在运行时进行编译，并在C ++中#include自定义标头

Question

I need to compile and link some code at runtime. I am using the approach suggested here: https://stackoverflow.com/a/10565120/3038460

Basically I am writing my code into a .cpp file and then compiling with

/usr/bin/g++ -shared mysource.cpp -o libname.so

Everything works fine while I only #include headers from standard libraries. But what if I need to use a custom class within the 'dynamic' code? How can include my own header? The .cpp file is temporary stored in the same location of my binary file and this might be different than the location of my source files. Is there a way to know at runtime where the original source code was located? I doubt.

Moreover I would like my code to work even if the original source code is not available.

To clarify, mysource.cpp might look like this:

#include <stdlib.h>
#include <vector>
#include "myheader.h"     <---- how can g++ find this file? At runtime,
                                when I create mysource.cpp, I have no idea
                                where myheader.h is located

void f(){
  // Code
}

What's the best solution to solve this problem?

Answer 1

Generate the contents of that header file into the generated source file.

Technically, an #include directive is just a directive to include the full contents of a header file in the body of a source file (specifically, at the point where you included it) at compile time. You can thus skip the middleman and generate the contents of that header file into your source file before compiling it instead of an #include