WARNING: This code is meant to drive you nuts!

Metaclasses change the way objects are created. That is, instead of simply calling __init__() something else might happen without you knowing it.

They are an ugly, incoherent, intransparent construct that exploits everything Python offers.

Valid Use Cases:

  • writing an ORM like Django models or SQLAlchemy
  • hijacking internal Python logic (e.g. like pytest does)
  • emulating JavaScript-like objects (the Prototype pattern)
  • showing off


  1. run the code
  2. admire what is happening
  3. try to understand what is happening
  4. return to 1

Here is the code

class CrazyMonkeyPack(type):

    def __new__(mcs, name, bases, dict):
        cls = type.__new__(mcs, name, bases, dict)

        def wrapper(*args):
            instance = []
            for i in range(1, args[0]+1):
                monkey = cls(f'monkey #{i}')  # calls __init__
                monkey.state = 'crazy'  # monkey-patches the state attribute
            return instance

        return wrapper

class CrazyMonkeys(metaclass=CrazyMonkeyPack):
    """A self-expanding horde of monkeys"""
    def __init__(self, name): = name

    def __repr__(self):
        return f"<{} ({self.state})>"

monkeys = CrazyMonkeys(3)  # calls __new__
print(monkeys)             # see what happens!

Final Warning

Don't try this at work, unless

  • you have excluded all alternatives
  • you really know what you are doing
  • you have talked to a developer more experienced than you

Dr. Kristian Rother

I am a professional Python trainer, developer and author based in Berlin. I believe everybody can learn programming.

Contact me via:
+49 176 3052 4691


BTW students were really impressed by my storytelling skills I demonstrated in class impromptu and was applausing at the end of the class... its all because of your excellent teaching, thank you!!

Kevin Wong, Data Scientist

See also