Testing is the systematic process of checking a program to determine whether it works as intended and meets its requirements.
Testing checks that a program produces the expected results when used correctly and attempts to reveal situations in which it fails.
Effective testing should include unusual, extreme, and invalid inputs rather than only typical inputs.
Destructive testing involves deliberately trying to make a program behave incorrectly or fail.

The two main types of testing are iterative testing and final testing, which is also called terminal testing.
Iterative testing takes place repeatedly while a program is being developed.
Individual sections or subprograms are tested as they are created rather than waiting until the complete program is finished.
Iterative testing makes errors easier to locate because recently added sections can be investigated separately.
Iterative testing can also check that newly developed sections work correctly with parts of the program created earlier.
Final or terminal testing takes place near the end of development after the individual parts of the program have been combined.
The complete program is tested as a whole to check that its features work together correctly and that it satisfies its overall requirements.

The two important types of programming error are syntax errors and logic errors.
A syntax error occurs when code breaks the grammatical rules of a programming language.
Syntax errors normally prevent the program from executing successfully.
Common causes of syntax errors include misspelt keywords, missing punctuation, incorrectly structured statements, and incorrect indentation.
A logic error occurs when code is syntactically valid but the algorithm does not produce the intended result.
Logic errors are identified by comparing the actual result produced with the expected result.

The four main categories of test data are normal, boundary, invalid, and erroneous data.
Normal test data is valid data of the correct type that represents a typical value within the accepted range rather than a boundary value.
Boundary test data is valid data at the extreme edge of an accepted range.
Boundary testing is useful because errors commonly occur around conditions such as \verb|<|, \verb|<=|, \verb|>|, and \verb|>=|.
Invalid test data has the correct data type but a value outside the permitted limits.
Values immediately outside the accepted boundaries are particularly useful for testing whether limits have been implemented correctly.
Erroneous test data is of the wrong data type for the input being tested.

A test plan records the tests that will be carried out so that testing is systematic.
A test plan normally includes the test data, type of test data, reason for choosing it, expected result, and actual result.
The expected result should be decided before the test is performed.
A test passes when the actual result matches the expected result and fails when the results differ.
If the actual result differs from the expected result, an error has been identified and the algorithm may need to be refined.

Refining an algorithm means improving it.
Refinement may involve correcting errors or improving an algorithm’s structure, efficiency, readability, or maintainability.
After an algorithm is refined, tests should be repeated to check that the change has corrected the problem without introducing new errors.
The refinement process can be viewed as test the algorithm, identify the problem, refine the algorithm, and retest the algorithm.
Refinement can also involve removing unnecessary repetition, making better use of iteration or subprograms, improving identifiers, indentation, or comments, simplifying logic, or improving maintainability.

Module Review