A test harness has a few features that beat the typical homemade test macro:
You need to test the failures. If a function is supposed to abort or exit with an error message, you need a facility to test that the program actually exited when you expected it to.
Each test is kept separate, so you don’t have to worry that test 3 affected the outcome of test 4. If you want to make sure the two procedures don’t interact badly, run them in sequence as an integration test after running them separately.
You probably need to build some data structures before you can run your tests. Setting up the scene for a test sometimes takes a good amount of work, so it would be ...
Share this highlighthttp://www.safaribooksonline.com/a/21st-century-c/20441993/