Coverage Complexity Chart - Sonar Plugin
The J3C Sonar Plugin is a visualization technique that is useful to quickly get an idea of how difficult it is to change / maintain your code. It combines the Cyclomatic Complexity and code Coverage from automated tests and reveals if the code contains areas that are dangerous to change.
The Cyclomatic Complexity and Coverage metrics were chosen because they can spot the code that is associated with an increased risk of breaking an existing functionality:
- excessive code complexity increases probability of introducing defects during maintenance
- without automated tests we cannot be sure that the changes have not introduced regressions
This plugin is inspired from crap4j project and uses the same heuristic Complexity / Coverage Thresholds in order to interpret and render the results.
| Complexity | Coverage % |
|---|---|
| 0 – 5 | 0% |
| 6 – 10 | 42% |
| 11 – 15 | 57% |
| 16 – 20 | 71% |
| 21 – 25 | 80% |
| 26 – 30 | 100% |
| 31+ | Time to refactor |
The J3C Sonar Plugin leverages JaCoCo library to analyze and extract methods coverage and complexity metrics (more details here and here), groups methods by complexity, computes the average coverage and aggregates the results as a bar chart.
The key elements in the chart above are:
- the green bars represent the average coverage for related complexity
- the red bars (if is the case) represent the difference between recommended coverage and the actual one
- the light yellow color is used to show the boundaries between desired and risky coverage areas
By analyzing the chart above we can see that even the total code line coverage reported by JaCoCo is 72%, there are code areas hard and risky to change.
In the chart below is presented another project (a smaller one) with a lower change code risk.
By examining this chart, the red bar immediately draws the attention to the methods having Cyclomatic Complexity equals with 7. Looking closer to the code, we found out 2 methods having this complexity. The first one has 45% and the second has 15% test coverage resulting in an average value equal to 30%. By increasing the test coverage for the second method we can lower the risk of introducing defects when change is requested.
How to build
Clone a copy of the main J3C Sonar Plugin git repository by running:
git clone https://github.com/cezarcoca/sonar-j3c.git
Enter the sonar-j3c directory
cd sonar-j3c
Run the build script:
mvn clean package
Installation
- Copy the plugin into the SONARQUBE_HOME/extensions/plugins directory
- Restart the SonarQube server (version 4.3+)
Usage
To launch JaCoCo as part of your Maven build and generate the jacoco.exec execution data file, use this command
mvn clean org.jacoco:jacoco-maven-plugin:prepare-agent install
To feed SonarQube with code coverage data, use below command
mvn sonar:sonar
For more information about how to configure SonarQube and JaCoCo see this documentation