Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Test Case Design - Triangle Problem

Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

Test cases key premises 3.

Test case design


We cannot test everything

The triangle problem


Good testing relies on a good selection of test cases Practical design of test cases
(most examples based on materials by Collard & Co.)

Lessons learned

How to specify a test case?


Testing is the process of executing a program with the intent of finding errors. Test failure occurs when fault executes How do we detect that failure has occurred? The actual output is different than the expected output How do we obtain the actual output? We execute a program with a given set of inputs

How to specify a test case?


For the full specification use a Test Case Specification Template (e.g. From IEEE 829 standard) Simple (tabulated) specification has to have as a minimum the following entries
Test case ID Specific values for all the inputs Expected output(s) Actual output(s) Observations

Test execution
Input1 Input2 ... Input n

The triangle problem (simple version)


The input to the TriangleType function are three numbers a, b and c that represent the lengths of the three sides of the triangle. Based on these inputs the function determines the type of the triangle, which can be
Equilateral (i.e. all three sides are equal) Isosceles (two equal sides) Scalene (three unequal sides)

Executable software
Actual output The same ?

Specification

Expected output

Test passed

Test failed

The function returns the result in the form of the character string, e.g. Equilateral if the triangle is equilateral.

Class exercise
Suggest a set of test cases for testing the TriangleType function

Minimal test?

Test case 1 2 3

Input values 3, 3, 3 3, 3, 2 3, 4, 5

Expected results Equilateral Isosceles Scalene

15 minutes

Automated test?
for i = minval to maxval by increment do for j = minval to maxval by increment do for k = minval to maxval by increment do enter { i, j, k } capture { result } if [ i = j = k ] and result = equilateral or if [ ( i = j ) and ( i not = k ) and result = isosceles ] or if [ ( j = k ) and ( j not = i ) and result = isosceles ] or if [ ( k = i ) and ( k not = j ) and result = isosceles ] or if [ ( i not = j ) and ( j not = k ) and ( k not = i ) and result = scalene ] then ok else write error_log [ i, j, k, result ] end end end

Automated Monte Carlo test?


do while we_want_to_keep_testing i = random ( i ) j = random ( j ) k = random ( k ) enter { i, j, k } capture { result } if [ i = j = k ] and result = equilateral or if [ ( i = j ) and ( i not = k ) and result = isosceles ] or if [ ( j = k ) and ( j not = i ) and result = isosceles ] or if [ ( k = i ) and ( k not = j ) and result = isosceles ] or if [ ( i not = j ) and ( j not = k ) and ( k not = i ) and result = scalene ] then ok else write error_log [ i, j, k, result ] end

More test cases


Input values -5 1 1 130 -5 10 2 140 -5 1 3 130 Expected results Invalid Invalid Invalid Isosceles Equilateral Invalid Actual results Equilateral Isosceles Scalene Invalid Scalene Equilateral

Lessons learned (1)

Specification must be as precise as possible

3.329951 3.330023 3.330050 4 3 3 3

Lessons learned (2)


Testing must include negative cases Question: Is it useful to have more negative than positive tests? Answer It depends on
The required reliability The perceived hostility of the environment

Lessons learned (3)


Effective testing requires careful selection of appropriate test cases Accurate and thorough specification is essential It helps to have background knowledge, both of the application area to be tested and of computer programming

Normally positive cases should outnumber negative

Homework

Next lecture

Improve the Automated Monte Carlo test given in the lecture

Test case design background, test plan, documentation and standards

You might also like