C#

George_George wrote:

Means a genetics type defined in one assembly, but used/referred from another assembly? So, C# can do binary level (IL) reference, but C++ could only do source code level (header file) reference?

George_George wrote:

Therefore, the C# compiler forces you to declare constraints" -- we can define a generics type without defining any constraints, right?

In what situations, do we have to define constraints when define a generics type?

class G<T>
{
   T t;
   public G(T t)
   {
     this.t = t;
   }

   public void Method()
   {
      t.SayHello();
   }
}

interface IHello
{
   void SayHello();
}

class Hello : IHello
{
   public void SayHello() { Console.WriteLine("Hello"); }
}

static void Main()
{
   Hello h = new Hello();
   G<Hello> g = new G<Hello>(h);
   g.Method();
}

class G<T> where T : IHello
{...}

George_George wrote:

I think the points from the document is, we can not use open constructed type, but only can use close constructed type as type parameter to generics, right? I think I already provide the open constructed type -- List<generic2><t>> gen1, but it could still compile.

template <typename T, template <typename> Container>
class A
{

}

George_George wrote:

Why you stress the point that a different assembly?

George_George wrote:

1.

I want to confirm that the answer to my question #3 is, C# is using constraints to check whether a type is supported or not, but C++ will generate the specific type to check whether the operation is supported or not, right?

George_George wrote:

But from your sample below, I learned more. It seems C# compiler will check more than just the constraints, in your sample, without constraints, it will parse the invoked method to see whether it is supported, right?

<br />
public void Compress(Stream source, Stream target)<br />