General Guidelines for C# Class Implementation

Introduction

While implementing my first projects using C# I found out that there were several issues to take into account if I wanted my classes to behave correctly and make good friends with .NET. This list is more about the "hows" and the "whats" and not the "whys" and while it is in no way complete, it contains the guidelines that I currently follow. Ah, and I almost forgot... it's guaranteed to be incomplete!

The Guidelines

1. Implement IComparable.CompareTo() if ordering of objects makes sense. Also implement operators <, <=, > and >= in terms of IComparable.CompareTo().

   int IComparable.CompareTo(object obj)
   { 
       return m_data - ((MyType)obj).m_data; 
   } 
   
   public static bool operator<(MyType lhs, MyType rhs)
   { 
       return ((IComparable)lhs).CompareTo(rhs) < 0;
   }
   
   public static bool operator<=(MyType lhs, MyType rhs)
   { 
       return ((IComparable)lhs).CompareTo(rhs) <= 0;
   }
   
   public static bool operator>(MyType lhs, MyType rhs)
   { 
       return ((IComparable)lhs).CompareTo(rhs) > 0;
   }
   
   public static bool operator>=(MyType lhs, MyType rhs)
   { 
       return ((IComparable)lhs).CompareTo(rhs) >= 0;
   }

2. Implement conversion functions only if they actually make sense. Make conversions explicit if the conversion might fail (throw an exception) or if the conversion shouldn’t be abused and it’s only necessary for low level code. Only make conversions implicit if it makes the code clear and easier to use and the conversion cannot fail.

   private MyType(int data)
   { 
       m_data = data;
   }
   
   public static explicit operator MyType(int from)
   { 
       return new MyType(from);
   }
   
   public static implicit operator int(MyType from)
   { 
       return from.m_data;
   }

3. Always implement Object.ToString() to return a significant textual representation.

   public override string ToString()
   { 
       return string.Format("MyType: {0}", m_data);
   }

4. Implement Object.GetHashCode() and Object.Equals() if object equality makes sense. If two objects are equal (Object.Equals() returns true) they should return the same hash code and this value should be immutable during the whole lifecycle of the object. The primary key is usually a good hash code for database objects.
   For reference types implement operators == and != in terms of Object.Equals().

    
   public override int GetHashCode()
   { 
       return m_data.GetHashCode();
   }
   
   public override bool Equals(object obj)
   { 
       // Call base.Equals() only if this class derives from a 
   
       // class that overrides Equals()
   
       if(!base.Equals(obj))
          return false;
   
       if(obj == null)
          return false; 
   
       // Make sure the cast that follows won't fail
   
       if(this.GetType() != obj.GetType())
          return false;
   
       // Call this if m_data is a value type
   
       MyType rhs = (MyType) obj;
       return m_data.Equals(rhs.m_data);  
   
       // Call this if m_data is a reference type
   
       //return Object.Equals(m_data, rhs.m_data); 
   
   }
   
   public static bool operator==(MyType lhs, MyType rhs)
   { 
       if(lhs == null) 
           return false; 
   
       return lhs.Equals(rhs);
   }
   
   public static bool operator!=(MyType lhs, MyType rhs)
   { 
       return !(lhs == rhs);
   }

   For value types, implement Object.Equals() in terms of a type-safe version of Equals() to avoid unnecessary boxing and unboxing.

   public override int GetHashCode()
   { 
       return m_data.GetHashCode();
   }
   
   public override bool Equals(object obj)
   { 
       if(!(obj is MyType))
           return false;
   
       return this.Equals((MyType) obj);
   }
   
   public bool Equals(MyType rhs)
   {
       // Call this if m_data is a value type
   
       return m_data.Equals(rhs.m_data);          
   
       // Call this if m_data is a reference type
   
       //return Object.Equals(m_data, rhs.m_data); 
   
   }
   
   public static bool operator==(MyType lhs, MyType rhs)
   { 
       return lhs.Equals(rhs);
   }
   
   public static bool operator!=(MyType lhs, MyType rhs)
   { 
       return !lhs.Equals(rhs);
   }

5. Enumerations that represent bit masks should have the [Flags] attribute.
   
   
6. All classes and public members should be documented using XML comments. Private members should be documented using normal comments. XML comments should at least include <summary>, <param> and <returns> elements.
   
   
7. If a class is just meant to be a "container" for static methods (has no state), it should declare a private parameter-less constructor so it can’t be instantiated.
   
   
8. All classes should be CLS compliant. Add an [assembly:CLSCompliant(true)] attribute in the AssemblyInfo.cs file. If it is convenient to add a non-CLS compliant public member add a [CLSCompliant(false)] attribute to it.
   
   
9. All implementation details should be declared as private members. If other classes in the same assembly need access, then declare them as internal members. Try to expose as little as possible without sacrificing usability.
   
   
10. strings are immutable objects and always create a new copy for all the mutating operations, which makes it inefficient for assembling strings. StringBuilder is a better choice for this task.
    
    
11. object.MemberWiseClone() provides shallow copying. Implement ICloneable to provide deep copy for classes. ICloneable.Clone() is usually implemented in terms of a copy constructor.

    public MyType(MyType rhs)
    { 
        m_data = rhs.m_data;
    } 
    
    public object Clone()
    { 
        return new MyType(this); 
    }

12. If a class represents a collection of objects implement one or more indexers. Indexers are a special kind of property so they can be read-only, write-only or read-write.

    public object this[int index]
    { 
        get { return m_data[index]; } 
        set { m_data[index] = value; }
    }

13. For a "collection" class to be used in a foreach loop it must implement IEnumerable. IEnumerable.GetEnumerator() returns a class the implements IEnumerator.

    public class MyCollection: IEnumerable
    { 
        public IEnumerator GetEnumerator()
        { 
            return new MyCollectionEnumerator(this);
        }
    }

14. The IEnumerator for a "collection" class is usually implemented in a private class. An enumerator remains valid as long as the collection remains unchanged. If changes are made to the collection, such as adding, modifying or deleting elements, the enumerator is irrecoverably invalidated and the next call to MoveNext or Reset throws an InvalidOperationException. If the collection is modified between MoveNext and Current, Current will return the element that it is set to, even if the enumerator is already invalidated.

    private class MyCollectionEnumerator: IEnumerator
    { 
        public MyCollectionEnumerator(MyCollection col)
        { 
            m_col = col;
            m_lastChanged = col.LastChanged;
        } 
    
        public bool MoveNext()
        { 
            if(m_lastChanged != m_col.LastChanged)
                throw new InvalidOperationException();
    
            if(++m_index >= m_col.Data.Count)
                return false; 
    
            return true;           
        }
    
        public void Reset()
        { 
            if(m_lastChanged != m_col.LastChanged)
                throw new InvalidOperationException();
    
            m_index = -1;
        }
    
        public object Current
        { 
            get { return m_col.Data[m_index]; } 
        } 
    }

15. There is no deterministic destruction in C# (gasp!), which means that the Garbage Collector will eventually destroy the unused objects. When this scenario is not ok, implement IDisposable...

    public class MyClass
    {
        ...
        public ~MyClass()
        {
            Dispose(false);
        }
    
        public void Dispose()
        {
            Dispose(true);
            GC.SuppressFinalize(this);  // Finalization is now unnecessary
    
        }
       
        protected virtual void Dispose(bool disposing)
        {
            if(!m_disposed)
            {
                if(disposing)
                {
                    // Dispose managed resources
    
                }
             
                // Dispose unmanaged resources
    
            }
          
            m_disposed = true;
        }
       
        private bool m_disposed = false;
    }
    

    And use the using statement to dispose resources as soon the object goes out of scope

    using(MyClass c = new MyClass())
    {
        ...
    } // The compiler will call Dispose() on c here

16. There is no way to avoid exceptions handling in .NET. There are several strategies when dealing with exceptions:
    
    
    • Catch the exception and absorb it

      try
      {
          ...
      }
      catch(Exception ex)
      {
          Console.WriteLine("Opps! Something failed: {0}", ex.Message);
      }

    • Ignore the exception and let the caller deal with it if there's no reasonable thing to do.

      public void DivByZero()
      {
          int x = 1 / 0; // Our caller better be ready to deal with this!
      
      }

    • Catch the exception, cleanup and re-throw

      try
      {
          ...
      }
      catch(Exception ex)
      {
          // do some cleanup
      
          throw;
      }

    • Catch the exception, add information and re-throw

      try
      {
          ...
      }
      catch(Exception ex)
      {
          throw new Exception("Something really bad happened!", ex);
      }

17. When catching exceptions, always try to catch the most specific type that you can handle.

    try
    {
        int i = 1 / 0;
    }
    catch(DivideByZeroException ex) // Instead of catch(Exception ex)
    
    {
        ...
    }
    

18. If you need to define your own exceptions, derive from System.ApplicationException, not from System.Exception.
    
    
19. Microsoft's FxCop design diagnostic tool is your friend. Use it regularly to validate your assemblies.
    
    
20. The definite guide for .NET class library designers is .NET Framework Design Guidelines.

Updates