In Python's multiprocessing package, why is there multiprocessing.Pool and multiprocessing.pool.Pool?

Question

While learning about Python's multiprocessing package (for Python 3.4 ), I noticed multiprocessing.Pool is defined in the Class BaseContext in context.py . This definition is

   def Pool(self, processes=None, initializer=None, initargs=(),
         maxtasksperchild=None):
    '''Returns a process pool object'''
    from .pool import Pool
    return Pool(processes, initializer, initargs, maxtasksperchild,
                context=self.get_context())

Thus, it ends up calling multiprocessing.pool.Pool defined in pool.py .

Why does the multiprocessing package define both multiprocessing.Pool and multiprocessing.pool.Pool ?

Answer 1

context.py contains OS-dependent code. The Pool and many other values are defined differently depending on whether sys.platform == 'win32' or not.

The pool.py module contains code related the creation of a Pool given a context .

By organizing the code in this way, the developer manages to write pool.py in an OS-agnostic way. There are no if sys.platform ... statements in pool.py, for example.

---

The __init__.py contains:

globals().update((name, getattr(context._default_context, name))
                 for name in context._default_context.__all__)
__all__ = context._default_context.__all__

which copies values from context._default_context into the multiprocessing packages' namespace. This is how multiprocessing.Pool gets defined.

As a user of the multiprocessing package, you are expected to access the Pool via multiprocessing.Pool , although it is possible to use multiprocessing.pool.Pool too.