Toggle navigation

.rst .pdf

Level 1: requirements.txt, “clone the repo” deployment

• Document the minimal Python version required in a README.md file or similar.

• Let the end user manage their Python environments with the required Python version. They are probably using Conda, and creating new Conda environments is not too difficult.

• List all dependencies in a requirements.txt file.

  If you want to force version numbers, use the pip freeze command. I’d advocate to replace the exact astropy==5.0.1 numbers to minimal version numbers such as astropy>=5.0.0.

  If you find that your requirements start to be complex, move to the Level 2 approach.

• People install your dependencies by running pip install -r requirements.txt.

• People reuse your code by coping the Python files in their own projects, or by adding your source folder to their Python path.