Protocols in Python (til.simonwillison.net)

submitted 1 year ago by rglullis@communick.news to c/python@programming.dev

20 comments fedilink hide all child comments

you are viewing a single comment's thread
view the rest of the comments

[-] lysdexic@programming.dev 1 points 1 year ago

In practice, Protocols are a way to make “superclasses” that you can never add features to (for example, readinto despite being critical for performance is utterly broken in Python).

Not really, and this interpretation is oblivious to the concept of protocols and misses the whole point of them.

The point of a protocol is to specify that a type supports a specific set of methods with specific signatures, aka duck typing, and provide the necessary and sufficient infrastructure to check if objects comply with a protocol and throw an error in case it doesn't.

Also, protocol classes can be inherited, and protocols can be extended.

https://peps.python.org/pep-0544/

[-] o11c@programming.dev 1 points 1 year ago

Then - ignoring dunders that have weird rules - what, pray tell, is the point of protocols, other than backward compatibility with historical fragile ducks (at the cost of future backwards compatibility)? Why are people afraid of using real base classes?

The fact that it is possible to subclass a Protocol is useless since you can't enforce subclassing, which is necessary for maintainable software refactoring, unless it's a purely internal interface (in which case the Union approach is probably still better).

That PEP link includes broken examples so it's really not worth much as a reference.

(for that matter, the Sequence interface is also broken in Python, in case you need another historical example of why protocols are a bad idea).

[-] lysdexic@programming.dev 1 points 1 year ago* (last edited 1 year ago)

(...) what, pray tell, is the point of protocols, other than backward compatibility with historical fragile ducks (at the cost of future backwards compatibility)?

Got both of those wrong. The point of protocols is to have a way to validate duck typing errors by adding a single definition of a duck. This is not something that only applies to backwards compatible, nor does it affects backwards compatibility.

Why are people afraid of using real base classes?

You're missing the whole point of prototypes. The point of a prototype is that you want duck typing, not inheritance. Those are entirely different things, and with prototypes you only need to specify a single prototype and use it in a function definition, and it automatically validates each and any object passed to it without having to touch it's definition or add a base class.

[-] o11c@programming.dev 1 points 1 year ago

That still doesn't explain why duck typing is ever a thing beyond "I'm too lazy to write extends BaseClass". There's simply no reason to want it.

[-] lysdexic@programming.dev 1 points 1 year ago

I already explained it to you: protocols apply to types you do not own or control, let alone can specify a base class.

You just specify a protocol, specify that a function requires it, and afterwards you can pass anything to it as-is and you'll be able to validate your calls.

[-] o11c@programming.dev 1 points 1 year ago

and I already explained that Union is a thing.

[-] lysdexic@programming.dev 1 points 1 year ago* (last edited 1 year ago)

I'm not sure you understand that what a union does or does not do is completely irrelevant and besides the point. Python's protocols add support for structural subtyping, and enable both runtime and build-time type checks without requiring major code changes. Don't you understand what that means?