Detail

Sourcerer: Slicing and dicing large amounts of open source code

Sourcerer is an engine for crawling, downloading, parsing, analyzing, searching, and retrieving open source code available on the internet. At its heart there is a static analyzer that mines information for further uses. This information is used both to conduct empirical studies of software as well as to support the design of tools that facilitate modern software development. Currently, Sourcerer holds a collection of close to 5,000 Java projects, totaling 37 million lines of code.

This talk focuses on static analysis in the context of mining large repositories of software, and the new forms of software science that the availability of large quantities of software enables.