Type Casting

I recently purchased the book C++ for C Programmers, 3rd Ed. by Ira Pohl. It was a $40 book for only $10. It's telling me that instead of something like

Code:

---

int i = 3;
cout << 2/double(i);

---

I should use the keyword static_cast like so

Code:

---

int i = 3;
cout << 2/static_cast<double>(i);

---

I don't really understand why this is. Of course, I'm just judging this on the fact that the latter method is longer to type. Please tell me why the latter method is perferable. Thanks.

It is a part of the new standard C++ set fourth by ANSI and ISO. Also you can use reinterpret_cast<> as well. They are prefered than doing the type() stye of C. Just a "Cleaner" way of doing it. It also allows the compiler to do type checking.

Quote:

---

Originally posted by velius
Also you can use reinterpret_cast<> as well.

---

My book really just glosses over this issue of casting. What makes reinterpret_cast different from static_cast?

Quote:

---

They are prefered than doing the type() stye of C. Just a "Cleaner" way of doing it. It also allows the compiler to do type checking.

---

How is it "cleaner"? And I don't understand what you mean by that last sentence. How was the language lacking before these new reserved words? I didn't know of a way to screw up a program by casting to an undefined type. I thought that an error like that would be caught at compile time.

Thanks for your help in advance.

A static_cast converts between related types, for instance int to float. A reinterpret_cast converts between to unrelated types, for instance an integral type to a pointer.

There is also a dynamic_cast, which allows you to cast from a polymorphic virtual base class to a derived class.

Quote:

---

How is it "cleaner"? And I don't understand what you mean by that last sentence. How was the language lacking before these new reserved words? I didn't know of a way to screw up a program by casting to an undefined type. I thought that an error like that would be caught at compile time.

---

The old C style cast operator could obscure some errors, particularly when you are dealing with polymorphism. The new keywords are also more expressive. You have a much better idea of what the programmer intended when they used the cast. I guess it also fits into C++'s stronger typechecking paradigm too.

There is also a const_cast which casts away the constness of a variable.

Avoid reinterpret_cast if at all possible. Essentially, it just takes the bit pattern of the given variable and interprets that bit pattern as the type specified. Very bad in terms of type-correctness, and can be avoided most of the time.

Thank you for all of your help thus far. I have another question. I know how to overload an operator for explicit type casting like double(fraction &f). Now, with this new way of casting, how do I do that?

When you do a static_cast, the compiler will search for the conversion operator (defined as normal) in your class to that particular type. The other casts are all defined in such a way that it would make no sense to be able to overload them.