Using C++ in Jupyter Notebooks on Windows

I run WIN11 and I want to do some tests in C++ in Jupyter Notebooks. I cannot easily install the xeus-cling kernel with conda under the WIN11 host because it is based on cling + LLVM blablabla…

I could use a Docker image but here I will use WSL

1. Check

Open a win terminal with Powerhell

wsl --list --verbose

You should see something similar to

If you don’t see Ubuntu, activate WSL2 on your WIN11 host (shame on you if it was not yet activated 😁) and install Ubuntu.

2. Run Ubuntu

3. Install Miniconda in Ubuntu

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

Keep in mind

At one point you will see a message saying something like :

If you’d prefer that conda’s base environment not be activated on startup, run the following command when conda is activated:

conda config --set auto_activate_base false You can undo this by running conda init --reverse $SHELL

4. Check conda

conda --version

If needed

export PATH="$HOME/miniconda3/bin:$PATH"
source ~/.bashrc
conda --version

5. Create conda virtual env & install

conda create -n cpp_jupyter -c conda-forge xeus-cling
conda activate cpp_jupyter
conda install -c conda-forge jupyterlab

For what I can remember

I used conda init at one point
I may have had to restart the terminal

6. Run Jupyterlab in Ubuntu

jupyter lab --no-browser --ip=0.0.0.0

Click on the line that says : ` http://127.0.0.1:8888/lab?token=cdbb449ab8b7f6f01e8cb8c8844cb4347b5c69c63dfe9940`

The browser should be opened on the WIN11 host

Select C++17 and live your live…

In the notebook above I first make a simple test (Hello…) then run the Dijkstra algorithm (read this page). Two points :

There is no main() function
So far, the notebook and the test_dijkstra_03.txt file which is opened by the C++ code are both on the Ubuntu host

What if…

I want to keep the Notebook on the WIN11 host ?

From the WIN11 host I can access my Ubuntu home directory via : \\wsl.localhost\Ubuntu\home\philippe

First (for testing purpose) I copy and paste both files (.ipynb and .txt files) in a new directory on the WIN11 host. Just to proove I’m not cheating I rename the Jupyter Notebook.

Then I File/Shut Down Jupyter because this instance is no longer useful. There is no need to CTRL+C in the terminal.

From the Ubuntu terminal I can access the files on the WIN11 host with : ls /mnt/c/Users/phili/OneDrive/Documents/Programmation/cpp_jupyter/dijkstra/

So, I move to the directory cd /mnt/c/Users/phili/OneDrive/Documents/Programmation/cpp_jupyter and run Jupyter from there :

This is exactly exactly what I want :

On the left hand side I see the directories and below I clicked on dijkstra to show the files
I can edit/copy/debug/create…
Thsi is similar to the Docker experience

Idea

From the home directory on Ubuntu launch VSCode (code .)
Open the .bashrc file
Add the line below
- alias jupytercpp='cd "/mnt/c/Users/phili/OneDrive/Documents/Programmation/cpp_jupyter" && jupyter lab --no-browser --ip=0.0.0.0'
Save and exit from VSCode
source ~/.bashrc
Now I can run Jupyter from the right directory by invoking jupytercpp

What is next ?

I’m starting to learn Rust (👍)
Obviously the next step is to follow a similar process the kernel evcxr_jupyter
Find here checklist for Rust