martincartledge.github.io

This is the seventh entry of my weekly series Learning Go. Last week I discussed Function Declarations, Arguments, Parameters, and Anonymous Functions. This week I will be talking about Function Literals and Closure.

Function Literals (Function Expressions)

Function Literals can be assigned to a variable or called (invoked) directly. They may refer to the variables defined in a surrounding function, making them a closure (we will talk more about this later in the post)

So, what is the difference between a Function Declaration and a Function Literal?

A Function Declaration binds an identifier (the function name) to a function. You can call this function by using its identifier.

A Function Literal is a closure, meaning they can reference variables that have been defined in a surrounding function. These variables can be shared between the function literal and the surrounding function. These variables persist as long as they are accessible.

Let’s start with a basic example and work our way up in complexity.

package main

import (
	"fmt"
)

func main() {
	f := func() {
		fmt.Println("I am a function literal!")
	}

	f()
	// I am a function literal!
}

inside of func main we declare the variable f and assign to an anonymous function
when this function is invoked, it uses the fmt package to print the string I am a function literal!
we invoke this function literal the same way we invoke function declarations: the identifier followed by arguments wrapped in parentheses ()
this function literal expects no parameters; therefore, we do not pass any arguments
once f is invoked, I am a function literal is printed and the program exits

Let’s see an example when a function literal has a parameter:

package main

import (
	"fmt"
)

func main() {
	f := func(x int) {
		fmt.Println("my birth year is ", x)
	}

	f(1990)
	// my birth year is 1990
}

inside of func main we declare the variable f and assign it to an anonymous function that takes one parameter, x, of type int
using the fmt package, we print the string my birth year is followed by the value of x
notice when f is invoked we pass a single argument 1990
f prints my birth year is 1990 and the programs exits

Next, let’s see how we can return a function from a Function Literal:

package main

import (
	"fmt"
)

func main() {
	f := bar()
	fmt.Println(f())
	// 2020
}

func bar() func() int {
	return func() int {
		return 2020
	}
}

bar:

below func main, using the func keyword, we create a function declaration with an identifier of bar with two return types: func() and int
these return types tell us that bar is expected to return a function and an int inside of that function
inside the function body of bar we return an anonymous function that has a return type of int
inside of this anonymous function, we return the value 2020 of type int

main:

inside of func main we declare the variable f and assign it to return value of the function declaration bar
note: f is assigned to the return value because we are invoking bar; therefore, what bar returns will be the value that f holds in memory. In this case, that return value is a function
f is invoked on the next line inside of the Println function from the fmt package
bar’s return value is a function that returns the value 2020 of type int: therefore, f() will print 2020

As you can see from a few of these examples - function literals can be very powerful and can be used very dynamically in your code. Remember a few things when you are thinking of using a function literal instead of a function declaration:

they are anonymous functions
variables are shared between a function literal and the surrounding function (closure)
variables “survive” as long as they are still accessible

Closure

the way that an anonymous function references variables declared outside of the anonymous function itself

A bit of a brain bender, huh?

The concept of closure can seem very abstract, which makes understanding how they work and the problems they solve difficult as well.

I am confident that seeing closure in action is the best way to learn how they work:

package main

import (
	"fmt"
)

func main() {
	a := incrementor()
	fmt.Println(a())
	// 1
	fmt.Println(a())
	// 2
	b := incrementor()
	fmt.Println(b())
	// 1
}

func incrementor() func() int {
	var x int
	return func() int {
		x++
		return x
	}
}

incrementor:

first we create incrementor, this should look familiar to bar in the last section
incrementor is a function declaration that returns a function and an int inside that function
using the var keyword we declare the variable x of type int
x is not assigned a value; therefore, it is given a zero value (0)
next, we return an anonymous function that is expected to return a value of type int
notice, using the ++ operator, we are incrementing the value of x by 1 - how is this possible? the answer is closure
after we increment x, we return x

main:

inside of func main we create the variable a and assign it to the return value of incrementor()
on the next line, a is invoked inside of the Println function from the fmt package
because the return value of a is the anonymous function inside incrementor(), we increment x by 1 and return the value 1; therefore, 1 is printed
we repeat this process by invoking a inside of the Println function again
since we have already invoked a the value of x is 1; therefore, when we increment x the value returned and printed will be 2

Notice when we assign incrementor() to the variable b it does not return 3, why is that?

Although a and b were assigned the same return value of incrementor, b has only been invoked once; therefore, it holds it’s own unique value of 1.

This is the power of closure, data isolation. Now, you can easily use common actions across multiple variables, and those variables can have their own, unique values.

In Summary

I hope you have enjoyed learning about Function Literals and Closure. With the power of closure, you are equipped with another powerful feature of the Go programming language that can make your code more modular, readable, and scalable. Next, I will discuss Recursion and how to apply those principles to your functions. Can’t wait, see you then!