql/2_generate_leaderboard.sql : remove leaderboard sp - http://git.io/yhwbdA 2011-09-16T00:49:08 *** Allied_Envy has quit IRC (Ping timeout: 260 seconds) 2011-09-16T01:01:33 *** Allied_Envy has joined #aichallenge 2011-09-16T01:23:40 aichallenge: McLeopold epsilon * r038f9dd / manager/jskills/TStest.txt : very difficult trueskill update - http://git.io/zsygiQ 2011-09-16T01:35:28 *** onensora has quit IRC () 2011-09-16T01:36:08 *** mleise has joined #aichallenge 2011-09-16T01:38:08 *** Allied_Envy has quit IRC (Ping timeout: 260 seconds) 2011-09-16T01:45:08 aichallenge: McLeopold epsilon * r01af5bc / website/sql.php : fix missing max_turns in sql - http://git.io/BhpBSA 2011-09-16T01:47:08 @later tell janzert 10 players seems to crash both jskills and phpskills, 9 players is fine, I think we found our limit 2011-09-16T01:47:08 McLeopold: OK 2011-09-16T01:50:17 *** Allied_Envy has joined #aichallenge 2011-09-16T01:57:28 aichallenge: McLeopold epsilon * r6476715 / (5 files in 2 dirs): update turns and game_length from results json - http://git.io/aaq6dg 2011-09-16T02:03:07 aichallenge: McLeopold epsilon * r8ab83fe / website/sql.php : fix sql - http://git.io/E30zEA 2011-09-16T02:10:10 aichallenge: McLeopold epsilon * rf72dc3a / website/api_game_result.php : fix json reference - http://git.io/FzkR6Q 2011-09-16T02:13:56 aichallenge: McLeopold epsilon * rabfc282 / website/sql.php : fix sql - http://git.io/Ts_7aQ 2011-09-16T02:19:02 aichallenge: McLeopold epsilon * r9fd5f2f / website/nice.php : added extra space to line up rankings - http://git.io/7NMk0w 2011-09-16T02:34:36 *** Palmik has joined #aichallenge 2011-09-16T02:40:56 *** onensora has joined #aichallenge 2011-09-16T02:41:54 aichallenge: McLeopold epsilon * r20e2ceb / sql/2_update_rankings.sql : fix tie skill levels - http://git.io/QfN9Wg 2011-09-16T02:44:15 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-09-16T03:00:26 *** Palmik has quit IRC (Read error: Connection reset by peer) 2011-09-16T03:01:09 *** aerique has joined #aichallenge 2011-09-16T03:01:22 *** Palmik has joined #aichallenge 2011-09-16T03:04:28 *** Allied_Envy has quit IRC (Ping timeout: 260 seconds) 2011-09-16T03:16:43 *** Allied_Envy has joined #aichallenge 2011-09-16T03:20:00 *** jako has joined #aichallenge 2011-09-16T03:20:10 *** onensora has quit IRC () 2011-09-16T03:21:42 *** ltriant has quit IRC (Quit: Get MacIrssi - http://www.sysctl.co.uk/projects/macirssi/) 2011-09-16T03:21:57 *** McLeopold has left #aichallenge 2011-09-16T03:24:53 *** Allied_Envy has quit IRC (Ping timeout: 260 seconds) 2011-09-16T03:25:35 *** sir_macelon has joined #aichallenge 2011-09-16T03:25:43 hi 2011-09-16T03:28:14 *** trololo has joined #aichallenge 2011-09-16T03:30:27 *** TTE has joined #aichallenge 2011-09-16T03:37:01 *** Allied_Envy has joined #aichallenge 2011-09-16T03:59:40 *** TTE has quit IRC (Quit: Page closed) 2011-09-16T04:01:48 *** jako has quit IRC (Ping timeout: 252 seconds) 2011-09-16T04:22:35 *** sigh has joined #aichallenge 2011-09-16T04:24:49 *** boegel has joined #aichallenge 2011-09-16T04:28:09 @later tell McLeopold sorry got busy with some work. Should do it today. 2011-09-16T04:28:09 delt0r__: Aye, aye, sir 2011-09-16T05:06:49 aichallenge: Marco Leise epsilon * r703b25c / (2 files in 2 dirs): visualizer: support Java applet in wiki sample maps - http://git.io/wJ0rGg 2011-09-16T05:21:38 *** cichociemno has joined #aichallenge 2011-09-16T05:30:23 *** mcstar has joined #aichallenge 2011-09-16T05:40:47 *** berak has joined #aichallenge 2011-09-16T05:41:53 current JSkills seems to be quite broken... ie some missing files --so this is slowing me down... 2011-09-16T05:42:48 *** kaemo0 has quit IRC (Ping timeout: 260 seconds) 2011-09-16T05:47:30 no--its eclipse that can't import from ant properly... dam tools, can't live without em, always curse em 2011-09-16T05:51:07 aichallenge: Marco Leise epsilon * r07992b4 / (2 files): fixed: fog of war button graphics where generated when only a map was displayed / moved the dark blue player color farther to the back of the list - http://git.io/hO_I4Q 2011-09-16T05:54:57 aichallenge: Marco Leise epsilon * r04bf61c / ants/visualizer/js/CanvasElement.js : 2011-09-16T05:54:57 aichallenge: fixed: the mouse over effect (circles) on the map could in some occassions stay 2011-09-16T05:54:57 aichallenge: even when the mouse moved out of the map / regression: the A/Z button has no 2011-09-16T05:54:57 aichallenge: visible effect if the visualizer isn't playing back - http://git.io/aMZwHg 2011-09-16T05:57:19 delt0r__: I wonder if I should add the Eclipse project for the visualizer to github. It's always a mess to set it up. 2011-09-16T05:57:39 mleise: I default to a work Ant file 2011-09-16T05:58:03 then its build tool "agnostic" at least as much as java is concerned 2011-09-16T05:58:05 a work(ing) build.xml? 2011-09-16T05:58:10 yea 2011-09-16T05:58:33 all the main ide should import ok 2011-09-16T05:58:41 I don't know why its not working for me 2011-09-16T05:58:47 Yes, is included. But when you want to edit stuff and have the correct imports and so on, it could be handy 2011-09-16T05:58:57 seems they use the type varable TValue rather than T 2011-09-16T05:59:31 well eclipse is suppose to import a ant file correctly to do all that 2011-09-16T05:59:45 ha ... haha 2011-09-16T05:59:57 but yea... eclipse folks will like it... i would just also have a ant build file 2011-09-16T06:00:15 eclipse exports them from .project files btw 2011-09-16T06:00:21 again ymmv 2011-09-16T06:00:38 I download Rhino from the web so that's difficult for Eclipse to figure out ;) 2011-09-16T06:00:42 NetBeans plays nicer with ant files --so i am told 2011-09-16T06:00:57 please say you don't use Maven... 2011-09-16T06:01:02 ;) 2011-09-16T06:01:05 not a fan 2011-09-16T06:01:16 I introduced Maven at work against protests of my boss 2011-09-16T06:01:25 hate the dam thing 2011-09-16T06:01:33 He refused to see the benefits ^^ 2011-09-16T06:01:53 almost as much as i hate maven trolls that go around telling everyone else they must use maven... cus they do 2011-09-16T06:01:55 ;) 2011-09-16T06:02:11 But the day we upgraded hibernate he understood 2011-09-16T06:02:16 if its so wonderfull... We don't need to all support it! 2011-09-16T06:02:34 aka the devs of packages etc 2011-09-16T06:03:06 its not version management when i have to do the work to "support" maven 2011-09-16T06:04:59 It has contious integration and other goodies as extensions. It's not only about version management. 2011-09-16T06:06:48 I found it to work well when someone checks out the project to a virgin environment as well and it would resolve and download all dependencies in Eclipse. Of course there are also things like the very verbose XML or unresolvable conflicts that turned me down a bit. 2011-09-16T06:07:50 yea-- then why do all these "maven is awesome" fans troll ever man and his dog to "support" maven... I will not... deal with it... if it as good as you claim, i don't need to support it! 2011-09-16T06:08:21 it more of the hack to dependency hell that is modern development... and imo is not the cure 2011-09-16T06:08:52 So if that is ok.. then i am ok with maven folk... but they just don't let sleeping dogs lie 2011-09-16T06:09:09 you can use... just don't try to make me use it 2011-09-16T06:20:40 use it! 2011-09-16T06:21:34 you cannot stop the future, 2011-09-16T06:21:59 i don't even have it installed here :) 2011-09-16T06:22:07 bloody maven trolls... :D 2011-09-16T06:23:04 btw is there somewhere i can push a public git whatever... so McLeopold can just pull from it? without the setup pain of github 2011-09-16T06:23:38 my uni give me a web page only... 2011-09-16T06:24:06 and my git stuff is local --mmm 2011-09-16T06:33:01 *** mceier has joined #aichallenge 2011-09-16T06:47:37 mleise: why you don't like yellow color? 2011-09-16T06:55:57 *** delt0r_ has joined #aichallenge 2011-09-16T06:58:01 git-daemon is refusing to work :( 2011-09-16T06:58:48 *** delt0r__ has quit IRC (Ping timeout: 252 seconds) 2011-09-16T07:05:03 delt0r_: what is it with you and your underscores? 2011-09-16T07:05:32 my ISP likes to change my IP address every few hours 2011-09-16T07:05:46 then some chans here don't loose my old nick 2011-09-16T07:06:02 so to stop it from barfing i need a long list of names... 2011-09-16T07:06:16 also i am connected at work too 2011-09-16T07:06:38 aichallenge: Marco Leise epsilon * rd0ba43f / (3 files in 2 dirs): Removed html_visualizer.zip from the deploy task in the ant script. It has been deprecated by the tools.zip package. - http://git.io/6pr2IQ 2011-09-16T07:06:40 every few hours, thats insane 2011-09-16T07:07:09 sir_macelon: because food is already yellow-ish 2011-09-16T07:07:30 i couldnt work had they gave me a new address every now and then 2011-09-16T07:07:41 what would happen to my ssh sshfs connections? 2011-09-16T07:08:24 Thor loved the yellow squares, didnt he? 2011-09-16T07:08:26 mcstar: You need a persistent connection to prevent disconnects and new adresses now and then 2011-09-16T07:08:48 dyndns 2011-09-16T07:08:51 ftw 2011-09-16T07:08:56 i used dyndns 2011-09-16T07:08:59 updates inside 30sec 2011-09-16T07:09:01 That said, my ISP hasn't changed my IPv4 for ages 2011-09-16T07:09:04 but thats different 2011-09-16T07:09:17 i work from home 2011-09-16T07:09:36 if the isp gives me a new dhcp address all my connections become stale 2011-09-16T07:09:43 yes I use dyndns as well. with an earlier ISP the router allowed me to schedule the disconnects for some time late at night 2011-09-16T07:09:47 fortunately this doesnt happen 2011-09-16T07:10:26 i connect to the machines of the institution, which has static ip's of course 2011-09-16T07:10:27 but yeah, fortunately we now get the same ip every time we connect to the internet :-) 2011-09-16T07:11:25 mcstar: how is the adsl ratio in hungary? we have 1:10 up:down here :p 2011-09-16T07:12:28 dunno, i have 1500/256 2011-09-16T07:12:36 mleise: but the water is blue and we have at least two other shades of blue and food is circles 2011-09-16T07:12:37 thats kbit/s 2011-09-16T07:13:25 I have 6Mbit down and 512kbit up 2011-09-16T07:13:50 a bit better 2011-09-16T07:13:59 i can get fiber for 5EU extra that puts it at 10Mbit both ways 2011-09-16T07:14:08 hey, i can use remote desktop with freenx 2011-09-16T07:14:09 but meh 2011-09-16T07:14:18 then i need to sign up for longer 2011-09-16T07:14:59 we have a local monopoly here, they keep the prices high, maybe when i get a job ill upgrade the connection 2011-09-16T07:15:51 dammit... git-daemon is fine... its the stupid fire wall 2011-09-16T07:16:19 mcstar: we had a Telekom monopoly here, too. But they were forced to open up the market. :D 2011-09-16T07:17:01 sir_macelon: hmm, there is some shade of orange that almost looks like yellow ! 2011-09-16T07:18:14 mleise: I would rather see more bright colors since those dark ones are harder to keep track off with a dark background, just a suggestion 2011-09-16T07:18:41 mleise: the market is really open, but the lines are theirs, and other providers would have to rent the lines form them, so they dont come here providing cheap service because they cant make profit 2011-09-16T07:19:27 yes, that's why I moved the dark blue already. I'll look over the colors once more. It seems I don't need to care about colors 10-26, because the ranking alorithm crashes with more than 9 players 2011-09-16T07:19:36 mleise: you could go on and try with some bright yellow instead of this dark blue and see how it looks 2011-09-16T07:22:29 I used a scripted web page that displays all the colors on blue and brown background, to find good colors, but it didn't seem to work too well in praxis 2011-09-16T07:22:30 *** delt0r_ has quit IRC (Ping timeout: 260 seconds) 2011-09-16T07:29:50 *** delt0r_ has joined #aichallenge 2011-09-16T07:30:21 @later tell McLeopold you can fetch/pull from git://delt0r.boldlygoingnowhere.org/home/git/aichallenge/ it is not a fast connection 2011-09-16T07:30:21 delt0r_: OK 2011-09-16T07:30:32 @later tell McLeopold I added a updated jar rather than replace the old 2011-09-16T07:30:32 delt0r_: Aye, aye, sir 2011-09-16T07:31:09 oh and the patch is included 2011-09-16T07:45:46 oh.. sorry its like 2 commits instead of one. I never typically bother fixing that sort of thing...just let git deal with it 2011-09-16T07:54:17 *** foRei has joined #aichallenge 2011-09-16T08:04:09 *** GeorgeSebastian has joined #aichallenge 2011-09-16T08:40:50 *** sir_macelon has quit IRC (Ping timeout: 252 seconds) 2011-09-16T08:48:28 *** mcstar has quit IRC (Quit: WeeChat 0.3.5) 2011-09-16T08:49:16 *** trololo has quit IRC (Ping timeout: 252 seconds) 2011-09-16T08:52:43 *** Allied_Envy has quit IRC (Ping timeout: 260 seconds) 2011-09-16T08:55:52 *** dr_terrible has quit IRC (Ping timeout: 260 seconds) 2011-09-16T08:57:02 *** dr_terrible has joined #aichallenge 2011-09-16T09:04:36 *** Allied_Envy has joined #aichallenge 2011-09-16T09:24:08 *** smiley1983 has quit IRC (Ping timeout: 260 seconds) 2011-09-16T09:27:23 *** mathis has joined #aichallenge 2011-09-16T09:38:37 *** eashoka has joined #aichallenge 2011-09-16T09:50:47 *** smiley1983 has joined #aichallenge 2011-09-16T10:07:06 contestbot: later tell McLeopold worker 198 is currently stuck trying to turn in the result of an 8 player game; I have a feeling it's just that more players are more likely to trigger the bug and there isn't a cutoff that is actually safe 2011-09-16T10:07:06 janzert: Ready to serve, my lord. 2011-09-16T10:09:04 anybody has an insight on the logic behind why this game was stopped at turn 255 ? looks bit odd to me. http://aichallengebeta.hypertriangle.com/visualizer.php?game=180379&user=264 2011-09-16T10:10:18 and worker 200 is stuck on a 7 player game 2011-09-16T10:16:49 eashoka: it looks like the max turn time was set to 255 for that game 2011-09-16T10:20:20 janzert : ah ok. hopefully, the turn limit will be determined dynamically in the finals so odd results like that one can be avoided. 2011-09-16T10:20:34 yes, that's the plan 2011-09-16T10:22:11 well dynamically as in a per map basis, there will still be a hard turn limit for a game but the idea is that 95% (or whatever is chosen) of games will finish naturally within that limit 2011-09-16T10:23:24 janzert : yea, understood. A totally fool proof solution may not be feasible there, we need to save cpu time as well. 2011-09-16T10:23:59 right, there are always going to be some games that just aren't going to end within reasonable lengths :( 2011-09-16T10:25:49 *** boegel has quit IRC (Ping timeout: 240 seconds) 2011-09-16T10:26:31 janzert :exactly. hopefully that will also encourage bots to do more fighting and resolve their differences among themselves, without taking chances by waiting till the cut off time to intervene. 2011-09-16T10:26:53 BTW, got a feature request. what do you think of this : 2011-09-16T10:28:27 displaying (in profile pages) the next game pairing and the map before it start, along with count down. that will add so much into suspense! 2011-09-16T10:29:04 the pairings are basically generated on demand now and aren't really known ahead of time 2011-09-16T10:29:06 *** aerique has quit IRC (Quit: ...) 2011-09-16T10:29:25 I suppose it could show what is in progress though 2011-09-16T10:30:20 but that won't last terribly long so probably doesn't add much 2011-09-16T10:30:40 what about breaking a game into few sessions ? 2011-09-16T10:31:13 you mean play X turns then come back later and play another X turns? 2011-09-16T10:31:26 that would add a lot of complications 2011-09-16T10:32:42 rather than ppl waiting full time interval to see the total game, they get to see(say) the first half of it in wuick time. made to wait is supspense for 10 more minutes and then will be shown the second half. :-) . 2011-09-16T10:33:32 seems like a fair bit of work for something that overall people are not going to like 2011-09-16T10:33:33 well, if it is really difficult for the worker to run few games in parallel (in chunks) then we can always fake it :-) 2011-09-16T10:33:57 I don't think people will actually appreciate the added suspense ;) 2011-09-16T10:34:14 oh you think ppl will not like it ? trust me, they will like :-) it's like a regular basketball or football match 2011-09-16T10:34:53 ppl are used to have 2 halves of a match separated by a little break. :-) 2011-09-16T10:34:55 not when they realize we are artificially withholding parts of the game to add it 2011-09-16T10:35:19 of course only if it doesn't add lot of additional last minute work for you guys. 2011-09-16T10:35:31 ok, take this scenario : 2011-09-16T10:35:54 I have to wait for an hour till my next game. 2011-09-16T10:36:59 so the worker execute first half of my game, save the state and moving on to the second half. letting me analyse and predict outcomes for the remaining 30 minutes 2011-09-16T10:37:32 which I will enjoy compared to being in total dark for an whole hour. think of the last few rounds in the final! 2011-09-16T10:38:11 but think of this counter scenario: you upload your latest submission, it plays its first game after 10 minutes and crashes after turn 857, but you only get to see the game in 100 turn increments released every 5 minutes. Are you going to be really mad when you wait an extra 40 minutes to find out that your current submission has a bug? 2011-09-16T10:39:03 I think the latter (mostly in less extreme variants) is going to be more common than the former 2011-09-16T10:40:30 janzert : I mean having only 2 halves of the game. 2011-09-16T10:41:15 eashoka: not all people are funs of football;-) Most people would prefer to see entire game or nothing. 2011-09-16T10:41:27 ok, it makes it less extreme but still has the same problem 2011-09-16T10:42:18 you can go ahead and add an issue on the bug tracker requesting it if you want, but because of the work involved making it happen and debatable advantages I doubt it will happen 2011-09-16T10:43:06 of course, that's why I ran it through you first. no point loggin it if it doesn't pass the sanity check first. :-) 2011-09-16T10:43:34 but what I really meant was not to keep playesr waiting for extra time. 2011-09-16T10:44:22 janzert: Some easy way to notifying people about latest games would be great. I mean something like RSS or status.php?player=1234 with text response just "2011:09:14 11:15". 2011-09-16T10:44:35 breaking the actual running of the game into sections would be really hard 2011-09-16T10:44:43 instead of making them wait a full hour to see the total thing. show the first half in 30 minutes, and then the second half in another 30. that way they will at least have something to do in the second 30 minute period before results. at least that was the idea. 2011-09-16T10:45:13 jazert : ok, in that case, it wouldn't probably worth it. 2011-09-16T10:45:18 *** sigh has quit IRC (Remote host closed the connection) 2011-09-16T10:46:25 http://nbickford.wordpress.com/ 2011-09-16T10:46:31 that's a blog for a 13 year old, wtf? 2011-09-16T10:46:44 cichociemno : you have a point. by giving up alerts, we may also reduce the amount of page refreshes from the users, that might do good for the server 2011-09-16T10:47:00 cichociemno: an rss feed per player would be rather nice and shouldn't be too difficult, it'd be a great project for someone not deeply involved already to jump in and add ;) 2011-09-16T10:48:56 antimatroid : may be for 13 year olds like Ender Wiggins and Bean :-) 2011-09-16T10:51:00 *** cichociemno has quit IRC (Ping timeout: 252 seconds) 2011-09-16T11:10:00 *** Akranis has joined #aichallenge 2011-09-16T11:23:06 *** FireFly has joined #aichallenge 2011-09-16T11:40:10 *** mathis has quit IRC (Disconnected by services) 2011-09-16T11:46:41 *** mathis_ has joined #aichallenge 2011-09-16T11:49:32 *** GeorgeSebastian has quit IRC (Read error: Connection reset by peer) 2011-09-16T11:55:44 *** dvladim has joined #aichallenge 2011-09-16T12:04:02 *** kaemo0 has joined #aichallenge 2011-09-16T12:17:57 *** mcstar has joined #aichallenge 2011-09-16T12:23:04 aichallenge: Marco Leise epsilon * r5496317 / (16 files in 5 dirs): changed constructor of Visualizer; it now takes a ready to use Options object instead of one parameter per option; this makes it easier to extend the available options - http://git.io/udhtYw 2011-09-16T12:41:00 *** eashoka has quit IRC (Ping timeout: 252 seconds) 2011-09-16T12:43:22 *** McLeopold has joined #aichallenge 2011-09-16T12:47:05 janzert: did the spot price jump, or are workers stuck? 2011-09-16T12:48:17 McLeopold: do you know a mechanism to pass multiple arguments to an interpreter on a shebang line? 2011-09-16T12:51:52 in linux? 2011-09-16T12:51:59 $@ or something 2011-09-16T12:52:55 actually, that's the answer to a different question 2011-09-16T12:53:40 yes 2011-09-16T12:53:50 #! 2011-09-16T12:54:10 i dont know if you faced this problem but i just did 2011-09-16T12:54:21 if you give a program on the shebang line with full path 2011-09-16T12:54:30 all the parameters afterwards arent plit 2011-09-16T12:54:32 split* 2011-09-16T12:54:58 that means, the interpreter or program gets all the arguments in argv[1] as 1 string 2011-09-16T12:56:01 this is very lame :S 2011-09-16T12:58:21 a shebang line should just work with arguments, what is your line? 2011-09-16T12:59:18 McLeopold: no it shouldnt, apparently not on linux 2011-09-16T12:59:27 there are some unixes where its ok 2011-09-16T12:59:54 well, not with /usr/bin/env 2011-09-16T13:00:01 try without that 2011-09-16T13:00:18 #!/usr/bin/ipython --qui=qt -i 2011-09-16T13:00:25 this doesnt work for example 2011-09-16T13:00:43 --gui 2011-09-16T13:01:21 ipython complains that it doesnt recognize the "--qui=qt -i" argument 2011-09-16T13:03:09 can those options be set within the program? 2011-09-16T13:04:43 no 2011-09-16T13:05:00 -i is for interactive, after the script is run i want to use the interpreter 2011-09-16T13:05:09 and the gui controls the thread management 2011-09-16T13:05:18 without it i cant use mlab/mayavi 2011-09-16T13:05:33 the obvious solution is a wrapper bash script 2011-09-16T13:05:49 but still, its awkward, to have 2 scripts 2011-09-16T13:12:03 shebang is only technically allowed one argument, some oses allow multiple though, I'm not sure which way linux comes down 2011-09-16T13:12:34 mcstar: ipython has profiles, you might be able to use that 2011-09-16T13:12:42 and the worker are stuck 2011-09-16T13:12:44 linux is one 2011-09-16T13:12:53 yeah, I noticed 2011-09-16T13:13:13 so, do we have the details of the game they are trying to post? 2011-09-16T13:13:22 one is stuck on an 8 player game and the other a 7 player 2011-09-16T13:13:26 list of user_ids and rank? 2011-09-16T13:13:40 just sec, I'm still logged in to both 2011-09-16T13:14:22 so, posting a game results should just copy the json to the main server and report back 2011-09-16T13:14:40 some other process should be processing the trueskill and updating the database 2011-09-16T13:15:04 which means we should probably go back to using the manager and have it watch a directory for json files 2011-09-16T13:15:09 http://pastebin.com/9KJcedSL 2011-09-16T13:16:24 delt0r's patch was changing a single number from 100 to 5000 2011-09-16T13:16:28 hmm, yeah. so don't have the script that accepts the result do any processing 2011-09-16T13:16:38 right 2011-09-16T13:16:39 I already changed that from 100 to 1000 2011-09-16T13:16:40 more robust 2011-09-16T13:16:47 on the version that is running on the server 2011-09-16T13:17:18 bumping it to 5000 might make it a bit rarer that it's a problem but isn't going to fix it I don't think 2011-09-16T13:17:32 at least I'm guessing it's the same change 2011-09-16T13:17:52 yeah, it probably is 2011-09-16T13:18:16 I never bothered checking it in since it wasn't really a fix 2011-09-16T13:18:27 your pastebin didn't give me ranks 2011-09-16T13:18:52 that's the json from the server, the worker doesn't get ranks and such 2011-09-16T13:18:57 at least so far as I know 2011-09-16T13:20:06 well, it does run the game, so it wouldn't be in the incoming json, but the outgoing json 2011-09-16T13:20:24 ahh, that never gets logged 2011-09-16T13:20:46 all that is logged is the hash 2011-09-16T13:21:25 you need --debug to get it, I think 2011-09-16T13:21:40 yeah, so if there are too many tie ranks, php takes too long 2011-09-16T13:21:57 even at 8 players 2011-09-16T13:22:09 even at 7 :} 2011-09-16T13:22:33 even at 6... 2011-09-16T13:26:31 *** dvladim has quit IRC (Ping timeout: 276 seconds) 2011-09-16T13:27:48 hmm 2011-09-16T13:28:27 if I change the draw probability to 10% the example that you checked in that crashes with a 1% draw probability works fine 2011-09-16T13:28:40 maybe we should just up the draw probability? 2011-09-16T13:28:49 I think 1% is too low anyway 2011-09-16T13:29:15 it will be too low next week 2011-09-16T13:29:29 it means sigma drops faster 2011-09-16T13:29:57 what about 5%? 2011-09-16T13:30:36 100 iteration limit and 5% still works fine with the given example 2011-09-16T13:30:49 *** dvladim has joined #aichallenge 2011-09-16T13:30:54 I'm going to try with php now... 2011-09-16T13:31:27 It would be better to have the processing separate, but I really don't want to delay anymore 2011-09-16T13:32:06 even 1.5% works, 1.1% hits the 100 iterations 2011-09-16T13:33:08 with the php implementation, it is easy to put the variables in the server_info.php 2011-09-16T13:33:20 * janzert nods 2011-09-16T13:33:22 so we can adjust without compiling TSUpdate 2011-09-16T13:34:42 did you adjust the beta server already? 2011-09-16T13:37:19 no 2011-09-16T13:38:36 want to me to manually adjust it now just to see if the currently stuck games pass 2011-09-16T13:40:19 sure 2011-09-16T13:40:38 set it at 5% for now 2011-09-16T13:40:39 ok, set to 5% 2011-09-16T13:41:02 so, by stuck, does that mean the worker is downloading and playing the same matchup over and over? 2011-09-16T13:41:19 no, it's trying to post the game result over and over 2011-09-16T13:41:28 okay, that's not so bad 2011-09-16T13:41:36 yay, it worked for the first one 2011-09-16T13:42:38 the other won't retry for another couple minutes 2011-09-16T13:42:55 did I put an increasing delay in? 2011-09-16T13:43:32 yes with a maximum of 5 minutes if I remember right 2011-09-16T13:43:56 I created an sql file to reset the rankings, can you run that now that we have a different tau? 2011-09-16T13:44:18 when did you change tau? ;) 2011-09-16T13:44:23 you just did 2011-09-16T13:44:30 *** dvladim has quit IRC (Ping timeout: 260 seconds) 2011-09-16T13:44:40 nope, just draw probability 2011-09-16T13:44:43 oh, no, wrong value 2011-09-16T13:44:47 that's separate 2011-09-16T13:44:49 *** smiley1983 has quit IRC (Quit: leaving) 2011-09-16T13:44:54 still, can you reset? 2011-09-16T13:44:58 but yeah we could reset anyway 2011-09-16T13:45:27 which sql file is it? 2011-09-16T13:46:51 both stuck games posted now, btw 2011-09-16T13:47:25 yeah 2011-09-16T13:47:34 reset_submission or reset_ranking 2011-09-16T13:47:51 the one where I set all the values to 50.0 and 16.6667 and null 2011-09-16T13:48:10 where would you have put it? I'm not seeing anything like that 2011-09-16T13:48:12 even a draw prob of 1.75 makes a world of difference in the speed 2011-09-16T13:48:25 ahh, /home/contest 2011-09-16T13:48:28 or that 2011-09-16T13:48:53 mysql -u contest aichallenge < reset_skills.sql right? 2011-09-16T13:49:07 -D aichallenge 2011-09-16T13:49:25 maybe you don't need the -D, I just always do it 2011-09-16T13:49:41 mysqldump doesn't accept it 2011-09-16T13:49:41 yeah, I'm pretty sure it's equivalent 2011-09-16T13:49:57 ok, reset 2011-09-16T13:50:27 I can now save 4 keystrokes... 2011-09-16T13:50:39 :) 2011-09-16T13:51:14 phpskills doesn't implement the max iterations 2011-09-16T13:51:19 I'm going to check in the 5% draw probability for now 2011-09-16T13:51:32 just goes forever if it doesn't converge 2011-09-16T13:52:36 aichallenge: janzert epsilon * r143c112 / website/api_game_result.php : Change draw probability to 5% - http://git.io/TfTaMQ 2011-09-16T13:52:38 okay, last major change is a nightly process to create maps and update the max turns 2011-09-16T13:52:57 did you make it come from the server_info? 2011-09-16T13:53:10 nope 2011-09-16T13:53:15 no, just manually changed it for now 2011-09-16T13:53:24 can you do the max_turns update? 2011-09-16T13:53:56 bleh 2011-09-16T13:54:04 error: Your local changes to the following files would be overwritten by merge: 2011-09-16T13:54:05 ants/visualizer/default.properties 2011-09-16T13:54:06 or give me the logic and I'll go it 2011-09-16T13:54:11 know anything about that? 2011-09-16T13:54:14 yeah 2011-09-16T13:54:31 I changed js-min to false, because it was causing errors 2011-09-16T13:54:41 can I just remove it or does it need to stay around? 2011-09-16T13:54:42 need to tell mleise 2011-09-16T13:54:59 do a "git checkout default.properties" 2011-09-16T13:55:12 just don't run ant 2011-09-16T13:55:14 yep, that's what I meant by remove it 2011-09-16T13:55:18 ok 2011-09-16T13:55:19 :) 2011-09-16T13:55:35 maybe revert is more exact :) 2011-09-16T13:55:53 yeah, although git revert seems to be something else entirely 2011-09-16T13:55:58 :( 2011-09-16T13:55:59 @later tell mleise minified script causes errors 2011-09-16T13:55:59 McLeopold: Ready to serve. 2011-09-16T13:56:07 oh 2011-09-16T13:56:23 hmm, there were several visualizer changes that hadn't been pulled to the server 2011-09-16T13:56:27 reset maybe? 2011-09-16T13:56:40 since when? I pulled last night 2011-09-16T13:56:42 git reset seems to be something else too 2011-09-16T13:57:00 I'm guessing mleise's earlier checkins 2011-09-16T13:57:14 oh, he was busy 2011-09-16T13:57:23 took me quite a while the first time to find the 'git checkout -- foo' the first time 2011-09-16T13:57:34 what is that? 2011-09-16T13:57:40 aichallenge: Marco Leise epsilon * re311542 / (6 files in 3 dirs): added new option "decorated"; if set to false only, the map view is visible - http://git.io/H-Udeg 2011-09-16T13:57:59 hmm, and now the current visualizer widget doesn't want to work 2011-09-16T13:58:06 :D 2011-09-16T13:58:10 did you minify? 2011-09-16T13:58:10 what's up? 2011-09-16T13:58:18 hey hey 2011-09-16T13:58:26 mleise: minified script thows errors 2011-09-16T13:58:27 want to rebuild the visualizer on the server? 2011-09-16T13:58:40 I can if you want me to 2011-09-16T13:58:43 or give me the magic invocation 2011-09-16T13:58:44 yeah, I think I put an sh script in for that 2011-09-16T13:58:48 ls 2011-09-16T13:58:51 deploy.sh 2011-09-16T13:59:34 just "./deploy.sh" with no args? 2011-09-16T13:59:43 yes 2011-09-16T13:59:47 or "ant deploy -Djava.plugin=/usr/share/icedtea-web/plugin.jar -Ddeploy.path=/home/contest/aichallenge/website" 2011-09-16T13:59:49 ok, building 2011-09-16T14:00:10 compile failed 2011-09-16T14:00:36 http://pastebin.com/fUidwksN 2011-09-16T14:00:51 there is a script in my home dir: update_visualizer.sh 2011-09-16T14:00:51 ehh, I think I know 2011-09-16T14:01:09 somebody did it from root at some point 2011-09-16T14:01:19 it has to be :) 2011-09-16T14:01:21 I thought I chowned it back to contest? 2011-09-16T14:01:40 still several directories owned by root 2011-09-16T14:01:53 well, do what mleise says then 2011-09-16T14:02:05 I run the script wait a sec 2011-09-16T14:02:25 There were problems in the past, so the script had to be run as root 2011-09-16T14:02:37 I don't know what it was exactly 2011-09-16T14:02:40 wait 2011-09-16T14:02:53 too late :p 2011-09-16T14:02:56 let me try it with everything reowned to contest and see 2011-09-16T14:02:58 ahh well 2011-09-16T14:03:11 ok I ctrl+z it 2011-09-16T14:03:53 doesn't that just suspend it? 2011-09-16T14:04:26 mleise: in the dynamic visualizer, I had the '#' check as a guard to prevent manipulating
 tags without it
2011-09-16T14:05:05   so, if the pre tag doesn't start with a #, then it should remain text
2011-09-16T14:05:07   janzert, now i ctrl+c ed it
2011-09-16T14:05:36   McLeopold: ok, sorry if I broke that
2011-09-16T14:05:38   yay, build successful
2011-09-16T14:05:53   mleise: np, just need to put the main code inside the if
2011-09-16T14:06:00   and it seems to actually work
2011-09-16T14:06:03   janzert: ? wait a sec, it has to succeed with my script, too :)
2011-09-16T14:06:15   actually, we should probably trim all the lines after the split as well
2011-09-16T14:06:27   janzert: should I run it as user contest?
2011-09-16T14:06:51   in general yes when manipulating the contest files
2011-09-16T14:07:01   i.e. anything under /home/contest
2011-09-16T14:07:06   McLeopold: why trim the lines?
2011-09-16T14:08:00   Uncaught TypeError: Cannot read property 'Va' of undefined
2011-09-16T14:08:09   visualizer broken for me in chrome
2011-09-16T14:08:29   visualizer-min.js:85
2011-09-16T14:08:30   aichallenge: Marco Leise epsilon * r803886e / website/visualizer_widget.php : fix for last commit - http://git.io/6DUqsQ
2011-09-16T14:08:51   mleise: so we don't have a leading space throw us off
2011-09-16T14:09:21   Ok, add that if you like. There is now also the option to run the visualizer undecorated.
2011-09-16T14:09:30   sweet
2011-09-16T14:09:38   janzert: how do I start a shell as user contest? su contest bash ... ?
2011-09-16T14:09:48   sudo su contest
2011-09-16T14:10:07   at least that's generally the way I do it
2011-09-16T14:10:30   there are a number of other ways to accomplish it of course
2011-09-16T14:10:34   janzert: go here and check for javascript errors http://aichallengebeta.hypertriangle.com/specification.php
2011-09-16T14:11:06   "a is undefined" from visualizer-min.js line 85
2011-09-16T14:11:29   :(
2011-09-16T14:11:44   and that only shows up with the minified version?
2011-09-16T14:11:48   ugh
2011-09-16T14:13:07   on FF7 here btw
2011-09-16T14:13:09   yeah, just minified
2011-09-16T14:15:28   ok the update script works as the contest user, but maybe the problem was with the downloads, not sure. I'll look into the JS problem
2011-09-16T14:18:42   McLeopold: back to the per map turn limits
2011-09-16T14:19:01   I thought I saw some stuff go in for that in your checkins, what is implemented now?
2011-09-16T14:19:08   Looks like I broke Google's minifier
2011-09-16T14:19:22   mleise: :(
2011-09-16T14:19:33   seems like something that could be hard to track down too
2011-09-16T14:20:05   janzert: I'm tracking the max_turns and game_length for each game, the map table has a turns value that is used by the worker
2011-09-16T14:20:19   so, we just need to set the value in the map table once a night
2011-09-16T14:20:33   ok
2011-09-16T14:20:46   mleise: it only breaks on the spec page, so maybe we are just missing something
2011-09-16T14:21:36   in the game table the field turns is the max turn limit and game_length is the actual length of the game?
2011-09-16T14:22:17   we'll also need to track whether a cutoff was invoked
2011-09-16T14:23:55   janzert: yes and yes
2011-09-16T14:24:04  *** amstan  has joined #aichallenge
2011-09-16T14:24:04  *** ChanServ sets mode: +o amstan
2011-09-16T14:24:50   janzert: cutoff is in the json, I just need to put it in the table
2011-09-16T14:25:04  * janzert nods
2011-09-16T14:25:07   which would be nice to display in the game list anyway
2011-09-16T14:25:18   yeah, probably a good idea
2011-09-16T14:26:42   btw, I'd like to change the game displayed on the frontpage. Instead of just displaying the most recent game I'd like to limit it to either most recent game of the top ranked player or involving one of the top say 10 players
2011-09-16T14:26:57   so we will most likely always have a good quality game on display
2011-09-16T14:28:32   I should switch to https://github.com/mishoo/UglifyJS
2011-09-16T14:28:48   It is faster, shaves off more bytes and more reliable
2011-09-16T14:29:09   sounds like no downside ;)
2011-09-16T14:29:22   ... well it has to be installed first :)
2011-09-16T14:29:35   is that a nodejs project?
2011-09-16T14:29:40   yes
2011-09-16T14:30:01   do you minify after combining the files?
2011-09-16T14:30:44   No I don't. Google supports multiple input files.
2011-09-16T14:30:57   With UglifyJS I'd have to cat them together.
2011-09-16T14:31:55   janzert: for the front page, I'd like to display a "10 most recent games", plus a sample game.  The api_game_result can choose when to switch the game, so we can only switch if it includes a rank <10 player.
2011-09-16T14:32:55   I'll need some game awesomeness metrics...
2011-09-16T14:32:59   I was thinking of giving the '10 most recent games' list its own page
2011-09-16T14:33:24   okay, with a prominant link on the frontpage
2011-09-16T14:33:30   yes
2011-09-16T14:33:47   I'm actually kinda worried about database performance
2011-09-16T14:34:04   pretty sure the visualizer for the sample game on the front page already gets pushed down into the fold anyway
2011-09-16T14:34:24   I've gotten the database to be extremely fast for single page queries, but I haven't done any load testing, and I removed memcache stuff for the rankings
2011-09-16T14:34:28   get memcache back in for the leaderboard results
2011-09-16T14:34:38   :)
2011-09-16T14:34:42   @rankings
2011-09-16T14:34:43   amstan: An error has occurred and has been logged. Please contact this bot's administrator for more information.
2011-09-16T14:34:47   @rankings
2011-09-16T14:34:48   amstan: An error has occurred and has been logged. Please contact this bot's administrator for more information.
2011-09-16T14:34:51   ...
2011-09-16T14:34:59   our current front page doesn't reflect what it would actually look like
2011-09-16T14:35:01   McLeopold: did you change the json page?
2011-09-16T14:35:01   try again, this time it works
2011-09-16T14:35:16   amstan: oh, yeah, that was part of removing the memcache stuff...
2011-09-16T14:35:29   McLeopold: is it modified or is it removed?
2011-09-16T14:35:29   it's kinda there, but I haven't tested it
2011-09-16T14:35:30   even if you scroll the top down to pretend the beta message aren't there
2011-09-16T14:35:45   well, the header needs to be shrunk
2011-09-16T14:35:51   quite a bit
2011-09-16T14:36:06   the visualizer sits right down to the fold for me and I've got 1920x1200
2011-09-16T14:36:24   amstan: Fatal error: Call to undefined function cache_key() in /home/contest/aichallenge/website/ranking.php on line 287
2011-09-16T14:36:29   and Opera is still behaving weird. Sometimes I see no text, or a comic font or the characters change size and have a negative spacing
2011-09-16T14:36:41   amstan: I'll fix it now
2011-09-16T14:36:56   we can remove google fonts
2011-09-16T14:37:43   :(
2011-09-16T14:37:48   hmm?
2011-09-16T14:37:56   removing google fonts
2011-09-16T14:38:03   McLeopold: yeah, why remove those?
2011-09-16T14:38:36   to see if it fixes Opera?
2011-09-16T14:38:39   btw, IE 6 is down to 2.4% of ai-contest visits
2011-09-16T14:39:24   I'm only interested in Σ IE<=8
2011-09-16T14:39:36   IE < 9 is ~4%
2011-09-16T14:40:06   I think back in July it was 5 or 6%?
2011-09-16T14:40:12   ok, I hope these 4% recognize my work on the applet :p
2011-09-16T14:40:25   :)
2011-09-16T14:40:59   0.07% IE with Chrome Frame
2011-09-16T14:42:40   ok, I'm pretty much offline for today. don't update the visualizer if you need the minified version to work until I get UglifyJS working :)
2011-09-16T14:43:12   mleise: can you describe the options first?
2011-09-16T14:43:19  *** Naktibalda  has joined #aichallenge
2011-09-16T14:43:41   ok, wait a sec
2011-09-16T14:44:08  *** mathis_  has quit IRC (Quit: Page closed)
2011-09-16T14:46:50   McLeopold: http://marco-leise.homedns.org/antgui/work/visualizer/jsdoc/symbols/Visualizer.html#constructor <- look for 'options'
2011-09-16T14:47:32   basically you type "var options = new Options(); var visualizer = new Visualizer(..., options, ...);"
2011-09-16T14:48:10   follow the link to http://marco-leise.homedns.org/antgui/work/visualizer/jsdoc/symbols/Options.html for details on the settings
2011-09-16T14:50:18   McLeopold: can you make the database and game update changes to record if a cutoff occured and I'll work on the script to update the map limits
2011-09-16T14:53:41   hmm
2011-09-16T14:54:15   ChrisH Go language submission was accepted but is now not compiling on the worker
2011-09-16T14:54:52   which looks like it is causing the worker to infinite loop trying to play a game with it
2011-09-16T14:54:54   I remember someone claiming that he got no output file
2011-09-16T14:55:28   ["Output file /home/contest/compiled/tmp0J6Zha/bot/MyBot was not created."]
2011-09-16T14:55:35   right
2011-09-16T14:55:55   that was for user 'jnwhiteh' (same nick here)
2011-09-16T14:56:14   that was apparently because the Go compiler doesn't report errors on stderr like every other compiler does
2011-09-16T14:56:15   correct
2011-09-16T14:56:36   I think McLeopold checked in a fix for that
2011-09-16T14:58:12   I did
2011-09-16T14:58:34   but if it fails on one worker, it should fail on all?
2011-09-16T14:59:01   will the system eventually give up on this game or just loop forever?
2011-09-16T14:59:17   we need to fix that
2011-09-16T14:59:30   obviously it needs to give up
2011-09-16T15:00:05   I think at first I was concerned the other players would loose their seed, but if the matchup is deleted, they would pop back to the front
2011-09-16T15:00:07   that would probably be good, although the forever loop makes the problem obvious
2011-09-16T15:00:27   and the submission needs to not be accepted to begin with
2011-09-16T15:01:16   I think we need to go back to a static version of go instead of tracking the latest stable
2011-09-16T15:01:41   the problem could be that it did compile and now a new version broke it
2011-09-16T15:02:20   i.e. I think the stable version is still changing too rapidly for us
2011-09-16T15:02:43   same for D :p
2011-09-16T15:03:44   we stick to a static version of D
2011-09-16T15:03:50   currently 2.053
2011-09-16T15:04:16   If you tell people what version you're on with Go
2011-09-16T15:04:18   taht's a non-problem
2011-09-16T15:04:19   we can write for it
2011-09-16T15:04:20   so, when a new worker was brought up, it broke a submission that was already marked as good?
2011-09-16T15:04:32   the Split/SplitN problem is a bit frustrating, but we can deal with it
2011-09-16T15:04:41   jnwhiteh: except it might change after a bot has already been submitted
2011-09-16T15:04:55   worse after the finals had started
2011-09-16T15:05:07   then pick release.r60
2011-09-16T15:05:10   janzert: I need to detect a compile error while running a game, and report back to mark the submission as broken
2011-09-16T15:05:11   and just say that's what you're going to use
2011-09-16T15:05:12   McLeopold: that's my guess yes
2011-09-16T15:05:20   people can code for that easily, and change their local go to match that if they need to
2011-09-16T15:05:29   that will frustrate some people, since it worked before, but not now
2011-09-16T15:05:30   McLeopold: yeah, that's probably needed
2011-09-16T15:05:53   jnwhiteh: that's what I was refering to above
2011-09-16T15:06:02   *nod*, absolutely
2011-09-16T15:06:35   part of the problem is that getting a static version of Go isn't easy
2011-09-16T15:06:55   we were compiling it from scratch before and that takes a fair amount of time
2011-09-16T15:07:01   O.o
2011-09-16T15:07:10   go takes about 30 seconds to compile
2011-09-16T15:07:18   if you run all of the tests/benchmarks it takes much longer, obviously
2011-09-16T15:07:26   I'm happy to help in any way
2011-09-16T15:07:30   hmm, maybe that's what it was doing
2011-09-16T15:07:49   ./all.bash will do all of that ./make.bash just makes/installs
2011-09-16T15:08:42   it would be great if you could take a look in the setup/worker_setup.py script at the install_ for examples and see if you could write one to install a static version of go quickly
2011-09-16T15:08:53   sure, lemme look
2011-09-16T15:10:01   generally they use curl to download a tarball or .deb and then install it from there
2011-09-16T15:10:20   go requires mercurial, if you can't rely on that, i'll try to find you a package that works
2011-09-16T15:10:42   yeah, it was checking out from mercurial before
2011-09-16T15:11:02   janzert: we could run gofix on entries :)
2011-09-16T15:11:06   if it could just download a tarball from somewhere would be preferred
2011-09-16T15:11:22   If you install golang-60 from this ppa (which you already seem to be doing there) that would be fine: https://launchpad.net/~gophers/+archive/go/+packages
2011-09-16T15:11:43   McLeopold: yeah, I'm not sure how reliable that is for fully automatic updates though
2011-09-16T15:11:56   janzert: it covers almost any language breaking change
2011-09-16T15:11:57   not at all
2011-09-16T15:11:58   jnwhiteh: ahh, yeah that should work
2011-09-16T15:12:00   that's why gofix is around =)
2011-09-16T15:12:33   let's just call it at r60
2011-09-16T15:12:58   it'll be a month until the next release, and if you stick at r60 that's easy enough
2011-09-16T15:13:26   hmm, is that package install going to break when the ppa moves to r61 though?
2011-09-16T15:13:41   that's a question I don't know how to answer =)
2011-09-16T15:14:26   janzert: so, you still want to use 10.10?
2011-09-16T15:14:30   I have a feeling it does, since I don't see any r59 package around :(
2011-09-16T15:14:33   it's almost time for 11.10 now
2011-09-16T15:14:44   no, we're on 11.04
2011-09-16T15:14:45   just stash the .deb locally?
2011-09-16T15:15:04   you could curl it from your own servers?
2011-09-16T15:15:06   janzert: i'm just referring to the time you were hesitant to update to 11.04
2011-09-16T15:15:18   hmm, putting the .deb on the main server might be an option
2011-09-16T15:16:06   well, if you think it's ok to have all the go submissions break in the middle of the finals then... :P
2011-09-16T15:17:54   hmm, I think I might move all the ppa installed languages to simply grab debs from the main server
2011-09-16T15:19:16  *** berak  has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443])
2011-09-16T15:19:25   actually all manually installed languages could have there package file (whether .deb or tarball) placed on the main server so we aren't sucking up the language sites bandwidth (and won't be reliant on there server staying up)
2011-09-16T15:20:17  *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge
2011-09-16T15:21:27   coffeescript would be a challenge to stop relying on third party sites since it uses nodejs package manager to install
2011-09-16T15:23:39  *** delt0r_  has quit IRC (Ping timeout: 260 seconds)
2011-09-16T15:28:59  *** mcstar  has quit IRC (Quit: WeeChat 0.3.5)
2011-09-16T15:32:27  *** Accoun  has quit IRC ()
2011-09-16T15:35:57  *** delt0r_  has joined #aichallenge
2011-09-16T15:43:46  *** Accoun  has joined #aichallenge
2011-09-16T15:56:22  *** Palmik  has quit IRC (Remote host closed the connection)
2011-09-16T16:08:55   McLeopold: something is still wacky with the pairing, a few players only have one game and a few have over 30
2011-09-16T16:14:06   is it the go submissions?
2011-09-16T16:14:30   nope
2011-09-16T16:15:17   it correlates to sigma
2011-09-16T16:15:49   doesn't seem to
2011-09-16T16:15:58   well, match quality always goes up with lower sigma
2011-09-16T16:16:13   sort by games, the low count has high sigma and vice versa
2011-09-16T16:17:50   well, more games would cause more certainty as well, but I think it is still the match quality factor
2011-09-16T16:18:06   hmm
2011-09-16T16:18:56   too bad the column sort doesn't go across pages ;)
2011-09-16T16:19:21   increase page_size to 400 for a quick glance
2011-09-16T16:19:26   ranking.php
2011-09-16T16:20:57   hmm, sigma *should* correlate so strongly with number of games played it's hard to tell whether it's the cause or effect :/
2011-09-16T16:21:09   yeah, it's both
2011-09-16T16:21:31   I know it is, that's how the match quality works, sigma is the biggest factor
2011-09-16T16:22:00   we want new entries to play well established entries
2011-09-16T16:22:15   but if every new entry picks the same established entry, that is bad
2011-09-16T16:23:07   shouldn't it be trying to minimize skill difference first?
2011-09-16T16:23:13   anyway gotta run
2011-09-16T16:23:22   no
2011-09-16T16:23:28   that's not how quality works
2011-09-16T16:23:36   it does play a factor though
2011-09-16T16:48:20  *** McLeopold  has quit IRC (Quit: Page closed)
2011-09-16T16:49:22   so if we have a seed player with mu=23 sigma=3 and two possible opponents A w/ mu=20 sigma=5 and B w/ mu=30 sigma=1
2011-09-16T16:49:30   it will choose B?
2011-09-16T16:50:05   seems backwards
2011-09-16T16:50:25   guess I should go read it again
2011-09-16T16:51:16  *** Allied_Envy is now known as AlliedEnvy
2011-09-16T16:56:35  *** nux67  has quit IRC (Ping timeout: 252 seconds)
2011-09-16T17:00:46  *** Naktibalda  has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 6.0.2/20110905175240])
2011-09-16T17:02:15   @later tell McLeopold The patch was more than that... Turns a "throw new RuntimeException("")" to a "System.err.println("Warrning");break;"
2011-09-16T17:02:15   delt0r_: Job's done.
2011-09-16T17:02:44   @later tell McLeopold Seriously... did you look at it? Or you just don't want anything to do with java?
2011-09-16T17:02:44   delt0r_: As you wish.
2011-09-16T17:23:57  *** Kingpin13  has joined #aichallenge
2011-09-16T17:35:55  *** onensora  has joined #aichallenge
2011-09-16T17:36:43  *** boegel  has joined #aichallenge
2011-09-16T17:37:21  *** boegel  has quit IRC (Client Quit)
2011-09-16T17:51:03  *** onensora  has quit IRC ()
2011-09-16T17:51:08  *** Akranis  has quit IRC (Read error: Connection reset by peer)
2011-09-16T18:19:01  *** kaemo0  has quit IRC (Ping timeout: 240 seconds)
2011-09-16T18:48:57  *** McLeopold  has joined #aichallenge
2011-09-16T18:51:13   delt0r_: I really don't like php, but I'm trying to minimize complexity by not introducing more languages.  If the php works good enough, I'd like to stick with that.
2011-09-16T18:51:58   McLeopold: Java is already in the game I guess :)
2011-09-16T18:52:02   delt0r_: it turns out my low draw prob was causing most of the slowdown.
2011-09-16T18:55:11   @later tell delt0r_ so the break code would then just cause the trueskill update to not be as accurate, but still succeed?
2011-09-16T18:55:11   McLeopold: OK
2011-09-16T18:57:52   @later tell janzert it should choose A, like you were probably thinking.  I think there is another issue with my match quality code...
2011-09-16T18:57:52   McLeopold: Yes master!
2011-09-16T19:01:11  *** ChrisH  has joined #aichallenge
2011-09-16T19:06:31   aichallenge: McLeopold epsilon * rdd6fb53 / (sql/0_schema.sql website/api_game_result.php): save cutoff boolean in game table - http://git.io/LTMRoA
2011-09-16T19:06:31   aichallenge: McLeopold epsilon * rb4fc8c3 / sql/opponent.sql : testing procedure for trueskill match quality - http://git.io/MlpKoA
2011-09-16T19:08:53  *** Kingpin13  has quit IRC (Quit: quit)
2011-09-16T19:09:38  *** mceier  has quit IRC (Quit: leaving)
2011-09-16T19:29:10  *** FireFly  has quit IRC (Quit: FireFly)
2011-09-16T19:44:22  *** ChrisH  has quit IRC (Quit: Page closed)
2011-09-16T19:56:22  *** amstan  has quit IRC (Remote host closed the connection)
2011-09-16T19:57:31  *** sigh  has joined #aichallenge
2011-09-16T20:13:59   aichallenge: McLeopold epsilon * rcfba38d / website/ranking.php : fix ranking_json, use memcache better - http://git.io/jWHvuA
2011-09-16T20:15:02   @later tell amstan you can pull to get the ranking_json back up
2011-09-16T20:15:02   McLeopold: Aye, aye, sir
2011-09-16T20:15:05   @ranking
2011-09-16T20:15:06   McLeopold: User error, it's not my fault.
2011-09-16T20:15:11   @rankings
2011-09-16T20:15:12   McLeopold: An error has occurred and has been logged. Please contact this bot's administrator for more information.
2011-09-16T20:16:29   McLeopold: I don't see that you're putting a time limit on the memcached version?
2011-09-16T20:17:28   e.g. it should invalidate itself after say 5 minutes
2011-09-16T20:19:08   60 seconds
2011-09-16T20:19:15   plus everytime a game is posted
2011-09-16T20:19:18   ahh, I see you're setting an expiration time
2011-09-16T20:19:24   @rankings
2011-09-16T20:19:25   McLeopold: An error has occurred and has been logged. Please contact this bot's administrator for more information.
2011-09-16T20:19:29   doh
2011-09-16T20:24:04   @rankings
2011-09-16T20:24:05   McLeopold: An error has occurred and has been logged. Please contact this bot's administrator for more information.
2011-09-16T20:24:14  *** McLeopold  has quit IRC (Quit: Page closed)
2011-09-16T20:36:00   aichallenge: janzert epsilon * rcadeed9 / (2 files): remove a few old unused scripts - http://git.io/SFMu8w
2011-09-16T21:04:03  <_flag> Why is there a limit of 255 turns on the server?
2011-09-16T21:45:27  *** amstan  has joined #aichallenge
2011-09-16T21:45:27  *** ChanServ sets mode: +o amstan
2011-09-16T21:51:47  *** mathis  has joined #aichallenge
2011-09-16T21:56:20  *** janzert1  has joined #aichallenge
2011-09-16T21:58:19  *** ChrisH  has joined #aichallenge
2011-09-16T21:59:17  *** a1k0n_  has joined #aichallenge
2011-09-16T22:00:06  *** janzert  has quit IRC (Ping timeout: 260 seconds)
2011-09-16T22:00:06  *** Lithosphere  has quit IRC (Ping timeout: 260 seconds)
2011-09-16T22:00:06  *** a1k0n  has quit IRC (Ping timeout: 260 seconds)
2011-09-16T22:00:08  *** dr_terrible  has quit IRC (Ping timeout: 260 seconds)
2011-09-16T22:00:36  *** dr_terrible  has joined #aichallenge
2011-09-16T22:02:47  *** Lithosphere  has joined #aichallenge
2011-09-16T22:02:47  *** Lithosphere  has joined #aichallenge
2011-09-16T22:15:41  *** mathis  has quit IRC (Quit: Page closed)
2011-09-16T22:33:39   interesting.. jeff wants to remove the leader void?
2011-09-16T22:51:26  *** McLeopold  has joined #aichallenge
2011-09-16T22:53:48   McLeopold: I saw in the log that my Go bot is causing some problems. I was wondering if there have been any more thoughts about dealing with the Go versioning issue.
2011-09-16T23:05:44   ChrisH: we are going to use r60
2011-09-16T23:06:08   ok, that's what I'm using now, so that's easy.
2011-09-16T23:06:22   any idea when that will be in place on the server for testing?
2011-09-16T23:11:02  *** greghaynes  has joined #aichallenge
2011-09-16T23:12:55   it's already using r60 at the moment
2011-09-16T23:13:09   but currently would keep upgrading with new releases
2011-09-16T23:13:47  *** eashoka  has joined #aichallenge
2011-09-16T23:15:31   McLeopold : are the workers going to get started again or is this the 'silence before the storm' ? :-)
2011-09-16T23:16:10   hmm, workers are running
2011-09-16T23:16:38  *** janzert1 is now known as janzert
2011-09-16T23:19:13   eashoka: they should be working, but I will be playing with rankings over the weekend
2011-09-16T23:19:28   and the games will be shorter so I can get good stats on how the final will go
2011-09-16T23:23:04   McLeopold : ok, tx.
2011-09-16T23:24:31  *** delt0r_  has quit IRC (Ping timeout: 276 seconds)
2011-09-16T23:26:02   janzert : I have a vastly modified version of the feature request from yesterday. the one you said could be done. Can't the visualizer play the game live ? I mean if a single turn takes a second, and if there are about 1000 turns per game. that would be 15 minutes of entertainment and added suspense! can the visualizer just play turn by turn as it receives them fro the server. (via an Ajax request from JQuery maybe, instead of
2011-09-16T23:26:36   the games aren't played on the same machine as the webserver
2011-09-16T23:26:51   so it would take quite a bit of work to stream them live back
2011-09-16T23:28:22   McLeopold: something weird is going on with the workers, I can see that they are both playing games but the stats page shows no activity
2011-09-16T23:30:08   janzert : ah ok. forget that too then. :-)
2011-09-16T23:30:40   McLeopold : yea, that's what I meant. no activity in the ranking page and profile pages too
2011-09-16T23:31:45   If the servers are already using golang r60, will my bot get a chance to play some games? I would like to verify golang is going to work before I sink too much time into the bot code.
2011-09-16T23:36:12  *** delt0r_  has joined #aichallenge
2011-09-16T23:46:39   janzert: it's probably the cutoff code in the api_game_result.php
2011-09-16T23:47:05   maybe a boolean to sql tinyint issue