Regression Testing

Testing to ensure new changes do not adversely affect existing functionality.

Also Known As

Non-Regression Testing
Retesting
Software Regression Testing

Addresses / Mitigates

Implementation Risk: Detects and prevents regressions in the software.
Reliability Risk: Regression testing helps prevent reliability breaks caused by software change.

Attendant Risks

Schedule Risk: Can be time-consuming and introduce delays.
Reliability Risk: Requires a stable testing environment, creating dependencies.
Complexity Risk: Managing extensive regression tests can add complexity.

Used By

Scrum: Regression testing is part of the Definition of Done to ensure changes don't break existing functionality.

Description

"Regression testing is a type of software testing that seeks to uncover software regressions: whether previously working software stops working or behaves differently after changes." - Regression testing, Wikipedia

Regression Testing compares outputs from version A and A* to ensure new changes haven't broken existing functionality. However, it's fraught with Hidden Risks: number format differences, ordering changes, schema changes, methodology changes, non-determinism, and human fatigue when reviewing exceptions. While it seems like a big win, the payback is often slight. Automated Testing is usually more effective—it surfaces problems you didn't know existed, and is far easier to debug.

Variations

Regression Approach	Description	Reference
Full Regression	Re-running the entire test suite after any change.	Regression Testing, Wikipedia
Partial Regression	Testing only the areas affected by changes.	Regression Testing, Wikipedia
Golden Master Testing	Comparing outputs against a known-good "golden" baseline.	Characterization Test, Wikipedia