ramsey, 16 days ago

@Crell I don’t even know how you do that in Python.

Crell, 16 days ago

@ramsey

1. Python has types, they're just elided. Basically just documentation, but use 'em.
2. Python has @dataclass definitions, and an extension library called Pydantic that's pretty nice.
3. Lean hard on "if you have a dict where you know the types in advance, make it a class." (Python version of "Never* Use Arrays".)

It's... more capable in this regard than I expected, honestly.

ramsey, 16 days ago

@Crell Since they’re elided, does the community have a universally agreed upon analyzer that everyone uses, like with TypeScript?

In other words, if I use types in my library, can I be certain they will prevent users of my library from making type mistakes with it?

Crell, 16 days ago

@ramsey I'm not sure. I'm just relying on my IDE. 🤷‍♂️

Girgias, 16 days ago

@ramsey @Crell The only one I know is FB's one, Pysa, in which they took the lesson learned from Zoncolan (like taint analysis, whose white paper @psalm used).

But I'm guessing there are some competing one.

https://engineering.fb.com/2020/08/07/security/pysa/