Virtual environments

Virtual environments will allow you to install and try out as many different Python packages as possible.

Creating a virtual environment

To create a virtual environment, execute the following shell command

% python -m venv test

This will create a directory named test.

Activating your virtual environment

To activate your virtual environment, execute the following shell command

% source test/bin/activate

You're now using an isolated/sandboxed Python installation under the test directory.

You should notice that which python now points to the Python executable that resided within your virtual environment.

deactivate

To deactivate the virtual environment, execute the deactivate shell command

(test) % deactivate

You should notice that which python now points to the system installed Python executable.

pip

The main package installer for Python is pip.

upgrade pip

For the rest of this tutorial, you should only run pip from within an activated virtual environment. Before you proceed any further, you should upgrade pip

(test) % pip install --upgrade pip

Now you can install external packages such as pydicom

(test) % pip install pydicom

dependency management

You may notice that when you install an external package such as pydicom, this triggers the installation of additional packages. This creates a dependency graph. If we were to explore the dependency graph for the yaxil library, you would see the following

The following graph was rendered with pipenv

yaxil==0.4.3
  - arrow [required: Any, installed: 0.15.5]
    - python-dateutil [required: Any, installed: 2.8.1]
      - six [required: >=1.5, installed: 1.14.0]
  - lxml [required: Any, installed: 4.5.0]
  - pyaml [required: Any, installed: 19.12.0]
    - PyYAML [required: Any, installed: 5.3]
  - pydicom [required: Any, installed: 1.4.1]
  - requests [required: Any, installed: 2.22.0]
    - certifi [required: >=2017.4.17, installed: 2019.11.28]
    - chardet [required: >=3.0.2,<3.1.0, installed: 3.0.4]
    - idna [required: >=2.5,<2.9, installed: 2.8]
    - urllib3 [required: >=1.21.1,<1.26,!=1.25.1,!=1.25.0, installed: 1.25.8]
  - six [required: Any, installed: 1.14.0]

You can see that yaxil depends on arrow, lxml, pydicom, requests, and six and in turn, each of those dependencies have their own dependencies.

Diving deeper, you'll see that arrow depends on python-dateutil which in turn depends on six >= 1.5. Coincidentally, yaxil also depends on six and will accept Any version. But what if it didn't? If these two packages had conflicting requirements, there would be a dependency conflict.