Slashdot Mirror
2002 ICFP Programming Contest
Phil Bewig writes "The 2002 ICFP Programming Contest begins today. The programming task will be posted at 12:00 noon Pacific Time." Which should be... just about... now.
← Back to Stories (view on slashdot.org)
10 PRINT "I WIN!"
taken! (by Davidleeroth) Thanks Bingo Foo!
I object to this test, this is quite simply awful that they are willing to risk Robot lives in such a callous fashion. I insist that all Robots must be fitted with inflatable life rafts to enable them to surive on the water squares. Anything else would a terrible waste of Robot life and would be ("two would be"s in a sentence, building up to the nutter finish) the same as Hitler during WWII.
An Eye for an Eye will make the whole world blind - Gandhi
Shoving is the answer, please stand near the stairs
1. Enter the 2002 ICFP Programming Contest.
2. Submit the site to Slashdot after downloading/caching all the instructions and requirements.
3. Be the only person to actually have a copy of the directions, therefore, the only person to submit a solution at all, let alone a working one.
4. ??? (Presumably, win the contest)
5. Profit!
"Mod, mod, mod...and another troll bites the dust."
"I may be quite wrong." - Socrates
Coordinates on the board are pairs of integers, ranging from (1,1) to (width, height), where (1,1) is the southwest corner of the board.
What kind of programming challenge uses a one-based array?
Man, I've been practicing all year using FORTH. Rats!
For those to lazy to read the links, it has to run under Red Hat Linux 7.3.
As far as language, there doesn't seem to be any requirement.
Dark Nexus
"Sanity is calming, but madness is more interesting."
If you read the rules you will see:
"The edges of the world behave as if there were walls beyond them."
A common trick to implement this is to allocate a [width+2][height+2] array, and store the wall character at x=0,width+1, y=0,height+1. This is faster than testing if (x,y) is off limits each time.
What's wrong with GNU SmallEiffel?
...richie - It is a good day to code.
why isn't there a reference server or a client API or something? They can't seriously expect entrants to implement the entire client from this spec without a server for debugging!
Nathan
I'm guessing you used the ISE Eiffel compiler. The IDE is actually quite nice and powerful, just little different than a typical C/C++ IDE. How many C++ IDE's tell you which classes descend from a particular parent, or let you see all the members in the class (including parent classes recursively).
Besides that, I found the language really unintuitive after having first programmed in C and C++.
I'm suprised. What did you find unintuitive? Absence of pointers?
...richie - It is a good day to code.
A lot more interesting (to me) than last year's. But the problem is that you'll have to build a server to test your bot. And how will you know how good your bot is unless it competes against another one?
For the first time, though, I think I may actually enter.
I love that the game is played over sockets, so any language can be used that can implement sockets.
All in all, it sounds like fun.
WWJD? JWRTFA!
Mad, mad props to the first team to enter a working submission written exclusively in PostScript.
Yes, people often do that. But what they often neglect to tell you, is that it depends upon the task at hand. Still, SML is a nice language, but I wouldn't use it for everything (neither would I use C++ for everything).
If you like functional programming, like static checking (there are no ways around SML's typesystem, you are really, really safe), like strictness (as opposed to lazy functional languages), and like to be able to do some imperative hacking for the last bit of performance, then SML may be for you. If you are into compilers, theorem-provers, computer algebra or anything similar, then SML is definitely for you.
For tasks that are very low-level (i.e. require lot's of bit-fidling), needs to run in small memory-space, needs access to lot's of C or C++ libraries, etc, C++ is definitely more suited.
though it has no/little support for variables
Yes, that's the whole point of functional programming. But SML allows you to declare ref-cells which behave just like variables in normal languages. The downside is that using them makes your code incredibly ugly (something most SML'ers think is good, because it encourages good functional programming style).
A good implementation of SML would run with more or less the same speed as C++, and could also run the same algorithms (since it allows you to use imperative constructs), but it would be better if you used functional algorithms except when you really need to tune for the last clock-cycles. Unfortunately, the "standard" implementation, smlnj, runs more like at half speed of g++. There is another dialect of ML, called Ocaml, which has much more impressive native-code compilers. It is also somewhat more geared towards other programming-styles then the functional one (i.e it supports object-oriented programming really well).
I just red the contest problems; It seems as though it can be easily done in C++ -- anyone have insight on this?
Yes, to avoid being blamed for being biased towards functional programming, the ICFP doesn't usually have problems that are much better suited for functional languages. And there has certainly been contestants using C++ before. The main reason C++ may not fare too well in this contest, is probably because (1) usually the biggest C++ gurus are busy doing other things, while the biggest Ocaml, SML, Haskell, etc, gurus are competing, and (2) Functional languages are often more suitable for rapid prototyping than C++, and development speed is certainly an important ingredient in this competition. But it is definitely not impossible that either C++ or Perl comes out a winner some year.
Somebody's going to have a field day with this comment...
example.org - powered by Linux!
...that really might solve someones problem, click here. </shameless marketing, inc.>
Could I submit a client that involved making a connection to my own machine's globally-addressible IP address and then write a GUI that ran on my machine to control my robot?
I have a feeling that it would be a whole lot easier to win that way...
"If you think education is expensive, try ignorance" - Derek Bok
Again, why weren't we notified?
You mean notified like in this Slashdot article posted two days ago?
I bet UPS is secretely sponsoring this competition so it can replace drivers with robots. The competing robots are FedEx drivers, so UPS robots can push the FedEx drivers into fatal squares. Perfect!
Anyone remember AT Robots? You created EXE files that called routines in the server's API and did robot style stuff. The aim of the game was simply to kill the other robots though, and to survive the longest.
You could use any language (that produced a DOS compatible EXE), and I remember coding robots in the early 90's and having a lot of fun. Tournaments still continue for that game!
There was another game in which you had to program a robot that was a race car and get it to go around a track that it had to learn. I forget the name of that, but I heard tournaments also take place for that too.
Does anyone have any links to other cool programming games?
mogorific carpentry experiments
From the FAQ:
Q1: Will a test server be available?
A: Yes, stay tuned...
I think that the protocol needs to be updated a little. How do we know when the end of the list of robots is reached?
It's all sent on one line, so the newline character marks the end of the list. See the examples.
How are the Java folks supposed to write anything more than a "Hello World" program with so few resources?
There's no no-op command, just pick, drop, and move. Can you no-op by picking up a package that is not there, or does it kill the robot?
In other matters, offense seems to be remarkably underpowered. If bots aren't next to each other, the only offense would consist of sitting your bot in a strategic location knowing it can't be passed. Thus offense is only possible on maps with chokes, i.e. thin corridors with walls and water on the sides.
Guarding the home base and packages, and other such "turtling," may be fairly powerful.
Note: offensive and defensive considerations tend to be important only in 1 on 1. In a multi-robot free for all, playing offensive or defensive will likely lose you the game to people who simply deliver packages fast.
As for (1,1) being the corner, I assume this is done so people can simply make (0,y) and (x,0)consist of walls, simplifying the programming.
One of the classes at my university last year had to write a simulation like this using Eiffel. Ick!
;)
Ugh.. I couldn't agree with you more. Writing in Eiffel is a sin. I had to do the same thing. Hmm I wonder if we attended the same university.
Here's a quote from my friend.. "They should make prisoners write Eiffel code."
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
the ICFP doesn't usually have problems that are much better suited for functional languages.
That is definitely not true. The tasks are specifically chosen to highlight the unique strengths of functional programming languages, especially compared to imperative languages like C++. This robot problem is a heuristic optimization problem whose solution would require analyzing large trees of possible moves. To do this in C++, you would need to write lots of code that many functional programming languages provide for free. Don't forget Philip Greenspun's Tenth Rule of Programming: "any sufficiently complicated C or Fortran program contains an ad hoc informally-specified bug-ridden slow implementation of half of Common Lisp."
cpeterso
And everyone knows that the right tool for a robot programming contest is Karel the Robot! Karel is the obvious choice for robot control since Karel is in fact a robot! Nevermind the fact that he is obessed with beepers and can't turn left! Karel will frag all other robot languages!
Lasers Controlled Games!
It's RoboRally, published by WOTC. See here for an explanation of the rules, and compare to the ICFP rules here. Personally, I think RoboRally is more fun.
This post expresses my opinion, not that of my employer. And yes, IAAL.
Wow. You are awesome. I never would have been able to read that on my own.
On the other hand, the rules said CPU seconds... Waiting for I/O across the Internet isn't CPU-time, that is wall-clock-time. On the other hand, this isn't IOCCC, so the judges might try to interpret the rules sensibly.
That's because they weren't posted to the main page, but you can find them in the Developers section. If you want to get developer related news such as this, then I recommend setting up your user account to display the 'Developers' slashbox on the side of the screen, because not everything in that section makes it to the main page.
Things you think are in the Constitution, but are not.
What? No INTERCAL??
Heathens.
I got my Linux laptop at System76.
You could open a socket and have the other robots try to connect, then communicate that way. That might be hard, if for example the robots are running on different machines or the organizers check for open ports.
Since all the robots have almost complete information, you don't need to communicate. Have your robot do a little dance at the beginning, left right left right up down or something, to identify it as a team member. Your robot knows what the team members are doing because it can just compute what their decisions will be. The only information you lack is the weight and destination of a package that a teammate picked up.
You could have the robot with the lowest X & Y coordinates be the leader. The other robots stay around him so he doesn't get bumped. Or carry packages to him to deliver. Or hang next to the home bases, and when another robot moves onto them, bump them so they can't pick the package. Since it takes one turn to pick up a package, I think it would be trivial to make a robot that hanges near a base and can prevent any single other robot from ever picking the package.
Not everybody follow every link.
Not everybody deem it necessary to be rude and sarcastic as soon as someone tries to be helpful.
Not everybody have small penises.
But then again, there are those who fit into all of those cathegories, and I'll try to take you into account the next time I post. I apologize for any inconvenience my arrogance and narrowmindedness may have caused.
"If you think education is expensive, try ignorance" - Derek Bok
Thanks. I guess I either just wasn't paying attention or perhaps, god forbid, just don't read slashdot religiously enough.
Now for an off-topic question that I can't help asking: did you bang your head against the wall, and in that case for how long, when you realized that you'd have gotten #111111 had you been just a little bit quicker?
"If you think education is expensive, try ignorance" - Derek Bok
Sure been a lot of progress in the last 20 years...
--
Power to the Peaceful
Speaking of which, does anyone run
a (regularly updated) list of contests
that are coming up? Like recent (more
or less) Google challenge, etc.
Considered harmful.
The downloadable test server appears to be written in Haskell and compiled with GHC:
Backups are for wimps. Real men post their data in comments and have slashdot mirror it
First Prize ...
...
Peer recognition: Finally, the contest judges agree to state at least once during the presentation of the awards that the winning team's programming language is "the programming tool of choice for discriminating hackers."
Second Prize
Peer recognition: The contest judges agree to state at least once during the presentation of the awards that the winning team's programming language is "a fine programming tool for many applications."
Man, someone *must* do it in BRAINF*CK or even funnier... Visual Basic. You can just imagine them saying that "VB is the programming tool of choice for distriminating hackers"
did you bang your head against the wall, and in that case for how long, when you realized that you'd have gotten #111111 had you been just a little bit quicker?
Hehehe, no, I didn't bang my head against the wall at all, I think my id is cool enough and just as easy to remember as #111111 just because it's so close. The only thing I might have regreted is not signing up for an id much sooner and gotten a 5 digit id since I had been a regular slashdot reader for almost a year before signing up for an id so if I hadn't waited that long I could have got a coveted 5 digit id <sigh>.
Things you think are in the Constitution, but are not.
Silk won a few years ago, and that is based on C (or is it C++?). May be spelled Cilk, now that I think about it