An xUnit framework has the following general architecture.1
A test case is the smallest part of a test that generally encodes a simple path through the software under test. The test case code prepares input data and environmental state, invokes the software under test and verifies expected results.
A programmer writes the code for each test case.
A test case is implemented with one or more assertions that validate expected results.
Generally, the framework provides assertion functionality. A framework may provide a way to use custom assertions.
A test suite is a collection of related test cases. They share a framework which allows for reuse of environment setup and cleanup code.
Generally, a test runner may run the cases of a suite in any order so the programmer should not depend on top-to-bottom execution order.
A test fixture (also known as a test context) provides the environment for each test case of a suite. Generally, a fixture is configured to setup a known, good, runtime environment before tests run, and to cleanup the environment after.
The fixture is configured with one or more functions that setup and cleanup state. The test runner runs each setup function before each case and runs each cleanup function after.
A test runner is a program that runs tests and reports results.2 The program is often part of a framework.
A test runner may produce results in various formats. Often, a common and default format is human-readable, plain-text. Additionally, the runner may produce structured output. Some xUnit adaptations (i.e. JUnit) can output XML that can be used by a continuous integration system such as Jenkins and Atlassian Bamboo.
Beck, Kent. "Simple Smalltalk Testing: With Patterns". Archived from the original on 15 March 2015. Retrieved 25 June 2015. https://web.archive.org/web/20150315073817/http://www.xprogramming.com/testfram.htm ↩
Meszaros, Gerard (2007) xUnit Test Patterns, Pearson Education, Inc./Addison Wesley ↩