Group does not implement Data (FooMethod method has pointer receiver) - go

This is my code:
package main
import "fmt"
type Group struct {
}
func (g *Group) FooMethod() string {
return "foo"
}
type Data interface {
FooMethod() string
}
func NewJsonResponse(d Data) Data {
return d
}
func main() {
var g Group
json := NewJsonResponse(g)
fmt.Println("vim-go")
}
but does not work as I expect.
$ go build main.go
# command-line-arguments
./main.go:22: cannot use g (type Group) as type Data in argument to NewJsonResponse:
Group does not implement Data (FooMethod method has pointer receiver)

If you want to use a struct receiver, remove the * from before Group in the definition of your function on line 8. As a convenience they do work the other way round (defined on struct works on a pointer receiver). See effective go for an explanation.
https://golang.org/doc/effective_go.html#pointers_vs_values
Modified version:
https://play.golang.org/p/ww6IYVPtIE

Related

How can I get the real pointer to a field inside a struct in Go?

I have a struct, say:
type sample struct{
data []float64
}
Now, I declare a method:
func (s sample) Get() *[]float64{
return &s.data
}
I am trying to append to this slice via the pointer I got via Get()
func main(){
example := sample{[]float64{1,2,3}}
//Here I will append:
pointerToArray := example.Get()
*pointerToArray = append(*pointerToArray, 4)
fmt.Println(example.Get()) // still shows only &{1,2,3}
}
I have a general idea of why this is happening: The function Get is returning the address of its local scope, and I have fixed it by change the struct itself to
type sample struct{
data *[]float64
}
for which the code returns the expected &{1,2,3,4}
Now, for my question:
Is there anyway to get the real pointer to the field in the struct through a getter method without using a pointer field directly in the struct?
The problem is that you've defined the Get method with a struct receiver, rather than on a pointer. That means when you return &s.data you get the address of the field of receiver, rather than from the original struct. You can fix this simply by using a pointer receiver:
func (s *sample) Get() *[]float64{
return &s.data
}
Here's a complete runnable example (https://play.golang.org/p/AIj8QOYfx85)
package main
import "fmt"
type sample struct {
data []float64
}
func (s *sample) Get() *[]float64 {
return &s.data
}
func main() {
example := sample{[]float64{1, 2, 3}}
p := example.Get()
*p = append(*p, 4)
fmt.Println(example.Get())
}

Effectively wrapping public sdk types in golang

I am using pagerduty go sdk to do a bunch of api requests.
Particularly I am making use of
func NewClient(authToken string) *Client
to create a new Client type. I want to add some utility functions to my own work to *Client. I tried doing this:
type BetterPdClient *pagerduty.Client
func NewClient(auth string) BetterPdClient {
return pagerduty.NewClient(auth)
}
func (b *BetterPdClient) DoSomething() {
b.GetIncident(....)
}
func main() {
pd_client := NewClient("")
fmt.Println(pd_client)
pd_client.DoSomething()
}
But I get the following error:
invalid receiver type *BetterPdClient (BetterPdClient is a pointer type)
I understand that DoSomething() is expecting a pointer as caller. Only other way I could think of is sending the ptr as a function argument:
func NewClient(auth string) *pagerduty.Client {
return pagerduty.NewClient(auth)
}
func DoSomething(cl *pagerduty.Client) {
fmt.Println(cl)
}
func main() {
pd_client := NewClient("")
fmt.Println(pd_client)
DoSomething(pd_client)
}
Is there a better way?
Declaring a type as a pointer to another type is almost never what you want because Go doesn't allow you to add methods to that new type, nor to the pointer of that type as you've already figured out yourself. This one doesn't compile either:
type T struct{}
type P *T
func (P) M() {}
If you want to "extend" a type without "hiding" it's existing functionality your best bet is to embed it in a struct.
type T struct{
// ...
}
func (T) M() {}
type U struct {
*T
}
func NewU() *U {
return &U{&T{}}
}
func (U) N() {}
func main() {
u := NewU()
u.M()
u.N()
}
And what I mean by "hiding existing functionality" is that when you define a new type in terms of another, already existing type, your new type will not get direct access to the methods of the existing type. All you're doing is just saying that your new type should have the same structure as the already existing type. Although it's worth pointing out that this property gives you the ability to convert one type to the other...
type T struct{
// ...
}
func (T) M() {}
type U T
func NewU() *U {
return &U{}
}
func (U) N() {}
func main() {
u := NewU()
u.M() // compile error
u.N()
// convert *U to *T and then call M
(*T)(u).M()
}

Why does method signature have to perfectly match interface method

I began to learn Go and have trouble understanding the following:
package main
import "fmt"
type I interface {
foo(interface {})
}
type S struct {}
func (s S) foo(i int) {
fmt.Println(i)
}
func main() {
var i I = S{}
i.foo(2)
}
This fails with:
cannot use S literal (type S) as type I in assignment:
S does not implement I (wrong type for foo method)
have foo(int)
want foo(interface {})
I don't understand why Go doesn't accept the foo(int) signature given the fact that int implements interface {}. Can anyone help with an explanation?
I think your understanding of interface isn't sound. Interface{} itself is a type. It consists of two things : underlying type and underlying value.
Golang doesn't have overloading. Golang type system matches by name and requires consistency in the types
So, when you are defining a function taking a interface type as a parameter:
foo(interface {})
This is a different function from a function taking int type:
foo(int)
So you should change the following line to
func (s S) foo(i interface{}) {
fmt.Println(i)
}
Or better yet to this:
type I interface {
foo()
}
type S struct {
I int
}
func (s S) foo() {
fmt.Println(s.I)
}
func main() {
var i I = S{2}
i.foo()
}
The error message itself is self-explaining:
"S does not implement I (wrong type for foo method)"
so S{} which is of S type cannot be used on the RHS of type I variable assignment.
To implement I interface, type S needs to define foo function with the exact same signature.
To achieve what you wanted, you can use empty interface as input parameter for your foo function in S and do type assertion
package main
import "fmt"
type I interface {
foo(interface {})
}
type S struct {}
func (s S) foo(i interface{}) {
if i, ok := i.(int); ok{
fmt.Println(i)
}
}
func main() {
var i I = S{}
i.foo(2)
i.foo("2")
}
Try it in go playground

Method vs Functions Usage golang

I know differences between func and method. But I am confusing for usages between:
prod:=Product{"title","brand","model"}
prod.Add()
or:
prod:=Product{"title","brand","model"}
products.Add(&prod) // products is package
These are two distinct cases, one which is a method belongs to Product instance and one is a global function belongs to products package.
type Product struct {
Title string
Brand string
Model string
}
// This method add value to a field in Product
func (p *Product) Add(field, value string) {
switch field {
case "Title":
p.Title = value
case "Brand":
p.Brand = value
case "Model":
p.Model = value
}
}
The above provide a method to add value to itself as an instance of Product, i.e.
product1 := &Product{}
product1.Add("Title", "first_title")
The second case is a public function exposed from a product package. In this case, an instance (or a pointer) of a Product must be supplied as an argument.
package products
func Add(p *Product, field, value string) {
// Same switch-case as above
}
Add function then can be used from any other package.
package main
import (
"path/to/products"
)
type Product struct {
// ...
}
func main() {
product1 := &Product{}
products.Add(product1, "Title", "first_title")
Normally in your scenario, the first approach is preferred since it encapsulates the functionality of managing its attributes to itself.
The second scenario might be seen as a "class method approach" (for those coming from OOP like Python or Java) where the package is similar to class and the exposed functions similar to class methods which are more generic and can be used across many types which implement the same interface, like so:
package products
// where p is a Product interface
func Add(p Product, field, value string) {
// Same switch-case as above
}
type Product interface {
someMethod()
}
And from another package:
package main
import (
"path/to/products"
)
type Car struct {
Title string
Brand string
Model string
}
type Ship struct {
// ...
}
type Airplane struct {
// ...
}
// All types implement `Product` and can be used in `products.Add`
func (c *Car) someMethod() {}
func (s *Ship) someMethod() {}
func (a *Airplane) someMethod() {}
func main() {
plane := &Airplane{}
products.Add(plane, "Model", "Boeing-747")
}
This is expected as per the spec:
The type of a method is the type of a function with the receiver as first argument.
See https://golang.org/ref/spec#Method_declarations
So when you declare the Add method on Product, you get a function that accepts a pointer to a Product as its first argument. So you end up with
func (p *Product) Add()
being translated to
func Add(p *Product)
So both your calls are valid and end up doing the same
Extending the fantastic answer by #Danilo:
package main
import "fmt"
type T struct {
i int
}
func (t *T) F() {
t = &T{1}
fmt.Println(t.i)
}
func F(t *T) {
fmt.Println(t.i)
}
func main() {
t := T{2}
(&t).F()
F(&t)
}
The type of the method func (t *T) F() is the type of the function func F(t *T) with the receiver (t *T) as first argument.

Private fields and methods for a struct

In the following test code I would like to have both mytype and the doPrivate method private, so that only members of mytype can access it, but not other types\functions in the scope of the mypackage package.
Can I do this in golang?
package mypackage
type mytype struct {
size string
hash uint32
}
func (r *mytype) doPrivate() string {
return r.size
}
func (r *mytype) Do() string {
return doPrivate("dsdsd")
}
Fields size and hash as well as the doPrivate method should be encapsulated and no other type should have access to them.
In Go, an identifier that starts with a capital letter is exported from the package, and can be accessed by anyone outside the package that declares it.
If an identifier starts with a lower case letter, it can only be accessed from within the package.
If you need members in a type to only be accessed by members of that type, you then need to place that type and its member functions in a separate package, as the only type in that package.
That's not how "privacy" works in Go: the granularity of privacy is the package.
If you really want only the members of mytype to access some fields, then you must isolate the struct and the functions in their own package.
But that's not the usual practice. Whether Go is OOP or not is debatable but clearly the practice isn't to encapsulate the code by a struct like you seem to want to do. Usually a package is small enough to be coherent: if you don't want to access fields from within the package, don't access them.
You can create an interface with the method you wish to expose and only access the object when wrapped into that interface.
package main
type mytype struct {
size string
hash uint32
}
// interface for exposed methods
type myinterface interface {
do() string
}
// constructor (optional)
func newMytype(size string, hash uint32) myinterface {
return &mytype{size, hash}
}
func (r *mytype) doPrivate() string {
return r.size
}
func (r *mytype) do() string {
return r.doPrivate()
}
func main() {
// with constructor
t := newMytype("100", 100)
t.do()
// t.doPrivate() // t.doPrivate undefined (type myinterface has no field or method doPrivate)
// without constructor
t2:= myinterface(&mytype{"100", 100})
t2.do()
// t.doPrivate() // t.doPrivate undefined (type myinterface has no field or method doPrivate)doPrivate)
}
https://play.golang.org/p/awjIIj8Kwms
You cannot do this in Go. Visibility is on a per package level only. But you may split your package into two.
In one module there can be any number of packages.
Public/Private works only across one package.
All public fields, methods and functions starts with uppercase char.
All private fields, methods and functions starts with lowercase char.
To add package to your module or program just create a lowercase folder and add package name to all files inside. Here is the example.
./main.go
./foo/foo.go
./foo/MyStruct.go
file ./foo/foo.go:
package foo
import "fmt"
func SomePublicFuncInFoo() {
somePrivateFuncInFoo()
}
func somePrivateFuncInFoo() {
fmt.Println("somePrivateFuncInFoo call")
}
file ./foo/MyStruct.go:
package foo
import "fmt"
type MyStruct struct {
MyPublicField string // starts with uppercase char
myPrivateField string // starts with lowercase char
}
func NewMyStruct(publicField string, privateField string) *MyStruct {
return &MyStruct{
MyPublicField: publicField,
myPrivateField: privateField,
}
}
func (self *MyStruct) SomePublicMethod() {
self.privateMethod()
}
func (self *MyStruct) privateMethod() {
fmt.Println("MyStruct", self.MyPublicField, self.myPrivateField)
}
file ./main.go:
package main
import (
"fmt"
"{your-module-name}/foo" // this line should be added by your IDE
)
func main() {
foo.SomePublicFuncInFoo()
myStruct := foo.NewMyStruct("string1", "string2")
fmt.Println("myStruct.MyPublicField=", myStruct.MyPublicField)
myStruct.SomePublicMethod()
}
You can have private variables and functions in Go, but the trick is that you simply don't define them in the struct. Bind them to the call stack of a closure, and simply don't return them.
package main
import (
"fmt"
)
type mytype struct {
Do func() string
}
func MyType(size string, hash uint32) mytype {
doPrivate := func() string {
return size
}
return mytype{
Do: func() string {
return doPrivate()
},
}
}
func main() {
instance := MyType("100", 100)
fmt.Println(instance.Do())
}

Resources