2011-03-01T00:03:22 "constant variables" is an http://en.wikipedia.org/wiki/Oxymoron 2011-03-01T00:03:49 i dunno 2011-03-01T00:03:56 a variable can be constant 2011-03-01T00:04:03 in maths as well 2011-03-01T00:04:19 i see what you're trying to say, but :P 2011-03-01T00:04:28 you and your definitions! you made that up, didn't you? 2011-03-01T00:04:37 antimatroid: git clone git://gist.github.com/847673.git 2011-03-01T00:04:52 antimatroid: no need to use the web interface 2011-03-01T00:05:26 but it renders all file contents in a conviniently formatted web page, amstan 2011-03-01T00:05:40 Frontier: ...to edit the files 2011-03-01T00:06:08 right, different story ^^ 2011-03-01T00:06:22 yeah meh for the moment :) 2011-03-01T00:06:34 antimatroid: you could be using this opportunity to learn version control 2011-03-01T00:06:47 So I can use these files to write a D starter bot? :p 2011-03-01T00:06:54 it's a small project, you're the only guy editing 2011-03-01T00:06:55 Frontier: go for it :P 2011-03-01T00:07:24 amstan: where do i learn about version control then? :P 2011-03-01T00:07:40 i was going to play around with perlin noise again this afternoon 2011-03-01T00:08:04 antimatroid: you could be using it for your starter package: git clone git://gist.github.com/847673.git 2011-03-01T00:08:23 amstand and I collaborated on a project last year and he made it absolutely hell by forcing me to use git from command line. I look back, it was worth it 2011-03-01T00:08:33 By working in a software company. That works best I think. Lots of conflict merging, branching stable and testing version. All the fun you can't have when you work alone. 2011-03-01T00:08:33 antimatroid: after you edit a file, you type git status to see what file changed, you type git add . to add everything, then git commit -m "your message" to commit 2011-03-01T00:08:47 i don't have one of those key things to use git 2011-03-01T00:09:03 antimatroid: you only need git, you already have a repo... any gist is a repo 2011-03-01T00:09:03 and everything on my computer is a mess, i really need to format, the command line can't even find gcc 2011-03-01T00:09:05 you need linux :o 2011-03-01T00:09:10 use putty 2011-03-01T00:09:17 it can export to openssh keys 2011-03-01T00:09:38 lol... 2011-03-01T00:10:18 frontier is right, i need to jump ship from linux 2011-03-01T00:10:21 from windows*** 2011-03-01T00:10:39 do it already? 2011-03-01T00:11:00 you know.. this is how i ended up using linux fulltime... 2011-03-01T00:11:04 ditched windows for mac 2011-03-01T00:11:13 yes, i have done full switches before 2011-03-01T00:11:16 antimatroid: suggestions: s/DEBUGGING/DEBUG 2011-03-01T00:11:25 i don't even have office installed anymore, if latex is too much notepad will do 2011-03-01T00:11:30 antimatroid: and don't use if(DEBUG), do a preprocessor if 2011-03-01T00:11:37 that way it's not a runtime check 2011-03-01T00:11:40 but a compile time 2011-03-01T00:12:01 how do i do that? oO (I'm a terrible programmer as you know) 2011-03-01T00:12:29 compile with -DDEBUG :) 2011-03-01T00:12:30 sec... 2011-03-01T00:12:34 Frontier: no 2011-03-01T00:12:37 Frontier: not yet... 2011-03-01T00:12:37 oh :/ 2011-03-01T00:12:58 antimatroid: read this: http://en.wikipedia.org/wiki/C_preprocessor 2011-03-01T00:13:17 antimatroid: might want to look at #define DEBUGPRINT2(...) fprintf(stderr, __VA_ARGS__) 2011-03-01T00:13:24 That's 19 laptop pages :O 2011-03-01T00:13:28 antimatroid: that way you can have those statments everywhere 2011-03-01T00:13:56 i don't need it anywhere else :P 2011-03-01T00:14:00 this is simple :\ 2011-03-01T00:14:19 antimatroid: this is a starter pack though 2011-03-01T00:14:28 the simpler the better 2011-03-01T00:14:29 advanced users will probably use this 2011-03-01T00:14:31 for a starter pack 2011-03-01T00:14:38 it's meant for learning 2011-03-01T00:14:44 this would qualify 2011-03-01T00:16:46 if (DEBUG) is really common and speeds up the executable a bit when debugging is turned off compared to a runtime if. 2011-03-01T00:19:51 done 2011-03-01T00:25:05 antimatroid: https://gist.github.com/848648 2011-03-01T00:26:39 amstan, this is a bit better: #define debug(...) { fprintf(stderr,__VA_ARGS__); fflush(stdout); } 2011-03-01T00:26:52 (putting multiple statements inside a block) 2011-03-01T00:26:57 oh.. ok 2011-03-01T00:27:16 that way "if(foo) debug();" still works :P 2011-03-01T00:27:19 sigh: you're right, if i put it in a if 2011-03-01T00:27:20 yeah 2011-03-01T00:27:22 fixed 2011-03-01T00:27:54 antimatroid: why is that awesome? because when you compile it with DEBUG=0, all debug print statments will not even get compiled in the program 2011-03-01T00:28:28 oh lol, and you are writing to stderr then flushing stdout :D 2011-03-01T00:28:45 sigh: yeah.. i copied this code from another project of mine.. 2011-03-01T00:28:57 fixed 2011-03-01T00:29:06 i don't get why i would want to send it to stderr? 2011-03-01T00:29:11 yeah, I know... I do that all the time too 2011-03-01T00:29:14 antimatroid: stdout then 2011-03-01T00:29:20 antimatroid: or a file.. 2011-03-01T00:29:21 that is blocked 2011-03-01T00:29:28 amstan: i already have it going to a file 2011-03-01T00:29:37 but is it compile time? 2011-03-01T00:29:44 it is now 2011-03-01T00:29:50 link? 2011-03-01T00:29:51 hang on, i'll update girhub 2011-03-01T00:30:36 amstan: https://gist.github.com/847673 2011-03-01T00:30:43 go to the bug.h/cc files 2011-03-01T00:31:36 antimatroid: i still don't like it though 2011-03-01T00:31:45 Bug can basically be treated like a filestream, it's wonderful 2011-03-01T00:32:04 antimatroid: right now.. if you do a for(int i=0;i it'll still have function calls and stuff 2011-03-01T00:32:29 if debugging is disabled.. 2011-03-01T00:32:43 if you do it the way i had it, there's no function calls if debug is 0 2011-03-01T00:32:43 wont you always? 2011-03-01T00:32:46 no 2011-03-01T00:33:25 if debug is 0, in my case, you get for(int i=0;i hmm, okay, i'll change it 2011-03-01T00:33:47 antimatroid: the preprocessor directives changes the code which gets sent to the compiler 2011-03-01T00:33:53 antimatroid: in your case.. you actually call Bug& operator<< MAXINT times 2011-03-01T00:34:06 antimatroid: which does nothing.. but you still call it 2011-03-01T00:34:10 yeah, i know my way will call that 2011-03-01T00:34:24 there might be a way to make it work your way.. 2011-03-01T00:34:28 but idk enough c++ for that 2011-03-01T00:34:31 i don't think so 2011-03-01T00:34:35 i'm a C kinda guy 2011-03-01T00:34:54 thing is that debug() is not a function, it's a macro 2011-03-01T00:35:18 maybe you could inline the Bug& operator<< call 2011-03-01T00:35:22 idk if it'll work or not 2011-03-01T00:35:23 in planet wars I made up something which is probably not very idomatic c++, so I could do stuff like: LOG( foo << bar << baz) 2011-03-01T00:35:36 https://github.com/sigh/Planet-Wars/blob/master/src/Log.h 2011-03-01T00:36:34 In D you write 'version(debug) stderr.writeln("something happened @", __FILE__, ":", __LINE__);' 2011-03-01T00:36:35 sigh: you didn't have a turn off debugging option, did you? 2011-03-01T00:36:48 Frontier: in python this is even nicer 2011-03-01T00:36:56 amstan, what do you mean? 2011-03-01T00:37:06 debugging was off by default 2011-03-01T00:37:07 Frontier: the __FILE__ stuff can be taken care for you automatically 2011-03-01T00:37:11 hehe, I think so. Python is really a nice language 2011-03-01T00:37:15 sigh: is there a way to turn off logging? 2011-03-01T00:37:24 sigh: while making everything faster? 2011-03-01T00:37:37 amstan, there is no logging is DEBUG is not defined 2011-03-01T00:37:59 sigh: won't you get a compile time error? 2011-03-01T00:38:00 line 6 starts the #ifdef 2011-03-01T00:38:12 sigh: since it doesn't know what the LOG macro does? 2011-03-01T00:38:30 oh.. nvm, you have an else 2011-03-01T00:38:32 nvm 2011-03-01T00:38:33 it's all good 2011-03-01T00:38:36 amstan, this was part of my pw submission, it compiled just fine 2011-03-01T00:39:11 antimatroid: so yeah.. look at sigh's log code: https://github.com/sigh/Planet-Wars/blob/master/src/Log.h 2011-03-01T00:39:14 it *worked*, I just doubt very much that it is idiomatic c++ or good practice in any way 2011-03-01T00:41:33 antimatroid: make yours inline, and i think it'll have the same speed 2011-03-01T00:41:41 amstan: that debug function is awful 2011-03-01T00:41:47 antimatroid: how? 2011-03-01T00:41:54 antimatroid: well.. yeah.. it's a c function 2011-03-01T00:41:55 it can't take multiple arguments 2011-03-01T00:41:59 yes it can 2011-03-01T00:42:06 or doubles, etc. 2011-03-01T00:42:08 that's what __VA_ARGS__ is for 2011-03-01T00:42:11 just treat it like printf 2011-03-01T00:42:20 it just copies the arguments you give it 2011-03-01T00:42:35 antimatroid: though.. you will need stdio.h 2011-03-01T00:43:34 it's still claiming too many arguments :\ 2011-03-01T00:44:30 amstan: " make yours inline, and i think it'll have the same speed" i don't get it? 2011-03-01T00:44:45 antimatroid: http://en.wikipedia.org/wiki/Inline_function 2011-03-01T00:45:17 ah i see 2011-03-01T00:45:56 antimatroid: how are you calling debug that causes the too many args error? 2011-03-01T00:46:12 debug("test ", "int"); 2011-03-01T00:46:22 antimatroid: no... 2011-03-01T00:46:29 oh, you need to use it like printf 2011-03-01T00:46:35 oh, rank 2011-03-01T00:46:44 antimatroid: http://en.wikipedia.org/wiki/Printf 2011-03-01T00:46:45 i'm going with inline 2011-03-01T00:48:41 i don't mind messy code for black boxes, but if i have to use something, i want it to look nice and be easy :) 2011-03-01T00:49:00 *** delt0r___ has joined #aichallenge 2011-03-01T00:49:04 antimatroid: anyone that knows C should be able to use it 2011-03-01T00:49:17 yes, but it wouldn't look nice and it'd take longer to type 2011-03-01T00:49:34 anyone that knows c++ will know how to use Bug 2011-03-01T00:50:06 produce the output: 10 has 2 prime factors. And the colour is red. 2011-03-01T00:50:37 *** delt0r_ has quit IRC (Ping timeout: 272 seconds) 2011-03-01T00:50:49 with c++ you have to do: cout << n << "has" << i << "prime factors. And the colour is " << colour << "." << endl; 2011-03-01T00:51:19 with printf("%d has %d prime factors. And the colour is %s.\n", n, i, colour); 2011-03-01T00:51:24 and have fun if you want to do any advanced formatting on those values :P 2011-03-01T00:51:25 which is longer? 2011-03-01T00:51:46 sigh: you mean hex << int << n? lol 2011-03-01T00:51:50 can you write a function to pass a state to printf? 2011-03-01T00:51:57 i can do bug << state << endl 2011-03-01T00:51:59 antimatroid: state? huh? 2011-03-01T00:52:11 amstan, yup :P 2011-03-01T00:52:14 and that will output the state grid to the file stream 2011-03-01T00:52:16 antimatroid: as in an int that's 0 or 1? 2011-03-01T00:52:28 it's be an mxn char map 2011-03-01T00:52:31 antimatroid: all you need is a function that returns a string 2011-03-01T00:52:32 it'll* 2011-03-01T00:52:41 over m lines 2011-03-01T00:53:17 antimatroid: so... you'll only use such a print function for debugging, since your bot doesn't output other strings like that 2011-03-01T00:53:37 amstan: what if i want to output the state to file? 2011-03-01T00:53:37 antimatroid: if you really want to make a function that returns such a state.. construct your function with debugs 2011-03-01T00:53:52 fprintf()? 2011-03-01T00:53:54 it's a c++ bot, why not use c++ syntax? 2011-03-01T00:54:02 i agree with that.. fine 2011-03-01T00:54:08 :) 2011-03-01T00:54:15 i will add the inline stuff though, i agree with you on that 2011-03-01T00:54:20 and that's why i don't like c++ though, lol 2011-03-01T00:54:27 :P 2011-03-01T00:56:13 i'm not sure inline functions will make this any faster 2011-03-01T00:56:38 it still has to call an empty function yeah? and i still need to return a Bug reference on << to compose stuff 2011-03-01T00:56:48 errr... inline means there is no function call 2011-03-01T00:56:49 antimatroid: if it's inline, there's no calling taking place 2011-03-01T00:57:12 (that is if c++ listens to you, it can ignore you and make it a function anyway) 2011-03-01T00:57:14 antimatroid: inline means.. copy this function in every place that it's called and expand it 2011-03-01T00:57:24 oh, err i am retarded :) 2011-03-01T00:57:44 but will i have trouble with bug << "test" << "this";? 2011-03-01T00:57:52 because wont it become << "this; 2011-03-01T00:57:53 should work, i think 2011-03-01T00:57:54 *** Zannick has quit IRC (Ping timeout: 250 seconds) 2011-03-01T00:58:18 no.. bug is the object 2011-03-01T00:58:36 *** Zannick has joined #aichallenge 2011-03-01T00:58:43 * antimatroid goes to try 2011-03-01T00:58:56 it sees bug << "test" << "this"; it inlines it.. you eventually end up with something looking like this: 2011-03-01T00:59:01 *** delt0r_ has joined #aichallenge 2011-03-01T00:59:21 bug.file << "test"; bug.file << "this"; 2011-03-01T00:59:33 but that's inside the #if, so you're good 2011-03-01T00:59:58 inlining operators works fine because the compiler is aware of what the code means, unlike #define which is just straight substitution 2011-03-01T01:00:09 *** delt0r___ has quit IRC (Ping timeout: 240 seconds) 2011-03-01T01:00:14 whiy wouldn't it go: 2011-03-01T01:00:14 ((bug.file << "test") << "this") -> (<< "this")? 2011-03-01T01:00:25 antimatroid: no.. 2011-03-01T01:00:31 antimatroid: because there's a return bug 2011-03-01T01:00:31 isn't that how those functions work? 2011-03-01T01:00:38 which also gets inlined 2011-03-01T01:00:45 ah, so still return the bug? 2011-03-01T01:00:48 that was my issue 2011-03-01T01:00:50 i'm all good now 2011-03-01T01:00:51 (<< "this") is not valid code 2011-03-01T01:01:08 sigh: that was my point, i thought it would break because i thought i was making the else function return nothing 2011-03-01T01:01:16 sigh: it won't, because of the return 2011-03-01T01:01:19 antimatroid: * 2011-03-01T01:01:20 it's all good now 2011-03-01T01:01:27 antimatroid: any inline functions act the same as non-inline 2011-03-01T01:01:35 they're just faster, and take more space 2011-03-01T01:01:37 amstan, I know, I was commenting on what antimatroid said 2011-03-01T01:01:47 sigh: wrong person 2011-03-01T01:01:51 so it would be ideal to use inline pretty much everywhere? 2011-03-01T01:01:59 antimatroid: yes and no 2011-03-01T01:02:08 yes for me, no for the server? 2011-03-01T01:02:10 antimatroid: it makes your binary bigger if you do it everywhere 2011-03-01T01:02:15 antimatroid: if your code is huge, the cpu has to get more stuff from the ram, you'll have a ton of cache misses 2011-03-01T01:02:21 antimatroid: therefore it'll actually go slower 2011-03-01T01:02:25 ah okay 2011-03-01T01:02:41 antimatroid: only inline if you call it in few places, but you call it a lot 2011-03-01T01:02:49 ex: inside a huge loop 2011-03-01T01:03:19 and if your inlined function is small of course... 2011-03-01T01:03:36 in your case.. your inline function has just the return, if debug is off 2011-03-01T01:04:34 Frontier & mcleopold: what's the diff between ants/viewer and ants/www? 2011-03-01T01:05:56 ants/viewer was McLeopold's canvas visualizer 2011-03-01T01:06:06 poo, it's now not working with endl 2011-03-01T01:06:15 Frontier: well.. which one is cooler/working? 2011-03-01T01:06:25 i don't think it likes that being inline 2011-03-01T01:06:28 Frontier: /keeping? 2011-03-01T01:06:39 amstan: Of course mine is cooler 2011-03-01T01:06:52 Frontier: does it work? any point to keeping the old one? 2011-03-01T01:07:03 ask McLeopold if he wants to remove his from git to avoid confusion 2011-03-01T01:07:55 contestbot: later tell mcleopold can you remove ants/viewer? i was told frontier's visualizer is cooler, and can replace yours 2011-03-01T01:07:55 amstan: As you wish. 2011-03-01T01:08:00 there :) 2011-03-01T01:08:01 It hasn't been updated in a while I would assume and is no longer working with the newer replays, has no keyboard input etc. 2011-03-01T01:08:52 amstan: McLeopold and I still have to work together, so don't make me look like a douche ^^. 2011-03-01T01:08:52 *** chris__0076 has joined #aichallenge 2011-03-01T01:08:59 i can't inline the function that does endl 2011-03-01T01:09:02 but i have the rest inlined 2011-03-01T01:09:04 Frontier: lol 2011-03-01T01:09:37 antimatroid: feel free to ask in #c++, i'm not exactly sure how the overloaders work 2011-03-01T01:10:13 *** Palmik has joined #aichallenge 2011-03-01T01:10:21 Frontier: about the canvas not working in ie, we can probably do excanvas 2011-03-01T01:10:26 There are some strict rules on inlining. The compiler must know what method get's called at runtime of course. 2011-03-01T01:10:54 amstan: Did you read the bug report I posted? 2011-03-01T01:10:56 Frontier: i expect it's the "std::ostream& (*manipulator)(std::ostream&)" parameter 2011-03-01T01:11:05 Frontier: yes, it's in front of me 2011-03-01T01:11:25 *** chris___0076 has quit IRC (Ping timeout: 240 seconds) 2011-03-01T01:11:52 *** amriedle has quit IRC (Quit: Lost terminal) 2011-03-01T01:12:03 Trouble is I don't boot to Windows often, so I would need a better way to test IE8. Maybe I can get it to run on wine 2011-03-01T01:13:26 Frontier: well.. if it's not 100%, it's fine.. as long as you see ants as pixels.. it's probably good 2011-03-01T01:13:52 amstan: I had that workin 2011-03-01T01:13:54 we can probably include a message to get a canvas supported browser 2011-03-01T01:14:31 what percentage of hits to the site are IE? 2011-03-01T01:14:51 let me bring up analytics 2011-03-01T01:14:56 I want to popup an 'alert('Go get a standards compliant browser to view this as it is meant to be!')' 2011-03-01T01:15:01 *** me0w has quit IRC (Quit: Quitte) 2011-03-01T01:15:04 ~5% 2011-03-01T01:15:06 Frontier: yep.. lol 2011-03-01T01:15:20 It is in the bug report :) 2011-03-01T01:15:32 only 5%? 2011-03-01T01:15:35 I looked it up this morning for Mcleopold :) 2011-03-01T01:15:36 who cares then? :P 2011-03-01T01:16:06 they say the same for linux, so.. if they do it for us, why should we care for them? lol 2011-03-01T01:16:17 sigh: Sometimes I think that myself - all the trouble for 5% of users who can easily install a second browser ^^. 2011-03-01T01:16:35 Frontier: i have ie6 handy :P 2011-03-01T01:16:41 this is why I avoid web dev :P 2011-03-01T01:16:41 6% IE 2011-03-01T01:16:54 i should start browsing the site for ie6 ehe 2011-03-01T01:16:58 from* 2011-03-01T01:17:03 70% of IE is IE8, 12% is IE7 2011-03-01T01:17:12 10% is IE9 2011-03-01T01:17:17 and 6% is IE6 2011-03-01T01:17:32 oh we have some 9ers already? that one has canvas support i believe 2011-03-01T01:17:53 Frontier: right!!! and microsoft supports standards properly.. 2011-03-01T01:18:10 they never had any compatibility issues to stuff they haven't made, lol 2011-03-01T01:18:44 Frontier: newvisits is 10% IE though 2011-03-01T01:19:01 Frontier: so maybe we should work on that 2011-03-01T01:19:56 amstan: newvisits = new ip? 2011-03-01T01:20:22 Frontier: i'm pretty sure there's some tracking cookie magick happening 2011-03-01T01:20:29 or just antimatroid testing all versions from ie 5 to 9? 2011-03-01T01:20:50 antimatroid: yes, make sure our side works in ie5! 2011-03-01T01:21:51 I'm going to start hitting the site with lynx 2011-03-01T01:21:56 support that 2011-03-01T01:22:01 sigh: lynx should work fine 2011-03-01T01:23:45 woah, my wikipedia page got imporved: http://en.wikipedia.org/wiki/Google_AI_Challenge 2011-03-01T01:24:29 amstan: https://gist.github.com/847673 fixed 2011-03-01T01:25:02 amstan: wow, I signed in using lynx 2011-03-01T01:25:16 antimatroid: and that's why i don't like c++, because of those monstrosities 2011-03-01T01:25:32 sigh: it's pretty standards compliant, the website 2011-03-01T01:25:35 amstan: but it's wonderful to use 2011-03-01T01:25:44 antimatroid: not really 2011-03-01T01:27:05 this is all I get for the planet wars visualiser: |# | # | |> | # | #| 2011-03-01T01:27:06 :( 2011-03-01T01:30:02 sigh: are you participating in ants? and python or c++? :P 2011-03-01T01:30:49 probably not, I'm likely moving to sydney in april 2011-03-01T01:31:00 to the dirty city? 2011-03-01T01:31:02 :O 2011-03-01T01:31:03 yeah 2011-03-01T01:32:45 sigh: so? why not? 2011-03-01T01:32:59 lack of time :P 2011-03-01T01:33:23 yeah... gimme your time 2011-03-01T01:33:29 damn uni students :P 2011-03-01T01:33:42 i have 5 hours of classes each week this semester aha 2011-03-01T01:33:50 +1, 2011-03-01T01:34:09 but i have thesis stuff to do as well, but i'm planning to do game theory stuff with an engine and bots etc. anyay :P 2011-03-01T01:34:20 hence why i need my process thing to at least work 2011-03-01T01:35:11 omg my cat is going nuts 2011-03-01T01:35:31 RUNRUNRUNRUNRUNRUNRUNRUNCHANGEDIRECTIONRUNRUNDOA180RUNRUNRUN 2011-03-01T01:35:35 contestbot: seen mcleopold 2011-03-01T01:35:35 amstan: mcleopold was last seen in #aichallenge 4 hours, 32 minutes, and 26 seconds ago: I'm out for awhile, cya 2011-03-01T01:35:39 that's it.. 2011-03-01T01:35:43 his 4 hours expired 2011-03-01T01:35:56 i'm going to watch tom and jerry, then go to bed 2011-03-01T01:36:02 since it's 2011-03-01T01:36:05 contestbot: time 2011-03-01T01:36:05 amstan: 01:36 AM, March 01, 2011 2011-03-01T01:36:08 at my place 2011-03-01T01:36:12 contestbot: time 2011-03-01T01:36:12 jmcarthur: 01:36 AM, March 01, 2011 2011-03-01T01:36:15 wrong 2011-03-01T01:36:28 oh wait.. 2011-03-01T01:36:29 omg 2011-03-01T01:36:32 12:20 2011-03-01T01:36:36 wtf? 4 minutes off 2011-03-01T01:36:41 that's not cool 2011-03-01T01:36:44 jmcarthur: my bot will be even faster now :P 2011-03-01T01:36:49 my server is not in good shape, lol 2011-03-01T01:36:55 antimatroid: i am defeated 2011-03-01T01:37:01 time to get ntp 2011-03-01T01:37:41 antimatroid: i could always just write bindings to yours :P 2011-03-01T01:37:53 i don't like my timer 2011-03-01T01:38:01 but i can't work out how else to get the time in milliseconds 2011-03-01T01:38:12 right.. the server ntpdate was using decided to not coopierate 2011-03-01T01:23:55 contestbot: time 2011-03-01T01:24:03 ... 2011-03-01T01:24:07 contestbot: hi 2011-03-01T01:24:11 contestbot: you there? 2011-03-01T01:24:19 * amstan kicks CIA-89 2011-03-01T01:24:20 ow 2011-03-01T01:24:25 ok.. so ... my irc works 2011-03-01T01:24:29 contestbot died.. 2011-03-01T01:25:54 *** contestbot has joined #aichallenge 2011-03-01T01:25:54 *** ChanServ sets mode: +o contestbot 2011-03-01T01:26:00 contestbot: time 2011-03-01T01:26:00 amstan: 01:26 AM, March 01, 2011 2011-03-01T01:26:04 there we go 2011-03-01T01:27:11 amstan: Does the contestbot work according to your local time or its time? 2011-03-01T01:27:21 JackyAlcine: system time 2011-03-01T01:27:34 That's a bit depressing. 2011-03-01T01:27:37 which is the same as my time 2011-03-01T01:27:47 which is the same as EST 2011-03-01T01:27:58 JackyAlcine: why? 2011-03-01T01:28:03 how would it work out your localtime o_O 2011-03-01T01:28:15 sigh: GeoIP, perhaps? 2011-03-01T01:28:24 amstan: It would have been a nifty feature. 2011-03-01T01:28:31 contestbot: apropos time 2011-03-01T01:28:31 amstan: Status uptime, Time ctime, and Time time 2011-03-01T01:28:33 yeah, that could give a reasonable guess 2011-03-01T01:28:41 And funny enough, I'm on EST as well. :) 2011-03-01T01:28:54 JackyAlcine: geoip would require some tought, and idk if it's that needed 2011-03-01T01:29:01 And _definitely_ have school tomorrow. 2011-03-01T01:29:01 contestbot: uptime 2011-03-01T01:29:01 amstan: I have been running for 3 minutes and 15 seconds. 2011-03-01T01:29:16 amstan: Okay, but it's just a though. 2011-03-01T01:29:20 *** McLeopold has joined #aichallenge 2011-03-01T01:29:20 too bad... i had a huge one though 2011-03-01T01:29:33 McLeopold: yay, i was contemplating leaving 2011-03-01T01:29:58 amstan: I want to leave my visualizer in. It will morph into a map editor at some point. 2011-03-01T01:30:03 McLeopold: i made my own contest now, submissions are open: http://home.hypertriangle.com/~alex/aichallenge/website/ 2011-03-01T01:30:31 McLeopold: ok, cool 2011-03-01T01:31:08 does mail work? 2011-03-01T01:31:12 no 2011-03-01T01:31:32 is it running planet wars? 2011-03-01T01:31:37 no 2011-03-01T01:31:40 backend is off 2011-03-01T01:31:44 oh 2011-03-01T01:31:49 i was just joking 2011-03-01T01:31:59 it'll give you an error when you try to upload stuff 2011-03-01T01:32:24 antimatroid: i need someone to write me content for the website 2011-03-01T01:32:41 amstan: at least my upload is working ;P 2011-03-01T01:33:40 amstan: which parts? 2011-03-01T01:33:49 the specs page is pretty much done 2011-03-01T01:33:52 antimatroid: anything that's planet wars, http://home.hypertriangle.com/~alex/aichallenge/website/ 2011-03-01T01:34:17 The description of "ants" can be ripped straight from the specs page 2011-03-01T01:34:30 no it can't 2011-03-01T01:34:47 unless you only want math majors to join 2011-03-01T01:35:05 how would you explain it? 2011-03-01T01:35:10 *** sigh has quit IRC (Ping timeout: 252 seconds) 2011-03-01T01:35:12 i tried to keep that non-mathsy 2011-03-01T01:35:18 antimatroid: http://pastebin.com/VaVxSUtu 2011-03-01T01:36:01 amstan: alright just in a .txt file? 2011-03-01T01:36:16 antimatroid: simple html would be nice 2011-03-01T01:36:24 i've never used html in my life :P 2011-03-01T01:36:44 .... 2011-03-01T01:36:47 really... 2011-03-01T01:36:48 i'll take a look at those pages and try to base it off those? 2011-03-01T01:37:01 antimatroid: make them wiki pages then 2011-03-01T01:37:18 okay easy 2011-03-01T01:37:22 antimatroid: it might be a better idea if we do those 2011-03-01T01:37:38 you did a good job on the game specs 2011-03-01T01:37:46 that was my first wiki attempt :P 2011-03-01T01:37:58 formatting at least, i haven't looked at the content, lol\ 2011-03-01T01:38:13 McLeopold: did you see my maze maps that i accidentally made? 2011-03-01T01:38:22 antimatroid: http://www.savagechickens.com/2009/08/something-new.html <-- a description like that 2011-03-01T01:38:39 McLeopold: cowboys and aliens! 2011-03-01T01:38:50 http://codepad.org/ZKrF837z <- maze maps 2011-03-01T01:39:13 antimatroid: why don't we use # for walls? 2011-03-01T01:39:17 wow, so much for radius 2011-03-01T01:39:18 ask mcleopold 2011-03-01T01:39:27 because it's for comments 2011-03-01T01:39:29 antimatroid: and space for passable 2011-03-01T01:39:41 so you can count it 2011-03-01T01:39:44 use ; for comments 2011-03-01T01:39:51 it's more visible 2011-03-01T01:40:02 or // 2011-03-01T01:40:03 where were you 1 month ago? 2011-03-01T01:40:06 lol 2011-03-01T01:40:08 fine 2011-03-01T01:40:17 amstan: i fought for #, it doesn't matter though 2011-03-01T01:40:22 it ain't code 2011-03-01T01:40:26 antimatroid: # all the way! 2011-03-01T01:40:29 and i don't like ' ' for land 2011-03-01T01:40:43 even though bots don't ever see char maps anymore 2011-03-01T01:41:23 McLeopold: what's https://github.com/McLeopold/aichallenge/blob/epsilon/install_server.sh? 2011-03-01T01:41:23 okay, tabs for walls, spaces for land 2011-03-01T01:41:44 McLeopold: then you get http://en.wikipedia.org/wiki/Whitespace_(programming_language) 2011-03-01T01:41:45 my notes, it's grown 2011-03-01T01:41:51 McLeopold: i like it :P 2011-03-01T01:42:18 McLeopold: like these? https://github.com/aichallenge/aichallenge/blob/epsilon/README.md 2011-03-01T01:44:08 McLeopold: "wow, so much for radius"? 2011-03-01T01:44:14 yeah, but I'm going for totally automatted 2011-03-01T01:44:27 oh, with the maze maps being single file 2011-03-01T01:44:42 McLeopold: why do you need totally automated db creation? 2011-03-01T01:44:45 a different tactic may be in order 2011-03-01T01:45:13 the more automated, the better 2011-03-01T01:45:27 McLeopold: you only do this once.. 2011-03-01T01:45:28 so I can blow away my server and set it up again with 1 command 2011-03-01T01:45:32 there is a better way to do symmetric maps, but i haven't quite worked out how to make them yet 2011-03-01T01:45:51 antimatroid: make a map, half the height, then mirror it 2011-03-01T01:46:12 the idea is to basically find rtranslate and ctranslate such that translating a position n (no players) times gets you back to the same position 2011-03-01T01:46:14 amstan: also, you said that last time, for the last contest, and here we are fighting with getting servers working 2011-03-01T01:46:42 McLeopold: ok 2011-03-01T01:47:55 i think if you start with a square nxn grid with players along the diagonal you can do random transformations to get any of the possible scenarios 2011-03-01T01:48:37 www -> website? a little bikeshedding? 2011-03-01T01:49:02 McLeopold: just to have the same name as zeta 2011-03-01T01:49:27 epsilon now has the same folder structure as zeta 2011-03-01T01:49:34 okay 2011-03-01T01:49:40 I have a git question 2011-03-01T01:49:45 fire 2011-03-01T01:49:54 when I pulled, I still have a www folder, what's the git way to clean it 2011-03-01T01:50:35 McLeopold: git clean -n . 2011-03-01T01:50:49 McLeopold: it'll output what it will remove, then use f instead of n if you're sure 2011-03-01T01:51:12 i usually do git add ., git reset --hard instead 2011-03-01T01:51:59 what's the difference? 2011-03-01T01:52:06 i think it's the same 2011-03-01T01:52:35 it's just that i remember the latter easier 2011-03-01T01:53:33 my trueskill dir is empty? 2011-03-01T01:53:33 what is coffeescript? 2011-03-01T01:53:41 compiles to javascript 2011-03-01T01:53:52 kk 2011-03-01T01:53:55 McLeopold: you have to init it: git submodule init; git submodule update 2011-03-01T01:53:58 an attempt to improve js syntax 2011-03-01T01:54:29 amstan: so then, it's not really in the repo? 2011-03-01T01:55:01 McLeopold: nope, it's an outside repo 2011-03-01T01:55:45 hmm... then in zeta, hopefully games can be submodules 2011-03-01T01:55:54 yeah, that would be cool 2011-03-01T01:56:33 submodules can be branches of the same repo i think 2011-03-01T01:57:29 I don't get the init command, was that the full syntax? 2011-03-01T01:58:10 oh, had to be at root 2011-03-01T01:58:11 git submodule init should work 2011-03-01T02:00:40 antimatroid: are you committing the map maker to git? 2011-03-01T02:00:50 McLeopold: which one? :P 2011-03-01T02:00:55 all of them 2011-03-01T02:01:20 they're all a mess, i can give you one if you need it for testing? otherwise i don't think there's a huge rush atm 2011-03-01T02:01:52 how many python files? 2011-03-01T02:02:00 a single file per map gen 2011-03-01T02:02:07 do you want asymmetric? 2011-03-01T02:02:48 do you have git write access? 2011-03-01T02:03:04 no, nor do i know how :P 2011-03-01T02:03:18 actually, asymmetric is the slowest 2011-03-01T02:03:23 do you want it? do you use something else? 2011-03-01T02:03:34 don't give me write access, i'd break things 2011-03-01T02:03:46 ok then 2011-03-01T02:03:57 antimatroid: too late 2011-03-01T02:03:58 do you have any vcs? 2011-03-01T02:04:09 i don't even know what that is :P 2011-03-01T02:04:18 someone should teach you 2011-03-01T02:04:20 scm? vcs? 2011-03-01T02:04:30 nope :P 2011-03-01T02:04:32 hg, git, svn, sourcesafe 2011-03-01T02:04:43 haha, got you, that last one doesn't count 2011-03-01T02:05:14 okay, it's time to drink the coolaid 2011-03-01T02:05:14 McLeopold: lol 2011-03-01T02:05:39 McLeopold: little did he know, he just used a scm: https://gist.github.com/847673 2011-03-01T02:06:37 McLeopold: http://codepad.org/gqEqucp2 add that for the moment if you want 2011-03-01T02:07:27 codepad should interface with github gists 2011-03-01T02:08:16 antimatroid: I don't need your code now. I just want to get you hooked into the system. 2011-03-01T02:08:23 ready to learn git? 2011-03-01T02:08:33 McLeopold: he's on windows 2011-03-01T02:08:36 last time i tried i failed at getting one of those passkeys 2011-03-01T02:08:38 so am i 2011-03-01T02:08:45 McLeopold: perfect then, 2011-03-01T02:08:46 go for it though :) 2011-03-01T02:09:26 antimatroid: just remember one thing.. with git, you can't break it... everything that you put in it will still be there 2011-03-01T02:11:03 http://code.google.com/p/msysgit/downloads/list 2011-03-01T02:11:11 that's what I use 2011-03-01T02:11:19 it's command line 2011-03-01T02:11:29 I also installed the unix tools from it, which is nice 2011-03-01T02:11:45 amstan: he has commit rights? 2011-03-01T02:12:01 McLeopold: yes, you'll have to guide him through key setup though 2011-03-01T02:12:03 get the top link, git-blabla-preview 2011-03-01T02:12:17 amstan: we'll start with password 2011-03-01T02:12:27 you can use just a password? 2011-03-01T02:12:30 yes 2011-03-01T02:12:37 i didn't know that, :O 2011-03-01T02:12:52 +1 McLeopold 2011-03-01T02:12:55 * amstan leaves 2011-03-01T02:13:13 antimatroid: you with me? 2011-03-01T02:13:22 yeah i'm doing stuff 2011-03-01T02:14:00 what's the one with the unix tools? 2011-03-01T02:14:16 antimatroid: cygwin? 2011-03-01T02:14:19 in the git-preview install, it's on one of the wizard pages 2011-03-01T02:14:25 ah okay 2011-03-01T02:14:43 amstan: it's installed, have never used it though 2011-03-01T02:14:43 amstan: this is a little better than cygwin 2011-03-01T02:15:07 i'm on a 1500k line, so it'll take a few minutes to download 2011-03-01T02:15:32 it's at about 45% 2011-03-01T02:15:36 k 2011-03-01T02:17:56 aichallenge: McLeopold epsilon * r41f8e8b / (ants/McMaps.py ants/engine.py): Small changes to engine output. - http://bit.ly/gzZFMd 2011-03-01T02:18:04 aichallenge: McLeopold epsilon * r6c8848b / (1389 files in 53 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://bit.ly/et2K2g 2011-03-01T02:19:28 antimatroid: do you know vi? 2011-03-01T02:19:33 nope :P 2011-03-01T02:19:36 i'm installing now 2011-03-01T02:19:39 ok 2011-03-01T02:20:02 so i want to run git and include unix tools from the command prompt? 2011-03-01T02:22:07 yes 2011-03-01T02:22:19 And check the box that says you know what you are doing 2011-03-01T02:22:28 the one the overwrites stuff 2011-03-01T02:22:35 :) 2011-03-01T02:22:46 antimatroid: another funny line in your code: getline(is, junk); 2011-03-01T02:23:23 Frontier: why is that funny? 2011-03-01T02:23:44 antimatroid: win7? 2011-03-01T02:23:45 because getline is junk makes a sentence 2011-03-01T02:24:56 xp :P 2011-03-01T02:25:01 ok 2011-03-01T02:25:15 hang on, i need to log off, explorer is dead 2011-03-01T02:25:50 *** antimatroid has quit IRC (Read error: Connection reset by peer) 2011-03-01T02:28:02 *** antimatroid has joined #aichallenge 2011-03-01T02:28:25 so, are you alive? 2011-03-01T02:28:39 i think so 2011-03-01T02:28:55 i have git bash open 2011-03-01T02:28:59 open a command line and type git 2011-03-01T02:29:36 .. 2011-03-01T02:30:04 I don't actually use gitbash or gitgui 2011-03-01T02:30:15 do you use the usage help info? 2011-03-01T02:30:49 me? what? :P 2011-03-01T02:30:52 yes you 2011-03-01T02:30:55 no 2011-03-01T02:30:58 i think 2011-03-01T02:38:08 *** sigh has joined #aichallenge 2011-03-01T02:46:22 amstan: I thought antimatroid was in the list of committers? 2011-03-01T02:47:25 McLeopold: he is, he was not publicized though 2011-03-01T02:47:51 * amstan is sleeping 2011-03-01T02:48:17 okay 2011-03-01T02:56:53 aichallenge: antimatroid epsilon * r0b29d7a / ants/mapgen/mapgen.py : added map gen - http://bit.ly/dJ6hWr 2011-03-01T03:02:34 aichallenge: McLeopold epsilon * r751cf74 / (ants/mapgen/McMaps.py ants/mapgen.py ants/McMaps.py): Moved McMaps.py into mapgen folder. - http://bit.ly/icoYFN 2011-03-01T03:32:00 *** JackyAlcine has quit IRC (Ping timeout: 240 seconds) 2011-03-01T03:39:10 *** fic885 has joined #aichallenge 2011-03-01T03:40:54 *** boegel has joined #aichallenge 2011-03-01T03:41:16 *** fic885 has left #aichallenge 2011-03-01T03:49:57 @later tell janzert where should jail_id_rsa be or where is it generated? 2011-03-01T03:49:57 McLeopold: Aye, aye, sir 2011-03-01T03:54:02 i guess i should probably comment the starter bot :\ 2011-03-01T03:54:12 comments are for pussies 2011-03-01T04:21:10 *** aerique has joined #aichallenge 2011-03-01T04:21:40 *** McLeopold has left #aichallenge 2011-03-01T04:21:49 aichallenge: McLeopold epsilon * r7612089 / (5 files in 3 dirs): Added paths to server_info.* files. - http://bit.ly/iiiQuO 2011-03-01T04:32:10 mega/bocsimaco been around lately? wondering if he's planning on writing the Common Lisp starter package 2011-03-01T04:34:22 aerique: i think he signs in sometimes 2011-03-01T04:34:30 i'll try to remember to prod him for one 2011-03-01T04:34:47 you could always use my c++ one :P 2011-03-01T04:42:34 *** amstan has quit IRC (Ping timeout: 276 seconds) 2011-03-01T04:45:38 Frontier: is it birth or spawn radius? 2011-03-01T04:45:58 i think spawn, just noticed i still have birth in my starter pack 2011-03-01T04:50:56 antimatroid: i'd rather use CL :) anyway, i'll checkout the repo and work on my own starter pack and if he doesn't write one i'll commit my substandard one 2011-03-01T04:52:26 sounds good :) 2011-03-01T04:52:50 there's a very unofficial competition for best starter pack 2011-03-01T04:53:30 and i think the unofficial judge will be the relative rankings of people right at the beginning, so maybe not much room for you :P 2011-03-01T04:53:36 as people don't really use lisp 2011-03-01T04:54:10 i don't know if i got ie working on wine and what version it is. it always throws 'Invalid parameter' errors when I enter a url and looks like ie6 although winetricks supposedly installed ie8 2011-03-01T04:54:19 i don't mind that, as long as i can use it. there's always been some people using the last two challenges 2011-03-01T04:54:34 that wont be an issue 2011-03-01T04:54:47 i'm pretty sure what's already there will be there again 2011-03-01T05:21:26 wine doesn't display the javascript errors in ie, but i found that I can get at the information if I override window.onerror(msg, source, line) 2011-03-01T05:24:44 the world would be a lot better off without ie 2011-03-01T05:36:14 antimatroid: watch this in ie ^^: http://marco-leise.homedns.org/antgui/work/visualizer.php?game_id=2#visualizer 2011-03-01T05:43:52 *** mega1 has joined #aichallenge 2011-03-01T05:45:31 it didn't load 2011-03-01T05:45:48 mega1: i was queried as to whether you might perhaps be writing a starter bot for lisp for ants? 2011-03-01T05:45:55 or might consider it? :P 2011-03-01T05:46:44 how is the concept coming along? 2011-03-01T05:47:54 it's pretty much there, the only real decision left is battle resolution i think, the rest is just setting everything up 2011-03-01T05:48:06 ETA? 2011-03-01T05:48:17 a few weeks max maybe? 2011-03-01T05:48:23 there isn't any major rush on bots 2011-03-01T05:48:52 I should have a bit more time next month, so it's possible. 2011-03-01T05:49:03 it is next month :P 2011-03-01T05:49:39 depends on how much you lag 2011-03-01T05:49:59 I mean April. 2011-03-01T05:50:44 i think we'd be hoping to be live by then 2011-03-01T05:50:54 i think aerique is happy to write one anyway 2011-03-01T05:51:03 quite possible 2011-03-01T05:51:21 how long will the contest itself be? 2011-03-01T05:51:31 i think we might go for 2 months 2011-03-01T05:51:37 via the goldy locks principle 2011-03-01T05:51:45 tron was too short, pw possibly too long? 2011-03-01T05:51:54 it's still undecided though 2011-03-01T05:53:02 have the game changed during the last couple of weeks? 2011-03-01T05:53:39 mega1: not a huge amount, mostly input etc., this wiki page is up to date bar viasuliser input 2011-03-01T05:53:50 https://github.com/aichallenge/aichallenge/wiki/Ants-Game-Specification 2011-03-01T05:54:14 Frontier: does it work in your ie? that was ie6 i tested it in (I don't bother updating it, i don't use it) 2011-03-01T05:56:07 the "starter strategy" we've mostly agreed on is pretty stupid and easy to do, iterate through the ants, moving them n, e, s, or w (first one that is land) if possible, and letting them collide (with the intention of that being the first change for beginners) 2011-03-01T05:56:17 "code that specifically implements anything related to the moves a bot would make"??? 2011-03-01T05:56:30 yeah, i need to phrase that better 2011-03-01T05:56:52 it's basically to say it would be better if you don't share your code for your actual bot, but i don't want it to sound like people can't share tools or general strategies 2011-03-01T05:57:09 okay 2011-03-01T05:57:15 how about this: "will either spawn into an ant or disappear, depending on whether there is a conflict"? 2011-03-01T05:57:26 what's a conflict in this context? 2011-03-01T05:57:39 a.*.b 2011-03-01T05:57:49 both are outside the battle distance, but both would spawn an ant 2011-03-01T05:57:53 so the food dies 2011-03-01T05:58:06 it's outlined better below, that's just a really general explanation about the game 2011-03-01T05:59:13 the turn input explanation is a bit of a mess too, we only recently decided on it, but the examples should make it clear 2011-03-01T06:04:11 antimatroid: it loads in what i believe is ie 8. 2011-03-01T06:04:44 but as i am editing this page all the time you might often hit it when it has errors 2011-03-01T06:05:14 it loaded in chrome for me 2011-03-01T06:05:25 i don't think ie6 should be much of a priority anyway 2011-03-01T06:06:00 could it be me not having something installed for java? 2011-03-01T06:06:02 like with flash? 2011-03-01T06:06:03 *** mceier has joined #aichallenge 2011-03-01T06:06:16 Have you thought more about the n-playerness and how it affects scoring, ranking? 2011-03-01T06:06:17 for ie** 2011-03-01T06:06:33 what's the range for n anyway? 2011-03-01T06:06:48 atm it's undecided, i've been using {4, 6, 9} 2011-03-01T06:07:04 i think dmj111 has done some stuff about rankings with trueskill 2011-03-01T06:07:30 i don't think anything has actually been done about adding it to the server, but it's on the list 2011-03-01T06:08:00 jeff only just emailed dmj about it a few days ago after having not really thought about it for a while 2011-03-01T06:12:44 *** boegel has quit IRC (Quit: *poof!*) 2011-03-01T06:17:23 *** boegel has joined #aichallenge 2011-03-01T06:18:06 mega1: yeah, i was wondering whether you had plans writing a starter package for this challenge, so we don't get in eachothers way 2011-03-01T06:19:25 aerique: I don't know yet. Most likely I won't have the time to do it before the contest starts. 2011-03-01T06:20:03 mega1: alright, i'll go make an initial package then 2011-03-01T06:20:25 with a ProxyBot? 2011-03-01T06:21:00 and asdf? :-) 2011-03-01T06:21:11 mega1: sure :) 2011-03-01T06:21:19 cool 2011-03-01T06:21:24 mega1: you can always update it to something better 2011-03-01T06:21:42 mega1: did you make any improvements to the proxybot last challenge? 2011-03-01T06:22:55 I put it into its own package, added a one-shot param, ported it to allegro cl since 2011-03-01T06:23:14 and probably a few tweaks here and there 2011-03-01T06:24:38 mega1: okey, good to know. i'll it out 2011-03-01T06:29:03 am i supposed to be able to play a local game once i've checked out aichallenge from github? (i'm running into some issues) 2011-03-01T06:36:14 aerique: i'm yet to run a game with an engine locally 2011-03-01T06:36:57 i just added a cpp starter bot to github, it should show up in a minute 2011-03-01T06:39:58 antimatroid: ah alright, good to know. so you guys are running against a server then? 2011-03-01T06:40:56 i haven't run any games 2011-03-01T06:40:56 *** j3camero has quit IRC (Read error: Connection reset by peer) 2011-03-01T06:41:08 i have passed my bot very simple sample game strings 2011-03-01T06:41:25 ok, that's what i'll do as well then, thanks again :) 2011-03-01T06:41:35 *** j3camero has joined #aichallenge 2011-03-01T06:41:42 np 2011-03-01T06:41:49 come on cia 2011-03-01T06:41:58 i will link you to my bot in a second 2011-03-01T06:43:06 if it ever turns up :\ 2011-03-01T06:45:57 hmmmm it didn't take this long last time 2011-03-01T06:47:39 mega1: actually, it looks like there was stuff put up in the last day about rankings 2011-03-01T06:47:47 https://github.com/aichallenge/trueskill/tree/0e83d87fb278eaceb0c234d680cafcb82befc0d0 2011-03-01T06:52:12 is trueskill timeless in the sense it doesn't matter what order the games are played? 2011-03-01T06:53:24 *** kaemo has quit IRC (Remote host closed the connection) 2011-03-01T06:55:18 i have no idea sorry 2011-03-01T06:55:29 i'm pretty sure the pairing algorithm throws that out anyway 2011-03-01T06:55:50 well, in this case grouping algorithm 2011-03-01T07:38:52 *** Naktibalda has joined #aichallenge 2011-03-01T07:45:14 *** sigh has quit IRC (Remote host closed the connection) 2011-03-01T08:05:03 *** jbroman has quit IRC (Ping timeout: 264 seconds) 2011-03-01T08:06:19 *** pgpaskar_ has quit IRC (Ping timeout: 250 seconds) 2011-03-01T08:06:20 *** j3camero has quit IRC (Read error: Connection reset by peer) 2011-03-01T08:06:31 *** Zepp has quit IRC (Ping timeout: 240 seconds) 2011-03-01T08:08:39 *** Zepp has joined #aichallenge 2011-03-01T08:09:08 *** jbroman has joined #aichallenge 2011-03-01T08:12:15 *** j3camero has joined #aichallenge 2011-03-01T08:13:38 *** pgpaskar_ has joined #aichallenge 2011-03-01T08:34:20 *** iFire` has joined #aichallenge 2011-03-01T08:34:30 *** iFire has quit IRC (Ping timeout: 272 seconds) 2011-03-01T08:39:18 *** Mathnerd314 has quit IRC (Ping timeout: 240 seconds) 2011-03-01T08:44:20 *** delt0r___ has joined #aichallenge 2011-03-01T08:45:39 *** delt0r_ has quit IRC (Ping timeout: 246 seconds) 2011-03-01T09:16:57 *** delt0r___ has quit IRC (Ping timeout: 272 seconds) 2011-03-01T09:55:00 *** helloworld has joined #aichallenge 2011-03-01T10:20:31 *** DanielVF has joined #aichallenge 2011-03-01T10:21:00 Simulator for simulating simulated battles. http://ai-contest.com/forum/viewtopic.php?f=20&t=1349 2011-03-01T10:24:29 *** DanielVF has quit IRC (Remote host closed the connection) 2011-03-01T10:25:13 *** DanielVF has joined #aichallenge 2011-03-01T10:31:12 @later tell McLeopold where are you seeing jail ssh keys used now? I'm pretty sure they're no longer used at all, and I think the latest cleanup pass by amstan should have removed all references 2011-03-01T10:31:12 janzert: Job's done. 2011-03-01T10:54:39 *** Naktibalda has quit IRC (Remote host closed the connection) 2011-03-01T10:56:03 *** Naktibalda has joined #aichallenge 2011-03-01T11:01:44 *** janzert1 has joined #aichallenge 2011-03-01T11:04:53 *** janzert has quit IRC (Ping timeout: 250 seconds) 2011-03-01T11:06:04 *** janzert1 has quit IRC (Ping timeout: 250 seconds) 2011-03-01T11:06:49 *** janzert has joined #aichallenge 2011-03-01T11:15:08 *** aerique has quit IRC (Remote host closed the connection) 2011-03-01T11:21:37 *** amstan has joined #aichallenge 2011-03-01T11:21:37 *** ChanServ sets mode: +o amstan 2011-03-01T11:34:14 *** McLeopold has joined #aichallenge 2011-03-01T11:36:08 janzert: the reference is in the readme file at the root of the repo 2011-03-01T11:36:26 yeah, I just found that a couple minutes ago 2011-03-01T11:36:46 I think amstan adapted those from my old notes 2011-03-01T11:37:21 the ssh keys are definitely outdated and overall they seem to mix worker and main server setup 2011-03-01T11:37:47 okay, still wrapping my head around how it all works. the more cleanup the better 2011-03-01T11:39:19 I did start working on getting a local setup so I can do it myself with the current layout and maybe give slightly better advice on the details no needed 2011-03-01T11:39:24 now needed 2011-03-01T11:40:02 but I probably won't actually get that finished till this evening 2011-03-01T11:40:06 *** aerique has joined #aichallenge 2011-03-01T11:40:21 okay 2011-03-01T11:41:03 *** aerique has quit IRC (Client Quit) 2011-03-01T11:44:00 amstan: I'm getting a 'sorry, wait 10 min' when trying to upload to my local server. You must have done something to the submit.php 2011-03-01T11:46:42 *** mceier has quit IRC (Quit: leaving) 2011-03-01T11:49:12 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2011-03-01T11:50:51 McLeopold: my guess would be you have a previous submission that hasn't been compiled 2011-03-01T11:51:20 any uncompiled submission or a submission more recent than 10 minutes in the past will trigger that 2011-03-01T11:51:43 I think it's probably related to a blank database 2011-03-01T11:52:52 hmm, possibly. the check is done in has_recent_submision in submission.php 2011-03-01T11:52:58 mega1: no, trueskill is order dependent. that's a downside 2011-03-01T11:53:26 yeah, it tickled my memory 2011-03-01T11:53:33 what's the upside? 2011-03-01T11:53:50 dmj111 had an idea to reduce the errors 2011-03-01T11:54:18 it only matters for a tie, which should be rare, and we can just put the lower ranked player first 2011-03-01T11:54:25 mega1: good matchmaking algorithm, supports >2 players, potential for not *completely* resetting score on resubmission 2011-03-01T11:54:36 *** mceier has joined #aichallenge 2011-03-01T11:54:42 not sure how a blank database would cause has_recent_submission to return true though 2011-03-01T11:54:42 mega1: and a nice climbing behavior instead of jumping around 2011-03-01T11:54:42 then, if you have a series of 2 player matches ending in draws, the errors should cancel somewhat 2011-03-01T11:54:54 more intuitive in that sense 2011-03-01T11:55:11 i lot of people complain that bayeselo makes no sense if you don't know how it works 2011-03-01T11:55:12 not completely resetting is a double edged sword 2011-03-01T11:55:17 i agree 2011-03-01T11:55:33 but at least with trueskill there is a more principled way to do it 2011-03-01T11:55:43 another upside is it handles n-player with ties 2011-03-01T11:56:02 I guess that's the main attraction. 2011-03-01T11:56:07 janzert: the sql is coming back with '1' 2011-03-01T11:56:23 upon resubmit, we could set sigma (uncertainty) to a large value, which means that the bot will start playing against bots of a similar skill to before the resubmission. might lead to a faster convergence and more interesting games 2011-03-01T11:56:36 janzert: ah, it is an uncompiled submission 2011-03-01T11:56:58 but since the uncertainty is high the bot can change rank much faster 2011-03-01T11:57:01 ahh, good (in the sense that I didn't see how it could return 1 with a blank database) :) 2011-03-01T11:57:09 uncertainty shrinks for each game played 2011-03-01T11:57:27 jmcarthur: trueskill is designed to add uncertainty over time (and you probably know this), so we just need to pick a nice value on resubmit 2011-03-01T11:58:03 McLeopold: well, we have to change it from the norm slightly in that we don't want to add (much?) uncertainty between games unless there is a resubmission 2011-03-01T11:58:29 and technically a resubmission can change nearly everything about the bot, so that would have to be a big change, technically 2011-03-01T11:58:57 anyway, yet another nice thing about doing it this way is that newly resubmitted bots will appear very low in the rankings and just converge more quickly, rather than appearing to not move at all 2011-03-01T11:58:58 Time can also change things because many bots will have been resubmitted. 2011-03-01T11:59:09 mega1: ah indeed 2011-03-01T11:59:13 jmcarthur: I'm thinking only add it on resubmit 2011-03-01T11:59:35 McLeopold: no i think we will need to give it a little adjustment each game, primarily due to what mega1 just said 2011-03-01T11:59:46 and then give it a much bigger adjustment on resubmission 2011-03-01T12:00:09 okay, small change for everyone each interval, large change for individual on resubmit 2011-03-01T12:00:19 yeah, that's what i'm thinking 2011-03-01T12:00:27 wait, each interval? 2011-03-01T12:00:33 it's typically each game 2011-03-01T12:00:44 if there are no resubmissions, then there is no need 2011-03-01T12:00:47 hmm... 2011-03-01T12:00:58 the best would be to add a little for everyone on any submission 2011-03-01T12:01:00 okay, change everyone a little on each resubmission 2011-03-01T12:01:03 yeah 2011-03-01T12:01:13 and the one that was resubmitted gets changed more 2011-03-01T12:01:22 but, that would be lost of work, so just count the resubmissions every 10 minutes or so 2011-03-01T12:01:29 sure, that works 2011-03-01T12:01:47 so much for incremental, though 2011-03-01T12:02:06 still more incremental than bayeselo at least 2011-03-01T12:02:38 i guess 10 minutes isn't going to be a big hit to accuracy, really. okay, i'm happy with that proposal 2011-03-01T12:02:48 McLeopold: hmm, idk 2011-03-01T12:02:49 near the deadline, uncertainty will spike :) 2011-03-01T12:02:59 McLeopold: know php? 2011-03-01T12:03:00 amstan: i figured it out 2011-03-01T12:03:08 McLeopold: heh. and nobody will know where they should really be ;) 2011-03-01T12:03:10 what was it? 2011-03-01T12:03:10 unfortunatley, yes 2011-03-01T12:03:21 I haven't turned on the background cron jobs yet 2011-03-01T12:03:31 they will only know with 95% certainty that they should be at least their current rank :P 2011-03-01T12:04:52 *** pyro-_ has quit IRC (Ping timeout: 240 seconds) 2011-03-01T12:06:08 McLeopold, antimatroid: for the people that want to make starter packages, what documentation can i give them? 2011-03-01T12:06:14 or example bots in other languages 2011-03-01T12:09:45 jmcarthur have you tested ranking bots with trueskill? 2011-03-01T12:11:12 DanielVF: it's hard to test because it's also dependent on the matching algorithm 2011-03-01T12:11:33 http://ai-contest.com/forum/viewtopic.php?f=20&t=1349 2011-03-01T12:11:43 DanielVF: i've been meaning to try it though, with a decent model for bots 2011-03-01T12:12:09 Yes, I agree, you have to test both the matching and the ranking together 2011-03-01T12:13:25 i was thinking of pregenerating a pool of bots with various skills and consistencies, then simulating matches with probabilistic outcomes 2011-03-01T12:13:41 *nods* 2011-03-01T12:13:50 gets even tougher if you want to also test the resubmissions idea 2011-03-01T12:14:06 but i think that idea by itself wouldn't be too hard to implement 2011-03-01T12:14:30 *nods* 2011-03-01T12:15:14 We do need to be sure to test it before the contest on a large enough scale. 2011-03-01T12:16:20 well we aren't wanting to use this for the final tournament anyway. this is basically just to provide the benefit of getting accurate rankings more quickly during development 2011-03-01T12:17:05 * janzert thinks the same thing should be used for the final tournament as the open tournament period 2011-03-01T12:17:23 the less changes that can be made between the two the better 2011-03-01T12:17:59 well then we definitely can't use trueskill 2011-03-01T12:18:19 although I do see the draw of using a direct elimination tournament for the finals 2011-03-01T12:18:22 it doesn't work well without the matching algorithm 2011-03-01T12:18:33 so we can't do anything like a bracket 2011-03-01T12:18:42 you would still use the matching algorithm during the finals too 2011-03-01T12:19:09 i just don't know. i've never seen something like trueskill used in a tournament 2011-03-01T12:19:12 only for ladders 2011-03-01T12:19:45 *** Mathnerd314 has joined #aichallenge 2011-03-01T12:21:16 so long as it's doing a good job at ranking it is working for the final tournament and if it's not then it's not really suitable for the open period either 2011-03-01T12:21:49 another way to look at it is that I'm pretty sure handling the open period ranking well is strictly harder than the final tournament 2011-03-01T12:22:59 McLeopold, janzert: anyone have any updates on readme.md? 2011-03-01T12:23:12 i heard that worker was out of date 2011-03-01T12:23:16 *** DanielVF_ has joined #aichallenge 2011-03-01T12:23:36 well, the installation instructions don't really make complete sense 2011-03-01T12:24:45 janzert: should i just remove the worker part for now? 2011-03-01T12:26:06 to me it looks mixed together between main and worker setup 2011-03-01T12:26:22 certainly everything about ssh keys is no longer relevant 2011-03-01T12:27:05 and the nproc limit stuff should be able to go away as well 2011-03-01T12:27:40 janzert: i separated it a little: https://github.com/aichallenge/aichallenge/wiki/Readme 2011-03-01T12:28:07 refresh 2011-03-01T12:28:20 darn 2011-03-01T12:28:45 that looks better anyway although I think it's lacking steps :) 2011-03-01T12:28:46 i added a bullet point 2011-03-01T12:28:49 well.. yeah 2011-03-01T12:29:29 really the worker setup should be run game_server_build.sh on a plain ubuntu install 2011-03-01T12:29:36 but that is broken right now 2011-03-01T12:29:40 ... 2011-03-01T12:29:42 lol 2011-03-01T12:29:53 ok.. well 2011-03-01T12:30:07 someone should go fix that, i just wanted to remove the false stuff from there 2011-03-01T12:30:12 I might have time tonight to work on fixing it 2011-03-01T12:30:42 and also change it so it will work on the same machine as a main server install 2011-03-01T12:31:00 work on fixing it tonight that is 2011-03-01T12:38:35 aichallenge: Alexandru Stan epsilon * r41f09c4 / (README.md trueskill): updated readme, updated trueskill to newer version - http://bit.ly/fX60j3 2011-03-01T12:39:30 McLeopold: do we have any bots already done? 2011-03-01T12:48:37 i did, but then the I/O protocol changed 2011-03-01T12:48:54 i think antimatroid's starter is up to date with it though, maybe 2011-03-01T12:49:11 McLeopold had a few in ants/bots before, but i don't know if they are up to date 2011-03-01T12:49:30 or if there are even in the same place now 2011-03-01T12:49:35 lots of changes last night that i didn't follow 2011-03-01T12:49:41 *if they are 2011-03-01T12:50:48 are we sandboxing the compilers this time? 2011-03-01T12:52:42 amstan is it possible for the trueskill package to live in the manager directory? 2011-03-01T12:53:43 I think everything that needs it will be there and if it's a subdirectory then it should be able to be directly imported 2011-03-01T12:55:58 janzert: yes 2011-03-01T12:56:05 *** DanielVF_ has quit IRC (Read error: Connection reset by peer) 2011-03-01T12:56:21 *** DanielVF_ has joined #aichallenge 2011-03-01T13:01:39 aichallenge: Alexandru Stan epsilon * r7a09d4d / (manager/trueskill trueskill): moved trueskill folder in manager - http://bit.ly/g95igt 2011-03-01T13:04:08 *** JackyAlcine has joined #aichallenge 2011-03-01T13:10:50 *** DanielVF_ has quit IRC (Quit: Rooms • iPhone IRC Client • http://www.roomsapp.mobi) 2011-03-01T13:13:45 3 java and python bots, they should be current 2011-03-01T13:15:53 McLeopold: i want to get some docs and some example bots for other people that want to make sample bots 2011-03-01T13:15:59 McLeopold: where can i find such things? 2011-03-01T13:17:17 ants/bots 2011-03-01T13:17:47 I'm behind on the replay format, but I'm pretty sure the bot input is solid. 2011-03-01T13:17:52 and docs? is the stuff in the wiki up to date? 2011-03-01T13:18:05 If someone could make a howto on using the engine, that would be good too. 2011-03-01T13:18:17 The wiki is a little messy still. 2011-03-01T13:18:24 But it should be up to date. 2011-03-01T13:20:11 *** janzert has quit IRC (Read error: Connection reset by peer) 2011-03-01T13:20:40 make_movie.cmd maps/random4.txt 1000 2011-03-01T13:20:44 there's your howto 2011-03-01T13:20:52 for windows :) 2011-03-01T13:21:32 *** janzert has joined #aichallenge 2011-03-01T13:23:02 any howto volunteers? 2011-03-01T13:23:33 *** boegel_ has joined #aichallenge 2011-03-01T13:24:00 McLeopold: this? https://github.com/aichallenge/aichallenge/wiki/How-to-Contribute 2011-03-01T13:24:02 it's up to date 2011-03-01T13:24:59 I was thinking more of an engine howto, that would be like this: http://ai-contest.com/starting_your_own.php 2011-03-01T13:25:15 the playgame instructions 2011-03-01T13:25:51 McLeopold: yeah, someone needs to write that, i think antimatroid was eager to do that 2011-03-01T13:25:59 he's already busy 2011-03-01T13:26:03 we need someone else 2011-03-01T13:26:10 well.. i have no idea how the engine works 2011-03-01T13:26:46 playgame.py -r 1 -o viewer -t %2 -m %1 "python bots/python/HunterBot.py" "python bots/python/LeftyBot.py" "java -jar bots/java/HunterBot.py" "java -jar bots/java/LeftyBot.py" 2011-03-01T13:26:51 how's that? 2011-03-01T13:47:37 *** helloworld has quit IRC (Quit: Leaving) 2011-03-01T13:49:03 *** choas has joined #aichallenge 2011-03-01T14:07:31 *** bhasker has joined #aichallenge 2011-03-01T14:29:55 *** bhasker has quit IRC (Quit: bhasker) 2011-03-01T14:43:03 *** jochy has joined #aichallenge 2011-03-01T14:43:15 hi 2011-03-01T14:43:26 what is GOOGLE IA Challenge? 2011-03-01T14:51:09 jochy: it's a programming contest 2011-03-01T14:51:24 about that? 2011-03-01T14:51:36 you make a bot that plays a game, on its own 2011-03-01T14:52:01 jochy: like this: http://ai-contest.com/visualizer.php?game_id=9559558 2011-03-01T14:53:27 ok 2011-03-01T14:55:53 what is the deadline? 2011-03-01T14:56:10 what is the prizze for the winner? 2011-03-01T14:56:14 jochy: there's no contest going on now actually, the one i liked you was finished in december 2011-03-01T14:56:40 jochy: you can find more info here: http://ai-contest.com/faq.php 2011-03-01T14:56:53 *** me0w has joined #aichallenge 2011-03-01T14:57:46 thanks amstan, bye 2011-03-01T14:57:51 bye 2011-03-01T15:00:51 *** SheV0 has joined #aichallenge 2011-03-01T15:01:35 *** Prillicy has joined #aichallenge 2011-03-01T15:03:29 *** Accoun has quit IRC () 2011-03-01T15:08:32 *** jochy has quit IRC (Quit: Page closed) 2011-03-01T15:17:58 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-03-01T15:30:44 *** nann has joined #aichallenge 2011-03-01T15:32:30 amstan: it looks like phpbb refuses to accept my mod, but the new version is done if you want it 2011-03-01T15:34:10 *** Apophis_ has quit IRC (Ping timeout: 260 seconds) 2011-03-01T15:36:06 Prillicy: link? 2011-03-01T15:36:27 just a sec, setting up a new page on my site to host it permanently 2011-03-01T15:36:39 put it on github 2011-03-01T15:38:15 *** Prillicy has quit IRC (Read error: Connection reset by peer) 2011-03-01T15:38:18 *** antimatroid1 has joined #aichallenge 2011-03-01T15:38:49 *** Prillicy has joined #aichallenge 2011-03-01T15:39:55 *** antimatroid has quit IRC (Ping timeout: 264 seconds) 2011-03-01T15:43:16 http://www.area-51-x.com/Downloads_Page/Our_Programs/SecurityBlanketSpamBlockerByArea-51-X.comV1.1.zip 2011-03-01T15:43:16 *** sigh has joined #aichallenge 2011-03-01T15:43:25 there ya go amstan, and my internet crashed? fuck 2011-03-01T15:44:47 we just got that fixed 2011-03-01T15:49:05 Prillicy: so.. what was changed? 2011-03-01T15:51:32 as I said, what I fixed was the large number of times it should have caught it and didn't. If you are concerned about differentiating between real spammers and fake ones, there's a change I could make to it, custom for you, to check if a user appears a certain # of times instead of my current check, if they appear at all 2011-03-01T15:51:41 but a username like sweet I'm sure will always get caught 2011-03-01T15:52:28 i disagree 2011-03-01T15:52:39 you should only filter by username if it's a really obscure username 2011-03-01T15:52:45 not something as common as sweet 2011-03-01T15:53:14 I could turn of username checks 2011-03-01T15:53:18 leaving ip & emails only 2011-03-01T15:53:32 or emails only if you really want, it would catch less, but you'd get almost no false positives 2011-03-01T15:55:04 Prillicy: can you find out how to create a user? 2011-03-01T15:55:32 *** sigh has quit IRC (Remote host closed the connection) 2011-03-01T15:55:57 in what? stopforumspam? 2011-03-01T15:56:03 Prillicy: i think i'll do it the other way.. so you can only create new users from the main site 2011-03-01T15:56:34 you mean, the admins have to create users? 2011-03-01T15:56:39 no 2011-03-01T15:56:48 i mean that they must have a normal site user to make a forum one 2011-03-01T15:57:04 your user dbs arent connected at all? 2011-03-01T15:57:17 that's trivial 2011-03-01T15:57:31 not as much as you'd think 2011-03-01T15:57:33 it's just an include.. 2011-03-01T15:57:55 I must admit, I thought it was strange I could be logged into the forums and not into the site 2011-03-01T15:58:06 not session sharing.. 2011-03-01T15:58:08 my site's "user" check uses the cookie from phpbb 2011-03-01T15:58:38 this way any spammer registration script is useless 2011-03-01T15:58:55 unless they write one for the site 2011-03-01T15:58:55 <_flag> What's wrong with captchas? 2011-03-01T15:58:58 they'll have to make new ones suited specifically for the site 2011-03-01T15:59:02 _flag: they don't work 2011-03-01T15:59:16 at all 2011-03-01T15:59:19 * 2011-03-01T15:59:23 that's why I made my mod 2011-03-01T15:59:30 <_flag> Since when? 2011-03-01T15:59:54 Zannick: if they do that, we can make it compulsory that you upload a starter package on registration 2011-03-01T15:59:58 it works great, but if you're concerned about users who actually think "sweet" is a cool username, then I don't know what to tell you, I'm sure "freepr0n" is also a spammer username 2011-03-01T16:00:03 *** mega1 has quit IRC (Remote host closed the connection) 2011-03-01T16:00:22 Prillicy: i also had the same issue with "jake" 2011-03-01T16:00:58 _flag: idk since when, but it's happening: http://ai-contest.com/forum/memberlist.php?start=1325 2011-03-01T16:01:17 January-ish 2011-03-01T16:01:40 they're all interested in things: Occupation:Consulting Interests:Museums and art 2011-03-01T16:01:57 <_flag> What if you add a hidden element to the page that must be left un-checked? That should at least defeat the non-specific ones 2011-03-01T16:02:54 _flag: i'll probably get to that, once i get the forums in git 2011-03-01T16:03:07 a good trick I heard was to leave a question to which hte correct answer was to leave the box blank 2011-03-01T16:03:18 like "please leave the following box alone and move on" 2011-03-01T16:03:22 Prillicy: that just confuses ppl 2011-03-01T16:03:33 lol 2011-03-01T16:03:36 <_flag> This is exactly the same Prillicy, except you just don't show the box 2011-03-01T16:03:53 <_flag> The computer can't "see" what elements are actually on the page, it just fills everything out 2011-03-01T16:03:55 right, but amstan isn't much of a web-coder from the sounds of things 2011-03-01T16:04:03 _flag: so... we have email registration, recaptcha, and stop forum spam scripts going.. and it's still a ton of people 2011-03-01T16:04:16 Prillicy: nah, i'm good 2011-03-01T16:04:23 :P 2011-03-01T16:04:26 <_flag> style="position: absolute; left: - 100; right: -100;" 2011-03-01T16:04:33 <_flag> It's not even coding 2011-03-01T16:04:35 Prillicy: i just hate most of the ways phpbb does stuff, so i try not to touch it 2011-03-01T16:04:41 <_flag> s/right/top 2011-03-01T16:04:42 I agree 2011-03-01T16:05:01 if you do choose to continue using mine, I recommend you switch to the new version 2011-03-01T16:05:03 yeah, i think i'll get git on the forums 2011-03-01T16:07:25 i'm disturbed that recaptcha is somehow not defeating the spam 2011-03-01T16:07:53 Zannick: all you need is to take a screenshot, put it on a porn site, then when you get a reply back, complete it 2011-03-01T16:07:54 but it's probably being defeated by captcha farms 2011-03-01T16:07:59 it only has to work 1% of the time 2011-03-01T16:08:25 amstan: it's more likely people are being paid pennies to fill out captchas 2011-03-01T16:08:30 Zannick: or that 2011-03-01T16:09:09 the porn site thing is pretty rare 2011-03-01T16:09:45 from what the recaptcha guys have said in the past 2011-03-01T16:12:30 *** Palmik has quit IRC (Remote host closed the connection) 2011-03-01T16:14:08 *** SheV0_ has joined #aichallenge 2011-03-01T16:15:30 *** SheV0 has quit IRC (Ping timeout: 250 seconds) 2011-03-01T16:15:36 *** SheV0_ is now known as SheV0 2011-03-01T16:17:40 Prillicy: i'm trying to see what stuff is specific to each forum and what stuff is general 2011-03-01T16:17:52 Prillicy: it's probably only the writable stuff and config.php, right? 2011-03-01T16:18:12 for phpbb? 2011-03-01T16:18:16 yes 2011-03-01T16:18:26 there's very little that's designed to be forum specific 2011-03-01T16:18:56 any mods input are where changes start to be made, and of course, any time you add code yourself or use a different motif 2011-03-01T16:19:14 so ys, you're probably right 2011-03-01T16:19:23 that and the db of course 2011-03-01T16:20:09 with db i won't bother 2011-03-01T16:21:55 I'm going to create a version 2.0 when I get around to it that also stores the spammer info to your database so that a) you put less load on stopforumspam.com because your db will be able to catch and kick spammers, and b) if stopforumspam.com ever shuts down, you still have a list to work with 2011-03-01T16:22:58 Prillicy: where can i find the list of folders that need to be readable again? 2011-03-01T16:23:03 Prillicy: it says during install.. but.. 2011-03-01T16:23:41 I'm not used to setting read/write permissions, all of my stuff is on an XP server 2011-03-01T16:23:49 so I have no friggin clue 2011-03-01T16:25:52 I'm not a phpbb expert. They want me to make the install edit at least 3 files for them to approve my mod 2011-03-01T16:26:10 wtf? why? 2011-03-01T16:26:48 1 for the mod, 1 for language pack so that there can be an error message in every language in case of failure, and 1 for ACP so that you can change it/turn it on/off from the Administration panel 2011-03-01T16:30:10 Prillicy: there we go: https://github.com/aichallenge/phpbb 2011-03-01T16:30:13 *** me0w has quit IRC (Quit: Quitte) 2011-03-01T16:30:45 rofl 2011-03-01T16:30:49 what's that for? 2011-03-01T16:31:00 every time i install a mod i'll commit 2011-03-01T16:31:06 nice 2011-03-01T16:31:09 that way uninstalling takes one command 2011-03-01T16:31:21 so where is your server hosted? 2011-03-01T16:31:25 server? 2011-03-01T16:31:29 what server? 2011-03-01T16:31:44 your website has to be hosted somewhere 2011-03-01T16:31:59 which website? 2011-03-01T16:32:14 let's start with whichever one hosts your forums 2011-03-01T16:32:22 which forums? i have 2 2011-03-01T16:32:28 the ones I care about? 2011-03-01T16:32:40 hosted in the same place as the main contest server 2011-03-01T16:32:46 which is..... 2011-03-01T16:32:53 not sure about details of the hoster, all i know is that i have root 2011-03-01T16:32:55 lol 2011-03-01T16:33:04 a box? a hosting service? 2011-03-01T16:33:08 207-150-202-47.aus.us.siteprotect.com 2011-03-01T16:33:12 a giant inflatable penus 2011-03-01T16:33:12 hosting service 2011-03-01T16:33:24 hmm 2011-03-01T16:33:27 odd 2011-03-01T16:33:30 how much they charge? 2011-03-01T16:33:30 contestbot: whois 207-150-202-47.aus.us.siteprotect.com 2011-03-01T16:33:30 amstan: Error: The command "whois" is available in the Internet and Network plugins. Please specify the plugin whose command you wish to call by using its name as a command before "whois". 2011-03-01T16:33:38 contestbot: internet whois 207-150-202-47.aus.us.siteprotect.com 2011-03-01T16:33:40 amstan: Error: I couldn't find such a domain. 2011-03-01T16:33:48 Prillicy: no idea, ask j3camero 2011-03-01T16:34:10 k, 'cause a while back I offered to upgrade your server and everyone told me you were hosted on a free cloud 2011-03-01T16:34:16 Prillicy: as for hypertriangle/cubicmatter, it's in my closet 2011-03-01T16:34:24 which is the wierdest fucking thing I've heard of 2011-03-01T16:34:31 hmm 2011-03-01T16:34:33 Prillicy: oh.. right.. the workers are in the cloud 2011-03-01T16:34:33 interesting 2011-03-01T16:34:40 amazon ec2 2011-03-01T16:34:53 and we're thinking of moving the main server in the cloud too 2011-03-01T16:34:57 it's cheaper 2011-03-01T16:35:05 I'm sure 2011-03-01T16:35:52 *** boegel_ has quit IRC (Quit: This computer has gone to sleep) 2011-03-01T16:35:56 Prillicy: can i have your mod in text form pls? none of that xml stuff? 2011-03-01T16:36:11 oh, sure, the xml is just cause of stupid phpbb 2011-03-01T16:36:53 Prillicy: ideally something like this: https://gist.github.com/797936 2011-03-01T16:39:15 gist updated 2011-03-01T16:39:27 though gists seem like more hassle than their worth to me atm 2011-03-01T16:39:36 maybe I just haven't had a real reason to need them 2011-03-01T16:39:47 can i just say now i'm quite against leaving peoples rankings on a resubmit 2011-03-01T16:40:01 ? 2011-03-01T16:40:04 Prillicy: ok.. use pastebin then! i want the whole addition of your mod 2011-03-01T16:40:09 Prillicy: a diff between no mod and mod 2011-03-01T16:40:15 Prillicy: not just 3 lines 2011-03-01T16:40:20 what? 2011-03-01T16:40:28 look at bix0r last time, it helped them artificially keep their ranking up which confused a lot of people when the final tournament ran and they played bots they hadn't in a while 2011-03-01T16:40:38 Prillicy: act like i don't have your mod, paste your mod somewhere! in plaintext 2011-03-01T16:40:40 I just edited the file and pasted the new one 2011-03-01T16:40:42 make people climb back up from wherever everyone has to 2011-03-01T16:40:43 I did 2011-03-01T16:40:45 it's on that gist 2011-03-01T16:40:49 is it? 2011-03-01T16:40:53 yes 2011-03-01T16:41:20 Prillicy: all i see is "if you're concerned about safe usernames/passwords, try:" 2011-03-01T16:41:39 Prillicy: do you have a gist i don't know about? 2011-03-01T16:42:21 *** entionar has joined #aichallenge 2011-03-01T16:42:37 q 2011-03-01T16:43:07 a 2011-03-01T16:43:20 z 2011-03-01T16:43:24 dudes.. 2011-03-01T16:43:31 *** Prillicy_ has joined #aichallenge 2011-03-01T16:43:40 direct link to the latest revision https://gist.github.com/797970/ad75588dae5bee7db2873f792c2ecd99a3ec08d0 2011-03-01T16:43:53 Prillicy_: there we go, thanks 2011-03-01T16:44:04 that's twice today 2011-03-01T16:44:12 what is? 2011-03-01T16:44:14 approximately 1 hour appart 2011-03-01T16:44:18 internet cras 2011-03-01T16:44:24 internets? yeah.. i hate that 2011-03-01T16:44:30 they came by the house and put a "reducer" on it to fix it 2011-03-01T16:44:42 sounds to me like they were trying to cut our bandwidth 2011-03-01T16:44:42 wtf is a reducer 2011-03-01T16:44:51 lol, "reducer" 2011-03-01T16:44:54 *** Prillicy has quit IRC (Ping timeout: 240 seconds) 2011-03-01T16:44:56 *** Prillicy_ is now known as Prillicy 2011-03-01T16:45:06 apparently it keeps from large bursts of speed followed by large drops 2011-03-01T16:45:54 amstan: my c++ bot is pretty much done, i went to push it to git last night and somehow failed 2011-03-01T16:46:14 antimatroid1: if you don't tell me the error i can't really help you 2011-03-01T16:46:53 *** Apophis_ has joined #aichallenge 2011-03-01T16:48:40 i never got an error, it just never showed up, i'll play around later 2011-03-01T16:48:45 i need to go to uni soonish though 2011-03-01T16:48:45 *** entionar has quit IRC (Quit: Page closed) 2011-03-01T16:49:04 antimatroid1: so what happened then?? what do you mean "somehow failed", what brought you to that conclusion? 2011-03-01T16:49:22 what commands did you do? 2011-03-01T16:49:47 amstan: the fact that it isn't in the repo? :P 2011-03-01T16:49:57 i went to the bots folder then did the following 2011-03-01T16:50:03 antimatroid1: did you git add, git commit, and git push? 2011-03-01T16:50:29 git add cpp 2011-03-01T16:50:30 git status 2011-03-01T16:50:30 git commit -a 2011-03-01T16:50:30 git push 2011-03-01T16:50:30 *** nann has quit IRC (Ping timeout: 250 seconds) 2011-03-01T16:50:42 now when i try to add it, it says there's nothing to add :\ 2011-03-01T16:50:46 antimatroid1: what did git push say? 2011-03-01T16:50:59 antimatroid1: actually.. type "git log" 2011-03-01T16:51:56 ergh, it says i merged something 2011-03-01T16:51:57 eek :P 2011-03-01T16:52:05 antimatroid1: can you pastebin pls? 2011-03-01T16:52:13 yeah hang on, i'm in the middle of that :P 2011-03-01T16:53:39 http://pastebin.com/7DFYjecj 2011-03-01T16:54:00 i have no idea how i managed that :P 2011-03-01T16:54:28 wtf? why does it say deleted? 2011-03-01T16:54:43 i did that, i wasn't sure whether that was meant to be private or not :P 2011-03-01T16:54:47 do you want that back in? 2011-03-01T16:54:56 yes 2011-03-01T16:55:34 http://pastebin.com/nfubfCNm 2011-03-01T16:55:40 antimatroid1: but this should work.. type git pull origin master, then git push origin master 2011-03-01T16:55:44 pastebin the outputs 2011-03-01T16:56:18 git is a terrible *first* VCS, in my opinion 2011-03-01T16:57:18 "couldn't find remote ref master" when i try to pull? 2011-03-01T16:58:02 s/master/epsilon/ 2011-03-01T16:58:11 antimatroid1: right.. 2011-03-01T16:58:18 antimatroid1: epsilon, sorry 2011-03-01T16:58:23 jmcarthur: was my first 2011-03-01T16:58:38 just epsilon? or s/master/epsilon? 2011-03-01T16:59:00 * Frontier now has Opera, Safari, Internet Explorer, Firefox and Chromium open at once in a Compiz Tab-Group. He hopes that this covers everything with > 3% user base 2011-03-01T16:59:00 s/master/epsilon means substitute master with epsilon 2011-03-01T16:59:28 yeah 2011-03-01T16:59:35 master is usually a branch name 2011-03-01T16:59:38 it isn't for us 2011-03-01T17:00:09 okay, hopefully that worked 2011-03-01T17:00:11 aichallenge: antimatroid epsilon * rc1434ce / (README.md manager/trueskill trueskill): Merge branch 'epsilon' of https://github.com/aichallenge/aichallenge into epsilon - http://bit.ly/icaieH 2011-03-01T17:00:21 bleh, that's not right :\ 2011-03-01T17:00:28 *** smellyhippy has quit IRC (Ping timeout: 240 seconds) 2011-03-01T17:00:34 amstan: doesn't mean it was a good first 2011-03-01T17:00:50 cp -r was my first 2011-03-01T17:01:34 my progression was: cp -r -> cvs -> svn -> darcs -> git -> darcs 2011-03-01T17:01:36 then cvs, then svn, then git, then hg, then p4 2011-03-01T17:02:12 now.. i wish git wouldn't worry about chmods anymore... 2011-03-01T17:02:18 amstan: http://pastebin.com/zTSEHBYs 2011-03-01T17:02:27 i wish darcs would even track chmods 2011-03-01T17:02:30 i swear it changed the wrong stuff wtf? 2011-03-01T17:02:40 antimatroid1: seems to work... ^^^ 2011-03-01T17:03:25 amstan: my bot is there now, but what's the deal with that link going to something about trueskill? 2011-03-01T17:04:01 antimatroid1: because you commited when you didn't have the latest version, so for a while there were 2 parallel versions at the same time, before you pushed, that pull merged them 2011-03-01T17:04:09 antimatroid1: https://github.com/aichallenge/aichallenge/network 2011-03-01T17:04:12 ahh okay 2011-03-01T17:04:14 i very roughly started the wiki's last night, but then went to bed, i'll do more on that later, do you want me to pretty much move all pages to wikis? 2011-03-01T17:04:34 like i started moving across even the 5 minute guide which should be reusable across competitions anyway 2011-03-01T17:05:07 antimatroid1: i don't think the 5 minute guide has anything about planet wars in there 2011-03-01T17:05:14 *** smellyhippy has joined #aichallenge 2011-03-01T17:05:14 *** smellyhippy has joined #aichallenge 2011-03-01T17:05:23 https://github.com/aichallenge/aichallenge/network 2011-03-01T17:05:23 no, that's my point, but do you want to move all the pages to a wiki format anyway? 2011-03-01T17:05:27 for a nice graph of the commits 2011-03-01T17:05:46 antimatroid1: not yet 2011-03-01T17:05:54 antimatroid1: i'm still contemplating 2011-03-01T17:06:10 antimatroid1: but feel free to make the new ones in github/wiki, they're easy to export to html 2011-03-01T17:06:39 yeah, i'll get onto that later today 2011-03-01T17:07:00 amstan: for the moment, people really ought to be able to make a starter bot with that as a guide and the example io on the specs page 2011-03-01T17:07:16 the specs page is up to date for everything except visualiser input right at the bottom 2011-03-01T17:07:59 *** sigh has joined #aichallenge 2011-03-01T17:13:13 janzert: where can i find http logs? 2011-03-01T17:19:38 somewhere near the www directory :} 2011-03-01T17:20:01 I think they are in a sibling directory of the www directory if I remember correctly 2011-03-01T17:20:18 found them in statistics, unfortunatelly my error is not getting logged 2011-03-01T17:21:49 testing... 2011-03-01T17:22:06 we're on the phone trying to demonstrate the internet issue 2011-03-01T17:32:02 *** Prillicy_ has joined #aichallenge 2011-03-01T17:34:23 *** Prillicy has quit IRC (Ping timeout: 250 seconds) 2011-03-01T17:34:27 *** Prillicy_ is now known as Prillicy 2011-03-01T17:40:17 Prillicy: k, i have your spam thing now 2011-03-01T17:40:37 https://github.com/aichallenge/phpbb/commit/f1f14800944ee1102e9543a38930da7ea2cf10d4 2011-03-01T17:43:03 *** Prillicy has quit IRC (Ping timeout: 250 seconds) 2011-03-01T17:43:46 *** mceier has quit IRC (Quit: leaving) 2011-03-01T17:44:23 *** SheV0 has quit IRC (Ping timeout: 240 seconds) 2011-03-01T17:45:36 *** antimatroid1 has quit IRC (Ping timeout: 240 seconds) 2011-03-01T17:49:29 *** Prillicy has joined #aichallenge 2011-03-01T17:49:42 *** bhasker has joined #aichallenge 2011-03-01T17:50:03 *** SheV0 has joined #aichallenge 2011-03-01T17:50:53 *** choas has quit IRC (Quit: leaving) 2011-03-01T17:53:16 *** antimatroid has joined #aichallenge 2011-03-01T17:56:23 *** SheV0 has quit IRC (Ping timeout: 250 seconds) 2011-03-01T17:56:23 *** Prillicy has quit IRC (Ping timeout: 250 seconds) 2011-03-01T18:05:43 Thinking 'bout the rankings 2011-03-01T18:06:18 Ranking players 1 to x, means that even if your bot is getting better 2011-03-01T18:06:32 you will probably be moving down in the rankins 2011-03-01T18:06:41 as the contest progresses. 2011-03-01T18:06:54 Might be better to turn that around a bit. 2011-03-01T18:07:43 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-03-01T18:09:06 what was the deal with ranking the lower of the drawing bots higher? 2011-03-01T18:09:13 i would have thought that is not what one wants to do 2011-03-01T18:10:07 No, nothing to do with ranking draws 2011-03-01T18:10:28 just if a players ranking is the number they watch to see how well they are doing 2011-03-01T18:10:44 *** Mathnerd314 has quit IRC (Ping timeout: 252 seconds) 2011-03-01T18:10:56 it's almost always going to look like they are doing worse 2011-03-01T18:11:01 as the contest goes by 2011-03-01T18:11:08 because of new bots being added 2011-03-01T18:12:08 how else are you meant to rank people? 2011-03-01T18:12:19 i see what you mean, but i'd still want to see where i'm currently ranked 2011-03-01T18:14:10 I'm wondering if we occasionally designate certain bots as benchmarks 2011-03-01T18:14:21 and when you can consistently beat it 2011-03-01T18:14:25 you go up a level 2011-03-01T18:14:48 The first couple levels could be pretty easy 2011-03-01T18:15:10 You should be able to move up two or three levels on your first day. 2011-03-01T18:17:28 what and no rankings for levels? 2011-03-01T18:17:33 i don't really like that to be honest :P 2011-03-01T18:17:57 anyway, i have to run or i'll be late for class 2011-03-01T18:20:28 no, you can still have rankings inside the level 2011-03-01T18:20:33 but turn them around 2011-03-01T18:20:43 so higher is better 2011-03-01T18:21:01 hmmm, not sure i agree that'd be better 2011-03-01T18:21:12 There can still be a global ranking, but it's just not the emphasized number 2011-03-01T18:21:51 it could give people more incremental goals to achieve 2011-03-01T18:22:18 right 2011-03-01T18:22:23 think I prefer the global ranking as well 2011-03-01T18:22:42 incremental achievements == happiness 2011-03-01T18:23:01 global ranking is surely more incremental than any levels that are set up 2011-03-01T18:23:05 we can still show benchmark bots in the global ranking 2011-03-01T18:23:38 i like that 2011-03-01T18:24:01 that way you can compare yourself relative to the benchmark bots which aren't changing 2011-03-01T18:24:03 janzert, the problem with global ranking, with 1# being best 2011-03-01T18:24:23 is that the vast majority of players are almost always moving down 2011-03-01T18:24:31 to worse 2011-03-01T18:24:34 more reason to try harder 2011-03-01T18:24:51 yes, but I think people completely understand that already 2011-03-01T18:24:54 you make a bot, and it comes in at number 4,000 2011-03-01T18:25:03 Then you make a better bot a week later 2011-03-01T18:25:08 it's better to be 100 out of 4000 than 9 out of 10 2011-03-01T18:25:10 i think that's something anyone should easily expect 2011-03-01T18:25:15 and by now you are at 4500 2011-03-01T18:25:48 your old bot would have dropped further in the meanwhile 2011-03-01T18:25:54 *nods* 2011-03-01T18:26:06 But you don't really know how much better the new was than the old 2011-03-01T18:26:08 which you'd presumably notice 2011-03-01T18:26:22 what if the main raink was your relative position 2011-03-01T18:26:24 we could also display a percentile rank if wanted, but I don't think it's a big deal. I don't remember anyone ever complaining about it, and they complained about just about everything :) 2011-03-01T18:26:42 ^^ that 2011-03-01T18:26:47 and it rubs in that you aren't gonna win. 2011-03-01T18:27:00 we have mega1 for that 2011-03-01T18:27:02 *** Mathnerd314 has joined #aichallenge 2011-03-01T18:27:04 anyway, gotta run 2011-03-01T18:27:15 laters 2011-03-01T18:38:02 *** pyro- has joined #aichallenge 2011-03-01T18:47:38 *** Prillicy has joined #aichallenge 2011-03-01T18:52:25 good freaking grief 2011-03-01T18:52:32 *** smellyhippy has quit IRC (Ping timeout: 260 seconds) 2011-03-01T18:52:42 i was thinking i had a space leak or something because my heap profile got big whever i tested it 2011-03-01T18:52:55 turns out it was just because the number of ants in the game increases by a lot 2011-03-01T18:53:29 it was never like *tons* of memory or anything, but i was thinking it could get worse with bigger inputs or something 2011-03-01T18:53:37 grr 2011-03-01T18:54:08 i extended my input file by like a million times. what do you know? completely flat 2011-03-01T18:57:12 *** smellyhippy has joined #aichallenge 2011-03-01T18:57:19 my take on the above discussion: i always thought it would have been nice to fix hunterbot to 0 and have everybody else vary around it. with elo it could have been simply a factor to add or subtract. relative values would have remained the same, and your absolute number would actually mean something 2011-03-01T18:58:11 a positive number being how much better you are than hunterbot, and a negative one being how much worse you are than hunterbot 2011-03-01T19:08:42 *** Meatkat has joined #aichallenge 2011-03-01T19:09:33 We probably don't want to hand anyone a negative score. 2011-03-01T19:09:39 That's just demoralizing. 2011-03-01T19:23:11 well, then we just can't have a static baseline then, i guess 2011-03-01T19:23:43 what if the baseline was just *really* bad? like... go left no matter what? 2011-03-01T19:26:11 that might be too bad to be meaningful at all though 2011-03-01T19:29:09 Yeah, base line could easily be a do-nothing 2011-03-01T19:29:16 or just cap at zero 2011-03-01T19:29:22 (floor) 2011-03-01T19:34:34 you mean if you're the same as or worse than the baseline you're zero? 2011-03-01T19:35:01 as long as the baseline is really bad that would be okay, but otherwise you wouldn't necessarily be able to tell you are improving until you surpass it 2011-03-01T19:36:46 yeah 2011-03-01T19:37:03 The zero level bot would be like something moving randomly 2011-03-01T19:37:09 for planetwars it initially adjusted either the mean or median player to be 2000 elo, at the point where that made the worst player less than 0 I switched it so the worst player was always 0 2011-03-01T19:37:31 and the bottom bots were the bots that just immediately crashed of course 2011-03-01T19:37:50 so random movers are way above the worst possible :) 2011-03-01T19:38:13 Yes but they should be the worst possible intelligence. :P 2011-03-01T19:38:29 One of the things concerns me about trueskill - and would need testing to check 2011-03-01T19:38:43 is that the best bots get better over the tounament 2011-03-01T19:38:46 much much better 2011-03-01T19:39:16 Wheres trueskill assumes a mostly static range of goodness and badness 2011-03-01T19:40:01 * DanielVF afk 2011-03-01T19:40:21 trueskill is intended for ranking, not for scoring 2011-03-01T19:43:09 man, i really want those graphs from the tron contest back 2011-03-01T19:43:13 those were the most motivating things ever 2011-03-01T19:43:31 anybody know what technology was behind those? 2011-03-01T19:50:19 jmcarthur: for ranking, i'd like to official list to have a skill level and a rank 2011-03-01T19:50:33 I'd like the skill level to be from 0 to 100 2011-03-01T19:50:40 rank will depend on number of entries. 2011-03-01T19:50:59 So, hunterbot, leftybot and so on can be in the list and have a skill level. 2011-03-01T19:51:46 TrueSkill will stay a positive number. 2011-03-01T19:52:51 DanielVF: I'm planning on ranking submissions, not players. A single submission will have a static skill, only varying by it's competition. 2011-03-01T19:56:50 *** dlila has joined #aichallenge 2011-03-01T19:57:22 I'll also display a +/- number over a few benchmark bots. 2011-03-01T19:57:57 So if you really need to know absolute improvement, you can check. 2011-03-01T19:58:23 McLeopold: so all submissions start at default mean and sigma? 2011-03-01T19:58:31 50 / 16.6667 2011-03-01T19:58:46 i mean resubmissions 2011-03-01T19:59:06 Yes, but we can gear the first match toward people of you last submissions level. 2011-03-01T20:00:36 jmcarthur: so, with multiple numbers being displayed (rank / skill / +- from hunter ), I'm conceding the fact that 1 number isn't going to give everyone the info they want to compare against. 2011-03-01T20:00:56 Some care about rank, some absolute skill, some relative skill. 2011-03-01T20:01:35 what might be cool is for new submissions to play against the previous submission first. might help converge on accurate ranks more quickly, at least if your new bot is an improvement on your old one 2011-03-01T20:01:42 The other thing we can do is continue to play 2 submissions for a player, so they can see the latest skill level against their last submission. 2011-03-01T20:01:54 That's a good idea. 2011-03-01T20:02:04 Maybe a 1v1 should be the test match? 2011-03-01T20:02:17 or maybe even your new bot against your n previous bots 2011-03-01T20:02:18 one game doesn't give you much information 2011-03-01T20:02:19 unless you last submission was a crash bot :) 2011-03-01T20:02:40 if it was a crash bot then it will be a fast, uninformative game ;) 2011-03-01T20:02:41 janzert: the first game gives you the biggest drop in uncertainty 2011-03-01T20:03:16 anyway, might be neat to play a new bot against your past three bots in one match or something 2011-03-01T20:03:27 it's also automatically gearing your first match against someone near your last rank... 2011-03-01T20:03:27 of course, just that a loss against your previous submission doesn't actually mean a whole lot 2011-03-01T20:03:32 get a guess of the relative rank there, then move into the main games 2011-03-01T20:03:45 janzert: but a win against it means a ton 2011-03-01T20:03:52 not really :) 2011-03-01T20:03:59 and a win against your last three means even more 2011-03-01T20:04:15 maybe not janzert's bots... 2011-03-01T20:04:18 self play is the worst way to check for improvement 2011-03-01T20:04:20 heh 2011-03-01T20:04:26 this isn't to check for improvement 2011-03-01T20:04:32 *** Mathnerd314 has quit IRC (Ping timeout: 240 seconds) 2011-03-01T20:04:32 janzert: no, that's not ture 2011-03-01T20:04:34 true 2011-03-01T20:04:44 this is to get a reasonable guess of rank going for your first match against another player 2011-03-01T20:05:06 You can miss big strategies, or not be prepared for different tactics, but you still get some idea. 2011-03-01T20:05:18 this would be particularly helpful for the top 50% of players 2011-03-01T20:05:20 Also, the previous bots will have been ranked against other people as well. 2011-03-01T20:05:22 look at go programmers, chess programmers or most any place where people spend lots of time working on game playing programs 2011-03-01T20:05:36 they all seem to agree that self play is a really bad measure 2011-03-01T20:05:38 i'm comparing this against just starting at the bottom 2011-03-01T20:05:50 if you beat the worst players, that's not helping to determine your rank much 2011-03-01T20:06:02 yeah, for sure don't start at the bottom 2011-03-01T20:06:12 at the worst case start at the median 2011-03-01T20:06:23 well actually trueskill would do that 2011-03-01T20:06:25 median 2011-03-01T20:06:31 yep 2011-03-01T20:06:54 i guess it's kind of like binary search in that sense 2011-03-01T20:07:06 we just need a more sophisticated way to display ranks of a player with multiple submissions 2011-03-01T20:07:09 but with statistical models and stuff 2011-03-01T20:07:18 McLeopold: hm? 2011-03-01T20:07:21 We aren't going to find an answer for the perfect number after a few games. 2011-03-01T20:07:26 i'm fine with resubmissions dropping your visible rank to zero 2011-03-01T20:07:44 I'm not 2011-03-01T20:07:56 if you're going to show skill as a separate number anyway, though... 2011-03-01T20:08:18 I want to see the last "real" rank of mega1 and other top contenders as their new submissions are being proven. 2011-03-01T20:08:21 oh right, we're resetting mu, too 2011-03-01T20:08:54 i really liked the graphs from tron :\ 2011-03-01T20:08:55 So, I'm not just concerned about my rank, but I want the ladder to be somewhat consistent. 2011-03-01T20:09:06 jmcarthur: do you have images to share? 2011-03-01T20:09:07 One solution would be to leave mu. Another would be to hold the old rank until a set number of matches have been played by the new submission. 2011-03-01T20:09:16 nope 2011-03-01T20:09:23 they were basically just showing your rank over time 2011-03-01T20:09:28 with drops at resubmissions 2011-03-01T20:09:41 but it still showed past ranks, so it wasn't frustrating or something 2011-03-01T20:09:47 jbroman: what I'm saying is, start the new submission at ground zero, but the rankings display can be smarter about showing ranks of past submissions. 2011-03-01T20:10:06 We don't need to find a solution to quickly ranking new bots. 2011-03-01T20:10:25 jmcarthur: my database design will have that 2011-03-01T20:10:30 McLeopold: Sorry; I meant to agree that it is a valid solution. 2011-03-01T20:10:42 ah, ok 2011-03-01T20:10:58 I understand you now. 2011-03-01T20:11:04 i'm wondering if there is any *principled* way to do it 2011-03-01T20:11:18 jmcarthur: each game records a bots before and after mu and sigma 2011-03-01T20:11:45 i mean about your rank immediately after resubmitting 2011-03-01T20:12:03 McLeopold: nice that the db has that. we can use it 2011-03-01T20:12:10 full round-robin? 2011-03-01T20:12:32 hmm? 2011-03-01T20:12:40 oh you mean of your own bots? 2011-03-01T20:12:52 that seems a bit heavy 2011-03-01T20:12:56 *** JackyAlcine has quit IRC (Ping timeout: 240 seconds) 2011-03-01T20:13:11 nope, I snarlingly meant the entire field 2011-03-01T20:13:18 ha 2011-03-01T20:14:01 how about an option to toggle showing current submissions only or *all* submissions? 2011-03-01T20:14:24 I would like that. If we can make it fast. 2011-03-01T20:14:28 and maybe if there is a run of a bunch of the same contestant's bots they get compressed into one 2011-03-01T20:15:12 or perhaps the last n submissions instead? 2011-03-01T20:15:18 for some n chosen to be efficient enough 2011-03-01T20:15:21 last n would work 2011-03-01T20:15:32 we move data to archive tables for speed 2011-03-01T20:15:54 last n would be vastly different for different people 2011-03-01T20:16:00 yeah... 2011-03-01T20:16:07 maybe it should be time instead of number 2011-03-01T20:16:12 *** Mathnerd314 has joined #aichallenge 2011-03-01T20:16:14 I had my first one live for 2.5 months 2011-03-01T20:16:14 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Read error: Connection reset by peer) 2011-03-01T20:16:18 last day's worth of submissions, say 2011-03-01T20:17:27 because surely a day is long enough to converge on a reasonable rank anyway 2011-03-01T20:17:43 maybe based on the gpm 2011-03-01T20:18:06 last 20 games * contestants 2011-03-01T20:18:29 i'm hesitant to allow things like gpm to artifically affect visible rankings like that 2011-03-01T20:19:07 I'm only suggesting bots that played in that timeframe are displayed. mu and sigma are not affected 2011-03-01T20:19:21 yeah i know 2011-03-01T20:19:56 the "full list" of submissions won't have "official" ranks 2011-03-01T20:20:09 just skill, and distance from hunter 2011-03-01T20:20:18 they will be in skill order 2011-03-01T20:20:41 well, here's an idea... how about each submission has its own stats like we've been planning, but also each player has another trueskill score. it has no affect on matchmaking or anything, but it can be used to rank players against each other 2011-03-01T20:20:55 and that score never changes even with resubmissions 2011-03-01T20:21:06 we can record the data 2011-03-01T20:21:08 well, i mean it changes for each game of course 2011-03-01T20:21:13 I'm not sure how useful it would be 2011-03-01T20:21:32 well it sounds like you are just wanting to prevent huge jumps in skill upon resubmission 2011-03-01T20:21:37 this would do that 2011-03-01T20:21:44 not really 2011-03-01T20:21:45 on a player basis, not a bot basis 2011-03-01T20:21:47 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-03-01T20:21:52 how not? 2011-03-01T20:21:58 I'm not thinking I want to prevent anything? 2011-03-01T20:22:24 you were wanting to prevent a player's rank from dropping to zero, no? 2011-03-01T20:22:27 I want the leader board to appear stable, even while resubmits are occuring. 2011-03-01T20:22:30 right 2011-03-01T20:22:41 this would make for a stable leaderboard 2011-03-01T20:22:56 but perhaps more inaccurate 2011-03-01T20:23:08 that's what you get for stability 2011-03-01T20:23:34 the player skill could perhaps be based on the last n games that the player had (recomputing the trueskill based on those or something) 2011-03-01T20:23:35 how about this, a bot highest skill level and current submission are show... 2011-03-01T20:23:52 and the non-current one has is flagged "retired" 2011-03-01T20:24:22 at some point, if the new one becomes the highest, you only display 1 2011-03-01T20:24:28 only meaningful if skill levels were absolute instead of relative 2011-03-01T20:24:44 since the retired one plays no more games and the scale can change 2011-03-01T20:24:56 the scale doesn't change 2011-03-01T20:25:27 skill is essentially percentile 2011-03-01T20:25:36 i just mean that if you're a level 95 bot today that doesn't mean the same bot should be level 95 a week from now 2011-03-01T20:25:53 exactly. it's relative 2011-03-01T20:26:06 right, but at some point, the retired submission will be removed 2011-03-01T20:26:22 it's only useful until the new bot converges 2011-03-01T20:26:59 "a bot highest skill level and current submission" <-- sounded like you don't just show the previous bot, but you show the highest ranking bot the player submitted of all time 2011-03-01T20:27:19 well, you show both, so you can see your new one climb 2011-03-01T20:27:20 i would agree with showing the most recently retired bot 2011-03-01T20:27:28 not the higher of all time 2011-03-01T20:27:30 *highest 2011-03-01T20:27:40 yes, that's what I meant 2011-03-01T20:27:43 okay 2011-03-01T20:27:46 it was a misunderstanding then 2011-03-01T20:27:52 i'm fine with this 2011-03-01T20:28:05 that's basically like my "last n" proposal 2011-03-01T20:28:08 for n = 1 2011-03-01T20:28:27 the retired bot can also be removed from the rankings, so you just see them in the right position, but they don't get a number 2011-03-01T20:28:43 yes, but only 1 of the last n? 2011-03-01T20:28:48 yes 2011-03-01T20:29:06 wait, what? 2011-03-01T20:30:18 1) McLeopold (v2) 2011-03-01T20:30:18 *) McLeopold (v1 *retired*) 2011-03-01T20:30:18 2) Bocsimako (v99) 2011-03-01T20:30:22 like that ^^ 2011-03-01T20:30:42 i was questioning what you meant by "yes, but only 1 of the last n?" 2011-03-01T20:30:47 oh 2011-03-01T20:31:09 show current bot and highest of last n bots 2011-03-01T20:31:18 ah 2011-03-01T20:31:32 yes, but i think it should be by time instead of number 2011-03-01T20:31:42 show current bot and highest of last day or so 2011-03-01T20:32:05 yeah, okay, whatever turns out to be the time for bots to converge correctly 2011-03-01T20:32:06 *** amstan has joined #aichallenge 2011-03-01T20:32:07 *** ChanServ sets mode: +o amstan 2011-03-01T20:32:10 yeah 2011-03-01T20:33:14 amstan: what technology was powering the rank-over-time graphs in the tron contest, and why were they pulled? were they just too slow? 2011-03-01T20:34:00 jmcarthur: it was just elo 2011-03-01T20:34:06 So I think this would be good. We aren't trying to solve the problem of a perfect number to start new bots on, and people can watch new submissions climb, plus see, in general, where they and their opponents will probably end up on the ladder. 2011-03-01T20:34:33 amstan: no, like the images that were generated 2011-03-01T20:34:45 (and it wasn't elo at the time, either) 2011-03-01T20:34:46 Assuming the new bot is near the old one. Which is a good assumptions. 2011-03-01T20:34:48 jmcarthur: what graphs then? the visual;izer? 2011-03-01T20:35:00 amstan: no, like... it showed your rank as time went by 2011-03-01T20:35:08 oh! 2011-03-01T20:35:15 Oh yeah. amriedle made those. 2011-03-01T20:35:20 contestbot: seen amreidle 2011-03-01T20:35:21 amstan: I have not seen amreidle. 2011-03-01T20:35:24 darn 2011-03-01T20:35:25 that guy... 2011-03-01T20:35:29 oh j3camero is here 2011-03-01T20:35:41 were they pulled because they were just too slow? 2011-03-01T20:35:43 amriedle has been here. Must have a differen nick. 2011-03-01T20:35:52 j3camero: nope 2011-03-01T20:35:57 contestbot: seen amriedle 2011-03-01T20:35:57 jbroman: amriedle was last seen in #aichallenge 3 weeks, 2 days, 22 hours, and 30 seconds ago: why would you use an iphone ... 2011-03-01T20:36:00 there 2011-03-01T20:36:01 they were very motivating. i'd love to have those back 2011-03-01T20:36:02 spelling :) 2011-03-01T20:36:05 or something like them 2011-03-01T20:36:17 jmcarthur: I pulled the rank-over-time charts because I thought they were causing trouble with overloading the database. In retrospect that probably wasn't true. 2011-03-01T20:36:30 j3camero: they won't be this time 2011-03-01T20:36:55 But, if we could have a visualizer, so we just push some data, that would be nice. 2011-03-01T20:36:56 Yeah I know. Especially with caching and such. 2011-03-01T20:36:58 it was nice because as soon as you submitted the starter, your graph started going 2011-03-01T20:37:06 and then you just *have* to improve it 2011-03-01T20:37:17 j3camero: have you looked over the issue tracker lately? 2011-03-01T20:37:21 For sure. That was the point of the graph. 2011-03-01T20:37:24 j3camero: i think we're doing good progress 2011-03-01T20:37:26 we track issues? 2011-03-01T20:37:42 McLeopold: yes 2011-03-01T20:38:02 McLeopold: /topic : Launch Checklist: https://github.com/aichallenge/aichallenge/issues/labels/launch 2011-03-01T20:38:02 jmcarthur: totally. It's visually addictive. Here's this chart that says YOU HAVE THIS MUCH E-PEEN: |--------| 2011-03-01T20:38:14 haha 2011-03-01T20:38:53 E-PEEN? 2011-03-01T20:39:02 it would also be really cool to have a few other visualizers like that. overlay graphs or something. like, maybe you want to compare yours to somebody else's 2011-03-01T20:39:26 yeah, don't you want to compare e-peens? :P 2011-03-01T20:39:32 *** JackyAlcine has joined #aichallenge 2011-03-01T20:40:44 urban dictionary to the rescue... 2011-03-01T20:40:59 Sometimes I don't want to learn something new everyday... 2011-03-01T20:43:46 amstan: are you prepared for major database changes? 2011-03-01T20:43:54 the php files will need some testing 2011-03-01T20:43:56 McLeopold: umm.. lol 2011-03-01T20:44:02 McLeopold: i can't dev today though 2011-03-01T20:44:21 hmm 2011-03-01T20:44:23 *** Prillicy_ has joined #aichallenge 2011-03-01T20:44:35 maybe I'll just break the code then... 2011-03-01T20:45:12 j3camero: do you have a working test server? 2011-03-01T20:45:29 *** Prillicy has quit IRC (Ping timeout: 276 seconds) 2011-03-01T20:45:33 *** Prillicy_ is now known as Prillicy 2011-03-01T20:46:06 McLeopold: nope. I believe it would be really helpful if someone created a script that automatically deployed the contest software. 2011-03-01T20:46:24 I'm halfway there 2011-03-01T20:46:30 Excellent. That would be HUGE. 2011-03-01T20:46:41 I'll give you what I have later tonight. 2011-03-01T20:46:46 will you be around? 2011-03-01T20:47:22 Ideally, you would kick off a freshly-installed Ubuntu inside a VM, then run inside the VM "wget some_script.sh; ./some_script.sh" and then have a complete working contest. 2011-03-01T20:47:46 I won't be very attentive tonight. I'm reading a bunch of papers about portfolio optimization. 2011-03-01T20:47:51 This would be much more fun to program with everyone in the same room. 2011-03-01T20:48:00 Can the contest money fly me out to toronto? 2011-03-01T20:48:03 IRC is the closest we can get haha. 2011-03-01T20:48:12 lol what contest money? 2011-03-01T20:48:21 McLeopold where are you located? 2011-03-01T20:48:23 ish 2011-03-01T20:48:26 Olympia 2011-03-01T20:48:43 McLeopold: you are working on a new worker deployment script or just one for a main server? 2011-03-01T20:48:45 I swear I thought I saw the number $10k floating around... 2011-03-01T20:48:52 janzert: main server 2011-03-01T20:49:01 but it is still murky in my mind 2011-03-01T20:49:09 dudes.. it's the simplest thing, takes less than 10 minutes 2011-03-01T20:49:15 just follow the readme 2011-03-01T20:49:34 ok, I'll probably give it a go to update the worker setup script later on 2011-03-01T20:49:35 amstan: sometimes you got to tinker to fully understand everything 2011-03-01T20:49:59 janzert: i've got a start on my fork of the project 2011-03-01T20:50:20 in any case the worker script needs to be fully automated so workers can be spawned automatically 2011-03-01T20:50:38 McLeopold: ok, based off the current one or something else? 2011-03-01T20:51:08 * janzert doesn't think it will be too bad to update the current one, but it's not flexible at all 2011-03-01T20:51:28 Current, ubuntu 10.10. 2011-03-01T20:51:33 still holes 2011-03-01T20:51:42 I mean the current setup script from planetwars 2011-03-01T20:52:00 https://github.com/McLeopold/aichallenge/blob/epsilon/install_server.sh 2011-03-01T20:52:08 oh, no 2011-03-01T20:52:11 that was a worker one 2011-03-01T20:52:39 this is to take a base ubuntu 10.10 server install and turn it into a main server 2011-03-01T20:52:41 how about a python installer? 2011-03-01T20:52:42 ahh, I thought you meant you had a worker one as well 2011-03-01T20:53:00 amstan: sudo apt-get install python? 2011-03-01T20:53:18 McLeopold: no, python is usually already installed on ubuntus 2011-03-01T20:53:35 oh, write the script in python? 2011-03-01T20:53:47 McLeopold: usually python libraries and stuff have a setup.py script, i think it's a standard thing 2011-03-01T20:53:47 only reason is it's easier to run shell commands from a shell script :) 2011-03-01T20:53:47 do we care about supporting windows? 2011-03-01T20:54:03 not to run a server off of I don't think 2011-03-01T20:54:12 yeah, I would still with shell scripts 2011-03-01T20:54:46 I have been tempted to use fabric though http://docs.fabfile.org/0.9.4/ 2011-03-01T20:54:58 but it's probably overkill 2011-03-01T20:55:03 janzert: actually that link is really no good. I've got a bunch of support files I haven't check in or made references to 2011-03-01T20:55:45 ok, don't worry I'm not planning on working automating the main server setup right now 2011-03-01T20:56:19 okay, I'll pull it later and try it out. 2011-03-01T20:56:26 I'm going to continue on the sql changes. 2011-03-01T20:56:38 janzert: make sure jeff can use it. :) 2011-03-01T20:56:48 *** McLeopold has quit IRC (Read error: Connection reset by peer) 2011-03-01T20:57:00 *** amriedle has joined #aichallenge 2011-03-01T20:57:14 janzert Fabric is great 2011-03-01T20:57:18 it's really low tech 2011-03-01T20:57:25 but makes things so much easier 2011-03-01T20:57:38 in that stuff stops running when something fails 2011-03-01T20:59:23 pyro-: umm 2011-03-01T20:59:27 pyro-: it's totally up to you 2011-03-01T20:59:55 i don't have access to your server? 2011-03-01T21:00:36 install mit-scheme? 2011-03-01T21:00:39 pyro-: so if you say you need version X, and it's possible to get version X on latest ubuntu, we'll get version X 2011-03-01T21:01:12 and the easier it is to get version X onto latest ubuntu the more likely it is to happen 2011-03-01T21:01:43 For sure. 2011-03-01T21:01:55 Also, it has to not cause problems for novices. 2011-03-01T21:01:57 pyro-: what's the executable called? 2011-03-01T21:02:04 janzert: if i don't have a garentee that it will happen, im not going to start writing 2011-03-01T21:02:14 amstan: http://www.gnu.org/software/mit-scheme/ run the command "mit-scheme" 2011-03-01T21:02:35 sorry, I almost certainly won't be doing it so for sure can't guarantee anything 2011-03-01T21:02:49 there's a mit-scheme-doc in the repo, but no mit-scheme 2011-03-01T21:03:00 np me neither. im looking from a garentee from the organisers not randoms lol 2011-03-01T21:03:02 pyro: any reason why people would prefer MIT Scheme over the seemingly much more popular Racket? 2011-03-01T21:03:10 that's weird 2011-03-01T21:03:11 j3camero: slow 2011-03-01T21:03:30 pyro-: amstan, j3camero qualify as organizers 2011-03-01T21:03:51 hm, my ubuntu seems to know about mit-scheme-doc but not mit-scheme 2011-03-01T21:03:57 Zannick: yeah, same 2011-03-01T21:04:17 amstan/ j3camero: you don't have to use your package manager for everything no? http://www.gnu.org/software/mit-scheme/documentation/mit-scheme-user/Unix-Installation.html 2011-03-01T21:04:30 pyro-: no, but it's nicer when it is 2011-03-01T21:05:07 gnu has mit-scheme binaries but i wouldn't necessary feel comfortable with that 2011-03-01T21:05:18 relying on archs 2011-03-01T21:05:31 I see. Here's the thing. If MIT Scheme differs from Racket in any significant way, then we probably have to go with Racket. There are two competing criteria here. One is speed for the power users, the other is the risk of infuriating incompatibilities that cause novices to give up. I am picturing a situation where a 1st year Waterloo student who has learned Racket in their classes complains "my code works on my machine, why does it throw an error and get 2011-03-01T21:05:34 debian also has the same problem: http://pkgs.org/debian-squeeze/debian-nonfree-i386/mit-scheme-doc_9.0.1-1_all.deb.html 2011-03-01T21:06:08 man this is tiring. you told me earlier if i said i needed version x you get it. i specify mit-scheme 9.0.1 and you start talking about racket. i don't give a shit what you use, but if you are going to make the decision yourself please make it and stop asking me 2011-03-01T21:06:22 2006-04-16 split off from mit-scheme package 2011-03-01T21:06:24 now two scheme implementations are really the same. scheme programmers just have to suck it up anyway. might as well go with a decent one 2011-03-01T21:06:31 *no two 2011-03-01T21:06:44 seems to imply that there should be one 2011-03-01T21:06:46 jmcarthur: exaxctly so :) 2011-03-01T21:07:47 packages.ubuntu.com has mit-scheme 2011-03-01T21:08:00 Zannick: not on amd64, sadly 2011-03-01T21:08:08 jbroman: oh, that's why! 2011-03-01T21:08:19 jbroman: aha 2011-03-01T21:08:26 that would explain it 2011-03-01T21:09:25 *** Prillicy has quit IRC (Ping timeout: 276 seconds) 2011-03-01T21:09:54 i just had my friend call it "Sheem" 2011-03-01T21:10:21 wow, lol 2011-03-01T21:10:37 I don't know much about the scheme world or various flavors of scheme, but as always I am tempted to go with whatever is most likely to be used by a novice programmer. 2011-03-01T21:10:53 We could always offer two versions. 2011-03-01T21:11:11 (e.g. .rkt -> Racket, .ss -> MIT Scheme) 2011-03-01T21:11:15 we'd need a way to differentiate, efb 2011-03-01T21:11:36 If we let people submit shell-script entries then people could submit shell scripts of the form "racket mybot.ss" or "mit-scheme mybot.ss" if they had a strong preference. I think that would be reasonable. Anyone see a problem? 2011-03-01T21:12:09 j3camero: that requires that "racket" and "mit-scheme" be valid commands 2011-03-01T21:12:30 What is the standard command for launching a scheme program? 2011-03-01T21:12:34 j3camero: I'd prefer using file extensions, or maybe some sort of Marker file. 2011-03-01T21:12:36 and that requires specific instructions for scheme 2011-03-01T21:12:42 j3camero: there isn't one 2011-03-01T21:12:49 Is ti compiled? 2011-03-01T21:12:59 j3camero: sometimes 2011-03-01T21:13:07 depends on your implementation 2011-03-01T21:13:18 j3camero: Scheme is a language only; how it's evaluated is completely implementation-dependent. 2011-03-01T21:13:19 'is why i cant start writing until someone tells me what to use 2011-03-01T21:13:38 yeah :/ 2011-03-01T21:13:53 on most modern systems you should be able to do #!/usr/bin/env scheme-script 2011-03-01T21:14:28 implementations are supposed to install a wrapper program called scheme-script which points to itself 2011-03-01T21:14:53 Okay. What I am getting is that it would be possible to have both side by side. 2011-03-01T21:15:04 but locally i always put #!/usr/bin/env ikarus 2011-03-01T21:15:27 j3camero: yeah.. but that would require run_anything.py to not actually prepent the commands with the executable 2011-03-01T21:15:30 that probably won't matter, since we don't execute scripts 2011-03-01T21:15:41 we pass them to interpreters 2011-03-01T21:15:44 j3camero: just chmod +x the file, then let the file have a #! line 2011-03-01T21:15:54 but 2011-03-01T21:15:56 Zannick: we should be executing them though 2011-03-01T21:15:58 that doesn't mean we can't 2011-03-01T21:15:59 j3camero: does this shell script idea imply that we will be able to do arbitrary binaries, too? 2011-03-01T21:16:34 Whatever is cool with me, as long as the standard file extension causes the scripts to get run as the most popular scheme implementation. If a few power users want something different, they should have to change their file extensions. Novices should get the behavior that they expect unless they do something unexpected. 2011-03-01T21:16:42 at the very least, if we do this, we also ought to allow shell scripts 2011-03-01T21:17:16 most popular being...? 2011-03-01T21:17:31 allowing binaries would sidestep this issue for any compiled language, basically 2011-03-01T21:17:40 jmcarthur: you always could submit binaries. Remember last time? Did you try submitting a python script that does os.system("./MyBot")? 2011-03-01T21:17:52 j3camero: i never did, no 2011-03-01T21:18:07 I bet anything it would have worked. I never put anything in that would forbid this. 2011-03-01T21:18:08 (exec would be nicer, and probably hook up pipes better) 2011-03-01T21:18:29 jbroman: probably yes. 2011-03-01T21:18:33 That said, we might as well officially support binary submissions if we want to do that. 2011-03-01T21:18:41 yeah, piping might prove difficult with that 2011-03-01T21:18:46 (trivial way would be to say that the extension ".bin" means binary) 2011-03-01T21:18:52 and there might be subprocess rules in place 2011-03-01T21:18:55 it's just that i'd like to at least have some official claim that it's *okay* and *supported* to do that 2011-03-01T21:19:18 j3camero: no... this is why we have starter packages.. 2011-03-01T21:19:23 jbroman: I hate the idea of officially supporting binary submissions. It means we have to commit to a certain architecture for the game servers. Maybe some day we will want to use many different architectures? 2011-03-01T21:19:29 j3camero: if they delete the #! line, that's there problem 2011-03-01T21:19:43 their* 2011-03-01T21:19:58 j3camero: That was my opposition as well. But if you're going to give tacit approval, might as well implement it properly. Otherwise say "Unsupported, at your own risk". 2011-03-01T21:20:02 amstan: okay cool that works, as long as scheme is generally interpreted instead of compiled. 2011-03-01T21:20:03 we're already trying to commit to a fixed architecture for fairness and ease of worker setup 2011-03-01T21:20:12 jbroman++ 2011-03-01T21:20:26 j3camero: scheme is not generally anything 2011-03-01T21:20:37 jbroman: unsupported at your own risk is what I want, I believe. I think that makes sense, given that we could very well switch architectures at any time. 2011-03-01T21:20:39 there is no dominant implementation 2011-03-01T21:20:39 we need to say it's officially supported or officially unsupported, not "try it if you want" 2011-03-01T21:21:34 j3camero: implementation choice is a major decision and i wouldn't want to make it without a high level of assurance that it will still work by the time the two or three month contest is over 2011-03-01T21:21:48 I think Racket and MIT Scheme is probably enough, and Racket recommends the .rkt extension anyways so there's no difficulty choosing the correct one. 2011-03-01T21:22:11 jbroman: cool. 2011-03-01T21:22:19 2011-03-01T21:22:37 (racket is a different language but it comes with a scheme compiler called plt-r6rs) 2011-03-01T21:22:41 and i just think that the benefits (no real need to support specific implementations of any compiled language) outweight the downsides (uniform architecture requirements for the duration of the contest) 2011-03-01T21:22:48 *outweigh 2011-03-01T21:23:08 racket is a derivitive of scheme 2011-03-01T21:23:26 pyro-: Most of the "newbie" people that Jeff is concerned about won't be using that – they'd want Racket proper. Otherwise, I'd think choosing some preferred Scheme implementation for .ss is fine. 2011-03-01T21:23:48 I got the impression that MIT Scheme was generally acceptable, but if plt-r6rs is better for n reasons, that could be substituted. 2011-03-01T21:23:48 yep 2011-03-01T21:24:23 jbroman: plt is slower 2011-03-01T21:24:32 Hence MIT Scheme. 2011-03-01T21:24:36 :) 2011-03-01T21:24:41 *** JackyAlcine has quit IRC (Read error: Connection reset by peer) 2011-03-01T21:24:53 anyway my point above was 2011-03-01T21:25:05 (As long as it's possible to run mit-scheme on amd64 – never tried this myself.) 2011-03-01T21:25:12 supporting racket is not to support scheme, its a scheme derivitive 2011-03-01T21:25:24 so its not like you'd have to schemes if you put mit-scheme as well 2011-03-01T21:25:28 two* 2011-03-01T21:26:53 i assume it is, given the source 2011-03-01T21:27:17 Zannick: It turns out they even provide a binary. So in any case, building a .deb is quite doable. 2011-03-01T21:27:17 i don't think we want to build from source per worker, of course 2011-03-01T21:29:41 jmcarthur: I don't think it's smart for the contest to commit to a specific architecture for the game servers. For one, it stops us from doing any cool things like decentralized game servers. I don't think it's worth the admittedly awesome advantage of letting power users compile their own entries. 2011-03-01T21:30:23 I know that in this IRC channel, filled with only power users, that is never going to be a popular stance :-) 2011-03-01T21:30:31 In practice, I think uniformity is nice. But I'd also like the freedom to say "we guarantee nothing buy what is described here". 2011-03-01T21:30:40 s/buy/but/ 2011-03-01T21:31:33 Well and suppose some organization has a large machine called corn-syrup that they want to lend the contest to bolster our games/minute throughput! 2011-03-01T21:32:20 (Supposing also, of course, that Ubuntu boots properly in the container without non-trivial fiddling.) 2011-03-01T21:32:31 If only. 2011-03-01T21:35:45 i apeart to still not have an answer. i don't know what amstan could possibly have meant when he said it was entirely up to me. perhaps life will be easier if i reduce myself to coding an entry in python :-) 2011-03-01T21:36:44 pyro-: I'm not even sure who the person that ultimately makes this decision is. If it were me, I'd say what I said above. 2011-03-01T21:36:44 *** JackyAlcine has joined #aichallenge 2011-03-01T21:37:27 I hate making promises, but it looks like we should be able to give you MIT scheme, pyro-. The caveat is that it's possible no one will feel like actually implementing said support. So you may have to clone the GitHub repo and make the changes yourself, then we will accept your patch. 2011-03-01T21:39:01 All you have to do is put compile and run rules into the two relevant scripts in the codebase. It's much more likely to happen if you send a pull request on GitHub. I love features where all I have to do is click Accept. 2011-03-01T21:40:48 yeah i've noticed you don't like making promises. in my view its perhaps not the best thing to have decisions outstanding and changes made beyond the open of any competition. that aside i'm unfortunatly not willing to work on your code (my time is going to be limited over the next few months) beyond starter packe writing. so perhaps i will look at writing an entry in python for real 2011-03-01T21:41:11 amstan or who is the owner ought to remove my name from the list of starter package writers 2011-03-01T21:41:33 pyro-: you wanna get removed? sure 2011-03-01T21:43:13 *** McLeopold has joined #aichallenge 2011-03-01T21:43:15 cheers 2011-03-01T21:49:19 pyro-: ubuntu 32bit has mit-scheme 2011-03-01T21:51:46 j3camero: i don't see why we can support 2 flavors of scheme 2011-03-01T21:52:00 I think we can, yep. 2011-03-01T21:52:25 I am going out for wings across the street. Catch y'all in a bit. 2011-03-01T21:52:44 pyro- thanks for bringing up MIT scheme. I think it's a good idea. 2011-03-01T21:53:27 *** SheV0 has joined #aichallenge 2011-03-01T21:54:17 *** Prillicy has joined #aichallenge 2011-03-01T21:54:54 pyro-: add it to the issue list 2011-03-01T21:55:57 pyro-: https://github.com/aichallenge/aichallenge/issues/labels/language%20support 2011-03-01T22:05:21 amstan: okay i will a bit later 2011-03-01T22:05:28 j3camero: not at all 2011-03-01T22:06:19 McLeopold: cool. i already run ikarus on FreeBSD on my machines though 2011-03-01T22:17:45 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-03-01T22:21:48 *** dlila has quit IRC (Quit: Leaving) 2011-03-01T22:40:27 *** amstan has joined #aichallenge 2011-03-01T22:40:27 *** ChanServ sets mode: +o amstan 2011-03-01T23:15:59 janzert: is the script done? 2011-03-01T23:22:23 I guess I have to do something about the rendering speed in IE... 0.25 FPS ain't exactly fast 2011-03-01T23:23:27 Frontier: i thought you couldn't support ID? 2011-03-01T23:23:52 *IE* 2011-03-01T23:27:28 McLeopold: Well.. that is the point. I'm using a JScript library from Google that generates a VML (~SVG from Microsoft) on the fly from the drawing commands. Surprisingly creating markup on the fly and modifying the DOM tree is not fast 2011-03-01T23:29:48 Well it was ok when I did not clear the vml each frame. but i cant really let the elements add up either. the buttons are semi transparent and maybe other stuff will be too so i should really clear each frame. adding even more complex code to decide when to clear and when not to clear would be an option ^^ 2011-03-01T23:30:23 maybe you could revert to a simple visualizer, like mine, for IE 2011-03-01T23:30:57 McLeopold: no sorry 2011-03-01T23:31:15 I got stuck on another problem at the moment 2011-03-01T23:31:24 an aichallenge problem? 2011-03-01T23:31:37 no, completely unrelated :( 2011-03-01T23:33:06 McLeopold: You are right, I'll just have to wait a day or two to realize that is the only way 2011-03-01T23:35:55 i wrote it amstan :D 2011-03-01T23:36:54 pyro-: you got a scheme bot? 2011-03-01T23:37:10 nope 2011-03-01T23:37:24 *** SheV0 has quit IRC (Quit: ChatZilla 0.9.86-rdmsoft [XULRunner 1.9.2.10/20100914130356]) 2011-03-01T23:37:30 janzert: create_matchups fails to create matchups with only 2 bots in the system 2011-03-01T23:37:35 i feel likes its about the 10th time i've said it today but i cant write until i know what im writing for :D 2011-03-01T23:37:43 *** Prillicy has quit IRC (Quit: I be bangin') 2011-03-01T23:37:48 so, what did you write? 2011-03-01T23:38:15 the issue amstan told me to :D 2011-03-01T23:39:00 pyro- I think we're all pretty much agreed that you can have your pick of either PLT or MIT scheme. The only caveat is that you might have to implement support for MIT yourself if nobody else cares enough. 2011-03-01T23:39:30 pyro-: I have a test server. I'm installing mit-scheme now. 2011-03-01T23:39:41 if you write it, I will test it. 2011-03-01T23:40:13 (9.0.1-1ubuntu1) 2011-03-01T23:41:00 You'll have to add the apt-get command to install MIT scheme to the server setup script, then also add MIT scheme support to the compile_anything.py script and run_anything.py script. 2011-03-01T23:41:07 j3camero: i haven't got time to work with caveats. McLeopold: yey :D i will write it to the wiki on github. is that page correct? 2011-03-01T23:41:27 a scheme bot? no 2011-03-01T23:41:28 j3camero: maybe i can do that mutch. soudns easy 2011-03-01T23:42:10 McLeopold: out of curiosity what was the error? 2011-03-01T23:42:16 McLeopold: is this page correct? https://github.com/aichallenge/aichallenge/wiki/Ants-Game-Specification 2011-03-01T23:42:50 pyro-: for the specs to write against? yes 2011-03-01T23:42:56 :) 2011-03-01T23:43:25 Setting 2 submissions with no provisional rank to 0 2011-03-01T23:43:25 After cutoff 2 available for pairing 2011-03-01T23:43:25 Traceback (most recent call last): 2011-03-01T23:43:25 File "aichallenge/manager/create_matchups.py", line 326, in 2011-03-01T23:43:25 main(runtime, high_buffer, low_buffer) 2011-03-01T23:43:25 File "aichallenge/manager/create_matchups.py", line 299, in main 2011-03-01T23:43:25 pairing_cutoff) 2011-03-01T23:43:26 File "aichallenge/manager/create_matchups.py", line 231, in add_matches 2011-03-01T23:43:26 p2 = choose_opponent(p1, total_ranking, recent) 2011-03-01T23:43:27 File "aichallenge/manager/create_matchups.py", line 130, in choose_opponent 2011-03-01T23:43:27 p1['submission_id'])) 2011-03-01T23:43:28 ValueError: No opponents left after removing recent, 196464 2011-03-01T23:43:30 sry for the spam 2011-03-01T23:44:21 i will write it to that page and using mit-scheme. and i will assume that will work despite the general vagueness of answers im getting 2011-03-01T23:44:42 pyro-: I'll test it, so you will know. 2011-03-01T23:46:05 okay :D 2011-03-01T23:49:14 ahh, yeah 2011-03-01T23:50:26 since they apparently already played a game against each other and the code won't repeat pairings within 50 games 2011-03-01T23:50:56 the magic number to change is line 228, that is how many games to count as recent 2011-03-01T23:51:38 it should have really been at least a constant but probably doesn't matter since you'll be throwing all that away anyway right? :) 2011-03-01T23:52:32 I haven't played any games. 2011-03-01T23:53:31 the games table is empty? 2011-03-01T23:54:00 McLeopold: do you have a magic main-server-setup script, or did you manually do a bunch of stuff to create a test server? 2011-03-01T23:54:52 oh, I see the problem 2011-03-01T23:55:05 actually the problem is that there have been no games played 2011-03-01T23:55:34 a submission without any games was forced to play it's first game against a submission that has previous games 2011-03-01T23:55:42 of course there aren't any with previous games 2011-03-01T23:56:38 so, in choose opponent, instead of raising an error, I can fall back on unacceptable opponenets. 2011-03-01T23:57:01 well, the acceptable opponents get wiped out on line 122, 123 2011-03-01T23:57:12 McLeopold: so what's going on in your fork? you going to merge? 2011-03-01T23:57:27 no, it's just a playground 2011-03-01T23:57:31 store that list comprehension in a temporary variable, then if the result is empty keep the original 2011-03-01T23:57:51 otherwise replace ranking with the new list 2011-03-01T23:59:57 that will get you one game :/