Testing

Software testing
Main issues:
There are a great many testing techniques
Often, only the final code is tested

2
Nasty question
 Suppose you are being asked to lead the team to
test the software that controls a new X-ray
machine. Would you take that job?
 Would you take it if you could name your own
price?
 What if the contract says you’ll be charged with
murder in case a patient dies because of a mal-
functioning of the software?

3
Overview
 Preliminaries
 All sorts of test techniques
 Comparison of test techniques
 Software reliability

4
State-of-the-Art
 30-85 errors are made per 1000 lines of source
code
 extensively tested software contains 0.5-3 errors
per 1000 lines of source code
 testing is postponed, as a consequence: the later
an error is discovered, the more it costs to fix it.
 error distribution: 60% design, 40%
implementation. 66% of the design errors are not
discovered until the software has become
operational.

5
Relative cost of error correction
1
2
5
10
20
50
100
RE design code test operation

6
Lessons
 Many errors are made in the early phases
 These errors are discovered late
 Repairing those errors is costly
 ⇒ It pays off to start testing real early

7
How then to proceed?
 Exhaustive testing most often is not feasible
 Random statistical testing does not work either if
you want to find errors
 Therefore, we look for systematic ways to proceed
during testing

8
Classification of testing techniques
 Classification based on the criterion to measure
the adequacy of a set of test cases:
 coverage-based testing
 fault-based testing
 error-based testing
 Classification based on the source of information
to derive test cases:
 black-box testing (functional, specification-based)
 white-box testing (structural, program-based)

9
Some preliminary questions
 What exactly is an error?
 How does the testing process look like?
 When is test technique A superior to test
technique B?
 What do we want to achieve during testing?
 When to stop testing?

10
Error, fault, failure
 an error is a human activity resulting in software
containing a fault
 a fault is the manifestation of an error
 a fault may result in a failure

11
When exactly is a failure a failure?
 Failure is a relative notion: e.g. a failure w.r.t. the
specification document
 Verification: evaluate a product to see whether it
satisfies the conditions specified at the start:
Have we built the system right?
 Validation: evaluate a product to see whether it
does what we think it should do:
Have we built the right system?

12
Point to ponder: maiden flight of Ariane 5

13
Testing process
oracle
P
P
test
strategy
compare
input
subset of
input
subset of
input
expected
output
real
output
test
results

14
Test adequacy criteria
 Specifies requirements for testing
 Can be used as stopping rule: stop testing if 100%
of the statements have been tested
 Can be used as measurement: a test set that
covers 80% of the test cases is better than one
which covers 70%
 Can be used as test case generator: look for a test
which exercises some statements not covered by
the tests so far
 A given test adequacy criterion and the
associated test technique are opposite sides of
the same coin

15
What is our goal during testing?
 Objective 1: find as many faults as possible
 Objective 2: make you feel confident that the
software works OK

16
Example constructive approach
 Task: test module that sorts an array A[1..n]. A
contains integers; n < 1000
 Solution: take n = 0, 1, 37, 999, 1000. For n = 37,
999, take A as follows:
 A contains random integers
 A contains increasing integers
 A contains decreasing integers
 These are equivalence classes: we assume that
one element from such a class suffices
 This works if the partition is perfect

17
Testing models
 Demonstration: make sure the software satisfies
the specs
 Destruction: try to make the software fail
 Evaluation: detect faults in early phases
 Prevention: prevent faults in early phases
time

18
Testing and the life cycle
 requirements engineering
 criteria: completeness, consistency, feasibility, and testability.
 typical errors: missing, wrong, and extra information
 determine testing strategy
 generate functional test cases
 test specification, through reviews and the like
 design
 functional and structural tests can be devised on the basis of
the decomposition
 the design itself can be tested (against the requirements)
 formal verification techniques
 the architecture can be evaluated

19
Testing and the life cycle (cnt’d)
 implementation
 check consistency implementation and previous documents
 code-inspection and code-walkthrough
 all kinds of functional and structural test techniques
 extensive tool support
 formal verification techniques
 maintenance
 regression testing: either retest all, or a more selective retest

20
Test-Driven Development (TDD)
First write the tests, then do the
design/implementation
Part of agile approaches like XP
Supported by tools, eg. JUnit
Is more than a mere test technique; it subsumes
part of the design work

21
Steps of TDD
1. Add a test
2. Run all tests, and see that the system fails
3. Make a small change to make the test work
4. Run all tests again, and see they all run properly
5. Refactor the system to improve its design and
remove redundancies

22
Test Stages
 module-unit testing and integration testing
 bottom-up versus top-down testing
 system testing
 acceptance testing
 installation testing

23
Test documentation (IEEE 928)
 Test plan
 Test design specification
 Test case specification
 Test procedure specification
 Test item transmittal report
 Test log
 Test incident report
 Test summary report

24
Overview
 Preliminaries
 manual techniques
 coverage-based techniques
 fault-based techniques
 error-based techniques

25
Manual Test Techniques
 static versus dynamic analysis
 compiler does a lot of static testing
 static test techniques
 reading, informal versus peer review
 walkthrough and inspections
 correctness proofs, e.g., pre-and post-conditions: {P} S {Q}
 stepwise abstraction

26
(Fagan) inspection
 Going through the code, statement by statement
 Team with ~4 members, with specific roles:
 moderator: organization, chairperson
 code author: silent observer
 (two) inspectors, readers: paraphrase the code
 Uses checklist of well-known faults
 Result: list of problems encountered

27
Example checklist
 Wrong use of data: variable not initialized,
dangling pointer, array index out of bounds, …
 Faults in declarations: undeclared variable,
variable declared twice, …
 Faults in computation: division by zero, mixed-
type expressions, wrong operator priorities, …
 Faults in relational expressions: incorrect
Boolean operator, wrong operator priorities, .
 Faults in control flow: infinite loops, loops that
execute n-1 or n+1 times instead of n, ...

28
Overview
 Preliminaries

29
Coverage-based testing
 Goodness is determined by the coverage of the
product by the test set so far: e.g., % of
statements or requirements tested
 Often based on control-flow graph of the program
 Three techniques:
 control-flow coverage
 data-flow coverage
 coverage-based testing of requirements

30
Example of control-flow coverage
procedure bubble (var a: array [1..n] of integer; n: integer);
var i, j: temp: integer;
begin
for i:= 2 to n do
if a[i] >= a[i-1] then goto next endif;
j:= i;
loop: if j <= 1 then goto next endif;
if a[j] >= a[j-1] then goto next endif;
temp:= a[j]; a[j]:= a[j-1]; a[j-1]:= temp; j:= j-1; goto loop;
next: skip;
enddo
end bubble;
input: n=2, a[1] = 5, a[2] = 3
√
√
√
√
√
√
√
√
√
√
√
√

31
Example of control-flow coverage (cnt’d)
procedure bubble (var a: array [1..n] of integer; n: integer);
var i, j: temp: integer;
begin
for i:= 2 to n do
if a[i] >= a[i-1] then goto next endif;
j:= i;
loop: if j <= 1 then goto next endif;
if a[j] >= a[j-1] then goto next endif;
temp:= a[j]; a[j]:= a[j-1]; a[j-1]:= temp; j:= j-1; goto loop;
next: skip;
enddo
end bubble;
input: n=2, a[1] = 5, a[2] = 3
√
√
√
√
√
√
√
√
√
√
√
√
a[i]=a[i-1]

32
Control-flow coverage
 This example is about All-Nodes coverage,
statement coverage
 A stronger criterion: All-Edges coverage, branch
coverage
 Variations exercise all combinations of
elementary predicates in a branch condition
 Strongest: All-Paths coverage (≡ exhaustive
testing)
 Special case: all linear independent paths, the
cyclomatic number criterion

33
Data-flow coverage
 Looks how variables are treated along paths
through the control graph.
 Variables are defined when they get a new value.
 A definition in statement X is alive in statement Y
if there is a path from X to Y in which this variable
is not defined anew. Such a path is called
definition-clear.
 We may now test all definition-clear paths
between each definition and each use of that
definition and each successor of that node: All-
Uses coverage.

34
Coverage-based testing of requirements
 Requirements may be represented as graphs,
where the nodes represent elementary
requirements, and the edges represent relations
(like yes/no) between requirements.
 And next we may apply the earlier coverage
criteria to this graph

35
Example translation of requirements to a graph
A user may order new
books. He is shown a
screen with fields to fill
in. Certain fields are
mandatory. One field is
used to check whether
the department’s budget
is large enough. If so,
the book is ordered and
the budget reduced
accordingly.
Enter fields
All mandatory
fields there?
Check budget
Order book
Notify
user
Notify user

36
Similarity with Use Case success scenario
1. User fills form
2. Book info checked
3. Dept budget checked
4. Order placed
5. User is informed
Enter fields
All mandatory
fields there?
Check budget
Order book
Notify
user
Notify user

37
Overview
 Preliminaries

38
Fault-based testing
 In coverage-based testing, we take the structure
of the artifact to be tested into account
 In fault-based testing, we do not directly consider
this artifact
 We just look for a test set with a high ability to
detect faults
 Two techniques:
 Fault seeding
 Mutation testing

40
Mutation testing
procedure insert(a, b, n, x);
begin bool found:= false;
for i:= 1 to n do
if a[i] = x
then found:= true; goto leave endif
enddo;
leave:
if found
then b[i]:= b[i] + 1
else n:= n+1; a[n]:= x; b[n]:= 1
endif
end insert;
2
-
n-1

41
Mutation testing (cnt’d)
procedure insert(a, b, n, x);
begin bool found:= false;
for i:= 1 to n do
if a[i] = x
then found:= true; goto leave endif
enddo;
leave:
if found
then b[i]:= b[i] + 1
else n:= n+1; a[n]:= x; b[n]:= 1
endif
end insert;
n-1

42
How tests are treated by mutants
 Let P be the original, and P’ the mutant
 Suppose we have two tests:
 T1 is a test, which inserts an element that equals a[k] with
k<n
 T2 is another test, which inserts an element that does not
equal an element a[k] with k<n
 Now P and P’ will behave the same on T1, while
they differ for T2
 In some sense, T2 is a “better” test, since it in a
way tests this upper bound of the for-loop, which
T1 does not

43
How to use mutants in testing
 If a test produces different results for one of the
mutants, that mutant is said to be dead
 If a test set leaves us with many live mutants, that
test set is of low quality
 If we have M mutants, and a test set results in D
dead mutants, then the mutation adequacy score
is D/M
 A larger mutation adequacy score means a better
test set

44
Strong vs weak mutation testing
 Suppose we have a program P with a component
T
 We have a mutant T’ of T
 Since T is part of P, we then also have a mutant P’
of P
 In weak mutation testing, we require that T and T’
produce different results, but P and P’ may still
produce the same results
 In strong mutation testing, we require that P and
P’ produce different results

45
Assumptions underlying mutation testing
 Competent Programmer Hypothesis: competent
programmers write programs that are
approximately correct
 Coupling Effect Hypothesis: tests that reveal
simple fault can also reveal complex faults

46
Overview
 Preliminaries

47
Error-based testing
 Decomposes input (such as requirements) in a
number of subdomains
 Tests inputs from each of these subdomains, and
especially points near and just on the boundaries
of these subdomains -- those being the spots
where we tend to make errors
 In fact, this is a systematic way of doing what
experienced programmers do: test for 0, 1, nil, etc

48
Error-based testing, example
Example requirement:
Library maintains a list of “hot” books. Each new
book is added to this list. After six months, it is
removed again. Also, if book is more than four
months on the list, and has not been borrowed
more than four times a month, or it is more than
two months old and has been borrowed at most
twice, it is removed from the list.

49
Example (cnt’d)
2 4 6
5
2
av # of
loans
age
A B

50
Strategies for error-based testing
 An ON point is a point on the border of a subdomain
 If a subdomain is open w.r.t. some border, then an
OFF point of that border is a point just inside that
border
 If a subdomain is closed w.r.t. some border, then an
OFF point of that border is a point just outside that
border
 So the circle on the line age=6 is an ON point of both
A and B
 The other circle is an OFF point of both A and B

51
Strategies for error-based testing (cnt’d)
 Suppose we have subdomains Di, i=1,..n
 Create test set with N test cases for ON points of
each border B of each subdomain Di, and at least
one test case for an OFF point of each border
 This set is called N∗1 domain adequate

52
Application to programs
if x < 6 then …
elsif x > 4 and y < 5 then …
elsif x > 2 and y <= 2 then …
else ...

53
Overview
 Preliminaries

54
Comparison of test adequacy criteria
 Criterion A is stronger than criterion B if, for all
programs P and all test sets T, X-adequacy
implies Y-adequacy
 In that sense, e.g., All-Edges is stronger that All-
Nodes coverage (All-Edges “subsumes” All-
Nodes)
 One problem: such criteria can only deal with
paths that can be executed (are feasible). So, if
you have dead code, you can never obtain 100%
statement coverage. Sometimes, the subsumes
relation only holds for the feasible version.

55
Desirable properties of adequacy criteria
 applicability property
 non-exhaustive applicability property
 monotonicity property
 inadequate empty set property
 antiextensionality property
 general multiplicity change property
 antidecomposition property
 anticomposition property
 renaming property
 complexity property
 statement coverage property

56
Experimental results
 There is no uniform best test technique
 The use of multiple techniques results in the
discovery of more faults
 (Fagan) inspections have been found to be very
cost effective
 Early attention to testing does pay off

57
Overview
 Preliminaries

58
Software reliability
Interested in expected number of failures (not
faults)
… in a certain period of time
… of a certain product
… running in a certain environment

59
Software reliability: definition
Probability that the system will not fail during a
certain period of time in a certain environment

60
Failure behavior
Subsequent failures are modeled by a stochastic
process
Failure behavior changes over time (e.g. because
errors are corrected) ⇒ stochastic process is
non-homogeneous
µ(τ) = average number of failures until time τ
λ(τ) = average number of failures at time τ (failure
intensity)
λ(τ) is the derivative of µ(τ)

61
Failure intensity λ(τ) and mean failures µ(τ)

62
Operational profile
Input results in the execution of a certain sequence
of instructions
Different input ⇒ (probably) different sequence
Input domain can thus be split in a series of
equivalence classes
Set of possible input classes together with their
probabilities

63
Two simple models
Basic execution time model (BM)
 Decrease in failure intensity is constant over time
 Assumes uniform operational profile
 Effectiveness of fault correction is constant over time
Logarithmic Poisson execution time model (LPM)
 First failures contribute more to decrease in failure intensity
than later failures
 Assumes non-uniform operational profile
 Effectiveness of fault correction decreases over time

64
Estimating model parameters (for BM)

65
Summary
 Do test as early as possible
 Testing is a continuous process
 Design with testability in mind
 Test activities must be carefully planned,
controlled and documented.
 No single reliability model performs best
consistently

Testing

More Related Content

What's hot(20)

Viewers also liked(7)

Similar to Testing(20)

Recently uploaded(20)

Testing

Editor's Notes