All geek questions in one place

Packing a Python application with PyPI + OS level dependencies

I would like to package a scientific Python application for PyPI. My problem is that it uses PyPI level fingerprints (e.g. numpy, scipy, etc.), as well as others that need to be processed at the OS level: wxPython and Python-VTK (e.g. from apt-get on Ubuntu , homebrew on OSX, etc.).

I would like to know what would be the ideal strategy for this, and in particular which packaging system is best suited (I currently prefer Distribute).

+6
source share
1 answer

distribute is fork setuptools with possibly the best documentation. You basically have distutils (stdlib) and setuptools . Since distutils does not allow you to specify dependencies, only setuptools remains.

As a rule, you specify all the dependencies and clearly document the installation procedure (including in the long_description field placed in PyPI). Include packages installed at the OS level; most distributions include information about the eggs when they are installed.

+3
source

Source: https://habr.com/ru/post/920609/

More articles:


All Articles