Trouble with updating a 2d array with ReasonML (compiling to JS)

#1

Hello everyone!

Sorry to bother you but I am having trouble updating a 2d array, and I think I have not screwed up anything obvious.

Here is the sketch (copied from @thangngoc89): https://sketch.sh/s/HqfcmU5IBWiSNWjClEZrVB/

I don’t know what’s going on, but I am quite sure it is a mistake on my part, just not sure what. If someone could help I would really appreciate it.

Thank you.

#2

Hey! The issue is that you’re initializing an array containing 3x a reference to the same array.
So effectively matrix[r][0] === matrix[r][1] === matrix[r][2] (try running it inside the sketch.sh file).

You can solve this using Array.make_matrix :slight_smile:

#3

Thank you @bsansouci!

Also, it doesn’t seem like make_matrix supports only 2d arrays right?

#4

I think Array.make_matrix only returns a 2D array. You can’t make it return an array(array(array('a))) for example.

#5

Wait, so how do you create 3d/higher multidimensional arrays in ReasonML/BuckleScript/OCaml?

#6

You can do it with a recursive function and a loop:

let make_3d_matrix = (dimx, dimy, dimz) => {
  let matrix = ref(Array.make(dimx, [||]));
  for (i in 0 to dimx - 1) {
    matrix.contents[i] = Array.make_matrix(dimy, dimz, 0);
  };
  matrix.contents;
};

And for higher dimensions you can nest them. I’m not sure you can make a function that returns a matrix of arbitrary nested arrays because the return type of the function cannot be different from one call to another, so any branch of the function returns array(int) and another branch returns array(array(int)) then the compiler will say “idk what to do here bro”. So you’d need to “unify” the different things that can be returned using a variant, in which case your simple matrix becomes a type t = MatrixOfLengthN | BaseArray(array(int)) kinda thing , and then you’re returning something of type t and of value MatrixOfLengthN(MatrixOfLengthN(BaseArray([||]))) for a 3d matrix etc… Which is probably not what you’d like. I’d recommend finding ways to avoid needing arbitrary nested arrays.