[RxJS] Stream Processing With RxJS vs Array Higher-Order Functions

Higher order Array functions such as filter, map and reduce are great for functional programming, but they can incur performance problems.

var ary = [1,2,3,4,5,6];

var res = ary.filter(function(x, i, arr){
  console.log("filter: " + x);
  console.log("create new array: " + (arr === ary));
  return x%2==0;
})
.map(function(x, i, arr){
  console.log("map: " + x);
  return x+"!";
})
.reduce(function(r, x, i, arr){
  console.log("reduce: " + x);
  return r+x;
});

console.log(res);

/*
"filter: 1"
"create new array: true"
"filter: 2"
"create new array: true"
"filter: 3"
"create new array: true"
"filter: 4"
"create new array: true"
"filter: 5"
"create new array: true"
"filter: 6"
"create new array: true"
"map: 2"
"map: 4"
"map: 6"
"reduce: 4!"
"reduce: 6!"
"2!4!6!"
*/

In the example, filter & map function will return a new array. That's good because it pushes forward the idea of immutability. However, it's bad because that means I'm allocating a new array. I'm iterating over it only once, and then I've got to garbage-collect it later. This could get really expensive if you're dealing with very large source arrays or you're doing this quite often.

Using RxJS:

var source = Rx.Observable.fromArray([1,2,3,4,5,6]);

source.filter(function(x){
  console.log("filter: " + x);
  return x%2==0;
})
.map(function(x){
  console.log("map: " + x);
  return x+"!";
})
.reduce(function(r, x){
  console.log("reduce: " + x);
  return r+x;
}).subscribe(function(res){
  console.log(res);
});

/*
"filter: 1"
"filter: 2"
"map: 2"
"filter: 3"
"filter: 4"
"map: 4"
"reduce: 4!"
"filter: 5"
"filter: 6"
"map: 6"
"reduce: 6!"
"2!4!6!"
*/

The biggest thing is that now you'll see it goes through each -- the filter, the map, and the reduce -- at each step.

Differences:

The first example: it creates two intermediary arrays (during filter and map). Those arrays needed to be iterated over each time, and now they'll also have to be garbage-collected.

The RxJS example:  it takes every item all the way through to the end without creating any intermediary arrays.