Function basics

This chapter covers basic Elm syntax that is important to get familiar with: functions, function signatures, partial application and the pipe operator.

Functions

Elm supports two kind of functions:

  • anonymous functions
  • named functions

Anonymous function

An anonymous function, as its name implies, is a function we create without a name:

\x -> x + 1

\x y -> x + y

Between the backslash and the arrow, you list the arguments of the function, and on the right of the arrow, you say what to do with those arguments.

Named functions

A named function in Elm looks like this:

addOne : Int -> Int
addOne x =
  x + 1
  • The first line in the example is the function signature. This signature is optional in Elm, but recommended because it makes the intention of your function clearer.
  • The rest is the implementation of the function. The implementation must follow the signature defined above.

In this case the signature is saying: Given an integer (Int) as argument return another integer.

You call it like:

addOne 3

In Elm we use whitespace to denote function application (instead of using parenthesis).

Here is another named function:

add : Int -> Int -> Int
add x y =
  x + y

This function takes two arguments (both Int) and returns another Int. You call this function like:

add 2 3

No arguments

A function that takes no arguments is a constant in Elm:

name =
  "Sam"

How functions are applied

As shown above a function that takes two arguments may look like:

divide : Float -> Float -> Float
divide x y =
    x / y

We can think of this signature as a function that takes two floats and returns another float:

divide 5 2 == 2.5

However, this is not quite true, in Elm all functions take exactly one argument and return a result. This result can be another function. Let's explain this using the function above.

-- When we do:

divide 5 2

-- This is evaluated as:

((divide 5) 2)

-- First `divide 5` is evaluated.
-- The argument `5` is applied to `divide`, resulting in an intermediate function.

divide 5 -- -> intermediate function

-- Let's call this intermediate function `divide5`.
-- If we could see the signature and body of this intermediate function, it would look like:

divide5 : Float -> Float
divide5 y =
  5 / y

-- So we have a function that has the `5` already applied.

-- Then the next argument is applied i.e. `2`

divide5 2

-- And this returns the final result

The reason we can avoid writing the parenthesis is because function application associates to the left.

Grouping with parentheses

When you want to call a function with the result of another function call you need to use parentheses for grouping the calls:

add 1 (divide 12 3)

Here the result of divide 12 3 is given to add as the second parameter.

In contrast, in many other languages it would be written:

add(1, divide(12, 3))

Partial application

As explained above every function takes only one argument and returns another function or a result. This means you can call a function like add above with only one argument, e.g. add 2 and get a partially applied function back. This resulting function has a signature Int -> Int.

add 2 returns another function with the value 2 bound as the first parameter. Calling the returned function with a second value returns 2 + the second value:

add2 = add 2
add2 3 -- result 5

Partial application is incredibly useful in Elm for making your code more readable and passing state between functions in your application.

The pipe operator

As shown above you can nest functions like:

add 1 (multiply 2 3)

This is a trivial example, but consider a more complex example:

sum (filter (isOver 100) (map getCost records))

This code is difficult to read, because it resolves inside out. The pipe operator allows us to write such expressions in a more readable way:

3
    |> multiply 2
    |> add 1

This relies heavily on partial application as explained before. In this example the value 3 is passed to a partially applied function multiply 2. Its result is in turn passed to another partially applied function add 1.

Using the pipe operator the complex example above would be written like:

records
    |> map getCost
    |> filter (isOver 100)
    |> sum

results matching ""

    No results matching ""