Installing Jupyter Lab on Windows Subsystem for Linux with Miniconda

Especially when installing Python packages to work with geographic data, the process on Windows can get a little bumpy. That’s why it might make sense to leverage the best of two worlds: using a programming environment with a GUI (Jupyter Lab) while using a Linux foundation from within Windows.

Why?

When working with specific Python packages, especially ones related to geoinformatics like fiona or GDAL, developers on Windows might encounter difficulties during installation. While the installation and bugfixing on Windows sometimes is cumbersome, installing said packages on Linux “just works” — at least from the perspective of the end user. Windows Subsystem for Linux (WSL) allows developers to run a Linux environment on Windows. While most of the time WSL takes place in the command line interface, there are possibilities to use Linux on Windows from within applications like Visual Studio Code (VSCode), Git or Docker.

If the question is “Why use Jupyter Notebooks or Jupyter Lab at all?”, Packt, amongst a lot of other sources, has ten good reasons why.

With this tutorial, we want to take creating a development environment a step further and use Jupyter Lab with Miniconda in WSL in addition to popular applications such as VSCode and Docker!

Prerequisites

WSL has to be installed (the distribution of your Linux Subsystem should not matter apart from the command to install packages), the official Documentation is very useful!

Checking and installing Python

Usually Python is installed by default. However, you should check the version with python --version in your cli. In case you are on Python 2 (checking the version would return Python 2.7.1.8), you should upgrade to Python 3 on your specific distribution (sudo apt install python3 for Ubuntu, pacman -S python3 for Arch Linux, …).

Installing Miniconda

Since we do not have a GUI at hand, we will not download the installer from the website but do everything from the cli. First we need to get the shell script which we have to run in order to install Miniconda:

sudo wget -c https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh

Since we want to run it to install Miniconda with that script, we need to make the script “executable”. If you type ls -l (which stands for “list”, the flag -l listing the contents in a table format) into your cli, you will see that there are a bunch of letters which are the permissions of each file and directory. Here you can learn more about “Permissions and Executables”. We make the script executable with the following command:

sudo chmod +x Miniconda3-latest-Linux-x86_64.sh

Now we can run the Miniconda installation script with:

./Miniconda3-latest-Linux-x86_64.sh

Now you are being greeted with a bunch of text which is the license agreement. Feel free to read it, otherwise use Enter or Space to skip ahead until you reach the line where you have to agree to the agreement. You have to answer with yes to proceed. Afterwards press Enter to install Miniconda at the default location. You did it!

Starting Miniconda and creating an environment (optional)

To start Miniconda, simply type:

conda activate

If (base) shows up at the beginning of your command line, you have successfully installed and activated Miniconda. The (base) tells us that we are in the global scope (or base) of environments.

To deactivate Miniconda (and go back to “normal”), simply type:

conda deactivate

which brings us to environments and how to create, activate and deactivate them. To create a new environment (which e.g. might be useful to use different versions of packages), type:

conda create -n jupyterenv

which would create a virutal conda environment with the name “jupyterenv”. To activate our newly created environment, we would use:

conda activate jupyterenv

Now (juypterenv) should show up at the beginning of your command line instead of (base). You can change inbetween environments without without deactivating the current environment first.

With conda info --envs you can list the installed environments.

Installing and using Jupyter Lab

Installing packages with conda is as easy as with pip: to install a package, in our case the package that will enable us to use Jupyter Lab, we have to type:

conda install jupyterlab

This will result in conda listing the packages needed to properly run Jupyter Lab. After we have confirmed the installation dialogue, we can try and start our Jupyter Lab by typing

jupyter lab

which will result in a dialogue that is supposed to redirect us to a Jupyter application. Since this has never worked for me, we press the buttons q (=quit) followed by y (=yes) button to quit the dialogue: et voilá, there is our IP to access our Jupyter Lab interface from our webbrowser as we are used to!

To not always activate the base environment by default, you can use

conda config --set auto_activate_base false

TL;DR

• sudo apt install python3 or similar, depending on your distribution
• sudo wget -c https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
• sudo chmod +x Miniconda3-latest-Linux-x86_64.sh
• ./Miniconda3-latest-Linux-x86_64.sh
• Press Enter or Space until you reach the bottom
• Type yes
• Press Enter
• Close and re-open the terminal window
• Type conda activate
• (base) should show at the beginning of the command line
• conda create -n jupyterenv
• conda install jupyterlab
• Confirm the dialogue
• jupyter lab
• q y
• Done :)

you can also find this on https://github.com/grndng/JupyterCondaWSL