Python packages Part 1

42. Python packages Part 1#

42.1. Introduction#

This notebook goes over how to install a single python file, like a301_lib.py so it can be imported into the a301 python environment. The general task of creating, selecting and installing files is called “package management”, and programs that do that are called “package managers”. None of this will be on the final exam, but if you are planning to work with python in the future, all of this should be on your to-learn list at some point relatively soon.

Python packaging is especially complicated, because python has an extensive history (python 1.0 was released in January 1994) that predates any formal way of distributing code. Python grew up along with the internet, and pioneered many of the ways that modern programming languages install libraries. That process also meant that there were many competing experiments and dead ends, and a semi-standardized way of packaging libraries has only evolved in the last 5 years or so.

Many modern languages that were created after python learned from python’s successes and failures, and designed a single installation manager from early in the language’s evolution (pkg for Julia, cargo for Rust, npm for javascript, etc.) Python instead has about 6 package managers that can all use the same configuration files but do the basic process of package and environment management slightly differently.

Python also is different in that there are two fundamentally different ways of approaching packaging, via the official python repository (pypi.org) or via conda and conda-forge.org. This notebook only considers creating a package that is compatible with pypi.org. A pypi package can also be wrapped as a conda package, but that topic is out of scope for A301.

42.2. Where to start#

There are an excellent set of tutorials developed by the pyopenscience.org and hosted as the python packaging guide. Their examples use a fairly new packaging tool called hatch. The a301_lib package instead uses the oldest python installation program called setuptools but for this simple example there is no real difference between the two approaches.

  • Step 1: read why create a python package

  • Step 2: read python packaging for scientists and [publishing packages]( (https://www.pyopensci.org/python-package-guide/package-structure-code/publish-python-package-pypi-conda.html)

  • Step 3: in class walk through of the a301_lib pyproject.toml

  • Step 4: how to use the library

    • install with pip

      pip install git+https://github.com/phaustin/a301_lib

    • alternatively, download the requirements.txt file from the week8 folder and do:

      pip install -r requirements.txt

  • Step 5: test the installation

    • the following should print out the path to a301_lib.py inside your miniconda install

      python -c 'import a301_lib;print(a301_lib.__file__)'

    • do conda list in the terminal and you should see this line:

      a301-lib                  0.1                      pypi_0    pypi

  • Step 5: More complex examples

Contents