Module 4, Interfaces for Abstraction, Topic 1.2, Interfaces. An interface is a concept used in Go and it helps us get polymorphisms. So we don't get inheritance, we don't need inheritance, we don't need overriding. We can use interfaces to basically accomplish the same thing. And I think it's in a better way. I think it's cleaner. But this is up to, this is up for argument, right? because people who are used to Java, or something like that, will like it the other way. They'll like their inheritance and want to fight to keep it, but Go does this in a different way. So an interface is basically a set of method signatures. So by signatures, I mean the name of the method, the parameters of the method, and their types, and their return values, and their types. So that's all it is, there's no implementations, implementation of a method, right? So it just defines the signatures to the method. So it says the methods have to have this name, these parameters, these return values. That's an interface. So it's not a type or anything, it's less than that. It's used to express conceptual similarity between types. So what I mean by that is say I've got an interface called Shape2D, right? And that's my interface, and it's supposed to represent two-dimensional shapes. So all two-dimensional shapes, I'm going to say they have to have two methods, okay? An area method and a perimeter method, right? If it's a two-dimensional shape, you gotta be able to compute the area, you gotta be able to compute perimeter. And then I call it a two-dimensional shape. So my two-dimensional shape interface just says look, gotta have these two methods with the arguments that I say. In this case, no arguments, right? You gotta have these two methods in order to be considered a two dimensional shape. But if you have those two methods, any type that has those two methods, it can be considered a two-dimensional shape. That's what an interface is saying. So it's saying that they are conceptually similar, a circle, a square, a rectangle, a triangle. You can compute the area of all those and the perimeter of all those. So I'm going to call them two-dimensional shapes So, satisfying interface. A type satisfies an interface if it actually defines all the methods specified in the interface. So remember that a method in an interface, an interface doesn't specify. It doesn't design, give you the method. It just gives you the signature for the method. It doesn't implement the method. So if a type actually implements all the methods in the interface with the same method signatures, so same arguments, same name, same return values, then that type is said to satisfy the interface. So for instance, I can have a Shape2D interface. And I have it two types, a Rectangle type, a Triangle type. And if the Rectangle type and the Triangle type both define an area and a parameter method, with the appropriate arguments and return values. Then you will say Rectangle and Triangle both satisfy the Shape2D interface. And so they can be considered to be two-dimensional shapes. Now Rectangle and Triangle can have lots of other methods besides the area and perimeter, right? Any number of other methods. Also, Rectangle and Triangle can have lots of other data. Maybe they're structs, and struct types, and they have x, y, z points. Who knows what they have, right? None of that matters. As long as they have the area and perimeter that is specified in the interface, then that's enough. And they can be considered to be satisfying the interface and considered to be two-dimensional shapes. So what that accomplishes is it basically accomplishes what you get from inheritance and overriding together, right? So now I got Rectangle and Triangle. If they're both Shape2D, if they satisfy that interface, then they both have area, they both have perimeter. But their area and perimeter methods can do completely different things, right? because when you compute the area of a rectangle, it is different than computing the area of a triangle, right? So their implementations can be different, but they have the same name. And at the high level they form the same thing, they compute area. So in this way, we're using interface to accomplish in Go lang what you would use inheritance and overriding to accomplish typically in a thing like JAVA or some other object oriented language. So how do you define an interface type? It's pretty straightforward, it looks sort of like a struct. There we got Shape2D, so I say type Shape2D interface. Just use the keyword interface after the name of the interface. In curly brackets, I start listing the signatures of the methods. So in this case is only two method and two signatures that I need, area and perimeter. Both of them area and perimeter take no arguments and they return a float, a 64 float. So that's it, that's how you define this interface. I just list all these method signatures that I want to put in the interface. Now then, say later on in my code I defined a type Triangle. And I don't even say what's in it. I just said it have an open curly bracket, close curly bracket, with some dots, just to say it doesn't matter what data I'm putting in that Triangle, right? Maybe that data Triangle, maybe it's a struct or something like that, who knows what it is. It doesn't matter, but whatever it is, as long as I define a function area whose receiver type is a Triangle and a function perimeter whose receiver type is also a Triangle. And the area and perimeter also take no arguments and return float64s, just like the interface. Then this type Triangle is said to satisfy that ShapeID interface, Shape2D interface, rather. So and it doesn't matter what other data is in Triangle, what other methods are using Triangle as a received type. As long as got area and perimeter and it matches and satisfies the interface. And one other thing before I go on, is that you don't have to state it explicitly. So in other languages that have interfaces, you often have to say explicitly the Triangle satisfies this interface, the Shape2D interface. You don't say that in Go. You just say here's the interface and here's my type Triangle, and here the methods for Triangle. And the compiler figure, it can do the matching automatically. So I see you have a area in perimeter, I will treat you just like I treat anything that satisfies the Shape2D interface. Thank you.
