Python type hints are probably "worth it" in the large for me
Chris Siebenmann finds Python type hints beneficial for code clarity and maintainability. He prefers NamedTuples and type aliases, emphasizing overall code quality improvement rather than mere compliance with mypy.
Read original article
Chris Siebenmann reflects on his experience with Python type hints, ultimately concluding that they are beneficial despite initial challenges. After a first attempt that left him uncertain, he reintroduced type hints in a more effective manner, opting for NamedTuples and type aliases instead of NewType(). This approach not only clarified the code but also prompted him to write more comments, enhancing future maintainability. He noted that accessing tuple elements by name improved code readability, and he made a necessary adjustment from 'int' to 'float' for accuracy. Siebenmann emphasizes that the goal of adding type hints should not merely be to silence mypy but to facilitate a broader code cleanup and improvement process. He expresses a desire to incorporate basic type hints during initial coding in the future, contingent on better integration with his GNU Emacs setup.
- Python type hints can enhance code clarity and maintainability.
- Using NamedTuples and type aliases can simplify type hinting.
- Adding type hints encourages better documentation through comments.
- The goal should be overall code quality improvement, not just compliance with mypy.
- Future integration of type hints during initial coding is a potential goal.
Related
The human typewriter, or why optimizing for typing is short-sighted
The article highlights the drawbacks of optimizing code for typing speed, advocating for readability and clarity over efficiency, as obfuscated code increases mental strain and confusion among developers.
Typing Lists and Tuples in Elixir
Elixir's new type system introduces sound gradual typing to prevent runtime errors, enhancing list and tuple handling while requiring developers to prove type adherence, potentially increasing boilerplate code.
TypedDicts are better than you think
TypedDicts in Python 3.8 enhance type annotations for dictionaries, allowing optional fields, improving function signatures, and offering better type safety. Upcoming PEPs will introduce extra and read-only item features.
Don't return named tuples in new APIs
The article argues against using named tuples in new APIs unless updating existing tuple-returning APIs, citing complexity and recommending alternatives like dataclasses for better clarity and usability.
Python type hints may not be not for me in practice
Chris Siebenmann expresses skepticism about Python's type hints for small tasks, finding them cumbersome. He prefers adding them post-stabilization for documentation, envisioning a system allowing distinct types without upfront details.
1 commentsRelated
The human typewriter, or why optimizing for typing is short-sighted
The article highlights the drawbacks of optimizing code for typing speed, advocating for readability and clarity over efficiency, as obfuscated code increases mental strain and confusion among developers.
Typing Lists and Tuples in Elixir
Elixir's new type system introduces sound gradual typing to prevent runtime errors, enhancing list and tuple handling while requiring developers to prove type adherence, potentially increasing boilerplate code.
TypedDicts are better than you think
TypedDicts in Python 3.8 enhance type annotations for dictionaries, allowing optional fields, improving function signatures, and offering better type safety. Upcoming PEPs will introduce extra and read-only item features.
Don't return named tuples in new APIs
The article argues against using named tuples in new APIs unless updating existing tuple-returning APIs, citing complexity and recommending alternatives like dataclasses for better clarity and usability.
Python type hints may not be not for me in practice
Chris Siebenmann expresses skepticism about Python's type hints for small tasks, finding them cumbersome. He prefers adding them post-stabilization for documentation, envisioning a system allowing distinct types without upfront details.