2011-01-10T00:05:05 *** tapwater has quit IRC (Quit: tapwater) 2011-01-10T01:34:14 *** Palmik has joined #aichallenge 2011-01-10T01:36:03 *** Utkarsh has quit IRC (Ping timeout: 250 seconds) 2011-01-10T01:41:04 *** Utkarsh has joined #aichallenge 2011-01-10T01:46:37 *** medrimonia2 has joined #aichallenge 2011-01-10T01:48:51 *** medrimonia has quit IRC (Ping timeout: 276 seconds) 2011-01-10T01:54:38 *** sigh has quit IRC (Ping timeout: 272 seconds) 2011-01-10T02:45:52 *** Mathnerd314 has quit IRC (Ping timeout: 240 seconds) 2011-01-10T02:53:28 *** sigh has joined #aichallenge 2011-01-10T03:11:55 *** Utkarsh has quit IRC (Ping timeout: 276 seconds) 2011-01-10T04:15:25 *** Naktibalda has joined #aichallenge 2011-01-10T04:16:59 *** perror has joined #aichallenge 2011-01-10T04:20:06 *** amstan__ has quit IRC (Ping timeout: 240 seconds) 2011-01-10T04:28:29 *** Utkarsh has joined #aichallenge 2011-01-10T05:52:14 *** mega1 has joined #aichallenge 2011-01-10T06:01:02 *** Utkarsh has quit IRC (Ping timeout: 246 seconds) 2011-01-10T06:04:00 *** Migi32 has joined #aichallenge 2011-01-10T06:05:50 *** Utkarsh has joined #aichallenge 2011-01-10T07:01:23 *** Zannick has quit IRC (Ping timeout: 276 seconds) 2011-01-10T07:11:55 *** delt0r_ has joined #aichallenge 2011-01-10T07:13:25 *** delt0r___ has quit IRC (Ping timeout: 255 seconds) 2011-01-10T07:28:25 *** Zannick has joined #aichallenge 2011-01-10T07:29:14 *** mega1 has quit IRC (Ping timeout: 246 seconds) 2011-01-10T07:30:12 *** mega1 has joined #aichallenge 2011-01-10T07:36:53 *** Zannick has quit IRC (Ping timeout: 260 seconds) 2011-01-10T07:40:17 *** amstan__ has joined #aichallenge 2011-01-10T07:40:17 *** ChanServ sets mode: +o amstan__ 2011-01-10T07:42:13 *** mega1_ has joined #aichallenge 2011-01-10T07:43:09 *** mega1 has quit IRC (Ping timeout: 276 seconds) 2011-01-10T07:45:03 *** Zannick has joined #aichallenge 2011-01-10T07:48:50 *** mega1_ has quit IRC (Quit: No Ping reply in 180 seconds.) 2011-01-10T07:50:39 *** sigh has quit IRC (Remote host closed the connection) 2011-01-10T07:51:33 *** mega1 has joined #aichallenge 2011-01-10T07:51:44 jmcarthur: I remain unconvinced that a well defined "records" format is not better 2011-01-10T07:51:53 *** Zannick has quit IRC (Ping timeout: 272 seconds) 2011-01-10T07:51:54 ie a line starting with a "g" 2011-01-10T07:51:57 is a go record 2011-01-10T07:52:10 a line starting with "f" is a fleet record 2011-01-10T07:52:33 *** mega1 has quit IRC (Client Quit) 2011-01-10T07:52:45 make it a requirement that all bots should ignore undefined records. 2011-01-10T07:53:17 and that a "line" is defined to work with unix, mac and windows linefeed/CR conventions 2011-01-10T07:53:58 *** Zannick has joined #aichallenge 2011-01-10T07:54:01 the bots are small even with parsing code... and no libs to setup for compiling or runing 2011-01-10T07:54:31 This is after atempting some JSON and other type formats for some work code btw 2011-01-10T08:10:01 i am on the fence about the format. If I recall, the idea was so game parameters could be passed to the bot. 2011-01-10T08:10:01 json/yaml may make that easier, but i am not sure it is worth the cost. 2011-01-10T08:10:01 if the target audience was more advanced, then sure. 2011-01-10T08:14:37 I don't see how using plain simple text fields makes it harder 2011-01-10T08:14:49 not text fields 2011-01-10T08:14:57 records 2011-01-10T08:15:28 I add a game init record... or records.. 2011-01-10T08:15:33 say "m" for map 2011-01-10T08:15:42 "t" for turns 2011-01-10T08:15:53 or "i" for a init record with all the details 2011-01-10T08:16:13 bots that don't have this... ignore it... and are no worse off 2011-01-10T08:16:32 *** amstan__ has quit IRC (Ping timeout: 240 seconds) 2011-01-10T08:17:06 what could be simpler... a pile of XML or JSON is not, thats for sure 2011-01-10T09:25:01 *** delt0r has quit IRC (Remote host closed the connection) 2011-01-10T09:26:42 *** delt0r has joined #aichallenge 2011-01-10T09:39:35 *** JamesMG has joined #aichallenge 2011-01-10T09:56:37 *** Migi32 has quit IRC (Quit: BOINC) 2011-01-10T10:22:28 *** Mathnerd314 has joined #aichallenge 2011-01-10T10:45:03 *** tapwater has joined #aichallenge 2011-01-10T10:52:38 *** mceier has joined #aichallenge 2011-01-10T10:57:30 *** nullkuhl_ has joined #aichallenge 2011-01-10T10:59:45 *** nullkuhl has quit IRC (Ping timeout: 240 seconds) 2011-01-10T11:30:08 what's so much harder about json? 2011-01-10T11:30:19 i've used it before and it was pretty easy 2011-01-10T11:33:38 pretty much you just pass a string to a function and get back a data structure or pass a data structure to a function and get a string 2011-01-10T11:46:39 not really 2011-01-10T11:46:59 at least in java you need to have a good idea of the "type" before hand 2011-01-10T11:47:06 otherwise you get exceptions 2011-01-10T11:47:20 exceptions? 2011-01-10T11:47:20 or you are back at traversing a DOM type parse tree 2011-01-10T11:47:31 both don't seem like fun 2011-01-10T11:47:41 parsing is no more fun 2011-01-10T11:47:49 its plain trival 2011-01-10T11:47:55 without libarays 2011-01-10T11:47:58 and still requires traversing a parse tree 2011-01-10T11:48:02 why add something and get nothing bacl 2011-01-10T11:48:18 I don't traverse any parse tree 2011-01-10T11:48:23 its a few lines of code 2011-01-10T11:48:28 and its expandable 2011-01-10T11:48:35 and not extra 3rd party libs 2011-01-10T11:48:37 no 2011-01-10T11:49:29 *** nullkuhl__ has joined #aichallenge 2011-01-10T11:49:33 you must be using much harder json libraries and on much more complicated json than i am 2011-01-10T11:49:42 I tried googles one 2011-01-10T11:49:45 and 2 others 2011-01-10T11:49:49 all were the same 2011-01-10T11:50:01 since there is no real "type" info in a JSON format 2011-01-10T11:50:05 seriously, it's just a key-value store, which is fundamentally the same as our custom format except that it doesn't require you to write you own parser and supports more types of data 2011-01-10T11:50:23 I know what it is 2011-01-10T11:50:31 and after useing it 2011-01-10T11:50:43 I am against it... 2011-01-10T11:50:57 i just don't see how parsing is so much easier than querying a key-value mapping 2011-01-10T11:51:17 It really doesn't add anything--expect a lot of extra support for starter packages since getting libs to work will be a real hassel 2011-01-10T11:52:12 also, how does the map fit into all this? do you just require that it be at the end of everything? 2011-01-10T11:52:21 no 2011-01-10T11:52:23 what? 2011-01-10T11:52:27 *** nullkuhl_ has quit IRC (Ping timeout: 264 seconds) 2011-01-10T11:52:30 what are you talking about> 2011-01-10T11:52:31 ? 2011-01-10T11:52:33 or give it its own key at the beginning o each row 2011-01-10T11:52:47 the map format is just a big chunk of text including newlines right now 2011-01-10T11:52:56 yea 2011-01-10T11:53:06 but was sugesting "records" 2011-01-10T11:53:17 lots of space catalogs have this type of thing 2011-01-10T11:53:19 so you propose we just stick a bunch of "m" keys at the beginning of each row of the map? 2011-01-10T11:53:24 works quite well 2011-01-10T11:53:44 a record starts with a char or whatever and tells you the format of the rest of the line 2011-01-10T11:53:52 ignore records you don't know about 2011-01-10T11:53:52 right. one line 2011-01-10T11:54:01 the map is more than one line 2011-01-10T11:54:13 have more that one map record 2011-01-10T11:54:21 okay that's all i was asking 2011-01-10T11:54:27 making sure i understood what you were getting at 2011-01-10T11:54:54 okay, so essentially, you're saying that a list of (char,string) pairs is better than a key-value mapping 2011-01-10T11:55:11 I just don't think using json + parsing libs adds anything 2011-01-10T11:55:18 not quite 2011-01-10T11:55:30 so in the last game 2011-01-10T11:55:33 we have fleets 2011-01-10T11:55:53 this was an "f" record... ie we now know the format of the rest of the line 2011-01-10T11:56:02 it can tell you anything. 2011-01-10T11:56:14 sure, but then you have to write a custom parser for it 2011-01-10T11:56:23 for every single type of record 2011-01-10T11:56:24 which is all of about 3 lines of code 2011-01-10T11:56:37 and does not require a 3rd party lib 2011-01-10T11:57:15 and is about the same amout of code as JSON would give anyway 2011-01-10T11:57:55 since i will still either have to querry the "keys" or have a "class" that matches what the JSON object it 2011-01-10T11:57:56 is 2011-01-10T11:58:09 it depends on the format, but if your language makes JSON so difficult to use i doubt it would only be able to do this in only 3 lines of code without it 2011-01-10T11:58:20 I tried in 3 langs 2011-01-10T11:58:31 the different libs all behave differently 2011-01-10T11:58:47 pyton was the easiest 2011-01-10T11:58:50 C was the worst 2011-01-10T11:59:09 i mean C++ 2011-01-10T11:59:52 but i am not a organizer... I am just voicing my perhaps quite strong opinion... 2011-01-10T12:00:28 And i won't have to answer all the question about getting the starter packs going either... or why jsonXXX.lib is not supported 2011-01-10T12:01:03 the whole point of the starter packs is to work out of the box. they shouldn't require anything special on part of the end user 2011-01-10T12:01:31 that is, any necessary libs should be included with the starter pack 2011-01-10T12:01:36 If the have a non standard lib dependency... good luck with that... 2011-01-10T12:01:53 yea with linking and including and compiling... 2011-01-10T12:02:16 * janzert agrees with that the ease of parsing the ad-hoc format trumps the extensibility and portability of json in this case 2011-01-10T12:02:20 it's not hard to write a one-line makefile 2011-01-10T12:02:25 two-line 2011-01-10T12:02:30 we are talking about folk you trying to code for the first time 2011-01-10T12:02:58 people coding for the first time aren't going to care about the serialization format and aren't going to write their own starters 2011-01-10T12:03:00 the very idea of a linking step is not the same as a compiling step is a new concept 2011-01-10T12:03:08 not 2011-01-10T12:03:09 no 2011-01-10T12:03:18 but they will want to compile them 2011-01-10T12:03:23 "make" 2011-01-10T12:03:27 lol 2011-01-10T12:03:31 or whatever visual studio wants you to do 2011-01-10T12:03:42 as i said... good luck 2011-01-10T12:04:45 well, i can see that there is strong enough opposition that i'm not going to win this 2011-01-10T12:04:49 jeff is against it too 2011-01-10T12:04:56 and he carries a lot of weight of course 2011-01-10T12:05:24 I carry no weight --soo feel free to ignore 2011-01-10T12:05:47 *** nullkuhl__ is now known as nullkuhl 2011-01-10T12:08:13 nor do i, for that matter 2011-01-10T12:19:11 jmcarthur: i think including our library in the starter package will work 2011-01-10T12:19:22 as long as it's easy to tell the ide what to do 2011-01-10T12:26:06 well, i've never had to make a project buildable out of the box with no dependencies on multiple platforms before, so maybe i'm not even qualified to argue for it 2011-01-10T12:27:33 its not that... 2011-01-10T12:27:35 i personally am fine either way. i kind of like writing parsers, even 2011-01-10T12:27:51 its the totally newb nature of a lot of the first time coders 2011-01-10T12:28:00 i just usually think libraries are a lot nicer 2011-01-10T12:28:10 in many ways they are 2011-01-10T12:28:11 delt0r: but if the build scripts are already set up.... 2011-01-10T12:28:18 yea-- 2011-01-10T12:28:25 there is always a better fool 2011-01-10T12:29:06 I would expect a lot of "why doesn't this compile when i use IDE XX" 2011-01-10T12:29:38 and then a lot of --why did you use lame libXX rather than awesome libYY 2011-01-10T12:30:59 I guess i am also coming from the angle that the data should be small/simple enough that manual parsing is no big deal 2011-01-10T12:32:21 well honestly i don't think i would mind parsing simplified JSON even without a lib, but most people are not me 2011-01-10T12:32:47 if its that simple I would like less {}"" etc 2011-01-10T12:32:58 however i can filter through a regex 2011-01-10T12:33:18 unless the format uses nested objects 2011-01-10T12:33:31 then the {} are significant 2011-01-10T12:33:48 again simple does not equal nested in my assumed game world 2011-01-10T12:33:59 :/ 2011-01-10T12:34:38 as i see it, simplicity is an ideal, not a criterion. that said, i also very strongly favor that ideal 2011-01-10T12:35:49 i'll just do a record-like format 2011-01-10T12:36:13 is this for tron "3D" ;) 2011-01-10T12:36:47 there's nothing 3d about it 2011-01-10T12:37:11 I am talk the piss out of Hollywood 2011-01-10T12:37:15 talking 2011-01-10T12:37:18 heh 2011-01-10T12:37:21 taking 2011-01-10T12:37:22 crap 2011-01-10T12:53:46 *** bhasker has joined #aichallenge 2011-01-10T12:55:09 so.. what we need is a json library that is one file big 2011-01-10T12:56:51 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2011-01-10T13:30:35 i'm just gonna go with a custom formet 2011-01-10T13:30:37 *format 2011-01-10T13:31:14 i'll expand slightly on the one suggested by delt0r though. allow multichar keys and stuff 2011-01-10T13:31:34 i think as long as we don't need nested data structures we'll be okay 2011-01-10T13:34:41 its something that can change to the official version easy enough 2011-01-10T13:35:38 here's what i'm toying around with looks like http://hpaste.org/paste/42914/simple_custom_format#p42930 2011-01-10T13:36:40 with lots of comments, anyway 2011-01-10T13:37:03 looks fine 2011-01-10T13:37:06 i like it 2011-01-10T13:37:07 mine is the same 2011-01-10T13:37:11 only 2011-01-10T13:37:14 map -> m 2011-01-10T13:37:48 which is a very small detail 2011-01-10T13:37:53 without comments http://hpaste.org/paste/42914/no_comments#p42931 2011-01-10T13:38:36 I have found that a "done" or "go" makes many thing much easier --even with json 2011-01-10T13:38:38 i'm conflicted about the redundant information. currently has map dimensions, but also has an explicit border around the map 2011-01-10T13:39:08 I don't think it matters if the maps are not too big 2011-01-10T13:39:25 in my ant case however the IO was a big performance bottle neck 2011-01-10T13:39:37 for maps at the 1000x1000 sizes 2011-01-10T13:39:42 eek 2011-01-10T13:39:52 well with 1000 ants ... 2011-01-10T13:39:55 what kind of buffering were you using? 2011-01-10T13:39:59 each getting its own map... 2011-01-10T13:40:10 oh 2011-01-10T13:40:18 this was where each ant was a seperate AI? 2011-01-10T13:40:24 gloabal vision would have fixed this 2011-01-10T13:40:30 more or less 2011-01-10T13:40:30 yeah 2011-01-10T13:40:49 so i think redundant is perhaps a good idea... 2011-01-10T13:40:50 *** perror has quit IRC (Quit: Bye all !) 2011-01-10T13:41:08 and pretty much the whole game state each turn... 2011-01-10T13:41:08 redundant at least means that lower level programmers can preallocate an array 2011-01-10T13:41:15 indeed 2011-01-10T13:41:26 and you can add assertions etc 2011-01-10T13:41:30 okay, then that leaves the border. necessary? 2011-01-10T13:41:31 sanity checks 2011-01-10T13:41:40 up to you 2011-01-10T13:42:01 i kind of see this the way i see null terminated strings vs. storing the length at the front of the string 2011-01-10T13:42:01 I would make it wraping... then anding a border turns off wraping... 2011-01-10T13:42:12 hmm 2011-01-10T13:42:17 but thats just because i like Tori topology 2011-01-10T13:42:23 i'm not a huge fan of wrapping 2011-01-10T13:42:28 anding->adding 2011-01-10T13:42:34 either way 2011-01-10T13:42:36 i find that it makes games harder for me to reason about 2011-01-10T13:42:44 My latest RTS does not have it 2011-01-10T13:42:59 also, i've already written most of the engine to not use wrapping and i'm lazy ;) 2011-01-10T13:43:03 but I have circular maps 2011-01-10T13:43:25 no "courner" cases... ;) 2011-01-10T13:43:40 ha 2011-01-10T13:43:54 also the whole map is know at the start --so its only sent once 2011-01-10T13:44:02 known 2011-01-10T13:45:25 the idea with this game is that the static parts of the map are known, just not the dynamic parts 2011-01-10T13:45:34 right 2011-01-10T13:46:17 i think it would be too hard if even static parts weren't known all the time 2011-01-10T13:47:10 the only thing i don't really like about this format is that the records are not commutative, but that's just being nitpicky i think 2011-01-10T13:48:08 it's better not being commutative, at least for the map part 2011-01-10T13:48:14 exactly 2011-01-10T13:48:33 but that was kind of my point. the fact that the map isn't just a single record 2011-01-10T13:49:04 well, you could do, instead of "map" on each line, "beginmap" above it and "endmap" below it 2011-01-10T13:49:08 or "startmap" 2011-01-10T13:49:13 perhaps a map "record" could include the row number 2011-01-10T13:49:16 ie 2011-01-10T13:49:27 map 3 #...........bob...............# 2011-01-10T13:49:32 this is how it is actually done in some roguelikes 2011-01-10T13:49:38 i like delt0r's idea more than Zannick's 2011-01-10T13:50:10 not sure if i like it more than just not making it commutative 2011-01-10T13:51:26 it occurs to me that whitespace insensitivity after the record type would mean you can't have a map like apocalytptic 2011-01-10T13:51:47 (its borders weren't just vertical/horizontal) 2011-01-10T13:52:09 oh but i guess you could also just stick more walls around the outside anyway 2011-01-10T13:52:11 wouldn't hurt 2011-01-10T13:53:31 for getting a core game for ideas... i would just go with whats fastest to develop 2011-01-10T13:53:54 well, you could have L-shapes and indentations, but it would be easiest to just add the square walls anyway 2011-01-10T13:53:55 I say this as the guy who has not yet released his RTS demo game 2011-01-10T13:53:56 honestly it doesn't make a big difference to me for dev time 2011-01-10T13:54:39 i'm using a parser combinator library, so it's pretty easy to do most reasonable things 2011-01-10T13:54:54 the hard part is what other people would want to do 2011-01-10T13:55:28 *** Naktibalda has joined #aichallenge 2011-01-10T13:55:53 also, i'm not just making a game here. i'm also making an engine that i plan to use for whatever game we ultimately use, and if we have a format in mind from the outset then i can build that into the engine to work for any game generically, as long as the format is used for other games as well 2011-01-10T13:56:11 roughly, at least 2011-01-10T13:56:15 right 2011-01-10T13:56:26 but... screw that, i'll just make the individual games have their own parsing 2011-01-10T13:56:40 I tried the generic code thing. 2011-01-10T13:56:57 but really the only code that got recycled was the IO process managing code 2011-01-10T13:57:02 heh 2011-01-10T13:57:12 i love generic coding in python 2011-01-10T13:57:23 but since that was still perhaps the bulk of the code... it was a win 2011-01-10T13:57:23 well, i've got a generic game running function too, which means i can also code in tournaments, network play, etc. 2011-01-10T13:57:52 so far the only IO i have is for local play though 2011-01-10T13:58:29 I found adding TCP very easy 2011-01-10T13:58:45 the turn structure makes it very easy 2011-01-10T13:59:05 instead of local processes with IO --you have sockets with IO 2011-01-10T13:59:06 *** SmkMnstr has left #aichallenge 2011-01-10T13:59:32 yeah, handles make it easy 2011-01-10T14:20:47 hmm, i just remembered another thing that might be worth looking to. we currently just give a certain amount of time on each turn. i remember we have discussed many times whether it might be good to have a total amount of time that's permissible instead of a per turn time, or perhaps start with some total amount of time available and add a little bit to it on each turn 2011-01-10T14:21:19 *total amount of time to think, that is 2011-01-10T14:22:14 so if both players start with 100 seconds of time, then if player 1 uses 10 seconds on the first turn and player two uses 0.5 seconds on the first turn, they would then have 90 seconds and 99.5 seconds respectively to use on the next turn 2011-01-10T14:28:02 the bad thing with giving just a fixed time for the full game is that every time a program locks up you have to wait the full game time before you can end the game, but some sort of hybrid might indeed be nice 2011-01-10T14:28:24 i think having a cumulative time and adding a small bit to it on each turn would be pretty good. could start with something like 10 seconds and then add something like 1 second on each turn. simple bots could just assume they have 1 second per turn and the 10 seconds could just be used to account for measurement inaccuracies 2011-01-10T14:29:19 most complex bots could try to use that extra 10 seconds more wisely 2011-01-10T14:29:23 *more complex 2011-01-10T14:29:53 and if it uses .1 second for 100 moves then locks up? 2011-01-10T14:29:54 this is how a chess ai assignment was done in college 2011-01-10T14:30:24 started with X time and received y more after each turn 2011-01-10T14:30:29 janzert: could cap it as well? 2011-01-10T14:30:36 "use it or lose it" 2011-01-10T14:31:07 since we play so many games, minimizing wasted time on locked up programs is an inordinately large concern compared to most other timecontrol situations 2011-01-10T14:31:49 how about the start time is also the maximum time? 2011-01-10T14:31:59 you never get more than 10 seconds saved up 2011-01-10T14:32:28 that could be lower, even 2011-01-10T14:32:45 but then the benefits of the system go down 2011-01-10T14:32:45 a fixed pool of 1 second * max game length time for the whole game and say 5 second limit for any one turn would work well I think 2011-01-10T14:33:02 ah yeah that could work 2011-01-10T14:33:12 or something along those lines anyway 2011-01-10T14:33:18 that may be equivalent, actually 2011-01-10T14:33:41 start with a 5 second pool, get a free second on every turn, but never save more than 5 2011-01-10T14:33:56 that never lets you use 5 seconds two moves in a row 2011-01-10T14:34:13 ah 2011-01-10T14:34:17 not equivalent, you're right 2011-01-10T14:35:07 yours may actually be better 2011-01-10T14:35:30 mine would probably encourage top bots to use more time for early/more critical moves 2011-01-10T14:35:34 mine would have a more flat profile 2011-01-10T14:35:44 so overall game times would probably be longer 2011-01-10T14:35:50 i'm not sure whether it's "better" that way 2011-01-10T14:36:25 i.e. with mine average game time is probably closer to maximum game time which can be seen as a negative 2011-01-10T14:36:31 yeah 2011-01-10T14:37:36 mine is basically good for cases where the bot needs a little extra "for emergencies" 2011-01-10T14:37:53 yep 2011-01-10T14:38:07 i'd say 10/+1/max 5 as a start, maybe 2011-01-10T14:38:23 well, starting with 10 and having a max 5 is pointless ;) 2011-01-10T14:38:36 oh 2011-01-10T14:38:40 you mean mixing *all* the ideas 2011-01-10T14:38:52 10 pool, add 1 each second, max 5 seconds per turn 2011-01-10T14:39:01 yes 2011-01-10T14:39:07 unlimited pool? 2011-01-10T14:39:16 this will end up with most bots averaging 1-2 seconds 2011-01-10T14:39:32 Arimaa has quite a nice, flexible time control system 2011-01-10T14:39:49 there should be a fixed upper limit on the pool for implementation reasons 2011-01-10T14:39:51 http://arimaa.com/arimaa/learn/matchRules.html 2011-01-10T14:39:59 but it doesn't need to be harsh 2011-01-10T14:40:01 Zannick: sounds pretty good 2011-01-10T14:41:39 1s/10s/100/5s is the Arimaa notation for that 2011-01-10T14:41:40 janzert: ah, arimaa rewards those who don't take the full time 2011-01-10T14:42:10 what's the 100? 2011-01-10T14:42:18 pool limit, i think 2011-01-10T14:42:23 1s per move / 10s initial reserve / 100 percent leftover is added to reserve 2011-01-10T14:42:29 ah 2011-01-10T14:42:30 and oops I got the last part wrong 2011-01-10T14:42:31 oh, i see 2011-01-10T14:42:55 it should be 1s/10s/100/0/0/5s 2011-01-10T14:43:39 arimaa time control looks complicated 2011-01-10T14:43:50 1s per move / 10s reserve / 100 percent leftover applied to reserve / no reserve cap / no game limit / 5s move limit 2011-01-10T14:44:15 *** bhasker has quit IRC (Ping timeout: 276 seconds) 2011-01-10T14:44:17 it seems more general than complicated 2011-01-10T14:44:22 they're quite flexible 2011-01-10T14:44:43 okay, yeah 2011-01-10T14:45:51 *** bhasker has joined #aichallenge 2011-01-10T14:46:36 well, this is an idea to add later then 2011-01-10T14:46:37 for most of the timecontrols actually used in arimaa you're only really concerned with the first 3 fields 2011-01-10T14:47:12 and the final per move limit field if it's used but that is somewhat less used 2011-01-10T14:50:19 *** sigh has joined #aichallenge 2011-01-10T15:10:03 *** delt0r_ has quit IRC (Read error: Connection reset by peer) 2011-01-10T15:12:00 *** delt0r_ has joined #aichallenge 2011-01-10T15:18:19 *** bhasker_ has joined #aichallenge 2011-01-10T15:20:01 *** bhasker has quit IRC (Ping timeout: 260 seconds) 2011-01-10T15:36:35 *** elimisteve has left #aichallenge ("Peace") 2011-01-10T15:59:11 *** dr- has joined #aichallenge 2011-01-10T16:02:26 *** sigh has quit IRC (Remote host closed the connection) 2011-01-10T16:06:46 *** bhasker has joined #aichallenge 2011-01-10T16:06:51 *** bhasker_ has quit IRC (Ping timeout: 265 seconds) 2011-01-10T16:59:55 *** sigh has joined #aichallenge 2011-01-10T17:04:17 *** mceier has quit IRC (Quit: leaving) 2011-01-10T17:08:51 *** bhasker has quit IRC (Quit: bhasker) 2011-01-10T17:42:31 *** dr- has quit IRC (Remote host closed the connection) 2011-01-10T18:21:13 *** Naktibalda has quit IRC (Remote host closed the connection) 2011-01-10T18:22:26 *** JamesMG has quit IRC (Quit: home) 2011-01-10T18:54:29 *** Palmik has quit IRC (Remote host closed the connection) 2011-01-10T19:13:25 *** amstan__ has joined #aichallenge 2011-01-10T19:13:25 *** ChanServ sets mode: +o amstan__ 2011-01-10T19:20:13 what's going on? 2011-01-10T19:20:55 *** amstan__ is now known as amstan_ 2011-01-10T19:21:39 jmcarthur: ping 2011-01-10T19:22:06 *** j3camero has joined #aichallenge 2011-01-10T19:22:11 Yooooo! 2011-01-10T19:22:27 nobody's up though, they were 2 hours ago 2011-01-10T19:22:35 jmcarthur's deving like crazy for this thing.. 2011-01-10T19:22:51 So what's this about doing Tron again? Wouldn't people not like that since we did it less than a year ago? 2011-01-10T19:22:55 delt0r and Zannick are also coding/thinking if i'm not mistaken 2011-01-10T19:23:01 Oh really? 2011-01-10T19:23:12 I didn't know that. I should be on IRC more. 2011-01-10T19:23:21 j3camero: check the logs.. that's why i keep them 2011-01-10T19:23:28 I gotta start working with them as well. 2011-01-10T19:23:35 http://contestbot.hypertriangle.com/channellogs/%23aichallenge/%23aichallenge.2011-01-10.log 2011-01-10T19:24:05 I should also ask them what they think about my proposed plan to do a stepping stone contest to give ourselves time to dev the next gen contest system. 2011-01-10T19:24:25 woah, it's that j3 guy, what's his name Jay, Jeff something like that :P 2011-01-10T19:24:27 you should post such grand plans on the forums 2011-01-10T19:24:47 Hey janzert. I hope you've been well the last few weeks. 2011-01-10T19:25:06 * amstan_ just found twizzlers on his desk, woah! 2011-01-10T19:25:14 yep, doing well thanks 2011-01-10T19:25:19 Yes amstan I should post the plan to the forums. 2011-01-10T19:25:24 beating my Arimaa bot back into submission 2011-01-10T19:25:33 j3camero: i already did phase a of the rankings algo 2011-01-10T19:25:47 amstan: nice 2011-01-10T19:25:57 janzert: how is the Arimaa going anyways? 2011-01-10T19:26:51 pretty well, really just tweaking what I got for this year since I'm working on a complete rewrite that won't be ready in time 2011-01-10T19:27:15 so what's phase A for the ranking algo? 2011-01-10T19:28:49 janzert: we need a new algo.. if we're going to implement multi player games 2011-01-10T19:28:55 by multi i mean >2 2011-01-10T19:28:58 btw, if the current backend is going to be used again I have a few pretty simple tweaks to make to help the pairing code a bit more 2011-01-10T19:29:01 right 2011-01-10T19:29:21 * janzert points at Trueskill again :) 2011-01-10T19:29:43 j3camero: yeah, trueskill seems like an interesting idea 2011-01-10T19:30:27 Was it Janzert who did an experiment with Trueskill during Galcon? 2011-01-10T19:30:37 pong 2011-01-10T19:30:38 I thought it was an epic fail. 2011-01-10T19:31:03 deltor or dmj111 ran the games from the galcon final through it recently 2011-01-10T19:31:15 *** bhasker has joined #aichallenge 2011-01-10T19:31:20 it was daniel, and the main failing was that the games weren't being scheduled optimally for trueskill, i think 2011-01-10T19:31:22 it came up with pretty similiar results as bayeselo at least at the top end 2011-01-10T19:31:30 dmj111 has run it more recently with good results 2011-01-10T19:31:30 jmcarthur: hey amstan tells me you're laying down a bunch of code. This is really cool. I'm going to do a forum post soon tonight to try to set out a roadmap for the upcoming contest. 2011-01-10T19:31:44 j3camero: cool 2011-01-10T19:32:02 i'm not necessarily intending for my code to be anything official, but if it gets picked up then i won't complain 2011-01-10T19:32:18 it's all in haskell, which would be a limiter to contributions 2011-01-10T19:32:25 What is the advantage of Trueskill then? Will it handle >2 players? 2011-01-10T19:32:34 trueskill is designed to handle that 2011-01-10T19:32:49 but the main advantage is that it's more predictable than bayeselo for a running ladder 2011-01-10T19:32:53 j3camero: trueskill was designed to rank xbox players 2011-01-10T19:32:54 Interesting. It may be just what we're looking for then. 2011-01-10T19:32:59 while still keeping much of the accuracy 2011-01-10T19:33:03 yooo 2011-01-10T19:33:19 i wouldn't use trueskill for a final tournament 2011-01-10T19:33:24 only for the dev period 2011-01-10T19:33:51 jmcarthur: well... we need an algo.. 2011-01-10T19:33:52 Okay I will refresh myself on Trueskill. I remember looking at it as one possibility back when I started using bayeselo for ranking. 2011-01-10T19:33:53 I think trueskill should be fine for the final as well 2011-01-10T19:34:18 but will take more work to use than bayeselo did 2011-01-10T19:34:28 like writing it :/ 2011-01-10T19:34:40 i have two player implemented, kind of. 2011-01-10T19:34:43 yep, that's the big thing :) 2011-01-10T19:34:48 We've been through a lot of ranking systems. What's most important is that it's widely perceived as being fair. 2011-01-10T19:34:52 multi-player, without teams, may not bee too bad. 2011-01-10T19:34:58 two player is all i care about. i hope we don't do a game for more than two players, personally 2011-01-10T19:35:06 multi-player, with teams is horrific. 2011-01-10T19:35:42 jmcarthur: ants will be interesting with >2 2011-01-10T19:35:49 j3camero: trueskill doesn't retroactively change your score, and your score gradually rises over time after submission rather than jumping around :) 2011-01-10T19:36:05 well, it can still jump around, but not as much as bayeselo 2011-01-10T19:36:08 and advantage of trueskill is that M$oft has a webpage you can use to verify that calculations, given the incoming player skill levels and the result. 2011-01-10T19:36:17 dmj111: +1, lol 2011-01-10T19:36:18 man, I cannot type. 2011-01-10T19:36:30 a big disadvantage of trueskill is lack of good implementations :\ 2011-01-10T19:36:38 who's with me for making a python implementation... or... making python bindings for a c++ implementation 2011-01-10T19:36:53 Maybe this is our chance to contribute a good quality trueskill library to the FOSS community. 2011-01-10T19:37:07 in python! 2011-01-10T19:37:09 a C implementation, if i had my druthers. easily bindable to other languages 2011-01-10T19:37:23 what's a druther? 2011-01-10T19:37:39 "I'drather it be in C" 2011-01-10T19:38:06 this guys web page probably has the most complete explanation : http://www.moserware.com/2010/03/computing-your-skill.html 2011-01-10T19:38:20 he has a link to a pdf with even a little more information 2011-01-10T19:38:31 yes, that part is probably the most important part. 2011-01-10T19:38:35 and a C# implementation. 2011-01-10T19:39:32 translate the c# one? 2011-01-10T19:39:44 well, my engine currently has 459 lines of code, according to sloccount, so i wouldn't say i've gone *too* crazy with the coding. this engine includes parallel implementations of tron and stratego 2011-01-10T19:39:48 maybe... it is definitely not written in a style I would like to emulate. 2011-01-10T19:39:59 ugh yeah, that C# code is pretty bad 2011-01-10T19:41:34 j3camero: regarding the tron game... this one is pretty different from the previous one 2011-01-10T19:41:48 i am mostly thinking at this point, no coding yet 2011-01-10T19:42:14 but i intend to when we get started 2011-01-10T19:42:20 I can take another look at the paper and C# code, and see what I think about a Python version. (probably in a style that could be easily ported to C) 2011-01-10T19:42:55 ok, not probably, but more so than the C# code. 2011-01-10T19:43:06 i'll be back later. 2011-01-10T19:43:22 j3camero: tentative example of what a game state could look like: http://hpaste.org/raw/42930/simple_custom_format 2011-01-10T19:43:28 dmj111: i want it to have a nice high level python interface 2011-01-10T19:43:29 with a bunch of comments 2011-01-10T19:43:32 that could be used to rank classes 2011-01-10T19:43:48 like, when the github repo is opened to contributors (and my account is Zannick, of course) 2011-01-10T19:43:54 jmcarthur: what is that? 2011-01-10T19:44:07 Zannick: you can dev on your own too, we can pull 2011-01-10T19:44:07 I guess if we used trueskill for multiplayer games we would have the option to not subtract 3 sigmas from people's scores. It seems like subtracting 3 sigmas would make people think that their bot earns points just by playing games. People may also come to believe that re-submitting your bot wipes out all your points. 2011-01-10T19:44:12 amstan_: tron 2011-01-10T19:44:31 yeah, but i'm less inclined to work on it while i have other things to work on :) 2011-01-10T19:44:34 jmcarthur: sorry.. i meant to say what file format is that? yaml? 2011-01-10T19:44:39 amstan_: custom 2011-01-10T19:45:03 j3camero: i agree 2011-01-10T19:45:19 jmcarthur: looks very cool. I like the multi-agent idea. People have been asking for a multi-agent game. 2011-01-10T19:45:23 amstan_: here's a YAMLy version http://hpaste.org/raw/42914/yaml 2011-01-10T19:45:27 jmcarthur: zomg, lasttime! that's interesting 2011-01-10T19:45:35 amstan_: i have a ton of ideas :) 2011-01-10T19:46:02 amstan_: a page full of a few versions: http://hpaste.org/42914/yaml 2011-01-10T19:46:15 suggestion for everyone, if you want things pasted that are slightly important, you can use github gists, they're git managed as well 2011-01-10T19:46:46 that makes me think of one thing I think should possibly change for the next contest, along with simple functional tests for submissions, pairing should not use least recently played submission but least recently played user (i.e. resubmission shouldn't bump you to the head of the line) 2011-01-10T19:46:48 i'll roll with that for the sake of convention 2011-01-10T19:47:19 janzert: agreed 2011-01-10T19:47:54 we had a nicer conversation earlier about timing 2011-01-10T19:48:05 we need a place to share ideas, while we can all edit at the same time, a wiki would be ideal, but.. we don't have one at this point 2011-01-10T19:48:26 maybe gists can do that... 2011-01-10T19:48:40 i'm thinking something like each bot gets a 10 second "reserve" to start with, on each turn you gain 1 second to add to the reserve, and you may use up to 5 seconds of your reserve per turn 2011-01-10T19:49:21 or whatever concrete numbers we end up with 2011-01-10T19:49:42 janzert and Zannick both contributed to the idea a lot 2011-01-10T19:50:20 s/nicer conversation/nice conversation/ 2011-01-10T19:51:18 can you guys test this? https://gist.github.com/773794 2011-01-10T19:51:34 yay or nay? 2011-01-10T19:51:54 i can comment... 2011-01-10T19:52:04 or fork 2011-01-10T19:52:06 yeah, but no editing.. 2011-01-10T19:52:15 can you merge? hmm? 2011-01-10T19:52:22 i'll fork and we'll see 2011-01-10T19:52:51 i'm thinking a forum thread in the behind the scenes section might be a better idea 2011-01-10T19:52:58 i don't see a way to merge aside from pulling it down with git 2011-01-10T19:53:10 i've forked and edited it here: https://gist.github.com/773795 2011-01-10T19:53:23 i agree on the forum thread(s) idea 2011-01-10T19:53:46 yeah... ok.. 2011-01-10T19:53:58 everyone here, if you don't have access to behind the scenes, here's your chance 2011-01-10T19:55:57 jmcarthur, janzert: http://www.ai-contest.com/forums/viewtopic.php?f=20&t=1237 2011-01-10T19:56:22 j3camero: where's those ants? i wanna see them 2011-01-10T19:59:35 I was reading about Trueskill 2011-01-10T20:00:26 I think I'll go for a run. When I'm back I will hack out the Ants simulation and try to crank out a video before the end of the night. 2011-01-10T20:01:10 So two things I want to get done tonight: (1) produce a video of the ants simulation and (2) post a roadmap for the next two contests to the forums. 2011-01-10T20:01:58 i would prefer the latter over the former if you have to choose 2011-01-10T20:03:47 Okay I'll do that first. But I am going for a run now. 2011-01-10T20:04:19 I am excited that there are people around who want to help in the initial stages! This is so cool! 2011-01-10T20:05:28 j3camero: it's all about the time after the contest.. there shouldn't be huge breaks in between 2011-01-10T20:05:37 hear hear 2011-01-10T20:05:46 well, at least not huge *silent* breaks 2011-01-10T20:06:03 there should always be something happening 2011-01-10T20:06:25 jmcarthur: so.. what's your opinion on django and the move to git? 2011-01-10T20:06:45 moving to a distributed VCS is a *must*, IMO 2011-01-10T20:06:53 Agreed. 2011-01-10T20:07:00 Also Django seems to be awesome. 2011-01-10T20:07:15 which one doesn't matter quite as much to me. i have my preferences (darcs), but git would be fine 2011-01-10T20:07:21 django beats php 2011-01-10T20:07:29 i don't really like most web frameworks 2011-01-10T20:07:38 s/django/python/ 2011-01-10T20:07:46 the fact that it's django doesn't matter much 2011-01-10T20:07:48 *** AlliedEnvy has quit IRC (Read error: Connection reset by peer) 2011-01-10T20:09:03 also, if my engine is to be supported in any *official* capacity, i'm going to insist on darcs since i'm already using it. but like i said, i find this scenario doubtful due to my language choice, and i don't really care ;) 2011-01-10T20:09:57 time to make dinner 2011-01-10T20:10:13 jmcarthur: that's not what i meant by small ideas, i meant that one liner that you said, same with zannick's 2011-01-10T20:10:40 the language has nothing to do with using it or not.. and aren't there darcs-git converters? 2011-01-10T20:10:54 i know the hg-git one is almost transparent 2011-01-10T20:11:15 *** bhasker has quit IRC (Quit: bhasker) 2011-01-10T20:12:36 @repo 2011-01-10T20:12:37 amstan_: repo = http://code.google.com/p/ai-contest/ 2011-01-10T20:12:42 *** bhasker has joined #aichallenge 2011-01-10T20:14:12 * amstan_ is converting the repo to git 2011-01-10T20:15:31 *** bhasker has joined #aichallenge 2011-01-10T20:28:00 janzert: i need your name and email address from github, so i can put you in the authors file 2011-01-10T20:38:55 dmj111: do you have a github account? 2011-01-10T20:43:16 also.. who's jokeserver 2011-01-10T20:43:20 me 2011-01-10T20:45:00 Zannick: you did commit, i need Full name that's valid on github 2011-01-10T20:45:16 i need to match the svn authors with those 2011-01-10T20:45:30 who's javache, did he ever svn commit? 2011-01-10T20:46:21 probably not 2011-01-10T20:46:22 do you actually need email? i have a github account https://github.com/zannick 2011-01-10T20:46:58 yeah, ideally one that github knows of 2011-01-10T20:47:06 or one that you can actually verify 2011-01-10T20:47:19 i think i verified my email with github 2011-01-10T20:47:32 i still need it, so i can make your commits by you 2011-01-10T20:47:37 that's how github maps commits to users 2011-01-10T20:47:40 ah 2011-01-10T20:47:41 by the email address in the commit 2011-01-10T20:47:57 i got that wrong once and half my project was made by "invalid email address" 2011-01-10T20:48:14 sent email address in pm 2011-01-10T20:48:43 so.. this won't be very private mind you 2011-01-10T20:48:58 yeah, well :/ 2011-01-10T20:49:03 anybody who clones the repo will see it 2011-01-10T20:49:10 my public repositories on bitbucket have my email in the clear 2011-01-10T20:49:16 ok 2011-01-10T20:49:40 should i keep your full name as Zannick or the real one? 2011-01-10T20:49:41 i'm not sure whether i have to put the +github in order to match in both my gitconfig or not 2011-01-10T20:49:51 hm. 2011-01-10T20:49:54 it should be the same one from gitconfig.. 2011-01-10T20:50:23 yeah, my gitconfig will contain Benjamin S Wolf so use that 2011-01-10T20:50:28 ok 2011-01-10T20:50:46 well.. i think that's all 2011-01-10T20:51:35 http://code.google.com/p/ai-contest/people/list -> https://gist.github.com/773851 2011-01-10T20:51:39 anyone that i missed? 2011-01-10T20:51:51 i mostly want to get the ppl that commited the most 2011-01-10T21:02:40 janzert@janzert.com is the more common address for me and the one github knows 2011-01-10T21:03:15 i had you as gmail 2011-01-10T21:03:18 ok.. 2011-01-10T21:03:24 the only one left is dmj 2011-01-10T21:03:26 i have his email 2011-01-10T21:03:31 i should get a name though 2011-01-10T21:03:37 should i just leave it as dmj? 2011-01-10T21:04:29 I imagine he'll give you a reply by morning 2011-01-10T21:05:38 i wanna finish this by then 2011-01-10T21:09:28 hey... just saw this. 2011-01-10T21:09:44 I don't have github yet... I'll sign up now. 2011-01-10T21:10:03 dmj111: use your dave@ email 2011-01-10T21:10:23 and tell me a full name to use for the git commits 2011-01-10T21:10:28 it can be dmj if you want 2011-01-10T21:13:36 amstan_: what one-liner? 2011-01-10T21:13:53 jmcarthur: your timeout stuff 2011-01-10T21:14:01 oh i see 2011-01-10T21:14:12 and janzert's new submission ranking stuf 2011-01-10T21:14:15 dmj is fine with me, but if you are using full names its Dave Jones. 2011-01-10T21:14:31 depends how you want to appear in your git log 2011-01-10T21:14:37 should be teh same as your gitconfig setting 2011-01-10T21:14:44 amstan_: there are indeed (very crappy) darcs-git converters 2011-01-10T21:15:12 amstan_: i will probably put my name in, not initials. 2011-01-10T21:16:28 *** nullkuhl has quit IRC (Read error: Connection reset by peer) 2011-01-10T21:16:47 i just don't really want to move away from darcs because (1) i'm selfish and (2) there's not really any reason to, as they are both DVCSs anyway, which is the most important thing 2011-01-10T21:18:07 i mean, it's not something i'd cry about. i just think it's silly and pointless, and it would *kind of* slow me down for a while as i adjust to using git 2011-01-10T21:18:22 and it would force me to use a different development model than i'm used to 2011-01-10T21:21:23 amstan_: the hg-git one is nearly transparent because hg and git are very similar ;) 2011-01-10T21:21:32 yeah... 2011-01-10T21:21:43 grrrr 2011-01-10T21:21:45 Author: AlexRiedler not defined in ../authors file 2011-01-10T21:23:08 who's andymatuszewski? 2011-01-10T21:24:24 oh well @gmail.com it is 2011-01-10T21:25:08 j3camero: i think the confidence interval stuff is a good reason to use it. It would add incentive to let your bot play a while, instead of add incentive to resubmit when you lose. 2011-01-10T21:25:36 but, there would need to be an explanation, I guess... 2011-01-10T21:25:43 *** bhasker has quit IRC (Ping timeout: 250 seconds) 2011-01-10T21:26:58 i don't think any deep explanation would be necessary. people expect their visible scores to drop when they resubmit anyway 2011-01-10T21:34:15 amstan_: okay i deleted my other post and wrote a new one 2011-01-10T21:34:27 jmcarthur: good man 2011-01-10T21:38:58 i'm up to r 400 2011-01-10T21:39:09 still stopping from time to time to add someone to the list 2011-01-10T21:39:53 just added another idea to my post that i hadn't mentioned here before 2011-01-10T21:39:56 i hope it'll fit on github 2011-01-10T21:40:06 shouldn't be anyone new past that I don't believe 2011-01-10T21:40:50 most of them are just me iterating through problems :} 2011-01-10T21:41:09 janzert: put your idea too 2011-01-10T21:41:55 yeah, I'm planning on posting a few tomorrow 2011-01-10T21:42:13 oh god.. how many maps do we have.. 2011-01-10T21:42:15 gees 2011-01-10T21:42:21 omg, it's done 2011-01-10T21:42:26 hehe 2011-01-10T21:42:39 68MB 2011-01-10T21:42:43 the whole git repo 2011-01-10T21:42:56 before git gc 2011-01-10T21:43:47 there is so much game specific stuff in there though I almost wonder if you don't want to try and strip out the game neutral stuff and start a new repo for the next contest 2011-01-10T21:43:49 aaaand i keep adding more 2011-01-10T21:43:57 18MB 2011-01-10T21:44:07 18MB is fine.. 2011-01-10T21:44:13 I guess 18MB of legacy isn't bad 2011-01-10T21:44:19 yeah, no 2011-01-10T21:44:27 and like.. we'll get the pretty graphs and everything as well 2011-01-10T21:45:30 hmm 2011-01-10T21:45:35 only one branch?? how does this work? 2011-01-10T21:45:55 what did you use to convert? 2011-01-10T21:46:17 git svn 2011-01-10T21:46:31 if you used git-svn you have to tell it what subdirectories branches and tags were supposed to be in 2011-01-10T21:46:37 git svn clone 2011-01-10T21:46:38 i did 2011-01-10T21:46:48 i used the -s flag 2011-01-10T21:46:59 git svn clone https://ai-contest.googlecode.com/svn/ -s --authors-file ../authors 2011-01-10T21:47:17 *** j3camero has quit IRC (Read error: Connection reset by peer) 2011-01-10T21:47:36 did you use -T, --t, and --b? 2011-01-10T21:48:01 *** j3camero has joined #aichallenge 2011-01-10T21:48:10 git svn clone -T --t= --b= 2011-01-10T21:48:32 Okay roadmap for the next two contests is posted. 2011-01-10T21:48:43 It's just my opinion, it's not binding law or anything. 2011-01-10T21:48:56 "-s is for --stdlayout which presumes the svn recommended layout for tags, trunk, and branches." 2011-01-10T21:49:06 People can feel free to tell me to buzz off and keep doing what theyre doing. 2011-01-10T21:49:22 woah, codenames! 2011-01-10T21:50:46 oh i misread the manpage then 2011-01-10T21:51:06 jmcarthur: oh well.. w/e, it's not like we had any worthwhile branches 2011-01-10T21:51:26 amstan: you really, really want the worker code branch 2011-01-10T21:51:38 that's where all the actually used engine code is 2011-01-10T21:51:46 ugh.. ok 2011-01-10T21:51:54 how do i get it then? lol 2011-01-10T21:52:00 http://www.kernel.org/pub/software/scm/git/docs/git-svn.html 2011-01-10T21:52:06 i already used -s, didn't like me 2011-01-10T21:52:24 * janzert shrugs 2011-01-10T21:52:36 lemme just give it a shot myself 2011-01-10T21:52:48 I have almost no experience with git nevermind converting an svn repo into it :/ 2011-01-10T21:53:09 look at that page, grep for If more than one --branches (or --tags) option was given to the init or clone command, you must provide the location of the branch 2011-01-10T21:53:10 doing that 2011-01-10T21:53:23 wait.. nvm 2011-01-10T21:53:33 i'm confused 2011-01-10T21:54:02 * janzert is the guy with DVCS experience in inverse proportion to popularity 2011-01-10T21:54:17 janzert: so which one do you use? 2011-01-10T21:54:21 well, at least for the big 3 2011-01-10T21:54:29 I don't go off the deep end to darcs :P 2011-01-10T21:54:41 the big three are git, hg, and bazaar? 2011-01-10T21:54:48 bzr more then hg more then git 2011-01-10T21:54:53 jmcarthur: yeah 2011-01-10T21:54:54 heh, darcs isn't even the least popular :P 2011-01-10T21:55:05 csv? 2011-01-10T21:55:08 cvs* 2011-01-10T21:55:08 yeah that's true 2011-01-10T21:55:10 it's probably number 4 2011-01-10T21:55:20 amstan_: the D is for distributed, which CVS is not ;) 2011-01-10T21:55:21 cvs isn't a DVCS 2011-01-10T21:55:40 although I did use it 15 years ago or so 2011-01-10T21:55:40 yeah, i was just trying to troll 2011-01-10T21:56:02 amstan_: git-svn is reporting that it's pulling branches, for me 2011-01-10T21:56:24 and actually I still type DVCS as DCVS half the time because of that 2011-01-10T21:56:26 jmcarthur: same here.. but it didn't make any git branches 2011-01-10T21:56:36 okay i'll just wait until the end then 2011-01-10T21:56:48 it takes forever 2011-01-10T21:56:50 i pointed out all the paths manually instead of using -s 2011-01-10T21:56:59 amstan: hacking tomorrow? 2011-01-10T21:57:00 yeah i'm only on r142 right now 2011-01-10T21:57:09 svn is so sloooow 2011-01-10T21:57:13 amstan: gchat me to set it up so I get the message when I come back. 2011-01-10T21:57:28 * janzert is off to bed, have fun converting :) 2011-01-10T21:57:56 j3camero: yes, let's meet 2011-01-10T21:58:38 janzert: did you see the proposal? 2011-01-10T21:58:42 to the coffee shop with the mad cute barista? 2011-01-10T21:59:15 @topid remove 5 2011-01-10T21:59:16 amstan_: User error, it's not my fault. 2011-01-10T21:59:23 @topic remove 5 2011-01-10T21:59:23 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || Congratulations to bocsimacko for first place! || Next Contest sometime in February" 2011-01-10T21:59:29 @topic remove 4 2011-01-10T21:59:30 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || Congratulations to bocsimacko for first place!" 2011-01-10T21:59:40 @topic add Roadmap for the next two contests: http://www.ai-contest.com/forums/viewtopic.php?f=20&t=1238 2011-01-10T21:59:40 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || Congratulations to bocsimacko for first place! || Roadmap for the next two contests: http://www.ai-contest.com/forums/viewtopic.php?f=20&t=1238" 2011-01-10T22:00:07 the roadmap? haven't read it yet no, will check it out in the morning 2011-01-10T22:00:45 but I don't expect to be very active for the next one anyway so just ignore me for now ;) 2011-01-10T22:01:01 janzert: ok.. so i have access to the branches now.. what do we need? 2011-01-10T22:01:30 -in-the-cloud or something like that 2011-01-10T22:01:37 janzert: sure 2011-01-10T22:01:56 that is the code that was actually running on the workers 2011-01-10T22:02:02 janzert: 20100929-games-in-the-cloud, can i rename it? 2011-01-10T22:02:11 sure :) 2011-01-10T22:02:21 maybe i could even merge it.. 2011-01-10T22:02:56 jmcarthur: it's all good, i can upload 2011-01-10T22:03:04 the download script needs to be updated to at the very least use git anyway 2011-01-10T22:03:05 okay cool 2011-01-10T22:03:29 mine got them all, btw 2011-01-10T22:03:44 jmcarthur: i forgot to -a or -r 2011-01-10T22:03:46 so i don't know what you did wrong :P 2011-01-10T22:03:51 oh, on git branch? 2011-01-10T22:04:02 jmcarthur: yep.. didn't realize they were repote 2011-01-10T22:04:06 remote* 2011-01-10T22:04:19 amstan_: yeah. when you push to github i think you would need to add --all to it 2011-01-10T22:04:33 jmcarthur: hmm? 2011-01-10T22:04:40 amstan_: it won't push remotes by default 2011-01-10T22:04:44 because, well, they're remotes 2011-01-10T22:05:03 actually just merging the backend directory of the worker branch into a worker directory of trunk would probably be fine 2011-01-10T22:05:59 amstan_: http://stackoverflow.com/questions/1584675/converting-svn-to-git-how-to-get-the-branches-not-to-be-just-remote-in-the-svn-r 2011-01-10T22:06:00 i'll keep 2 branches for now 2011-01-10T22:06:01 pw 2011-01-10T22:06:03 and pw/worker 2011-01-10T22:06:18 jmcarthur: i don't want to pollute the git with all the svn branches 2011-01-10T22:06:30 * jmcarthur shrugs 2011-01-10T22:06:32 alright 2011-01-10T22:08:22 i find myself not liking many of the ideas j3camero listed under Zeta, including the ones that i know have been talked about before. i disagree with having multiple contests at a time and allowing users to upload new games. and i find the idea to allow users to develop their bots in the browser kind of frivolous. 2011-01-10T22:08:35 pw-worker 2011-01-10T22:09:00 jmcarthur: yeah.. he has some wild ideas that might not work 2011-01-10T22:09:02 i mainly think any from-scratch effort should be to make what we already have awesome and more featureful 2011-01-10T22:09:29 and stable, well-specified, etc. 2011-01-10T22:09:57 another thing we were orgasming over was git repo integration.. 2011-01-10T22:10:02 multiple contests at once sounds like a great way to make each contest individually meaningless 2011-01-10T22:10:12 make the contest linked to a git repo's branch 2011-01-10T22:10:17 and each commit is a submission 2011-01-10T22:10:20 ... integration? you mean like, git push to the server and it uses your code? 2011-01-10T22:10:23 yep! 2011-01-10T22:10:24 *each* commit? 2011-01-10T22:10:33 well.. from a specific branch 2011-01-10T22:10:36 ah 2011-01-10T22:10:48 well, i don't think the idea is all that bad, but i think it sounds frivolous too 2011-01-10T22:11:18 it would be easier to just include a script with the start packs to automatically zip and upload your bot 2011-01-10T22:11:22 *starter 2011-01-10T22:11:43 so you don't have to navigate through the web site and stuff to do it 2011-01-10T22:13:21 hmm 2011-01-10T22:13:29 so.. i pushed everything 2011-01-10T22:13:33 and it finished in like a second 2011-01-10T22:13:48 oh, that's why: Writing objects: 100% (3221/3221), 5.64 MiB | 5108 KiB/s, done. 2011-01-10T22:13:52 5MB/s 2011-01-10T22:13:55 this internet's weird 2011-01-10T22:14:18 and cia didn't do anything either 2011-01-10T22:14:26 * amstan_ kicks CIA-6 2011-01-10T22:14:26 ow 2011-01-10T22:16:09 amstan_: when to meet tomorrow? 2011-01-10T22:16:17 j3camero: second cup 2011-01-10T22:16:38 when? 2011-01-10T22:16:49 5:20? 2011-01-10T22:18:08 5:45 is good for me 2011-01-10T22:18:25 I may show up earlier though 2011-01-10T22:18:32 But 5:45 for sure. 2011-01-10T22:18:43 ok 2011-01-10T22:18:44 Imma watch a movie with the gf now. See y'all later. 2011-01-10T22:18:51 i guess i'll just hang out at the office till then 2011-01-10T22:18:52 cya 2011-01-10T22:19:11 Instead, you could hang out at the shop and leer at the cute staff. 2011-01-10T22:19:14 *** j3camero has quit IRC (Quit: leaving) 2011-01-10T22:19:25 seems to work: https://github.com/amstan/aichallenge/commits/pw 2011-01-10T22:19:31 janzert's in a lot of commits 2011-01-10T22:19:34 jmcarthur: cool eh? 2011-01-10T22:21:33 github is pretty awesome 2011-01-10T22:22:31 the impact graph is not updating though.. 2011-01-10T22:23:02 aichallenge: Alexandru Stan master * rf55d91a / django/settings.py.sample : added blank settings.py file - http://bit.ly/eYCOyV 2011-01-10T22:23:02 aichallenge: Alexandru Stan pw * r06a0128 / authors : imported the svn repo into git, authors file included (+383 more commits...) - http://bit.ly/dOugrN 2011-01-10T22:23:02 aichallenge: Janzert pw-worker * r7ece9fa / (75 files): 2011-01-10T22:23:02 aichallenge: Second set of maps for finals 2011-01-10T22:23:02 aichallenge: git-svn-id: https://ai-contest.googlecode.com/svn/branches/20100929-games-in-the-cloud@639 9d0e4ec5-6d14-e29b-3149-cabf4038ef1e (+55 more commits...) - http://bit.ly/fWD1Xn 2011-01-10T22:23:29 woot! cia finally said something 2011-01-10T22:23:44 383 more commits... woot 2011-01-10T22:29:42 *** amriedle has joined #aichallenge 2011-01-10T22:30:40 *** closedbracket has joined #aichallenge 2011-01-10T22:32:43 *** acieroid` has joined #aichallenge 2011-01-10T22:35:43 *** amriedle has quit IRC (Ping timeout: 276 seconds) 2011-01-10T22:36:36 forever alone!!! https://github.com/amstan/aichallenge/graphs/impact :'( 2011-01-10T22:37:03 *** amriedle has joined #aichallenge 2011-01-10T22:38:46 Zannick: your account doesn't seem to be liked, add it to your github emails 2011-01-10T22:41:09 *** acieroid has quit IRC (*.net *.split) 2011-01-10T22:41:09 *** pgpaskar_ has quit IRC (*.net *.split) 2011-01-10T22:41:09 *** UndiFineD has quit IRC (*.net *.split) 2011-01-10T22:43:37 *** pgpaskar_ has joined #aichallenge 2011-01-10T22:43:37 *** UndiFineD has joined #aichallenge 2011-01-10T23:11:27 *** jmcarthur has quit IRC (Quit: WeeChat 0.3.2) 2011-01-10T23:12:31 *** delt0r___ has joined #aichallenge 2011-01-10T23:13:35 amstan_: done 2011-01-10T23:13:42 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-01-10T23:13:53 you guys all have commit access btw 2011-01-10T23:14:59 Zannick: yep: https://github.com/amstan/aichallenge/commit/24a4904efec57854d92a475ffa7f219edbe8f4db 2011-01-10T23:16:57 *** jmcarthur has joined #aichallenge 2011-01-10T23:17:35 @forget repo 2011-01-10T23:17:35 amstan_: Ready to serve. 2011-01-10T23:17:37 @forget repository 2011-01-10T23:17:37 amstan_: I come to serve. 2011-01-10T23:17:47 @learn https://github.com/amstan/aichallenge/ as repo 2011-01-10T23:17:47 amstan_: Ready to serve. 2011-01-10T23:17:49 @learn https://github.com/amstan/aichallenge/ as repository 2011-01-10T23:17:49 amstan_: I think that worked... 2011-01-10T23:17:51 @repo 2011-01-10T23:17:52 amstan_: I do not know about 'repo', but I do know about these similar topics: 'botrepo' 2011-01-10T23:17:58 @where repo 2011-01-10T23:17:58 amstan_: I do not know about 'repo', but I do know about these similar topics: 'botrepo' 2011-01-10T23:18:02 nooo 2011-01-10T23:18:08 @forget https://github.com/amstan/aichallenge/ 2011-01-10T23:18:08 amstan_: Error: 2 factoids have that key. Please specify which one to remove, or use * to designate all of them. 2011-01-10T23:18:13 @forget https://github.com/amstan/aichallenge/ * 2011-01-10T23:18:13 amstan_: I come to serve. 2011-01-10T23:18:17 @learn repo as https://github.com/amstan/aichallenge/ 2011-01-10T23:18:17 amstan_: OK 2011-01-10T23:18:21 @learn repository as https://github.com/amstan/aichallenge/ 2011-01-10T23:18:21 amstan_: I think that worked... 2011-01-10T23:18:23 @repo 2011-01-10T23:18:24 amstan_: repo = https://github.com/amstan/aichallenge/ 2011-01-10T23:38:30 *** amriedle has quit IRC (Quit: Lost terminal) 2011-01-10T23:48:03 *** medrimonia2 has quit IRC (Ping timeout: 255 seconds) 2011-01-10T23:50:21 *** tapwater has quit IRC (Quit: tapwater)