[Home]RoboRumble/ParticipantsChatArchive20070918

Robo Home | RoboRumble | Changes | Preferences | AllPages

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

Changed: 1c1,585
oh...would someone plz paste the archive (using diff) since i can't due to the links or something...
Vic, you were right. There are other bots with two versions in the RR :
gjr.Cephalosporin 0.2 , kvk.HebusLeTroll
i dont know who these bots wrote but please remove one version you like so that only one stays. --deathcon

Of course I'm right ;-p Actually I observed that this happens also with more current bots. But I found out this is a glitch in the RR system where sometimes previous versions of a bot magically reappear and after a while disappear again. So if I'm really honest I should say I was wrong. Which I just did. --Vic

Well the two bots posted above stand both in the participants list with two different version and i want the authors to remove one version. What you mean comes from the RR thats true. RR cant delete old versions from the ranking, that is automatically done by clients. But sometimes it happens that the old versions appeares for some time, too. afaik, this comes from running the client with the option download=NOT . --deathcon

Almost. But it's more like "upload_results=NOT" I think. Never mind those, they dissappear once a client starts a new iteration. I think I counted to four or five CassiusClay's the other day. =) -- PEZ


Now 5(!) bots that have not yet 500 battles fought. We need more clients :D --deathcon

Funny. And just two months ago or so I thought maybe Robocode was half dead. =) -- PEZ

*Maybe we all were paralyzed by RaikoMX slaughtering all of us .
Do you remember the title of the eternal rumble site : ROBOCODE FOREVER !!! --deathcon

I can't think of a better description of the EternalRumble. -- Kawigi



Removed wiki.CigareT? - i created this bot one day (cigaret with a gun of mine), but it was not really successful. Added BoxingVirus. Shivas movement with CCs gun. I dont really want this bot to stay in the rankings - just want to see how well it performs and find out how good my movement is.I will it remove when it reached 600 battles. --deathcon

Ok, now i removed BoxingVirus again as promised. --deathcon




I'm leaving two computers running RR clients through the night (10 to 11 june). My bot has only 200 more battles to fight, so there's plenty of room for other updates this night :-) --Vic

Added new version of Shiva and BoxingVirus. BoxingVirus will be removed when it gets 600+ battles if it wont become #2, which i dont expect it to do :D --deathcon
* removed again. --deathcon

I pulled out SilverFist. At the moment it only proves that if you blend bot #2 and #3 you can create a new #2 bot. =) We'll assemble a new version soon when Axe finds some time for it. -- PEZ




I have removed two blank lines:

...
davidalves.net.YALT 1.3,2084


dft.Axraelic 1.25,1244
...

The client was throwing an error and refusing to download bots because of that blank lines... -- Axe


Not really about participants, but i cant think of the right place to ask. What does the total%wins column in the premier league table mean?? I'm guessing it relects the ELO score, but im not sure... --Brainfade

jeje ... I can't remember. Did I place it there? Anyway, it's not used at all to calculate the ranking. PL works just like a real world sports league: 2 points for the winner, 0 for the loser. The bot that beats more participants wins the league :-) Albert

I believe you've said in the past that it's the tie-break condition and basically a reflection of how much they usually beat opponents by. Mathematically I'm not sure what it is (average % of score?). -- Kuuran

Yeah! That's it. I'm getting old and my memory gets worst :-) You are right. It is calculated just as the sum of the % of score you get agaist each enemy (it like "goals"). -- Albert

I pulled Resin because it caused out-of-memory conditions. -- PEZ

Added a new Resin (version 0.3). Please feel free to remove it if it causes problems for your client. -- PEZ


Someone added the SampleBots. Is that serious? -- Jonathan

I removed them from 1v1. They don't belong there. Dunno if they should be left around in melee though. Not too crowded there. -- PEZ



I removed a duplicated "shinh.Entangled 0.3,1070" line in MeleeParticipants?. Was it getting double the number of battles than the other bots? -- MN?

I think not, but not sure -- Albert


Should BeeRRGC still be in the rumble? -- Alcatraz

No, it shouldn't IMO. It's in the way for lower ranked bots. I'll remove it. -- PEZ


Can I get the US flag for grybgoofy? (I re-added GoofyBot 0.10) --Goofy


The number if 'wiki' bots is increasing again. Isn't it time to remove some of them. For example, there are currently two versions of DevilFISH in the rumble (and i hate this bot ;-) ) --Loki

But one of them is a nano... -- PEZ

Mmmm right. But both of them are a pain in the ##. It was a nice try... --Loki

Though they _are_ both behaving and performing very similar (the 5 point rating diff could just be random noice). We could remove the non-nano one is my vote. Not to help you with removing a problem bot, mind you. But we need to keep the list reasonably tidy as it has grown rather large. I'll do it right away. -- PEZ

Goofy, update the RoboRumble/CountryFlags table so that David Alves can prepare your flag in the new RR server. -- PEZ

why are more and more people using direct links to their bots? The RobocodeRepository is still up, isn't it? --Loki

It wasn't the other day. I think I downed it when i cleaned my robot jars. And some bot jars are just too big for the repository. -- PEZ

That big bots is SS! :) Repository denies uploads to bots bigger than ~250Kb... And some editions of my Godzilla Bot actually go bigger than that. -- Axe

@PEZ: (U probably knew that i was going to complain about it, dont u?:) Why? Marcelo.DevilFISH is similar, but not the same bot... -- Axe

It's the same functionality packed in different jars. They behave the same way. They perform the same way. The list of participants is growing fast enough anyway. The only difference I can see is that one is a mirco and the other is a nano. Removing the nano would mean no DF in the nano rumble so I figured let's remove the other one. Check this comparison if your still not convinced "it's the same bot": http://pezius.com/rumble/servlet/RatingDetailsComparison?game=microrumble&name1=Marcelo.Alpha.DevilFISH%201.3&name2=wiki.nano.DevilFISH%201.0 -- PEZ

Your word is enought to me. If it is the same, surelly u r right. I'm in mourning :(... -- Axe

Why would you miss the duplicate DF? -- PEZ

Sentimental reasons... I grew fond of Marcelo.DevilFISH. :) -- Axe

Where did PulsarNano go? -- Alcatraz

According to the revision history Pulsar removed it himself. But it was in the same edit where he updated PulsarMax so probably it was a mistake. I'll put it back. We can't have a nano top-3 without a Swedish flag, can we? =) -- PEZ

Hi guys
I have a little question, How can i put the flag of my country on my bot? (like a lot in the rankings) -- F4kill

See /CountryFlags and PEZ will fix it as soon as he sees it. And welcome! -- Pulsar

Thanks :) -- F4kill


Removed the following bots because they are not packaged according RR@Home standards.
b0lt.Obliterator,2567 --> This is not a jar file
rpf.RPFRobot 1.0,2577 --> This bot misses a <name>.properties file in the jar
-- GrubbmGait



@Krabb,
is it correct you entered Doboh v1.2 in the RoboRumble/Participants and v1.3 in the ImplementsDroidCompetition/Participants? Shouldn't they both be v1.3? --Loki

Removed amk.advancedstrike.AdvancedStrike 0.5,2601 because it could not be downloaded.
Removed kinsen.ScanFire 1.0,http://www.geocities.com/seakinsen/kinsen.scanfire1.0.jar because it could not be downloaded.
Changed Krabb.doboh.DoBoh 1.2 to 1.3 because "Downloaded file is wrong or corrupted:Krabb.doboh.[1]?.2.jar". --Loki

Thanks, Loki! --Krabb


Hey, why isnt the Sample team in the teams competition? -- Jokester

Because SandboxLump can beat it (not a team of SandboxLumps, but just SandboxLump). -- Kawigi


Just thought I'd give ya guys a heads up. Dont expect to much though, I've just come back again after loosing all my previuos code.(damn HD's they should warn before they crash) Jimpa



I don't feel it's my place to remove florent.Froggy 0.1.1PM but I hope its author will consider it. It pegs my CPU and stops the RoboRumble. Not a single result has been uploaded since its arrival. Sorry, but ratings will cease if this bot stays. --Corbos

Just do it.. I think if a bot disables the RoboRumble you are allowed to delete it! --Krabb


To the author of florent.*: please quit uploading versions so quickly - especially the versions that take forever (version 1.6 of FloatingTadpole is slow, slow, slow). If this continues, I see no reason to run RoboRumble. RoboRumble has become the florent show. That sucks. Please test your bots and compete with stable and significant changes. If you're careless, you can ruin this for everyone. --Corbos

I'll remove it now and upload a new one in a few days when it will be quicker --florent

Florent, After reading what I wrote yesterday, I see that I owe you an apology. Your bots are getting better and better and I don't want you to remove them from the Rumble. They provide excellent pressure for Chomsky. Both FloatingTadpole and Froggy beat him. I was just frustrated that by the time one version's rating stabilized a new version was uploaded. It felt like my client spent all of its time stabilizing your bots. Can we strike a compromise that keeps your bots in the Rumble? If you keep at it, you're going to give those untouchable top bots a run for their money. Please accept an apology from this crabby old code monkey. --Corbos

There is no need for an appology, I just spend to much on my bots now because I am on holliday, so as soon as I have implemented something that seems to be good enough to gain a few places(dynamic distancing, fixing a huge bug in my waves, a forward pattern matching gun (that is slow and not nearly as effective as i supposed it would be)) I upload it. I will try to test more my bots before uploading them and clean the code of unessary part that tend to slow down the bot (PM gun in a VG array). --florent

Dunno what it is about today but there have been several bots submitted / updated. I'm running rumbles on 4 cpu's trying to catch them (and particularly Ugluk) up. I'm pretty stoked too because Ugluk had a score of 1700 after 95 duels and it keeps climbing. -- Martin

Yep, I noticed the surge myself... Perhaps everyone's thinking the same thing I was: I want to get a new version up tonight since I won't be working on it for probably a week or so! (I'll be running the RoboRumble momentarily here, and then all night, FWIW.) And congrats on the ratings hike, Martin. -- Voidious


Loki, congratulations with the entry of bot number 400: bvh.micro.Freya 0.1 -- GrubbmGait

thanks for the congratulations! but as it holds almost the 400th position, i don't want to draw to much attention to this bot :) It is not designed for 1-vs-1 though, so let's see where it ends in the melee ranking. I just wanted a competitor in the micro melee rumble and ended up with the minimum risk movement of mini.Freya and a simple gun with zero bytes to spare. Maybe i can find some extra bytes for a better gun. --Loki

Bayen, my RR@Home client said it couldn't download your SamSpin? .jar file. I tried downloading it from the Robocode Repository, and noticed that you have it as "...Samspin...jar", with different capitals than the title of the tank ("SamSpin?"). That might be the problem, although I'm not completely sure. Also, the general convention is to always use the same top level package for all your tanks, so you may want to change it to 'bayen' when you get a chance. -- Voidious

Actually, Samspin is not my robot, it is my brother's. joe.ADinosaur is my other brother's. So that explains the seperate packages. I think the captialization is the problem, thanks. --Bayen

Is there any logical explanation as to why Dookious has faced Shadow 49 times? (Dookious details page.) I don't know what that would do to which rating, but Dookious is overperforming against him, in any case. I imagine it would bring Shadow's down. ABC, should one of us rerelease? -- Voidious

It does not have an influence on the rating, except being more stable. Once you have fought one battle against someone, the results of the next battle against the same opponent are compared against the (weighted) previous results.
If your first battle was 60-40 and the second 65-35, you will slightly go up and the next 'expected' result would be approx 62-38. If the third result is 60-40 again, you would slightly go down, even if the other one is higher ranked. Only for the first battle the expected result is depending on the rankingdifference. After that, the previous results is the guideline. About facing another bot a lot, it can happen when it was the last fight to complete the priority battles. Then sometimes that same battle is fought as many times as the number of battles in RR@Home is set. -- GrubbmGait

Ah, OK, cool. Thanks for the info. -- Voidious

Out of curiosity I just checked the average ranking of all participants, and it came to 1591.61, pretty close to the 1600 'average' mark. I imagine this is influenced by people swapping out new versions of bots frequently, yet it is still holding relatively steady. -- Martin

Sorry about the double entry for PowerHouse tonight guys, I will remove one of them as soon as they stabilize. I know I should only have one version up at once, if its a problem I'll remove one right away. --wcsv

Whoah, does someone have PulsarMax set to TC or MC mode in their RoboRumble directory? There's no way Dookious could ever beat him by a score of 90%. (Dookious details) What should I do about this? -- Voidious

I ran a few battles to test this, Dookious is a very strong bot, but I don't think it's quite *that* strong (yet). (My tests show scores of 45-50% vs PulsarMax) --wcsv

Unless someone can point out why I shouldn't, I'm going to set my "battlesperbot" to 2,000 and run RR@Home for the rest of the day. (I'll be gone, anyway.) Hopefully that will mean Dookious fights PulsarMax at least one more time. -- Voidious

Hm.. Voidous brings to light a growing reliance on Mohammed Ali references. Perhaps we should branch out with StampedeGoneIn91Seconds, GrubbmGrill (knocks out the fat), ChomskyEar (has a nice ring to it), and the obvious .. ShadowBoxer. -- Martin / Ugluk

Not sure what you mean about growing reliance. But anyway, the latest too Dooki-bots are tests with CassiusClay and Dookius guns and movements used together. -- PEZ

I was attempting to be funny. -- Martin

Hehe, yeah, I was with you that far dude. =) -- PEZ

I think it would be funnier if I could tell what combinations of bots those are supposed to be, Martin :-) If anyone is annoyed at the extra entries, feel free to remove them or set your "BATTLESPERBOT" to less than how many battles they have. -- Voidious

Well the 91 seconds thing is how long one of the Mike Tyson fights lasted. Imagine buying tickets to that fight. On one hand, you saw a legendary fight. On the other hand .. it only lasted 91 seconds... -- Martin (who didn't really have any movements / targeting in mind when coming up with the hybrid names)

Hey guys, sorry about the onslaught of Dookious releases this week. I won't be making a habit of it or anything. I'm taking a short break from Dookious to explore some other stuff after 0.71, and I wanted to give a good push at hitting 2k first... -- Voidious

It seems to me that the Rumble gives priority to higher ranked bots, since Shadow has processed more battles and PowerHouse is close, even though they were both released 6 hours after Ugluk. Ah well... -- Martin

Well, don't worry about releasing loads of versions in sequence. Look at CasssiusClay/UpdateHistory? and you'll see that this has been done before and in my book it is how it should be. It is why Albert gave us the RoboRumble@Home to begin with even. Martin, no priority is given to higher ranked bots. But randomness and timing of releases and different settings on RR@H clients and stuff like that sometimes give tilted battle counts like that. -- PEZ

I have had my 'battlesperbot' set on 500 today, to boost the relatively large number of updates. It seems that this was not beneficial to both Ugluk and ds.Oo. Such a low number for 'battlesperbot' does help to fill the PL-ranking quicker. -- GrubbmGait

I have my suspicions, but is there any plausible reason why my client is still processing pez.mini.Pugilist long after his battle count has passed the BATTLESPERBOT threshold? It is happening on at least two machines where the setting was changed to 500. Meanwhile Pugilist has 618 battles, he's in well over 50% of battles, and there are around 7 other bots to process... -- Martin

It has to get up to 500 in the MiniRumble?, too, where it's still under 300. -- Voidious

Yeah. Be happy Pugilist isn't a NanoBot! =) It sure is a SlowBot. Appologies for that. But it's hard to get surfing and statistical targeting into a mini and some corners are definately cut in the case of Pugilist. -- PEZ

Out of curiosity, how many people put bots in the RoboRumble without running the RR@Home client? As I can see there are a lot of bots to stabilize, I'll run my client for at least a little while here. I suppose watching the Lakers could take the place of Robocode for a little while =) -- Voidious

I do not know how good a litmus test it is, but I assume that the people who commonly discuss / comment on things are all running the Rumble when they have bots that have not reached the plateau, and sometimes when they don't have active participants. I suspect that people who don't feel as involved are less apt to crunch the numbers. -- Martin

I used to have it running 365/7/24. But lately my machine has some problems making it restart itself now and then like twice a day or so. So now I mainly run it when I have a bot I want to see the rating on quicker. And sometimes I don't run it even when I have bots that need battles. This because running robots seems to make my machine restart more often. But think of it like this. Now it never takes a full day to get an answer on how strong a bot is even when quite a few bots are updated. Back in the days of the EternalRumble we had to wait weeks. And we never knew when the next ratings where going to get determined. Sometimes it took months. If you're really eager to see if you have improved your bot, te TargetingChallenge2K6 and MovementChallenge2K6 are there for you. If those shows improvements it is pretty much certain you have a better RR@H rating coming. -- PEZ

I'm definitely grateful for the RoboRumble setup, and I have no complaints. I definitely rely on the TC and MC challenges, as well as a combination of stable tanks and ProblemBots, to benchmark my tanks before release. -- Voidious

Once I realized that I screwed something up with the movement, I prepared 0.75b with the intention of posting it in the morning. Anyway, got up in the middle of the night and noticed 0.75 was past 1k at 1989, so I went ahead and posted 0.75b. It restores the movement of 0.72, leaving only gun changes. (I was getting pretty good about only updating gun or movement at once; I shouldn't have pushed my luck!) -- Voidious

It isn't so much about luck as it is about knowing what you can deduce from a set of changes and their results. Even if you would have ssen an improvement of 10 points you couldn't know if it was really a movement improvement of 20 points and a deteriation of the gun of 10 points. Unless, of course, your local tests have showed something that can be of guidance here. The important thing is to test gun and movement changes seperately. Says someone that releases loads of bots with changes everywhere... -- PEZ

Why "mrm.MightyMoose .1" doesn't participate in RoboRumble? This bot has been downloaded successfully! It is in list, but doesn't fight. Why? -- DemetriX

First of all, the name is not correct, the bot is named "mrm.MightyMoose 1.0". Secondly, this bot does nothing except keeping a radarlock. Therefor it gets a score of 0 (zero) against xxx, which is not accepted by RoboRumble. If it was named properly it would only consume valuable processortime without any result. ad.last.Bottom always scores 1 point against anybody and therefor it is a valid entry (and also last in the ranking). -- GrubbmGait

OK. The problem was simle. This bot has version ".2", but in the list named ".1". I dare to correct this and now it's all right! Thanks for helping me to solve this problem. :-) -- DemetriX

No... It fights normally. Just bot had wrong version in participants list. -- DemetriX

Then it is already corrected. I tried it the first time I saw it was an invalid entry, and then the bot did nothing. After that I have not checked if a new .jar was placed there. The .jar file has no version, so there is no way to check that other than to try it. This wiki is intended to help others, this includes fixing small mistakes and typos. Thanks for correcting the entry! -- GrubbmGait

I have been looking at specialization indices and only 12 of the top 25 robots have a spec. index under 20. They are also all bots that have been around for thousands of matches. The venerable SandboxDT has the lowest at 6.8, with Quest at 9.5 and RaikoMX at 9.8. Sandbox's main problem bots are ram bots, but even then they are not serious problems. -- Martin

It's worth noting that there's something wrong with the SPI calculation - according to Excel, the average squared value of DT's PBIs is 29.513. Really old bots seem to have this issue; one of us really should fix that script sometime... -- Voidious

Sorry, I know that Dookious is among the tanks that isn't stable yet, but I really want/need to use my CPU without the RR client running for a bit. I let it run all night and most of today (until it hung), but man, X2 is just incredibly slow (sorry Florent, I know you know that), and I think I was the only one running the client when I last was. I'll try and run it for a bit later in the evening, and certainly during the night... -- Voidious

Hey Florent, that was very thoughtful of you, and I'm sure wcsv appreciates it... Do you want me to restore 0.11 to the list when I start my client later this evening? -- Voidious

Thanks Florent, but you didn't have to do that. I know i've been making a lot of updates, so I don't mind running X2 for a while. Thanks again though, the client runs slowly on my machine even when i'm running relatively fast bots. --wcsv

There's no need to restore 0.11, I'll published a new one when I can get a faster version of it. -- Florent

By the way, PEZ told me it was cool with him if I posted a version of BeeRRGC with the latest Bee gun. For some reason, the data reloading was failing every time on my machine, I'm guessing because the gun is in a different base package than the bot itself; I remember in this situation with a version of DookiBee? my RR client crashed the bot every time, so I just disabled the VG stats stuff in Bee for this version. Shouldn't make a huge difference, and it's noted in the version number. -- Voidious


how do you think about removing the wiki.*-bots from the Participants list (and MeleeParticipants? list)? They are often experiments, e.g. a combination of gun and movement from two bots. Some are removed after some time by their creators, e.g. wiki.Tyranius, others remain in the rumble for ever. I am in favour of removing them. what do you think? --Loki

What about ones that aren't experiments and are in fact collaborations between two authors? I would leave those in. However, I've been wondering about the legitimacy of SilverFist as a KingOfRR?. If SilverFist is listed there, why not Tyranius? --Alcatraz

I think the main difference between SilverFist and Tyranius is that PEZ and Axe took the throne from ABC with SilverFist, while Tyranius is a combo of the two current top tanks. I don't speak for mue, but it seemed like we both just wanted to post it as an experiment, and not have it as a throne contender. That said, though, I don't personally have any problem with it being listed there if others think it should be. -- Voidious

I guess I also am partly wishing that SilverFist hadn't just disappeared after losing it's title. It felt very much like an experiment to me. And in regards to Tyranius, I just find it interesting that there exists a bot stronger than the king. But this appears to the mythology-lover in me that envisions Tyranius as the mysterious warrior who out of nowhere came and usurped the throne and then went just as quickly back into the night. -- Alcatraz

i am afraid it is not as easy as a the distinction between 'experiment' and 'cooperation'. Most wiki-bots are a cooperation and most cooperations are also an experiment, take for example again Tyranius. But some wiki-bit authors decide to remove their entry, while others don't. If this is a list with all bots ever made, then all wiki-bots may remain listed. SilverFist is a good example as it also was a succesfull entry. So maybe i try to determine some criteria when to keep wiki-bots and when not.

We already have the first criteria: ;) --Loki
* personal choice and
* freedom of entry

Well, personally, I do try to keep my entries in the rumbles pretty clean, which means removing such experimental tanks. (I might even remove Shaakious.) However, since it's such a common practice for tanks to build upon one another, I'm not sure I'd feel right removing such combo tanks without the authors' permission. I do understand where you're coming from, don't get me wrong... something like BlackDestroyer, which is outranked by each of the tanks that it was a combo of, does seem like a good example of one that maybe should be removed. And the whole combo tank thing does have the potential to get out of hand. But I guess I'd still say the author(s) should be the ones to do it, and the resulting may still be an interesting and unique tank to have in the rumble. -- Voidious

I agree that a cleanup of wiki.* bots can be usefull, therefor I removed the unsuccessfull experiment ARAMtocles. I will check if some of my bots still have some added value (like GrubbmTwo) and if not, remove them also. With more than 400 bots in the rumble, any reduction of the number of bots seems welcome. -- GrubbmGait

It's up to the authors of the wiki.* bots of course. If they have left them there I can't see that it means anything else than that they want them there. Some bots might just have been forgotten by their authors, but that goes for a lot of bots. Rather than focusing on wiki.* bots we might want to shout out somehow that people check the participants and question if all their entries have reason enough for being there. Personally I have removed more bots from that list than most of you have ever entered. =) I feel like the wiki.* bots in the list that I am involved in belong there. And of course Tyranious should have an entry in the KingsOfRR? table. It would add to it's mystical features too. Where's T now and what's she planning? =) -- PEZ

By the way, there seems to be a pretty high number of bots to process in the RoboRumble right now. (7 under 1,000 battles, 5 under 500 battles, 3 of them <= 10 battles... and I want to post an RRGC test tonight or soon.) I'm setting my BATTLESPERBOT to 500 and running as much as I can tonight and tomorrow. To anyone that doesn't run the RR@H client, it's very helpful to have as many people as possible running it to get the rankings updated, so check out the /StartingWithRoboRumble page if you can spare some CPU cycles. Martin also posted his RR@Home settings to the /RankingChat page, if that helps in getting it setup. -- Voidious

My harddisk crashed last weekend, but this weekend (hopefully this evening) I will be up and running again. -- GrubbmGait

This RR@Home situation reminds me of an idea that's occurred to me from time to time. I'm not sure how practical it is, and I'm not sure if it would bring enough attention to our RoboRumble to invite security risks (ie, sending bad results), but... I think it would be really cool if we could make a screensaver-type program that would run Robocode battles for the RR@Home and display them on a user's screen (a la SETI@Home). It would actually be a lot slower than our RR@Home clients, since it wouldn't be running minimized, but if it were used by even a relatively small number of people it could really help with the processing. And, IMO, it seems like it might appeal to some people as a cool screensaver. -- Voidious

How about caching/saving the battles via some sort of 'recorder' and playing them back without the burden of real-time processing? Maybe important battles (surprise victories, etc) would rank higher and re-run more often. --Corbos

Well, I was also thinking of this screensaver as being an active RR@Home client, so it would also help with the RoboRumble processing. Or do you just mean it would run battles in the background, but display different battles as the screensaver aspect? (Robocode already can record/playback battles, I think, although I've never actually tried it.) I do think that any kind of Robocode-based screensaver would be really cool, but I was more specifically talking about it as a RR@Home client, as well. -- Voidious

I was thinking graphic-less real-time processing for the client (very fast) with recorded playback in the screensaver (also very fast with some dramatic value). I'm always looking for the best of several worlds ;) --Corbos

Kev, if you want to hit my GMail account with your latest update (voidious at), I'll gladly host it on my web space for you and re-post it to the rumble when I do. -- Voidious

Thanks a lot Voidious, I would really appreciate that. I'll send the latest version of Vyper to you now. -- Kev




My friggin' computer crashed pretty early in the day while I was at work today, so unfortunately, I missed a lot of potential RR processing there. By the way, I edited line 206 of roborumble/netengine/BotsDownload?.java to fail on any RobocodeRepository downloads, otherwise it's a lot slower to process right now with the repository down. (Basically, just "return false" instead of creating the repository URL on that line. You compile it from the 'robots' dir, and include robocode.jar in the classpath.) -- Voidious

Thanks for the tip, now I can run on full speed again (although full speed is not that fast on a P3 1GHz) -- GrubbmGait

Gah, my internet connection is so flaky... If Dookious is holding up anyone else with tanks that need battles, feel free to revert him to 1.06 temporarily. I won't be contributing RR@Home battles until I get home again this evening. -- Voidious



There are some other bots in the rumble that may be questionable: Albert's bot apv.test.Virus is literaly based on jam.RaikoMX.


public class Virus extends jam.RaikoMX {
.. snip ..
}


I think extending this/a strong bot is a clever idea, but as the package name describes: "it's a test". And it has lasted long enough in my opinion.
--Loki

Hmm... That's an interesting point. Personally, I like Virus, as it has a very unique gun. Since RaikoMX is open source, anyway, I kind of think it's fair to use the movement in another tank. A lot of MiniBots take guns or movements from each other - the first version of Komarious had an unaltered version of RaikoMicro's gun. But that's just my 2 cents... If others think it should be removed, that's fine. Hopefully Albert will chime in if he's around. -- Voidious

Using someone else's gun or movement has been done tons of times. Personally, I don't like it much, but I think it is a common enough practice that we can't start ruling out bots based on it. Just don't use someone else's gun AND movement... that's a bit much. --David Alves

I think Virus should stay for two reasons. First, it would change the competitive landscape too much. Virus has been in the rumble since I started. It seems strange to remove him now. Second, he should stay out of respect for Albert. Albert contributed a ton to the wiki content, explored novel pattern matching and neural networks in his guns, and gave us RoboRumble@home! You’d have a hard time convincing me Virus is a lazy coder’s exploitation of the system. Let’s stick with the 'extend gun or movement – but not both' rule. Cheers. –Corbos

It's true that exact movement/gun code has been re-used in many bots and i don't dispute Albert's contributions to the rumble and his robots (think of Aspid, LauLectrik and ScruchiPu!!), i only mentioned Virus because of the definition ("extends jam.RaikoMX") and it's package-name ("test"). I think it was an (excelent!) exercise to show how a succesfull bot can be created based on another succesfull bot. But i think it was an exercise. So I hope Albert still ocasionally visits this site, thinks the same way as i do (and maybe enrich us with some more contributions). --Loki

Anyone know what's going on with Phoenix 0.67? It's got over 5,000 battles, but has only faced about half the RR contestants (on average like 20+ times each). I think if a RR client starts but doesn't finish uploading results, it will re-send all those same results next time, so I figure that must be part of what's going on there... -- Voidious

Um, that's weird. I left my client going this morning and went to school. I was only running one client so it shouldn't be able to run that many battles in ~8 hours anyway. No idea what happened. --David Alves

I figured it out. Apparently this is what happens if you run two RRAH clients that are both using the same directory. =P --David Alves

Was about to tell you that WaveSerpent is throwing errors, but I realize that I just need to upgrade my clients to the new version. Doh!


=========================
Round 6 of 35
=========================
kc.serpent.WaveSerpent 1.0: Throwable: java.lang.NoSuchMethodError: kc.serpent.WaveSerpent.setBodyColor(Ljava/awt/Color;)V
java.lang.NoSuchMethodError: kc.serpent.WaveSerpent.setBodyColor(Ljava/awt/Color;)V
at kc.serpent.WaveSerpent.run(WaveSerpent.java:35)
at robocode.peer.RobotPeer.run(RobotPeer.java:633)
at java.lang.Thread.run(Thread.java:613)


-- Voidious

Kev, you might consider checking if a method is present before calling it if you're going to use methods from the newer versions of robocode. For example:

if(TeamRobot.class.getMethod("onPaint", new Class[]{Graphics2D.class}) == null){
debug = false;
} else {
debug = true;
info("Graphical debugging enabled, click 'Debug Paint' in robot console to view");
}

--David Alves

Thanks guys, I don't think I've ever seen a -80 PBI before :). I re-released WaveSerpent using setColors(), which is compatible with all versions of robocode, so the problem won't happen again. -- Kev

Is there a way to see who turned in which battles? Someone turned in a battle of Phoenix 0.7 vs. KanoGamma where Phoenix scored 56%. This is very very wrong, I've run a bunch of battles and the results are more like:

1st: davidalves.Phoenix 0.7 4455 1700 340 2020 395 0 0 34 1 0
2nd: kano.gamma.KanoGamma 1.8 723 50 10 654 8 0 0 1 34 0

I don't know if it's skipping a lot of turns on that system or what, but something is really screwed up.
--David Alves

A few months ago I mentioned that the average rating was 1591.61. Now it's at 1591.91. A pretty small bump for at least 4 months of rumbles, with at least 20 new bots added. My assumption is that running the Rumble enough times without changing the participants would get it even closer to 1600, but I don't know how long it would take or if it would even have a noticable effect on ratings. It's hard to say how the 8 point average increase would be distributed, though I suspect it would be larger in the middle than at the fringes. In the end it doesn't really matter. I think the Rankings would not be affected. -- Martin

Why do you expect the average tends to 1600? The distribution is centered on 1600, but the function used to calculate ratings is not linear, so if the distribution of the bots on the upper side is different to the distribution on the lower side it is possible is stabilizes in a point different from 1600. -- Albert

Step, are you having problems with packaging robots for the rumble? There is a handy tool inside the Robocode client to 'package robot for upload' that should make things go smoothly. -- Martin

ebp.NanoBruno? is a class file, that type of file can not be handled by RoboRumble@Home. Only jar-files (correctly packaged) are supported. -- GrubbmGait

I removed DM.Titan _3 from the rumble. Please do not use underscores in your version number. It makes the RRAH client have errors:

Fighting battle 12 ... myl.micro.Troodon 1.10,DM.Titan_ 3
Exception in thread "Application Thread" java.lang.ArrayIndexOutOfBoundsExceptio
n: 1 >= 1
at java.util.Vector.elementAt(Unknown Source)
at robocode.battle.BattleResultsTableModel.getValueAt(Unknown Source)
at roborumble.battlesengine.BattlesRunner.runBattles(BattlesRunner.java:
95)
at roborumble.RoboRumbleAtHome.main(RoboRumbleAtHome.java:94)

--David Alves

I don't expect anyone else to do so, but I've bumped the priority battles threshold from 1200 to 2000 for all of my RoboRumble clients. We've nearly 500 bots in the Rumble now and I'd like to have a better chance of getting a consistent final rating. At present only the two bots using the new interface are below the 2000 mark (since I cannot process them), plus very new releases. -- Martin

I'll definitely join you with the 2000 min battles. I often set mine to 1399 or 1599 once everyone's past 1000, but I've been lazy and just leaving them at 999 lately. -- Voidious

Good idea, also set my meleeclient from 1500 to 2000, although I don't run that one that often. -- GrubbmGait

Here's the error message from "Exploit" under 1.1.2, which my RR@H clients are running:


=========================
pedersen.Exploit 1.2.1beta: Throwable: java.lang.IncompatibleClassChangeError
java.lang.IncompatibleClassChangeError
at pedersen.core.Foundation.cheat(Foundation.java:138)
at pedersen.core.Foundation.processTurn(Foundation.java:113)
at pedersen.core.Foundation.run(Foundation.java:59)
at pedersen.Exploit.run(Exploit.java:30)
at robocode.peer.RobotPeer.run(Unknown Source)
at java.lang.Thread.run(Thread.java:613)


So it just gets a zero score every time.

-- Voidious

I built a new roborumble with the 1.2.1beta robocode.jar and it is also getting an error, though a different one. So at least for now the bug in question cannot allow someone to dominate the ratings. The robot is still available to demonstrate the bug though. It is linked from the News page. I pulled the robot from the rumble a few mintues ago. -- Martin

I hate to be a party pooper, but I just have to voice my opinion that it seems like a bad idea to continue testing these exploits in the rumble. If there is a bug in Robocode, it should be fixed in the main program, anyway, and that will fix it in the RR client as well. Putting these exploits in the rumble seems like unnecessary distortion of the ratings of every tank in the rumble, not to mention wasted processing time. (E.g., it starts at 1600, wins 99% against a 2000+ tank, you're looking at a -85 PBI and probably 3-4 RR points lost on just that battle.) -- Voidious

Voidious, I have a very very very good reason for testing chase.Dodger, because unlike pedersen.Exploit 1.2.1beta, chase.Dodger does not use any of the Set commands in the Manager, it uses only get commands to direction every day robot hardware into attempting to dodge bullets. After 100 battles (hopefully thats not to big a waste to test a potential loophole) I plan to remove it. -- Chase-san

I hear you. I do think these things should be tested, I just don't see why putting them in the rumble makes a difference - it uses the same engine to run battles as Robocode itself. Dodger seems to get 0 every battle on my system, so it is going to take a lot of battles before it gets to 100 RR battles. Anyway, the one client I have running is currently hacked to not download Exploit or Dodger... -- Voidious

One of the advantages of having a less competitive bot is not feeling the need to take things so seriously. -- Martin

Apparently I've been doing Robocode for over two years now. I was checking some old emails and when I started a friend and I would email out robots back and forth one-upping each other. The last one he sent was October 10th, 2004. I've held off on submitting it in the rubmle on his behalf, but curiosity overcame me. After making the jar file I decided to test it against Ugluk, which was modified to not move for the creation of the targeting challenge challenger. Even without moving, Ugluk beat him 5800 to 4300 or so. He must not be using 3.0 power bullets. He used to be hard to beat for me. I guess Ugluk has come a long ways in 2 years. -- Martin

Hey Simonton, I have one tiny request - could you use a "real" broken URL instead of what you're currently using for the bots you aren't posting for download? My clients seem to hang on this kind of URL instead of just moving on like they would with a 404 URL. I just hacked the client for now, so no worries, but some others might be hanging on it =) -- Voidious

Oh, sure, no problem. Sorry about that. Would it work if I just put a "http://" in front of it? Maybe it's trying to download a bot with that ID from the repository without the http prefix? -- Simonton

Ah, yes, that's probably what it's doing. (Actually, the client just checks for the "://"...) Just unhacked my client and it looks good. Thanks man! -- Voidious

Hey Incom, your fw.Number bot isn't working in my RoboRumble client because it is a .class file instead of a .jar file. What you need to do to package it is this: in Robocode, click on Robot >> Package Robot for Upload, select your bot, and then give it a version number (and other info, if you'd like). This will create a .jar file in your "robots" folder which you could then upload to the repository. (Note that you can upload a new file for an existing entry on the RobocodeRepository, you don't need to make a new one.) The RoboRumble client only handles these .jar files. Once you've got that done, you should make a page for yourself on the wiki and request the appropriate flag for your bots' package. :) (I found the name Incom? in your RobocodeRepository page...) -- Voidious

* Thanks! I was wondering why it didn't show up in the statistics. I just started with Robocode so I will need some time to figure out how all this stuff works ;-) -- Incom
* I'm still having a problem with the new .jar file - it downloads fine, but I get this error when I try to run it:

SYSTEM: Could not load fw.Number1 1.0 : java.lang.UnsupportedClassVersionError: Bad version number in .class file
java.lang.UnsupportedClassVersionError: Bad version number in .class file
at java.lang.ClassLoader.defineClass1(Native Method)
at java.lang.ClassLoader.defineClass(ClassLoader.java:620)
at robocode.security.RobocodeClassLoader.loadRobotClass(Unknown Source)
at robocode.battle.Battle.initialize(Unknown Source)
at robocode.battle.Battle.run(Unknown Source)
at java.lang.Thread.run(Thread.java:613)

I'm not sure what caused this - are you able to run this version of your robot on your system? It would appear with your bot's name and version number, with no asterisk next to the version number. I was able to compile it from the source that you included, and that ran fine. (I just loaded the .java file in the Robocode editor and went to Compiler >> Compile.) I packaged that as 1.0b and uploaded it to my web space if you'd like to use that version in the rumble for now: [fw.Number1_1.0b.jar] ... But I'm sure you'd like to actually figure out why that happened, too. :) -- Voidious

My guess is that Incom built it with Java 6 and you tried to run it with Java 5? That's the kind of message I've seen having to do with that. -- Kawigi

I'm using Java 6 on a Linux System. For now I'll use Voidius jar (thanks for it!), but I think I'll try Jikes in the future. -- INcom

You could also instruct the compiler to generate 'java 5 compatible code'. How you could make such a setting should be easy to find. -- GrubbmGait

=========================



Man, I hate to bring this up, especially since I've been away from the scene for a while, but what's up with http://notforjustanyclient ???? That pretty much stinks. The Rumble is a peer-processed competition. If you don't release your bot for peer processing, it really sends up a warning sign. Why not? Simonton, please reconsider. It just looks bad and makes everyone a bit uneasy. --Corbos

I totally understand. You can find the reason discussed on /ReportedProblems (just search for my name, it starts the conversation). Basically, my bot does great for one battle. Uses about 100-ish megs of memory for itself. But then, if you don't re-start the JVM, for some reason that memory doesn't get re-claimed & OutOfMemoryErrors? ensue. I was thinking it was because Java's String class hangs on to every string it ever sees, but after looking at the Java source recently I'm wondering if that's true. Whatever the reason is - my bot only works on a fresh JVM (well ... one that it hasn't run on itself before). SO, I modded my rumble client to re-start after running any battle that my bot is in. If you call that foul play, I would understand. If you're skeptical of my integrity, I'll gladly send you the bot to try yourself, it does very well. Actually, my bots are RWCL, I just haven't posted them to the repository lately. Sorry, I hate to make a bad entrace to the community! -- Simonton

I'm definitely not skeptical of your integrity. I've really enjoyed your contributions to the wiki and the Rumble. However, this seems like a dangerous stretching of the rules. If you need to modify your rumble client for your bot to perform well, what's to stop other people from doing the same and isn't that making the playing field a bit less even? Even without modding a client, I know Chalk skips turns more often on weaker machines and his performance suffers. Given that, I could limit it so that only the client on my beefy machine processes his battles. In theory, his score in the Rumble would improve. My question is: is that an accurate and valid score given the limitations that were built into the Rumble's peer-processing model?

I'd like to hear what others have to say. Please don't change anything until we get some feedback. You're a great member of this community and you certainly haven't done anything wrong. I'd just like consensus as to how we handle modded clients or exclusive battle processing. Cheers. --Corbos

Well in Simonton's case its not a speed or turn skipping problem, its a memory problem. As making a robot with his pattern matching method in a mini requires the use of memory devouring methods, however I don't see the problem if Simonton were to make a Mega, as they not having size restrictions could use a less memory destructive method. So I think its mote the fact of limited space vs memory problems. I believe he was just saving us from tons of frustration having to kill roborumble@home after it ran a battle against one of his bots. So I don't think its a so much a question of should we allow exculsive battle processing so much as how much we sould limits a robots allocation of memory, which in this case is persistant.

As far as I know it could be a robocode problem, as I have noticed its running size gets bigger after running battle after battle and doesn't seem to dump the data between rounds, forcing me to restart after a few battles against the big bots. Perhaps if robocode called directly to the garbage collector after a battle to clean up the threads and free the memory (i'll experiment with my cvs build later, as I have a copy ofa newer WeeksOnEnd to test it with). --Chase-san

Well, before this circumstance, I would definitely have said that all bots should always be peer processed and to do otherwise is unfair. (And I think I have said/implied exactly this in some previous situations.) I still have that feeling about it, but this circumstance has made me think twice... I also think it's unfair to Simonton that his tank should crash in the middle of some battles, based on what and how many previous battles were run in that iteration of the RoboRumble client. I've tested his tank on my machine and it is clearly not using excessive memory in a single battle... I guess the ideal situation would be to fix the issue with Robocode itself, if possible, and forbid the exclusive battle processing? Perhaps I'll test some more with WeeksOnEnd to try and get a better understanding of the situation, as well. -- Voidious

Yeah, I've never really been sure if I bought the thing about string caching, but I've seen similar problems with Robocode before, particularly that it doesn't seem to let go of references to instances of bots. I've seen this as a bigger problem within a battle (stuff that's allocated every round as a member variable of the robot's class never gets released), and that tends to be fixable (either make it static or nullify your larger structures in your Death / Win events). As for between battles, it would sound like Robocode isn't letting go of entire classloaders. Running System.gc() after a battle is something that sounds like it would work, but probably wouldn't :-) -- Kawigi

Robocode uses Thread.stop() to stop the threads, where as it should be done via a viariable within the threads themselves tos top it, I am unsure to if theya re ever avctually deleted, via join or interrupt (I think, i'm not a expert on threads). A simple thread = null should allow the garbage collector to remove the thread(I think, again no expert on threads), if not then perhaps run a call by robocode that sends a call to the robot to finalize itself(which is far as I know impossible or dangerous). --Chase-san

Is there some problem with the most recent version of StrengthBee?? Or is it normal these days not to see a single battle run for a new entry after about 12 hours? -- Greywhind

Looks like a problem downloading it from the repository: [StrengthBee Details page]. Try downloading the JAR from there, it shows an error. Try re-uploading it I guess? -- Voidious

I just tried re-uploading it - but the error still says it's not found... Is this a bug with the repository? For now, I'm switching the download location. EDIT: I apparently uploaded a carbon copy of StrengthBee? 0.6 as 0.6.1. I will fix that now with an "0.6.2." Hopefully this time, I'll actually do it right. -- Greywhind

Hey Chase-san, looks like a broken link you've got up for Genesis Demo3 right now. My client requests that you upload it so that it can run your battles. :) -- Voidious

Hum, thats odd, I thought it uploaded. The one time I forget to double check. Also you can call me Chase the san is just formal. Uploaded. --Chase-san

Ah, that explains the "-san". You either know or are Japanese, right? -- Nfwu

If I remmeber correctly "-san" is used in more languages then just japanese, which i'm not. I however do know some Japanese, but not enough to fluently talk with someone even half the time. --Chase-san



And who will have the honour (or the nerve) to enter participant #500 ?? -- GrubbmGait


winamp32.micro.MicroMacro? 1.0, and krillr.mega.Psyche 0.0.3 aren't availiable to download for me, can anyone mirror them?

No problem: [winamp32.micro.MicroMacro_1.0.jar], [krillr.mega.Psyche_0.0.3.jar]. -- Voidious

Thanks, and I think this page is due for an archival, because its become massive :P -- Baal

Simonton's LifelongObsession = uh oh! ;) I think it's great you adopted that name. -- Voidious

:) Hey, that was a great name! You can't argue with that. Thanks for suggesting it! Let me know if it runs in the clients ok (e.g. whether it runs you out of memory). -- Simonton

I wish I could take credit, but it was David Alves who suggested it, I believe. -- Voidious

Bleh, I already see a bad result for LifelongObsession that I know came from my 1.1.3 client. I am upgrading it to 1.1.5 right now, hopefully it will work fine then. I thought the system discarded results with a 0 score, though, I don't know why it didn't just discard it. Anyway, you might want to repost as 0.11 :-\ -- Voidious

Oh shoot I forgot about that. Darn. I know I can't release that stuff, too! Sorry about that. I should probably take it off the participant's list until I fix it, unless you're the only one running the old client. -- Simonton

I really don't know who's running what :-\. I'd probably just skip the Rules class if it were me, as it's easy enough to code that functionality yourself in a MegaBot. In any case, lots of people are using Rules now, so it's probably about time we all start running 1.1.5 or later. -- Voidious

Oops, sorry for the mis-placed credit, David (if you still look at this)! So then, David Alves, that was a great name suggestion! -- Simonton

Hey Simonton - what you did will work fine for new results, but the old / bad results will not be erased. You'd have to change the version number to start fresh and lose those. You're likely to get more battles against those 2 bots that will partially offset the 0 scores, but not completely. -- Voidious




I started working on updating the RRAH client code to work with version 1.2.5A of Robocode. After deleting the RoboRumble .class files and recompiling, these are the 10 errors I'm seeing. Many of them seem easy to fix, but I'm just going to post them here for other people (hopefully Fnl =)) to give input and point out any flaws I'm making.

./roborumble/netengine/BotsDownload.java:223: cannot find symbol
symbol : method copy(java.io.File,java.io.File)
location: class robocode.util.Utils
try { Utils.copy(new File(filed), new File(finald)); }
^
// This method is no longer in the Robocode API, but it is easy enough to write our own
// method based off the old source, which I do have.

./roborumble/battlesengine/AtHomeListener.java:7: package apv does not exist
import apv.TeamCompetition;
^
// I'm not sure about this one yet...

./roborumble/battlesengine/RobocodeEngineAtHome.java:9: cannot find symbol
symbol : class FileSpecificationVector
location: package robocode.repository
import robocode.repository.FileSpecificationVector;
^
// This internal structure has changed and the class file used is different now.
// Should be a straightforward change, I think.

./roborumble/battlesengine/RobocodeEngineAtHome.java:14: cannot find symbol
symbol : class Constants
location: package robocode.util
import robocode.util.Constants;
^
// This stuff's in the Rules class now, another simple fix.

./roborumble/netengine/ResultsUpload.java:156: cannot find symbol
symbol : method copy(java.io.File,java.io.File)
location: class robocode.util.Utils
try { Utils.copy(new File(tempdir+"results.txt"), new File(resultsfile)); }
^
// Same as #1.

./roborumble/battlesengine/RobocodeEngineAtHome.java:49: cannot find symbol
symbol : variable Constants
location: class roborumble.battlesengine.RobocodeEngineAtHome
Constants.setWorkingDirectory(robocodeHome);
^
// Same as #4.

./roborumble/battlesengine/RobocodeEngineAtHome.java:85: cannot find symbol
symbol : class FileSpecificationVector
location: class roborumble.battlesengine.RobocodeEngineAtHome
FileSpecificationVector v = robotRepository.getRobotSpecificationsVector(false, false, false, false, true, false);
^
// Same as #3.

./roborumble/battlesengine/RobocodeEngineAtHome.java:85: cannot find symbol
symbol : method getRobotSpecificationsVector(boolean,boolean,boolean,boolean,boolean,boolean)
location: class robocode.repository.Repository
FileSpecificationVector v = robotRepository.getRobotSpecificationsVector(false, false, false, false, true, false);
^
// This is probably similar to #3, but I need to dig a little more to be sure...

./roborumble/battlesengine/RobocodeEngineAtHome.java:94: cannot find symbol
symbol : method setLogListener(robocode.control.RobocodeListener)
location: class robocode.util.Utils
Utils.setLogListener(listener);
^
// Not sure about this one yet.

./roborumble/battlesengine/RobocodeEngineAtHome.java:97: cannot find symbol
symbol : method startNewBattle(robocode.battle.BattleProperties,boolean)
location: class robocode.manager.BattleManager
manager.getBattleManager().startNewBattle(battleProperties, false);
^
// Not sure about this one yet.

Note: ./roborumble/battlesengine/RobocodeEngineAtHome.java uses or overrides a deprecated API.
Note: Recompile with -Xlint:deprecation for details.
Note: Some input files use unchecked or unsafe operations.
Note: Recompile with -Xlint:unchecked for details.
10 errors

I have an old Robocode source tree, probably from 1.07, but not the latest one. Grabbing that should give me the answers I don't have. Fnl, if you see this, I wouldn't mind any corrections or advice you might have. =) Cheers.

-- Voidious

To the author of Flower, your bot is crashing on the RoboRumble@Home clients because it uses the Rules class. There is not yet a RR-compatible version of Robocode that implements that class. You can easily implement the functionality of those methods yourself and re-release, and your bot will then get a real score. Make a page for yourself and your bot, too! -- Voidious

Same goes for GottesKrieger?. It seems a quite decent bot, but that does not show in the rankings. Re-release without using the Rules-class and the real strenght of your bot is revealed! -- GrubbmGait

Hey, is FrankTheTank? compiled for Java 6? I get an "unsupported version number" error when trying to run it. You should compile with 1.5 (or 1.5 compatibility) to be sure everyone can run it, or you'll get some very low scores. ;) -- Voidious

hennes.SimpleBot? also seems to be java 6, it runs fine under my newer machine but on the machine that hasn't had an update the roborumble client reports scores of 350-0 100% of the time. --Skilgannon

I've removed SimpleBot? from the Participants list. It's been almost 2 weeks since we brought this up and the author has neither updated it nor commented on the wiki. To the author - please make your bots Java 5 compatible; Java 6 is not yet required for Robocode and it's not fair to expect people to use it in order to get battles run against your bot. I'd normally never do this, but it's been almost 2 weeks since we mentioned it with no response from the author and I just noticed it has wasted hours of my RoboRumble client's CPU cycles. -- Voidious

Noticing your comments...How do you compile bots in a lower version? Is it (from the compiler preferences): Compiler Classpath: spath "C:\Program Files\Java\jre1.*4*.0_01/lib/rt.jar";libs/robocode.jar;robots or some other version number there (the asterisks werent there i added them for emphasis)? And do you have to install the older jvm? --Starrynte

I'd have to Google the compiler options, but you definitely don't need the older JVM. I use Eclipse which has a very simple preference setting for compatibility level to use. I found the answer on [this page]: the command to pass to javac is -target 1.4 (or 1.5, 1.2, 1.1, etc.). -- Voidious
*Jikes has the same option: -target 1.4.2. I am not sure if 1.5 (as older target) already is supported by Jikes 1.22. -- GrubbmGait

Kev seems to be on vacation from Robocode for a while and I am pretty sure he never meant to permanently have two WaveSerpents in the rumble, so I'm going ahead and removing the lower ranked one from the participants list. -- Voidious

Vic, you were right. There are other bots with two versions in the RR :
gjr.Cephalosporin 0.2 , kvk.HebusLeTroll
i dont know who these bots wrote but please remove one version you like so that only one stays. --deathcon

Of course I'm right ;-p Actually I observed that this happens also with more current bots. But I found out this is a glitch in the RR system where sometimes previous versions of a bot magically reappear and after a while disappear again. So if I'm really honest I should say I was wrong. Which I just did. --Vic

Well the two bots posted above stand both in the participants list with two different version and i want the authors to remove one version. What you mean comes from the RR thats true. RR cant delete old versions from the ranking, that is automatically done by clients. But sometimes it happens that the old versions appeares for some time, too. afaik, this comes from running the client with the option download=NOT . --deathcon

Almost. But it's more like "upload_results=NOT" I think. Never mind those, they dissappear once a client starts a new iteration. I think I counted to four or five CassiusClay's the other day. =) -- PEZ


Now 5(!) bots that have not yet 500 battles fought. We need more clients :D --deathcon

Funny. And just two months ago or so I thought maybe Robocode was half dead. =) -- PEZ

Do you remember the title of the eternal rumble site : ROBOCODE FOREVER !!! --deathcon

I can't think of a better description of the EternalRumble. -- Kawigi


Removed wiki.CigareT? - i created this bot one day (cigaret with a gun of mine), but it was not really successful. Added BoxingVirus. Shivas movement with CCs gun. I dont really want this bot to stay in the rankings - just want to see how well it performs and find out how good my movement is.I will it remove when it reached 600 battles. --deathcon

Ok, now i removed BoxingVirus again as promised. --deathcon


I'm leaving two computers running RR clients through the night (10 to 11 june). My bot has only 200 more battles to fight, so there's plenty of room for other updates this night :-) --Vic

Added new version of Shiva and BoxingVirus. BoxingVirus will be removed when it gets 600+ battles if it wont become #2, which i dont expect it to do :D --deathcon

I pulled out SilverFist. At the moment it only proves that if you blend bot #2 and #3 you can create a new #2 bot. =) We'll assemble a new version soon when Axe finds some time for it. -- PEZ


I have removed two blank lines:

...
davidalves.net.YALT 1.3,2084


dft.Axraelic 1.25,1244
...
The client was throwing an error and refusing to download bots because of that blank lines... -- Axe
Not really about participants, but i cant think of the right place to ask. What does the total%wins column in the premier league table mean?? I'm guessing it relects the ELO score, but im not sure... --Brainfade

jeje ... I can't remember. Did I place it there? Anyway, it's not used at all to calculate the ranking. PL works just like a real world sports league: 2 points for the winner, 0 for the loser. The bot that beats more participants wins the league :-) Albert

I believe you've said in the past that it's the tie-break condition and basically a reflection of how much they usually beat opponents by. Mathematically I'm not sure what it is (average % of score?). -- Kuuran

Yeah! That's it. I'm getting old and my memory gets worst :-) You are right. It is calculated just as the sum of the % of score you get agaist each enemy (it like "goals"). -- Albert

I pulled Resin because it caused out-of-memory conditions. -- PEZ

Added a new Resin (version 0.3). Please feel free to remove it if it causes problems for your client. -- PEZ


Someone added the SampleBots. Is that serious? -- Jonathan

I removed them from 1v1. They don't belong there. Dunno if they should be left around in melee though. Not too crowded there. -- PEZ


I removed a duplicated "shinh.Entangled 0.3,1070" line in MeleeParticipants?. Was it getting double the number of battles than the other bots? -- MN?

I think not, but not sure -- Albert


Should BeeRRGC still be in the rumble? -- Alcatraz

No, it shouldn't IMO. It's in the way for lower ranked bots. I'll remove it. -- PEZ


Can I get the US flag for grybgoofy? (I re-added GoofyBot 0.10) --Goofy
The number if 'wiki' bots is increasing again. Isn't it time to remove some of them. For example, there are currently two versions of DevilFISH in the rumble (and i hate this bot ;-) ) --Loki

But one of them is a nano... -- PEZ

Mmmm right. But both of them are a pain in the ##. It was a nice try... --Loki

Though they _are_ both behaving and performing very similar (the 5 point rating diff could just be random noice). We could remove the non-nano one is my vote. Not to help you with removing a problem bot, mind you. But we need to keep the list reasonably tidy as it has grown rather large. I'll do it right away. -- PEZ

Goofy, update the RoboRumble/CountryFlags table so that David Alves can prepare your flag in the new RR server. -- PEZ

why are more and more people using direct links to their bots? The RobocodeRepository is still up, isn't it? --Loki

It wasn't the other day. I think I downed it when i cleaned my robot jars. And some bot jars are just too big for the repository. -- PEZ

That big bots is SS! :) Repository denies uploads to bots bigger than ~250Kb... And some editions of my Godzilla Bot actually go bigger than that. -- Axe

@PEZ: (U probably knew that i was going to complain about it, dont u?:) Why? Marcelo.DevilFISH is similar, but not the same bot... -- Axe

It's the same functionality packed in different jars. They behave the same way. They perform the same way. The list of participants is growing fast enough anyway. The only difference I can see is that one is a mirco and the other is a nano. Removing the nano would mean no DF in the nano rumble so I figured let's remove the other one. Check this comparison if your still not convinced "it's the same bot": http://pezius.com/rumble/servlet/RatingDetailsComparison?game=microrumble&name1=Marcelo.Alpha.DevilFISH%201.3&name2=wiki.nano.DevilFISH%201.0 -- PEZ

Your word is enought to me. If it is the same, surelly u r right. I'm in mourning :(... -- Axe

Why would you miss the duplicate DF? -- PEZ

Sentimental reasons... I grew fond of Marcelo.DevilFISH. :) -- Axe

Where did PulsarNano go? -- Alcatraz

According to the revision history Pulsar removed it himself. But it was in the same edit where he updated PulsarMax so probably it was a mistake. I'll put it back. We can't have a nano top-3 without a Swedish flag, can we? =) -- PEZ

Hi guys I have a little question, How can i put the flag of my country on my bot? (like a lot in the rankings) -- F4kill

See /CountryFlags and PEZ will fix it as soon as he sees it. And welcome! -- Pulsar

Thanks :) -- F4kill


Removed the following bots because they are not packaged according RR@Home standards.
 b0lt.Obliterator,2567     --> This is not a jar file
 rpf.RPFRobot 1.0,2577     --> This bot misses a <name>.properties file in the jar
-- GrubbmGait


@Krabb, is it correct you entered Doboh v1.2 in the RoboRumble/Participants and v1.3 in the ImplementsDroidCompetition/Participants? Shouldn't they both be v1.3? --Loki

Removed amk.advancedstrike.AdvancedStrike 0.5,2601 because it could not be downloaded. Removed kinsen.ScanFire 1.0,http://www.geocities.com/seakinsen/kinsen.scanfire1.0.jar because it could not be downloaded. Changed Krabb.doboh.DoBoh 1.2 to 1.3 because "Downloaded file is wrong or corrupted:Krabb.doboh.[1]?.2.jar". --Loki

Thanks, Loki! --Krabb


Hey, why isnt the Sample team in the teams competition? -- Jokester

Because SandboxLump can beat it (not a team of SandboxLumps, but just SandboxLump). -- Kawigi


Just thought I'd give ya guys a heads up. Dont expect to much though, I've just come back again after loosing all my previuos code.(damn HD's they should warn before they crash) Jimpa


I don't feel it's my place to remove florent.Froggy 0.1.1PM but I hope its author will consider it. It pegs my CPU and stops the RoboRumble. Not a single result has been uploaded since its arrival. Sorry, but ratings will cease if this bot stays. --Corbos

Just do it.. I think if a bot disables the RoboRumble you are allowed to delete it! --Krabb


To the author of florent.*: please quit uploading versions so quickly - especially the versions that take forever (version 1.6 of FloatingTadpole is slow, slow, slow). If this continues, I see no reason to run RoboRumble. RoboRumble has become the florent show. That sucks. Please test your bots and compete with stable and significant changes. If you're careless, you can ruin this for everyone. --Corbos

I'll remove it now and upload a new one in a few days when it will be quicker --florent

Florent, After reading what I wrote yesterday, I see that I owe you an apology. Your bots are getting better and better and I don't want you to remove them from the Rumble. They provide excellent pressure for Chomsky. Both FloatingTadpole and Froggy beat him. I was just frustrated that by the time one version's rating stabilized a new version was uploaded. It felt like my client spent all of its time stabilizing your bots. Can we strike a compromise that keeps your bots in the Rumble? If you keep at it, you're going to give those untouchable top bots a run for their money. Please accept an apology from this crabby old code monkey. --Corbos

There is no need for an appology, I just spend to much on my bots now because I am on holliday, so as soon as I have implemented something that seems to be good enough to gain a few places(dynamic distancing, fixing a huge bug in my waves, a forward pattern matching gun (that is slow and not nearly as effective as i supposed it would be)) I upload it. I will try to test more my bots before uploading them and clean the code of unessary part that tend to slow down the bot (PM gun in a VG array). --florent

Dunno what it is about today but there have been several bots submitted / updated. I'm running rumbles on 4 cpu's trying to catch them (and particularly Ugluk) up. I'm pretty stoked too because Ugluk had a score of 1700 after 95 duels and it keeps climbing. -- Martin

Yep, I noticed the surge myself... Perhaps everyone's thinking the same thing I was: I want to get a new version up tonight since I won't be working on it for probably a week or so! (I'll be running the RoboRumble momentarily here, and then all night, FWIW.) And congrats on the ratings hike, Martin. -- Voidious


Loki, congratulations with the entry of bot number 400: bvh.micro.Freya 0.1 -- GrubbmGait

thanks for the congratulations! but as it holds almost the 400th position, i don't want to draw to much attention to this bot :) It is not designed for 1-vs-1 though, so let's see where it ends in the melee ranking. I just wanted a competitor in the micro melee rumble and ended up with the minimum risk movement of mini.Freya and a simple gun with zero bytes to spare. Maybe i can find some extra bytes for a better gun. --Loki

Bayen, my RR@Home client said it couldn't download your SamSpin? .jar file. I tried downloading it from the Robocode Repository, and noticed that you have it as "...Samspin...jar", with different capitals than the title of the tank ("SamSpin?"). That might be the problem, although I'm not completely sure. Also, the general convention is to always use the same top level package for all your tanks, so you may want to change it to 'bayen' when you get a chance. -- Voidious

Actually, Samspin is not my robot, it is my brother's. joe.ADinosaur is my other brother's. So that explains the seperate packages. I think the captialization is the problem, thanks. --Bayen

Is there any logical explanation as to why Dookious has faced Shadow 49 times? (Dookious details page.) I don't know what that would do to which rating, but Dookious is overperforming against him, in any case. I imagine it would bring Shadow's down. ABC, should one of us rerelease? -- Voidious

It does not have an influence on the rating, except being more stable. Once you have fought one battle against someone, the results of the next battle against the same opponent are compared against the (weighted) previous results. If your first battle was 60-40 and the second 65-35, you will slightly go up and the next 'expected' result would be approx 62-38. If the third result is 60-40 again, you would slightly go down, even if the other one is higher ranked. Only for the first battle the expected result is depending on the rankingdifference. After that, the previous results is the guideline. About facing another bot a lot, it can happen when it was the last fight to complete the priority battles. Then sometimes that same battle is fought as many times as the number of battles in RR@Home is set. -- GrubbmGait

Ah, OK, cool. Thanks for the info. -- Voidious

Out of curiosity I just checked the average ranking of all participants, and it came to 1591.61, pretty close to the 1600 'average' mark. I imagine this is influenced by people swapping out new versions of bots frequently, yet it is still holding relatively steady. -- Martin

Sorry about the double entry for PowerHouse tonight guys, I will remove one of them as soon as they stabilize. I know I should only have one version up at once, if its a problem I'll remove one right away. --wcsv

Whoah, does someone have PulsarMax set to TC or MC mode in their RoboRumble directory? There's no way Dookious could ever beat him by a score of 90%. (Dookious details) What should I do about this? -- Voidious

I ran a few battles to test this, Dookious is a very strong bot, but I don't think it's quite *that* strong (yet). (My tests show scores of 45-50% vs PulsarMax) --wcsv

Unless someone can point out why I shouldn't, I'm going to set my "battlesperbot" to 2,000 and run RR@Home for the rest of the day. (I'll be gone, anyway.) Hopefully that will mean Dookious fights PulsarMax at least one more time. -- Voidious

Hm.. Voidous brings to light a growing reliance on Mohammed Ali references. Perhaps we should branch out with StampedeGoneIn91Seconds, GrubbmGrill (knocks out the fat), ChomskyEar (has a nice ring to it), and the obvious .. ShadowBoxer. -- Martin / Ugluk

Not sure what you mean about growing reliance. But anyway, the latest too Dooki-bots are tests with CassiusClay and Dookius guns and movements used together. -- PEZ

I was attempting to be funny. -- Martin

Hehe, yeah, I was with you that far dude. =) -- PEZ

I think it would be funnier if I could tell what combinations of bots those are supposed to be, Martin :-) If anyone is annoyed at the extra entries, feel free to remove them or set your "BATTLESPERBOT" to less than how many battles they have. -- Voidious

Well the 91 seconds thing is how long one of the Mike Tyson fights lasted. Imagine buying tickets to that fight. On one hand, you saw a legendary fight. On the other hand .. it only lasted 91 seconds... -- Martin (who didn't really have any movements / targeting in mind when coming up with the hybrid names)

Hey guys, sorry about the onslaught of Dookious releases this week. I won't be making a habit of it or anything. I'm taking a short break from Dookious to explore some other stuff after 0.71, and I wanted to give a good push at hitting 2k first... -- Voidious

It seems to me that the Rumble gives priority to higher ranked bots, since Shadow has processed more battles and PowerHouse is close, even though they were both released 6 hours after Ugluk. Ah well... -- Martin

Well, don't worry about releasing loads of versions in sequence. Look at CasssiusClay/UpdateHistory? and you'll see that this has been done before and in my book it is how it should be. It is why Albert gave us the RoboRumble@Home to begin with even. Martin, no priority is given to higher ranked bots. But randomness and timing of releases and different settings on RR@H clients and stuff like that sometimes give tilted battle counts like that. -- PEZ

I have had my 'battlesperbot' set on 500 today, to boost the relatively large number of updates. It seems that this was not beneficial to both Ugluk and ds.Oo. Such a low number for 'battlesperbot' does help to fill the PL-ranking quicker. -- GrubbmGait

I have my suspicions, but is there any plausible reason why my client is still processing pez.mini.Pugilist long after his battle count has passed the BATTLESPERBOT threshold? It is happening on at least two machines where the setting was changed to 500. Meanwhile Pugilist has 618 battles, he's in well over 50% of battles, and there are around 7 other bots to process... -- Martin

It has to get up to 500 in the MiniRumble?, too, where it's still under 300. -- Voidious

Yeah. Be happy Pugilist isn't a NanoBot! =) It sure is a SlowBot. Appologies for that. But it's hard to get surfing and statistical targeting into a mini and some corners are definately cut in the case of Pugilist. -- PEZ

Out of curiosity, how many people put bots in the RoboRumble without running the RR@Home client? As I can see there are a lot of bots to stabilize, I'll run my client for at least a little while here. I suppose watching the Lakers could take the place of Robocode for a little while =) -- Voidious

I do not know how good a litmus test it is, but I assume that the people who commonly discuss / comment on things are all running the Rumble when they have bots that have not reached the plateau, and sometimes when they don't have active participants. I suspect that people who don't feel as involved are less apt to crunch the numbers. -- Martin

I used to have it running 365/7/24. But lately my machine has some problems making it restart itself now and then like twice a day or so. So now I mainly run it when I have a bot I want to see the rating on quicker. And sometimes I don't run it even when I have bots that need battles. This because running robots seems to make my machine restart more often. But think of it like this. Now it never takes a full day to get an answer on how strong a bot is even when quite a few bots are updated. Back in the days of the EternalRumble we had to wait weeks. And we never knew when the next ratings where going to get determined. Sometimes it took months. If you're really eager to see if you have improved your bot, te TargetingChallenge2K6 and MovementChallenge2K6 are there for you. If those shows improvements it is pretty much certain you have a better RR@H rating coming. -- PEZ

I'm definitely grateful for the RoboRumble setup, and I have no complaints. I definitely rely on the TC and MC challenges, as well as a combination of stable tanks and ProblemBots, to benchmark my tanks before release. -- Voidious

Once I realized that I screwed something up with the movement, I prepared 0.75b with the intention of posting it in the morning. Anyway, got up in the middle of the night and noticed 0.75 was past 1k at 1989, so I went ahead and posted 0.75b. It restores the movement of 0.72, leaving only gun changes. (I was getting pretty good about only updating gun or movement at once; I shouldn't have pushed my luck!) -- Voidious

It isn't so much about luck as it is about knowing what you can deduce from a set of changes and their results. Even if you would have ssen an improvement of 10 points you couldn't know if it was really a movement improvement of 20 points and a deteriation of the gun of 10 points. Unless, of course, your local tests have showed something that can be of guidance here. The important thing is to test gun and movement changes seperately. Says someone that releases loads of bots with changes everywhere... -- PEZ

Why "mrm.MightyMoose .1" doesn't participate in RoboRumble? This bot has been downloaded successfully! It is in list, but doesn't fight. Why? -- DemetriX

First of all, the name is not correct, the bot is named "mrm.MightyMoose 1.0". Secondly, this bot does nothing except keeping a radarlock. Therefor it gets a score of 0 (zero) against xxx, which is not accepted by RoboRumble. If it was named properly it would only consume valuable processortime without any result. ad.last.Bottom always scores 1 point against anybody and therefor it is a valid entry (and also last in the ranking). -- GrubbmGait

OK. The problem was simle. This bot has version ".2", but in the list named ".1". I dare to correct this and now it's all right! Thanks for helping me to solve this problem. :-) -- DemetriX

No... It fights normally. Just bot had wrong version in participants list. -- DemetriX

Then it is already corrected. I tried it the first time I saw it was an invalid entry, and then the bot did nothing. After that I have not checked if a new .jar was placed there. The .jar file has no version, so there is no way to check that other than to try it. This wiki is intended to help others, this includes fixing small mistakes and typos. Thanks for correcting the entry! -- GrubbmGait

I have been looking at specialization indices and only 12 of the top 25 robots have a spec. index under 20. They are also all bots that have been around for thousands of matches. The venerable SandboxDT has the lowest at 6.8, with Quest at 9.5 and RaikoMX at 9.8. Sandbox's main problem bots are ram bots, but even then they are not serious problems. -- Martin

It's worth noting that there's something wrong with the SPI calculation - according to Excel, the average squared value of DT's PBIs is 29.513. Really old bots seem to have this issue; one of us really should fix that script sometime... -- Voidious

Sorry, I know that Dookious is among the tanks that isn't stable yet, but I really want/need to use my CPU without the RR client running for a bit. I let it run all night and most of today (until it hung), but man, X2 is just incredibly slow (sorry Florent, I know you know that), and I think I was the only one running the client when I last was. I'll try and run it for a bit later in the evening, and certainly during the night... -- Voidious

Hey Florent, that was very thoughtful of you, and I'm sure wcsv appreciates it... Do you want me to restore 0.11 to the list when I start my client later this evening? -- Voidious

Thanks Florent, but you didn't have to do that. I know i've been making a lot of updates, so I don't mind running X2 for a while. Thanks again though, the client runs slowly on my machine even when i'm running relatively fast bots. --wcsv

There's no need to restore 0.11, I'll published a new one when I can get a faster version of it. -- Florent

By the way, PEZ told me it was cool with him if I posted a version of BeeRRGC with the latest Bee gun. For some reason, the data reloading was failing every time on my machine, I'm guessing because the gun is in a different base package than the bot itself; I remember in this situation with a version of DookiBee? my RR client crashed the bot every time, so I just disabled the VG stats stuff in Bee for this version. Shouldn't make a huge difference, and it's noted in the version number. -- Voidious


how do you think about removing the wiki.*-bots from the Participants list (and MeleeParticipants? list)? They are often experiments, e.g. a combination of gun and movement from two bots. Some are removed after some time by their creators, e.g. wiki.Tyranius, others remain in the rumble for ever. I am in favour of removing them. what do you think? --Loki

What about ones that aren't experiments and are in fact collaborations between two authors? I would leave those in. However, I've been wondering about the legitimacy of SilverFist as a KingOfRR?. If SilverFist is listed there, why not Tyranius? --Alcatraz

I think the main difference between SilverFist and Tyranius is that PEZ and Axe took the throne from ABC with SilverFist, while Tyranius is a combo of the two current top tanks. I don't speak for mue, but it seemed like we both just wanted to post it as an experiment, and not have it as a throne contender. That said, though, I don't personally have any problem with it being listed there if others think it should be. -- Voidious

I guess I also am partly wishing that SilverFist hadn't just disappeared after losing it's title. It felt very much like an experiment to me. And in regards to Tyranius, I just find it interesting that there exists a bot stronger than the king. But this appears to the mythology-lover in me that envisions Tyranius as the mysterious warrior who out of nowhere came and usurped the throne and then went just as quickly back into the night. -- Alcatraz

i am afraid it is not as easy as a the distinction between 'experiment' and 'cooperation'. Most wiki-bots are a cooperation and most cooperations are also an experiment, take for example again Tyranius. But some wiki-bit authors decide to remove their entry, while others don't. If this is a list with all bots ever made, then all wiki-bots may remain listed. SilverFist is a good example as it also was a succesfull entry. So maybe i try to determine some criteria when to keep wiki-bots and when not.

We already have the first criteria: ;) --Loki

Well, personally, I do try to keep my entries in the rumbles pretty clean, which means removing such experimental tanks. (I might even remove Shaakious.) However, since it's such a common practice for tanks to build upon one another, I'm not sure I'd feel right removing such combo tanks without the authors' permission. I do understand where you're coming from, don't get me wrong... something like BlackDestroyer, which is outranked by each of the tanks that it was a combo of, does seem like a good example of one that maybe should be removed. And the whole combo tank thing does have the potential to get out of hand. But I guess I'd still say the author(s) should be the ones to do it, and the resulting may still be an interesting and unique tank to have in the rumble. -- Voidious

I agree that a cleanup of wiki.* bots can be usefull, therefor I removed the unsuccessfull experiment ARAMtocles. I will check if some of my bots still have some added value (like GrubbmTwo) and if not, remove them also. With more than 400 bots in the rumble, any reduction of the number of bots seems welcome. -- GrubbmGait

It's up to the authors of the wiki.* bots of course. If they have left them there I can't see that it means anything else than that they want them there. Some bots might just have been forgotten by their authors, but that goes for a lot of bots. Rather than focusing on wiki.* bots we might want to shout out somehow that people check the participants and question if all their entries have reason enough for being there. Personally I have removed more bots from that list than most of you have ever entered. =) I feel like the wiki.* bots in the list that I am involved in belong there. And of course Tyranious should have an entry in the KingsOfRR? table. It would add to it's mystical features too. Where's T now and what's she planning? =) -- PEZ

By the way, there seems to be a pretty high number of bots to process in the RoboRumble right now. (7 under 1,000 battles, 5 under 500 battles, 3 of them <= 10 battles... and I want to post an RRGC test tonight or soon.) I'm setting my BATTLESPERBOT to 500 and running as much as I can tonight and tomorrow. To anyone that doesn't run the RR@H client, it's very helpful to have as many people as possible running it to get the rankings updated, so check out the /StartingWithRoboRumble page if you can spare some CPU cycles. Martin also posted his RR@Home settings to the /RankingChat page, if that helps in getting it setup. -- Voidious

My harddisk crashed last weekend, but this weekend (hopefully this evening) I will be up and running again. -- GrubbmGait

This RR@Home situation reminds me of an idea that's occurred to me from time to time. I'm not sure how practical it is, and I'm not sure if it would bring enough attention to our RoboRumble to invite security risks (ie, sending bad results), but... I think it would be really cool if we could make a screensaver-type program that would run Robocode battles for the RR@Home and display them on a user's screen (a la SETI@Home). It would actually be a lot slower than our RR@Home clients, since it wouldn't be running minimized, but if it were used by even a relatively small number of people it could really help with the processing. And, IMO, it seems like it might appeal to some people as a cool screensaver. -- Voidious

How about caching/saving the battles via some sort of 'recorder' and playing them back without the burden of real-time processing? Maybe important battles (surprise victories, etc) would rank higher and re-run more often. --Corbos

Well, I was also thinking of this screensaver as being an active RR@Home client, so it would also help with the RoboRumble processing. Or do you just mean it would run battles in the background, but display different battles as the screensaver aspect? (Robocode already can record/playback battles, I think, although I've never actually tried it.) I do think that any kind of Robocode-based screensaver would be really cool, but I was more specifically talking about it as a RR@Home client, as well. -- Voidious

I was thinking graphic-less real-time processing for the client (very fast) with recorded playback in the screensaver (also very fast with some dramatic value). I'm always looking for the best of several worlds ;) --Corbos

Kev, if you want to hit my GMail account with your latest update (voidious at), I'll gladly host it on my web space for you and re-post it to the rumble when I do. -- Voidious

Thanks a lot Voidious, I would really appreciate that. I'll send the latest version of Vyper to you now. -- Kev


My friggin' computer crashed pretty early in the day while I was at work today, so unfortunately, I missed a lot of potential RR processing there. By the way, I edited line 206 of roborumble/netengine/BotsDownload?.java to fail on any RobocodeRepository downloads, otherwise it's a lot slower to process right now with the repository down. (Basically, just "return false" instead of creating the repository URL on that line. You compile it from the 'robots' dir, and include robocode.jar in the classpath.) -- Voidious

Thanks for the tip, now I can run on full speed again (although full speed is not that fast on a P3 1GHz) -- GrubbmGait

Gah, my internet connection is so flaky... If Dookious is holding up anyone else with tanks that need battles, feel free to revert him to 1.06 temporarily. I won't be contributing RR@Home battles until I get home again this evening. -- Voidious


There are some other bots in the rumble that may be questionable: Albert's bot apv.test.Virus is literaly based on jam.RaikoMX.

public class Virus extends jam.RaikoMX {
  .. snip ..
}

I think extending this/a strong bot is a clever idea, but as the package name describes: "it's a test". And it has lasted long enough in my opinion. --Loki

Hmm... That's an interesting point. Personally, I like Virus, as it has a very unique gun. Since RaikoMX is open source, anyway, I kind of think it's fair to use the movement in another tank. A lot of MiniBots take guns or movements from each other - the first version of Komarious had an unaltered version of RaikoMicro's gun. But that's just my 2 cents... If others think it should be removed, that's fine. Hopefully Albert will chime in if he's around. -- Voidious

Using someone else's gun or movement has been done tons of times. Personally, I don't like it much, but I think it is a common enough practice that we can't start ruling out bots based on it. Just don't use someone else's gun AND movement... that's a bit much. --David Alves

I think Virus should stay for two reasons. First, it would change the competitive landscape too much. Virus has been in the rumble since I started. It seems strange to remove him now. Second, he should stay out of respect for Albert. Albert contributed a ton to the wiki content, explored novel pattern matching and neural networks in his guns, and gave us RoboRumble@home! You’d have a hard time convincing me Virus is a lazy coder’s exploitation of the system. Let’s stick with the 'extend gun or movement – but not both' rule. Cheers. –Corbos

It's true that exact movement/gun code has been re-used in many bots and i don't dispute Albert's contributions to the rumble and his robots (think of Aspid, LauLectrik and ScruchiPu!!), i only mentioned Virus because of the definition ("extends jam.RaikoMX") and it's package-name ("test"). I think it was an (excelent!) exercise to show how a succesfull bot can be created based on another succesfull bot. But i think it was an exercise. So I hope Albert still ocasionally visits this site, thinks the same way as i do (and maybe enrich us with some more contributions). --Loki

Anyone know what's going on with Phoenix 0.67? It's got over 5,000 battles, but has only faced about half the RR contestants (on average like 20+ times each). I think if a RR client starts but doesn't finish uploading results, it will re-send all those same results next time, so I figure that must be part of what's going on there... -- Voidious

Um, that's weird. I left my client going this morning and went to school. I was only running one client so it shouldn't be able to run that many battles in ~8 hours anyway. No idea what happened. --David Alves

I figured it out. Apparently this is what happens if you run two RRAH clients that are both using the same directory. =P --David Alves

Was about to tell you that WaveSerpent is throwing errors, but I realize that I just need to upgrade my clients to the new version. Doh!

=========================
Round 6 of 35
=========================
kc.serpent.WaveSerpent 1.0: Throwable: java.lang.NoSuchMethodError: kc.serpent.WaveSerpent.setBodyColor(Ljava/awt/Color;)V
java.lang.NoSuchMethodError: kc.serpent.WaveSerpent.setBodyColor(Ljava/awt/Color;)V
    at kc.serpent.WaveSerpent.run(WaveSerpent.java:35)
    at robocode.peer.RobotPeer.run(RobotPeer.java:633)
    at java.lang.Thread.run(Thread.java:613)

-- Voidious

Kev, you might consider checking if a method is present before calling it if you're going to use methods from the newer versions of robocode. For example:

if(TeamRobot.class.getMethod("onPaint", new Class[]{Graphics2D.class}) == null){
	debug = false;
} else {
	debug = true;
	info("Graphical debugging enabled, click 'Debug Paint' in robot console to view");
}
--David Alves

Thanks guys, I don't think I've ever seen a -80 PBI before :). I re-released WaveSerpent using setColors(), which is compatible with all versions of robocode, so the problem won't happen again. -- Kev

Is there a way to see who turned in which battles? Someone turned in a battle of Phoenix 0.7 vs. KanoGamma where Phoenix scored 56%. This is very very wrong, I've run a bunch of battles and the results are more like:

1st: davidalves.Phoenix 0.7	4455	1700	340	2020	395	0	0	34	1	0
2nd: kano.gamma.KanoGamma 1.8	723	50	10	654	8	0	0	1	34	0
I don't know if it's skipping a lot of turns on that system or what, but something is really screwed up. --David Alves

A few months ago I mentioned that the average rating was 1591.61. Now it's at 1591.91. A pretty small bump for at least 4 months of rumbles, with at least 20 new bots added. My assumption is that running the Rumble enough times without changing the participants would get it even closer to 1600, but I don't know how long it would take or if it would even have a noticable effect on ratings. It's hard to say how the 8 point average increase would be distributed, though I suspect it would be larger in the middle than at the fringes. In the end it doesn't really matter. I think the Rankings would not be affected. -- Martin

Why do you expect the average tends to 1600? The distribution is centered on 1600, but the function used to calculate ratings is not linear, so if the distribution of the bots on the upper side is different to the distribution on the lower side it is possible is stabilizes in a point different from 1600. -- Albert

Step, are you having problems with packaging robots for the rumble? There is a handy tool inside the Robocode client to 'package robot for upload' that should make things go smoothly. -- Martin

ebp.NanoBruno? is a class file, that type of file can not be handled by RoboRumble@Home. Only jar-files (correctly packaged) are supported. -- GrubbmGait

I removed DM.Titan _3 from the rumble. Please do not use underscores in your version number. It makes the RRAH client have errors:

Fighting battle 12 ... myl.micro.Troodon 1.10,DM.Titan_ 3
Exception in thread "Application Thread" java.lang.ArrayIndexOutOfBoundsExceptio
n: 1 >= 1
        at java.util.Vector.elementAt(Unknown Source)
        at robocode.battle.BattleResultsTableModel.getValueAt(Unknown Source)
        at roborumble.battlesengine.BattlesRunner.runBattles(BattlesRunner.java:
95)
        at roborumble.RoboRumbleAtHome.main(RoboRumbleAtHome.java:94)
--David Alves

I don't expect anyone else to do so, but I've bumped the priority battles threshold from 1200 to 2000 for all of my RoboRumble clients. We've nearly 500 bots in the Rumble now and I'd like to have a better chance of getting a consistent final rating. At present only the two bots using the new interface are below the 2000 mark (since I cannot process them), plus very new releases. -- Martin

I'll definitely join you with the 2000 min battles. I often set mine to 1399 or 1599 once everyone's past 1000, but I've been lazy and just leaving them at 999 lately. -- Voidious

Good idea, also set my meleeclient from 1500 to 2000, although I don't run that one that often. -- GrubbmGait

Here's the error message from "Exploit" under 1.1.2, which my RR@H clients are running:

=========================
pedersen.Exploit 1.2.1beta: Throwable: java.lang.IncompatibleClassChangeError
java.lang.IncompatibleClassChangeError
    at pedersen.core.Foundation.cheat(Foundation.java:138)
    at pedersen.core.Foundation.processTurn(Foundation.java:113)
    at pedersen.core.Foundation.run(Foundation.java:59)
    at pedersen.Exploit.run(Exploit.java:30)
    at robocode.peer.RobotPeer.run(Unknown Source)
    at java.lang.Thread.run(Thread.java:613)

So it just gets a zero score every time.

-- Voidious

I built a new roborumble with the 1.2.1beta robocode.jar and it is also getting an error, though a different one. So at least for now the bug in question cannot allow someone to dominate the ratings. The robot is still available to demonstrate the bug though. It is linked from the News page. I pulled the robot from the rumble a few mintues ago. -- Martin

I hate to be a party pooper, but I just have to voice my opinion that it seems like a bad idea to continue testing these exploits in the rumble. If there is a bug in Robocode, it should be fixed in the main program, anyway, and that will fix it in the RR client as well. Putting these exploits in the rumble seems like unnecessary distortion of the ratings of every tank in the rumble, not to mention wasted processing time. (E.g., it starts at 1600, wins 99% against a 2000+ tank, you're looking at a -85 PBI and probably 3-4 RR points lost on just that battle.) -- Voidious

Voidious, I have a very very very good reason for testing chase.Dodger, because unlike pedersen.Exploit 1.2.1beta, chase.Dodger does not use any of the Set commands in the Manager, it uses only get commands to direction every day robot hardware into attempting to dodge bullets. After 100 battles (hopefully thats not to big a waste to test a potential loophole) I plan to remove it. -- Chase-san

I hear you. I do think these things should be tested, I just don't see why putting them in the rumble makes a difference - it uses the same engine to run battles as Robocode itself. Dodger seems to get 0 every battle on my system, so it is going to take a lot of battles before it gets to 100 RR battles. Anyway, the one client I have running is currently hacked to not download Exploit or Dodger... -- Voidious

One of the advantages of having a less competitive bot is not feeling the need to take things so seriously. -- Martin

Apparently I've been doing Robocode for over two years now. I was checking some old emails and when I started a friend and I would email out robots back and forth one-upping each other. The last one he sent was October 10th, 2004. I've held off on submitting it in the rubmle on his behalf, but curiosity overcame me. After making the jar file I decided to test it against Ugluk, which was modified to not move for the creation of the targeting challenge challenger. Even without moving, Ugluk beat him 5800 to 4300 or so. He must not be using 3.0 power bullets. He used to be hard to beat for me. I guess Ugluk has come a long ways in 2 years. -- Martin

Hey Simonton, I have one tiny request - could you use a "real" broken URL instead of what you're currently using for the bots you aren't posting for download? My clients seem to hang on this kind of URL instead of just moving on like they would with a 404 URL. I just hacked the client for now, so no worries, but some others might be hanging on it =) -- Voidious

Oh, sure, no problem. Sorry about that. Would it work if I just put a "http://" in front of it? Maybe it's trying to download a bot with that ID from the repository without the http prefix? -- Simonton

Ah, yes, that's probably what it's doing. (Actually, the client just checks for the "://"...) Just unhacked my client and it looks good. Thanks man! -- Voidious

Hey Incom, your fw.Number bot isn't working in my RoboRumble client because it is a .class file instead of a .jar file. What you need to do to package it is this: in Robocode, click on Robot >> Package Robot for Upload, select your bot, and then give it a version number (and other info, if you'd like). This will create a .jar file in your "robots" folder which you could then upload to the repository. (Note that you can upload a new file for an existing entry on the RobocodeRepository, you don't need to make a new one.) The RoboRumble client only handles these .jar files. Once you've got that done, you should make a page for yourself on the wiki and request the appropriate flag for your bots' package. :) (I found the name Incom? in your RobocodeRepository page...) -- Voidious

SYSTEM: Could not load fw.Number1 1.0 : java.lang.UnsupportedClassVersionError: Bad version number in .class file
java.lang.UnsupportedClassVersionError: Bad version number in .class file
    at java.lang.ClassLoader.defineClass1(Native Method)
    at java.lang.ClassLoader.defineClass(ClassLoader.java:620)
    at robocode.security.RobocodeClassLoader.loadRobotClass(Unknown Source)
    at robocode.battle.Battle.initialize(Unknown Source)
    at robocode.battle.Battle.run(Unknown Source)
    at java.lang.Thread.run(Thread.java:613)
I'm not sure what caused this - are you able to run this version of your robot on your system? It would appear with your bot's name and version number, with no asterisk next to the version number. I was able to compile it from the source that you included, and that ran fine. (I just loaded the .java file in the Robocode editor and went to Compiler >> Compile.) I packaged that as 1.0b and uploaded it to my web space if you'd like to use that version in the rumble for now: [fw.Number1_1.0b.jar] ... But I'm sure you'd like to actually figure out why that happened, too. :) -- Voidious

My guess is that Incom built it with Java 6 and you tried to run it with Java 5? That's the kind of message I've seen having to do with that. -- Kawigi

I'm using Java 6 on a Linux System. For now I'll use Voidius jar (thanks for it!), but I think I'll try Jikes in the future. -- INcom

You could also instruct the compiler to generate 'java 5 compatible code'. How you could make such a setting should be easy to find. -- GrubbmGait

=========================


Man, I hate to bring this up, especially since I've been away from the scene for a while, but what's up with http://notforjustanyclient ???? That pretty much stinks. The Rumble is a peer-processed competition. If you don't release your bot for peer processing, it really sends up a warning sign. Why not? Simonton, please reconsider. It just looks bad and makes everyone a bit uneasy. --Corbos

I totally understand. You can find the reason discussed on /ReportedProblems (just search for my name, it starts the conversation). Basically, my bot does great for one battle. Uses about 100-ish megs of memory for itself. But then, if you don't re-start the JVM, for some reason that memory doesn't get re-claimed & OutOfMemoryErrors? ensue. I was thinking it was because Java's String class hangs on to every string it ever sees, but after looking at the Java source recently I'm wondering if that's true. Whatever the reason is - my bot only works on a fresh JVM (well ... one that it hasn't run on itself before). SO, I modded my rumble client to re-start after running any battle that my bot is in. If you call that foul play, I would understand. If you're skeptical of my integrity, I'll gladly send you the bot to try yourself, it does very well. Actually, my bots are RWCL, I just haven't posted them to the repository lately. Sorry, I hate to make a bad entrace to the community! -- Simonton

I'm definitely not skeptical of your integrity. I've really enjoyed your contributions to the wiki and the Rumble. However, this seems like a dangerous stretching of the rules. If you need to modify your rumble client for your bot to perform well, what's to stop other people from doing the same and isn't that making the playing field a bit less even? Even without modding a client, I know Chalk skips turns more often on weaker machines and his performance suffers. Given that, I could limit it so that only the client on my beefy machine processes his battles. In theory, his score in the Rumble would improve. My question is: is that an accurate and valid score given the limitations that were built into the Rumble's peer-processing model?

I'd like to hear what others have to say. Please don't change anything until we get some feedback. You're a great member of this community and you certainly haven't done anything wrong. I'd just like consensus as to how we handle modded clients or exclusive battle processing. Cheers. --Corbos

Well in Simonton's case its not a speed or turn skipping problem, its a memory problem. As making a robot with his pattern matching method in a mini requires the use of memory devouring methods, however I don't see the problem if Simonton were to make a Mega, as they not having size restrictions could use a less memory destructive method. So I think its mote the fact of limited space vs memory problems. I believe he was just saving us from tons of frustration having to kill roborumble@home after it ran a battle against one of his bots. So I don't think its a so much a question of should we allow exculsive battle processing so much as how much we sould limits a robots allocation of memory, which in this case is persistant.

As far as I know it could be a robocode problem, as I have noticed its running size gets bigger after running battle after battle and doesn't seem to dump the data between rounds, forcing me to restart after a few battles against the big bots. Perhaps if robocode called directly to the garbage collector after a battle to clean up the threads and free the memory (i'll experiment with my cvs build later, as I have a copy ofa newer WeeksOnEnd to test it with). --Chase-san

Well, before this circumstance, I would definitely have said that all bots should always be peer processed and to do otherwise is unfair. (And I think I have said/implied exactly this in some previous situations.) I still have that feeling about it, but this circumstance has made me think twice... I also think it's unfair to Simonton that his tank should crash in the middle of some battles, based on what and how many previous battles were run in that iteration of the RoboRumble client. I've tested his tank on my machine and it is clearly not using excessive memory in a single battle... I guess the ideal situation would be to fix the issue with Robocode itself, if possible, and forbid the exclusive battle processing? Perhaps I'll test some more with WeeksOnEnd to try and get a better understanding of the situation, as well. -- Voidious

Yeah, I've never really been sure if I bought the thing about string caching, but I've seen similar problems with Robocode before, particularly that it doesn't seem to let go of references to instances of bots. I've seen this as a bigger problem within a battle (stuff that's allocated every round as a member variable of the robot's class never gets released), and that tends to be fixable (either make it static or nullify your larger structures in your Death / Win events). As for between battles, it would sound like Robocode isn't letting go of entire classloaders. Running System.gc() after a battle is something that sounds like it would work, but probably wouldn't :-) -- Kawigi

Robocode uses Thread.stop() to stop the threads, where as it should be done via a viariable within the threads themselves tos top it, I am unsure to if theya re ever avctually deleted, via join or interrupt (I think, i'm not a expert on threads). A simple thread = null should allow the garbage collector to remove the thread(I think, again no expert on threads), if not then perhaps run a call by robocode that sends a call to the robot to finalize itself(which is far as I know impossible or dangerous). --Chase-san

Is there some problem with the most recent version of StrengthBee?? Or is it normal these days not to see a single battle run for a new entry after about 12 hours? -- Greywhind

Looks like a problem downloading it from the repository: [StrengthBee Details page]. Try downloading the JAR from there, it shows an error. Try re-uploading it I guess? -- Voidious

I just tried re-uploading it - but the error still says it's not found... Is this a bug with the repository? For now, I'm switching the download location. EDIT: I apparently uploaded a carbon copy of StrengthBee? 0.6 as 0.6.1. I will fix that now with an "0.6.2." Hopefully this time, I'll actually do it right. -- Greywhind

Hey Chase-san, looks like a broken link you've got up for Genesis Demo3 right now. My client requests that you upload it so that it can run your battles. :) -- Voidious

Hum, thats odd, I thought it uploaded. The one time I forget to double check. Also you can call me Chase the san is just formal. Uploaded. --Chase-san

Ah, that explains the "-san". You either know or are Japanese, right? -- Nfwu

If I remmeber correctly "-san" is used in more languages then just japanese, which i'm not. I however do know some Japanese, but not enough to fluently talk with someone even half the time. --Chase-san


And who will have the honour (or the nerve) to enter participant #500 ?? -- GrubbmGait

winamp32.micro.MicroMacro? 1.0, and krillr.mega.Psyche 0.0.3 aren't availiable to download for me, can anyone mirror them?

No problem: [winamp32.micro.MicroMacro_1.0.jar], [krillr.mega.Psyche_0.0.3.jar]. -- Voidious

Thanks, and I think this page is due for an archival, because its become massive :P -- Baal

Simonton's LifelongObsession = uh oh! ;) I think it's great you adopted that name. -- Voidious

:) Hey, that was a great name! You can't argue with that. Thanks for suggesting it! Let me know if it runs in the clients ok (e.g. whether it runs you out of memory). -- Simonton

I wish I could take credit, but it was David Alves who suggested it, I believe. -- Voidious

Bleh, I already see a bad result for LifelongObsession that I know came from my 1.1.3 client. I am upgrading it to 1.1.5 right now, hopefully it will work fine then. I thought the system discarded results with a 0 score, though, I don't know why it didn't just discard it. Anyway, you might want to repost as 0.11 :-\ -- Voidious

Oh shoot I forgot about that. Darn. I know I can't release that stuff, too! Sorry about that. I should probably take it off the participant's list until I fix it, unless you're the only one running the old client. -- Simonton

I really don't know who's running what :-\. I'd probably just skip the Rules class if it were me, as it's easy enough to code that functionality yourself in a MegaBot. In any case, lots of people are using Rules now, so it's probably about time we all start running 1.1.5 or later. -- Voidious

Oops, sorry for the mis-placed credit, David (if you still look at this)! So then, David Alves, that was a great name suggestion! -- Simonton

Hey Simonton - what you did will work fine for new results, but the old / bad results will not be erased. You'd have to change the version number to start fresh and lose those. You're likely to get more battles against those 2 bots that will partially offset the 0 scores, but not completely. -- Voidious


I started working on updating the RRAH client code to work with version 1.2.5A of Robocode. After deleting the RoboRumble .class files and recompiling, these are the 10 errors I'm seeing. Many of them seem easy to fix, but I'm just going to post them here for other people (hopefully Fnl =)) to give input and point out any flaws I'm making.

./roborumble/netengine/BotsDownload.java:223: cannot find symbol
symbol  : method copy(java.io.File,java.io.File)
location: class robocode.util.Utils
                        try { Utils.copy(new File(filed), new File(finald)); }
                                   ^
// This method is no longer in the Robocode API, but it is easy enough to write our own
// method based off the old source, which I do have.

./roborumble/battlesengine/AtHomeListener.java:7: package apv does not exist
import apv.TeamCompetition;
           ^
// I'm not sure about this one yet...

./roborumble/battlesengine/RobocodeEngineAtHome.java:9: cannot find symbol
symbol  : class FileSpecificationVector
location: package robocode.repository
import robocode.repository.FileSpecificationVector;
                           ^
// This internal structure has changed and the class file used is different now.
// Should be a straightforward change, I think.

./roborumble/battlesengine/RobocodeEngineAtHome.java:14: cannot find symbol
symbol  : class Constants
location: package robocode.util
import robocode.util.Constants;
                     ^
// This stuff's in the Rules class now, another simple fix.

./roborumble/netengine/ResultsUpload.java:156: cannot find symbol
symbol  : method copy(java.io.File,java.io.File)
location: class robocode.util.Utils
                        try { Utils.copy(new File(tempdir+"results.txt"), new File(resultsfile)); }
                                   ^
// Same as #1.

./roborumble/battlesengine/RobocodeEngineAtHome.java:49: cannot find symbol
symbol  : variable Constants
location: class roborumble.battlesengine.RobocodeEngineAtHome
            Constants.setWorkingDirectory(robocodeHome);
            ^
// Same as #4.

./roborumble/battlesengine/RobocodeEngineAtHome.java:85: cannot find symbol
symbol  : class FileSpecificationVector
location: class roborumble.battlesengine.RobocodeEngineAtHome
                FileSpecificationVector v = robotRepository.getRobotSpecificationsVector(false, false, false, false, true, false);
                ^
// Same as #3.

./roborumble/battlesengine/RobocodeEngineAtHome.java:85: cannot find symbol
symbol  : method getRobotSpecificationsVector(boolean,boolean,boolean,boolean,boolean,boolean)
location: class robocode.repository.Repository
                FileSpecificationVector v = robotRepository.getRobotSpecificationsVector(false, false, false, false, true, false);
                                                           ^
// This is probably similar to #3, but I need to dig a little more to be sure...

./roborumble/battlesengine/RobocodeEngineAtHome.java:94: cannot find symbol
symbol  : method setLogListener(robocode.control.RobocodeListener)
location: class robocode.util.Utils
                Utils.setLogListener(listener);
                     ^
// Not sure about this one yet.

./roborumble/battlesengine/RobocodeEngineAtHome.java:97: cannot find symbol
symbol  : method startNewBattle(robocode.battle.BattleProperties,boolean)
location: class robocode.manager.BattleManager
                manager.getBattleManager().startNewBattle(battleProperties, false); 
                                        ^
// Not sure about this one yet.

Note: ./roborumble/battlesengine/RobocodeEngineAtHome.java uses or overrides a deprecated API.
Note: Recompile with -Xlint:deprecation for details.
Note: Some input files use unchecked or unsafe operations.
Note: Recompile with -Xlint:unchecked for details.
10 errors
I have an old Robocode source tree, probably from 1.07, but not the latest one. Grabbing that should give me the answers I don't have. Fnl, if you see this, I wouldn't mind any corrections or advice you might have. =) Cheers.

-- Voidious

To the author of Flower, your bot is crashing on the RoboRumble@Home clients because it uses the Rules class. There is not yet a RR-compatible version of Robocode that implements that class. You can easily implement the functionality of those methods yourself and re-release, and your bot will then get a real score. Make a page for yourself and your bot, too! -- Voidious

Same goes for GottesKrieger?. It seems a quite decent bot, but that does not show in the rankings. Re-release without using the Rules-class and the real strenght of your bot is revealed! -- GrubbmGait

Hey, is FrankTheTank? compiled for Java 6? I get an "unsupported version number" error when trying to run it. You should compile with 1.5 (or 1.5 compatibility) to be sure everyone can run it, or you'll get some very low scores. ;) -- Voidious

hennes.SimpleBot? also seems to be java 6, it runs fine under my newer machine but on the machine that hasn't had an update the roborumble client reports scores of 350-0 100% of the time. --Skilgannon

I've removed SimpleBot? from the Participants list. It's been almost 2 weeks since we brought this up and the author has neither updated it nor commented on the wiki. To the author - please make your bots Java 5 compatible; Java 6 is not yet required for Robocode and it's not fair to expect people to use it in order to get battles run against your bot. I'd normally never do this, but it's been almost 2 weeks since we mentioned it with no response from the author and I just noticed it has wasted hours of my RoboRumble client's CPU cycles. -- Voidious

Noticing your comments...How do you compile bots in a lower version? Is it (from the compiler preferences): Compiler Classpath: spath "C:\Program Files\Java\jre1.*4*.0_01/lib/rt.jar";libs/robocode.jar;robots or some other version number there (the asterisks werent there i added them for emphasis)? And do you have to install the older jvm? --Starrynte

I'd have to Google the compiler options, but you definitely don't need the older JVM. I use Eclipse which has a very simple preference setting for compatibility level to use. I found the answer on [this page]: the command to pass to javac is -target 1.4 (or 1.5, 1.2, 1.1, etc.). -- Voidious

Kev seems to be on vacation from Robocode for a while and I am pretty sure he never meant to permanently have two WaveSerpents in the rumble, so I'm going ahead and removing the lower ranked one from the participants list. -- Voidious


Robo Home | RoboRumble | Changes | Preferences | AllPages
Edit text of this page | View other revisions
Last edited September 19, 2007 14:06 EST by Voidious (diff)
Search: