Topic analysis

Are you expected to run five Python type-checkers now?

This blog post advises Python library maintainers on handling the growing number of available Python type checkers (Mypy, Pyrefly, Pyright, ty, Zuban), recommending they prioritize running multiple type checkers on their test suites rather than internal source code to ensure public API compatibility for end users. The author shares their experience adding Pyrefly type checking to the Polars dataframe library's CI workflows, noting that testing public API usage across type checkers requires far fewer code modifications than enforcing all checker rules on internal implementation code.