I used a Visual Studio Code’s Polyglot notebook and wrote the following in 3 separate cells, in the order indicated below.
Cell 1:
class Z;
Cell 2:
class A
{
public static void f()
{
Console.WriteLine(Z.x);
}
}
Cell 3:
class Z: A
{
public static int x = 42;
}
I then executed the cells in order, and got the following error message.
'Z' does not contain a definition for 'x'
My original question was:
Is it possible, in C#, to write a (possibly abstract) class
Athat has a (possibly static) function with signaturevoid f(B b), whose parameter’s typeBis a (possibly abstract) class that derives fromA(class B: A {...})? If not, why not?
and it was titled Can a C# class use a derived class?
I accepted the answer below based on this formulation. This formulation was later deleted by another user who edited my question, and changed its title.
16
The notebook executes the cells one by one. So when you execute cell 2, Z is nothing but a class, without any notion of x. That’s why you get the error.
It’s not clear what behaviour you actually want. Of course you can just write the entire code into a single cell:
class A
{
public static void f()
{
Console.WriteLine(Z.x);
}
}
class Z: A
{
public static int x = 42;
}
So your problem is not related to inheritance, but to how Polyglot executes the cells. Be aware, though, that your approach is pretty strange. A base-class should never make any assumptions about derived classes, this it it shouldn’t even know that there exist any derived classes in the first place. However that’s more of a conceptual problem, which is way too broad for this questions scope.
1
Yes, it is possible.
Here’s an example that follows exactly what you requested:
using System;
class A
{
public void f(B b)
{
Console.WriteLine("in A.f. b.Prop=" + b.Prop.ToString());
}
}
class B : A
{
public int Prop { get; set; } = 0;
}
class Program
{
static void Main(string[] args)
{
B b1 = new B();
b1.Prop = 1;
B b2 = new B();
b2.Prop = 2;
b1.f(b2);
A a = new A();
a.f(b1);
}
}
Output:
in A.f. b.Prop=2
in A.f. b.Prop=1
Live demo
Note:
Since you asked, A and/or B can also be abstract.
Update:
The questiom was editted after this answer was given, with additional details about the usage of Polyglot notebook.
Regarding this additional information: the problem is related to the way the notebook executes the cells one by one. It is not an issue related to C# in general or the usage of inheritance in particular.
Yes, sure. This is fully valid:
class A
{
public B GetDerived() => new B();
}
class B : A
{
// ...
}
Of course, the two classes need to be in the same assembly, because they need to know each other. The other question is for a relevant use case, which I find hard to imagine. Normally, if A is abstract, you would return A’s (which then need to be derived types of A), and not B’s, since that prevents you from later creating a class C that derives from A and which would replace B.