C / C++ / MFC

25-Jul-07 23:52

What is the correct sintax to solve this specialization problem?
I'm trying but for now I have only errors while compiling or linking.

template<typename T> class CMyClass{
   ...
   int    Foo(){    //First  specialization: if T is int
   }
   double Foo(){    //Second specialization: if T is double
   }
   void   Foo(){    //elseway
   }
   ...
}

Thanks to everyone

Russell

CPallini25-Jul-07 23:57

25-Jul-07 23:57

As far as I can remember you cannot declare function templates based on return value types.
Smile | :)

If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.

Russell^'26-Jul-07 0:03

26-Jul-07 0:03

ops, really?
Of course, I don't want to overload the functions: I only want to specify completely different functions in that 2 cases (T=int and T=double) using specialization (if it is possible).
Ciao

Russell

CPallini26-Jul-07 0:26

26-Jul-07 0:26

If I remember well, you cannot do such specializations, i.e. functions having the same name, the same number and type of arguments but different return types.
Smile | :)

If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.

CPallini26-Jul-07 0:33

26-Jul-07 0:33

For instance, you cannot do the following:

int foo(){return 1;}
double foo(){return 1.;}

while the code snippet below:

int foo(int i){return 1;}
double foo(char c){return 1.;}

compiles fine.

~~If I remeber well this is related to the C++ Mangling scheme (return parameters are not used when mangling).~~

[added] There is a better argument: how the hell the compiler will distinguish which flavour of the function you're calling?
Smile | :)

If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.

Russell^'26-Jul-07 0:43

26-Jul-07 0:43

probally you are rjght!
I was only hoping that templates could be very very powerful, but probally doesn't.
tnx

Russell

CPallini26-Jul-07 1:00

26-Jul-07 1:00

In fact they are very powerful. But, anyway, you cannot do that.
Smile | :)

If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.

Russell^'26-Jul-07 0:54

26-Jul-07 0:54

CPallini wrote:
how the hell the compiler will distinguish which flavour of the function you're calling?

mmm...but I think that you are not understanding that I'm using templates , Am I rigth?
Templates are elaborated during pre-compiler step! There you can write different member definitions according to some conditions, like #if #else ...
The compiler comes after that processing.
Smile | :)

Russell

CPallini26-Jul-07 0:59

26-Jul-07 0:59

I understand, understand...
But template specializations are eventaully standard C++ functions (or methods) hence what you cannot do with the latter you cannot do with the former.
Smile | :)

_Russell_ wrote:
Templates are elaborated during pre-compiler step! There you can write different member definitions according to some conditions, like #if #else ...
The compiler comes after that processing

I don't agree: templates, AFAIK, are not pre-processor stuff.
Smile | :)

If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.

Russell^'26-Jul-07 2:10

26-Jul-07 2:10

Ok,
I'll surrender.
Unsure | :~

Russell