Integrating multiple third-party libraries in a C++ project often leads to a common yet tricky issue: macro conflicts. Since macros are processed by the preprocessor and are not restricted by namespaces or scopes, conflicts are easy to run into. This article aims to share several strategies to resolve such conflicts effectively.

Prefixing Macros with Library Names

A best practice recommendation is to prefix your macros with the library name or a unique identifier to reduce the likelihood of conflicts. For instance, if your library is called MyLib, then the macro should be defined as MYLIB_SOME_MACRO instead of just SOME_MACRO. This approach’s simplicity and directness are its main advantages, but it requires that you can modify the library source code.

Pros: Easy to implement; enhances code readability and maintainability.

Cons: Only applicable to code bases you can modify.

In cases where modifying third-party libraries isn’t an option—a scenario all too familiar to developers—a viable workaround involves the creation of a wrapper header file, commonly referred to as fwd.h. This file serves as a conduit for including the actual library headers, with a strategic implementation of push_macro and pop_macro directives. These directives are instrumental in preserving and restoring macro definitions that might otherwise clash, and are universally supported by the three major C++ compilers.

Consider a typical scenario where two libraries, A and B, both define CHECK macro - a frequent source of conflict in my experience. In this case, if you prefer to retain B’s definition of CHECK, you would encapsulate A’s headers within A_fwd.h as follows:

// A_fwd.h
#pragma once

#pragma push_macro("CHECK")
#undef CHECK

#include "A.h" // And other A's headers

#pragma pop_macro("CHECK")

Whenever your project requires access to library A’s API, simply include A_fwd.h instead of directly including A’s headers. This approach not only circumvents macro conflicts with elegance but also ensures seamless integration of multiple libraries, even when they share macro names.

Pros: No need to modify third-party library code; provides a flexible isolation mechanism.

Cons: Requires additional maintenance; may obscure some problems leaking from the library.

PImpl Idiom

The PImpl Idiom presents a solution only in very limited cases. We still use our A/B’s example. This trick will shine when B’s CHECK is essential within a single class, and the class’s implementation is independent of A. This technique allows for the encapsulation of a class’s implementation details, minimizing header dependencies and thereby isolating potential macro conflicts to the implementation side.

// MyClass.h - Interface using PImpl idiom
#include <memory>

class MyClassImpl; // Forward declaration of the implementation class

class MyClass {
    ~MyClass(); // Destructor must be defined where MyClassImpl is fully defined, typically in the .cpp file
    // Other public interface methods

    std::unique_ptr<MyClassImpl> pImpl;

// MyClassImpl.h
#include "B.h" // Include library B's header where CHECK macro is defined

class MyClassImpl {
    // Implementation details that utilize B's CHECK macro

#include "MyClass.h"
#include "MyClassImpl.h" // Include the PImpl implementation details

MyClass::MyClass() : pImpl(new MyClassImpl()) {}

This arrangement ensures that the macro conflict is neatly avoided by confining the inclusion of conflicting headers to the source file of the PImpl class, rather than the original translation unit.

Pros: Reduces compile-time dependencies; improves code encapsulation and isolation; applicable to unmodifiable code.

Cons: Limited applications; increases implementation complexity; might introduce a slight performance overhead.


Macro conflicts in C++ can be a hurdle, but with the right strategies, they are manageable. Each method has its context where it shines, offering a blend of maintainability, readability, and encapsulation. Experiment with these strategies in your projects to find the best fit, and keep your C++ codebase healthy and robust.