fratambot/my-package

A project showcasing how to create and use a conda package

Conda packaging

This project showcases how to create and use a conda package.

Usage

• Create a conda virtual environment with python 3.9 , e.g. :

  conda create --name my_package python=3.9
  

• Install the package from the tarball archive already built and living in the /dist folder. You can conveniently use the make command:

  make install
  

That's it ! You can now use the package modules and their methods by importing them as you would do with any other python package or library.

You can test it directly from the python shell :

python
>>> from my_package import my_module
>>> print(my_module.add_one(41))
42
>>>

Or via the interactive script:

python my_script.py

You can uninstall the package as any other python package with the command:

conda remove my_package 

Rebuild the package

You can also rebuild the package if you want.

In this case you have to:

• Create a conda environment from the build_environment.yaml file since some other packages such as conda-build and conda-verify are required:

  conda env create -f build_environment.yaml
  

  A conda environment called my_package will be created and all the required packages installed.

  If you have already created the my_package environment (conda create --name my_package python=3.9) you can update it with the requirements necessary for building:

  conda env update --name my_package --file build_environment.yaml
  

• Use the make command to build the package:

  make build
  

The content of the /dist folder will be removed and the package will be built if some tests are passed.

It might take few minutes depending on your machine.

my-package's Issues

use cross-platform build

The tarball path points to the dist/osx-64 folder (see .env file) but that won't work if someone is building on a Windows machine.

Use noarch instead.

Use pyproject.toml instead of setup.py

Avoid this warning during build:

SetuptoolsDeprecationWarning: setup.py install is deprecated. Use build and pip and other standards-based tools.

Build the package using pip instead of conda-build...

...since there are still fools using pyenv+pip instead of conda 🙊