Build Stuff'14

Fans of different programming languages always argue about benefits of their language of choice. It is difficult to use objective criteria in a debate like this. Terms like 'clarity' or 'maintainability' are too vague and subjective. What if we used some tools from machine learning and network science to compare projects written in different languages?

In the talk, I’ll examine how the choice of programming language affects dependency structure of code. I will interactively compare different projects written in C# and F# languages. The two languages use the same execution runtime, both are statically typed, but they represent different programming paradigms - object-oriented C# and functional F#.

I’ll show that F# and C# implicitly lead to different graph structures and local network patterns. Specifically, I'll describe specific types of cyclic dependencies that frequently appear only in C# projects. Seeing the local network patterns can give more insights than calculating measures like cyclometric complexity of code. I’ll also show how anyone can run the analysis on a their own software project.