Today’s Agenda
Quiz 1
Quick Review
Continue on Graph-Based Testing
Software Testing and Maintenance 1
Quick Review
What is a basic block?
Software Testing and Maintenance 2
Graph-Based Testing
Introduction
Basic Concepts
Control Flow Testing
Data Flow Testing
Summary
Software Testing and Maintenance 3
Motivation
Graph-based testing first builds a graph model
for the program under test, and then tries to
cover certain elements in the graph model.
Graph is one of the most widely used
structures for abstraction.
Transportation network, social network, molecular
structure, geographic modeling, etc.
Graph is a well-defined, well-studied structure
Many algorithms have been reported that allow for
easy manipulation of graphs.
Software Testing and Maintenance 4
Major Steps
Step 1: Build a graph model
What information to be captured, and how to
represent those information?
Step 2: Identify test requirements
A test requirement is a structural entity in the graph
model that must be covered during testing
Step 3: Select test paths to cover those
requirements
Step 4: Derive test data so that those test
paths can be executed
Software Testing and Maintenance 5
Graph Models
Control flow graph: Captures information about
how the control is transferred in a program.
Data flow graph: Augments a CFG with data
flow information
Dependency graph: Captures the data/control
dependencies among program statements
Cause-effect graph: Modeling relationships
among program input conditions, known as
causes, and output conditions, known as effects
Software Testing and Maintenance 6
Graph-Based Testing
Introduction
Basic Concepts
Control Flow Testing
Data Flow Testing
Summary
Software Testing and Maintenance 7
Graph
A graph consists of a set of nodes and edges
that connect pairs of nodes.
Formally, a graph G = <N, N0, Nf, E):
N: a set of nodes
N0 N: a set of initial nodes
Nf N: a set of final nodes
E N N: a set of edges
In our context, N, N0, and Nf contain at least
one node.
Software Testing and Maintenance 8
Example
n0 n0 n1 n2
n1 n2 n3 n4 n5 n6
n3 n7 n8 n9
N = {n0, n1, n2, n3} N = {n0, n1, n2, n3 , n4, n5, n6 , n7, n8, n9}
N0 = {n0} N0 = {n0 , n1, n2}
Nf = {n3} Nf = {n7, n8, n9}
E = {(n0, n1), (n0, n2), (n1, n3), (n2, n3)} E = {(n0, n3), (n0, n4), (n1, n4), (n1, n5), …}
Software Testing and Maintenance 9
Path, Subpath, Test Path
A path is a sequence [n1, n2, …, nM] of nodes,
where each pair of adjacent nodes (ni, ni+1) is an
edge.
The length of a path refers to the number of edges in
the path
A subpath of a path p is a subsequence of p,
possibly p itself.
A test path is a path, possibly of length zero,
that starts at an initial node, and ends at a final
node
Represents a path that is executed during a test run
Software Testing and Maintenance 10
Reachability
A node n is syntactically reachable from node
n’ if there exists a path from n’ to n.
A node n is semantically reachable from node
n’ if it is possible to execute a path from n’ to n
with some input.
reach(n): the set of nodes and edges that can
be syntactically reached from node n.
Software Testing and Maintenance 11
Example
n0 n1 n2
n3 n4 n5 n6
n7 n8 n9
p1 = [n0, n3, n7]
p2 = [n1, n4, n8, n5, n1]
p3 = [n4, n8, n5]
reach(n0) = ?
reach(n5) = ?
Software Testing and Maintenance 12
SESE Graph
n0
n1 n2
n3
n4 n5
n6
Software Testing and Maintenance 13
Visit & Tour
A test path p is said to visit a node n (or an
edge e) if node n (or edge e) is in path p.
A test path p is said to tour a path q if q is a
subpath of p.
Software Testing and Maintenance 14
Test Case vs Test Path
a<b
n0 n1
a=b
a>b
n2 n3
t1: (a = 0, b = 1) => p1 = [n0, n1, n3, n2]
t2: (a = 1, b = 1) => p2 = [n0, n3, n2]
t3: (a = 2, b = 1) => p3 = [n0, n2]
Software Testing and Maintenance 15
Graph-Based Testing
Introduction
Basic Concepts
Control Flow Testing
Data Flow Testing
Summary
Software Testing and Maintenance 16
Basic Block
A basic block, or simply a block, is a sequence
of consecutive statements with a single entry
and a single exit point.
Control always enters a basic block at its entry
point, and exits from its exit point.
No entry, halt, or exit inside a basic block
If a basic block contains a single statement,
then the entry and exit points coincide.
Software Testing and Maintenance 17
Example
1. begin
2. int x, y, power;
3. float z; Block Lines Entry Exit
4. input (x, y);
5. if (y < 0) 1 2, 3, 4, 5 2 5
6. power = -y; 2 6 6 6
7. else 3 8 8 8
8. power = y;
9. z = 1; 4 9 9 9
10. while (power != 0) { 5 10 10 10
11. z = z * x; 6 11, 12 11 12
12. power = power – 1;
7 14 14 14
13. }
14. if (y < 0) 8 15 15 15
15. z = 1/z; 9 16 16 16
16. output (z);
17. end;
Software Testing and Maintenance 18
Function Calls
Should a function call be treated like a regular
statement or as a separate block of its own?
Software Testing and Maintenance 19
Control Flow Graph
A control flow graph is a graph with two
distinguished nodes, start and end.
Node start has no incoming edges, and node end has
no outgoing edges.
Every node can be reached from start, and can reach
end.
In a CFG, a node is typically a basic block, and
an edge indicates the flow of control from one
block to another.
Software Testing and Maintenance 20
Example
start
1
true false
2 3
false
5
true
6
7 true
false 8
end
Software Testing and Maintenance 21
Node Coverage
A test set T satisfies Node Coverage on graph
G if and only if for every syntactically reachable
node n in N, there is some path p in path(T) such
that p visits n.
path(T): the set of paths that are exercised by the
execution of T
In other words, the set TR of test requirements
for Node Coverage contains each reachable node
in G.
Software Testing and Maintenance 22
Edge Coverage
The TR for Edge Coverage contains each
reachable path of length up to 1, inclusive, in a
graph G.
Note that Edge Coverage subsumes Node
Coverage.
Software Testing and Maintenance 23
Node vs Edge Coverage
n0
n1
n2
Software Testing and Maintenance 24
Edge-Pair Coverage
The TR for Edge-Pair Coverage contains each
reachable path of length up to 2, inclusive, in a
graph G.
This definition can be easily extended to paths
of any length, although possibly with diminishing
returns.
Software Testing and Maintenance 25
Edge-Pair vs Edge Coverage
n0
a b
n1 n2
c d
n3
e f
n4 n5
g h
n6
Software Testing and Maintenance 26
Complete Path Coverage
The TR for Complete Path Coverage contain all
paths in a graph.
n0
n3
n1
n4
n2
How many paths do we need to cover in the above graph?
Software Testing and Maintenance 27
Simple & Prime Path
A path is simple if no node appears more than
once in the path, with the exception that the first
and last nodes may be identical.
A path is a prime path if it is a simple path,
and it does not appear as a proper subpath of
any other simple path.
Software Testing and Maintenance 28
Prime Path Coverage
The TR for Prime Path Coverage contains every
prime path in a graph.
Software Testing and Maintenance 29
Example
n0
n3
n1
n4
n2
Prime paths = {[n0, n1, n2], [n0, n1, n3, n4], [n1, n3,
n4, n1], [n3, n4, n1, n3], [n4, n1, n3, n4], [n3, n4, n1,
n2]}
Path (t1) = [n0, n1, n2]
Path (t2) = [n0, n1, n3, n4, n1, n3, n4, n1, n2]
T = {t1, t2}
Software Testing and Maintenance 30
Infeasible Test Requirements
The notion of “tour” is rather strict.
S0 a b d Sf
Let q = [a, b, d], and p = [S0, a, b, c, d, Sf].
Does path p tour path q?
Software Testing and Maintenance 31
Sidetrips/Detours
Tour: Test path p is said to tour path q if and
only if q is a subpath of p.
Tour with sidetrips: Test path p is said to tour
path q with sidetrips if and only if every edge in
q is also in p in the same order.
Tour with detours: Test path p is said to tour
path q with detours if and only if every node in q
is also in p in the same order
Software Testing and Maintenance 32
Example
1 2 5 6
S0 a b d Sf
3 4
1 2 5
S0 a b d Sf
4
c
Software Testing and Maintenance 33
Best Effort Touring
If a test requirement can be met without a
sidetrip (or detour), then it should be done so.
In other words, sidetrips or detours should be
allowed only if necessary.
Software Testing and Maintenance 34
Computing Prime Paths
Step 1: Find all the simple paths
Find all simple paths of length 0, extend them to
length 1, and then to length 2, and so on
Step 2: Select those that are maximal
Software Testing and Maintenance 35
Example
0 1
4 5
Software Testing and Maintenance 36
Example – Simple Paths (2)
len = 0 len = 1 len = 2 len = 3
1. [0] 8. [0, 1] 17. [0, 1, 2] 25. [0, 1, 2, 3]!
2. [1] 9. [0, 4] 18. [0, 1, 5] 26. [0, 1, 5, 6]!
3. [2] 10. [1, 2] 19. [0, 4, 6]! 27. [1, 2, 3, 1]*
4. [3] 11. [1, 5] 20. [1, 2, 3] 28. [2, 3, 1, 2]*
5. [4] 12. [2, 3] 21. [1, 5, 6]! 29. [2, 3, 1, 5]
6. [5] 13. [3, 1] 22. [2, 3, 1] 30. [3, 1, 2, 3]*
7. [6]! 14. [4, 4]* 23. [3, 1, 2] 31. [3, 1, 5, 6]
15. [4, 6]! 24. [3, 1, 5]
16. [5, 6]!
len = 4
32. [2, 3, 1, 5, 6]!
Software Testing and Maintenance 37
Example – Prime Paths
14. [4, 4]*
19. [0, 4, 6]!
25. [0, 1, 2, 3]!
26. [0, 1, 5, 6]!
27. [1, 2, 3, 1]*
28. [2, 3, 1, 2]*
30. [3, 1, 2, 3]*
32. [2, 3, 1, 5, 6]!
Software Testing and Maintenance 38
Example – Test Paths
Start with the longest prime paths and extend
them to the start and end nodes of the graph
1) [0, 1, 2, 3, 1, 5, 6]
2) [0, 1, 2, 3, 1, 2, 3, 1, 5, 6]
3) [0, 1, 5, 6]
4) [0, 4, 6]
5) [0, 4, 4, 6]
Software Testing and Maintenance 39
Graph-Based Testing
Introduction
Basic Concepts
Control Flow Testing
Data Flow Testing
Summary
Software Testing and Maintenance 40
Definition/Use
A definition is a location where a value for a
variable is stored into memory.
Assignment, input, parameter passing, etc.
A use is a location where a variable’s value is
accessed.
p-use: a use that occurs in a predicate expression,
i.e., an expression used as a condition in a branch
statement
c-use: a use that occurs in an expression that is used
to perform certain computation
Software Testing and Maintenance 41
Data Flow Graph
A data flow graph (DFG) captures the flow of
data in a program
To build a DFG, we first build a CFG and then
annotate each node n in the CFG with the
following two sets:
def(n): the set of variables defined in node n
use(n): the set of variables used in node n
Software Testing and Maintenance 42
Example (1)
1. begin
2. float x, y, z = 0.0;
3. int count;
Node Lines
4. input (x, y, count);
5. do { 1 1, 2, 3, 4
6. if (x <= 0) { 2 5, 6
7. if (y >= 0) {
8. z = y * z + 1; 3 7
9. } 4 8, 9, 10
10. }
5 11, 12,
11. else {
13
12. z = 1/x;
13. } 6 14, 15,
14. y = x * y + z; 16
15. count = count – 1; 7 17, 18
16. while (count > 0)
17. output (z);
18. end
Software Testing and Maintenance 43
Example (2)
1 def={x, y, z, count}
def={}
use = {x}
2
x <= 0 x>0
def={}
use = {y}
3 5 def={z}
y<0 use = {x}
y >= 0
def={z} 4 6
use = {y, z} def={count, y}
use = {count, x, y, z}
count == 0
7 def={}
use = {z}
Software Testing and Maintenance 44
DU-pair & DU-path
A du-pair is a pair of locations (i, j) such that a
variable v is defined in i and used in j.
Suppose that variable v is defined at node i,
and there is a use of v at node j. A path p = (i,
n1, n2, …, nk, j) is def-clear w.r.t. v if v is not
defined along the subpath n1, n2, …, nk.
A definition of a variable v reaches a use of v if
there is a def-clear path from the definition to
the use w.r.t. v.
A du-path for a variable v is a simple path from
a definition of v to a use of v that is def-clear
w.r.t. v.
Software Testing and Maintenance 45
Example
Consider the previous example:
Path p = (1, 2, 5, 6) is def-clear w.r.t variables x, y
and count, but is not def-clear w.r.t. variable z.
Path q = (6, 2, 5, 6) is def-clear w.r.t variables count
and y.
Path r = (1, 2, 3, 4) is def-clear w.r.t variables y and
z.
Software Testing and Maintenance 46
Notations
Def-path set du(n, v): the set of du-paths w.r.t
variable v that start at node n.
Def-pair set du(n, n’, v): the set of du-paths
w.r.t variable v that start at node n and end at
node n’.
Note that du(n, v) = n’ du(n, n’, v).
Software Testing and Maintenance 47
All-Defs Coverage
For each def-path set S = du(n, v), the TR for
All-Defs Coverage contains at least one path in S.
Informally, for each def, we need to tour at
least one path to at least one use.
Software Testing and Maintenance 48
All-Uses Coverage
For each def-pair set S = du(n, n’, v), the TR
for All-Uses Coverage contains at least one path
in S.
Informally, it requires us to tour at least one
path for every def-use pair.
Software Testing and Maintenance 49
All-DU-Paths Coverage
For each def-pair set S = du(n, n’, v), the TR
for All-DU-Paths Coverage contains every path in
S.
Informally, this requires to tour every du-path.
Software Testing and Maintenance 50
Example
all-defs
n0 def(0) = {x} 0-1-3-4
all-uses
n1 n2
0-1-3-4
0-1-3-5
n3
all-du-paths
0-1-3-4
use(4) = {x} n4 n5 use(5) = {x}
0-1-3-5
0-2-3-4
0-2-3-5
n6
Software Testing and Maintenance 51
Why data flow?
Consider the previous example. Assume that
there is a fault in line 14, which is supposed to be y
= x + y + z.
Does the following test set satisfy edge
coverage? Can the test set detect the above fault?
x y count
t1 -2 2 1
t2 -2 -2 1
t3 2 2 1
t4 2 2 2
Software Testing and Maintenance 52
Graph-Based Testing
Introduction
Basic Concepts
Control Flow Testing
Data Flow Testing
Summary
Software Testing and Maintenance 53
Subsumption Hierarchy
Complete Path
Coverage
Prime Path
Coverage
All-du-paths
Coverage Edge-pair
Coverage
All-Uses
Coverage Edge Coverage
All-Defs
Coverage Node Coverage
Software Testing and Maintenance 54
Recap
Graph provides a good basis for systematic
test selection.
Control flow testing focuses on the transfer of
control, while data flow testing focuses on the
definitions of data and their subsequent use.
Control flow coverage is defined in terms of
nodes, edges, and paths; data flow coverage is
defined in terms of def, use, and du-path.
Software Testing and Maintenance 55
