Creating a Cross Platform Program Using C++ - c++

Question

Let's say I have a simple project in MSVC++ 2010.
All there is in it is main.cpp, its code being something simple like this:
// include macros
#define WIN32
#include <FL/Fl.H>
#include <FL/Fl_Window.H>
#include <FL/fl_ask.H>
// main function
int main()
{
// init window
Fl_Window *window = new Fl_Window(250, 250, "Derp Window");
// show window, run window
window->show();
int result = Fl::run();
fl_message("Goodbye...");
// release pointers
delete window;
// return
return result;
}
It's easy to make a Windows compatible version, all I have to do is set the mode to release and the build it. But, as I have recently found out, the generated .exe file would not work on a Mac or Linux OS. This surprised me because all I am using is plain old c++ and FLTK, which is cross platform.
So, my question is, how would I take this code and compile it in a way that it would work on a Linux OS, and the a Mac OS? Or, is it even possible to keep the same code and compile it in a different way so it works on another OS? If it is not possible, what would I have to change?
PS. The code is pretty straight forward but if you're wondering the #define WIN32 is there because without it, the compiler freaks out about a missing header file, something like "X/X11.h"

Answer 1

The code is compatible, but that doesn't mean the generated binary will be. By and large, this is never the case.
Compilers take your code and translate it into lower-level code that your specific architecture and platform can understand, and Windows is not the same as Mac. This lower-level code has basically nothing to do with C++, FLTK or the compatibility promises of either of them.
The analogy here is that driving a car on the left hand side of the road is, mechanically, the same as driving a car on the right hand side of the road (so let's ignore things like navigation differences and the fact that you're probably steering in different directions), so your knowledge of how to drive a car fits both scenarios identically … but that doesn't mean you can simply plonk your car on the left hand side of the road in France or the United States. You'd cause a pretty gnarly accident. When you apply your knowledge of driving a car to a specific environment you have to fit that around the local rules of the road.
You can re-compile the same code under the target environment, or use a cross-compiler.

Answer 2

If you don't have access to a Linux or Mac OSX computer, you have to cross-compile it. To do this you have to either find a existing cross-compiler, you download the source to e.g. GCC and build it your self. Do some searching from "cross compiler" (or similar) and you will find some easy to follow tutorials.
If you do have access to a Linux or Mac, then just copy the code and build it in that environment. Be careful with Linux through, as different distributions have different versions of some libraries.
And finally, there are environments such as Wine which will allow Windows programs to run on other platforms.

Answer 3

If you use FLTK, you need to include its headers on every platform you compile your program, not just WIN32.
The error message you get about not finding X11/X.h is either caused by missing compiler flags (see 1), prerequisites that need to be included first (see 2), or missing headers (see 3).
Use fltk-config --cflags to get the required compiler flags, fltk-config --libs for the flags you need to pass to the linker.
Doesn't seem to apply here, but you might encounter libraries that require it.
If you use Ubuntu or Debian, install the X headers by executing apt-get install x11proto-core-dev, but that shouldn't be required if you already installed libfltk1.3-dev (or some other version), as it should pull in all the required dependencies. If you use a different distribution, make sure you have the X headers package installed.

Answer 4

The problem is probably the #define WIN32. From what you describe, I would guess that the FLTK library is using this to conditionally compile its headers either for Windows or for Unix (X). So you'd need it defined when compiling for Windows, and undefined when compiling for Unix. The usual way of handling such issues is to add a /DWIN32 option to the compiler invocation when compiling for Windows. You'll probably also have to do something to get the right libraries.
And FWIW: it's perfectly possible to design the library so that the header files work for both systems, without any conditional compilation. And if you want conditional compilation, you should probably use the system's predefined symbols (_WIN32, for example).