Test case generator
Describe one input field and get the test cases two classic techniques produce: the values at and just beyond each boundary, one representative of each class of valid and invalid input, and the inputs that break parsers. Each case says what it tests and whether the field should accept it. Runs in your browser.
Boundary value analysis, in one paragraph
Defects cluster at the edges of a valid range, because that is where a comparison is written and where < and <= get confused. Boundary value analysis tests each edge three ways: the edge itself, the value one step inside it and the value one step outside it. For a price from 1.00 to 100.00 with two decimals, that is 0.99, 1.00, 1.01, 99.99, 100.00 and 100.01. Six values find the off-by-one that a hundred random values usually miss.
Equivalence partitioning
Inputs that the system should treat the same way form one class, and one value from the class tests the whole class. A numeric field has at least three: below the range, inside it and above it; a text field has too short, acceptable and too long. The generator adds the classes most forms forget: empty, whitespace only, letters in a number, exponent notation that parseFloat accepts, and characters that take more bytes than they do characters.
What the generator does not know
- Business rules beyond the range: a quantity may need to be in stock, a date in the future, a username unique. Add those cases by hand.
- Whether the maximum is a character count or a byte count. The unicode cases are there to make the implementation tell you.
- Combinations of fields. For two or more inputs that interact, a decision table or pairwise set is the next technique.
- The generated rows are inputs and expectations, not automated tests. A recorded flow can replay them as a data-driven run, and the API studio can send them as request bodies.