Protocol vs Subclassing - objective-c

I have found that we can create model classes in Objective-C using following methods.
Create a protocol with some properties and methods. Implementing this protocol in any of the classes provide us an inheritance feature of OOP concept.
Create a NSObject class with interface and implementation. Add properties and method declarations to that class. SubClassing of this class also provide us an inheritance feature of OOP concept.
First of all I have to know that, these two methods will work as it stated.
If it will work what is difference between them, pros and cons.
Thanks in Advance.

Related

What is Protocol Oriented Programming in Swift? What added value does it bring?

From Apple's own website: "At the heart of Swift's design are two incredibly powerful ideas: protocol-oriented programming and first class value semantics."
Can someone please elaborate what exactly is protocol oriented programming, and what added value does it bring?
I have read this and watched the Protocol-Oriented Programming in Swift video, but coming from an Objective-C background still haven't understood it. I kindly ask for a very plain English answer along with code snippets & technical details about how it's different from Objective-C.
Just one of the confusions I have is using <tableViewDelegate, CustomDelegate> Couldn't we also conform to multiple protocols in Objective-C as well? So again how is Swift new?
EDIT: See Protocol-Oriented Views video. I find this video to be more basic and easier to grasp a meaningful use case. The WWDC video itself is a bit advanced and requires more breadth. Additionally the answers here are somewhat abstract.
Preface: POP and OOP are not mutually exclusive. They're design paradigms that are greatly related.
The primary aspect of POP over OOP is that is prefers composition over inheritance. There are several benefits to this.
In large inheritance hierarchies, the ancestor classes tend to contain most of the (generalized) functionality, with the leaf subclasses making only minimal contributions. The issue here is that the ancestor classes end up doing a lot of things. For example, a Car drives, stores cargo, seats passengers, plays music, etc. These are many functionalities that are each quite distinct, but they all get indivisibly lumped into the Car class. Descendants of Car, such as Ferrari, Toyota, BMW, etc. all make minimal modifications to this base class.
The consequence of this is that there is reduced code reuse. My BoomBox also plays music, but it's not a car. Inheriting the music-playing functionality from Car isn't possible.
What Swift encourages instead is that these large monolithic classes be broken down into a composition of smaller components. These components can then be more easily reused. Both Car and BoomBox can use MusicPlayer.
Swift offers multiple features to achieve this, but the most important by far are protocol extensions. They allow implementation of a protocol to exist separate of its implementing class, so that many classes may simply implement this protocol and instantly gain its functionality.
It surprised me that none of the answers mentioned value type in POP.
To understand what is protocol oriented programming, you need to understand what are drawbacks of objected oriented programming.
It (Objc) has only one inheritance. If we have very complicated hierarchy of inheritance, the bottom class may have a lot of unnecessary state to hold.
It uses class which is a reference type. Reference type may cause code unsafe. e.g. Processing collection of reference types while they are being modified.
While in protocol oriented programming in swift:
It can conform multiple protocols.
It can be used by not only class, but also structures and enumerations.
It has protocol extension which gives us common functionality to all types that conforms to a protocol.
It prefers to use value type instead of reference type. Have a look at the standard swift library here, you can find majority of types are structures which is value type. But this doesn't mean you don't use class at all, in some situation, you have to use class.
So protocol oriented programming is nothing but just an another programming paradigm that try to solve the OOP drawbacks.
In Objective C protocol is the same thing as interface in most languages. So in Objective C protocol's usage is limited to SOLID principle "Depend upon Abstractions. Do not depend upon concretions."
In Swift protocols were improved so seriously that since they still could be used as interfaces in fact they are closer to classes (like Abstract classes in C++)
In Objective C the only way to share functionality between classes is an inheritance. And you could inherit the only one parent class. In Swift you could also adopt as many protocols as you want. And since protocols in Swift can have default methods implementation they give us a fully-functional Multiple inheritance. More flexibility, better code reuse - awesome!
Conclusion:
Protocol Oriented Programming is mostly the same as OOP but it pays additional attention to functionality sharing not only via inheritance but also via protocol adoption (Composition over inheritance).
Worth to mention that in C++ abstract classes are very similar to protocols in Swift but no one says C++ supports some specific type of OOP. So in general POP is a one of the versions of OOP if we speak about programming paradigms. For Swift POP is an improved version of OOP.
Adding to the above answer
Protocol is a interface in which signature of methods and properties are declared and any class/struct/enum subclassing the enum must have to obey the contract means they have to implement all the methods and properties declared in superclass protocol.
Reason to use Protocol
Classes provide single inheritance and struct doesn't support inheritance. Thus protocols was introduced.
Extension The methods declare inside the protocol can be implemented inside the extension to avoid the redundancy of the code in case protocol is being inherited in multiple class / struct having same method implementation. We can call the method by simply declaring the object of struct/enums. Even we can restrict the extension to a list of classes, only restricted class will be able to use the method implemented inside the extension while rest of the classes have to implement method inside own class.
Example
protocol validator{
var id : String{ get }
func capitialise()-> (String)
}
extension validator where Self : test{
func capitialise() -> String{
return id.capitalized
}
}
class test : validator {
var id: String
init(name:String) {
id = name
}
}
let t = test(name: "Ankit")
t.capitialise()
When to use In OOP suppose we have a vehicle base class which is inherited by the airplane, bike, car etc. Here break, acceleration may be common method among three subclass but not the flyable method of airplane. Thus if we are declaring flyable method also in OOP, the bike and car subclass also have the inherit flyable method which is of no use for those class. Thus in the POP we can declare two protocols one is for flyable objects and other is for break and acceleration methods. And flyable protocol can be restricted to use by only the airplane
Protocol Oriented Programming(POP)
protocol-first approach
Protocol as a key point of OOP concept. abstraction, inheritance, polymorphism, encapsulation.
Protocol as a base for SOLID[About]
Protocol instead of class hierarchy tree. Its is hard to support class inheritance. Moreover it has some performance impact
Class/struct can implements multiple protocols(a kind of multiple inheritance)
Composition over inheritance.
extension MyClass: MyProtocol {
}
Default method. Shared implementation for all implementators
extension MyProtocol {
func foo() {
//logic
}
}
Protocol inheritance. One protocol can extends another protocol. Implementator of protocol one should implements all from first and the second protocols
protocol ProtocolB: ProtocolA {
}
value type implement protocol(as usual reference type)[About]
Protocol Oriented Programming (POP)
Came since Swift 2.0
class (OOP)
is reference type
memory leak, incorrect data stored,race condition to access in complex multi-thread environments
can be large by inheriting members of super classes at chain time
struct (POP)
is value type - each time a fresh copy is made when needed
provides multi inheritance - inherits protocols
Protocol :
Defines what Methods, Properties and Initializes are required o Can
inherit another Protocol(s)
Don’t have to use override keyword to implement protocol functions
Extensions:
Default value and
The default implementation for protocol
Can add extra members to
protocol
what is protocol oriented programming? What’s POP?
is a new programming paradigm
we start designing our system by defining protocols. We rely on new concepts: protocol extensions, protocol inheritance, and protocol compositions.
value types can inherit from protocols, even multiple protocols. Thus, with POP, value types have become first class citizens in Swift. value types like enums, structs
*POP lets you to add abilities to a class or struct or enum with protocols which supports multiple implementations.
Apple tells us:
“Don’t start with a class, start with a protocol.”
Why? Protocols serve as better abstractions than classes.
Protocols: are a fundamental feature of Swift. They play a leading role in the structure of the Swift standard library and are a common method of abstraction.
Protocols are used to define a “blueprint of methods, properties, and other requirements that suit a particular task or piece of functionality.”
Benefits of Protocol-Oriented Programming:
All classes are decoupled from each other
Separating the concerns of declaration from implementation
Reusability
Testability

Can i use inheritance instead of implement an interface in strategy pattern?

From a picture, Can i use inheritance instead of implement an interface?
I mean change from "ConcreteStrategyA and ConcreteStrategyB implements Strategy Interface" to "ConcreteStrategyA and ConcreteStrategyB extends Strategy Class"
Is it still work well or have some problem?
If it still work well my next question is "Why most people prefer to use interface?"
Well technically from a Design pattern perspective with the Strategy pattern, the concrete Strategies need to implement (I mean write code for, not the interface implements thing) a common contract which the Strategy Context is aware of. This is the primary backbone of Strategy pattern philosophy. The adherence to the common contract is what allows the Strategy context to replace the concrete strategies based on some runtime feature. This pattern ideology is what we loosely call Polymorphism in OOP parlance.
Now in Java you can implement his polymorphic strategy either as an interface or as inheritance. For interface you have given the example in the question itself. For inheritance as long as the contract holds between subclasses (something like a base abstract class with an abstract contract which subclasses implement to provide concrete strategy implementations) you can implement Strategy pattern in inheritance as well.
Now thinking about it from OOP perspective. For OOP inheritance is something which a subclass inherits from a super class. The subclass automatically owns and thus demonstrates that inherited generic behavior but it has a choice to make that behavior more specific to its own type. Thus multiple subclasses can override the same behavior and make bits of it more specific to their use. But this chain becomes cumbersome to manage when it gets too long or when subclasses try to inherit the behaviors which don't apply to them logically.
Thus it makes more sense to implement Strategy pattern using interfaces as against inheritance.
Absolutely. Inheritance is most often used with an abstract base class, when you want your derived strategies to share some common code.
People prefer to use interfaces or abstract classes over concrete base classes because :
With a Dependency Inversion approach, a class needs loose coupling to its Strategy. It only needs to know that the Strategy fulfills a contract, but doesn't want to know about its implementation details. Interfaces and abstract classes are an elegant and minimal way to define a contract without specifying the implementation.
It doesn't make sense to instantiate the base Strategy class most of the time, because it's a general, abstract concept -- in fact, it's better if you forbid instantiating it.
There are no technical problems.
However, a class can only extend one base class but it can implement multiple interfaces. So if you want to, let's say, change your inheritance structure in the future it is easier if you choose to implement an interface instead.
As you know a design pattern is "a general solution to a commonly occurring problem". It just describe a general solution without indications concerning implementation details.
If your problem requires a class in place of the interface, there is nothing wrong replacing it with a concrete (or abstract) class.
Using an interface in the pattern UML is a way to say: "you have to expose this set of public methods".
So, no problem using your approach. As an alternative you could leave the Strategy interface and implement it in a StrategyImpl class, then you can inherit this class in your ConcreteStrategyA and ConcreteStrategyB classes.

iOS: how to make an interface cannot be instantiated?

First I designed a protocol as my base class, which could not be instantiated. Later I found there should be some shared methods between classes conformed to this protocol, so I'd like to add some default implementation to the protocol.
AFAIK, Obj-C doesn't support protocol with default behaviors so I have to make it an interface. Then came up with my question, how to make an interface cannot be instantiated?
Thanks.
Interfaces can't be instantiated, ever. Maybe you're thinking about abstract classes, which aren't really available in Objective-C. Maybe you should rethink your program structure and try to use more inheritance. If the class you are inheriting from conforms to a certain protocol ( which is somewhat of the Objective-C equivalent of Java Interfaces ), then you can instantiate that, or subclass it and then instantiate it.
Have a look here for more information about inheritance: http://www.techotopia.com/index.php/Objective-C_Inheritance

#protocol vs Class Cluster

What are those major pro and contra for #protocol and Class Clusters concepts in Objective-C ?
Both of them introduce Loose Coupling in program architecture. Are they conceptually almost equal, or is there something else worth to know ?
Caveat: Not a cocoa pro, but I don't believe they are equal at all.
With Class Clusters you subclass.
Class clusters are a design pattern that the Foundation framework makes extensive use of. Class clusters group a number of private concrete subclasses under a public abstract superclass. The grouping of classes in this way simplifies the publicly visible architecture of an object-oriented framework without reducing its functional richness. Class clusters are based on the Abstract Factory design pattern discussed in “Cocoa Design Patterns.”
#protocols on the other hand, are more like Java interfaces.
The Objective-C extension called a protocol is very much like an interface in Java. Both are simply a list of method declarations publishing an interface that any class can choose to implement. The methods in the protocol are invoked by messages sent by an instance of some other class.
In short, Class Clusters are subclass/superclass where the subclass conforms to the entire identity of the superclass so that the implementation can be hidden from the user. This is apparent in the case of NSArray where the compiler uses context to choose the best type of data structure to use. You don't call NSTree or NSLinkedList like you might in Java. You can see how NSNumber is implemented here, especially the part where it says:
// NSNumber instance methods -- which will never be called...
#protocols are like client/server relationship where the client class adopts a protocol of the server class, so the server can call functionality on the client. <NSAppDelegate> and <UIAlertViewDelegate> are great examples of the use of protocols.

Supressing warning for a class (trying to simulate an Abstract class in objc)

This is the situation. I've been a C++ programmer for ages. I like abstract classes and "interfaces" so I would likt to do the same using objc.
I use a protocol for my interfaces, but the problem is that when my abstract class inherits from the protocol (I don't want to implement it here) I get warnings like:
warning: method definition for 'XXXXX' not found and 'XXXXX' class does not fully implement the 'XXXXXX' protocol.
Is there anyway to supress this? I hope child classes of this ones will throw "correct warnings" if base class did not implemented the protocol.
Another option is to inherit from the protocol just when needed, but I like to force this in the base class to make sure inherited implementes the interface.
Any tip?,
Thanks in advance.
When you implement a protocol in an Objective-C class, you have to implement all the methods. However, you can provide stub implementations. This mailing list post describes how to use doesNotRecognizeSelector: in "abstract" classes.
I don't think there is a solution in a way you are looking for. You must define all methods declared in the protocol, at least implement them as an empty methods.
I understand that you are looking for a C++ like code, but Obj-C is different and we must live with it. Also, gcc supports c++/obj-c mix, so you can write some part of the project in pure C++ what is great when you need some low-level code or want something easy to port.