Mozilla wants Iodide to breath life into data science documents

Mozilla wants Iodide to breath life into data science documents

Firefox makers Mozilla have found another way to use their favourite browser, by tricking it out for (data) science documentation. Not completely satisfied with the possibilities of Google Colaboratory and Co, the company has come up with Iodide, a  – still very early stages – tool for generating interactive documents for scientific communication.

While it is already possible to use so called notebooks to build interactive text documents with embedded code and environments to share them within a browser environment exist (something quite popular for sharing examples in the machine learning community), Mozilla wanted a more elegant way for its scientists to display their information.

Iodide takes inspiration from Matlab and R Markdown and lets users create more homogeneous looking documents, making it accessible for less technical audiences. While a typical Iodide document may look like a standard website with interactive elements, an explore button placed in the upper right corner lets those interested in the code switch to a programming environment for closer inspection.

The idea isn’t terribly new though. Websites like OpenProcessing.org have been around for quite a while and let users jump between code and a rendered view in the browser. Iodide’s creators however focus their project on a more scientific audience and are also trying to get languages familiar to data scientists into the mix.

For now it’s only JavaScript and Python, but the project description already mentions R and Julia as possible future additions, as well as looking into bringing C/C++ science libraries to the browser via Webassembly and JavaScript APIs.

In the upcoming months the team will focus on getting Iodide stable, but it has grand plans for its new tool: adding ways of exporting notebook archives for example, editing notebooks with several people at a time, suggesting changes similar to the way it is done in GitHub or adding comment threads.
You can find the Iodide code on GitHub and take the project for a spin on its website. While you don’t need an account to try things in a notebook, saving your experiments is only possible after logging in via GitHub for now.