Slashdot Mirror
2004 ICFP Contest Spinoff Game
TheRealFoxFire writes "Taking a page from the popular Corewars competitive programming game,
the 2004
ICFP Programming Contest task has been turned into an online
competition: Ant Wars. In the game, programmers create
state-machine ant "brains" which battle against each other for food and
programmer glory. And just in time for the ICFP contest itself, where this year's winners will be announced."
http://www.corewar.info/ Is maintained by Christrian Schmidt, one of the great corewars men. Or you could go to http://www.corewar.co.uk/ Maintained by John Metcalf, corewar hall of famer.
Or better yet, you could try www.koth.org or sal.math.ualberta.ca and go straight to the hills.
The winners will be announced at the ICFP (International Conference on Functional Programming) itself, not the ICFP Programming Contest. The contest took place in June. The conference is in Snowbird, Utah, September 19-22. See you all there.
"In total there are 182 users online :: 2 Registered, 0 Hidden and 180 Guests"
Looks like a promising audience...
Have you seen my stapler?
Here is a very simple introduction for anyone new to the game and interested in playing.
http://www.koth.org/info/corewars_for_dummies/dum
corewars.org is a quite unlucky link for corewar. Much better sources are:
/ /www.corewar.co.uk
http://koth.org
http://www.corewar.info
http:
Fizmo
Slashdot submitters: try to explain acronyms you use in your submission. Same should go to people who make the original websites too. It took a lot of poking around to figure out ICFP=International Conference on Functional Programming.
The corewars site linked to in the story has never been known to be active or complete.
The KOTH server is home to the "pro" hills of which 94NOP is the most active.
The most up-to-date site for info & links is Fizmo's.
There are beginner's hills and others at
SAL hills
Yellow hills
There's also an IRC channel at irc://irc.koth.org#corewars
Ant wars looks interesting - pity the event is over
As has been mentioned, the high level language is written to be easily parsed and compiled (its recursive and LISPy). The target is actually bytecode, so there is nothing stopping people from writing any sort of syntax, or higher level frontends capable of things that aren't possible in the provided language.
We don't use Java, one, because we don't want people to have to use Java, and more importantly we have to avoid turing completeness both to keep the game simple and elegant, and to ensure a reasonable amount of time to simulate games.
Besides, the Java Security Manager isn't so fine grained that you can turn off arbitrary parts of the Java library. It just prevents certain known dangerous things like I/O, etc.
Robocode is still very much alive.
Proof of this is a very active wiki.
Join, we could use some new cannon fodder :)
Part of the challenge of the IFPC contest was that the ant language is significantly less powerful than what people are accustomed to coding in. In order to write anything that isn't horribly painful, you have to write your own compiler.
The only per-ant state that is remembered is the state number (ants are limited to 10000 states). For instance, in order to remember which way your ant is pointing you can't just remember it in some variable, you have to make six copies of your program and jump between copies any time you turn right or left.
What this means for the game is that when an ant reaches food, it can never remember how to get back without actually marking some sort of trail (which might lead nowhere). Coordinating complex behavior between multiple ants is difficult, though not impossible.
-jim