View mode: basic / threaded / horizontal-split · Log in · Help
February 01, 2005
Implicit casting issue
The following C++ code correctly compiles and prints "C1":

  #include <cstdio>
  class C0{};
  class C1 : public C0{};
  class C2 : public C1{};
  class V{
  public:
    void v(C0& i){printf("C0");}
    void v(C1& i){printf("C1");}
  };
  int main(){
    V i;	
    C2 j;
    i.v(j);
    return 0;
  }

Here's the equivalent D version:

  class C0{}
  class C1 : C0{}
  class C2 : C1{}
  class V{
    void v(C0 i){printf("C0");}
    void v(C1 i){printf("C1");}
  }
  int main(){
    V i = new V;
    C2 j = new C2;
    i.v(j);
    return 0;
  }

... which does *not* compile.  DMD complains:
"function test.V.v overloads void(C0 i) and void(C1 i) both match 
argument list for v"

Why doesn't DMD implicitly cast C2 to C1 just like C++ compilers do?
Is this intended or a bug?
February 01, 2005
Re: Implicit casting issue
zwang wrote:

[...]
> ... which does *not* compile.  DMD complains:
> "function test.V.v overloads void(C0 i) and void(C1 i) both match 
> argument list for v"
> 
> Why doesn't DMD implicitly cast C2 to C1 just like C++ compilers
> do? Is this intended or a bug?

Congratulions for not reading the docs:
http://www.digitalmars.com/d/function.html#overloading

| In D, function overloading is simple. It matches exactly, it
| matches with implicit conversions, or it does not match. If there
| is more than one match, it is an error.

C2 can be implicitely casted to C1 and to C0. So there is more than one 
match and therefore an error.

-manfred
February 02, 2005
Re: Implicit casting issue
Thanks! I guess I'm still writing code in a c++ mindset.

Manfred Nowak wrote:
> Congratulions for not reading the docs:
> http://www.digitalmars.com/d/function.html#overloading
> 
> | In D, function overloading is simple. It matches exactly, it
> | matches with implicit conversions, or it does not match. If there
> | is more than one match, it is an error.
> 
> C2 can be implicitely casted to C1 and to C0. So there is more than one 
> match and therefore an error.
> 
> -manfred
Top | Discussion index | About this forum | D home