Serious Ben
  1. Posts
  2. Curated Links

Some golang gotchas

April 01, 2018 Me

Go is such a simple and elegant language. There is no magic happening and reading go code is so straightforward. But like any language there are some gotchas.

This post is about some general Golang gotchas (and neat tricks I have used or seen people use).

Range and iterators

This gotcha is a pretty important one that I see people do all the time. It is a very tricky one to track down as well.

In one sentence: iterations reuses addresses.

Simple example

package main

import "fmt"

type someStruct struct {
	str string
}

func main() {
	vals := []someStruct{someStruct{"1"}, someStruct{"2"}, someStruct{"3"}, someStruct{"4"}, someStruct{"5"}}

	var myVal *someStruct
	for _, val := range vals {
		if val.str == "1" {
			myVal = &val
		}
	}
	fmt.Println("myVal of value 1 ==", myVal.str)
}

Go Playground

myVal will equal 5 and not 1.

This is because val is always pointing to the same address in the iteration. Every iteration use the same address resulting in val to contain different data.

A way to fix this would be to copy values.

package main

import "fmt"

type someStruct struct {
	str string
}

func main() {
	vals := []someStruct{someStruct{"1"}, someStruct{"2"}, someStruct{"3"}, someStruct{"4"}, someStruct{"5"}}

	var myVal *someStruct
	for _, val := range vals {
		if val.str == "1" {
			tmp := val
			myVal = &tmp
		}
	}
	fmt.Println("myVal of value 1 ==", myVal.str)
}

Go Playground

Here val is copied within tmp resulting in a new address.

Goroutine example

When using goroutines within a loop, you can end up in the same situation as well.

package main

import (
	"fmt"
	"time"
)

func main() {
	for i := 0; i != 10; i++ {
		go func() {
			fmt.Println(i)
		}()
	}
	time.Sleep(5) // Waiting for goroutine to end
}

Go Playground

In this example, we see that simple for i iterations (even without using a range operator) also reuses the same address for i.

This will print:

10
10
10
10
10
10
10
10
10
10

To fix this problem, only rely on the address of an iteration when doing work within the loop. Both using the variable outside the loop or in a goroutine will result in unexpected behavior.

You could do this instead:

package main

import (
	"fmt"
	"time"
)

func main() {
	for i := 0; i != 10; i++ {
		go func(num int) {
			fmt.Println(num)
		}(i)
	}
	time.Sleep(5) // Waiting for goroutine to end
}

Go Playground

Fun with golang nils

Pointers in go are typed. That is they are a pointer to a value of a specific type. This means that nil for a pointer of a specific type is NOT nil for a pointer of another type.

This example shows that pretty clearly:

package main

import (
	"fmt"
)

type customError struct{}

func (ce *customError) Error() string {
	return "customError string"
}

type customStuff struct {
	error *customError
}

func main() {
	var err error = customStuff{}.error
	if err == nil {
		fmt.Println("err IS nil ->", err)
	} else {
		fmt.Println("err IS NOT nil ->", err)
	}
	if err == (*customError)(nil) {
		fmt.Println("err IS (A*customError)(nil) ->", err)
	} else {
		fmt.Println("err IS NOT nil ->", err)
	}

	err2 := customStuff{}.error
	if err2 == nil {
		fmt.Println("err2 IS nil ->", err2)
	} else {
		fmt.Println("err2 IS NOT nil ->", err2)
	}

}

Go Playground

The output of this example will be:

err IS NOT nil -> customError string
err IS (*customError)(nil) -> customError string
err2 IS nil -> customError string

You can see that even though my customError implements the Error interface it is not equal to Error(nil).

This gotchas is the reason why you should always return a nil of type error and never of you custom type.

Fun with golang scopes and := assignment operator

Guess what this little program will print.

package main

import (
	"errors"
	"fmt"
)

func main() {
	err1 := errors.New("err1")
	err2 := errors.New("err2")
	a, err := "a", err1

	for {
		a, err := "b", err2
		_, _ = a, err // Bypassing "not used" error
		break
	}

	fmt.Println("Is err == err2? \n> ", err == err2)
	fmt.Println("Is a == \"b\"? \n> ", a == "b")
}

Go Playground

Now guess what this little program will print.

package main

import (
	"errors"
	"fmt"
)

func main() {
	err1 := errors.New("err1")
	err2 := errors.New("err2")
	a, err := "a", err1

	for {
		a, err = "b", err2
		_, _ = a, err // Bypassing "not used" error
		break
	}

	fmt.Println("Is err == err2? \n> ", err == err2)
	fmt.Println("Is a == \"b\"? \n> ", a == "b")
}

Go Playground

Spoilers

Program 1 will print:

Is err == err2?
>  false
Is a == "b"?
>  false

Program 2 will print:

Is err == err2?
>  true
Is a == "b"?
>  true

The difference is the assignment done at line 14.

I often see this when dealing with errors. Just be aware that might be creating a brand new variable within a scope OR that you might be overwriting the value of a parent scope.

Functions can implement interfaces too

// A Handler responds to an HTTP request.
type Handler interface {
	ServeHTTP(ResponseWriter, *Request)
}
...
// The HandlerFunc type is an adapter to allow the use of
// ordinary functions as HTTP handlers. If f is a function
// with the appropriate signature, HandlerFunc(f) is a
// Handler that calls f.
type HandlerFunc func(ResponseWriter, *Request)

// ServeHTTP calls f(w, r).
func (f HandlerFunc) ServeHTTP(w ResponseWriter, r *Request) {
    f(w, r)
}

This snippet is how functions can be handlers in the core library of go.

Extra reading