Abstract
GitHub is the most widely used social, distributed version control system. It
has around 10 million registered users and hosts over 16 million public
repositories. Its user base is also very active as GitHub ranks in the top 100
Alexa most popular websites. In this study, we collect GitHub's state in its
entirety. Doing so, allows us to study new aspects of the ecosystem. Although
GitHub is the home to millions of users and repositories, the analysis of
users' activity time-series reveals that only around 10% of them can be
considered active. The collected dataset allows us to investigate the
popularity of programming languages and existence of pattens in the relations
between users, repositories, and programming languages.
By, applying a k-means clustering method to the users-repositories commits
matrix, we find that two clear clusters of programming languages separate from
the remaining. One cluster forms for "web programming" languages (Java Script,
Ruby, PHP, CSS), and a second for "system oriented programming" languages (C,
C++, Python). Further classification, allow us to build a phylogenetic tree of
the use of programming languages in GitHub. Additionally, we study the main and
the auxiliary programming languages of the top 1000 repositories in more
detail. We provide a ranking of these auxiliary programming languages using
various metrics, such as percentage of lines of code, and PageRank.
Description
On GitHub's Programming Languages
Links and resources
Tags