Understanding Python: Metaclasses 

Thank you and please let me know if there is a topic you'd like me to make a video on. As for my release cadence, my job keeps me quite busy so I'm currently releasing one video a month.

Thank you, you're much too kind. That's a fantastic question, bringing together those two pieces! For abstract methods, that happens in __new__ which happens after __call__. With that in mind, you may be able to exploit that behavior for even deeper checks than just abstract methods. Here's a quick gist if you wanted to play around with your idea. It is a modification of the NoisyMeta class. gist.github.com/JacobCallahan/c81e4d7572b2007372a77608e00899f8

 ah I probably missed that part, so the __new__ method is also called when an instance is created not just when the class is created?

Be careful to remember which __new__ is being called because there are 2. The first is the one on the class' metaclass that created the class object. The second is the one on the class object that creates the class instance. In your question, __call__ is ran after the first (because the class object must exist before it can be called) but before the second. __call__ -> __new__ -> __init__

​@@JakeCallahan sorry I'm a lil bit confused, when we create the class MyClass with meta class MyMetaClass the following method will be executed : __prepare__(method from MyMetaClass) -> __new__(method from MyMetaClass) -> (execute MyClass body) -> __init__(method from MyMetaClass) all these methods are from the metaclass MyMetaClass is this correct? Then when we create instance of MyClass such that my_class_instance = MyClass() the following will happen: __call__(method from the MyMetaClass) -> __new__(method from MyClass) -> __init__(method from MyClass)

__new__ creates the class instance that is then passed in to __init__. This instance object is what __init__ accepts most commonly as "self". You're correct that in the third example, I didn't define a __new__ method, so it used Python's default implementation.

Great catch! The reason this still works, and wouldn't if the inverse were true, is that Python is passing the meta class as the first positional argument. So while this is decorated with staticmethod, there is no real effect on the behavior since Python itself is the consumer. We can verify this by looking at the output at 18:09

@@JakeCallahan yep, great channel!!

While not a full notebook, I Python is what Jupyter uses under the hood. You can do the same as you see me doing by installing IPython with pip, then running from your terminal. pip install IPython Then the command is just: ipython

@@JakeCallahan thank you