It takes only a few minutes to get comfortable with the web Jupyter interface and the main gotcha to be aware of is the order of the cells (the value between branckets).Īs described here, there is a new interface of Jupyter that is in active development: it is called JupyterLab The interesting part is the first directory ( /home/jovyan/work) that will be later used as a mount point so that you can use files on your local disk.
The sys.path confirms -if it was really needed- that all the dependencies are contained inside the docker image filesystem. There are a tons of documentation online that describes the basic navigation inside a Jupyter document, but it is really easy to create your first python document: This version of Juptyer comes only with the Python 3 kernel, and using the New menu you can create your first document: If you open the URL in your browser you are in your notebook! To use that container, you use the docker CLI: docker run -p 8888:8888 jupyter/minimal-notebookĪt the end of the installation, the image outputs the information to access your notebook: To access the notebook, open this file in a browser:įile:///home/jovyan/.local/share/jupyter/runtime/nbserver-7-open.html The jupyter/minimal-notebook is a good starting point to test Jupyter. You could build your own container but there are plenty of docker distributions available on the web to choose from. The second step is to choose a Jupyter docker container image to use. The first step is to have docker installed on your machine, but that’s a no brainer to install. Icing on the cake, the setup of Jupyter takes less than 5 mn! How to test Jupyter in less than 5 minutes
I used intensively Jupyter first to gain expertise in Python, but also to visualize the results of data processing algorithms. Docker makes it easy to keep your machine clean.
Jupyter is even more powerful when you use it inside Docker: as a developer you don’t want to mess up your global environments with unexpected dependencies.
0 kommentar(er)
