Not sure why you have two type parameters on the Repository - what is the point?
*Here is the classic example of a .NET Repository using Generics: *
*First, the Repository Interface: *
public interface IRepository<T> where T : class
{
T FindSingle(Expression<Func<T,bool>> predicate);
IQueryable<T> FindAll(); // optional - matter of preference
void Add(T entity);
void Remove(T entity);
}
*Second, the Generic Repository Implementation (EF as the example): *
public abstract class GenericRepository<T> : IRepository<T>
{
private IObjectSet<T> _ObjectSet; // get this in via DI (for example)
public T FindSingle(Expression<T,bool>> predicate)
{
return _ObjectSet.SingleOrDefault(predicate);
}
// you can figure out how to do the other implementation methods
}
*Then, the Specific Repository (you should have one per aggregate root, and also an interface for each specific repository detailing specific methods): *
public EmployeeRepository : GenericRepository<Employee>, IRepository<Employee>
{
// all regular methods (Find, Add, Remove) inherited - make use of them
public Employee FindEmployeeByName(string name)
{
return FindAll().SingleOrDefault(x => x.Name == name);
// or you could do: return FindSingle(x => x.Name == name);
}
}
Usage:
IRepository<Employee> repository = new EmployeeRepository<Employee>();
Don't go out looking to go too crazy with generics - the only one you need is to constrain the Repository to be used by a entity that is encapsulated behind the Repository.
I simply use where T : class.
Other's use where T : IDomainAggregate or similar, to put constraints on the actual type of entity which is allowed.