Related
Since multiple inheritance is bad (it makes the source more complicated) C# does not provide such a pattern directly. But sometimes it would be helpful to have this ability.
For instance I'm able to implement the missing multiple inheritance pattern using interfaces and three classes like that:
public interface IFirst { void FirstMethod(); }
public interface ISecond { void SecondMethod(); }
public class First:IFirst
{
public void FirstMethod() { Console.WriteLine("First"); }
}
public class Second:ISecond
{
public void SecondMethod() { Console.WriteLine("Second"); }
}
public class FirstAndSecond: IFirst, ISecond
{
First first = new First();
Second second = new Second();
public void FirstMethod() { first.FirstMethod(); }
public void SecondMethod() { second.SecondMethod(); }
}
Every time I add a method to one of the interfaces I need to change the class FirstAndSecond as well.
Is there a way to inject multiple existing classes into one new class like it is possible in C++?
Maybe there is a solution using some kind of code generation?
Or it may look like this (imaginary c# syntax):
public class FirstAndSecond: IFirst from First, ISecond from Second
{ }
So that there won't be a need to update the class FirstAndSecond when I modify one of the interfaces.
EDIT
Maybe it would be better to consider a practical example:
You have an existing class (e.g. a text based TCP client based on ITextTcpClient) which you do already use at different locations inside your project. Now you feel the need to create a component of your class to be easy accessible for windows forms developers.
As far as I know you currently have two ways to do this:
Write a new class that is inherited from components and implements the interface of the TextTcpClient class using an instance of the class itself as shown with FirstAndSecond.
Write a new class that inherits from TextTcpClient and somehow implements IComponent (haven't actually tried this yet).
In both cases you need to do work per method and not per class. Since you know that we will need all the methods of TextTcpClient and Component it would be the easiest solution to just combine those two into one class.
To avoid conflicts this may be done by code generation where the result could be altered afterwards but typing this by hand is a pure pain in the ass.
Consider just using composition instead of trying to simulate Multiple Inheritance. You can use Interfaces to define what classes make up the composition, eg: ISteerable implies a property of type SteeringWheel, IBrakable implies a property of type BrakePedal, etc.
Once you've done that, you could use the Extension Methods feature added to C# 3.0 to further simplify calling methods on those implied properties, eg:
public interface ISteerable { SteeringWheel wheel { get; set; } }
public interface IBrakable { BrakePedal brake { get; set; } }
public class Vehicle : ISteerable, IBrakable
{
public SteeringWheel wheel { get; set; }
public BrakePedal brake { get; set; }
public Vehicle() { wheel = new SteeringWheel(); brake = new BrakePedal(); }
}
public static class SteeringExtensions
{
public static void SteerLeft(this ISteerable vehicle)
{
vehicle.wheel.SteerLeft();
}
}
public static class BrakeExtensions
{
public static void Stop(this IBrakable vehicle)
{
vehicle.brake.ApplyUntilStop();
}
}
public class Main
{
Vehicle myCar = new Vehicle();
public void main()
{
myCar.SteerLeft();
myCar.Stop();
}
}
Since multiple inheritance is bad (it makes the source more complicated) C# does not provide such a pattern directly. But sometimes it would be helpful to have this ability.
C# and the .net CLR have not implemented MI because they have not concluded how it would inter-operate between C#, VB.net and the other languages yet, not because "it would make source more complex"
MI is a useful concept, the un-answered questions are ones like:- "What do you do when you have multiple common base classes in the different superclasses?
Perl is the only language I've ever worked with where MI works and works well. .Net may well introduce it one day but not yet, the CLR does already support MI but as I've said, there are no language constructs for it beyond that yet.
Until then you are stuck with Proxy objects and multiple Interfaces instead :(
I created a C# post-compiler that enables this kind of thing:
using NRoles;
public interface IFirst { void FirstMethod(); }
public interface ISecond { void SecondMethod(); }
public class RFirst : IFirst, Role {
public void FirstMethod() { Console.WriteLine("First"); }
}
public class RSecond : ISecond, Role {
public void SecondMethod() { Console.WriteLine("Second"); }
}
public class FirstAndSecond : Does<RFirst>, Does<RSecond> { }
You can run the post-compiler as a Visual Studio post-build-event:
C:\some_path\nroles-v0.1.0-bin\nutate.exe "$(TargetPath)"
In the same assembly you use it like this:
var fas = new FirstAndSecond();
fas.As<RFirst>().FirstMethod();
fas.As<RSecond>().SecondMethod();
In another assembly you use it like this:
var fas = new FirstAndSecond();
fas.FirstMethod();
fas.SecondMethod();
You could have one abstract base class that implements both IFirst and ISecond, and then inherit from just that base.
With C# 8 now you practically have multiple inheritance via default implementation of interface members:
interface ILogger
{
void Log(LogLevel level, string message);
void Log(Exception ex) => Log(LogLevel.Error, ex.ToString()); // New overload
}
class ConsoleLogger : ILogger
{
public void Log(LogLevel level, string message) { ... }
// Log(Exception) gets default implementation
}
This is along the lines of Lawrence Wenham's answer, but depending on your use case, it may or may not be an improvement -- you don't need the setters.
public interface IPerson {
int GetAge();
string GetName();
}
public interface IGetPerson {
IPerson GetPerson();
}
public static class IGetPersonAdditions {
public static int GetAgeViaPerson(this IGetPerson getPerson) { // I prefer to have the "ViaPerson" in the name in case the object has another Age property.
IPerson person = getPerson.GetPersion();
return person.GetAge();
}
public static string GetNameViaPerson(this IGetPerson getPerson) {
return getPerson.GetPerson().GetName();
}
}
public class Person: IPerson, IGetPerson {
private int Age {get;set;}
private string Name {get;set;}
public IPerson GetPerson() {
return this;
}
public int GetAge() { return Age; }
public string GetName() { return Name; }
}
Now any object that knows how to get a person can implement IGetPerson, and it will automatically have the GetAgeViaPerson() and GetNameViaPerson() methods. From this point, basically all Person code goes into IGetPerson, not into IPerson, other than new ivars, which have to go into both. And in using such code, you don't have to be concerned about whether or not your IGetPerson object is itself actually an IPerson.
In my own implementation I found that using classes/interfaces for MI, although "good form", tended to be a massive over complication since you need to set up all that multiple inheritance for only a few necessary function calls, and in my case, needed to be done literally dozens of times redundantly.
Instead it was easier to simply make static "functions that call functions that call functions" in different modular varieties as a sort of OOP replacement. The solution I was working on was the "spell system" for a RPG where effects need to heavily mix-and-match function calling to give an extreme variety of spells without re-writing code, much like the example seems to indicate.
Most of the functions can now be static because I don't necessarily need an instance for spell logic, whereas class inheritance can't even use virtual or abstract keywords while static. Interfaces can't use them at all.
Coding seems way faster and cleaner this way IMO. If you're just doing functions, and don't need inherited properties, use functions.
If you can live with the restriction that the methods of IFirst and ISecond must only interact with the contract of IFirst and ISecond (like in your example)... you can do what you ask with extension methods. In practice, this is rarely the case.
public interface IFirst {}
public interface ISecond {}
public class FirstAndSecond : IFirst, ISecond
{
}
public static MultipleInheritenceExtensions
{
public static void First(this IFirst theFirst)
{
Console.WriteLine("First");
}
public static void Second(this ISecond theSecond)
{
Console.WriteLine("Second");
}
}
///
public void Test()
{
FirstAndSecond fas = new FirstAndSecond();
fas.First();
fas.Second();
}
So the basic idea is that you define the required implementation in the interfaces... this required stuff should support the flexible implementation in the extension methods. Anytime you need to "add methods to the interface" instead you add an extension method.
Yes using Interface is a hassle because anytime we add a method in the class we have to add the signature in the interface. Also, what if we already have a class with a bunch of methods but no Interface for it? we have to manually create Interface for all the classes that we want to inherit from. And the worst thing is, we have to implement all methods in the Interfaces in the child class if the child class is to inherit from the multiple interface.
By following Facade design pattern we can simulate inheriting from multiple classes using accessors. Declare the classes as properties with {get;set;} inside the class that need to inherit and all public properties and methods are from that class, and in the constructor of the child class instantiate the parent classes.
For example:
namespace OOP
{
class Program
{
static void Main(string[] args)
{
Child somechild = new Child();
somechild.DoHomeWork();
somechild.CheckingAround();
Console.ReadLine();
}
}
public class Father
{
public Father() { }
public void Work()
{
Console.WriteLine("working...");
}
public void Moonlight()
{
Console.WriteLine("moonlighting...");
}
}
public class Mother
{
public Mother() { }
public void Cook()
{
Console.WriteLine("cooking...");
}
public void Clean()
{
Console.WriteLine("cleaning...");
}
}
public class Child
{
public Father MyFather { get; set; }
public Mother MyMother { get; set; }
public Child()
{
MyFather = new Father();
MyMother = new Mother();
}
public void GoToSchool()
{
Console.WriteLine("go to school...");
}
public void DoHomeWork()
{
Console.WriteLine("doing homework...");
}
public void CheckingAround()
{
MyFather.Work();
MyMother.Cook();
}
}
}
with this structure class Child will have access to all methods and properties of Class Father and Mother, simulating multiple inheritance, inheriting an instance of the parent classes. Not quite the same but it is practical.
Multiple inheritance is one of those things that generally causes more problems than it solves. In C++ it fits the pattern of giving you enough rope to hang yourself, but Java and C# have chosen to go the safer route of not giving you the option. The biggest problem is what to do if you inherit multiple classes that have a method with the same signature that the inheritee doesn't implement. Which class's method should it choose? Or should that not compile? There is generally another way to implement most things that doesn't rely on multiple inheritance.
If X inherits from Y, that has two somewhat orthogonal effects:
Y will provide default functionality for X, so the code for X only has to include stuff which is different from Y.
Almost anyplace a Y would be expected, an X may be used instead.
Although inheritance provides for both features, it is not hard to imagine circumstances where either could be of use without the other. No .net language I know of has a direct way of implementing the first without the second, though one could obtain such functionality by defining a base class which is never used directly, and having one or more classes that inherit directly from it without adding anything new (such classes could share all their code, but would not be substitutable for each other). Any CLR-compliant language, however, will allow the use of interfaces which provide the second feature of interfaces (substitutability) without the first (member reuse).
i know i know
even though its not allowed and so on, sometime u actualy need it so for the those:
class a {}
class b : a {}
class c : b {}
like in my case i wanted to do this
class b : Form (yep the windows.forms)
class c : b {}
cause half of the function were identical and with interface u must rewrite them all
Since the question of multiple inheritance (MI) pops up from time to time, I'd like to add an approach which addresses some problems with the composition pattern.
I build upon the IFirst, ISecond,First, Second, FirstAndSecond approach, as it was presented in the question. I reduce sample code to IFirst, since the pattern stays the same regardless of the number of interfaces / MI base classes.
Lets assume, that with MI First and Second would both derive from the same base class BaseClass, using only public interface elements from BaseClass
This can be expressed, by adding a container reference to BaseClass in the First and Second implementation:
class First : IFirst {
private BaseClass ContainerInstance;
First(BaseClass container) { ContainerInstance = container; }
public void FirstMethod() { Console.WriteLine("First"); ContainerInstance.DoStuff(); }
}
...
Things become more complicated, when protected interface elements from BaseClass are referenced or when First and Second would be abstract classes in MI, requiring their subclasses to implement some abstract parts.
class BaseClass {
protected void DoStuff();
}
abstract class First : IFirst {
public void FirstMethod() { DoStuff(); DoSubClassStuff(); }
protected abstract void DoStuff(); // base class reference in MI
protected abstract void DoSubClassStuff(); // sub class responsibility
}
C# allows nested classes to access protected/private elements of their containing classes, so this can be used to link the abstract bits from the First implementation.
class FirstAndSecond : BaseClass, IFirst, ISecond {
// link interface
private class PartFirst : First {
private FirstAndSecond ContainerInstance;
public PartFirst(FirstAndSecond container) {
ContainerInstance = container;
}
// forwarded references to emulate access as it would be with MI
protected override void DoStuff() { ContainerInstance.DoStuff(); }
protected override void DoSubClassStuff() { ContainerInstance.DoSubClassStuff(); }
}
private IFirst partFirstInstance; // composition object
public FirstMethod() { partFirstInstance.FirstMethod(); } // forwarded implementation
public FirstAndSecond() {
partFirstInstance = new PartFirst(this); // composition in constructor
}
// same stuff for Second
//...
// implementation of DoSubClassStuff
private void DoSubClassStuff() { Console.WriteLine("Private method accessed"); }
}
There is quite some boilerplate involved, but if the actual implementation of FirstMethod and SecondMethod are sufficiently complex and the amount of accessed private/protected methods is moderate, then this pattern may help to overcome lacking multiple inheritance.
We define interface as below:
interface IMyInterface
{
void MethodToImplement();
}
And impliments as below:
class InterfaceImplementer : IMyInterface
{
static void Main()
{
InterfaceImplementer iImp = new InterfaceImplementer();
iImp.MethodToImplement();
}
public void MethodToImplement()
{
Console.WriteLine("MethodToImplement() called.");
}
}
instead of creating a interface , why can we use the function directly like below :-)
class InterfaceImplementer
{
static void Main()
{
InterfaceImplementer iImp = new InterfaceImplementer();
iImp.MethodToImplement();
}
public void MethodToImplement()
{
Console.WriteLine("MethodToImplement() called.");
}
}
Any thoughts?
You are not implementing the interface in the bottom example, you are simply creating an object of InterfaceImplementer
EDIT: In this example an interface is not needed. However, they are extremely useful when trying to write loosely coupled code where you don't have to depend on concrete objects. They are also used to define contracts where anything implementing them has to also implement each method that it defines.
There is lots of information out there, here is just a brief intro http://www.csharp-station.com/Tutorials/Lesson13.aspx
If you really want to understand more about interfaces and how they can help to write good code, I would recommend the Head First Design Patterns book. Amazon Link
instead of creating a interface , why
can we use the function directly like
below
Are you asking what the point of the interface is?
Creating an interface allows you to decouple your program from a specific class, and instead code against an abstraction.
When your class is coded against an interface, classes that use your class can inject whichever class they want that implements this interface. This facilitates unit testing since not-easily-testable modules can be substituted with mocks and stubs.
The purpose of the interface is for some other class to be able to use the type without knowing the specific implementation, so long as that type conforms to a set of methods and properties defined in the interface contract.
public class SomeOtherClass
{
public void DoSomething(IMyInterface something)
{
something.MethodToImplement();
}
}
public class Program
{
public static void Main(string[] args)
{
if(args != null)
new SomeOtherClass().DoSomething(new ImplementationOne());
else
new SomeOtherClass().DoSomething(new ImplementationTwo());
}
}
Your example doesn't really follow that pattern, however; if one that one class implements the interface, then there really isn't much of a point. You can call it either way; it just depends on what kind of object hierarchy you have and what you intend to do for us to say whether using an interface is a good choice or not.
To sum: Both snippets you provide are valid code options. We'd need context to determine which is a 'better' solution.
Interfaces are not required, there is nothing wrong with the last section of code you posted. It is simply a class and you call one of it's public methods. It has no knowledge that an interface exists that this class happens to satisfy.
However, there are advantages:
Multiple Inheritance - A class can only extend one parent class, but can implement any number of interfaces.
Freedom of class use - If your code is written so that it only cares that it has an instance of SomethingI, you are not tied to a specific Something class. If tomorrow you decide that your method should return a class that works differently, it can return SomethingA and any calling code will not need to be changed.
The purpose of interfaces isn't found in instantiating objects, but in referencing them. Consider if your example is changed to this:
static void Main()
{
IMyInterface iImp = new InterfaceImplementer();
iImp.MethodToImplement();
}
Now the iTmp object is of the type IMyInterface. Its specific implementation is InterfaceImplementer, but there may be times where the implementation is unimportant (or unwanted). Consider something like this:
interface IVehicle
{
void MoveForward();
}
class Car : IVehicle
{
public void MoveForward()
{
ApplyGasPedal();
}
private void ApplyGasPedal()
{
// some stuff
}
}
class Bike : IVehicle
{
public void MoveForward()
{
CrankPedals();
}
private void CrankPedals()
{
// some stuff
}
}
Now say you have a method like this somewhere:
void DoSomething(IVehicle)
{
IVehicle.MoveForward();
}
The purpose of the interface becomes more clear here. You can pass any implementation of IVehicle to that method. The implementation doesn't matter, only that it can be referenced by the interface. Otherwise, you'd need a DoSomething() method for each possible implementation, which can get messy fast.
Interfaces make it possible for an object to work with a variety of objects that have no common base type but have certain common abilities. If a number of classes implement IDoSomething, a method can accept a parameter of type IDoSomething, and an object of any of those classes can be passed to it. The method can then use all of the methods and properties applicable to an IDoSomething without having to worry about the actual underlying type of the object.
The point of the interface is to define a contract that your implementing class abides by.
This allows you to program to a specification rather than an implementation.
Imagine we have the following:
public class Dog
{
public string Speak()
{
return "woof!";
}
}
And want to see what he says:
public string MakeSomeNoise(Dog dog)
{
return dog.Speak();
}
We really don't benefit from the Interface, however if we also wanted to be able to see what kind of noise a Cat makes, we would need another MakeSomeNoise() overload that could accept a Cat, however with an interface we can have the following:
public interface IAnimal
{
public string Speak();
}
public class Dog : IAnimal
{
public string Speak()
{
return "woof!";
}
}
public class Cat : IAnimal
{
public string Speak()
{
return "meow!";
}
}
And run them both through:
public string MakeSomeNoise(IAnimal animal)
{
return animal.Speak();
}
To put this in concise language...
Aim:
To create a class which can load and save a list of objects to/from a data source.
Current Method:
I have created a class which accepts two delegates as construction parameters:
private class Foo
{
public delegate List<object> LoadObjectsDelegate();
public delegate void SaveObjectsDelegate(List<object> data);
private LoadObjectsDelegate _loadDelegate;
private SaveObjectsDelegate _saveDelegate;
public Foo(LoadObjectsDelegate loadDelegate, SaveObjectsDelegate saveDelegate)
{
_loadDelegate = loadDelegate;
_saveDelegate = saveDelegate;
}
public List<object> Objects {get; private set;}
public void Load()
{
Objects = _loadDelegate.Invoke();
}
public void Save()
{
_saveDelegate.Invoke(Objects);
}
}
i was wondering if there was a cleaner way to do this.
By the looks of it i'd say you're trying to implement the Repository pattern, with some extended functionality. I also see no point in trying to inject the load and save logic, as it's the repository's job to implement them. If the logic of loading and saving objects is the same for all objects in your domain, have it implemented in a base class, and override it in derived classes if needed. This article on SO could give you some ideas on how to deal with this pattern.
Well for one thing I'd make it generic (Foo<T>) - your callers can always specify object if they really want, and it means they can use it in a more strongly typed way if necessary.
Currently I'm not sure the class is really adding much value, to be honest. It's really just grouping the three values together - and that's all. If that's what you intended, that's fine... but it feels a bit anaemic to me.
It is not cleaner, just another approach. If the behaviour which you are specifying in the delegates does not vary too much, consider the Template Pattern.
private abstract class AbstractFoo
{
public List<object> Objects { get; private set; }
public List<object> Load();
public abstract void Save(List<object> data);
// add common functionality
}
private class ConcreteFoo : AbstractFoo
{
public override List<object> Load()
{
// do specific stuff
}
public override void Save(List<object> data)
{
// do specific stuff
}
}
The disadvantage is that you need a class for each specific behaviour. Thus: If the behaviour given by the delegates varies a lot or you need to dynamically load different behaviours, I find your delegate approach is suitable. I usually find the Template Pattern easier to understand and easier to unit test.
You can go one step further: Behavior is specified by inheritance in the Template Pattern. Instead of inheritance you can specify the behaviour by reference if you choose a more abstract approach with an interface:
private interface IRepository
{
List<object> Load();
void Save(List<object> data);
}
private class FooBar
{
private IRepository repository;
public List<object> Objects { get; private set; }
public FooBar(IRepository repository)
{
this.repository = repository;
}
public void Load()
{
Objects = repository.Load();
}
public void Save()
{
repository.Save(Objects);
}
}
If you need some specific behaviour you need to implement a class derived from IRepository. This approach is even more easier to understand and unit test than the Template Pattern. I often end up replacing the Template Pattern with the latter approach.
I'm contemplating two different class designs for handling a situation where some repositories are read-only while others are read-write. (I don't foresee any need for a write-only repository.)
Class Design 1 -- provide all functionality in a base class, then expose applicable functionality publicly in sub classes
public abstract class RepositoryBase
{
protected virtual void SelectBase() { // implementation... }
protected virtual void InsertBase() { // implementation... }
protected virtual void UpdateBase() { // implementation... }
protected virtual void DeleteBase() { // implementation... }
}
public class ReadOnlyRepository : RepositoryBase
{
public void Select() { SelectBase(); }
}
public class ReadWriteRepository : RepositoryBase
{
public void Select() { SelectBase(); }
public void Insert() { InsertBase(); }
public void Update() { UpdateBase(); }
public void Delete() { DeleteBase(); }
}
Class Design 2 - read-write class inherits from read-only class
public class ReadOnlyRepository
{
public void Select() { // implementation... }
}
public class ReadWriteRepository : ReadOnlyRepository
{
public void Insert() { // implementation... }
public void Update() { // implementation... }
public void Delete() { // implementation... }
}
Is one of these designs clearly stronger than the other? If so, which one and why?
P.S. If this sounds like a homework question, it's not, but feel free to use it as one if you want :)
How about a third option, closely related to the first, but using interfaces instead:
public interface IReadRepository {
public void Select();
}
public interface IWriteRepository {
public void Insert();
public void Update();
public void Delete();
}
// Optional
public interface IRepository : IReadRepository, IWriteRepository {
}
public class Repository : IRepository {
// Implementation
}
This way the implementation is (or can be) all in one place, and the distinction is made only by which interface you are looking at.
(Edit: I think Eric Petroelje offers a very nice interface-based solution in his answer. I would probably vote for his suggestion, first of all.)
From your two choices, I would clearly vote for design #2.
With design #1, I think it doesn't make sense to have a "read-only" class that internally isn't read-only at all:
The read-only class is "heavier" than it needs to be.
Anyone can derive from your read-only class and then call any of the base class' modification methods. At the very least, with design #1, you ought to make the read-only class sealed.
With design #2, it's much clearer than the read-only class is a reduced version (base class) of the full-featured class, or phrased differently.
I would say design #2, but then you ought to change the name of the ReadOnlyRepository to something like ReadRepository.
Inheritance defines an IS-A relation between the classes, ans saying 'a ReadWriteRepository is a ReadOnlyRepository' doesn't sound logical. But 'ReadWriteRepository is a ReadingRepository' does.
First let me admit that I am making some assumptions about what you might intend doing. If this misses the point then let me know.
I am not sure how usefull the classes would be in either of your two options. I assume you would have calling code that would use an instance of a readonly repository and at othertimes an instance of a read/write repository, but the interfaces do not match so you would have to differenciate in your code anyway?
It might be better to provide a common interface and then throw exceptions if you try to write to the repository when it is readony and have your code handle the exceptions.
I would certainly say that design 2 is the strongest. If you want to have a read-only implementation, it does not need to know anything about writing. It makes perfect sense to extend the read-only implementation with the insert, update, and delete methods. I also think this design is the one that conforms the most to the Open-Closed principle.
I would suggest having a base class ReadableFoo, with a sealed derived class ImmutableFoo whose constructor takes a ReadableFoo, and a possibly-inheritable derived class MutableFoo, and possibly a class ReadableShadowFoo whose constructor would take a ReadableFoo (which may or may not be mutable), but which would act as a read-only wrapper to it.
The answer by Eric means SOLID Principle ISP. It´s so easy and basic to use.
Since multiple inheritance is bad (it makes the source more complicated) C# does not provide such a pattern directly. But sometimes it would be helpful to have this ability.
For instance I'm able to implement the missing multiple inheritance pattern using interfaces and three classes like that:
public interface IFirst { void FirstMethod(); }
public interface ISecond { void SecondMethod(); }
public class First:IFirst
{
public void FirstMethod() { Console.WriteLine("First"); }
}
public class Second:ISecond
{
public void SecondMethod() { Console.WriteLine("Second"); }
}
public class FirstAndSecond: IFirst, ISecond
{
First first = new First();
Second second = new Second();
public void FirstMethod() { first.FirstMethod(); }
public void SecondMethod() { second.SecondMethod(); }
}
Every time I add a method to one of the interfaces I need to change the class FirstAndSecond as well.
Is there a way to inject multiple existing classes into one new class like it is possible in C++?
Maybe there is a solution using some kind of code generation?
Or it may look like this (imaginary c# syntax):
public class FirstAndSecond: IFirst from First, ISecond from Second
{ }
So that there won't be a need to update the class FirstAndSecond when I modify one of the interfaces.
EDIT
Maybe it would be better to consider a practical example:
You have an existing class (e.g. a text based TCP client based on ITextTcpClient) which you do already use at different locations inside your project. Now you feel the need to create a component of your class to be easy accessible for windows forms developers.
As far as I know you currently have two ways to do this:
Write a new class that is inherited from components and implements the interface of the TextTcpClient class using an instance of the class itself as shown with FirstAndSecond.
Write a new class that inherits from TextTcpClient and somehow implements IComponent (haven't actually tried this yet).
In both cases you need to do work per method and not per class. Since you know that we will need all the methods of TextTcpClient and Component it would be the easiest solution to just combine those two into one class.
To avoid conflicts this may be done by code generation where the result could be altered afterwards but typing this by hand is a pure pain in the ass.
Consider just using composition instead of trying to simulate Multiple Inheritance. You can use Interfaces to define what classes make up the composition, eg: ISteerable implies a property of type SteeringWheel, IBrakable implies a property of type BrakePedal, etc.
Once you've done that, you could use the Extension Methods feature added to C# 3.0 to further simplify calling methods on those implied properties, eg:
public interface ISteerable { SteeringWheel wheel { get; set; } }
public interface IBrakable { BrakePedal brake { get; set; } }
public class Vehicle : ISteerable, IBrakable
{
public SteeringWheel wheel { get; set; }
public BrakePedal brake { get; set; }
public Vehicle() { wheel = new SteeringWheel(); brake = new BrakePedal(); }
}
public static class SteeringExtensions
{
public static void SteerLeft(this ISteerable vehicle)
{
vehicle.wheel.SteerLeft();
}
}
public static class BrakeExtensions
{
public static void Stop(this IBrakable vehicle)
{
vehicle.brake.ApplyUntilStop();
}
}
public class Main
{
Vehicle myCar = new Vehicle();
public void main()
{
myCar.SteerLeft();
myCar.Stop();
}
}
Since multiple inheritance is bad (it makes the source more complicated) C# does not provide such a pattern directly. But sometimes it would be helpful to have this ability.
C# and the .net CLR have not implemented MI because they have not concluded how it would inter-operate between C#, VB.net and the other languages yet, not because "it would make source more complex"
MI is a useful concept, the un-answered questions are ones like:- "What do you do when you have multiple common base classes in the different superclasses?
Perl is the only language I've ever worked with where MI works and works well. .Net may well introduce it one day but not yet, the CLR does already support MI but as I've said, there are no language constructs for it beyond that yet.
Until then you are stuck with Proxy objects and multiple Interfaces instead :(
I created a C# post-compiler that enables this kind of thing:
using NRoles;
public interface IFirst { void FirstMethod(); }
public interface ISecond { void SecondMethod(); }
public class RFirst : IFirst, Role {
public void FirstMethod() { Console.WriteLine("First"); }
}
public class RSecond : ISecond, Role {
public void SecondMethod() { Console.WriteLine("Second"); }
}
public class FirstAndSecond : Does<RFirst>, Does<RSecond> { }
You can run the post-compiler as a Visual Studio post-build-event:
C:\some_path\nroles-v0.1.0-bin\nutate.exe "$(TargetPath)"
In the same assembly you use it like this:
var fas = new FirstAndSecond();
fas.As<RFirst>().FirstMethod();
fas.As<RSecond>().SecondMethod();
In another assembly you use it like this:
var fas = new FirstAndSecond();
fas.FirstMethod();
fas.SecondMethod();
You could have one abstract base class that implements both IFirst and ISecond, and then inherit from just that base.
With C# 8 now you practically have multiple inheritance via default implementation of interface members:
interface ILogger
{
void Log(LogLevel level, string message);
void Log(Exception ex) => Log(LogLevel.Error, ex.ToString()); // New overload
}
class ConsoleLogger : ILogger
{
public void Log(LogLevel level, string message) { ... }
// Log(Exception) gets default implementation
}
This is along the lines of Lawrence Wenham's answer, but depending on your use case, it may or may not be an improvement -- you don't need the setters.
public interface IPerson {
int GetAge();
string GetName();
}
public interface IGetPerson {
IPerson GetPerson();
}
public static class IGetPersonAdditions {
public static int GetAgeViaPerson(this IGetPerson getPerson) { // I prefer to have the "ViaPerson" in the name in case the object has another Age property.
IPerson person = getPerson.GetPersion();
return person.GetAge();
}
public static string GetNameViaPerson(this IGetPerson getPerson) {
return getPerson.GetPerson().GetName();
}
}
public class Person: IPerson, IGetPerson {
private int Age {get;set;}
private string Name {get;set;}
public IPerson GetPerson() {
return this;
}
public int GetAge() { return Age; }
public string GetName() { return Name; }
}
Now any object that knows how to get a person can implement IGetPerson, and it will automatically have the GetAgeViaPerson() and GetNameViaPerson() methods. From this point, basically all Person code goes into IGetPerson, not into IPerson, other than new ivars, which have to go into both. And in using such code, you don't have to be concerned about whether or not your IGetPerson object is itself actually an IPerson.
In my own implementation I found that using classes/interfaces for MI, although "good form", tended to be a massive over complication since you need to set up all that multiple inheritance for only a few necessary function calls, and in my case, needed to be done literally dozens of times redundantly.
Instead it was easier to simply make static "functions that call functions that call functions" in different modular varieties as a sort of OOP replacement. The solution I was working on was the "spell system" for a RPG where effects need to heavily mix-and-match function calling to give an extreme variety of spells without re-writing code, much like the example seems to indicate.
Most of the functions can now be static because I don't necessarily need an instance for spell logic, whereas class inheritance can't even use virtual or abstract keywords while static. Interfaces can't use them at all.
Coding seems way faster and cleaner this way IMO. If you're just doing functions, and don't need inherited properties, use functions.
If you can live with the restriction that the methods of IFirst and ISecond must only interact with the contract of IFirst and ISecond (like in your example)... you can do what you ask with extension methods. In practice, this is rarely the case.
public interface IFirst {}
public interface ISecond {}
public class FirstAndSecond : IFirst, ISecond
{
}
public static MultipleInheritenceExtensions
{
public static void First(this IFirst theFirst)
{
Console.WriteLine("First");
}
public static void Second(this ISecond theSecond)
{
Console.WriteLine("Second");
}
}
///
public void Test()
{
FirstAndSecond fas = new FirstAndSecond();
fas.First();
fas.Second();
}
So the basic idea is that you define the required implementation in the interfaces... this required stuff should support the flexible implementation in the extension methods. Anytime you need to "add methods to the interface" instead you add an extension method.
Yes using Interface is a hassle because anytime we add a method in the class we have to add the signature in the interface. Also, what if we already have a class with a bunch of methods but no Interface for it? we have to manually create Interface for all the classes that we want to inherit from. And the worst thing is, we have to implement all methods in the Interfaces in the child class if the child class is to inherit from the multiple interface.
By following Facade design pattern we can simulate inheriting from multiple classes using accessors. Declare the classes as properties with {get;set;} inside the class that need to inherit and all public properties and methods are from that class, and in the constructor of the child class instantiate the parent classes.
For example:
namespace OOP
{
class Program
{
static void Main(string[] args)
{
Child somechild = new Child();
somechild.DoHomeWork();
somechild.CheckingAround();
Console.ReadLine();
}
}
public class Father
{
public Father() { }
public void Work()
{
Console.WriteLine("working...");
}
public void Moonlight()
{
Console.WriteLine("moonlighting...");
}
}
public class Mother
{
public Mother() { }
public void Cook()
{
Console.WriteLine("cooking...");
}
public void Clean()
{
Console.WriteLine("cleaning...");
}
}
public class Child
{
public Father MyFather { get; set; }
public Mother MyMother { get; set; }
public Child()
{
MyFather = new Father();
MyMother = new Mother();
}
public void GoToSchool()
{
Console.WriteLine("go to school...");
}
public void DoHomeWork()
{
Console.WriteLine("doing homework...");
}
public void CheckingAround()
{
MyFather.Work();
MyMother.Cook();
}
}
}
with this structure class Child will have access to all methods and properties of Class Father and Mother, simulating multiple inheritance, inheriting an instance of the parent classes. Not quite the same but it is practical.
Multiple inheritance is one of those things that generally causes more problems than it solves. In C++ it fits the pattern of giving you enough rope to hang yourself, but Java and C# have chosen to go the safer route of not giving you the option. The biggest problem is what to do if you inherit multiple classes that have a method with the same signature that the inheritee doesn't implement. Which class's method should it choose? Or should that not compile? There is generally another way to implement most things that doesn't rely on multiple inheritance.
If X inherits from Y, that has two somewhat orthogonal effects:
Y will provide default functionality for X, so the code for X only has to include stuff which is different from Y.
Almost anyplace a Y would be expected, an X may be used instead.
Although inheritance provides for both features, it is not hard to imagine circumstances where either could be of use without the other. No .net language I know of has a direct way of implementing the first without the second, though one could obtain such functionality by defining a base class which is never used directly, and having one or more classes that inherit directly from it without adding anything new (such classes could share all their code, but would not be substitutable for each other). Any CLR-compliant language, however, will allow the use of interfaces which provide the second feature of interfaces (substitutability) without the first (member reuse).
i know i know
even though its not allowed and so on, sometime u actualy need it so for the those:
class a {}
class b : a {}
class c : b {}
like in my case i wanted to do this
class b : Form (yep the windows.forms)
class c : b {}
cause half of the function were identical and with interface u must rewrite them all
Since the question of multiple inheritance (MI) pops up from time to time, I'd like to add an approach which addresses some problems with the composition pattern.
I build upon the IFirst, ISecond,First, Second, FirstAndSecond approach, as it was presented in the question. I reduce sample code to IFirst, since the pattern stays the same regardless of the number of interfaces / MI base classes.
Lets assume, that with MI First and Second would both derive from the same base class BaseClass, using only public interface elements from BaseClass
This can be expressed, by adding a container reference to BaseClass in the First and Second implementation:
class First : IFirst {
private BaseClass ContainerInstance;
First(BaseClass container) { ContainerInstance = container; }
public void FirstMethod() { Console.WriteLine("First"); ContainerInstance.DoStuff(); }
}
...
Things become more complicated, when protected interface elements from BaseClass are referenced or when First and Second would be abstract classes in MI, requiring their subclasses to implement some abstract parts.
class BaseClass {
protected void DoStuff();
}
abstract class First : IFirst {
public void FirstMethod() { DoStuff(); DoSubClassStuff(); }
protected abstract void DoStuff(); // base class reference in MI
protected abstract void DoSubClassStuff(); // sub class responsibility
}
C# allows nested classes to access protected/private elements of their containing classes, so this can be used to link the abstract bits from the First implementation.
class FirstAndSecond : BaseClass, IFirst, ISecond {
// link interface
private class PartFirst : First {
private FirstAndSecond ContainerInstance;
public PartFirst(FirstAndSecond container) {
ContainerInstance = container;
}
// forwarded references to emulate access as it would be with MI
protected override void DoStuff() { ContainerInstance.DoStuff(); }
protected override void DoSubClassStuff() { ContainerInstance.DoSubClassStuff(); }
}
private IFirst partFirstInstance; // composition object
public FirstMethod() { partFirstInstance.FirstMethod(); } // forwarded implementation
public FirstAndSecond() {
partFirstInstance = new PartFirst(this); // composition in constructor
}
// same stuff for Second
//...
// implementation of DoSubClassStuff
private void DoSubClassStuff() { Console.WriteLine("Private method accessed"); }
}
There is quite some boilerplate involved, but if the actual implementation of FirstMethod and SecondMethod are sufficiently complex and the amount of accessed private/protected methods is moderate, then this pattern may help to overcome lacking multiple inheritance.