POST Python

POST Python is a statically typed, performance-optimized subset of Python defined by a normative specification and accompanied by a reference ahead-of-time (AOT) compiler. Designed to address the need for a formal, compilable subset of Python, it enables developers to write code that remains valid Python—runnable under the standard CPython interpreter—while also being directly compilable to native code without embedding a Python runtime in the resulting binary.

A key feature of POST Python is its focus on clear subset enforcement, with the reference implementation providing explicit diagnostics for unsupported semantics rather than silently altering behavior. The language is organized into conformance profiles such as POST Core, POST Array, POST Ufunc ABI, and CPython Extension, and the compiler in the repository serves as a reference implementation rather than the language's definition. The toolchain supports features like structural checking, scalar kernel compilation, control flow, module constants, and decorators for vectorization that produce NumPy-conformant generalized ufuncs (gufuncs). ufuncs with full broadcasting and dtype support.

POST Python is particularly aimed at scientific computing use cases, as demonstrated by the PostSciPy initiative, which seeks to recreate SciPy subpackages as pure POST Python libraries. The ppspecial package, for example, already implements 26 special mathematical functions and compiles into both a single native library and an importable NumPy extension. 2 draft specification and ongoing development to expand the feature set based on real-world numerical code requirements.

Distribution is source-only via PyPI, with binaries available through package managers that handle native code appropriately. The platform is in an early stage of development, and its interfaces are subject to change as the specification matures.