“I tested it” usually means “I ran it once with a number I liked”. A test plan is the written version of the question what would have to happen for this to be wrong? — decided before you run anything, so that the program cannot talk you out of it.
The four columns
| Scenario | Input | Expected | Actual |
|---|---|---|---|
| Typical adult ticket | 35 | $12.00 | |
| Child boundary, below | 12 | $8.00 | |
| Child boundary, at | 13 | $12.00 | |
| Senior boundary | 65 | $9.00 | |
| Not a number | twelve | Asks again, does not crash | |
| Nothing typed | (Return) | Asks again, does not crash | |
| Impossible age | -4 | Rejected with a message |
The expected column is filled in before running, by hand or from the specification. That order is the whole method: a value you work out after seeing the output is not a test, it is a transcript.
Choosing the rows
- One typical case per branch of the program, so every path runs at least once.
- The boundaries, and both sides of each. If the rule changes at 13, test 12 and 13. Nearly every off-by-one error in this course lives at exactly that pair.
- The empty case: no input, an empty file, an empty list.
- The hostile case: letters where a number belongs, a missing file, a number far outside anything sensible.
Seven rows chosen this way beat thirty typical ones.
Running it
Fill the Actual column honestly, including the rows that pass. A row where actual differs from expected is not a failure of the plan — it is the plan doing its only job. Fix, then re-run every row, because fixes break neighbours.
Keep the table with the program. When you hand work in, it shows what you checked; when you come back in three weeks, it tells you what “working” meant. The Bad Input Hunt is where these rows come from for a program somebody else wrote, and Testing and Debugging is the wider habit this belongs to.
Curriculum connection
B4.4
use a test plan to test programs (i.e., identify test scenarios, identify suitable input data, calculate expected outcomes, record actual outcomes, and conclude ‘pass’ or ‘fail’) by comparing expected to actual outcomes;
Link to original
A4.5
demonstrate the ability to validate a program using a full range of test cases.
Link to original