compiling with SDL issues

[eatcomics]

Don't include headers in headers, they're bad...

You do realize that while it is definitely good practice to keep header includes from headers to a minimal, simple class/function prototyping often isn't enough? Consider class A inheriting from class B. Class A REQUIRES Class B to be included, so that the compiler knows how much space to allocate in memory for an instance of class A.

I actually don't prototype classes, just functions. I have no idea why I put that, must not have been paying enough attention

[Live-Dimension]

A header file should be able to compile with it's cpp file left empty.

A header file has absolutely no dependence on a .cpp file whatsoever (unless you're #including a .cpp from within the header). This doesn't make any sense.

What I meant was that something like the following would not compile until you include <string> in the .cpp file (and it would have to be included before including "A.h"). Which is why you should include <string> inside the header instead, unless you can get away with a forward declaration instead. Am I wrong about this?

A.h

Code: Select all

#ifndef A_H
#define A_H

class A {
public:
    void doShit(const std::string& myString);
};

#endif

A.cpp

Code: Select all

#include "A.h"

That's perfectly fine and valid. =]