eolas/zk/Python_package_management.md

categories	tags
Programming Languages	python data-types

Package management

• You can use conda (the package manager that comes with anaconda). This makes it easier to work with conflicting package libraries (a bit like a package lock).

• The alternative is the native pip but you have to create virtual environments (venv) to manage packages at different versions.

To make use of virtual environments in pip you have to create the virtual environment before installing anything:

python3 -m venv venv

This will create a venv directory in your project that will manage the handling of modules.

This is especially important in ArchLinux since without a virtual environment it will ask you to install packages with pacman which is annoying for local packages.

You then activate the environment with:

source venv3/bin/activate

Now you can install packages:

pip [library_name]

requirements.txt

The requirements.txt file is similar to the package.json in Node projects.

Each line of the `requirements.txt`` file specifies a package that your project depends on and, optionally, the desired version of that package. When you share your project with others or move it between different environments (e.g., from development to production), this file makes it easy to set up your environment with the right dependencies.

Example

Flask==1.1.2
requests>=2.24.0
numpy~=1.19.2
pandas

Generate requirements file

pip freeze > requirements.txt

Pypy

• pypi.org is package registry like NPM