1 comments

  • conmak 17 minutes ago

    Type Enforced is a zero-dependency Python library (plus an automatic C++ build with fallback if you have a compiler) for high-throughput runtime type validation. It verifies standard and nested type annotations on callables very fast.

    For a simple int validation, it adds ~10 nanoseconds of overhead (less than a frame call) by manipulating the code via AST. Other performance gains if using the C++ variant include loop unrolling for complete nested checks and much more.

    On a practical level, it speeds up development for humans and AI agents alike.

    It's ~15x faster than beartype for sampled validation It's ~20x faster than pydantic for full validation

    Benchmarks: https://github.com/connor-makowski/type_enforced/blob/main/b...

    Example: ```python from type_enforced import Enforcer

    @Enforcer def process_point(x: float) -> dict[str,float|int]: return {"point": x}

    process_point(1.5) process_point("invalid") # Raises error ```