FP vs. OOP: input

I often think about how Functional Programming (FP) compares to Object Oriented Programming (OOP).

One aspect is how do we provide input to functions. FP promotes small functions that given the same input return the same output. It seems that being forced to pass input as argument results in smaller functions because more params usually means more work, which means larger functions. Larger functions are harder for client code to use because you have to remember what are all those arguments.

One other side-effect is that if a function get its input through arguments, it’s also easier to reuse in other contexts.

One example would be the pluck function:

function pluck(propertyName) {
  return function(object) {
    return object[propertyName];
  };
}

I can use this to get the amount for items in a cart:

cart.items.map(pluck('amount'));
// [1, 2, 1, 3, 1]

or, to get the list of comment IDs for an article:

article.comments.map(pluck('id'));
// [523, 524, 580, 587]

Small size seems to result in a more generic function, which does less things — ideally one, and which is more reusable.

In OOP, functions are bound with the data that they operate on. So if I’d have an object for every context, I would have a separate method for each of them. For example, with a Cart class I would have something like this:

Cart.prototype.itemAmounts = function() {
  return this.items.map(function(item) {
    return item.amount;
  });
};

and then, for with the Article class I would have something like this:

Article.prototype.commentIds = function() {
  return this.comments.map(function(comment) {
    return comment.id;
  });
};

Both of these methods: Cart.prototype.itemAmounts and Article.prototype.commentIds do a very similar thing: iterate over a collection of items, and for every item collect a given property. This is (incidental?) duplication.

With OOP methods do not explicitly receive the input, it’s enclosed in the instance context. This requires more effort to understand the code: you have to look into it body to see what it does.

On the other hand, having article.commentIds() is more concise than article.comments.map(pluck('id')). It also results in better encapsulation: it doesn’t expose how comment IDs are collected, which leads to less coupling.

I have used both approaches and they both work. The choice seems to come down to which is more familiar to the programmer.

Tags