Important metrics for Measuring Code Health
3 likes1,051 views
The document discusses important metrics for measuring code health, including churn, code coverage, and complexity, while explaining how these metrics can indicate potential areas of concern in codebases. It emphasizes the role of churn in predicting future code changes, the safety of code changes based on coverage, and the dangers associated with complex code. The document also outlines methods for collecting these metrics and offers guidance on interpreting their values to assess code quality.
Important metrics for Measuring Code Health
- 1. IMPORTANT METRICS @mscottford BY M. SCOTT FORD CO-FOUNDER & CHIEF CODE WHISPERER, CORGIBYTES FOR MEASURING CODE HEALTH
- 2. @mscottford
- 3. @mscottford METRICS YOUR DOCTOR USES
- 7. @mscottford AGE? WEIGHT? HEIGHT? METRICS YOUR DOCTOR USES TEMPERATURE?
- 8. @mscottford AGE? WEIGHT? HEIGHT? BLOOD O2 LEVEL? METRICS YOUR DOCTOR USES TEMPERATURE?
- 9. @mscottford AGE? WEIGHT? HEIGHT? BLOOD O2 LEVEL? METRICS YOUR DOCTOR USES TEMPERATURE? BLOOD PRESSURE?
- 10. @mscottford AGE? WEIGHT? HEIGHT? BLOOD O2 LEVEL? ALL OF THESE ARE USED TOGETHER TEMPERATURE? BLOOD PRESSURE?
- 11. @mscottford AGE? WEIGHT? HEIGHT? BLOOD O2 LEVEL? TOGETHER THEY PROVIDE SOME HINTS TEMPERATURE? BLOOD PRESSURE?
- 14. @mscottford WE PERFORM CODE INSPECTIONS
- 15. @mscottford WE’VE LOOKED AT A LOT OF PROJECTS
- 16. WHAT METRICS DO WE LOOK AT? @mscottford
- 19. @mscottford CHURN A measure of how many times file has changed over a specific amount of time
- 20. @mscottford LIFETIME CHURN A measure of how many times file has changed over the life of a project
- 21. @mscottford RECENT CHURN A measure of how many times file has changed during the last 3 months
- 24. @mscottford CODE COVERAGE A measure of how much of the application is executed when a test is performed
- 25. @mscottford ACCEPTANCE TEST CODE COVERAGE A measure of how much of the application is executed when the acceptance tests are run
- 26. @mscottford INTEGRATION TEST CODE COVERAGE A measure of how much of the application is executed when the integration tests are run
- 27. @mscottford UNIT TEST CODE COVERAGE A measure of how much of the application is executed when the unit tests are run
- 30. @mscottford COMPLEXITY A measure of how difficult a chunk of code is to work with
- 31. @mscottford A measure of unique code paths for a block of code CYCLOMATIC COMPLEXITY
- 32. @mscottford A measure of the cognitive effort required to understand a block of code COGNITIVE COMPLEXITY
- 33. @mscottford HOW DO WE COLLECT THESE METRICS?
- 34. CHURN • git-churn script by Corey Haines & Gary Bernhardt • https://github.com/garybernhardt/dotfiles/blob/master/bin/git- churn @mscottford
- 35. CODE COVERAGE • Run tests with code coverage collection tool • These vary by language •Ruby: simplecov, rcov •Python: coverage.py • C#/.NET: dotCover, NCover, OpenCover •Java: Clover, JCov, JCoCo, OpenClover, Cobertura, •JavaScript: Istanbul, Jest @mscottford
- 36. COMPLEXITY • Run a static analysis tool to collect the data • These also vary by language • Ruby: rubocop, flog • Python: radon • C#/.NET: MS Visual Studio, NDepend, ReSharper, SonarQube • Java: PMD, GMetrics, IntelliJ, SonarQube, • JavaScript: ESLint, JSHint, ccmj @mscottford
- 38. @mscottford CHURN Code that has changed in the past is more likely to change in the future. A good stand-in for single responsibility principle.
- 39. @mscottford CODE COVERAGE Code that is tested is safer to change.
- 40. @mscottford COMPLEXITY The more complex code is the more dangerous to change.
- 41. @mscottford WHAT ARE ‘GOOD’ METRIC VALUES?
- 42. @mscottford CHURN Low is better. High is not bad, but outliers are. Recent churn is more important than lifetime churn.
- 43. @mscottford CODE COVERAGE 100% across the entire codebase is asymptotic. 85% is a reasonable target. Unit > Integration > Acceptance
- 44. @mscottford COMPLEXITY Provides best info at method level: 1 - 5 : Good 6 - 10 : Complex > 10 : Extremely Complex
- 45. @mscottford USING THESE METRICS TOGETHER
- 46. @mscottford
- 47. @mscottford
- 48. LET’S EXPLORE COMMON SCENARIOS @mscottford
- 50. @mscottford COMPLEXITY LOW CODE COVERAGE HIGH CHURN LOW HEALTHY
- 52. @mscottford COMPLEXITY LOW CODE COVERAGE HIGH CHURN HIGH DESIGN SMELL
- 54. @mscottford COMPLEXITY LOW CODE COVERAGE LOW CHURN LOW TESTING NOT A FOCUS
- 56. @mscottford COMPLEXITY HIGH CODE COVERAGE HIGH CHURN LOW REFACTORING NOT A FOCUS
- 58. @mscottford COMPLEXITY HIGH CODE COVERAGE LOW CHURN LOW UNHEALTHY: LOW PRIORITY
- 60. @mscottford COMPLEXITY HIGH CODE COVERAGE LOW CHURN HIGH UNHEALTHY: HIGH PRIORITY