I've been working with Python for a few years now, mainly focusing on machine learning and deep learning, so my foundational understanding of Python as a programming language isn't as deep as I'd like it to be. Recently, I started reading **Python Distilled** (definitely a must-read!), and I'm currently on chapter 4, which discusses objects, types, and protocols. However, I find myself a bit stuck on the concept of **Protocols**. I grasp what the book explains, especially regarding methods like __add__() and __radd__(), which dictate how the add() function operates, but I'm uncertain about how protocols can actually improve my coding practices. The book doesn't dive into practical uses of protocols. Am I missing something? Any suggestions for resources like pages, blogs, or chapters that delve deeper into understanding Python protocols?
2 Answers
Yes, operator overloading in Python is often what we refer to as using protocols. Essentially, if you define how your custom object should behave with these operations, you're effectively using protocols! It’s a neat way to make your custom classes behave like built-in types.
Protocols in Python allow you to define how your classes interact with built-in operations, like when you implement custom behavior for operators like + or when they are indexed. For instance, when you create your own class, you can specify how it handles addition or iteration using dunder methods. Consider NumPy arrays: when you add two arrays, the addition is element-wise due to these methods. You might not use this often, but should you build a custom class for something like an AI model chunk, you might want to overload operators to allow combining models seamlessly with symbols like +. Don't worry if it feels complex now; it's a powerful concept that comes in handy as you advance!
Thanks mate, that really clears things up! I now also see why Andrej Karpathy discussed protocols in his neural networks video - it makes so much more sense!
Exactly, I’m starting to see how this ties into practical applications. Thanks for the insights!