How can I hide a base class public property in the derived class

I'm going to attempt to explain with examples why this is a bad idea, rather than using cryptic terms.

Your proposal would be to have code that looks like this:

public class Base
{
    public int Item1 { get; set; }
    public int Item2 { get; set; }
}


public class WithHidden : Base
{
    hide Item1; // Assuming some new feature "hide" in C#
}

public class WithoutHidden : Base { }

This would then make the following code invalid:

WithHidden a = new WithHidden();
a.Item1 = 10; // Invalid - cannot access property Item1
int i = a.Item1; // Invalid - cannot access property Item1

And that would be just what you wanted. However, suppose we now have the following code:

Base withHidden = new WithHidden();
Base withoutHidden = new WithoutHidden();

SetItem1(withHidden);
SetItem1(withoutHidden);

public void SetItem1(Base base)
{
    base.Item1 = 10;
}

The compiler doesn't know what runtime type the argument base in SetItem1 will be, only that it is at least of type Base (or some type derived from Base, but it can't tell which -- it may be obvious looking at the code snippet, but more complex scenarios make it practically impossible).

So the compiler will not, in a large percentage of the cases, be able to give a compiler error that Item1 is in fact inaccessible. So that leaves the possibility of a runtime check. When you try and set Item1 on an object which is in fact of type WithHidden it would throw an exception.

Now accessing any member, any property on any non-sealed class (which is most of them) may throw an exception because it was actually a derived class which hid the member. Any library which exposes any non-sealed types would have to write defensive code when accessing any member just because someone may have hidden it.

A potential solution to this is to write the feature such that only members which declare themselves hideable can be hidden. The compiler would then disallow any access to the hidden member on variables of that type (compile time), and also include runtime checks so that a FieldAccessException is thrown if it is cast to the base type and tried to be accessed from that (runtime).

But even if the C# developers did go to the huge trouble and expense of this feature (remember, features are expensive, especially in language design) defensive code still has to be written to avoid the problems of potential FieldAccessExceptions being thrown, so what advantage over reorganising your inheritance hierarchy have you gained? With the new member hiding feature there would be a huge number of potential places for bugs to creep into your application and libraries, increasing development and testing time.

What you want to do goes directly against the grain of OO, you can't 'unpublish' members as this violates the substitution principle. You have to refactor this into something else.

Vadim's response reminded me of how MS achieve this in the Framework in certain places. The general strategy is to hide the member from Intellisense using the EditorBrowsable attribute. (N.B. This only hides it if it is in another assembly) Whilst it does not stop anyone from using the attribute, and they can see it if they cast to the base type (see my previous explination) it makes it far less discoverable as it doesn't appear in Intellisense and keeps the interface of the class clean.

It should be used sparingly though, only when other options like restructuring the inheritance hierarchy would make it a lot more complex. It's a last resort rather than the first solution to think of.

If you use an interface instead of a base class for defining the property, you could implement the property explicitly. The would require an explicit cast to the interface to use the property.

public interface IMyInterface
{
    string Name { get; set; }
}

public class MyClass : IMyInterface
{

    string IMyInterface.Name { get; set; }

}

You can find more out here.