Wednesday, October 05, 2005

icfp 2005

Haskell rocks. Iam still learning the langugage and am yet to write a full-blown application in this language but I already have an unshakeable opinion that this is "the" language for rapid prototyping. Some years back (2002) when I first heard of this competetion I thought this would get me off the ground with my haskell fundaes. I didnt have a team and made very slow progress. Needless to say I didnt submit a thing. Three years hence and I am still learning the language. Not that the language is too hard or anything- I simply dont use it often. Competing in the icfp contest in haskell is impossible in the present circumstances. When I speak of Haskell here the respones were not encouraging: :Ohh Pascal, we had it in our school but why are you learning Pascal now?

This year I participated in the ICFP 2005 in the C language -- which everybody knows with a modestly (?) sized team of 6. Being the first time for us I never expected a position in the final results. We only were here to experience first-hand 72 hrs of mad-frenzied programming. ContestCrackers we called ourselves (uncool, you say but it shows some of the chutzpah we carrried).

Atul had already done a couple of icfps before, so we had a workhorse amongst us whom we could trust. Myself and Jatin(of the Enduro-3 fame) worked on the Cop programs. Navendu wrote the basic data structures and the skeleton code for everybody to use. Atul worked on the robber and tejinder and adrija on an alternative robber. We adopted 'C' which everybody knew. Atul went about programming the robber rather fast and kept the quality of his program very high. By Sunday afternoon he had a program ready and tested. Jatin and myself kept discussing the strategy for cop for 2 days. On the third day we implemented something but quality issues kept us from submitting a intelligent cop. We only managed to get a no-brainer cop working -- even this failed the initial rounds.

The couple of weeks later towards Friday I worked on the cop again and brought it to a stable enough state. The cops could now catch Atuls robber working in tandem. On Saturday when the twisted specs where announced, I was still tweaking my old cop. Jatin, Tejinder and Navendu and Adrija were kept busy by other issues. On Sunday, I had a look at the twisted specs tweaking it just so that it satisfied the new specs and added no new intelligence to the cop. So also with the robber which Atul got working independently at his office. The results are out and as expected we didnt do well in the first round. But the twist results are encouraging. The results show two submissions, each almost as good as the other. They only differed in that one cop program used a car and the other didnt. I dont know why I thought a cop-on-car would give us some-edge. Well, anyway.

Comments: Post a Comment



<< Home

This page is powered by Blogger. Isn't yours?