I'm attempting to use the PIMPL idiom. This is my public header file in include/foo.h:

class FooPrivate;

class Foo {
public:
        Foo();
        ~Foo();

private:
        FooPrivate* p_impl;
};

The private class is defined in: src/foo_private.h and used throughout src/*.cpp:

class FooPrivate {
public:
        ...
private:
        ...
};

My question is, what are the reasons to distinguish public: or private: members and functions in FooPrivate since the entire class is private? This is from the perspective of distributing foo.h and libfoo.so to end users who will use it my library.

I figure that paying attention to public and private members would enforce correct coding practices, i.e. making me think about exposing an API to use around my private code.

On the other hand, by having all public members in FooPrivate, I would have an easier time writing unit tests for it, and also pass it around more easily in my code without having to think about APIs, accessors, etc. Also, the whole thing is private so the end user won't be impacted by this sloppiness, right?

Which would you prefer?

Note that I'm still a beginner at C++ so I could be wrong about my assumptions.