nil-errors-that-are-non-nil-errors.md

title	description
Nil errors that are non-nil errors	Difference between nil interface value and interface holding a nil value

Nil errors that are non-nil errors

Consider the following code:

package main

import "fmt"

type E struct {}

func (*E) Error() string { return "error" }

func f() error {
    var err *E = nil
    return err
}

func main() {
    if err := f(); err != nil {
        panic(err)
    }
    fmt.Println("success")
}

You can easily check that although f() error always returns nil, f() != nil so this code panics.

Why?

An interface in Go is a tuple of (type, value). In the following code:

var err error

err has nil value since we didn't assign anything to it, but in:

var (
    specific *E
    err error = specific
)

We're setting the err's value to a (*E, nil) tuple, there's a type, but it doesn't point to any value.

You can check that out in the following example:

package main

import "fmt"

type Printer interface{ Print() }

type StringPrinter string

func (s *StringPrinter) Print() {
	if s == nil {
		fmt.Println("nil value")
		return
	}
	fmt.Println(*s)
}

func main() {
	var (
		sp      *StringPrinter
		printer Printer = sp
	)

	printer.Print()
}

Since printer already has a type, we can already call pointer receiver functions on it, although it's value is nil. Obviously you can't do that if you don't specify the type, because there's nowhere to call.