Skip to content

Latest commit

 

History

History
83 lines (60 loc) · 5.04 KB

VIRTUAL_ENVIRONMENTS.adoc

File metadata and controls

83 lines (60 loc) · 5.04 KB
Raw

Using Python Virtual Environments on Windows

Python has a feature that allows you to set up multiple virtual environments on a single computer. Each environment has its own copy of Python (e.g. Python 3.6.5 64-bit in one, Python 2.7 32-bit in another). Each environment has its own site-packages folder containing only the modules needed for that environment. Each environment has its own Scripts folder containing any executables that are installed by pip that go along with a module that is installed. And so on.

(Don’t worry, each virtual environment that you create only uses up 20-50MB of disk space, depending on the number of modules needed.)

Preparing to Create the Virtual Environment

First, install the version of Python that you’ll want to use with that environment (e.g. C:\Python38\python.exe), if not already.

Next, make sure each of your projects have a requirements.txt file that names the modules needed for that project.

  1. Navigate to the project’s root folder.

  2. pip freeze > requirements.txt

  3. Edit the requirements.txt file, paring it down to only the modules that you care about.

  4. You’ll probably want to change all of the version specs from == to >=, unless you have a particular reason to keep one or more frozen.

  5. Commit the requirements.txt file to version control.

Then, identify the best place to install the virtual environment. If you only have one project, then use the root folder for that project. If you have multiple projects that you want to share the same virtual environment, and the project folders have a common parent folder, then set up the virtual environment in the parent folder. For example, say you have C:\work\web-crawlers\craigslist and C:\work\web-crawlers\nextdoor). The craigslist project and the nextdoor project can each have their own virtual environment (C:\work\web-crawlers\craigslist\.venv and C:\work\web-crawlers\nextdoor\.venv), or they can have a common one (C:\work\web-crawlers\.venv).

  1. Open a terminal window (CMD or PowerShell).

  2. Navigate to the root folder.

  3. Invoke the venv command to create a new virtual environment, as follows:

CD c:\work\web-crawlers
C:\Python38\python -m venv .venv

What this does, specifically:

  • Creates the target directory (naming it .venv is the common convention). If an existing directory is specified, it will be re-used.

  • Creates pyvenv.cfg that points to the Python installation from which the command was run (among other things).

  • Copies the Python executable into the virtual environment. (Whichever version of Python you used to run the venv command.)

  • Installs pip into the environment.

  • Creates an (initially empty) Lib\site-packages for pip to use.

  • Creates a Scripts subdirectory containing a copy (or symlink) of the Python binaries.

  • When pip installs a module that includes command-line executables, it places them in the environment’s Scripts folder. If that executable is a Python script, then the she-bang line is automatically tweaked to point to the Python executable for that virtual environment.

Activate the Environment

When venv creates the .venv\Scripts folder, it places an activate batch file in there. (Two actually, activate.bat and activate.ps1 for PowerShell.) Running the activate script ensures that the virtual environment controls how Python operates. (That’s why we made sure to run the pip freeze > requirements.txt command before activating the virtual environment; otherwise, we would have gotten an empty list.)

Tip
 If you are using the VSCode IDE, you can select an environment using the Python: Select Interpreter command in the command palette. Now, when you open the terminal pane (ctrl-), you’ll see that the command prompt starts with "(.venv)". That selection is then saved in the VSCode workspace settings (in the `.vscode folder in the root of your workspace). So, any time thereafter, when you open that project folder, VSCode will know to use the corresponding virtual environment. Again, when you open the terminal pane (ctrl-`), you should see that the command prompt starts with "(.venv)".

Activate it now.

  1. .\.venv\Scripts\activate

  2. Use python --version to verify that the correct version of Python is tied to the environment.

  3. python -m pip install --upgrade pip

  4. At this point, if you run pip freeze, it’ll come back empty.

  5. pip install -r requirements.txt

  6. Now, if you run pip freeze, it’ll come back with a list of all of the modules you have listed in requirements.txt, plus their automatically-installed dependencies.

Notes

  • venv used to be called pyvenv until Python 3.6.

  • On Windows, it may be required to enable the Activate.ps1 script by setting the execution policy for the user. You can do this by issuing the following PowerShell command: PS C:> Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope CurrentUser

References

Next Topic: Design Notes