# Function Composition

Recall that a function signature is comprised of two main parts. Everything to the left of the arrow -> indicates the types of the function's arguments, and the right indicates the function's return type.

Thus far, we've mostly focused on the argument types. Let's turn our attention to the return type.

; number->string : number -> string
> (number->string 524)
"524"

In this example, we've used the number->string function to convert a number, 524, into its string representation, "524" (remember, they are different types, and therefore different data).

Notice what happens here:

> (string-append "ted cruz"
524
"zodiac killer")
string-append: expects a string as 2nd argument, given 524

> (string-append "ted cruz"
(number->string 524)
"zodiac killer")
"ted cruz524zodiac killer"

What just happened? Let's deconstruct this in the stepper.

First, we evaluating the innermost expression. (number->string 524) simplifies to "524":

(string-append "ted cruz"
(number->string 524)
"zodiac killer")

becomes

(string-append "ted cruz"
"524"
"zodiac killer")

Now, we can string-append everything together:

"ted cruz524zodiac killer"

Fair enough. Let's take a look at the function signatures for number->string and string-append:

• number->string : number -> string
• string-append : string ... -> string

Notice how our initial attempt fails because we've violated the string-append function signature by passing in a number.

Our second try succeeds, because we first convert 524 to a string. Then, when string-append is called, all of its arguments are strings.

How do we know this? Because the function signature for number->string guarantees it will return a string.

In this sense, we can think of function signatures as "contracts" for function usage.

Function signatures dictate that

• as long as we provide arguments of the correct type, in the correct order,
• we can count on the function to return a specific type.

This contract, coupled with the rules for program evaluation we've already seen, form the basis for programming in Racket. If you internalize these two things, everything in this class will make sense.