Should Pure Functions Be Self-contained

satansdeer profile image Maksim Ivanov ・1 min read

I my last video about pure functions I said that to be pure a function should be self-contained i.e it should only access values from it's own scope.

function pureFunction(){
  return 'Some string literal';
  // Doesn't perform side effects, is deterministic
  // thus we can be sure that it's pure 

function thisWontBePure(){
  return pureFunction();
  // Isn't pure because calls function from global scope.
  // Even though the function we are calling here is pure.

So if you want a function to be pure it should only access values from it's arguments:

function nowItsPure(anotherPureFunction){
  return anotherPureFunction();
  // Isn't pure because calls function from global scope.

What do you think about this? Can pure function call other pure functions from global scope and still remain pure?

Posted on by:

satansdeer profile

Maksim Ivanov


I write tutorials for developers at https://maksimivanov.com


markdown guide

In the first snippet i would say that thisWontBePure is pure. As far as i know pureFunction is a constant. What could possibly change the value returned by pureFunction at runtime?

Consider this:

function add(a) {
  return function inner(b) {
    return a + b;

I would say that inner is still a pure function because the value of a can't change once is locked in the closure.

If a function doesn't have side effects and mantains referential transparency, to me that is a pure function.


mmm you got me thinking. What if :

var c = 0;
var obj1 = {
        valueOf: function () {
            return 1;
var obj2 = {
        valueOf: function () {
            return 2;
function add(a) {
  return function inner(b) {
    return a + b;

Fair question. This is javascript, right? In a dynamic language you can only write code in a functional style and hope that no one does a trick like the one you did there.

In your example inner becomes impure because valueOf is impure (it changes some state outside of its scope). So if an impure function gets executed inside of a "pure" function than it is no longer pure.


I messed up the code format. Used from stack overflow... Sorry for asking this here, how can I format code in a comment?

The syntax highlight can be activated with a block closed in triple backticks, and optionaly a language name. Like this:

// some code

Thank you sir. Sorry for the irrelevant question again.


As you know a pure function has the following two requirements:

  1. It must not have any side effects, i.e. it is not allowed to change any state of the program or of external ressources.
  2. It must be deterministic, i.e. for a given set of inputs it must almost return the same output. This is also called memoizable.

That does not imply that you cannot call other functions but if the called functions change behaviour you are breaking rule #2.

Functional programming languages have immutable values (at least by default) so you cannot "override" functions or change variables. By definition an immutable variable is not really a variable but just a value.

If you are trying to do functional programming in JavaScript or C# which have references, well, then you have to enforce those rules by coding style and self-discipline. The runtime will not help you.


Well If the functions called are pure as well, then yes, I think it doesnt defeat the purpose.


But what if that function in global scope will be overridden by non-pure function?

Global scope can be altered from other parts of the code.


Yes then this function became not pure. So just like a var can change from a number to function, I believe a function can be rendered pure or not while it changes.

Then it looks like it's not the function that can be pure or not, but a function call.

So we can have pure function calls and impure function calls.

Would be awesome to summon some FP pro here to clarify the terminology.