Evaluating the dynamic behaviour of Python applications

The Python programming language is typical among dynamic languages in that programs written in it are not susceptible to static analysis. This makes efficient static program compilation difficult, as well as limiting the amount of early error detection that can be performed. Prior research in this area tends to make assumptions about the nature of programs written in Python, restricting the expressiveness of the language. One may question why programmers are drawn to these languages at all, if only to use them in a static-friendly style. In this paper we present our results after measuring the dynamic behaviour of 24 production-stage open source Python programs. The programs tested included arcade games, GUI applications and non-interactive batch programs. We found that while most dynamic activity occurs during program startup, dynamic activity after startup cannot be discounted entirely.