Nython

Build Python Extension Modules for Nim libraries.

Synopsis

This is using Nim's compileToC command to generate C code that Python can then package as an extension module and compile natively when your package is installed elsewhere. Note that this does not solve the fact that Python wheels with C code are specific the platform they are built on. So a wheel built on a mac will run on another mac. Check out the manylinux project for ideas on how to build wheels for more platforms.

System Reqs

• Nim
• Nimpy

Install

Using your favorite Python package manager, this library lives on pypi

pip install nython

Usage

See the example folder of a working project that uses nython (and runs all the tests).

Poetry

• Add nython as package dependency
• In the [tool.poetry] section of the pyproject.toml, add `build = "build.py"
• Create the file build.py in top level of your project. This will be called by poetry when creating the package, essentially it just needs to have a build function that takes a dict of setup kwargs and adds to it.
• Add your Nim modules, nythonize them, and pass them back
• Note: you must pass in nimbase.h
• Note: your Nim code must live in a directory that is included in your package build process. It can live side by side with your python.

# Example build.py script
from nython import nythonize
from os.path import expanduser


def build(setup_kwargs):
    """Called by poetry, the args are added to the kwargs for setup."""
    nimbase = expanduser("~") + "/.choosenim/toolchains/nim-1.0.4/lib/nimbase.h"
    setup_kwargs.update(
        {
            "ext_modules": nythonize(
                nimbase, [{"name": "adder", "path": "ponim/adder.nim"}]
            ),
        }
    )

• Please see the example project for one method of merging your namespaces.
  • The tests directory shows how the functions can be run (nothing special)

Setuptools

Todo - but basically just add ext_modules = nythonize(nimbase, [{"name": "adder", "path": "ponim/adder.nim"}]) to your setup call

Showcase

A list of projects using nython to reference as how-tos

• ny_lapper

Acknowledgements

• nimpy: this is an amazing project that 'just works' and makes working with Nim and Python easy. nython is just the last 5% in getting the packaging to work. nimpy is the fir 95% of the work.
• faster-than-requests, I looked a lot at how the build system for this package was set up. Essentially, nython is just a streamlined version of the setup used in faster-than-requests.

Development

Goals

Create a seamless development experience for integrating Nim code with Python. Nim should be so easy to use that eventually you just end up writing Nim-only modules for Python, and then realize you don't really need Python and migrate to just Nim. This package should enable Nim in places and companies where it can't be selected as the primary language for a project, but it can be reached for when performance is needed. This should be easier to use than Cython.

Tests

Currently this is tested by running the code in the example project. I would like to find a better way to do this, so if you have a good idea, feel free to contribute!

Currently:

cd example
poetry shell
poetry install
poetry run py_test

And that is it.

TODOs

• Support Nimble / full Nim projects with dependancies
• Allow for fine-grained compiler option tuning
• Remove the spurious .so file that ends up in your project root dir.
• Remove the dep fo passing in nimbase.h and find it on the system somehow
• Possibly create a Nim install in your local virtualenv somehow, with nimpy
• Generate some performance tests, although that is more on nimpy than this package

Ponim

If you are interested in this, and want to see a nice local dev way of doing things, check out this example I put together: https://github.com/sstadick/ponim