venv (virtual environments) in python

You can download the project of this blog post clicking here.

Virtual environments in python (venv) are a way that we can isolate different projects from each other. Sometimes When we work with python, each time we don’t have a dependency, what we do is:

sudo pip install DEPENDENCY; for example

sudo pip install sympy

This way, we have the library available globally, with all the consequences. If we have two projects that use that library, and one of them needs a new feature of a new version, if we have it globally, we update globally, with the consequent risk for the other application. A good practice is to control the versions of the software that you have installed, and before upgrading versions (especially if it is a major version update) do all the testing. With virtual environments we can have the application environment a with the version x of that library and an application b with a z version of that library. Usually globally we only want to have, setuptools, pip and the virtual environments themselves.

Controlling versions

A good practice is for each project to have a requirements.txt file, this way when someone download a project, they can create a virtual environment, install all the dependencies with which the project is working, and start working on the project.

A requirements.txt file is something as easy as:

library==version

For example

sympy==1.0

Using virtual environments

To install them, we type

sudo pip install virtualenv

Once it is installed, we can create a virtual environment with:

virtualenv environment_name

For example:

virtualenv env3

Using base prefix '/usr'
New python executable in /home/dio/envexample/env3/bin/python3.5
Also creating executable in /home/dio/envexample/env3/bin/python
Installing setuptools, pip, wheel...done.

After that, some files and folders have been created

ls -la
total 32
drwxr-xr-x 5 dio dio 4096 jun 18 17:08 .
drwxr-xr-x 3 dio dio 4096 jun 18 17:08 ..
drwxr-xr-x 2 dio dio 4096 jun 18 17:08 bin
drwxr-xr-x 2 dio dio 4096 jun 18 17:08 include
drwxr-xr-x 3 dio dio 4096 jun 18 17:08 lib
-rw-r--r-- 1 dio dio   60 jun 18 17:08 pip-selfcheck.json

The bin folder has scripts for interacting with the virtual environment and a copy of the python and pip interpreter among other libraries. The include folder contains support libraries if needed. The lib directory is where all of the python libraries installed on that virtual environment are.

Right now, we have a virtual environment, but we are not using it yet, we have to activate it first. To do this, we do (within the virtual environment folder):

cd env3
source bin/activate
(env3) ➜  env3

The terminal prompt changes to tell us that we are using the virtual environment. Now, everything we use is within our virtual environment. If we list the packages we have installed:

pip list --format=columns         
Package    Version
---------- -------
pip        9.0.1  
setuptools 36.0.1 
wheel      0.29.0

We see that we have a “basic” python installation in which we do not have any extra packages.

When we no longer need the virtual environment, or need to work with another, we have to deactivate it, if we don’t deactivate the virtual environment is going to be active during the session. To deactivate it we do:

deactivate

And we see the the prompt as we had it before the activation of the virtual environment.

With the environment active, if we need to install a library (for example pytest), we do the same as we would do to install globally, but we do not need to be superusers:

(env3) ➜  env3 pip install pytest
Collecting pytest
  Using cached pytest-3.1.2-py2.py3-none-any.whl
Requirement already satisfied: setuptools in ./lib/python3.5/site-packages (from pytest)
Collecting py>=1.4.33 (from pytest)
  Using cached py-1.4.34-py2.py3-none-any.whl
Installing collected packages: py, pytest
Successfully installed py-1.4.34 pytest-3.1.2
(env3) ➜  env3

If we download the tutorial project, and try to run the main.py, it will give us error since we do not have the necessary dependency. As we have said, if all dependencies of our project are in the requirements.txt file, it is as simple as run pip install -r requirements.txt to get all the dependencies.

(env3) ➜  python_venv_example git:(master) ✗ pip install -r requirements.txt 
Collecting sympy==1.0 (from -r requirements.txt (line 1))
  Downloading sympy-1.0.tar.gz (4.3MB)
    100% |████████████████████████████████| 4.3MB 434kB/s 
Collecting mpmath>=0.19 (from sympy==1.0->-r requirements.txt (line 1))
  Downloading mpmath-0.19.tar.gz (498kB)
    100% |████████████████████████████████| 501kB 2.9MB/s 
Building wheels for collected packages: sympy, mpmath
  Running setup.py bdist_wheel for sympy ... done
  Stored in directory: /home/dio/.cache/pip/wheels/05/93/22/2d0f59d842347b1f38df0d3f7a3870586df60568d2a49d94c5
  Running setup.py bdist_wheel for mpmath ... done
  Stored in directory: /home/dio/.cache/pip/wheels/02/2b/99/cd867d5da48d951118a8020e86c0c12a65022702426582d4b8
Successfully built sympy mpmath
Installing collected packages: mpmath, sympy
Successfully installed mpmath-0.19 sympy-1.0
(env3) ➜  python_venv_example git:(master) ✗

If we check the dependencies we see that we have installed the ones that we had in the requirements.txt file

pip list --format=columns      
Package    Version
---------- -------
mpmath     0.19   
pip        9.0.1  
py         1.4.34 
pytest     3.1.2  
setuptools 36.0.1 
sympy      1.0    
wheel      0.29.0