Analysing the complexity of a git repository's history

Question

I am sure most developers are familiar with cyclomatic complexity which is used to indicate the complexity of code.

I was wondering if there is a tool that analyses the complexity of git history. Not the actual code or commits themselves but how the commit graph looks.

Something like having a "clean" history would give a low score and having a complex history would give a high score.

Any ideas if this exists?

Answer 1

A simple solution could be to get all your branches, generate all combination of two branches, and for each couple test if one branch is contained into the other (git branch --contains). If that is the case, then you can increment a counter. At the end, you can provide a score, that would be the counter divided by the number of combination:

• A score close to 100% likely means that all your branches are merged one into the other
• A score closes to 50% means that all your branches are probably all merged into master
• A score of 0% means that your changes are systematically rebased (or that all your branches are un active).

This is quiet limited, but there is a code sample here on Github with an example of how to integrate it into your repo.