Python Virtual Environments
What a Python Virtual Environment Is
A Python virtual environment is an isolated directory that holds its own installed packages, separate from the packages used by the rest of the system. Each project can then use its own versions of libraries without interfering with other projects.
Python 3 includes the venv module for creating these environments. A typical command, python -m venv env, creates a folder named env. Each environment has its own Python link and its own site-packages folder, where pip installs libraries. Activating an environment makes its Python and pip the defaults for that terminal session.
Why Isolation Matters on Shared and Production Servers
Without isolation, every project on a machine shares one set of packages. Suppose one site needs one version of a library and another site needs a different version. Installing both globally is impossible, and upgrading for one project can break the other. A separate virtual environment for each project avoids this conflict.
Servers add more reasons. Many Linux distributions depend on the system Python for their own tools, and recent releases often block global pip installs to protect it. Installing into a project environment is the accepted approach. Each application can also be deployed with its own dependency list, which makes upgrades and rollbacks easier to control. Web applications written in Python are usually deployed this way, as described in Python for web development.
How requirements.txt Recreates an Environment
A project records its dependencies in a file, conventionally named requirements.txt, which lists each package and optionally its version. Pip reads this file and installs the listed packages into the active environment. Because the list is plain text, it can be stored in version control and used to rebuild the same environment on a server or on another developer's machine.
The virtual environment folder itself is normally not committed. It contains machine-specific files and can be large. The requirements list is the reference, and each machine builds its own environment from it. Settings that differ between machines, such as secrets, belong in environment variables rather than in this file. For the general concepts behind dependency lists, see what a package manager is.
Frequently asked questions
Is a virtual environment the same as a Docker container?
No. A virtual environment isolates Python packages only, and it still relies on the operating system and the Python interpreter installed on the machine. A container also isolates system files, system libraries and network settings. Containers are often used for deployment, while virtual environments are the everyday tool for managing Python dependencies.
Do I need a virtual environment for a small Python project?
It is still a good habit. Even a small script can conflict with other projects over time, and a clear requirements file makes it easier to move the project to a server later. Creating one environment per project is a low-effort default that prevents a common class of problems.