I’ve been learning Elixir for about 10 months now in my spare time and I would welcome some constructive feedback on my Maze Generator library.
Thanks!
Kurt Symanzik
I’ve been learning Elixir for about 10 months now in my spare time and I would welcome some constructive feedback on my Maze Generator library.
Thanks!
Kurt Symanzik
Hello and welcome,
Disclaimer… I did not read the book Mazes for programmers.
I do not like the use of Agent to store visited cells, just to manage a map. It is started/stopped in the same function.
It’s possible to do without processes. I translated your code to Rust, for fun (and learning).
pub struct RecursiveBacktrack {
visited: HashMap<Coordinate, bool>
}
impl RecursiveBacktrack {
pub fn new() -> Self {
Self {
visited: HashMap::new(),
}
}
pub fn carve(&mut self, grid: &mut Grid) {
let mut rng = thread_rng();
let x: usize = rng.gen_range(0, grid.width);
let y: usize = rng.gen_range(0, grid.height);
let starting_coordinate = Coordinate(x, y);
self.do_carve(grid, starting_coordinate, starting_coordinate)
}
fn do_carve(&mut self, grid: &mut Grid, current: Coordinate, last: Coordinate) {
match self.visited.get(¤t) {
Some(_) => (),
None => {
self.visited.insert(current, true);
grid.open_passage(current, last);
let mut neighbors = grid.neighbors(¤t);
neighbors.shuffle(&mut thread_rng());
for neighbor in neighbors {
self.do_carve(grid, neighbor, current);
}
},
}
}
}
and the result
+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+
| | | | | | | | | | | | | | | | | | | | |
+---+---+ + + +---+ + + + + +---+ + + + +---+ + + +
| | | | | | | | | | |
+---+ +---+---+ +---+ +---+ +---+ + +---+---+ + + +---+ + +
| | | | | | | | | | | |
+---+---+ +---+ + + +---+ +---+ +---+ +---+ +---+ + +---+ +
| | | | | | | | | | | |
+---+---+---+ + + +---+ +---+ + +---+ + + +---+ + + + +
| | | | | | | | | | | | | | | |
+---+---+---+ + + +---+ +---+ + + +---+ + +---+ +---+---+---+
| | | | | | | |
+---+ +---+---+ +---+ + +---+---+ +---+---+ +---+ +---+---+ +---+
| | | | | | | | | |
+---+---+---+ +---+---+ + + +---+---+ + + +---+ +---+---+---+ +
| | | | | | | | | |
+---+ + +---+---+ +---+---+---+ +---+ + +---+ + + +---+---+---+
| | | | | | | |
+---+---+---+---+---+ + +---+---+---+ +---+---+ +---+ + + + +---+
| | | | | | | | | | |
+---+---+ + +---+---+ +---+---+ + +---+ +---+ +---+---+---+ + +
| | | | | | | | | |
+---+ +---+ + + + +---+ + +---+ +---+ +---+---+---+ + + +
| | | | | | | | | | | | | |
+---+---+ + + +---+ +---+ +---+ +---+---+ + + +---+---+ + +
| | | | | | | | | | | |
+---+ + + +---+ +---+ +---+---+ +---+ + + + + + + +---+
| | | | | | | | | | | | | | | |
+---+ + +---+ +---+---+ + +---+---+ + + + + + +---+---+ +
| | | | | | | | | |
+---+---+ + +---+---+---+ +---+ + +---+---+ +---+ + + +---+ +
| | | | | | | | | | |
+---+---+ +---+ + +---+ + + +---+ + +---+---+ +---+ +---+---+
| | | | | | | | | |
+---+ +---+ +---+ +---+---+---+ +---+---+ + +---+ +---+---+ + +
| | | | | | | | | |
+---+---+ +---+ +---+ + +---+---+---+---+---+ + +---+---+---+ +---+
| | | | | |
+---+ +---+ + +---+ +---+ +---+ +---+ +---+---+---+---+---+---+---+
| | | | | | | | | | | | | |
+---+ + + + + + + + + + + + + +---+ + +---+ + +
I don’t know if Agent are used to solve the maze, but You can generate without
Thank you for the suggestion - I will do that. I’m curious to find out how that affects performance.
Thanks
I did not benchmark, but I am sure calling a pure function will be faster than starting and passing messages to an Agent
I refactored to use a map instead of an agent and it confirmed your suggestion. The result was over 3 times faster. Thanks!
Whenever You want speed, real speed, You can use an ETS table.
It is blazing fast.