Beyond Linear Notebooks: Implementing Reactivity with IPython

Jupyter notebooks are ubiquitous in data science workflows today due to their power and flexibility. However, they do have some drawbacks, including the much discussed issues of hidden state and legibility of logic flow. To address these problems at Hex, we've augmented the IPython kernel to allow for reactive execution of both Python and SQL. This talk will discuss the pros and cons of the traditional IPython execution model versus a reactive model, as well as dive into our implementation of reactive notebooks under the hood.