native_jit

native_jit is a Cython-powered JIT decorator for Python functions and classes.

It tries to compile decorated objects for speed and automatically falls back to the original Python object if compilation is not available in the current environment.

Features

• @jit for functions and classes
• Cython compilation pipeline (.pyx -> .c -> extension module)
• In-memory and disk cache for compiled modules
• Concurrent compilation deduplication
• Cross-platform compiler detection (CC / system compiler fallback)
• Safe fallback mode: no hard failure when compilation is unavailable

Install

pip install native-jit

For local development:

pip install -e .[dev]

Quick Start

from native_jit import jit

@jit
def add(a: int, b: int) -> int:
    return a + b

print(add(1, 2))

Advanced usage:

from native_jit import jit

@jit(nopython=True)
def fast_sum(n: int) -> int:
    s = 0
    for i in range(n):
        s += i
    return s

print(fast_sum(1000000))

API

• jit(...): decorate function/class/method
• jit.from_pyx(func_name, pyx_code, **kwargs): compile raw Cython source
• jit.warmup(): warm up compiler flag probing
• jit.cache_info(): inspect cache and compiler flags
• jit.clear_cache(disk=False): clear in-memory/disk cache

Notes on Compilers

Cython does not bundle a C compiler. A system C toolchain is still required for native compilation.

native_jit detects compiler command from:

1. CC environment variable
2. Python build configuration (sysconfig)
3. common fallback commands (cc, clang, gcc, cl)

If no compiler is available, decorated code still runs with Python fallback mode.

Publish to GitHub

git init
git add .
git commit -m "chore: prepare package for release"
git branch -M main
git remote add origin https://github.com/<your-user>/<your-repo>.git
git push -u origin main

Build and Publish to PyPI

1. Create PyPI account and API token.
2. Export token:

export TWINE_USERNAME=__token__
export TWINE_PASSWORD=<pypi-token>

3. Build package:

python -m build

4. Verify artifacts:

python -m twine check dist/*

5. Upload to TestPyPI (recommended first):

python -m twine upload --repository testpypi dist/*

6. Upload to PyPI:

python -m twine upload dist/*

Verify Install

pip install native-jit
python -c "from native_jit import jit; print(callable(jit))"