Equivalence Class Testing
Chapter 6
Boundary value problems
What problems does boundary value testing have?
ECT–2
Boundary value problems – 2
Generates test cases with
Serious gaps
Massive redundancy
ECT–3
Motivation for equivalence class testing
What are the motivations for equivalence class testing?
ECT–4
Motivation for equivalence class testing – 2
Avoid redundancy
Have fewer test cases
Complete testing
Remove gaps
ECT–5
Addressing the motivation
How do equivalence classes meet the motivations of
complete testing and avoiding redundancy?
ECT–6
Addressing the motivation– 2
The variable domain is partitioned into disjoint sub-sets
ECT–7
Assumptions
What assumptions are made?
ECT–8
Assumptions – 2
Program is a function from input to output
Input and/or output variables have well defined intervals
For a two-variable function F(x1,x2)
a ≤ x1 ≤ d, with intervals [a,b), [b,c), [c,d]
e ≤ x2 ≤ g, with intervals [e,f), [f,g]
ECT–9
Assumptions – 3
Completeness
The entire set is represented by the union of the sub-sets
Redundancy
The disjointness of the sets assures a form of non-
redundancy
ECT–10
Selecting test values
How are test values for a single variable selected?
ECT–11
Selecting test values – 2
How are test values for a single variable selected?
Choose one test value from each sub-set
ECT–12
Variations
What variations are used for equivalence class testing?
ECT–13
Variations – 2
Use the same two orthogonal dimensions as in boundary
value analysis
Robustness
Robust-normal
Distinguishes valid data from invalid data
Single/Multiple Fault Assumption
Weak-strong
Distinguishes single from multiple faults
Combinations give four variations.
ECT–14
Weak-Normal ECT
What is the number of test cases for weak-normal
testing?
ECT–15
Weak-Normal ECT – 2
Number of test cases =
max / [[ v : 1 .. #variables
• number_equivalence_classes (variable v) ]]
x2
e
x1
a b c d
ECT–16
Strong-Normal ECT
What is the number of test cases for strong-normal
testing?
ECT–17
Strong-Normal ECT – 2
Number of test cases =
× / [[ v : 1 .. #variables
• number_equivalence_classes (variablev) ]]
x2
e
x1
a b c d
ECT–18
Weak-Robust ECT
What is the number of test cases for weak-robust
testing?
ECT–19
Weak-Robust ECT – 2
Number of test cases =
max / [[ v : 1 .. #variables
• number_equivalence_classes (variablev) ]]
+
+/ [[ v : 1 .. #variables
• number_invalid_bounds (variablev) ]]
x2
f
Figure 6.3 in
e the textbook
is incorrect
x1
a b c d
ECT–20
Strong-Robust ECT
What is the number of test cases for strong-robust
testing?
ECT–21
Strong-Robust ECT – 2
Number of test cases =
× / [[ v : 1 .. #variables
• number_equivalence_classes (variablev)
+ number_invalid_bounds (variablev) ]]
x2
e
x1
a b c d
ECT–22
Limitations of ECT
What are the limitations of equivalence class testing?
ECT–23
Limitations of ECT – 2
The same as those for boundary value testing
Does not work well for Boolean variables
Does not work well for logical variables
When variables are not independent – i.e. are dependent
ECT–24
Limitations of ECT – 3
For robust variations same as for boundary value testing
Difficult or impossible to determine expected values for
invalid variable values
Not useful for strongly-typed languages
ECT–25
Triangle Output Equivalence Classes
Four possible outputs
Not a Triangle
Isosceles
Equilateral
Scalene
ECT–26
Triangle Output Equivalence Classes – 2
Output (range) equivalence classes
O1 = {a, b, c : 0 .. 200 | equilateral_triangle ( <a,b,c> ) }
O2 = {a, b, c : 0 .. 200 | isoceles_triangle ( <a,b,c> ) }
O3 = {a, b, c : 0 .. 200 | scalene_triangle ( <a,b,c> ) }
O4 = {a, b, c : 0 .. 200 | not_a_triangle ( <a,b,c> ) }
What is the number of test cases for weak-normal?
ECT–27
Triangle – Weak-Normal
Does the number of cases follow the formula in slide 16?
Expected
Test Case a b c
Output
WN1 5 5 5 Equilateral
WN2 2 2 3 Isosceles
WN3 3 4 5 Scalene
Not a
WN4 4 1 2
Triangle
ECT–28
Triangle Strong-Normal
Output (range) equivalence classes
O1 = {a, b, c : 0 .. 200 | equilateral_triangle ( <a,b,c> ) }
O2 = {a, b, c : 0 .. 200 | isoceles_triangle ( <a,b,c> ) }
O3 = {a, b, c : 0 .. 200 | scalene_triangle ( <a,b,c> ) }
O4 = {a, b, c : 0 .. 200 | not_a_triangle ( <a,b,c> ) }
What is the number of test cases for strong-normal?
ECT–29
Triangle – Strong-Normal = Weak-Normal
Does the number of cases follow the formula in slide 18?
Expected
Test Case a b c
Output
WN1 5 5 5 Equilateral
WN2 2 2 3 Isosceles
WN3 3 4 5 Scalene
Not a
WN4 4 1 2
Triangle
ECT–30
Triangle Weak-Robust
Output (range) equivalence classes
O1 = {a, b, c : 0 .. 200 | equilateral_triangle ( <a,b,c> ) }
O2 = {a, b, c : 0 .. 200 | isoceles_triangle ( <a,b,c> ) }
O3 = {a, b, c : 0 .. 200 | scalene_triangle ( <a,b,c> ) }
O4 = {a, b, c : 0 .. 200 | not_a_triangle ( <a,b,c> ) }
What are the number of test cases for weak-robust?
ECT–31
Triangle – Weak Robust Test Cases
Weak-normal cases + following error cases
Does the count follow the formula in slide 20?
Expected
Test Case a b c
Output
WR1 -1 5 5 a not in range
WR2 5 -1 5 b not in range
WR3 5 5 -1 c not in range
WR4 201 5 5 a not in range
WR5 5 201 5 b not in range
WR6 5 5 201 c not in range
ECT–32
Triangle Strong-Robust
Output (range) equivalence classes
O1 = {a, b, c : 0 .. 200 | equilateral_triangle ( <a,b,c> ) }
O2 = {a, b, c : 0 .. 200 | isoceles_triangle ( <a,b,c> ) }
O3 = {a, b, c : 0 .. 200 | scalene_triangle ( <a,b,c> ) }
O4 = {a, b, c : 0 .. 200 | not_a_triangle ( <a,b,c> ) }
What are the number of test cases for strong-robust?
ECT–33
Triangle – Strong-Robust Test Cases
#strong-normal = #weak-normal = 4
#Error cases = all combinations of errors in one or more of
a, b and c.
Each of a, b c have 3 values
too low • normal • too high
All combinations with at least one error
3^3 – 1 = 26 Remove normal-normal-normal
Total = 30
Triangle – input equivalence classes
What are the number
of test cases for
D1 = { a,b,c : 1..200 | a = b = c • <a,b,c> }
• weak-normal?
D2 = { a,b,c : 1..200 | a = b, a ≠ c • <a,b,c> } • strong-normal?
• weak-robust?
D3 = { a,b,c : 1..200 | a = c, a ≠ b • <a,b,c> }
• strong-robust?
D4 = { a,b,c : 1..200 | b = c, a ≠ b • <a,b,c> }
D5 = { a,b,c : 1..200 | a ≠ b, a ≠ c, b ≠ c • <a,b,c> }
D6 = { a,b,c : 1..200 | a ≥ b+c • <a,b,c> }
D7 = { a,b,c : 1..200 | b ≥ a+c • <a,b,c> } Is this a good set of
D8 = { a,b,c : 1..200 | c ≥ a+b • <a,b,c> } equivalence classes
to use or is there a
problem?
ECT–35
NextDate – naive equivalence classes
M1 = { month : 1 .. 12 } What are the number
D1 = { day : 1 .. 31 } of test cases for
• weak-normal?
Y1 = { year : 1812 .. 2012 } • strong-normal?
• weak-robust?
Invalid data
• strong-robust?
M2 = { month : Integer | month < 1 }
M3 = { month : Integer | month > 12 }
D2 = { day : Integer | day < 1 }
D3 = { day : Integer | day > 31 } What is the problem
Y2 = { year : Integer | year < 1812 } with using these
equivalence classes?
Y3 = { year : Integer | year > 2012 }
ECT–36
NextDate – improved equivalence classes
M1 = {month : 1 .. 12 | days(month) = 30 }
M2 = {month : 1 .. 12 | days(month) = 31 }
M3 = {month : {2} }
D1 = {day : 1 .. 28}
What is good and bad
D2 = {day : {29} } with using these
equivalence classes?
D3 = {day : {30} }
D4 = {day : {31} }
Y1 = {year : {2000} }
Y2 = {year : 1812 .. 2012 | leap_year (year) ∧ year ≠ 2000 }
Y3 = {year : 1812 .. 2012 | common_year (year) }
ECT–37
Weak Normal Test Cases
Expected
Test Case Month Day Year
Output
WN1 6 14 1900 6/15/1900
WN2 7 29 1996 7/30/1996
Invalid input
WN3 2 30 2002
date
Invalid input
WN4 6 31 1900
date
ECT–38
NextDate strong test cases
What are the number of test cases for
strong-normal testing?
What are the number of test cases for
strong-robust testing?
ECT–39
NextDate strong test cases – 2
There are 36 strong-normal test cases (3 x 4 x 3)
Some redundancy creeps in
Testing February 30 and 31 for three different types of
years seems unlikely to reveal errors
There are 150 strong-robust test cases (5 x 6 x 5)
ECT–40
Commission problem – input classes
L1 = {locks : 1 .. 70 } What are the number
of test cases for
L2 = {locks : { -1 } }
• weak-normal?
S1 = {stocks : 1 .. 80 }
• strong-normal?
B1 = {barrels : 1 .. 90} • weak-robust?
Invalid data • strong-robust?
L3 = {locks : Integer | locks ≤ 0 ∧ locks ≠ -1}
L4 = {locks : Integer | locks > 70 }
S2 = {stocks : Integer | stocks < 1 }
What is good and
S3 = {stocks : Integer | stocks > 80 } not good about
using these classes?
B2 = {barrels : Integer | barrels < 1 }
B3 = {barrels : Integer | barrels > 90 }
ECT–41
Commission problem – output classes
What are the number
Sales = 45 × locks + 30 × stocks + 25 × barrels of test cases for
• weak-normal?
• strong-normal?
S1 = {sales : 0 .. 1000 } • weak-robust?
S2 = {sales : 1001 .. 1800 } • strong-robust?
S3 = {sales : Integer | sales > 1800 }
Invalid data
What is good and
S4 = {sales : Integer | sales < 0} not good about
using these classes?
Figure 5.6, page 84 shows the
classes pictorially
ECT–42
Guidelines and observations
Equivalence Class Testing is appropriate when input data is
defined in terms of intervals and sets of discrete values.
Equivalence Class Testing is strengthened when combined
with Boundary Value Testing
Strong equivalence makes the presumption that variables
are independent.
If that is not the case, redundant test cases may be
generated
ECT–43
Guidelines and observations – 2
Complex functions, such as the NextDate program, are well-
suited for Equivalence Class Testing
Several tries may be required before the “right” equivalence
relation is discovered
If the equivalence classes are chosen wisely, the potential
redundancy among test cases is greatly reduced.
The key point in equivalence class testing is the choice of
the equivalence relation that determines the classes.
ECT–44
