Bonsai: Reactive Tools for Data Science

Опубликовано 22 июня 2016, 2:45

I will present Bonsai, a general purpose visual programming language (GPVPL) for rapid prototyping of reactive data acquisition and control systems. We will start with a live demo of how the system works and I will then briefly describe how we leveraged the .NET Reactive Extensions (Rx) combinators as well as runtime code generation to develop flexible, compact and efficient graphical specifications of multi-modal reactive systems for neuroscience and beyond. Finally, I will discuss two recurrent key concerns in GPVPLs that have proven hard to get right in visual programming languages: flexible specification of asynchronous computations; and reactive control structures for discrete event systems (e.g. state machines; behavior trees). I will present how these two problems were addressed in Bonsai by leveraging LINQ and Rx combinators and hopefully we will finish with a brainstorm about future directions (virtual reality, embedded systems, and more).