Interface segregation principle

svg viewer

ISP: Only inherit applicable methods

“Many client-specific interfaces are better than one general-purpose interface.”

The interface segregation principle (ISP) requires that classes only be able to perform behaviors that are useful to achieve its end functionality. In other words, classes do not include behaviors they do not use.

This relates to our first SOLID principle in that together these two principles strip a class of all variables, methods, or behaviors that do not directly contribute to their role. Methods must contribute to the end goal in their entirety.

Any unused part of the method should be removed or split into a separate method.

The advantage of ISP is that it splits large methods into smaller, more specific methods. This makes the program easier to debug for three reasons:

There is less code carried between classes. Less code means fewer bugs.
A single method is responsible for a smaller variety of behaviors. If there is a problem with a behavior, you only need to look over the smaller methods.
If a general method with multiple behaviors is passed to a class that doesn't support all behaviors (such as calling for a property that the class doesn't have), there will be a bug if the class tries to use the unsupported behavior.

Implementation

To see how the ISP principle looks in code, let's see how a program changes with and without following the ISP principle.

First, the program that doesn't follow ISP:

// Not following the Interface Segregation Principle  
  
  public interface IWorker  
  {  
      string ID { get; set; }  
      string Name { get; set; }  
      string Email { get; set; }  
      float MonthlySalary { get; set; }  
      float OtherBenefits { get; set; }  
      float HourlyRate { get; set; }  
      float HoursInMonth { get; set; }  
      float CalculateNetSalary();  
      float CalculateWorkedSalary();  
  }  
  
  public class FullTimeEmployee : IWorker  
  {  
      public string ID { get; set; }  
      public string Name { get; set; }  
      public string Email { get; set; }  
      public float MonthlySalary { get; set; }  
      public float OtherBenefits { get; set; }  
      public float HourlyRate { get; set; }  
      public float HoursInMonth { get; set; }  
      public float CalculateNetSalary() => MonthlySalary + OtherBenefits;  
      public float CalculateWorkedSalary() => throw new NotImplementedException();  
  }  
  
  public class ContractEmployee : IWorker  
  {  
      public string ID { get; set; }  
      public string Name { get; set; }  
      public string Email { get; set; }  
      public float MonthlySalary { get; set; }  
      public float OtherBenefits { get; set; }  
      public float HourlyRate { get; set; }  
      public float HoursInMonth { get; set; }  
      public float CalculateNetSalary() => throw new NotImplementedException();  
      public float CalculateWorkedSalary() => HourlyRate * HoursInMonth;  
  }

This program does not follow ISP because the FullTimeEmployee class does not need the CalculateWorkedSalary() function, and the ContractEmployeeclass does not need the CalculateNetSalary().

Neither of these methods advance the goal of these classes. Instead, they are implemented because they are derived classes of the IWorker interface.

Here's how we could refactor the program to follow the ISP principle:

// Following the Interface Segregation Principle  
  
    public interface IBaseWorker  
    {  
        string ID { get; set; }  
        string Name { get; set; }  
        string Email { get; set; }  
         
         
    }  
  
    public interface IFullTimeWorkerSalary : IBaseWorker  
    {  
        float MonthlySalary { get; set; }  
        float OtherBenefits { get; set; }  
        float CalculateNetSalary();  
    }  
  
    public interface IContractWorkerSalary : IBaseWorker  
    {  
        float HourlyRate { get; set; }  
        float HoursInMonth { get; set; }  
        float CalculateWorkedSalary();  
    }  
  
    public class FullTimeEmployeeFixed : IFullTimeWorkerSalary  
    {  
        public string ID { get; set; }  
        public string Name { get; set; }  
        public string Email { get; set; }  
        public float MonthlySalary { get; set; }  
        public float OtherBenefits { get; set; }  
        public float CalculateNetSalary() => MonthlySalary + OtherBenefits;  
    }  
  
    public class ContractEmployeeFixed : IContractWorkerSalary  
    {  
        public string ID { get; set; }  
        public string Name { get; set; }  
        public string Email { get; set; }  
        public float HourlyRate { get; set; }  
        public float HoursInMonth { get; set; }  
        public float CalculateWorkedSalary() => HourlyRate * HoursInMonth;  
    }

In this version, we've split the general interface IWorker into one base interface, IBaseWorker, and two child interfaces IFullTimeWorkerSalary and IContractWorkerSalary.

The general interface contains methods that all workers share. The child interfaces split up methods by worker type, FullTime with a salary or Contract that gets paid hourly.

Now, our classes can implement the interface for that type of worker to access all methods and properties in the base class and the worker-specific interface.

The end classes now only contain methods and properties that further their goal and thus achieve the ISP principle.