=?windows-1252?Q?Abstract_classes_=93interfaces=94_and_const_methods?=

or messengers, they don?t do the job, but they ask someone else to do.
The same for member functions pointers.
Considering this, I don?t want to add into the message signature a
contract about the implementers if they will change or not their data
members.
In the object implementation I know if the method is const or not, and
I am happy to define this properly. I can use these members direclty
and the const will be usefull.

So what is the problem?
The problem is that virtual functions signature will depend on the
const.

For instance;
struct Interface {
   virtual int GetSize() = 0;
};

struct Object : public Interface {
     virtual int GetSize() const { return 1; }
}; // Error cannot instantiate abstract class

The solution adopted in COM is that the const is left out in the
object implementation.

I was thinking... I can use mutable in data members to say ?I don?t
care about const?.
So, should we have mutable for functions?

struct Interface {
   virtual int GetSize() mutable = 0;
};

struct Object : public Interface {
     virtual int GetSize() const { return 1; } //ok
};

struct Object2 : public Interface {
     virtual int GetSize() { return 1; } //ok
};

struct Object : public Interface {
     virtual int GetSize() { return 1; } //error
     virtual int GetSize() const { return 1; }
};

Is there a better solution ?

--
      [ See http://www.gotw.ca/resources/clcm.htm for info about ]
      [ comp.lang.c++.moderated. First time posters: Do this! ]