Software testing basics: In software engineering, what is the best definition of unit testing?

Introduction / Context:
Unit testing is a core practice in quality assurance. By verifying the smallest testable parts of software—functions, classes, or modules—engineers detect defects early, document intended behavior, and enable safe refactoring. Clear understanding of unit testing distinguishes it from integration, system, and user acceptance testing.

Given Data / Assumptions:

• “Unit” refers to a small, isolated piece of code with a defined interface.
• Tests run against the unit without external dependencies when possible (using stubs/mocks).
• Pass/fail criteria come from the module’s specification.

Concept / Approach:
The essence of unit testing is isolation. Inputs are provided and outputs or side effects are asserted. External services (files, networks, databases) are simulated to keep tests deterministic and fast. This differs from system testing (end-to-end), acceptance testing (business validation), and regression testing (guarding against reintroduction of bugs). Strong unit tests reduce defect propagation to later phases where fixes are costlier.

Step-by-Step Solution:

Verification / Alternative check:
Code coverage metrics, mutation testing, and continuous integration results provide evidence that unit tests exercise the module thoroughly and consistently.

Why Other Options Are Wrong:

• Live end-user runs describe system or acceptance testing.
• Ensuring transactions process during user acceptance is not unit isolation.
• Testing only recent changes without isolation risks masking defects and is not a definition.

Common Pitfalls:
Over-mocking (tests mirror implementation not behavior), flaky tests due to hidden dependencies, and insufficient negative test cases.

Final Answer:
Checking the logic and behavior of an individual module or program in isolation