|
|
| (3 intermediate revisions by 3 users not shown) |
| Line 1: |
Line 1: |
| − | Welcome and best of luck with your bots! =) --[[User:Voidious|Voidious]] 14:50, 4 February 2010 (UTC)
| |
| | | | |
| − | Welcome to Robowiki! Looking at what you've put in the rumble, one quick suggestion is to have all of your bots (WhoAmI.WhoAmI and uccc.Orbiter) under a common package. This is recommended so authors can easily be sorted out, and also one can get one's country flag in the rumble if one wishes. Best of luck with your Robocoding! :) --[[User:Rednaxela|Rednaxela]] 01:59, 5 February 2010 (UTC)
| |
| − |
| |
| − | Thanks for opening a dialogue folks! Rednaxela, I agree about the package name. I'll be removing my bots in a day or so once the results stabilize. I've proposed the uccc.* package name to the club. While I don't see a reason why we wouldn't use it, I want to hold off until our next meeting before going forward with that namespace -- just in the event someone makes a good case for us not to use it. We will have a decision made by 02-10-2010. Anything we submit between now and then will be removed and repackaged under the new name. I'll drop a post in [[Talk:RoboRumble/Country_Flags]] once we have decided. Thank you again for your input, we appreciate it! [[User:Christopher.Hilla|Christopher.Hilla]] 02:17, 5 February 2010 (UTC)
| |
| − |
| |
| − | Group effort, eh? Sounds like it could be very successful and fairly frustrating. I've had two friends who did Robocode. The one who introduced me (and was retired from it) and another that I competed back and forth with, emailing zipped up class files back and forth as we one-upped each other. He also dropped out after a few months. Would be cool to interest someone else in it to collaborate with but it hasn't happened yet. --[[User:Pedersen|Martin]] 03:06, 5 February 2010 (UTC)
| |
| − |
| |
| − | We're 12 students in the SUNY Ulster computer science program that have decided to compete in RoboCode as a function of the computer club. We have a pretty simple mission -- collaborate to get a bot as high up in the rankings as we can before the semester is up. After this round is over with, who knows where we'll go with it. At least you've got our company for the next 12 weeks. Thanks for the well wishes :) [[User:Christopher.Hilla|Christopher.Hilla]] 01:55, 6 February 2010 (UTC)
| |
| − |
| |
| − | == Re: Subject line comment comment ==
| |
| − |
| |
| − | Nah, we sometimes use them on Robowiki. I personally try to when starting a new thread of discussion. One note is part of why you may not see it on some talk pages is the ones migrated from the old wiki where that wasn't done so much. --[[User:Rednaxela|Rednaxela]] 04:14, 6 February 2010 (UTC)
| |
| − |
| |
| − | == Bad Rumble Results ==
| |
| − |
| |
| − | Hi Christopher. Could you take a look at what happens with Midboss 1r in your robocode install please? I noticed your client upload some [http://darkcanuck.net/rumble/BattleDetails?game=roborumble&name=ags.Midboss%201r&vs=ap.Frederick%201.1 bad] [http://darkcanuck.net/rumble/BattleDetails?game=roborumble&name=ags.Midboss%201r&vs=synnalagma.test.MiniNeural%201.1 results]. Thanks. --[[User:Rednaxela|Rednaxela]] 21:18, 10 February 2010 (UTC)
| |
| − |
| |
| − | ap.Frederick doesn't shoot at all and MiniNeural hardly ever lands a shot, so Midboss 1r wins every time in both cases --[[User:Christopher.Hilla|Christopher.Hilla]] 03:55, 22 February 2010 (UTC)
| |
| − |
| |
| − | I wonder why then a uccc client uploaded results to the contrary a couple times =\ --[[User:Rednaxela|Rednaxela]] 04:00, 22 February 2010 (UTC)
| |
| − |
| |
| − | I've deleted and made a new roborumble client from the zip available on the wiki on my main computer. I have one other box that is running the client as well and I'll reload that one tomorrow, maybe that'll clear up the problem. Any suggestions as to why this would happen? --[[User:Christopher.Hilla|Christopher.Hilla]] 05:23, 22 February 2010 (UTC)
| |
| − |
| |
| − | Tough to guess without reproducing it and checking the robot console for errors. I hit some bad results for Midboss recently that were the result of a memory leak in Robocode, but I thought Rednaxela found a workaround in his bot. If it only happens with Midboss, you could exclude it using roborumble.txt for now. --[[User:Voidious|Voidious]] 15:22, 22 February 2010 (UTC)
| |
| − |
| |
| − | : The leak is that pre-"1.7.2.0 Beta" versions of robocode, including 1.6.1.4, '''never''' free static memory stored in bots '''ever''', so it's more than a little leak, and affects all bots, not just Midboss. Latest version of Midboss both started using lazy-allocation for it's huge VCS array, reducing memory usage 10x, plus it also nulls out all static data upon receiving the battle ended event and also calls System.gc() then too. Other advanced learning bots should currently be affected much much more than Midboss by the leak issues actually... --[[User:Rednaxela|Rednaxela]] 15:38, 22 February 2010 (UTC)
| |
| − |
| |
| − | == 0 wins ==
| |
| − |
| |
| − | I have a confusing problem related to uccc.MilkyWay_1.00. Although it is an improvement over our previous bots, it always loses every round in the rumble. As of right now it has an Elo of -914.5 and an APS of 0.195% because it loses 2100 to 0 against every bot it fights. It is now dead last in the rankings (a privileged spot, no doubt). What gives?
| |
| − |
| |
| − | I removed MilkyWay from the participants list after about 500 pairings. I also noticed that my client is almost exclusively performing battles against MilkyWay, there are only about 5 other cases when two bots that I didn't write were battled. Any thoughts as to why this is happening?
| |
| − |
| |
| − | --[[User:Christopher.Hilla|Christopher.Hilla]] 04:20, 25 February 2010 (UTC)
| |
| − |
| |
| − | Did you try starting up Robocode in the robocode install and running it? I have to guess it's throwing an error in 1.6.1.4 or something... maybe using some Robocode API added in 1.7+? --[[User:Voidious|Voidious]] 04:24, 25 February 2010 (UTC)
| |
| − |
| |
| − | That was exactly it. 1.7.1.6 has util.Utils.normalRelativeAngleDegrees and 1.6.1.4 doesn't. Thanks for the help! --[[User:Christopher.Hilla|Christopher.Hilla]] 04:36, 25 February 2010 (UTC)
| |
| − |
| |
| − | Double edit conflict: Beat me to answering it. Also about why the client was almost exclusively running battles with it: Rumble clients have a strong preference towards bots which have not yet reached full pairings. If all bots have reached full pairings, they favor bots that have not yet had 2000 total battles (or another number if the config is changed). If all bots have reached that condition, only then is it at random. --[[User:Rednaxela|Rednaxela]] 04:49, 25 February 2010 (UTC)
| |
| − |
| |
| − | == Bot pages ==
| |
| − |
| |
| − | Hey, why not make bot pages for your bots? =) Always interesting to see what people are up to. And just curious, are you the author of all of them, or are different people working on them? --[[User:Voidious|Voidious]] 16:30, 25 February 2010 (UTC)
| |
