SVN Update Notes
If you update from SVN, be ready to change your .cfg files. The example "run.cfg" is updated - it's still easy. Perhaps easier than they were! --Simonton 04:25, 16 September 2008 (UTC)
HSQLDB No More
Beware of abrupt shutdowns of the database server. I just lost all my scores between Simonton/PFResearch 0073 and 0083, apparently because I closed Eclipse which simply killed the database process, instead of shutting it down nicely. I certainly thought it would handle such things better than that, and I have done it many times before with no consequence, but apparently my timing was bad this time. Does anyone have a good recommendation for a more reliable database that can be run without firing off a separate process, if desired? --Simonton 07:51, 18 September 2008 (UTC)
It looks like the Apache Derby implementation distributed with Java 6 would be the natural choice! Expect that change to come in the near future, for fear of losing more of my data otherwise! --Simonton 15:04, 18 September 2008 (UTC)
It happened again. This time I only lost some data for 0088. I wasn't going to be upset if I lost alot more, but it just confirms this is a switch that has to be made. For now I'm making a copy of the database directory before I kill the database server. Sending the sever the shutdown command via SQL would work, but that's too much effort. They should make it accept a clean shutdown command from the terminal in which you launch the server, that would be handy. --Simonton 02:21, 23 September 2008 (UTC)
Buggy when running from GUI
Hey Simonton if you're around. I found a bit of a odd bug when trying to run the TCRM here. Under the GUI it always fails with:
Thread 1: Unrecognized output from robocode, "Aborting battle, could not find robot: apv.AspidMovement 1.0". Killing battle.
whereas in the CLI it works just fine. Not sure why all this is but if I find a fix I'll tell let you know. --Rednaxela 07:10, 21 December 2008 (UTC)
Hmm, okay, I discovered that the problem disappaered if I cleared the working_dirs directory, and renamed all the bot jars to match the version number they have in the properties file. It seems that RoboResearch doesn't like jars that are differently named than the default, and some such non-default names are included in the zip files for challenge bots. Anyways I have it working now, and maybe this info posted here might help someone else. --Rednaxela 19:06, 21 December 2008 (UTC)
- Ran into this - thanks for the info --Miked0801 02:13, 22 May 2009 (UTC)
Has anyone used RoboResearch to run melee battles? Looking through the code, it seemed like the part that runs battles would support running a melee battle, but I didn't see any way to configure this (from a .rrc file or otherwise). My melee Test Bed process is currently pathetic and verging on useless... I'd love some RoboResearch support to help me out. --Voidious 15:06, 18 May 2009 (UTC)
I'll take that as a no on anyone using it for Melee? I'm going to try and get it going, then. I'm already like a fish out of water in the Melee arena, a powerful testing tool would help me big time. --Voidious 15:41, 19 May 2009 (UTC)
Sweet, got Melee working (thanks to a lot of the guts already supporting it), though it's not incredibly elegant and the output isn't quite right yet. Good enough to be my new Melee testing utility, though. =) I'll post it or contribute it back to Simonton if/when I clean it up a bit. --Voidious 02:04, 20 May 2009 (UTC)
Got a much more polished version of the Melee support working, doing it in a way that seems mostly elegant in the RoboResearch code. My previous hack was ugly and just didn't always work right. Most of the internals actually already supported Melee, so it wasn't that hard once I figured out the overall structure of the code.
- If you have "Melee" in your challenge name, it uses Melee mode. This probably should be done another way, but this makes sense, works fine, and it's easy to make it something else if we want.
- If it's Melee, all the bots in a category are put in the battle at once. Multiple categories gives you multiple battles per season.
- 10 bots per battle max, a limitation in the existing code / database structure. Also, it uses 1000x1000 for battles with > 2 bots, that was already in there too.
- Scoring is displayed just like it would be in the MeleeRumble: one battle yields 9 separate scores, with the category score being the average of those, and overall score being average of categories.
What I don't have done:
- When the CLI is showing what battle it's running, it just says "<challenger> vs <first reference bot>".
- I haven't worked with the GUI at all because I am using the CLI. The only change I needed to make to the CLI (besides the previous cosmetic change) was for the JAR copying: before, it was copying the first bot from each battle, while obviously I need to copy all bots from each battle.
I'll share this and/or share it back to the SourceForge project shortly. I feel so much less lost in the Melee world being able to run some serious benchmarks. =)
- Please do - Infinity and DustBunny would love some number tweaking help to unseat Kawigi's monster Lib. The GUI stuff is my current preference, but beggar's can't be choosers :) --Miked0801 16:26, 28 May 2009 (UTC)
--Voidious 03:12, 28 May 2009 (UTC)
Alright, just made a few tiny cosmetic changes to the CLI and the GUI (the "<your bot> vs <reference bot>" stuff). Viewing results for 45 bots (my current melee test bed) in the GUI isn't pretty, but you can do the "Copy as wiki" into a text editor if you need it more compact. Not that it's that pretty as pure text, either, but at least it's not a 2500 pixel wide window.
Anyway, here it is: void_roboresearch_melee_01.zip. That's the same base stuff you'd get from the SVN checkout plus the binaries compiled for 1.5. You can just overwrite your 'bin' dir with the one from this .zip if you already have RoboResearch setup. (Hopefully Simonton doesn't mind me distributing it like this, but I saw no such mentions in the source and I'm going to get this stuff back to him sometime...) Let me know if you hit any problems - enjoy. =)
--Voidious 01:08, 29 May 2009 (UTC)
- Grabbed and in use - thanks! --Miked0801 10:23, 29 May 2009 (UTC)
- Is this still the latest version? I'd like to merge this with my hacked-for-annoying-thread-stoppage version and do some melee testing =) --Darkcanuck 21:58, 5 September 2009 (UTC)
- Yep, I haven't made any more changes since then. Let me know if you have any issues. --Voidious 22:35, 5 September 2009 (UTC)
Getting started instructions
Has anyone here got this to work at all?!? It'd sure help with testing... --Miked0801 23:36, 18 May 2009 (UTC)
You mean like to work for 1v1 / not Melee? Yeah, I've been using it. What kind of problems are you having? I tried long ago and hit issues which I attributed to the Mac JVM. Since I've been Robocoding again and using SoyLatte Java, it's worked fine, but it's also a different version of Robocode, so it could be that too. --Voidious 02:35, 19 May 2009 (UTC)
- Mainly, I'm not a java hack. There are no step by step instructions nor a downloadable distribution so I didn't even know where to start. Being able to run batch battles locally would be awesome. --Miked0801 05:04, 19 May 2009 (UTC)
- I bumbled through a lot of stuff, myself, but I did get it working. And it's great once it's working. =) I've posted a quick and dirty write-up of the process below. It's for Mac / Unix, while for Windows you'd use ; instead of : for Java classpath stuff. But I also zipped and uploaded my roboresearch dir (minus a few dozen of my dev versions) here, if that helps. Indeed we should update the page with some polished instructions, but I just wanted to get something up before bed here...
- Thanks for the files, I bumbled around and actually installed the sun java compiler, then decided to just use your files. Everything seems to run just fine, though the CLI output is ugly. I'll dork around with the GUI next. --Miked0801 01:05, 22 May 2009 (UTC)
- And I have fallen in love with the GUI version of this. This will allow me to test locally instead of testing distributed over the internet. Wee! --Miked0801 02:13, 22 May 2009 (UTC)
to install from scratch: svn checkout https://roboresearch.svn.sourceforge.net/svnroot/roboresearch cd ./roboresearch/trunk mkdir bin javac -cp src:hsqldb.jar -d bin src/roboResearch/CLI.java install a copy of Robocode to roboresearch/trunk/robocode_install put your bots in robocode_bots run battles with something like: java -cp bin:hsqldb.jar roboResearch.CLI myBot.cfg for multi-threaded, setup server with database_server.sh, then do something like: java -cp bin:hsqldb.jar roboResearch.CLI -S -t 2 myBot.cfg docs/getting_started_running.txt has some info, but some is outdated (boo)
--Voidious 05:28, 19 May 2009 (UTC)
- Oops, I nearly forgot, one more important detail: I did have to actually edit the source to fix one thing. Some Robocode message had its wording changed, and RoboResearch was aborting if it didn't recognize the message. (I should find that and fire it back to Simonton.) So anyway, you should definitely grab my .zip and use that or compile from that source. --Voidious 05:33, 19 May 2009 (UTC)
- Well, or instead of the CLI thing, you can use the newish GUI version. I find it's rather nice, and if you want to do multi-threaded it can manage all the threads and has no need to seperately start the database server etc. I find it rather nice to be able to compare versions directly in the GUI version as the battles run... ;) --Rednaxela 07:21, 19 May 2009 (UTC)
- Again, put a distribution or something together with instructions. I'm a C/C++/Asm programmer, not a Java dude - as you'd know if you look at my very C like code :) --Miked0801 16:10, 19 May 2009 (UTC)
Hi! I'm also interested in RoboResearch but I don't get on well with it so far. :(
- I downloaded voidious's zip and tried to run the run.bat. : 
- Then I modified TUI to CLI in the batch file. : 
--HUNRobar 17:13, 13 June 2009 (UTC)
- run.bat is out of date and only runs some of Simonton's tests anyway. Best way to get started is to use the GUI: "java -cp bin:hsqldb.jar roboResearch.GUI" (mac/linux) or "java -cp bin;hsqldb.jar roboResearch.GUI" (windows). Remember to start this from inside the roboresearch dir. --Darkcanuck 17:27, 13 June 2009 (UTC)
- Argh, classNotFoundException again.. --HUNRobar 17:34, 13 June 2009 (UTC)
- Just downloaded Voidious's zip and I see that everything's actually inside the "trunk" dir. So you'll need to cd into roboresearch/trunk before running the command I suggested. I have mine setup slightly differently but this should fix that exception... --Darkcanuck 21:08, 13 June 2009 (UTC)
When running multi-threaded I didn't setup server with database_server.sh nor used the -S option as suggested in the instructions above, I just used the -t 2 option, is there any problem in doing so? --Navajo 21:54, 23 July 2009 (UTC)
- Have you tried the GUI? No command line options and multi-threading is just a few clicks. I'm never going back... --Darkcanuck 04:34, 24 July 2009 (UTC)
- I tried the GUI, but I like the CLI, to run multi-threaded instead of a few clicks I only need a few digits. I just wanted to know if there was any problem in just using -t 2 to run two threads instead of setup the server. I did it both ways and didn't notice any difference. --Navajo 13:59, 24 July 2009 (UTC)
- Hmm, I have no idea. I thought it just wouldn't even let you do it. =) Maybe he updated the code at some point, but not the instructions? Also, just to be sure, do you have a dual core CPU or multiple CPUs? If not, multi-thread won't really do much for you, but it might make bots skip turns since they have less CPU time... --Voidious 15:55, 24 July 2009 (UTC)
- I'm using the zip file you posted here, and it allows multi-threaded this way. I have a dual core CPU, so this multi thread is realy useful to me. --Navajo 16:17, 24 July 2009 (UTC)
I just trued running robo research on ubuntu 9.10 and for some misterious reason I'm unable to run more than 1 thread. Does anyone have any idea of how to solve this? --Navajo 23:45, 17 January 2010 (UTC)
- What error are you getting or what is the symptom? Are you running the database by itself and using -S when you run RoboResearch? (I know someone, maybe you, said they didn't need to, but the instructions still say to do that for multiple threads.) I haven't personally tried it multi-threaded on Linux, but I can't imagine why it wouldn't work. --Voidious 01:58, 18 January 2010 (UTC)
- I'm using only the GUI to run everything. What happens is that one thread runs normaly while the other gets stuck at starting forever --Navajo 03:49, 18 January 2010 (UTC)
- What version of robocode is being used? Perhaps it has to do with that. I haven't used multi-threading much myself, but I recall it working in the past (I personally avoid using the multi-threading though for two reasons: 1) My laptop's CPU gets hotter than I'm comfortable with, and 2) I can't do other things on the system at the same time and be sure I'm not causing skipped turns) --Rednaxela 04:01, 18 January 2010 (UTC)
- I've tried both 188.8.131.52 and 184.108.40.206, and none of them worked. I usually leave robo research running at night, so there is no problem with doing other things on the system, and I don't care much about the heat (what is the point of having a good computer if I can't use its resources...)--Navajo 22:15, 18 January 2010 (UTC)
- Is there a roboresearch for dummies page ? when i type "javac -cp src;hsqldb.jar -d bin src/roboResearch/CLI.java" It tells me:
"javac is not reconised as an internal or external command..." -Jlm0924 20:37, 14 May 2010 (UTC)
One thing that is sorely missing is a standard deviation calcuation per battle (and/or overall) so that you can get a feel for just how accurate your results are. It would go a long ways towards telling someone just how many seasons they should run to get stable results. --Miked0801 20:56, 2 June 2009 (UTC)
- I believe it shows something like that if you hover the mouse over a score for a moment in the GUI. ;) --Rednaxela 23:05, 2 June 2009 (UTC)
Is there a way to not have it bring a thread down on an exception? MosquitoPM cost me 1/2 a night of processing as it sabotaged all 3 of my threads :) --Miked0801 16:45, 16 June 2009 (UTC)
- No, I think now. When I ran RoboResearch, I need to get up every 3 hours or so to check if they are running correctly. » Nat | Talk » 16:58, 16 June 2009 (UTC)
- I hacked my version to handle a few more error messages so that it doesn't stop as often. I think the robocode console output has gotten more wordy than when Simonton first put this together. The relevant code is easy to spot in LineListener, near the bottom of src/roboResearch/engine/BattleRunner.java. I should add exceptions though, I'm constantly restarting my threads... --Darkcanuck 05:15, 17 June 2009 (UTC)
- Personally... I really find it preferable to stop running a thread if there's an exception, because 1) if my bot is throwing an exception, I've got a bigger problem then benchmarks would help with, and 2) If a bot in the test bed is throwing an exception, then well, it's a bad test bed... --Rednaxela 06:31, 17 June 2009 (UTC)
I temporarily deleted the package toys (on the local copy), as it was giving me errors, will it still work? --Starrynte 02:06, 23 September 2009 (UTC)
Update for 1.7.*
I've used it with multiple 1.7.x versions, the latest being 220.127.116.11. Are you hitting any issues? --Voidious 17:16, 14 July 2011 (UTC)
- Yes, I am using the one pulled from the SVN. While I have been fixing the problems as I go. I just sorta want to use it. Is there a newer version I don't know about? — Chase-san 18:22, 14 July 2011 (UTC)
- Nothing official. I ran my changes by Simonton at one point but we agreed they needed some polishing. But, of course, polishing goes on the back burner when what you have works and you're in the throes of addiction. =) I did post my binaries under the "Melee" section of this page, with my hacked Melee support and whatever else I had done: void_roboresearch_melee_01.zip. --Voidious 18:27, 14 July 2011 (UTC)
I've been using RoboResearch some lately, and have a few suggestions.
- This one is just my own opinion, and others may see it differently. In the Results window, the "Total" is an average of the "Sub" group totals. I think this is misleading, and should be an average of all individual values. As it is now, I don't think it is intuitive, and you have to be very careful to balance each of the sub groups equally for the total to be meaningful. I know I can get around it by just eliminating all of my sub groups and running every robot under a single group so that I get a total that is of value to me, but I don't think I should have to do this.
- The results and seasions windows should have the table in a JScrollPane.
- Threads should have an option to set a number of retries when an exception occurs. Furthermore, if the number of retries is exceeded, it would be very nice if there was an option to just ignore the robot and continue on.
- There needs to be an obvious way to reset challenges. For example, when I was having trouble with some robots throwing exceptions and pausing the threads (see previous item), the best thing to do was just remove the trouble robot from the challenge. However, when starting back up, even if the challenge is removed and readded, the old data for the challenge still gets picked up. It is very much not clear how to reset this to start fresh.
- Side note to this: I discovered that you can update the groups in the rrc file, remove the challenges from the GUI, then readd them. When you do this, it picks up the old battle data and puts it in the new grouping. This is nice, though I had to discover it by accident. And I still think it would be nice if there was an obvious way to clear out data so you don't have to keep changing the version numbers for every quick little bug/issue fix.
- When there is only a single thread in shown in the GUI, and you try to remove it, it does not clear out the thread controls box.
-- Skotty 21:43, 18 August 2011 (UTC)
- This isn't how challenges usually worked on the wiki, which what RoboResearch was designed with in mind. Just create that other challenge and open it in roboresearch after you have run the battles to get the other result.
- Yes, it probably should. :)
- Yes, again it probably should.
- There was at one point a way to goto the additional add bot screen and erase a robot and delete all its values from the database as well, but that doesn't seem to work anymore (or yet?). It stores all battles, and just plugs them into whatever challenge you decide to add/run later.
- This is expected behavior. Whats the point of having an empty thread box?
- Regarding "This is expected behavior. Whats the point of having an empty thread box?" -- If you look closer, you'll find that clicking the Remove button on the thread when it is the only thread in the box, the Remove button keeps it's depressed appearance and the controls on it no longer work. This happens because it's not actually there; the window needs a repaint. This can be demonstrated by then resizing the window, which causes a repaint, and the thread box disappears. -- Skotty 00:36, 19 August 2011 (UTC)
Re: #3, I hacked my version to restart the thread after a bot exception (the result still gets thrown away). Probably not too hard to implement a retry either. For some reason, bots such as Phoenix crash occasionally on my system -- without this hack, I was constantly restarting threads. I keep an eye on the first challenge round to make sure my bot runs ok, just in case. --Darkcanuck 05:14, 19 August 2011 (UTC)
- I suppose that wouldn't be too hard to do since I am running it from Eclipse. Pretty quick and easy to make a code change. It definitely needs it. I hate coming back to my PC 4 or so hours after starting a challendge to find that some bot I added to a challenge threw an exception and the thread(s) have been sitting idle for hours. I ran into this most recently with LemonDrop, which runs fine most of the time, but on infrequent occasion throws an exception (array index out of bounds, iirc).
New Version 1.2 (finally)
I have updated RoboResearch to version 1.2. It has been a long time since it has had much done on it.
- Resolved numerous repaint errors in the GUI.
- Now with configuration pane and working button. :)
- Allows you to change the challenge name and number of seasons running.
- Yes, it does save these changes between open/closing.
- The result and season panes display a vertical scroll bar when needed.
- Threads now automatically restart after 5 seconds when they fail.
- Now runs robocode in Debug mode, to avoid skipping turns when you are doing other things as well as testing.
- Now produces tabulated data when you copy from the results and seasons tables (instead of java references)
- Updated for 1.7.X (awhile ago)
- The configuration pane doesn't resize correctly.
- Add an option menu (or configuration file actually)
- Fix the "Delete Bot" and "Delete Versions" buttons in the Version Browser (to delete all results for that bot/version)
- Move the "Remove" button to the bottom of the button pane, away from the results button.
- Ability to access the Version Browser from the main menu
- Ability to automatically find open results browsers for the same challenge and add results to that when clicking results
- This is a touch more involved then you may think.
Cool - I've been thinking for a while someone needs to fork this and start updating it again. I don't think I agree with running in Debug mode by default - I don't really want skip turns masked in my tests, nor do I want my tests to run slower against bots that skip turns (I run a lot of tests against large fields of bots). I think Darkcanuck, I, and some other people might have some misc fixes that should be integrated, too, I'll look into my changes. --Voidious 12:34, 25 August 2011 (UTC)
- It'll be the first thing I add to the configuration settings. (I felt this way too) — Chase-san 13:05, 25 August 2011 (UTC)
- Can you make the command line string part of the configuration? That's one of the first things I change. I think the rest of my hacks were superseded by Voidious' melee version plus your thread restart modification. Thanks for taking this on! --Darkcanuck 16:38, 25 August 2011 (UTC)
I currently work on distributed analog of RoboResearch. It's SOAP-based web service for battles execution and client. So we will can contribute out CPUs. It's now supports .rrc challenges format and will support wiki format for output results, so there're no big migration problems. I plan to release alpha version of it under open source licence in next few days. I need help, especially in UI part. So, may be, our community can switch development effort to this tool? --Jdev 12:54, 25 August 2011 (UTC)
- Are you sure it wouldn't be easier to modify RoboResearch to run in a distributed manner? I would be better to consolidate our efforts if possible... --Skilgannon 13:12, 25 August 2011 (UTC)
- IMHO roboresearch has too big design restriction - run robocode through command line for every battle. my tool use robocode API to run battles, instead of run it through command line. It allows avoid overhead for startup robocode and terrible robocode output parsing --Jdev 13:30, 25 August 2011 (UTC)
- One thing to keep in mind... is we already have a client for distributed execution of battles. The roborumble client. Making a new one seems really silly to me when the existing one could be adapted/extended. I'd much rather see an extension/revamp of the roborumble client protocol than some new SOAP-based thing. --Rednaxela 13:19, 25 August 2011 (UTC)
- how you suggest to post battle request to execute by this way? --Jdev 13:30, 25 August 2011 (UTC)
- The roborumble protocol already has an existing mechanism for that sort of thing. The "priority battles" mechanism allows a rumble server to ask the client to execute specific pairings. The current version of Darkcanuck's rumble server recently started making use of this long-existing feature.
- The two main modifications/extensions that need to be made to the client/protocol are 1) a configuration that doesn't execute any battles *except* "priority battles" requested by the server, and 2) Allow the rumble server to specify parameters like field size and number of rounds, instead of it being in a client-side configuration file.
- --Rednaxela 14:34, 25 August 2011 (UTC)
- There're many disadvantages:
- 1. You need client which will post requests anywhere, monitor they's state and show results.
- 2. I wish to have possibility to local execution, so roborumble's client must modified to support several servers (common in internet and private in LAN).
- 3. I did not try, but i think roborumble's server is more heavyweight and difficult to run and maintain
- 4. What to do with battles, which fetched by roborumble's client and then this client is shut down? There're no feedback from clients
- 5. You need to solve problem with bot's code hosting and versioning bot's with same version, but different code (you will not increment version for every build of your's robot)
- 6. the system with one server is less fault free
- 7. i wish to have possibility to run my battles on my clients in prioritet order - but roborumble does not support it now
- 8. 3 step system (challenge client -> roborumble server -> execution client) is more slow
- 9. i think there're many other issues, which appears, when tools misused
- --Jdev 16:11, 25 August 2011 (UTC)
- Interesting idea (using the rumble client) but would require some heavy client modifications. Jdev, your central app would replace the role of the server (you wouldn't use the rumble server), clients would request battles and your server would delegate the battles to be run.
- For the record, my server has always used the priority battles feature. The recent updates make much heavier use of this feature, but it was there since day 1. --Darkcanuck 16:38, 25 August 2011 (UTC)
- hmmm... currently i can not imagine this solution. How you will initiate battles execution? By third app (it's what about i tell) or by modifying roborumble's client, so that it can be runned in active mode? Any way, suggested tool is open source and with open interface, so there're may be many clients:)) --Jdev 16:50, 25 August 2011 (UTC)
- Early access (i hope ready to run) links: client , server . Just try it, may be you will like it. (requires Java 7!) --Jdev 13:41, 25 August 2011 (UTC)
Have you updated the parsing so that it handles 1.7.* style printing and results? --Skilgannon 13:12, 25 August 2011 (UTC)
- No source??? If you can't get Simonton to give you write access to the sourceforge project, I can host an svn repository for you on darkcanuck.net. --Darkcanuck 16:44, 25 August 2011 (UTC)
- Yeah, it would be good to get a new source tree up on GitHub or something. First and foremost, I think we should point BattleRunner to robocode_install/libs/robocode.jar instead of lib/robocode.jar! If we make you install Robocode to a specific dir, it's pretty silly to look elsewhere for the JARs. I'd also patch the issues related to the CLI ("unrecognized output" for a bunch of new messages) and so the Mac doesn't pop an icon into your dock every single battle. --Voidious 20:21, 27 August 2011 (UTC)
- In the meantime, if you want to cherry pick from my changes, I uploaded my patch (vs the latest Simonton source) here: . Includes both source trees and a .txt of the diff -r -w output. What I've added:
- Melee support (hacky but functional)
- Looks at robocode_install/libs/robocode.jar instead of lib/robocode.jar.
- Make the CLI work with newer Robocodes - adding lots of acceptable outputs to BattleRunner, and tweaking Result.java to ignore the " (%score)" added to the raw score before parsing it.
- On Apple's JVM, doesn't pop up a Java icon in the dock every single battle (which also steals focus, driving you insane).
- IIRC, the one change to Database.java was a bug fix I found while adding Melee support.
- --Voidious 21:18, 27 August 2011 (UTC)
- In the meantime, if you want to cherry pick from my changes, I uploaded my patch (vs the latest Simonton source) here: . Includes both source trees and a .txt of the diff -r -w output. What I've added:
Another thing that I think it needs -- the preferred challenge slider should be updated to slide from 1 to the number of challenges in the queue, instead of fixed from 1 to 9. And even more useful, have an option to run the challenge battles in random order. I made this change in my local copy. It allows me to run a variety of challenges more or less in parallel. This is better, because I can set up a set of challenges with a high number of seasons, let it go, and when I check on it the next day I am gauranteed having as many results as possible from all challenges.
|Thread title||Replies||Last modified|
|Buggy debug mode||4||18:08, 5 August 2012|
|new Robocode batch battle runner||18||17:48, 25 July 2012|
|Development||2||18:46, 28 September 2011|
|Wacky Scores||6||06:03, 11 September 2011|
RoboResearch is not starting JVMs in debug mode. It passes "-Ddebug" as parameter, but it should be "-Ddebug=true".
So that's just an argument to the JVM, right? Custom JVM args is near the top of my list for RoboRunner, so I'm working on it now. But there are some things like "-nodisplay" that are arguments to Robocode itself, which wouldn't work because I don't launch Robocode (just use the control API from my own program). So I'd have to add arguments myself if I wanted to let you configure it to actually display battles, for instance.
I'll try RoboRunner someday.
I'm strongly considering writing up a new Robocode battle runner to replace my RoboResearch usage. I think I could even have something workable over the weekend. Here's my short list of design goals:
- Easy to setup / configure!
- Use Robocode control API instead of running a fresh Robocode engine / JVM from the command line for each battle.
- Support existing .rrc files.
- Support any battle configuration (battle size, Melee, Twin Duel, etc).
Anyone have any thoughts / requests? I think I would go command line only to start, but maybe if I really get something good I'll consider a GUI.
Now that I can crank out 1,000 battles per hour, the thought of also restarting the JVM 1,000 times per hour kind of makes me cringe. =)
I really should take a look at what you have there, I've never tried it. Though I think I'm looking for something a bit simpler and lightweight. Does the client do multiple threads with multiple Robocode engines? Or do you just have to configure one client per thread, like with RoboRumble clients?
I just whipped up the core battle runner, which takes paths to multiple Robocode install directories, then runs a collection of battles across the threads using the control API. Though that's pretty much the easy part... =)
It uses one thread per server.
Yes, but in my case, i have more CPU power at my work, so support of remote servers was critical for me:) Latter today, i will publish soures on github.
Yeah, support for remote clients is a killer feature in that situation!
Great feature list. Support of current .rrc files is a good idea. You'd need some sort of config file to parse anyway, and it's a nice, easy to read config format anyway.
Remote servers would be fun for me too. I've got my main home desktop, and an identical machine running as our living room media center.. so both of them could run battles for me. *evil grin*
I said it over in my talk page, Voidious, but I'd love to help contribute to this with code and testing. As a teacher my free time will contract somewhat during the school year.
My recent experience with RR has me interested in helping to make something a bit easier to get off the ground.
Hmm. It seems that Robocode itself is not thread safe for running multiple instances? I have multiple RobocodeEngine's going in separate threads, each with its own directory, and I'm still hitting some weird concurrency issues. Bummer! I wonder if I can do something with ClassLoaders to deal with it? I've never really played with them before, and not sure if it would kill performance or not.
Oh, yes i also faced with this problem and also think about class loaders but in result came to current architecture
Yep, the engine is not thread safe.
Creating ClassLoaders should be faster than creating separate JVMs, but don't know by how much.
Another approach would be staying with separate JVMs, but not restarting them before each battle. Exchanging messages between JVMs and using the API inside a loop instead, much like how the RoboRumble client works.
Creating ClassLoaders once and reusing them on many battles is also another approach. I remember trying this approach in a custom multi-threaded RoboRumble client a long time ago. Didn't remember exactly what went wrong. Maybe the need of a custom JAR loading mechanism.
Yeah, having each engine in its own process and doing IPC seemed like a huge ordeal that I wanted to avoid, but now I'm thinking it might not be so bad, and it would be easier and safer than fussing with classloaders. I brushed up on some RMI stuff last night, but even that requires some setup to get the RMI registry going and stuff, which I don't really want to force on people. I think just starting the separate processes with Runtime.exec and communicating with them over STDIN and STDOUT may be the simplest solution all around.
Cool, that wasn't so hard. =) I have this pretty much working now, including reading .rrc files, with option additional stuff like multiple bots per line for Melee battles and custom battle field sizes, and simple/fast/human readable results output to .properties files instead of a database. Still a few things left to do for usability, which is a high priority for me, but I think I can have something posted soon.
~29 mins for one season of a 500-bot tested, vs ~40 mins with RoboResearch. Yum. =)
I've almost got deBroglie's score on the testbed you made for me back to the level of rev108 after my hitrate normalization fixes. It's been a great boon to have a working batch battle runner!
Looking forward to testing this work of yours out!
Sounds awesome. When there get to be a lot of robots in the install directory RoboResearch gets really slow starting each battle... try running a single season against the entire rumble and you'll see what I'm talking about.
I also see big slowdowns once the database gets too big, say 15k battles, which is a pretty frequent occurrence for me these days. It's pretty obvious when the first 6 battles fire right up at startup vs trickling in with half a second between them.
I'm not sure I'll have time to get it polished and posted tonight, but if not, definitely within the next day or two. And I'll post it on GitHub as public domain. It's not even that much code, if anyone wants to learn it to fork it or add a GUI or whatever.
There will still be a bunch of bells and whistles left on the to-do list as of the first release, but for general APS bot benchmarking it's in great shape. I'm using it for my own battles now and it ran overnight with no issues. I think I have the setup process pretty smoothed out, too (albeit Unix only, using a bash script).
Voidious, i like yours solution and think, that 3 tools for batch battles execution is too much. But i still need in execution of battles on remote machines, so i think it will be cool to combine some way of our solutions. Rigth now i have no inspiration for work on it, but if you make good interface for yours tool, i can add my client-server solution in future. What do you think? We can contact via email or skype to discuss interface.
Well, since my new tool is more of a replacement for RoboResearch, that only leaves us with 2, and I don't think 2 is too many. =) If you're only using one machine, then a fast, powerful tool that's easy to use for multi-threaded benchmarks is great. If you have multiple machines, no matter what you do it's going to require some setup, so having a tool for that is great too. My multi-process BattleRunner is pretty simple, though, so it shouldn't be hard to integrate with that if we want to use that in the client part of your system.
Problem in my solution is that you need run several servers on same machine to utilize cpu completly and i wish to run only one server, which will run several engines. Also even 2 is too much. It means that every feature (ui, hit rate colculation, melee support etc.) must be doubled and i think, that better for community to concentrate all resources on one tool. I will review your tool, when you publish it and contact you, if detect troubles in integration or impruvement ideas.
Is Chase leading the support for RoboResearch development now? Just let me know if you would like extra help. I have plenty of experience in Java and Swing.
I might end up redoing a bit of the GUI when I have the chance. But I don't even have time to really work on my robots at the moment, let alone on RoboResearch. :)
With my latest versions of XanderCat 9.0 and 9.0.1, they seem to perform just fine with 18.104.22.168, but I am now getting really wacky scores in RoboResearch. There are no exceptions happening, but the scores are coming up a good 20% to 30% lower than the scores with 22.214.171.124 (e.g. score against Rapture.Rapture is coming up at about 65% in RoboResearch, but a consistent 90% in 126.96.36.199 test runs and in the RoboRumble). I checked to ensure my robot is not throwing any exceptions. Any thoughts on what might be causing this? It is preventing me from using RoboResearch for testing.
It might be due to me setting it to debug mode (it keeps robots from skipping turns, so if enemies were skipping, they are not in RR).
I don't think so. I might have to spend some time digging into the code more. A lot of the docs and pages discussing RoboResearch seem to be outdated. For example, I keep seeing references to having a copy of Robocode installed in the robocode_install directory, but from what I can see, there doesn't seem to be any robocode_install directory. I assumed it just used robocode.jar from the lib directory. Quite confusing. So I don't even know what version of Robocode RoboResearch is using.
AFAIK you're supposed to install a copy of Robocode into robocode_install. But the original RoboResearch also came with a lib/robocode.jar that it used, which I only just recently realized it uses instead of the one in robocode_install/libs. I'm not sure what the problem is, but I'd install Robocode 1.7.3.x to robocode_install, start it up and recalculate the CPU constant, and copy that robocode.jar into roboresearch/lib. See if you still see a score discrepancy.
RoboResearch docs could definitely use a lot of work, but hopefully with Chase taking charge to start updating it again recently, we'll resolve some of the usability issues and keep it better maintained.
I dug into the code a little to figure it out. I didn't realize that RoboResearch just runs Robocode normally using "-nodisplay". That made it easy to debug, because I could just comment out the line that added "-nodisplay" and watch the battle as it ran. The version of Robocode in the lib directory that it is using is 188.8.131.52. For whatever reason, it appears as though the onRoundEnded() event never gets called, which is messing up some of the round data I keep track of. Ultimately, this lead to one of my multi-mode scenarios getting incorrectly activated much more often than it should have been, resulting in low scores. Now I need to either develop a work-around or try to set up RoboResearch to use 184.108.40.206.
Ah, cool. I wondered if it might be some robot method added in 1.7.x, which I think onRoundEnded is. You should be able to copy robocode.jar from robocode/libs to roboresearch/lib to get whatever Robocode version you want.
I modified my RoboResearch to use robocode_install/libs/robocode.jar, so whichever Robocode version I install there is used. That and the other patches I'm using are available here, and described/linked a bit further up this talk page. Recompile with
javac -cp src:hsqldb.jar -d bin src/roboResearch/CLI.java (or GUI.java).