Bringing Esterel out of its Shell – RETROFIT

The language Esterel has proven successful for real-time, safety-critical applications, and is used in many industrial systems including Airbus and Dassault Aviation airplanes. This success of Esterel in domains requiring strong safety guarantees is attributable to its synchronous reactive programming model. Esterel treats computation as a sequence of reactions to external stimuli; even though each reaction includes both state and parallelism, the result is completely deterministic.

The semantic difference between Esterel and conventional programming languages has led to an entirely separate Esterel compiler development. Accordingly, Esterel's use has been limited to domains that require high-reliability guarantees, where the high cost of a separate software development can be tolerated. Thus, Esterel remains a niche language and the vast majority of programmers miss out on Esterel's reliability benefits.

We propose to explore and improve two different implementation strategies for Esterel that retrofit Esterel with with mainstream language integration, using two different techniques. We will explore and improve these two techniques writing Esterel-based systems in three domains: graphical user interfaces (GUIs), the Internet of Things (IoT), and medical prescriptions. In the medical prescription domain, we will enlist M.D. and medical researcher Dr. Stephen Belknap, with whom we have collborated before, to build a prescription programming language.

The two PIs, with G. Berry, have been working together on Esterel for several years. They have been a part of an Inria Associated Team where they have obtained preliminary results using Esterel-like language for designing and implementing a medical prescription programming language. The results are promising and, to the PIs, demand follow up.