9/11/2023

1 Design & Implementation of Human-Computer Interfaces

NPTEL-MOOCS
Dr Samit Bhattacharya
Computer Science and Engineering
IIT Guwahati
2 L28 - Black Box Testing (II)
Dr Samit Bhattacharya
Computer Science and Engineering
IIT Guwahati
3

4 Recap
• Code testing - TWO ways
• Review-based
• Execution-based
5 Execution Based Testing
• Two types
• Black box testing (functionality testing)
• White box testing (structural testing)
6 BBT - Idea
• Design test cases based on input/output values ONLY [no knowledge of design or
code required]

7 Idea
• TWO main approaches to design test cases
• Equivalence class partitioning
• Boundary value analysis

8 Equivalence Class & Partitioning

• Domain of input values partitioned into sets, each called „equivalence class‟, such
that program behaves the same for all the member of an (equivalent) class
9 Equivalence Class & Partitioning
• Testing code with any ONE value of an equivalence class is as good as testing with
ALL input values belonging to that class
10 Examples
11 Example 1
• Consider a code to compute square root of an input integer in the range [0, 5000]
• Input – integer within the range, output – square root
• What is/are the equivalence class(es)?
12 Obvious Choice
• Let us define ONE equivalence class – the most obvious one!
• Set of all integers within the range [0, 5000], both inclusive

13 Example 1 1
 • Consider a code to compute square root of an input integer in the range [0, 5000] 9/11/2023
• Input – integer within the range, output – square root
• What is/are the equivalence class(es)?
12 Obvious Choice
• Let us define ONE equivalence class – the most obvious one!
• Set of all integers within the range [0, 5000], both inclusive

13 Example 1
• Is that the best thing to do?
• Let‟s see
14 Example 1
• Recollect the definition of equivalence class – for all elements in the class, the
program should behave in the same way
• So, we can take any one element from the set to test the code

15 Example 1
• What is our test suite then?
• It should be of the form <input, output>
• Input = a number taken from the equivalence class
• Output = the expected output of the program for the input

16 Example 1
• So, let‟s take one pair of input/output as the test suite
• E.g. <4, 2>
17 Example 1
• If we use only this single test case, are we going to get a comprehensive idea of the
program behavior?
• Ideally, yes – we have taken a representative case from the equivalence class
18 Example 1
• Now suppose a user inputs -9!
• Is our program able to deal with this input in real usage scenario?
19 Example 1
• Answer – a big NO!
• We never tested for this input – it‟s outside the range we considered
• So, we missed something in our equivalence class formulation
20 Next Obvious Choice
• To take care, let‟s us redefine our equivalence class definition – now, the class is
defined as “the set of all integers, both positive and negative” – another obvious
one!
• Will it serve our purpose?
21 Example 1
• Answer is still NO!!
22 Example 1
• With this new definition, we take one test case
• Let us say, it is <9, 3>
• Since the entire domain of integers is now equivalence class, any one element
(here, 4) should be representative and sufficient to capture program behavior
23 Example 1
• Clearly, that is NOT the case 2
 • Answer is still NO!! 9/11/2023
22 Example 1
• With this new definition, we take one test case
• Let us say, it is <9, 3>
• Since the entire domain of integers is now equivalence class, any one element
(here, 4) should be representative and sufficient to capture program behavior
23 Example 1
• Clearly, that is NOT the case
• It still can‟t handle user inputs like -9.
• Also, we don‟t know what happens if user inputs 5001 onwards (outside the range)
24 Note
• So, we have to be very careful
• We shouldn‟t simply consider the most obvious range as the equivalence class
• It may affect our ability to form suitable test suit for comprehensive testing
25 A Non-Obvious Solution
• Now consider THREE equivalence classes
• Instead of one earlier
26 Example 1
• Set 1 - Set of negative integers (INVALID input range)
27 Example 1
• Set 2 - set of integers in the range of [0, 5000], both inclusive (VALID input range)
28 Example 1
• Set 3 – set of ALL integers larger than 5000 (INVALID input range)
29 Example 1
• Test suite must include representative input (and corresponding output) for each of
the three equivalence classes
30 Example 1
• e.g., a test suite: {(-5,output),(500,output),(6000,output)}
• Note
• It is a set of THREE elements
• Each element is a representative of the corresponding equivalence set (set 1, set
2 and set 3)
• Each element of the form (input, output)
31 Note
• Thus, with a more granular classes, we could take care of different situations
• And accordingly modify the code to take care of those
32

33 Example 2
• Let‟s consider another program for testing
• It computes intersection point of two straight lines and displays the result
34 Example 2
• Input - two integer pairs (m1, c1) and (m2, c2);
• Each pair defines a straight line of the form y=mx + c
• Output – intersection point

• What is/are the equivalence class(es)?

35 First (Obvious) Choice
• Set of ALL lines! 3
34 Example 2 9/11/2023
• Input - two integer pairs (m1, c1) and (m2, c2);
• Each pair defines a straight line of the form y=mx + c
• Output – intersection point

• What is/are the equivalence class(es)?

35 First (Obvious) Choice
• Set of ALL lines!
• Clearly, it will not work
• Why?
36 First (Obvious) Choice
• For the same reason - we‟ll not be able to create a test suite to capture variations in
input
• What variations we are looking at?
37 Input Variation
• Input is pair of straight line
• We should look to capture different relationships between the pair from the point of
view of “intersection”
38 Relationships
• THREE possibilities
39 Relationship
• Parallel lines (m1=m2, c1≠c2)
• Intersecting lines (m1≠m2)
• Coincident lines (m1=m2, c1=c2)
40 Equivalence Class
• THREE
• Set of all parallel lines
• Set of all intersecting lines
• Set of all coincidental lines
41 Equivalence Class
• At least ONE representative from each class
• Ex [(2, 2) (2, 5); (5, 5) (7, 7); (10, 10) (10, 10)]
42 Food for Thought
• What would be the equivalence classes if, instead of lines, we now consider line
segments?
• Input of the form: (x11,y11),(x21,y21); (x12,y12),(x22,y22) [each pair indicate
one end point]
• Output – intersection point or NULL

43

44 Example 3
• Let‟s consider a slightly different example
• You have developed an user authentication system, as part of the a bigger system.
The authentication system/function takes as input user ID and produces
success/failure message (which can be used to proceed further).
• Let‟s try to design the test suite for the example
45 Test Suite Design
• Input = user ID
• Not sufficient, need more clarity 4
44 Example 3
• Let‟s consider a slightly different example
9/11/2023
• You have developed an user authentication system, as part of the a bigger system.
The authentication system/function takes as input user ID and produces
success/failure message (which can be used to proceed further).
• Let‟s try to design the test suite for the example
45 Test Suite Design
• Input = user ID
• Not sufficient, need more clarity
46 First Thing – Clearly Specified Input
• Clarified input – alphanumeric string
• Still not sufficient
• Need length information (string of any length acceptable)

• Suppose you specified “5 character string” as valid ID

47 Test Suite Design
• Obvious/intuitive solution (equivalence classes)
• Set of all alphanumeric strings
• Is it sufficient?
48 Test Suite Design
• NO!!
• Program behavior for a 10 character string and 5 character string should not be the
same
49 A Better Solution
• Equivalence class 1 – set of all alphanumeric strings of length 5
• Equivalence class 2 – all other strings

• Will it work?
50 An Even Better Solution
• Class 1 – set of all 5 character alphabetic (only) strings
• Class 2 – set of all 5 character numeric (only) strings
• Class 3 – set of all 5 character alphanumeric strings
• Class 4 – set of all alphanumeric strings with length >5
• Class 5 – set of all alphanumeric strings with length < 5
• Class 6 – set of all strings with special characters

51 Note
• Test suite
• One (input, output) pair from each set
• At least six test cases
• Can we do any better (take home exercise)!!
52

53 Boundary Value Problem

• Programming error frequently occurs at the boundaries of equivalence classes
• Due to oversight - programmers fail to notice special processing required for inputs
at class boundaries
• E.g., may improperly use < instead of <= or conversely <= for <
54 Boundary Value Analysis
• Selection of test cases at the boundaries (of equivalence classes)
55 Boundary Value Analysis 5
53 Boundary Value Problem 9/11/2023
• Programming error frequently occurs at the boundaries of equivalence classes
• Due to oversight - programmers fail to notice special processing required for inputs
at class boundaries
• E.g., may improperly use < instead of <= or conversely <= for <
54 Boundary Value Analysis
• Selection of test cases at the boundaries (of equivalence classes)
55 Boundary Value Analysis
• Ex - reconsider function to compute square root of integer values in the range of 0
and 5000
56 Boundary Value Analysis
• Earlier we considered test cases for 3 classes
• Set of negative integers
• Set of integers in the range of [0, 5000]
• Integers larger than 5000
57 Boundary Value Analysis
• Along with those, we should include test cases in our suit
• <-1, output>
• <0, output>
• <5000, output>
• <5001, output>
58

59 Note
• Equivalence classes are sets 🡪 should use set notations to represent
• Test cases should contain both input and expected output
• Remember to include test case(s) from valid class(es), invalid class(es) and
boundary cases in black-box test suites
60 Note
• Next lecture – case study
61 Reference
• Rajib Mall (2018). Fundamentals of Software Engineering, 5th ed, PHI Learning Pvt
Ltd. Chapter 10
• Roger S Pressman (2020). Software Engineering: A Practitioner's Approach, 9th ed,
McGraw-Hill Education, New York, Chapter 19-21