Create a new folder for your problem, as in problems/yourProblemName
Inside of the yourProblemName folder, you should create two new folders called solutions and tests.
The solutions folder should contain the source code for any solutions to the problem.
Typically we create a folder for each language (so you might have a solutions/python3/ folder or a solutions/typescript/ folder).
Also inside the solutions folder are one or more .run files. This is simply a bash script that executes one of the solutions. For example, if you had a Python 3 solution inside of solutions/python3/main.py, you could add a file named solutions/python3.run with the following contents:
#!/bin/bash
python3 $(dirname $0)/python3/main.py
Note: You must mark the
.runfile as executable (as inchmod +x python3.run) or else you will get a permission denied problem when trying to run the tests.
Your input and expected outputs should go inside of the tests folder. The files inside this folder must follow a specific naming pattern. For inputs, you should name them with a number as the base name with .in as the extension, as in 0.in, 1.in, 2.in, etc. Similarly, for the corresponding expected output file, you should use numbers as the base name with .out as the extension (as in 0.out, 1.out, 2.out, etc).
Each problem requires a clear description. Typically the descriptions follow the following rough format:
- Introducing the Problem: Explain the backstory of the problem and give the reader an idea of what problem needs to be solved.
- Input / Output: You should explain clearly the format of the input and the expected output format.
- Make certain that your output format is completely unambiguous. For every input, there should be exactly one correct output. Common tricks include adding up numbers to a single sum or ordering the output in ascending order. This doesn't always work, so just make sure there is only one correct answer.
- If you have multiple lines of input, it can simplify things for some students if the first line of input tells them how many lines of input to expect, especially if they are using terrible languages like C++ (we've learned this from years past)
- Constraints: Explain how big the input size can be, how big the numbers can get, and any critical disambiguating rules.
To run the tests for a problem, run this fancy make command from the root of the repo.
make test-yourProblemName
This will run your .run program against each pair of inputs and outputs. You may have to install the colordiff program to have that work.
On Ubuntu, something like this:
sudo apt-get install colordiff
On Fedora/RHEL, something like this:
sudo dnf install colordiff
On Windows or Mac, I have no foggy clue how to install colordiff. If anyone cares, feel free to add a page for that.