2011-09-24T00:00:53 *** Keth_ has joined #aichallenge 2011-09-24T00:00:54 *** Keth has quit IRC (Read error: Connection reset by peer) 2011-09-24T00:07:25 *** dvladim has joined #aichallenge 2011-09-24T00:24:48 *** sigh has joined #aichallenge 2011-09-24T00:33:51 *** dr0b3rts has quit IRC (Ping timeout: 248 seconds) 2011-09-24T00:36:15 *** ChrisH has joined #aichallenge 2011-09-24T00:40:37 aichallenge: janzert epsilon * r88a9742 / (3 files): Have workers retrieve non-distribution languages from main server - http://git.io/zHXfRQ 2011-09-24T00:47:18 *** dr0b3rts has joined #aichallenge 2011-09-24T00:55:11 *** dvladim has quit IRC (Ping timeout: 245 seconds) 2011-09-24T00:57:22 *** ChrisH has quit IRC (Quit: Page closed) 2011-09-24T01:04:11 *** Maligo has quit IRC (Ping timeout: 252 seconds) 2011-09-24T01:10:19 *** eashoka has joined #aichallenge 2011-09-24T01:10:35 *** Maligo has joined #aichallenge 2011-09-24T01:33:17 *** dvladim has joined #aichallenge 2011-09-24T01:35:44 *** Accoun has quit IRC (Remote host closed the connection) 2011-09-24T01:42:02 *** Accoun has joined #aichallenge 2011-09-24T01:56:20 *** denebil has quit IRC (Ping timeout: 252 seconds) 2011-09-24T02:04:12 *** Palmik has joined #aichallenge 2011-09-24T02:06:26 *** dvladim has quit IRC (Ping timeout: 245 seconds) 2011-09-24T02:13:56 *** eashoka has quit IRC (Ping timeout: 252 seconds) 2011-09-24T02:17:50 *** Maligo has quit IRC (Ping timeout: 256 seconds) 2011-09-24T02:18:09 *** Maligo has joined #aichallenge 2011-09-24T02:18:12 *** Maligo has left #aichallenge 2011-09-24T02:18:55 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 248 seconds) 2011-09-24T03:06:25 *** UncleVasya has joined #aichallenge 2011-09-24T03:07:26 *** dr0b3rts has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 6.0.1/20110830202826]) 2011-09-24T03:12:33 *** chris__0076 has joined #aichallenge 2011-09-24T03:15:59 *** chris_0076 has quit IRC (Ping timeout: 248 seconds) 2011-09-24T03:49:55 *** acieroid` is now known as acieroid 2011-09-24T04:06:11 *** gattschardo has joined #aichallenge 2011-09-24T04:08:58 *** UncleVasya has quit IRC (Ping timeout: 244 seconds) 2011-09-24T04:10:54 good morning 2011-09-24T04:16:41 *** UncleVasya has joined #aichallenge 2011-09-24T04:21:31 *** dvladim has joined #aichallenge 2011-09-24T04:24:20 @later tell eashoka I just saw your blog post from planet wars, was a good read thanks :D 2011-09-24T04:24:20 antimatroid: OK 2011-09-24T04:25:19 *** delt0r_ has joined #aichallenge 2011-09-24T04:28:51 antimatroid: spam: http://ai-contest.com/forum/viewtopic.php?f=21&t=1467 2011-09-24T04:29:25 UncleVasya: I swear i just looked for spam 5 minutes ago and missed it all! thanks :) 2011-09-24T04:29:49 you're welcome. 2011-09-24T04:30:17 *** Akranis has joined #aichallenge 2011-09-24T04:39:11 *** amstan has quit IRC (Ping timeout: 248 seconds) 2011-09-24T04:42:09 antimatroid, I think it's time to make Misc forum for Ants because people begin to share replays and do not know where to put them. 2011-09-24T04:44:10 Also it's good to replace thread about contestbot from PlanetWars forum: http://ai-contest.com/forum/viewtopic.php?f=19&t=436 2011-09-24T04:52:50 UncleVasya: I think amstan would be the best person to do that 2011-09-24T04:52:59 i'm not entirely sure of what he's done in the past 2011-09-24T04:53:03 ok 2011-09-24T04:58:22 *** ibdknox_ has quit IRC (Remote host closed the connection) 2011-09-24T04:58:53 *** ibdknox has joined #aichallenge 2011-09-24T05:03:19 *** ibdknox has quit IRC (Ping timeout: 255 seconds) 2011-09-24T05:24:31 *** antimatroid has quit IRC (Ping timeout: 248 seconds) 2011-09-24T05:24:50 *** dvladim has quit IRC (Ping timeout: 256 seconds) 2011-09-24T05:29:49 *** antimatroid has joined #aichallenge 2011-09-24T05:31:38 *** olexs has joined #aichallenge 2011-09-24T06:02:08 *** arlequ1 has joined #aichallenge 2011-09-24T06:02:24 `http://antsbeta.janzert.com/visualizer.php?game=207961 :) 2011-09-24T06:19:14 *** olexs has quit IRC (Quit: Leaving.) 2011-09-24T06:27:00 Hmmm, it seems I missed a change of rules. Are you allow bot to kill his ants by moving them to one cell? It was an incorrect order earlier. 2011-09-24T06:29:25 I mean this: http://antsbeta.janzert.com/visualizer.php?game=207776&user=308&&row=4&&col=18&&turn=252 2011-09-24T06:35:53 *** dvladim has joined #aichallenge 2011-09-24T07:01:40 *** Kingpin13 has joined #aichallenge 2011-09-24T07:05:42 *** eashoka has joined #aichallenge 2011-09-24T07:13:23 *** arlequ1 has quit IRC (Quit: Page closed) 2011-09-24T07:23:45 *** mleise has quit IRC (Ping timeout: 260 seconds) 2011-09-24T07:29:17 *** delt0r_ has quit IRC (Ping timeout: 276 seconds) 2011-09-24T07:32:12 *** eashoka has quit IRC (Ping timeout: 252 seconds) 2011-09-24T07:37:02 *** olexs has joined #aichallenge 2011-09-24T07:41:09 *** delt0r_ has joined #aichallenge 2011-09-24T07:41:54 *** eashoka has joined #aichallenge 2011-09-24T07:48:48 *** olexs has quit IRC (Quit: Leaving.) 2011-09-24T07:48:59 antimatroid : you are welcome! thanks for the compliments :-) 2011-09-24T07:55:40 *** eashoka has quit IRC (Ping timeout: 252 seconds) 2011-09-24T08:03:23 *** eashoka has joined #aichallenge 2011-09-24T08:29:49 *** TTE has joined #aichallenge 2011-09-24T08:33:32 *** TTE has quit IRC (Client Quit) 2011-09-24T08:47:07 hi 2011-09-24T08:47:22 I suggest to change game specifications 2011-09-24T08:48:12 1. The timeouted bots should receive scores if his ant kills the enemy ant 2011-09-24T08:48:51 2. The quantity of ants should influence to total scores 2011-09-24T08:49:15 e.g. total score = score + k*ants 2011-09-24T08:51:51 where k = 2 - scores of killing one ant (at battle one against one) 2011-09-24T08:52:24 *** Akranis has quit IRC (Read error: Connection reset by peer) 2011-09-24T08:53:09 see that game: http://aichallengebeta.hypertriangle.com/visualizer.php?game=203221 2011-09-24T08:54:29 at turns 585-588 FlagCapper jump from 3rd place to 1st 2011-09-24T08:56:00 with essential decrease of ants 2011-09-24T08:58:11 *** xathis has joined #aichallenge 2011-09-24T08:59:34 at k = 2 it will exclude tactics of end-games when ants enter battles one against one 2011-09-24T09:07:18 I want to exclude tactics of "stupid battles" at end-game, when bot rapidly increase scores at the expense of not necessary ants 2011-09-24T09:11:02 dvladim: Why? 2011-09-24T09:11:17 why do you want to exclude battles at the end 2011-09-24T09:11:29 I want to encourage them 2011-09-24T09:11:46 also if you time out i think you get no score. 2011-09-24T09:11:54 which is the way it should be 2011-09-24T09:13:40 if you time out - you easy opponent 2011-09-24T09:14:47 may be disqualified timeouted bots with no scores and last place? 2011-09-24T09:15:46 yea 2011-09-24T09:16:07 you timed out or crashed... you quit --you get zero score 2011-09-24T09:16:11 or you should 2011-09-24T09:16:20 you are last... and a loser! :) 2011-09-24T09:16:53 you don't get a "winning at the time they gave up in Marathon" medal 2011-09-24T09:17:43 as for this adjusting score so x y and z don't happen... i have yet to hear why z y or z shouldn't happen... 2011-09-24T09:17:56 If the boat is disqualified his ants should are removed from a map 2011-09-24T09:18:04 seriously... why should there not be epic battles at the end? 2011-09-24T09:18:32 i say "stupid battles" 2011-09-24T09:18:37 dvladim: they because NPC much like quite a few other games in fact... i think its a good trade off 2011-09-24T09:19:06 So why are they stupid? why should we discorade battle at the end? 2011-09-24T09:19:15 discourage 2011-09-24T09:19:30 why have battles at all? 2011-09-24T09:19:40 if they are stupid 2011-09-24T09:20:07 in this game at the end FlagCapeer with 15 ants potentialy has weak position 2011-09-24T09:20:09 and if they are not stupid at one point in the game, what makes them stupid at a different time in the game 2011-09-24T09:20:31 *potentially* being the key work 2011-09-24T09:20:33 word 2011-09-24T09:21:06 have a better both than the others... you can make it work in battle for example... 2011-09-24T09:21:17 don't and you lose .... not seeing a problem here. 2011-09-24T09:21:48 brb gotta restart my x win 2011-09-24T09:21:51 *** delt0r has quit IRC (Remote host closed the connection) 2011-09-24T09:22:25 stupid - because the purpose - increase score without caring of quantity of ants 2011-09-24T09:23:50 *** delt0r has joined #aichallenge 2011-09-24T09:24:13 *** Lithosphere has quit IRC (Remote host closed the connection) 2011-09-24T09:24:28 *** Lithosphere has joined #aichallenge 2011-09-24T09:24:28 *** Lithosphere has joined #aichallenge 2011-09-24T09:25:17 quantity of ants at end-game it doesn't matter 2011-09-24T09:25:51 It is not necessary to protect ants 2011-09-24T09:25:59 score does... 2011-09-24T09:26:16 every ant needlessly lost is points someone else gets 2011-09-24T09:26:43 It is better to lose ants, but to receive scores 2011-09-24T09:27:01 it is better to maximize your score gain 2011-09-24T09:27:15 you can't do that without some ant to use... 2011-09-24T09:27:28 its not a given that "just let em all die" will work 2011-09-24T09:28:04 in fact it is trivial to show that in 2 player game this is useless... you must kill more than you lose if you want to gain anything 2011-09-24T09:28:17 Yes, therefore the quantity of ants should influence quantity of scores 2011-09-24T09:29:05 It does! you get a point for each one spawned .. 2011-09-24T09:29:38 no, quantity of ants at endgame 2011-09-24T09:30:01 should influence to scores 2011-09-24T09:30:13 its mathematically the same as give a -1 to each ant you lose 2011-09-24T09:30:17 nothing more 2011-09-24T09:30:37 and no--if we do that then just keep it at "the most ants at the end wins" 2011-09-24T09:30:51 which will result in nothing more than harvest fests 2011-09-24T09:30:58 aka why keep battles at all 2011-09-24T09:31:45 you this has been trashed out for many months 2011-09-24T09:31:50 since like Feb 2011-09-24T09:34:23 *** McLeopold has joined #aichallenge 2011-09-24T09:35:36 Also a bot with bigger ants amount with right battle resolution will not give his loosing opponent an opportunity to kill ants without losses. 2011-09-24T09:36:55 basically if you don't have the most ants close to end game, and your == matched bots, you have a very good chance of getting first 2011-09-24T09:37:01 sorry 2011-09-24T09:37:13 very good chance of losing 2011-09-24T09:38:16 From another point, if loosing in quantity bot can fight well enough to kill another bot's ants without losses so this is the problem of that bot. 2011-09-24T09:40:51 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-09-24T09:41:45 _flag: how's your matchups been lately? 2011-09-24T09:44:47 <_flag> McLeopold: Better, although every once in a while I have a game like this: http://antsbeta.janzert.com/visualizer.php?game=208377 2011-09-24T09:45:55 <_flag> I actually lost points for winning that one, not sure how 2011-09-24T09:46:29 <_flag> But usually the matchups are pretty good 2011-09-24T09:47:08 your sigma increased a little, your mu is the same 2011-09-24T09:47:17 hover over the skill to see 2011-09-24T09:47:42 <_flag> My skill went down 2011-09-24T09:47:52 skill is mu - sigma*3 2011-09-24T09:48:57 <_flag> Okay, I see 2011-09-24T09:49:02 it just means your next match gets you a bigger update, if you win 2011-09-24T09:49:10 unfortunately, your lost ;) 2011-09-24T09:49:43 <_flag> ;) 2011-09-24T09:51:24 McLeopold: what tests do you run on submitted bots? 2011-09-24T09:51:39 i'm still working on the erlang starter package and i keep getting timeouts 2011-09-24T09:51:59 i cloned the git repo though, and my bot runs through the test_bot.sh script just fine 2011-09-24T09:52:05 gattschardo: test_bot.sh 2011-09-24T09:52:14 I might be able to help you, hold on 2011-09-24T09:52:16 thats strange then 2011-09-24T09:55:38 *** mcstar has joined #aichallenge 2011-09-24T09:58:08 *** eashoka has quit IRC (Ping timeout: 252 seconds) 2011-09-24T09:58:10 aichallenge: McLeopold epsilon * r63ec92e / worker/engine.py : capture timeout errors - http://git.io/AJK35w 2011-09-24T09:58:41 gattschardo: make sure to write to stderr and flush often, now the engine will report errors after a timeout occurs 2011-09-24T09:59:37 i will get all stderr output? 2011-09-24T09:59:53 locally, yes 2011-09-24T10:00:01 on the official server, just 1024 bytes 2011-09-24T10:00:08 ok, that should be fine 2011-09-24T10:00:10 thanks 2011-09-24T10:04:22 _flag: are you testing beraks tcp server again today? 2011-09-24T10:09:38 <_flag> xathis: Probably not today. I may try setting it up on a seperate computer, I'm just a little scared that there may be some possible security bugs still around 2011-09-24T10:09:58 <_flag> Anyone can set it up though 2011-09-24T10:10:49 <_flag> It's fairly easy, all you have to do is run main.py and forward ports 2080 and 2081 in your router to your computer 2011-09-24T10:11:05 _flag: i'm really bad with server stuff 2011-09-24T10:11:57 is the security issue from yesterday fixed on github? 2011-09-24T10:13:34 McLeopold: i get no output at all and still a timeout on turn 0, i suspect it doesn't start properly 2011-09-24T10:13:56 <_flag> xathis: It would appear to be 2011-09-24T10:13:59 can you (or someone) check for the files that are created on the server? 2011-09-24T10:14:36 _flag: okay maybe i try to set it up later 2011-09-24T10:14:39 i think i must've messed something up in compiler.py, because everything else works, locally 2011-09-24T10:17:46 gattschardo: is this the starter_bot? 2011-09-24T10:18:25 what, this? 2011-09-24T10:18:51 i upload the stuff to the beta server as 'ydinic' 2011-09-24T10:18:57 ok 2011-09-24T10:20:07 there should be ants, map and my_bot, a .erl and a .beam of each, after successful compile (works locally, but i really don't know...) 2011-09-24T10:20:19 ok 2011-09-24T10:35:01 also, could you run "erl -noshell -s init stop" on the server where the code runs and tell me whether it prints anything to stdout or stderr (it should not) 2011-09-24T10:43:11 * gattschardo is going for a walk with the dog 2011-09-24T10:43:12 bbl 2011-09-24T10:54:05 gattschardo: you are writing to stderr, but you must not be properly flushing it. 2011-09-24T10:54:25 gattschardo: I can update a bot that times out with an error message and the server captures it 2011-09-24T10:55:23 gattschardo: {"init terminating in do_boot",{undef,[{my_bot,start,[]},{init,start_it,1},{init,start_em,1}]}} Crash dump was written to: erl_crash.dump init terminating in do_boot () 2011-09-24T10:56:41 Hi McLeopold. Are you allow bot to kill his ants by moving them to one cell? It was an incorrect order earlier. 2011-09-24T10:57:26 yes 2011-09-24T10:57:38 if you step on your own ant, they both die 2011-09-24T10:57:44 but there is no score 2011-09-24T10:58:00 also, you can't walk on food 2011-09-24T10:58:20 I saw it in one game and was confused. Now I know that it's okay and I just missed this change. Thanks 2011-09-24T11:09:32 *** UncleVasya has quit IRC (Ping timeout: 244 seconds) 2011-09-24T11:12:17 *** mceier has joined #aichallenge 2011-09-24T11:13:37 what change? 2011-09-24T11:14:08 It never changed 2011-09-24T11:21:30 *** detryo has joined #aichallenge 2011-09-24T11:45:06 McLeopold: thanks for that error message, that explains what's happening 2011-09-24T11:45:50 it says that my_bot:start/0 is undefined, which means it can't find the my_bot module 2011-09-24T11:46:16 so either the my_bot.beam file is not in the working directory of where the vm runs, or it was not compiled at all 2011-09-24T11:56:34 aichallenge: janzert epsilon * r7e589c2 / setup/worker_setup.py : Add missing dependency for nodejs - http://git.io/BQ0Wew 2011-09-24T12:00:47 *** Keth_ has quit IRC (Read error: Connection reset by peer) 2011-09-24T12:01:01 *** Keth has joined #aichallenge 2011-09-24T12:06:33 *** detryo has quit IRC (Ping timeout: 260 seconds) 2011-09-24T12:08:18 *** detryo has joined #aichallenge 2011-09-24T12:22:16 *** mleise has joined #aichallenge 2011-09-24T12:23:25 *** sigh has quit IRC (Remote host closed the connection) 2011-09-24T12:29:15 *** okayzed is now known as okay| 2011-09-24T12:44:33 *** mceier has quit IRC (Ping timeout: 256 seconds) 2011-09-24T12:46:05 *** mceier has joined #aichallenge 2011-09-24T12:46:24 *** berak has joined #aichallenge 2011-09-24T12:52:14 *** amstan has joined #aichallenge 2011-09-24T12:52:14 *** ChanServ sets mode: +o amstan 2011-09-24T13:07:43 aichallenge: janzert epsilon * rb5be63d / worker/worker.py : Have worker check for a stop file after each task when executing forever - http://git.io/I2FAMQ 2011-09-24T13:35:11 *** Kingpin13 has quit IRC (Quit: quit) 2011-09-24T13:36:11 Any clojure users around? looking for opinions on upgrading the version used for the contest to 1.3 2011-09-24T13:38:01 soon to be clojure user 2011-09-24T13:38:19 i think that is what everybody uses 2011-09-24T13:38:42 i guess 1.2.1 is the prev. one right? 2011-09-24T13:38:44 well it was just released 19 hours ago 2011-09-24T13:38:56 well, 1.3 is available for quite some time 2011-09-24T13:39:07 it is called master snapshot 2011-09-24T13:39:08 https://groups.google.com/forum/#!topic/clojure/w5Nmx5rPaQs 2011-09-24T13:39:46 yes but not the actual release though :} 2011-09-24T13:40:04 yeah, i guess it was beta then 2011-09-24T13:40:44 is the jvm ver 6 or ver 7? 2011-09-24T13:41:16 i installed 7 for myself, seems to work with it 2011-09-24T13:41:25 pretty sure it's 6 2011-09-24T13:41:32 * mcstar isnt sure whats the difference of the 2 versions though 2011-09-24T13:41:37 whatever comes in ubuntu 11.04 natty 2011-09-24T13:42:06 and jdk7 didn't come out till after that I'm pretty sure 2011-09-24T13:43:56 i guess java has the biggest startup delay of all the languages, right? 2011-09-24T13:44:24 *** Keth has quit IRC (Read error: Connection reset by peer) 2011-09-24T13:44:29 that would be my guess as well, but I don't have any real measurements 2011-09-24T13:44:55 must take some time to start a game with 8 players all of java :) 2011-09-24T13:45:30 ignoring of course that languages on top of the JVM will pretty much by definition be longer than straight java (e.g. clojure and scala) 2011-09-24T13:46:44 i dont know, does starting the jvm and starting a java app take the same time? 2011-09-24T13:47:38 (im a nub regarding java, i can imagine that a clean jvm, is not the same as a full java runtime) 2011-09-24T13:48:01 mcstar: AFAIK, the Java runtime is pretty much just the JVM. 2011-09-24T13:48:26 aham, so all the jvm languages are written explicitely in java 2011-09-24T13:48:46 well, they compile down to the same bytecode 2011-09-24T13:48:47 mcstar: Not necessarily, but they're compiled down to Java bytecode. 2011-09-24T13:48:54 I think some use self hosting compilers 2011-09-24T13:49:18 that is neat :) 2011-09-24T13:49:19 But Java doesn't need much of a runtime. 2011-09-24T13:49:25 (beyond JVM facilities) 2011-09-24T13:50:19 yeah, the JVM and Java were built to fit together quite tightly, all the other languages have an 'impedance' mismatch to some extent or another 2011-09-24T13:50:32 Though things like invokedynamic are making it better. 2011-09-24T13:50:45 although JDK7 has features meant to help it... yes 2011-09-24T13:51:14 But yeah, the semantics of the native JVM types match the basic Java types pretty closely. 2011-09-24T13:51:58 *** Keth has joined #aichallenge 2011-09-24T13:52:36 i wish the jvm will become more lightweight, until then the languages on top of it cant really be used for scriptinh 2011-09-24T13:52:38 g 2011-09-24T13:53:17 mcstar: What do you mean by scripting? (The term is so overloaded nowadays.) 2011-09-24T13:53:37 *** dvladim has quit IRC (Quit: Konversation terminated!) 2011-09-24T13:53:51 i mean the usual linux/unix scripting way 2011-09-24T13:53:56 Do you mean as a drop-in for a Perl script (typically short-lived code that executes for a short amount of time then terminates, and is developed similarly quickly)? 2011-09-24T13:54:07 couple hunderd lines tops 2011-09-24T13:54:24 like using bash commands, you know 2011-09-24T13:54:42 one program for one purpose, can get invoked many times 2011-09-24T13:54:49 and terminates fast 2011-09-24T13:55:07 i dont know perl 2011-09-24T13:55:13 but lets say python 2011-09-24T13:55:21 So you're referring to the startup time of the JVM. 2011-09-24T13:55:28 yeah, thats a bummer 2011-09-24T13:55:41 i cant imagine doing "scripting" in clojure 2011-09-24T13:55:48 but i can do it in sbcl 2011-09-24T13:56:14 (actually, that has a pretty long startup time already) 2011-09-24T13:56:30 A lot of it is disk IO reading in the standard library classes. 2011-09-24T13:56:38 i heard that before 2011-09-24T13:56:41 This is why Java shipped with the oh-so-unpopular "quick starter". 2011-09-24T13:56:53 Which more or less loads them into memory and sits there, so they're loaded out of cache next time. 2011-09-24T13:57:17 but why is that? java isnt that modular as python? 2011-09-24T13:57:20 or what? 2011-09-24T13:57:32 i guess python is not a bit less "frameworked" 2011-09-24T13:57:47 (bad term, i mean it has the same functionality, doesnt it?) 2011-09-24T13:58:24 My understanding of the comparative startup procedure of the two VMs isn't good enough to give a satisfactory answer to that. 2011-09-24T13:58:54 jbroman: does java load everything it has? 2011-09-24T13:59:06 it must have a module system too 2011-09-24T13:59:17 or namespace or package i dont know what it is called 2011-09-24T13:59:29 mcstar: It is modular (packages, classes, etc), but a number of classes must be initialized at startup. 2011-09-24T13:59:41 (The string, much of java.lang, etc.) 2011-09-24T13:59:53 I read an article on this at some point but cannot find it right now. 2011-09-24T14:00:56 what always surprised me, is that java runs on mobile devices, so they must have a lot of experience dumbing things down, fitting things in less memory, disk, time whatever, yet its the slowest starting vm for a pc 2011-09-24T14:01:49 mcstar: It does get quite efficient once running, though, and for many of Java's applications this is a good tradeoff. 2011-09-24T14:04:05 I mean, for a small application the startup time is quite sub-second. 2011-09-24T14:04:19 (With the client JVM, not the server one.) 2011-09-24T14:04:32 Though -server is comparable. 2011-09-24T14:04:43 how do you do that? 2011-09-24T14:04:48 Do what? 2011-09-24T14:04:52 launch a server process of java? 2011-09-24T14:04:58 and let scripts use that? 2011-09-24T14:05:09 That is a misleading name, I'll admit. 2011-09-24T14:05:17 "java -server" launches the HotSpot Server VM. 2011-09-24T14:05:24 Which is a version of the JVM intended for use on servers. 2011-09-24T14:07:30 is that a launch once, reuse many times version? 2011-09-24T14:07:40 or no 2011-09-24T14:07:51 No, it's still an independent JVM. 2011-09-24T14:08:03 Just one with different performance characteristics. 2011-09-24T14:08:12 so you are saying it just takes less time to start? 2011-09-24T14:08:14 aha 2011-09-24T14:08:27 i just tried it, and it took the same time 2011-09-24T14:08:29 I think it takes more, actually. But does more aggressive optimization in some cases. 2011-09-24T14:08:36 i think im missing something here.. 2011-09-24T14:08:40 As I said, intended for long-running processes. 2011-09-24T14:08:47 Though you could probably construct such a Java process if you wanted. 2011-09-24T14:09:00 Would be pretty hacky though; you'd have to hook all over the place. 2011-09-24T14:09:17 jbroman: i think i simply misunderstood your sentence: I mean, for a small application the startup time is quite sub-second. 2011-09-24T14:09:56 Oh, sorry. I just meant that starting a small Java application doesn't take very long. 2011-09-24T14:10:26 starting clojure takes 10 sec real time, 18 user time for either 2011-09-24T14:10:59 Wait, really? 2011-09-24T14:11:12 yes 2011-09-24T14:11:15 :) 2011-09-24T14:11:47 im not sure how much is the clojure part, but i dont think its too much, its a 2-3MB file 2011-09-24T14:12:21 I can start the Clojure interpreter in far, far less than that, so that large a time surprises me. 2011-09-24T14:12:41 then help me out 2011-09-24T14:12:46 what am i doing wrong? 2011-09-24T14:12:57 I really know even less about Clojure than Java. :P 2011-09-24T14:14:23 mcstar: Anyhow, I just stumbled upon this. May suck, but still you might be interested: https://github.com/Neronus/clj-server 2011-09-24T14:14:26 jbroman: you can start it in less time than me, so you seem to know it a hell lot better 2011-09-24T14:14:41 mcstar: I just apt-get installed it and ran it? 2011-09-24T14:15:08 how do you start it? 2011-09-24T14:15:19 mcstar: "clojure"? 2011-09-24T14:15:46 i think i can get a pre-made script,to look at it 2011-09-24T14:15:59 which more or less does: 2011-09-24T14:16:00 i know arch has a clojure package but im not using it 2011-09-24T14:16:06 java -cp clojure.jar clojure.main … 2011-09-24T14:16:08 java -server -cp .clojure/clojure.jar clojure.main 2011-09-24T14:16:14 this is how i start it 2011-09-24T14:16:34 *** ibdknox has joined #aichallenge 2011-09-24T14:16:53 jbroman: how can i start something else with java? 2011-09-24T14:17:02 some other jar file 2011-09-24T14:17:12 i dont have anyhting, recommend me something 2011-09-24T14:17:21 "something"? 2011-09-24T14:17:27 yeah 2011-09-24T14:17:28 I was just running a hello world app. 2011-09-24T14:17:37 java hello world? 2011-09-24T14:17:44 i want to skip clojure 2011-09-24T14:17:52 Hold, I'll paste it. 2011-09-24T14:18:44 https://gist.github.com/1239662 2011-09-24T14:18:59 Save as HelloWorld.java, run "javac HelloWorld.java" to compile, then "java HelloWorld" to run. 2011-09-24T14:19:12 k 2011-09-24T14:19:33 I get about 0.4 seconds to run the Java hello world, 1.0 seconds to run the Clojure equivalent. 2011-09-24T14:20:47 real0m0.177s 2011-09-24T14:20:50 user0m0.137s 2011-09-24T14:20:52 sys0m0.037s 2011-09-24T14:21:14 this is interesting 2011-09-24T14:21:38 so when i run this with "java" the hotspot just starts up 2011-09-24T14:21:48 then its pretty fast 2011-09-24T14:22:34 Actually, I ran Java on a different machine. Actual Java runtime for me is 0.08 s. 2011-09-24T14:23:05 It's related to the complexity of your app, though. 2011-09-24T14:23:38 If your app is one class, linked to a bit of java.io and not much else, it's pretty fast. I think it loads your dependencies pretty aggressively at startup, though. 2011-09-24T14:24:06 ok, so what is youre clojure equivalent? 2011-09-24T14:24:28 when i used the above java -cp invokation, it brings up the repl 2011-09-24T14:24:39 Probably a cleaner way of doing it, but for equivalency, I put this in HelloWorld.clj: 2011-09-24T14:24:40 (. (. System out) println "Hello world") 2011-09-24T14:24:43 so i guess you mean, you run a clojure script with it, without interactivity 2011-09-24T14:24:51 And then java -cp clojure.jar clojure.main HelloWorld.clj 2011-09-24T14:25:58 I think you can compile Clojure, though, which would be more fair if possible. 2011-09-24T14:26:11 real0m7.791s 2011-09-24T14:26:13 user0m12.766s 2011-09-24T14:26:15 sys0m0.200s 2011-09-24T14:26:28 Wow. 2011-09-24T14:29:00 *** gattschardo has quit IRC (Quit: Кіровоград) 2011-09-24T14:30:05 jbroman: all you need is (println "hello world") 2011-09-24T14:30:25 ibdknox: See, I knew there would be a function in global scope that would do it. 2011-09-24T14:30:51 No effect on startup time, though. :) 2011-09-24T14:30:59 no haha 2011-09-24T14:31:24 What are you guys trying to figure out? 2011-09-24T14:31:41 ibdknox: mcstar wants to know why Clojure takes 10 seconds to start for him. 2011-09-24T14:31:56 I imagine it's disk I/O speed, maybe CPU speed, but I'm surprised its 10x slower than what I'm seeing. 2011-09-24T14:32:01 real0m1.745s 2011-09-24T14:32:02 *** okay| is now known as okayzed 2011-09-24T14:32:03 user0m1.960s 2011-09-24T14:32:05 sys0m0.097s 2011-09-24T14:32:14 i downloaded the 1.3.0 release 2011-09-24T14:32:28 the previous times were for 1.3.0-master-SNAPSHOT 2011-09-24T14:32:41 jbroman: are you measuring already warmed up disk caches possibly? 2011-09-24T14:32:43 jbroman: are those comparable to yours? 2011-09-24T14:32:58 mcstar: I got about 1 second real. So close enough. 2011-09-24T14:33:10 janzert: Well, sure, but mcstar has run it enough times that it ought to be in his cache now, too. 2011-09-24T14:33:27 yeah, i guess my caches are warm, only my feet are cold 2011-09-24T14:33:54 ibdknox: hi, seems that im ready to join the cult of hickey 2011-09-24T14:34:03 mcstar: excellent 2011-09-24T14:34:30 mcstar: use leiningen to run clojure, it'll make your life much happier 2011-09-24T14:34:36 i have it 2011-09-24T14:34:40 good deal 2011-09-24T14:34:46 it just took soo long to start up 2011-09-24T14:34:54 you can use cake if you want 2011-09-24T14:34:59 that gives you a persistent JVM 2011-09-24T14:35:24 ibdknox: Awesome, eminently ungoogleable name. 2011-09-24T14:35:24 im using slime with clojure-jack-in 2011-09-24T14:35:38 ibdknox: Tried "cake", then "java cake", then finally "java cake jvm". 2011-09-24T14:36:10 ibdknox: have any advantages or disadvantages for the contest moving to clojure 1.3? 2011-09-24T14:36:10 Perhaps "clojure cake" would've worked better. 2011-09-24T14:36:16 https://github.com/flatland/cake 2011-09-24T14:36:17 clojure cake first hit 2011-09-24T14:36:34 janzert: shouldn't really make a difference 2011-09-24T14:37:01 janzert: maybe if you rely really heavily on very large number computation 2011-09-24T14:37:02 lol 2011-09-24T14:37:11 :) 2011-09-24T14:37:29 otherwise it's marginal speed gains 2011-09-24T14:37:36 ibdknox: seems to be exactly liek leininggen 2011-09-24T14:37:43 mcstar: it is 2011-09-24T14:37:54 mcstar: the difference is the use of a persistent JVM 2011-09-24T14:37:56