• Auto-complete and contextual help while coding.
• Rich content display: HTML, markdown (with latex), images, javascript, svg, videos, etc.
• Uses standard Go compiler: 100% compatibility with projects, even those using CGO. It also supports arbitrary Go compilation flags to be used when executing the cells.
• Faster execution than interpreted Go, used in other similar kernels -- at the cost of imperceptible increased start up, since each cell is compiled.
• Run cell's Test* and Benchmark* functions with go test, simply adding %test to cell.
• Compile and execute the Go code as WASM: allows one to do interactive widgets in notebooks. See %wasm (EXPERIMENTAL).
• Support for go.mod and go.work, to allow local development. Including importing specific versions of libraries.
• Shell command executions with ! -- handy at times, for instance to install packages.
• Reported to work with Github Codespace, VSCode, Binder, Google's Colab, etc.
• Very well documented and supported.
• Great for data-science, testing, writing reports, live demos, etc.
• Includes a pre-built docker, that includes JupyterLab and GoNB, that can be used to easily try it out. Alternatively, there is a Google's Colab, that bootstraps GoNB and can be used online.
• Online help and much more, see %help.

Go is a compiled language, but with very fast compilation, that allows one to use it in a REPL (Read-Eval-Print-Loop) fashion, by inserting a "Compile" step in the middle of the loop -- so it's a Read-Compile-Run-Print-Loop — while still feeling very interactive.

GoNB leverages that compilation speed to implement a full-featured (at least it's getting there) Jupyter notebook kernel. As a side benefit it works with packages that use CGO — although it won't parse C code in the cells, so it can't be used as a C kernel.

It already includes many goodies: cache between cell of results, contextual help and auto-complete (with gopls), compilation error context (by mousing over), bash command execution, images, html, etc. See the tutorial.

It's been heavily used by the author (in developing GoMLX, a machine learning framework for Go), but should still be seen as experimental — if we hear success stories from others, we can change this.

Reports of issues as well as fixes are always welcome.

There is also a live version in Google's Colab that one can interact with (make a copy first) — if the link doesn't work (Google Drive sharing publicly is odd), download it from GitHub and upload it to Google's Colab.

Only for Linux and macOS. In Windows, it works in WSL or inside a Docker

GoNB offers a pre-built docker, that includes JupyterLab and GoNB. To use it, go to a directory that you want to make available to the Jupyter notebook (your home directory, or a directory where to store the notebook files). It will be mounted on the work/ subdirectory in JupyterLab.

To start it:

docker pull janpfeifer/gonb_jupyterlab:latest
docker run -it --rm -p 8888:8888 -v "${PWD}":/home/jovyan/work janpfeifer/gonb_jupyterlab:latest

Then copy&paste the URL that it outputs in your browser.

You need to install (if not yet there), GoNB, goimports and gopls (for auto-complete), and then run gonb --install.

go install github.com/janpfeifer/gonb@latest && \
  go install golang.org/x/tools/cmd/goimports@latest && \
  go install golang.org/x/tools/gopls@latest && \
  gonb --install

And then (re-)start Jupyter (if it is already running).

In GitHub's Codespace, if Jupyter is already started, restart the docker — it will also restart Jupyter.

Note: for go.work to be parsed correctly for auto-complete, you need gopls version greater or equal to v0.12.4 (or at least v0.12.0?). You can check it with gopls version.

The recommendation is to use WSL (Windows Subsystem for Linux) or WSL2, and run Jupyter and the GoNB kernel in the Linux/WSL environment. Install there as if it were in a linux machine.

A pure Windows installation is not supported at this time — but contributions to add support for it would be welcome :)

GoNB opens a named pipe (set in environment variable GONB_PIPE) that a program can use to directly display any type of HTML content.

For most cases though, one can simply use the github.com/janpfeifer/gonb/gonbui: library, it offers a convenient API to everything available. Examples of use in the tutorial.

If implementing some new mime type (or some other form of interaction), see kernel/display.go for the protocol details.

• What is the %% symbol seen everywhere?
  • It is a special commands for GoNB that means "insert a func main {...} here". There are many other special commands, see %help for the complete list, or check out the tutorial.
• Go error handling is verbose and annoying for things interactive as a notebook. Can we do something ?
  • Yes! Error handling for small scripts in a notebook can get in the way at times. There are various solutions to this, one simple one is to use this simple package ("must")

Contributions are welcome!

• Windows version:
  • Installation.
  • Named-pipe implementation in kernel/pipeexec.go.
• Create a JupyterLab extension to allow the Go code to create and interact with widgets. Alternatively, open a WebSocket from the widget to the kernel. Some links:
  • https://github.com/jupyterlab/extension-examples
  • https://jupyter-notebook.readthedocs.io/en/4.x/comms.html
  • https://jupyter-client.readthedocs.io/en/latest/api/jupyter_client.asynchronous.html#jupyter_client.asynchronous.client.AsyncKernelClient.comm_info
  • https://discourse.jupyter.org/c/jupyterlab/extensions/43

The Jupyter kernel started from gophernotes implementation, but was heavily modified and little from it is left. Also, the execution loop and mechanisms are completely different and new: GoNB compiles and executes on-the-fly, instead of using a REPL engine.