Exact difference between overriding and hiding
Take a look at this answer to a different question by Eric Lippert.
To paraphrase (to the limits of my comprehension), these methods go into "slots". A
has two slots: one for Test1
and one for Test2
.
Since A.Test1
is marked as virtual
and B.Test1
is marked as override
, B
's implementation of Test1
does not create its own slot but overwrites A
's implementation. Whether you treat an instance of B
as a B
or cast it to an A
, the same implementation is in that slot, so you always get the result of B.Test1
.
By contrast, since B.Test2
is marked new
, it creates its own new slot. (As it would if it wasn't marked new
but was given a different name.) A
's implementation of Test2
is still "there" in its own slot; it's been hidden rather than overwritten. If you treat an instance of B
as a B
, you get B.Test2
; if you cast it to an A
, you can't see the new slot, and A.Test2
gets called.
Already answered at here
Overriding is the definition of multiple possible implementations of the same method signature, such that the implementation is determined by the runtime type of the zeroth argument (generally identified by the name this in C#).
Hiding is the definition of a method in a derived type with a signature identical to that in one of its base types without overriding.
The practical difference between overriding and hiding is as follows:
Hiding is for all other members (static methods , instance members, static members). It is based on the early binding . More clearly , the method or member to be called or used is decided during compile time.
•If a method is overridden, the implementation to call is based on the run-time type of the argument this. •If a method is simply hidden, the implementation to call is based on the compile-time type of the argument this.
Here are some samples : Example # 1. and Example # 2
To add to @Rawling's answer, practical examples could be shown using an example such as this:
class Base
{
// base property
public virtual string Name
{
get { return "Base"; }
}
}
class Overriden : Base
{
// overriden property
public override string Name
{
get { return "Overriden"; }
}
}
class New : Base
{
// new property, hides the base property
public new string Name
{
get { return "New"; }
}
}
1. Overriding
In case of the overriden property, base class' virtual method's slot is replaced by a different implementation. Compiler sees the method as virtual, and must resolve its implementation during run-time using the object's virtual table.
{
Base b = new Base();
Console.WriteLine(b.Name); // prints "Base"
b = new Overriden();
// Base.Name is virtual, so the vtable determines its implementation
Console.WriteLine(b.Name); // prints "Overriden"
Overriden o = new Overriden();
// Overriden.Name is virtual, so the vtable determines its implementation
Console.WriteLine(o.Name); // prints "Overriden"
}
2. Hiding
When a method or a property is hidden using the new
keyword, the compiler creates a new non-virtual method for the derived class only; base class' method remains untouched.
If the type of the variable is Base
(i.e. only contains the virtual method), its implementation will be resolved through the vtable. If the type of the variable is New
, then the non-virtual method or property will be invoked.
{
Base b = new Base();
Console.WriteLine(b.Name); // prints "Base"
b = new New();
// type of `b` variable is `Base`, and `Base.Name` is virtual,
// so compiler resolves its implementation through the virtual table
Console.WriteLine(b.Name); // prints "Base"
New n = new New();
// type of `n` variable is `New`, and `New.Name` is not virtual,
// so compiler sees `n.Name` as a completely different property
Console.WriteLine(n.Name); // prints "New"
}
3. Summary
If a part of your code accepts the base type, it will always use the virtual table during run-time. For most OOP scenarios, this means that marking a method as new
is very similar to giving it a completely different name.
4. Object sizes after instantiation
Note that instantiating any of these types doesn't create a copy of the virtual table. Each .NET object has a couple of bytes of header and a pointer to the virtual table of table of its type (class
).
Regarding the new
property (the one which is not virtual), it is basically compiled as a static method with thiscall semantics, meaning that it also doesn't add anything to the size of the instance in memory.