User talk:DavidR

Bem-vindo ao Robocode! Não tinha reparado que havia um novo membro português por aqui, espero que tenhas sucesso por estes lados. Cuidado é com o vício, isto come tempo como tudo ;). Btw, swarm intelligence? a teamrumble é capaz de dar para testar uma ou outra teoria disso... --ABC 00:12, 3 December 2009 (UTC)

Bem, no tempo passado, nós costumávamos usar Babelfish para traduzir o seu Português. Hoje em dia, eu pensei que usando o Google Translate é uma idéia melhor. Além disso, como lembrou ABC tem muitos falantes de Português no passado, que a língua principal do wiki é o Inglês, então não escreva Português muito. Ou teremos de usar Bebelfish ou o Google Translate para cada mensagem que você fez. Parece que somente Inglês e Português tem sido falado sobre o wiki, talvez porque o ABC é muito influência Robocoder. Enfim, eu não conheço nenhum Português e esta mensagem foi escrita usando o Google Translate ... --Nat Pavasant 00:30, 3 December 2009 (UTC)

Obrigado ABC / Thank You ABC.

I just started playing with robocode, but have already been infected with the bug... and I still have lot's of things to learn before I can have a competitive bot. For now I'm just starting to get the fundamentals and probably during 2010 I'll move to the TeamRumble competition to test those ideas from my PhD.--DavidR 08:54, 3 December 2009 (UTC)

Please consider making a page for your robot(s) instead of list it on your user page. Thank you. --Nat Pavasant 17:58, 8 December 2009 (UTC)

Good advice! Done! --DavidR 18:02, 8 December 2009 (UTC)

Melee client problem?

DavidR, I'm seeing some very strange results from your melee client... See User_talk:CrazyBassoonist and I've been trying to figure out how B26354 dropped from the top-15 to below 80 and all the battles I've found so far are from your client? Did you use the superpack? How much memory did you set and are you running other programs at the same time? --Darkcanuck 05:42, 3 December 2009 (UTC)

Hi, Yes, I am using the superpack. The only change I did was to increase the memory from 512 to 768 with

java -Xmx768M -Dsun.io.useCanonCaches=false -cp libs/robocode.jar (...)

This was done because I was getting some Out of Memory errors stopping the client almost every 24h.

Also I tried to understand what was happening with the oog.melee.Capulet bot this morning and when running it against MiniR101 v0.4 in v.1.6.1.4, Capulet gets disabled always because of a SetColor out of place... At least that's what the clint prints out:

My Java:

$ java -version
java version "1.6.0_15"
Java(TM) SE Runtime Environment (build 1.6.0_15-b03-226)
Java HotSpot(TM) 64-Bit Server VM (build 14.1-b02-92, mixed mode)

The output from the client:

oog.melee.Capulet 0.1: Throwable: robocode.exception.RobotException: 
You cannot call the setColors() method before your run() method is called, or 
you are using a Robot object that the game doesn't know about.

oog.melee.Capulet 0.1 still has not started after 16947 ms... giving up.

I also tried Capulet 0.1 against other bots and the result is the same, so its not something my MiniR101 is doing :)

On the B26354 Problem I did the same thing and run it against some other bots and he is getting disabled in my client with the message:

darkcanuck.B26354 1.04: Throwable: robocode.exception.RobotException: 
You cannot call the getDataDirectory() method before your run() method is called, 
or you are using a Robot object that the game doesn't know about.

darkcanuck.B26354 1.04 still has not started after 16947 ms... giving up.

So in this case it is the getDataDirectory method that is giving the error.

Hope this helps debug the problem. Do you have any ideas? --DavidR 08:50, 3 December 2009 (UTC)

Ok, this is odd. Both of those bots have been in the rumble for months without issues and it's not a java compilation problem this time. I also see strange results for justin.Mallais [1], YersiniaPestis [2] and there are likely others... Your client seemed fine before yesterday -- has anything changed? --Darkcanuck 15:43, 3 December 2009 (UTC)

A correction on my last comment: the suspect battles go back to at least 20Nov. PallasHawk [3] is another affected bot. Also, I've checked my source code and getDataDirectory() is only called from within the run() method for B26354. --Darkcanuck 16:22, 3 December 2009 (UTC)

Perhaps the problem is:

Java HotSpot(TM) 64-Bit Server VM (build 14.1-b02-92, mixed mode)

? Because I use

java version "1.6.0_15"
Java(TM) SE Runtime Environment (build 1.6.0_15-b03)
Java HotSpot(TM) Client VM (build 14.1-b02, mixed mode, sharing)

and everything work fine. --Nat Pavasant 16:51, 3 December 2009 (UTC)

I just tried with java 1.5 32bits

java version "1.5.0_20"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0_20-b02-315)
Java HotSpot(TM) Client VM (build 1.5.0_20-141, mixed mode, sharing)

And got the same problem At the Terminal:

darkcanuck.B26354 1.04 still has not started after 16947 ms... giving up.

oog.melee.Capulet 0.1 still has not started after 16947 ms... giving up.

And at the console of each robot:

SYSTEM: Skipping robot: xxx.... 

Next Step: I'm going back to 64Bit mode and I'll redownload the Super Pack and see if it works from there. I'll try to report back during this night... --DavidR 17:15, 3 December 2009 (UTC)

New report: There was some problem with my previous install... I re-downloaded the super pack and now both bots work as they should. Possibly something got corrupted in my previous install. I'll try to run it for a while with upoads OFF to see what happens. --DavidR 17:33, 3 December 2009 (UTC)

Do you know what the problem was? Finding the cause can help us from preventing others having the same problem. I'd also like to know when you did this install so I can flag all of the suspect battles for cleanup. It looks like back to Nov-20 at least. --Darkcanuck 04:32, 4 December 2009 (UTC)

Just checked and this buggy install is indeed from 20 Nov 09:37 GMT! I'll try to runs some test in this install to see what was wrong with it. Is there a way to put the client in DEBUG mode? --DavidR 08:44, 4 December 2009 (UTC)

To get more detailed output? No, I don't think so. The best you can do is turn off uploads and scan through the results file, but that probably won't tell you what was wrong. I've flagged all battles from your client back to 19-Nov and am in the process of rebuilding the scores. Unfortunately, you were the biggest melee contributor during that period, so there are 175k uploads to rescore -- it will take me almost two weeks to rescore those at my current rate! --Darkcanuck 15:19, 4 December 2009 (UTC)

Oh, it looks like the problem might go even further back: 16-Nov. This overlaps the period where we thought it was a java 1.5 problem... --Darkcanuck 16:41, 4 December 2009 (UTC)

That's very bad news! It means all results are broken! I'm very sorry this happened... Something must got corrupted during the download of the superpack... Does anyone know any ideas on how to Debug the old install? I still have it but don't know what eles I can do. The Console erros are those I've reported already. --DavidR 00:46, 6 December 2009 (UTC)

Well, the problem could perhaps be narrowed down if you make a copy of the broken old install, and inside it, delete everything in the robots folder except the jar files themselves (i.e. delete .robotcache etc). Then check if that fixes that install or not. My guess is it will. In that case it means the problem was after download, not the download/unzip itself since robotcache is not included in the zip. --Rednaxela 04:35, 6 December 2009 (UTC)