Mismatched types float64 and int

Question

I'm trying to create random grades and adding them to a test_scores array. Then calculate the average.

This program:

package main

import (
    "fmt"
    "math/rand"
)

func main() {
    i := 0
    var test_scores [5]float64
    for i < len(test_scores) {
        test_scores[i] = rand.Float64()
        i++
    }
    fmt.Println(test_scores)

    var total float64 = 0
    i = 0
    for i < len(test_scores) {
        total += test_scores[i]
        i++
    }
    fmt.Println(total)
    fmt.Println(total / len(test_scores))
}

produces:

main.go:24: invalid operation: total / 5 (mismatched types float64 and int)

This one works fine:

package main

import (
    "fmt"
    "math/rand"
)

func main() {
    i := 0
    var test_scores [5]float64
    for i < len(test_scores) {
        test_scores[i] = rand.Float64()
        i++
    }
    fmt.Println(test_scores)

    var total float64 = 0
    i = 0
    for i < len(test_scores) {
        total += test_scores[i]
        i++
    }
    fmt.Println(total)
    fmt.Println(total / 5)
}

---

The only difference being that in the final line, I'm using a fixed 5 and on the non-working one, I'm using the len(test_scores) call.

Len returns an integer as well, so what's up?

Answer 1

When you write 5 directly in the source-code that's called a constant. Same goes for writing true. The only difference is that the former is an untyped constant and the latter a typed constant.

The difference lies in that there's no ambiguity about what type true should have – it'll always be bool but in the case of 5 that's not so obvious and depends on the context.

The Go compiler will figure out what type to give the constant on compilation. The details of this are described in Go's language specification.

Edit:

I realized that there's a mistake in my answer: true is in fact also untyped according to the spec because it may be utilized anywhere where a type deriving from bool is expected. That means:

type MyBool bool

func DoNothing(b MyBool) {}

DoNothing(true) // true is coerced to MyBool

The answer is still valid, though. The distinction between typed and untyped constants holds.

Answer 2

float64 and int are different types, but conversions are allowed under specific circumstances. (http://golang.org/ref/spec#Conversions)

Your literal 5 in the code is an untyped constant (http://golang.org/ref/spec#Constants), and the proper type is determined by the expression during compilation.

Simply use float64(len(test_scores))

Answer 3

This lines

fmt.Printf("%T\n", total)
fmt.Printf("%T\n", 5)
fmt.Printf("%T\n", 5.0)
fmt.Printf("%T\n", len(test_scores))

prints

float64
int
float64
int

Maybe compiler perceives 5 as 5.0.. Anyway you should use conversion to float64.