We competed in the 24h lightning division, June 28—29, 2024. This was our first time participating in this contest.
Our choice programming language was Python.
We also had a Discord integration that posted our scores to our team server (in .github/).
First we wrote a few different translators for the ICFP language. But the translators are simple, and do not include evaluation of ICFP. However, they were sufficient to complete the hello challenge, and get us enrolled in lambdaman and spaceship courses. During the hello challenge, we also discovered the backdoor command to enroll in 3d course.
Our ICFP translators are in index.py and parser.py.
We also had a utility to ease communication with the server in senq-req.py.
The initial messages we exchanged are in messages/.
Instruction for the various courses are in the repository root.
We tried a few different approaches to solve the lambdaman puzzles.
One attempt is in lambdaman_solver.py and another strategy is in simple_lm.py.
The second is very naive, and finds non-optimal solutions, but it does find solutions for all mazes. It works by putting the pills in order, using a simple distance measure. Then, for the ordered sequence of pills, it finds the shortest path between each pair of pills.
We were unable to solve problems #6, #9 and #21 since we did not complete evaluation for ICFP. We could solve problem #10, because it was possible to construct the maze by reading the clues contained in the server response.
We first solved a few of the simple problems by hand (#1 and #3). We tried to come up with a general solution, but this did not work out.
In the end we had a simple programmatic solution in space.py, with a heavy restriction, that works for some problems.
It puts the coordinates in order so that the distance between coordinates is minimal.
Then it calculates, at each timestep, which phone dial button will move the spaceship to the next coordinate.
The built-in restriction is that a steps between coordinates must take exactly one unit of time, meaning it fails for problems where progress cannot be made in every timestep.
We started work on this problem but were unable to arrive to a solution.
The stater solution is in 3d_prog.py and an example input in 3d_input.txt.
We never discovered how to enroll in this course 🙁.