I know that this is an old question, but one very valid reason to use a class as a namespace (a non-static one at that) is that C# does not support the definition of parametric or generic namespaces. I've written up a blog post on this very topic here: http://tyreejackson.com/generics-net-part5-generic-namespaces/.
The gist of it is, when using generics to abstract huge swaths of boilerplate code, it is sometimes necessary to share multiple generic parameters between related classes and interfaces. The conventional way to do this would be to redefine the generic parameters, constraints and all in each interface and class signature. Over time this can lead to a proliferation of parameters and constraints not to mention constantly having to qualify the related types by forwarding the type parameters from one type to the type arguments of the related type.
Using an outer Generic class and nesting the related types within can dramatically DRY up the code and simplify its abstraction. One can then derive the parametric namespace class with a concrete implementation that supplies all of the concrete details.
Here is a trivial example:
public class Entity
<
TEntity,
TDataObject,
TDataObjectList,
TIBusiness,
TIDataAccess,
TIdKey
>
where TEntity : Entity<TEntity, TDataObject, TDataObjectList, TIBusiness, TIDataAccess, TIdKey>, subclassed
where TDataObject : Entity<TEntity, TDataObject, TDataObjectList, TIBusiness, TIDataAccess, TIdKey>.BaseDataObject, subclassed
where TDataObjectList : Entity<TEntity, TDataObject, TDataObjectList, TIBusiness, TIDataAccess, TIdKey>.BaseDataObjectList, subclassed
where TIBusiness : Entity<TEntity, TDataObject, TDataObjectList, TIBusiness, TIDataAccess, TIdKey>.IBaseBusiness
where TIDataAccess : Entity<TEntity, TDataObject, TDataObjectList, TIBusiness, TIDataAccess, TIdKey>.IBaseDataAccess
{
public class BaseDataObject
{
public TIdKey Id { get; set; }
}
public class BaseDataObjectList : Collection<TDataObject> {}
public interface IBaseBusiness
{
TDataObject LoadById(TIdKey id);
TDataObjectList LoadAll();
void Save(TDataObject item);
void Save(TDataObjectList items);
void DeleteById(TIdKey id);
bool Validate(TDataObject item);
bool Validate(TDataObjectList items);
}
public interface IBaseDataAccess
{
TDataObject LoadById(TIdKey id);
TDataObjectList LoadAll();
void Save(TDataObject item);
void Save(TDataObjectList items);
void DeleteById(TIdKey id);
}
}
Used like this:
public class User
:
Entity
<
User,
User.DataObject,
User.DataObjectList,
User.IBusiness,
User.IDataAccess,
Guid
>
{
public class DataObject : BaseDataObject
{
public string FirstName { get; set; }
public string LastName { get; set; }
}
public class DataObjectList : BaseDataObjectList {}
public interface IBusiness : IBaseBusiness
{
void DeactivateUserById(Guid id);
}
public interface IDataAcccess : IBaseDataAccess {}
}
Consume the derivatives like this:
public class EntityConsumer
{
private User.IBusiness userBusiness;
private Permission.IBusiness permissionBusiness;
public EntityConsumer(User.IBusiness userBusiness, Permission.IBusiness permissionBusiness) { /* assign dependencies */ }
public void ConsumeEntities()
{
var users = new User.DataObjectList();
var permissions = this.permissionBusiness.LoadAll();
users.Add
(new User.DataObject()
{
// Assign property values
});
this.userBusiness.Save(users);
}
}
The benefit to writing the types this way is added type safety and less casting of types in the abstract classes. Its the equivalent of ArrayList vs List<T> on a larger scale.
