Lambda the Ultimate

Another HOPL-III paper: Statecharts in the Making: A Personal Account by David Harel. This paper reads much different than most of the others, as the author admits to being mostly an accidental PL designer - backing into it from a mathematical perspective.

This paper is a highly personal and subjective account of how the language of statecharts came into being. The main novelty of the language is in being a fully executable visual formalism intended for capturing the behavior of complex real-world systems, and an interesting aspect of its history is that it illustrates the advantages of theoreticians venturing out into the trenches of the real world, "dirtying their hands" and working closely with the system's engineers. The story is told in a way that puts statecharts into perspective and discusses the role of the language in the emergence of broader concepts, such as visual formalisms in general, reactive systems, model-driven development, model executability and code generation.

The Statecharts language arose from the domain of avionics and real-time state modeling. The author's main goal was to turn what were visual doodles into executable models - finite-state-automata. Both UML and Rhapsody use parts of the Statecharts engine. The paper provides a good background for the subject of visual programming languages - a topic that periodically crops up on LtU. I found the emphasis on topology, as opposed to geometry, as the mathematical basis of visual programming to be of interest (though perhaps obvious to those who are more familiar with the subject):

When it comes to visuality, encapsulation and side-by-side adjacency are topological notions, just like edge connectivity, and are therefore worthy companions to edges in hierarchical extensions of graphs. Indeed, I believe that topology should be used first when designing a graphical language and only then one should move on to geometry. Topological features are a lot more fundamental than geometric ones, in that topology is a more basic branch of mathematics than geometry in terms of symmetries and mappings. One thing being inside another is more basic than it being smaller or larger than the other, or than one being a rectangle and the other a circle. Being connected to something is more basic than being green or yellow or being drawn with a thick line or with a thin line. I think the brain understands topological features given visually much better than it grasps geometrical ones. The mind can see easily and immediately whether things are connected or not, whether one thing encompasses another, or intersects it, etc.

Provides a nice refutation for the recent brouhaha of those who think math is irrelevant for process modeling - a solid mathematical foundation is even more critical for languages that concentrate on expression in unique fashions.
(Previous LtU links to HOPL-III papers)