[Home]RoboRumble/Chat

Robo Home | RoboRumble | Changes | Preferences | AllPages

Difference (from prior major revision) (no other diffs)

Added: 273a274,275


To the author of as.xbots: The filename should be <as.xbots_1.0.jar> (note the underscore instead of space). Although Robocode itself has no problem with the current name of the jar-file, RoboRumble is not that intelligent and can not handle it. And ofcourse welcome at the wiki, it is always nice to see a fresh rammer trying to take the ram-throne. Take a peek at RamBots and/or RambotChallenge2K6. -- GrubbmGait

We need somewhere for general chat, so here is somewhere.

Archives: RoboRumble/Chat06012007


Looking at the rankings a while ago, I think I see a lot of bots missing. At least 30 from the top 120s or so, because Carruthers is about 30 places higher that I've expected. Is it me, or do I need to go to sleep... -- Nfwu

There are 84 missing in all that I can tell at this time. Thats really wierd, nothing I can think of can explain this. Seraphim only shot up about 5 [places as far as I know, however looking at the premier league results of a older version of seraphim there was more bots, not to mention Ascendant and SandboxDT are missing. --Chase-san

Probably something happened to the participants page. I'll check it out and will definitely have 3 clients running overnight, so hopefully it will be fixed up before long... -- Voidious

Spoke too soon, the participants page looks OK to me. I don't see any text in my RR clients that says it's removing them, which is what would happen if they were removed from the participants page. Maybe somebody setup a new client and had it access the wrong participants page in the setup file? I've read of that happening before. Anyway, I just posted a new Komarious and I will have my 3 clients going all night - it should face everyone, and with each bot it faces they should re-appear in the listing. Here's to hoping! -- Voidious

The rankings did disappear again. They will re-appear again, but it is somewhat annoying. Will anyone who setup a new client check the url's in the roborumble.txt file. You can find a correct example here. -- GrubbmGait


Ever wanted a "who's client's running now" type of utility? =) Now you have one! http://thekandieman.com/nfwu/rr-uploads.php -- Nfwu

Neat! Thanks for posting that. -- Voidious

How do you read the output?
Uploaders for the last 20000 bytes:
Reading Range: 162958718 to 162978718
array(3) { [0]=> string(4) "NFWU" [1]=> string(8) "Voidious" [200]=> string(0) "" }
-=End Of Output=-
--Starrynte

It is a bit cryptic, but it's basically this line:

array(3) { [0]=> string(4) "NFWU" [1]=> string(8) "Voidious" [200]=> string(0) "" } 
that shows that Voidious , some anonymous guy and I are uploading. --Nfwu

Source code: Use and abuse! http://www.thekandieman.com/nfwu/rr-uploads.txt -- Nfwu

Nice feature! But isn't that anonymous guy just a 'border-issue'? Can you see how much each of them uploaded? -- GrubbmGait

Three questions:

-Starrynte

@Starrynte:

  1. Updated for other types of games, example: http://thekandieman.com/nfwu/rr-uploads.php?game=minirumble
  2. If the number of battles is less than the BATTLESPERBOT parameter of your client, those bots have priority. And, if i remember correctly, the client also asks the server what robots to battle. If unable to connect to server or DOWNLOAD=NOT, randomly battle bots together.
    1. Typically, BATTLESPERBOT = 999.
  3. It's a typo in the source code. Line 102, roborumble/RoboRumbleAtHome?.java.
-- Nfwu

Other than a minor typo (Who's uplodaing now), very cool! --Starrynte

RR-uploaders script: Repaired socket opening error. I was hardcoding the IP, so when the server changed IP, it didn't catch it. -- Nfwu

Seems that you're only counting the one-on-one battles, as I am not on the list while running melee? -- GrubbmGait

Does anybody here successfully run version 1.2+ of Robocode in their RRAH clients? I have this problem when I try to upgrade to 1.2. Technically, I should just be able to replace the .jar file, I believe; that's what I did to upgrade to 1.1.3. But when I do that for 1.2, I start getting this error:

Exception in thread "Application Thread" java.lang.NumberFormatException: For input string: "pez.mini.Gouldingi 1.5"
        at java.lang.NumberFormatException.forInputString(NumberFormatException.java:48)
        at java.lang.Integer.parseInt(Integer.java:447)
        at java.lang.Integer.parseInt(Integer.java:497)
        at roborumble.battlesengine.BattlesRunner.runBattles(BattlesRunner.java:92)
        at roborumble.RoboRumbleAtHome.main(RoboRumbleAtHome.java:94)
That line in BattlesRunner?.java looks like this:
	int PointsFirst = Integer.parseInt((String) resultsTable.getValueAt(0,1));
Anyone else had this problem while running 1.2+ versions of Robocode in their clients? I don't see what I could possibly be doing differently, so I'll be very confused if I'm alone in this problem...

-- Voidious

Could it mayhaps be the change from the hashtable to a hashmap? --Chase-san

Where is that change? I did delete/rebuild the robot.database file and .robotcache directory immediately after switching, too, forgot to mention that. -- Voidious

We really need to backlog this pages, tis huge. And also all the hashtables in robocode were replaced. Though i'm not sure why it would throw that exception, unless the input was null or simular, try printing the string to the console before parsing and see what it is. --Chase-san

I did, it's like "voidious.mini.Komarious 1.0", IIRC, package/botname + space + version number. I'll do some more investigating, like comparing it to the 1.1.3 client's behavior. Maybe the indices got offset by 1 or something... -- Voidious

Hmm, does indeed look like something might be offset by 1. I didn't force the same 2 bots to fight each time, but... Here's the output for resultsTable.getValueAt(0,1) + " :: " + resultsTable.getValueAt(0,2) + " :: " + resultsTable.getValueAt(0,3) in 1.1.3 and 1.2:

From 1.1.3:
Fighting battle 0 ... stelo.MatchupMicro 1.0,dummy.micro.Sparrow 2.5
2936 :: 1050 :: 210
RESULT = stelo.MatchupMicro 1.0 wins 2936 to 2145
Fighting battle 1 ... stelo.MatchupMicro 1.0,timmit.nano.TimDog 0.33
3941 :: 1300 :: 260

From 1.2:
Fighting battle 0 ... stelo.MatchupMicro 1.0,wiki.mini.GouldingiHT 1.0
wiki.mini.GouldingiHT 1.0 :: 2963 :: 1100
2936 :: 1050 :: 210
I couldn't say immediately if this is a true "bug" in 1.2 or if the RRAH client just had a dependency on something that wasn't really guaranteed to be a certain way in Robocode, but we may need to update one or the other to get them to work nicely together again.

-- Voidious

For each batle in meleerumble, is it fought only once? Because it's taking way longer than a 'normal' battle on most of them...unless it's because i've never seen a battle that lasts that long --Starrynte

It's only fought once, but there are 10 bots to process instead of 2 (in 1v1), so each individual battle will take longer. Each battle counts as 9 battles for each bot, though, usings its score % vs every other bot. -- Voidious


Do you think we could change the server over to redirect and use my LRP, my new host as 16 times more bandwidth, so no worries. The other page didn't wanna display this. I think the wiki is having problems. --Chase-san


I have a problem running nexus.Prototype 1.0 with RR@Home, the probable cause seems the same as with the first version of QuantumChromodynamics, namely Java 1.6. See the remarks at the end of the QuantumChromodynamics page. -- GrubbmGait

Java 6 is the current standard for most everyone, but you gotta remember that must of us just never really get around to updating, and that robocode itself is built on Java 5.0, I myself have the jdk (and the jre of course) of 1.3.1, 1.4.2, 5.0 and 6 installed along with their documentation to give me flexability on my builds. --Chase-san

Well, Mac OS X is still only up to Java 5, so I would strongly protest making Robocode require 1.6 until I can actually run it. =) -- Voidious

Seems only I use the Sun defined numbers. 5.0 had a .0 and 6 they removed that so its just Java 6 now, funky but interesting. They just gotta complecate things huh. --Chase-san


I think someone's client is returning bad results, this score just is not possible (especially over 2 battles). From PowerHouse2's details:

  pe.[SandboxDT 3]?.02 | 99.9 | 2 | 26-4-2007:12:4 | 52.7 | 47.2 
--wcsv

That would be me, I have been running RR@Home with 1.3Beta, and for some reason SandboxDT is getting an exception. It should have a score of 0 though, marking the result as invalid. The bot Kabuto is getting other strange results, a lot of 0.0% and 0.1%. In this case I am not the cause. Tonight I will try to see what is going on. -- GrubbmGait

So who else is using Robocode 1.3.5 as their RoboRumble client? So far, I haven't noticed any drastic rating fluctuations or anything, which is good. I would like to try re-releasing one of my bots and having most battles processed by 1.3.5 just to semi-confirm that it's basically "compatible" with older versions. (For those that weren't around, 1.1.4 and 1.1.5 were a little unstable and seemed to cause a 10-20 point rating boost for new bots processed by them.) -- Voidious

I am currently using 1.3.2 as client, without any problems. I had no time to upgrade to 1.3.5 yet. -- GrubbmGait

1.3.5 is my Rumble client, I'm using 1.3.2 for dev. --Skilgannon

Is it safe to use the Rules class now? -- Skilgannon

It seems that the PL is only updated once every 24 hours. Am I right? -- Skilgannon

jgap.JGAP15300_1.0 has 120 battles vs simonton.micro.WeeklongObsession 3.2.2. Do you have a clue how this could happen? --Krabb

Hmm, my client runs most battles with a JGAP twice or triply. It looks like this:

OK. Krabb.sliNk.Garm 0.8o vs. tcf.Repat3 2 received
OK. Krabb.sliNk.Garm 0.8o vs. elloco.Kabuto 0.2r received
OK. stelo.Wang 1.0 vs. jgap.JGAP130166 1.0 received
OK. stelo.Wang 1.0 vs. jgap.JGAP130166 1.0 received
OK. rz.Artist 0.2 vs. jgap.JGAP130166 1.0 received
OK. rz.Artist 0.2 vs. jgap.JGAP130166 1.0 received
OK. rz.Artist 0.2 vs. jgap.JGAP130166 1.0 received
OK. lk.nano.Avesnar 1.1 vs. jgap.JGAP7958 1.0 received
OK. lk.nano.Avesnar 1.1 vs. jgap.JGAP7958 1.0 received
OK. lk.nano.Avesnar 1.1 vs. jgap.JGAP7958 1.0 received
OK. ar.QuantumChromodynamics 1.2.1 vs. jgap.JGAP130166 1.0 received
OK. ar.QuantumChromodynamics 1.2.1 vs. jgap.JGAP130166 1.0 received
OK. bing2.Melody 1.3.1 vs. jgap.JGAP6139 1.0 received
OK. Krabb.sliNk.Garm 0.8o vs. amk.ChumbaWumba 0.3 received
OK. pi.Dark 10 vs. jgap.JGAP7958 1.0 received
OK. jgap.JGAP6139 1.0 vs. gg.Wolverine 2.0 received

May be its because of his strange name (many mumbers) or because there are many JGAPxxxx versions in the rumble? Thats the only possibility i can think of. --Krabb

Actually, the issues are unrelated. =) I believe the 120 pairings vs WeeklongObsession (based on something GrubbmGait once told me, same situation) is because JGAP had faced everyone but still needed to get up to the BATTLESPERBOT set on a given client; sometimes, a client will fall into a loop against the last bot that JGAP had faced and just run that battle until the bot hits the minimum battles. As for the uploading more than once, that's because JGAP is a MicroBot; for a match vs a MegaBot, it's only uploaded to the General rumble, while for MiniBots or MicroBots, it will upload two or three results (of the same battle), one for each rumble. -- Voidious

Thanks! --Krabb

I must unlearn what I learned in the past. My previous bots only went up against other nanos and were scored thus. In this competition, quite often I am up against Melee master bots who I cannot win against in close combat. In these cases, it is more important to be able run away and "hide" in a corner. This gets survival to around 5 bots left. It then becomes important to have movement that is not to easily pattern matched against from afar and to really save power to try and get down to the last 2 or 3 bots. This is the mistake I made with DustBunny 3.2 - it ran towards the center with a few bots left to attack the remaining bots. It then preceded to get pummeled by superior aim routines. Head on Aim just doesn't cut it at close range :)

I now understand why MeleeSeed is such a dang good bot. It gets to a corner immediately and has a movement pattern that is just unusual enough that most true melee bots won't be able to easily lock onto in 35 rounds. It also is very good at spreading out over a large area to foil no lead and full linear lead bots. It also has a decent radar lock to target the closest bots, which Kawigi's bot is a bit deficient. My bunny has hard to hit movement, a great radar, but tends to attract too much attention to itself whilst wandering around in and around corners. It also has a lame 1v1 pattern which gets it killed by nearly any bot at the end. My work is definately cut out for me here. --Miked0801

So I reverted back to the previous DustBunny (3.1) and behold, it causes Infinity to take the lead!?! Um, ok. Go underhanded tactics I guess. --Miked0801

I have noticed that new versions of nanos are only fighting other nanos now. Is that normal? -- Miked0801

Is it allowed to save data to file between battles in roborumble? is there a memory limit? -- Marco M B


Somebody's client is removing all (or most) of the bots from the rumble....check your config files, guys! -- Skilgannon

I think you missed some discussion on RoboRumble/RankingChat =) It's actually on its way back towards being fixed now. -- Voidious

Ah - thanks. I've been out of the scene for a few days (just moved house, and the DSL line took a while to catch up). -- Skilgannon


I am trying to run a RR@H server to locally compare a set of robots, with little success. Is it possible to put local addresses in roborumble.txt instead of web links/repos nums? -- BigPete

Solved. For others, the address in roborumble.txt is only used if the jar can't be found under ./robots -- BigPete


Wow so many robots needing battles (2 of them are mine), i'm firing up my client and i'm gonna let it run for a nice long time. (luckly its on a remote system so it doesn't bother me) --Chase-san

serenity.nonSense_1.36 test crashes my client with an IndexOutOfBoundsException?, junkly its on a shell loop, but it I want to ask that either the owner fix it or remove it. Its good to see a old robocoder back and about though. --Chase-san

It works once in awhile, thought it does often, meaning that I get a nice long buildup of battles to upload when it actually manages to do so, so i'm gonna watcha movie while this run (on a high end core 2 duo with plunty of ram and no GUI to load it down). I hope this can catch up the battles. --Chase-san

I didn't expect to get the 3rd degree, and when I did well, it feels bad, its not something I enjoy, so i'll just leave for awhile and perhaps in time people will forgive and forget, or atleast forget. My greatest apologizes to GrubbmGait, your probably right after all, you all are, i'm just a Rookie who hacks things togeather afterall. Maybe in a few months, I might have something worth contributing to the community after I correct my attitude. --Chase


Testing RoboRumble on my own pc with packaged teams gives some interesting results. When running a team which saves data to a file the robots don't work, but the jars work fine with robocode! Any ideas? -- BigPete


I was hoping someOne could help with getting my bot in the roboRumble /melee ... it says to take note of the "id" # when uploading it to the Repository , but I can't locate it :(

- my bot is jlm.javaDisturance. -Justin (jlm0924).

You go to the [[Robocode Repository]] and click the "Bots" link, from there, find your bot's page and look at the URL which should be something like "http://www.robocoderepository.com/BotDetail.jsp?id=3645", and the id number there is the number. In the case of jlm.javaDisturbance 0.59 it's 3645. -- Rednaxela

I didn't realise it was the url .. Thx Rednaxela

I added my bot to the Melee / participants list above .. several hours ago. I don't see it the melee /general rankings. Was there something else I needed to do ? Not sure how this works, thx guys .. -Justin

Well, now it is in the [[Rankings]]. There wasn't anything you did wrong, you just had to wait longer. RoboRumble battles are run not on the server but on the computers of individual robocoders who donate some CPU time to the task. If you want it to show up quicker, you could always run the RoboRumble@Home client yourself too :) Also note, rankings take longer to show up for Melee than for the other leagues, because melee battles take longer to run, and because I suspect most RoboRumble clients are not running melee as well. One other quick note, is that don't take your initial ranking too seriously yet, generally rankings don't stabilize until a bot has reached a decently large number of battles. Some of what I said you may already know, but I'm just making these points just in case. Best of luck in your Robocode endeavors :) -- Rednaxela

How do you get the RoboRumble to loop after sending the results... --Starrynte

Type this in the .bat file:

:run
java -Xmx512M -Dsun.io.useCanonCaches=false -cp libs/robocode.jar;libs/codesize.jar;libs/roborumble.jar roborumble.RoboRumbleAtHome ./roborumble/roborumble.txt
goto run

or set "ITERATE=YES" in roborumple.txt

I hope that helps --Krabb

Thanks, now how do you get roborumble to "focus" on a bot, if it can? --Starrynte


To the author of as.xbots: The filename should be <as.xbots_1.0.jar> (note the underscore instead of space). Although Robocode itself has no problem with the current name of the jar-file, RoboRumble is not that intelligent and can not handle it. And ofcourse welcome at the wiki, it is always nice to see a fresh rammer trying to take the ram-throne. Take a peek at RamBots and/or RambotChallenge2K6. -- GrubbmGait

Robo Home | RoboRumble | Changes | Preferences | AllPages
Edit text of this page | View other revisions
Last edited November 8, 2008 23:52 EST by GrubbmGait (diff)
Search: