Generic Functions in Go

I'm in the process of learning Go and the documentation and interactive lessons say that an empty interface can hold any type, as it requires no additionally implemented methods.

So for an example:

func describe(i interface{}) {
    fmt.Printf("Type: %T | Value: %v\n", i, i)
}

...would print out...

"Type: int | Value: 5" // for i := 5
"Type: string | Value: test" // for i := "test"
... etc

So I guess my question is if this is Go's way of implementing generic functions or if there is another, more suitable, way of doing them.

As of Go 1.18 you can write a generic function Print as below:

package main

import (
	"fmt"
)

// T can be any type
func Print[T any](s []T) {
	for _, v := range s {
		fmt.Print(v)
	}
}

func main() {
	// Passing list of string works
	Print([]string{"Hello, ", "world\n"})

	// You can pass a list of int to the same function as well
	Print([]int{1, 2})
}

Output:

Hello, world
12

The Go paradigm is generally to avoid this by implementing the behavior in non-empty interfaces. For example, say you wanted to print something with type-specific formatting:

func Print(i interface{}) {
    switch o := i.(type) {
        case int64:
            fmt.Printf("%5d\n", o)
        case float64:
            fmt.Printf("%7.3f\n", o)
        case string:
            fmt.Printf("%s\n", o)
        default: // covers structs and such
            fmt.Printf("%+v\n", o)
    }
}

Alternatively, you could define an interface for things that know how to string themselves (this exists in the base library as an fmt.Stringer), and use that:

type Stringer interface {
    String() string
}

func Print(o Stringer) {
    fmt.Println(o.String())
}

type Foo struct {
    a, b int
}

func (f Foo) String() string {
    // Let's use a custom output format that differs from %+v
    return fmt.Sprintf("%d(%d)", f.a, f.b) 
}

type Bar struct {
    t bool
}

func (b Bar) String() string {
    if b.t {
        return "TRUE! =D"
    }
    return "false =("
}

https://play.golang.org/p/Ez6Hez6cAv

This lets you have a generic-like functionality, but still retain type safety and have the behavior itself defined by the types, rather than your generic function.

Go encourages you to think of types in this way, based on their behavior, what they can do rather than what they contain.

Golang doesn't have a generic type, so the way you get around this is passing interface type and using type switches within the function.