A simple mining software repository tool that extracts information
from a source code repository and then allows you to query it using
plain old SQL! ;)

What informations does it collect?

It retrieves repository information such as commit author, date, message, file diffs, and source code;

How does the configuration work?

You should write a .properties configuration file. In this file, you will configure things like repository, repository’s path, build mechanism (if needed), and all
metric tools you want to run together.

You have an example in /config-examples folder.

How does it persist information?

It uses Hibernate to persist information. It means that rEvolution can save data in any database. You configure it at the .properties file.

It is also easy to create a new metric. The persistence mechanism will recognize your new plugin and will automatically persist the data from your own metric tool.

My tool needs to build the source code. How can I do it?

rEvolution currently supports “ant” builds. All you have to do is to configure your ant task.

How can I create a new metric tool?

Create a class that implements the Tool interface. You will be required to implement a calculate method.

If your tool will persist information (most of them do), all you have to do is to implement ToolThatPersists. It will give you a Hibernate Session object which allows
you to save your objects.

What about logs?

It uses log4j to log stuff, so you can redirect it to anywhere you want.

Is there any visualizations up to now?

No, but you can check a few queries that are already prepared.

I will share it in a moment.

Besides getting basic information from the repository, it also calculates the following:

• Number of files per commit
• JDepend metrics
• Word count in Diff (it allows you to measure how many “@Test” or “TODO” were written, for example)
• Number of added/removed lines per commit
• Bugged commits 1

1 It tries to predict bugged commits by looking to the commit message. If the commit message says something about
“bug fixed”, it then starts looking all removed lines in the diff, and search for the first commit that added that line.
As soon as it finds it, it marks this commit as a “bugged”. The algorithm was based on this paper: http://patricklam.ca/papers/11.msr.time-of-day.pdf)

All you have to do is to have a .properties file, and then use: java -jar rEvolution.jar my-configs.properties

• Download the source code.
• Resolve dependencies: ant resolve
• Change the code
• Build a dist: ant dist

I wrote this README in the Agile 2011 Conference. It is not enough to fully understand all
metrics, and parameters you can use. You still need to look at the code. However, the configuration
example is very helpful and will help you at least get started with the tool.

My next steps are:

• Write a manual
• Implement agile metrics
• Implement SVN access