Software Testing and Quality
Assurance
Lecture 7 - Software Testing
Techniques
1

Lecture Outline

Discuss Equivalence Partitioning

How to select test cases based on equivalence Partitioning.

Discuss Boundary-value Analysis

How to select test cases based on boundary-value analysis.
2

Equivalence Partitioning

Partition the input of a program using the functional requirements.

Test inputs are executed on the program, and

The tester evaluates whether the output of the program is in the expected output domain.
3

Choosing Equivalence
Classes

Aim is to minimize the number of test cases required to cover all of the Equivalence
Classes (EC).

Identify the initial EC.

Identify overlapping EC, and

Eliminate them by making the overlapping part a new equivalence classes.

Select one element from each EC as the first test input; and

Work out the expected result for that test input.
4

Identify initial equivalence classes

No clear formula

We need to build up some judgment and intuition.

Guidelines to identify potential EC.

We may need to choose extra cases in order to explore difficult subsets of the input domain.
5

Identify initial equivalence classes - Guidelines


Input condition specifies range of vales,

Identify one valid EC

Two invalid EC

One for the set of values below the range

One for the set of values above the range.
For example, range of values 1 …. 99,

The valid EC 1 …99; and

Two invalid EC i.e. {x | x < 1} and {x | x > 99}
6

Identify initial equivalence classes - Guidelines

Input condition specifies a set of input values.

A valid EC for each element of the set, and

One EC class for the elements not in the set.

For example, input selected from a set of N inputs, we need N + 1 EC;

One valid EC for each element of set,and

One invalid EC for element outside the set.
7

Identify initial equivalence classes - Guidelines

Input condition specifies a ‘must be’ situation,

Identify one valid EC and one invalid EC.

For example, if character input must be numeric , we need two EC

A valid EC --- { s | the first character of s is a numeric}

One invalid EC --- {s | the first character of s is not a numeric}
8

Identify initial equivalence classes - Guidelines

If program handles each valid input differently, then
 define one valid EC per valid input.

For example, input from a menu

Define one valid EC for each menu item.

If elements in EC are handled differently by the program, then

Split the EC into smaller EC
Some of the guidelines are very general in nature. Domain knowledge and experience plays a key role in identifying EC.
9

Eliminating Overlapping EC
EC 1
Overlapping
EC
EC 2
10

Eliminating Overlapping EC -
Example
Disjoint EC

A - an empty list

B - list with one element

C - list with at least two element

D - a sorted list; and

E - an unsorted list
Two or more elements are either sorted or unsorted.
• Unsorted list with at least 2 elements.
• Sorted list with at least 2 elements.
11

Selecting Test Cases

Once EC have been identified

Any value from EC is identified to be as likely to produce a failure as any other value in that class.

Thus, any element of the class servers as a test input.

However, selecting just any value in EC may not be optimal for finding faults.

For example, faulty implementation of square function.
12

Important Question?

How to resolve the overlap so that we do not derive too many test cases?
13

Boundary-value Analysis
(BVA)


Greater number of errors occur at the boundaries of the input domain.

Select test cases to explore the boundary conditions of a program.
BVA complements equivalence partitioning.

Rather then selecting any element from
EC,

BVA leads to the selection of test cases at the ‘edges’ of the class.
14

Boundary-value Analysis

Faults are more likely to introduced at boundary because

Unsure of the correct boundary for an input condition;

Have incorrectly tested the boundary.
15

Example - Triangle Program

A program reads floating point values from the standard input.

The three values are interpreted as representing the lengths of the sides of a triangle.

Program prints a message that states whether the triangle is

Equilateral, isosceles, scalene or invalid.
16

Example - Triangle Program

Equilateral triangle

Only one boundary where x = y = z.

Test cases to explore this boundary

(3,3,3) On Point

(2.99, 3, 3) Off Point Below

(3.001,3,3) Off Point Above
17

Example - Triangle Program

Isosceles Triangle
 x = y AND y ≠z, y = z AND z ≠x …..

Test cases to explore the boundary x = y AND y ≠z

(3,3,4) On Point

(2.99, 3, 4) Off Point Below

(3.001,3,4) Off Point Above
18

Boundary-value Analysis -
Guidelines



Input condition specifies a range bounded by values a and b , test cases should designed with values a and b as well as just above and just below a and b .

Input condition specifies a number of values,
Test cases should check the minimum and maximum numbers.
19

Boundary-value Analysis -
Guidelines



Guidelines 1 and 2 are applied to output domain as well.
A program generates temperature vs pressure table.
Test case should create output that produces maximum (and minimum) allowable number of table entries.
20

Boundary-value Analysis and
Equivalence Partitioning

One or more test cases be selected from the edge of the EC or close to the
EC.

Simply requires that any element in the equivalence class will do.

Test cases be derived from the output conditions.

Only the input domain is usually considered.
21

Key points

Equivalence partitioning is one of the oldest and still most widely used method for selecting test cases based on a partitioning of the input domain.

Boundary conditions are predicates that apply directly on, above, and beneath the input boundaries of input and output EC.
22