Politehnica University of Timisoara Mobile Computing, Sensors Network and Embedded Systems Laboratory Embedded systems testing Testing design techniques instructor: Razvan BOGDAN
Outlines 1. The test development process 2. Categories of Test Design Techniques 3. Specification-based or Black-box Techniques 4. Structure-based or White-box Techniques 5. Experience-based Techniques 6. Choosing Test Techniques
THE TEST DEVELOPMENT PROCESS • Deriving test cases from requirements • Designing test cases must be a controlled process. • Test cases can be crated in a formal way or in an informal way, depending on the project delimitations and on the maturity of the process that is in use. Test scenarios test case 6 test case 5 test case 4 test object and requirements on the test object delivering test requirements and test criteria test case 6 test case 5 test case 4 test case 3 test case 2 test case 1
THE TEST DEVELOPMENT PROCESS • Traceability • Tests should be traceable: which test case was included in the test portfolio based on which requirement? • The consequences of changes in the requirements on the tests to be made can be identified directly. • Traceability also helps to determine requirement coverage. Test scenarios test case 6 test case 5 test case 4 test object and requirements on the test object delivering test requirements and test criteria test case 6 test case 5 test case 4 test case 3 test case 2 test case 1
THE TEST DEVELOPMENT PROCESS • Definitions • Test object: The subject to be examined: a document or a piece of software in the software development process. • Test condition: An item or an event: a function, a transaction, a quality criterion or an element in the system. • Test criteria: The test object has to confirm the test criteria in order to pass the test. • Test execution schedule: A scheme for the execution of the test procedures. The test procedures are included in the test execution schedule in their context and in their order in which they are to be executed.
THE TEST DEVELOPMENT PROCESS • Test case description according to IEEE 829 • Preconditions: situation previous to test execution or characteristics of the test object before conducting the test case. • Input values: description of the input data on the test object. • Expected results: output data that the test object is expected to produce. • Post conditions: characteristics of the test object after test execution, description of its situation after the test. • Dependencies: order of execution of the test cases, reason for dependencies. • Distinct identification: Id or key in order to link, for example, an error report to the test case where it appeared. • Requirements: characteristics of the test object that the test case will examine.
THE TEST DEVELOPMENT PROCESS • Combining test cases • Test cases may be combined to test suites and test scenarios. • A test procedure specification: defines the sequence of actions for the execution of a single test case or a test suite. It is a script or screenplay for the test describing the steps, handlings and/or activities required for test execution. • With the use of adequate tools, test suites can be coded to be executed automatically. • The (dynamic) test plan states the sequence of tests planned, who is to do them and when. Constraints that have to be considered are priorities, availability of resources, test infrastructure etc. • The test execution schedule defines the order of the execution of test procedures and automation test scripts using for prioritization, regression tests etc.
THE TEST DEVELOPMENT PROCESS • Summary • Test cases and test suites are derived from the test object requirements or characteristics • Components of a test case description are: • key/Id • input values • pre-conditions • expected results • post-conditions • dependencies • requirements out of which the test case was derived
Outlines 1. The test development process 2. Categories of Test Design Techniques 3. Specification-based or Black-box Techniques 4. Structure-based or White-box Techniques 5. Experience-based Techniques 6. Choosing Test Techniques
CATEGORIES OF TEST DESIGN TECHNIQUES • Black box and white box testing black box Equivalence partitioning Boundary value analysis State transition testing Decision tables Use case based testing experience-based techniques white box dynamic White-Box Statement coverage Branch coverage Condition coverage Path coverage Reviews / walkthroughs static Black-Box Analytical QA • Dynamic testing is divided in two categories/groups • The grouping is done on the bases of the method to derive test cases • Every group has its own set of methods for designing test cases Control flow analysis Data flow analysis Compiler metrics/analyzer
CATEGORIES OF TEST DESIGN TECHNIQUES • Specification-based or Black-box Techniques • The tester looks at the test object as a black box • Internal structure of the test object is irrelevant or unknown • Test cases are derived/selected based on specification analysis (functional and non-functional) of a component system • Testing of input/output behaviour • Functionality is the focus of attention! • Black-box technique is also called functional testing or specification oriented testing Test case design Test cases on the basis of specifications Internal program structure irrelevant Test of all selected resp chosen combinations of input/output data Test object Black-Box
CATEGORIES OF TEST DESIGN TECHNIQUES • Structure-bases or White-box Techniques • The tester knows the internal structure of the program/code • i. e. component hierarchy, control flow, data flow, etc. • Test cases are selected on the basis of internal program code/program structure • During testing, interference with the test execution is possible • Structure of the program is the focus of attention! • White-box technique is also called structure-based testing or control flow based testing Test case design Test cases on the basis of program structure Test process controlled externally Analyze control flow within the test object during test execution
CATEGORIES OF TEST DESIGN TECHNIQUES • Categories of test design methods – overview • Specification-based methods • Test objects have been selected in accordance with the functional software model. • The coverage of specification can be measured (e. g. which percentage of specification covered by test cases). • Structure-based methods • The internal structure of the test object is used to design test cases (code/statements, menus, calls, etc. ) • The coverage percentage is measured and used as a bases for creating additional test cases • Experience-bases methods • Knowledge and experience about the test object and its environment are the bases for designing test cases • Knowledge and experience about possible weak spots, probable errors and former errors are used to determine test cases.
CATEGORIES OF TEST DESIGN TECHNIQUES • Summary • Test cases can be designed using different methods • If specification functionality is the focus of testing, the methods used are called specification oriented methods or black-box methods. • If the internal structure of an object is investigated, the methods used are called structure oriented methods or white-box methods. • Experience based methods use knowledge and skills of the personnel involved in test case design.
Outlines 1. The test development process 2. Categories of Test Design Techniques 3. Specification-based or Black-box Techniques 4. Structure-based or White-box Techniques 5. Experience-based Techniques 6. Choosing Test Techniques
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Overview • The following black-box methods will be explained in detail: • Equivalence partitioning • Boundary value analysis • Decision table testing and cause-effect graphing • State transition testing • Use case testing • This accounts for the most important and popular methods • Other black box methods are for instance • Statistical testing • Pairwise testing • Smoke testing
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • General • Functional testing is targeted at verifying the correctness and the completeness of a function • Are all specified function available within the module? • Do executed functions give correct results? • The execution of the test cases should be done without high redundancy, but nevertheless comprehensive • Test as little as possible but • Test as much as necessary
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning • Equivalence class partitioning is what most testers do intuitively: they divide the possible values into classes. Hereby they look at • Input values of a program (usual use of EC-method) • Output values of a program (rarely used EC-method) • The range of defined values is grouped into equivalence classes, for which the following rules apply: • All values, for which a common behaviour of the program is expected, are grouped together in one equivalence class • Equivalence classes may not overlap and may not contain any gaps • Equivalence classes may contain a range of values: (e. g. 0<x<10) or a single value (e. g. x = “Yes”)
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – valid and invalid • The equivalence classes of each variable (element) may be divided further • Valid EC: all values within the definition range are combined into one equivalence class, if they are handles identically by the test object • Invalid EC: we distinguish two cases for values outside of the definition range: • Values with a correct format but outside of the value range can be combined into one or more equivalence classes • Values with a wrong format generally build a separate EC • Tests are performed using a single representative from each EC • For every other value from the EC the same behaviour as for the chosen value is expected
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – example • Equivalence classes are chosen for valid and invalid inputs • If a value x is defined as 0 ≤ x ≤ 100, then we can initially identify three equivalence classes: 1. x<0 (invalid input values) 2. 0 ≤ x ≤ 100 (valid input values) 3. X > 100 (invalid input values) • Further invalid EC can be defined, containing, but not limited to: • Non-numerical inputs • Numbers too big or too small • Non-supported format for numbers
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – input variables • All input values (elements) of the test objective were identified e. g. • Fields of a GUI (e. g. system test) • Parameters of a function (e. g. component test) • A range for each input value is defined • this range defines the sum of all valid Equivalence Classes (v. EC) • Invalid equivalence classes (i. EC) result from the values outside of this range • values expected to be handled differently (known or suspected) are assigned to a different separate equivalence class.
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – example • A program expects a percentage value according to the following requirements: • Only integer values are allowed • 0 is the valid lower boundary of the range • 100 is the valid upper boundary of the range • Valid are all the numbers from 0 to 100, invalid are all the negative numbers, all numbers grater than 100, all decimal numbers and all non numerical values (e. g. “fred”) • one valid equivalence class: 0 ≤ x ≤ 100 • 1 st invalid equivalence class: 0 < x • 2 nd invalid equivalence class: x > 100 • 3 rd invalid equivalence class: x = no integer • 4 th invalid equivalence class: x = non numeric (n. n. )
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – example • The percentage value will now be displayed in a bar chart. The following additional requirements apply (both values included): • Values between 0 and 15: grey bar • Values between 16 and 50: green bar • Values between 51 and 85: yellow bar • Values between 86 and 100: red bar • Now there are four instead of one valid equivalence classes: • 1 st valid equivalence class: 0 ≤ x ≤ 15 • 2 nd valid equivalence class: 16 ≤ x ≤ 50 • 3 rd valid equivalence class: 51 ≤ x ≤ 85 • 4 th valid equivalence class: 86 ≤ x ≤ 100 <0 0 -15 16 -50 51 -85 86 -100 >100
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – picking representatives • On the last step, one representative of each EC is determined as well as the result to be expected for it Variable Equivalence class Percentage value (valid) EC 1: 0≤x≤ 15 +10 EC 2: 16≤x≤ 50 +20 EC 3: 51≤x≤ 85 +80 EC 4: 86≤x≤ 100 +90 EC 5: x<0 -10 Percentage value (invalid) EC 6: x>100 Representatives +200 EC 7: x no integer 1. 5 EC 8: x non numeric fred
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Exercise: Equivalence Partitioning • From a given specification (online shop) please extract: • All input values • Equivalence classes for each of the input values • Valid equivalence classes • Invalid equivalence classes
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – example 2/1 • Analysing the specification • A piece of code computes the price of a product based on its value, a discount in % and shipping costs (6, 9 or 12 EURO, depending on shipping mode) Variable Equivalence class Value of goods EC 11: x≥ 0 valid 1000. 00 EC 12: x<0 invalid -1000. 00 EC 13: x non-numerical value invalid fred valid 10% EC 22: x<0% invalid -10% EC 23: x>100% invalid 200% EC 24: x non-numerical value invalid fred EC 31: x=6 valid 6 EC 32: x=9 valid 9 EC 33: x=12 valid 12 EC 34: x={6, 9, 12} invalid 4 EC 35: x non-numerical value invalid fred Discount Shipping costs EC 21: 0%≤x≤ 100% Status Representative • Assumptions: • Value of goods is give as a positive number with 2 decimal places • Discount is a percentage value without decimal places between 0% and 100%. • Shipping costs can only be 6, 9 or 12
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – example 2/2 • Test cases for valid EC: • Valid equivalence classes provide the following combinations or test cases: T 01, T 02 and T 03 Variable Equivalence class Value of goods EC 11: x≥ 0 T 01 T 02 T 03 valid 1000. 00 * * * EC 12: x<0 invalid -1000. 00 invalid fred valid 10% * * * EC 22: x<0% invalid -10% EC 23: x>100% invalid 200% EC 24: x non-numerical value Shipping costs Representative EC 13: x non-numerical value Discount invalid fred EC 31: x=6 valid 6 EC 32: x=9 valid 9 EC 33: x=12 valid 12 EC 34: x={6, 9, 12} invalid 4 EC 35: x non-numerical value invalid fred EC 21: 0%≤x≤ 100% Status * * *
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – example 2/3 • Test cases for invalid EC: • The following test cases were created using the invalid EC, each in combination with valid ECs of other elements: Variable Equivalence class Value of goods EC 11 valid 1000. 00 EC 12 invalid -1000. 00 invalid fred EC 21 valid 10% EC 22 invalid -10% EC 23 invalid 200% EC 24 Shipping costs Representative EC 13 Discount Status invalid fred EC 31 valid 6 EC 32 valid 9 EC 33 valid 12 EC 34 invalid 4 EC 35 invalid fred T 04 T 05 T 06 T 07 T 08 T 09 T 10 * * * *
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – example 2/4 • Test cases for invalid EC: • 10 test cases are derived: 3 positive (valid values) and 7 negative (invalid values) test cases: Variable Status Representative T 01 T 02 T 03 Value of goods valid 1000. 00 * * * invalid -1000. 00 invalid fred valid 10% invalid -10% invalid 200% invalid fred valid 6 valid 9 valid 12 invalid 4 invalid fred Discount Shipping costs T 04 T 05 T 06 T 07 T 08 T 09 T 10 * * * * * * *
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – output based • Equivalence classes can also be built, based on the expected output values • The method is used in analogy, applied to the output values • The variable (element) is thus the output (for example, a field values on the GUI) • Equivalence classes are built for all possible outputs defined • A representative is determined for each equivalence class of output values • The input leading to the representative value is than acquired • A higher costs and effort because the input values have to be found for the given output recursively
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – in general /1 • Partitioning • The quality of the test depends on precisely segmented variables/elements in equivalence classes • EC that were not identified hold the risk of overlooking possible defects, since the representatives used did not cover all possibilities • Test cases • Equivalence class method provides test cases for which a representative still has to be chosen • Test data combinations are selected by defining the representative or representatives of each equivalence class
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – in general /2 • Choosing representatives • Any value within EC can be a representative. Optimal are: • Typical values (used often) • Problem values (suspected failures) • Boundary values (on the edge of the EC) • Representatives of valid EC may be combined • Representatives of invalid EC may not be combined • Representatives of invalid EC may only be combined with valid representatives of other EC • For test cases, representatives of invalid EC should be combined with always the same values of other valid EC (standard combinations) • Choosing representatives implies that the function within the program uses compare operations
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – coverage • Equivalence class coverage can be used as exit criteria to end testing activities • EC – Coverage = Number of EC tested / Number of EC defined * 100
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – transition • The transition from the specification or definition of functionality to the creation of equivalence classes • Often a difficult task due to the lack of precise and complete documentation • Boundaries that are not defined of missing descriptions make it difficult to define equivalence classes • Often, contact with the customer is needed to complete information
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Equivalence Partitioning – benefits • Benefits • Systematic test case design method, i. e. with a minimum amount of test case a defined coverage can be expected • Partitioning value ranges in equivalence classes on the basis of specifications covers the functional requirements • Prioritizing equivalence classes can be used to prioritize test cases (Inputs that are used rarely are to be tested last) • Test of known exceptions is covered by test cases on the basis of negative equivalence classes • Equivalence partitioning is applicable at all levels of testing • Can be used to achieve input and output coverage goals • Can be applied to human input or via interfaces to a system or interface parameters in integration testing
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Boundary analysis /1 • Boundary analysis extends equivalence class partitioning by introducing a rule for the choice of representatives • The edge values of the equivalence class are to be tested intensively • Why put more attention to the edges? • Often, the boundaries of value ranges are not well defined or lead to different interpretations • Checking if the boundaries were programmed correctly • Please note: • Experience shows that errors occur very frequently on the boundaries of values ranges! • Boundary value analysis can be applied at all test levels. It is easy to apply and its defect finding capability is high on base of detailed specifications.
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Boundary analysis /2 • Boundary analysis assume that: • The equivalence class is composed of a continuous value range (not a single value or a group of discrete values ) • That boundaries can be defined for the value range • As an extension to equivalence class partitioning, boundary value analysis is a method suggesting the choice of representatives • Equivalence class partitioning: • Examines one (typical)value of the equivalence class • Boundary analysis • Examines the boundaries and the neighbouring values • Uses the following scheme: Value range: Bottom Value ≤ x ≤ Top Value bottom value –δ Lower boundary +δ Top value –δ Higher boundary Top value +δ δ is the smallest step defined for the value For example: 1 - for integer values
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Defining boundary values • Basic scheme can only be applied when the value range was defined accordingly • In this case additional testing is needed for a valued in the middle of the value range • If the EC is defined as single numerical value, for example x=5, the neighbouring value will be used as well • The representatives (of the class and its neighbouring values) are: 4, 5 and 6
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Boundary analysis for invalid EC • Boundary value make little sense for invalid equivalence classes • Representatives of the invalid EC at the boundary to a valid EC are already covered through the basic schema • For value ranges defined as a set of values, no boundaries can be created generally • For example: single, married, divorced, widowed
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Boundary analysis example 3 a • Example 3 a: • Value range for a discount in %: 0. 00 ≤ x≤ 100. 00 • Definition of EC • 3 classes: • EC: x <0 • EC: 0. 00 ≤ x≤ 100. 00 • EC: x > 100 • Boundary analysis • Extends the representative to: • 2. EC: -0. 01; 0. 00; 0. 01; 99. 99; 100. 00; 100. 01 • Please note • Instead of one representative for the valid EC, there are now six representatives (four valid and two invalid)
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Boundary analysis example 3 b • Basic scheme: choose three values to be tested – the exact boundary and two neighbouring values (within and outside the EC) • Alternative point of view: since the boundary value belongs to the EC, only two values are needed for testing: one within and one outside the EC • Example 3 b • value range for a discount in %: 0. 00≤ x ≤ 100. 00 • valid EC: 0. 00≤ x ≤ 100. 00 • Boundary analysis additional representatives are: -0. 01; 0. 00; 100. 01 • 0. 01 – same behaviour as 0. 00 • 99. 99 – same behaviour as 100. 00 • A programming error caused by a wrong comparison operator will be found with the two boundary values
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Decision table testing /1 • Equivalence class partitioning and boundary analysis deal with isolated input conditions • However, an input condition may have an effect only in combination with other input conditions • All previously described methods do not take into account the effects of dependencies and combinations • Using the full set of combination of all input equivalence classes usually leads to a very high number of test cases (test case explosion) • With the help of cause-and-effect graphs and the decision tables derived from them, the amount of possible combinations can be reduced systematically to a subset
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Decision table testing /2 • A cause-and-effect diagram use a formal language • A cause-and-effect diagram is created by translating the (mostly informal) specification of a test object into a formal language • The test object is subject to a certain amount of effect that are traced back to their respective causes • Elements 7 symbols: • Assertion (if cause A – then effect E) • Negation (if cause A – then no effect E) • Or (if cause A or B – then effect E) • And (if cause A and B – then effect E)
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Decision table testing /3 • Other elements of a cause-and-effect diagram: • Exclusive (Either cause A or cause B) • Inclusive (At least one of two causes: A or B) • One and only one (One and exactly one of two causes: A or B) • Required (If cause A then also cause B)
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Decision table testing /4 • Example 5: Online-Banking • The user has identified himself via account number and PIN. If having sufficient coverage, he is able to set a transferal. To do this, he must input the correct details of the recipient and a valid TAN.
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Decision table testing /5 • Example 5: Online-Banking T 01 T 02 T 05 Enough coverage Yes No - - Yes - No - Valid TAN Yes - - No Do transferal Yes No No No Mark TAN as used Yes No No No Deny transferal No Yes No Request again TAN Activities (Effects) T 04 Correct recipient Preconditions (Causes) T 03 No No No Yes • Each table column represents a test case • Creating a decision table: • Choose an effect • Trace back along the diagram to identify the cause • Each combination of causes represent a column of the decision table (a test case) • Identical combinations of causes, leading to a different effects, may merged, to form a single test cases.
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Decision table testing /6 • Practical use • The specification is divided into manageable parts, thus leading to decision tables of a practical size • It is difficult to deduce boundary values out of a cause-and-effect diagram or decision table • It is recommended to combine test cases derived from decision tables with values derived from boundary analysis • The number of causes and effects that are examined will determine the complexity of the cause-and-effect diagram: for n preconditions that may be true or false, 2 n test cases can be created • On systems of larger size, this method is only manageable with tool support
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Decision table testing /7 • Benefits • Systematical identification of input combinations (combined causes) that might not be found using other methods • Test cases are easily derived from the decision table • Easy to determine sufficient test case coverage, e. g. at least one test case created for each column of the decision table • The number of test cases can be reduced by systematically merging columns of the decision table • Drawbacks • Setting up a large number of causes leads to complex and extensive results • Thus, many errors can occur when applying this method • This makes it necessary to use a tool
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • State Transition Testing /1 • Many methods only take into account the system behaviour in terms of input data and output data • Different states that a test object might take on are not taken into account • For example, results of actions that happened in the past- actions that caused the test object to be in a certain internal state • The different states that a test object can take on are modelled using state transition diagrams • State transition analysis is used to define state transition based test cases • State transition testing is much used within embedded software industry and technical automation in general
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • State Transition Testing /2 • To determine test cases using a state transition diagram, a transition tree is constructed: • The initial state is the root of the tree • For every state that may be reached from the initial state, a node is created which is connected to the root by a branch • This operation is repeated and comes to an end if • The state of the node is an end state (a leaf of the tree) Or • The same node with the same state is already part of the tree dead married “to marry” dead “to die” divorced widowed married “to die” married “d. o. p ” “to marry” dead “to die” “to marry” single dead “to die” “be single” unborn Event “d. o. p. ”: death of partner
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • State Transition Testing /3 • Every path from the root to a leaf then represents a test case for state transition testing • The state transition tree for this example leads to the following six test cases: State 1 State 2 State 3 State 4 State 5 End state unborn single dead unborn single married widowed married unborn single married divorced dead unborn leidg married divorced married D D M M V W M D S D dead U
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES dead married “to marry” dead “to die” married “to die” widowed divorced “to marry” “d. o. p” “getting divorced” married dead “to die” “to marry” error “to die” single dead “getting divorced” “be single” “d. o. p” • The transition tree of our example may now be extended using invalid transitions (negative test cases, robustness testing). • Example: two possible invalid transitions – there are more • Impossible transitions between states can not be tested. error unborn Event “d. o. p. ”: death of partner
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • State Transition Testing – Summary • Test exit criteria • Every state has to be entered at least once • Every transitions has to be executed at least once • Benefits/Drawbacks of this method • Good testing method for test objects than can be described as state machines • Good testing method to test classes, only if the object life cycle is available • Very often, states are rather complex, i. e. a lot of parameters are necessary to describe the state • Designing test cases and analysing test results can be difficult and time consuming in this cases • Only covering all states does not guarantee complete test coverage
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Use Case Testing /1 • Test cases are derived directly from the use cases of the test object • The test object is seen as a system reacting with actors • A use case describes the interaction of all involved actors leading to an end result of the system • Every use case has pre-conditions that have to be met in order to execute the use case (the test case) successfully • Every use case has post-conditions describing the system after execution of the use case (the test case) • Use cases are elements of the Unified Modelling Language UML* • Use case diagrams are one of 13 different types of diagrams used by UML • A use case diagram is a diagram describing a behaviour, it does not describe the sequence of events • It shows the system reaction from the viewpoint of a user *: * UML is a non-proprietary specification language for object modeling
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Use Case Testing /2 • Example of a simple use case diagram (source: Wikipedia) • The diagram on the left describes the functionality of a sample Restaurant System. • Use cases are represented by ovals and the actors are represented by stick figures • The Patron actor can Eat Food, Pay for Food, or Drink Wine. • Only the Chef actor can Prepare Food. Note that both the Patron and the Cashier are involved in the Pay for Food use case. • The box defines the boundaries of the Restaurant System, i. e. the use cases shown are part of the system being modelled, the actors are not.
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Use Case Testing /3 • Every use case describes a certain task (user-system-interaction) • Use-case descriptions include, but are not limited to: • Pre-conditions • Expected results/System behaviour • Post-conditions • These descriptive elements are also used to define the corresponding test cases. • Every use case may be used as the basis for a test case. • Every alternative within the diagram corresponds to a separate test case. • Typically, information provided with a use case has not enough detail to define the test cases directly. Additional data is needed (input data, expected results) to make up the test case.
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Use Case Testing –Summary • Benefits • Well suited for acceptance testing and system testing, because each use case describes a user scenario to be tested • Useful for designing acceptance tests with customer/user participation • Well suited if system specifications are available in UML • May be combined with other specification-based test techniques • Drawbacks • No derivation of additional test cases beyond the information provided by the use case • Therefore, this method should be used only combined with other methods of systematic test case design
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Specification-based or Black-box techniques- General conclusions /1 • Testing system functionality is the main goal of black box testing • Therefore, the test result depends on the quality of the system specification (e. g. completeness, missing or wrong specifications lead to bad test cases) • If specifications are wrong, tests will be wrong too. Testing is only performed for described functions. Missing specification of required functionality will not be discovered during testing. • If the test object holds functions that have not been specified, they will not be examined • Such superfluous functions may cause problems in the area of stability and security (e. g. software for automated teller machines)
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Specification-based or Black-box techniques- General conclusions /2 • In spite of these drawbacks, functional testing is still the most important testing activity • Black-box methods are always used in testing • The drawbacks can be compensated using additional methods of test case design, e. g. white box testing or experience based testing
SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES • Specification-based or Black-box techniques - Summary • Black-box methods: • Equivalence class partitioning • Boundary analysis • Cause Effect graphing and decision tables • State-transition testing • Use case based testing • Black-box testing verifies the specified functions: if functions are not specified, they are not tested. • Additional code (i. e. code that should not be there) cannot be detected using black box testing
Outlines 1. The test development process 2. Categories of Test Design Techniques 3. Specification-based or Black-box Techniques 4. Structure-based or White-box Techniques 5. Experience-based Techniques 6. Choosing Test Techniques
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Structure-based or white box techniques /1 • The following techniques will be explained in detail: • Statement Testing and Coverage • Decision Testing and Coverage (=branch) • Condition Testing and Coverage • Path Testing Coverage • Remark • These techniques represent the most important and the most widely used dynamic testing techniques. They relate to the static analysis techniques which were described earlier. • Other white box testing techniques include but are not limited to: • LCSAJ (Linear Code Sequence and Jump) • Data flow based techniques
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Structure-based or white box techniques /2 • It is based on an identified structure of the software or the system • Component-level: the structure of a software component, i. e. statements, decisions, branches, distinct paths • Integration-level: the structure may be a call tree (a diagram in which modules call other modules) • System-level: the structure may be a menu structure, business process or web page structure • Three code-related structural test design techniques for code coverage based on statements, branches and decisions.
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Structure-based or white box techniques -Tools /1 • During white-box testing, the program under test is executed, same as in black box testing. Both make up dynamic testing. • Theory states that all parts of a program should be executed at least once during testing. • The rate of coverage of the program is measured using tools (e. g. coverage analysers): • Code instrumentation is performed in order to count execution paths, i. e. counters are inserted into the program code of the test object • These counters are initialized holding the value of zero, each execution path increments the respective counter. • Counters that remain zero after testing indicate program parts that have not been executed.
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Structure-based or white box techniques -Tools /2 • White-box techniques need the support of tools in many areas, there are: • Test case specification • Automatically generating a control flow graph from program source code • Test execution • Tools to monitor and control the program flow inside the test objects • Tool support ensures the quality of the tests and increases efficiency • Because of the complexity of the necessary measures for white box testing, manual test execution is • Time consuming, resource consuming • Difficult to implement and prone to errors
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • The main types of coverage • Statement Testing and Coverage • The percentage of executable statements that have been exercised by the test cases • Can also be applied to modules, classes, menu points etc. • Decision Testing and Coverage (=branch coverage) • The percentage of decision outcomes, that have been exercised by the test cases • Path Testing and Coverage • The percentage of execution paths, that have been exercised by the test cases • Condition Testing and Coverage • The percentage of all single condition outcomes independently affecting a decision outcome, that have been exercised by the test cases • Condition coverage comes in various degrees, e. g. single, multiple and minimal multiple condition coverage.
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Statement Testing and Coverage • The statement of the program code is the focus of attention • What test cases are necessary in order to execute all (or a certain percentage) of the existing code statements? • Basis of this investigation in the control flow graph • All instructions are represented as nodes and the control flow between the instructions is represented as an edge (arrow) • Multiple instructions are combined in a single node if they can only be executed in one particular sequence • Aim of the test (test exit criteria) is to achieve the coverage of selected percentage of all statements, called the statement coverage (C 0 code coverage). • Statement coverage (C 0) = Number of executed statements / Total number of statements * 100%
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Statement Testing and Coverage – Example 1/1 • We are assessing the following segment of program code, which is represented by the control flow graph (see right side): if(i>0) { j=f(i); if(j>10){ while(k>10){ … } } }
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Statement Testing and Coverage – Example 1 /2 • Consider the program represented by the control flow graph on the right • Contains two if-statements and a loop (do-while) inside the second if-statement • There are three different “routes” through the program segment • The first if-statements allows two directions • The right-hand direction of the first if-statement is divided again using the second if-statement • All the statements of this program can be reached using the route to the right • A single test case will be enough to reach 100% statement coverage
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Statement Testing and Coverage – Example 2 • In this example the graph is slightly more complex: • The program contains the if-statements and a loop (inside one if statement) • Four different “routes” lead through this program segment • The first if-statement allows two directions • In both branches of the if-statement another ifstatement allows for again two different directions • For a 100% statement coverage four test cases are needed
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Statement Testing and Coverage – General Conclusions • The measurement of coverage is done using specifically designed tools • These tools are called Coverage Analysis Tools or Coverage Analysers • Benefits/drawbacks of this method • Dead code, that is, code made up of statements that are never executed, will be discovered • If there is dead code within a program, a 100% coverage cannot be achieved • Missing instructions, that is, code which is necessary in order to fulfil the specification, cannot be detected • Testing is only done with respect to the executed statements: can all code be reached/executed? • Missing code cannot be detected using white box test techniques (coverage analysis)
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Decision Testing and Coverage* • Instead of statements, decision coverage focuses on the control flow with a program segment (not the nodes, but the edges of a control flow graph) • All edges of the control flow graph have to be covered at least once • Which test cases are necessary to cover each edge of the control flow graph at least once? • Aim of this test (test exit criteria) is to achieve the coverage of a selected percentage of all decisions , called the decision coverage (C 1 code coverage) • Decision coverage (C 1 ) = Number of executed decisions/ Total number of all decisions * 100% Which is synonymous to: • Branch coverage (C 1 ) = Number of covered decisions/ Total number of all branches * 100% * Also referred to as “branch coverage”
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Decision Testing and Coverage – Example 1 • The control flow graph on the right represents the program segment to be inspected • Three different “routes” lead through the graph of this program segment • The first if-statements leads onto two different directions • One path of the first if-statement is divided again in two different paths, one of which holds a loop. • All edges can only be reached via a combination of three possible paths • Three test cases are needed to achieve a decision coverage of 100 % • Only using the two directions on the right, nine out of ten edges can be covered (C 1 -value=90%)
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Decision Testing and Coverage – Example 2 • In this example the graph is slightly more complex • Four different “routes” lead through the graph of this program segment • The first if-statements allows two directions. • In both branches of the if-statement another ifstatement allows again for two different directions • In this example, the loop is not counted as an additional decision • For a 100% decision coverage four test cases are needed • In this example, the same set of test cases is also required for 100% statement coverage!
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Decision Testing and Coverage – General Conclusions • Achieving 100% decision coverage requires at least as many test cases as 100% statement coverage – in most cases more • A 100% decision coverage always includes a 100% statement coverage • In most cases edges are covered multiple times • Drawbacks: • Missing statements cannot be detected • Not sufficient to test complex conditions • Not sufficient to test loops extensively • No consideration of dependencies between loops
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Statement Testing and Coverage and Decision Testing and Coverage • Both methods relate to the paths through the control flow graph • They differ in the amount of test cases necessary to achieve 100% coverage • Only the final result of a condition is considered, although the resulting condition can be made up of several atomic conditions • The condition If((a>2) OR (b<6)) may only be true or false • Which path of the program is executed depends only on the final outcome of the combined condition • Failures due to a wrong implementation of parts of a combined condition may not be detected
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Condition Testing and Coverage • The complexity of a condition that is made up of several atomic conditions is taken into account • An atomic condition cannot be divided further into smaller condition statements • This method aims at finding defects resulting from the implementation of multiple conditions(combined conditions) • Multiple conditions are made up of atomic conditions which are combined using logical operators like OR, AND, XOR, etc. Example: ((a>2)OR(b<6)) • Atomic conditions do not contain logical operators but only relational operators and the NOT-operator (=, >, <, etc. ) • There are three types of condition coverage • Simple condition coverage • Multiple condition coverage • Minimal multiple condition coverage
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Simple Condition Testing and Coverage • Every atomic sub-condition of a combined condition statement has to take at least once the logical values true as well as false Example 2 Consider the following condition: a >0 OR b<6 Test cases for simple condition coverage could be for example: -a=3 (true) b=7(false) a>2 OR b<6 (true) a=1 (false) b=5 (true) a>2 OR b<6 (true) • This example is used to explain condition coverage, using a multiple condition expression. • With only two test cases, a simple condition coverage can be achieved • Each sub-condition has taken on the value true and the value false. • However, the combined result is true in both cases • True OR false = true • False OR true = true
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Multiple Condition Testing and Coverage • All combinations that can be created using permutation of the atomic sub conditions must be part of the tests. • This example is used to explain condition coverage, using a multiple condition expression. Example 2 • With four test cases, the multiple Consider the following condition: condition coverage can be achieved a >0 OR b<6 Test cases for multiple condition coverage could be for • All possible combinations of true example: and false were created. a=3 (true) b=7(false) a>2 OR b<6 (true) • All possible results of the multiple a=3 (true) b=5 (true) a>2 OR b<6 (true) conditions were achieved. a=1 (false) b=5 (true) a>2 OR b<6 (true) • The number of test cases increases a=1 (false) b=7 (false) a>2 OR b<6 (false) exponentially: • n =number of atomic conditions n • 2 =number of test cases
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Minimal Multiple Condition Coverage (defined c. c. ) • All combinations that can be created using the logical results of the subconditions must be part of the test, only if the change of the outcome of one sub -condition changes the result of the combined condition Example 2 Consider the following condition: a >0 OR b<6 Test cases for multiple condition coverage could be for example: a=3 (true) b=7(false) a>2 OR b<6 (true) a=3 (true) b=5 (true) a>2 OR b<6 (true) a=1 (false) b=7 (false) a>2 OR b<6 (false) • This example is used to explain condition coverage, using a multiple condition expression. • For three out of the four test cases the change of a sub-condition changes the overall result • Only for case no. 2 (true OR true =true) the change of a sub-condition will not result in a change of the overall condition. This test case can be omitted! • The number of test cases can be reduced to a value between n+1 and 2 n
• STRUCTURE-BASED OR WHITE-BOX TECHNIQUES All complex decisions must be tested – the minimal multiple condition coverage is a suitable method to achieve this goal.
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Path Testing and Coverage /1 • Path coverage focuses on the execution of all possible paths through a program • A path is a combination of program segments (in a control flow graph: an alternating sequence of nodes and edges) • For decision coverage, a single path through a loop is sufficient. For path coverage they are additional test cases: • One test case not entering the loop • One additional test case of loop executions • This may easily lead to a very high number of test cases
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Path Testing and Coverage /2 • Focus of the coverage analysis is the control flow graph • Statements are nodes • Control flow is represented by the edges • Every path is a unique way from the beginning to the end of the control flow graph • The aim of this test(test exit criteria) is to reach a defined path coverage percentage: Path coverage = Number of covered paths / Total number of paths * 100%
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Path Testing and Coverage – Example 1 • The control flow graph on the right represents the program segment to be inspected. It contains three if-statements • Three different paths leading through the graph of this program segment achieve full decision coverage. • However, five different possible paths may be executed • Five test cases are required to achieve 100% path coverage • Only two are needed for 100% C 0 -three are needed for 100% C 1 -coverage
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Path Testing and Coverage – Example 2 • The control flow graph on the right represents the program segment to be inspected. It contains two if-statements and a loop inside the second if-statement • Three different paths leading through the graph of this program segment achieve full decision coverage. • Four different paths are possible, if the loop is executed twice • Every increment of the loop counter adds a new test case.
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Path Testing and Coverage – General conclusions • 100% path coverage can only be achieved for very simple programs • A single loop can lead to a test case explosion every possible number of loop execution constitutes a new test case • Theoretically an indefinite of paths is possible • Path coverage is much more comprehensive that statement or decision coverage • Every possible path through the program is executed • 100% path coverage includes 100% decisions coverage, which again contains 100% statement coverage
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Exercise: White-box Techniques • Using a control flow graph (part 1) and pseudo code (part 2) the minimum number of test cases has to be determined to achieve 100% coverage of: • Statements • Branches/decisions • Paths
STRUCTURE-BASED OR WHITE-BOX TECHNIQUES • Summary • White box and black box methods are dynamic methods, the test object is executed during test • White box comprise: • Statement coverage • Decision coverage • Path coverage • Condition coverage(single, multiple, minimum multiple) • Only existing code can be tested. If functions are missing this fact cannot be discovered. Dead and superfluous code, however, can be discovered using white box testing. • White box methods are used mainly in lower test levels like component testing or integration testing • The methods differ in their intensity of test (test depth) • Depending on the method, the number of test cases differ
Outlines 1. The test development process 2. Categories of Test Design Techniques 3. Specification-based or Black-box Techniques 4. Structure-based or White-box Techniques 5. Experience-based Techniques 6. Choosing Test Techniques
EXPERIENCE-BASED TECHNIQUES black box Boundary value analysis State transition testing Decision tables Use case based testing ✔ ✔ ✔ experience-based techniques white box dynamic Equivalence partitioning Statement coverage Branch coverage Condition coverage Path coverage Reviews / walkthroughs static • Practice of creating test cases without a clear methodical approach based on the intuition and experience of the tester. • Test cases are based on intuition and experience • Where have errors accumulated in the past? • Where does software often fail? Analytical QA • Definition of Experience-based Techniques Control flow analysis Data flow analysis Compiler metrics/analyzer
EXPERIENCE-BASED TECHNIQUES • Fundamentals • Experience based testing is also called intuitive testing and includes: error guessing (weak point oriented testing) and exploratory testing (iterative testing based on gained knowledge about the system) • Mostly applied in order to complement other, more formally created test cases • Does not meet the criteria for systematical testing • Often produce additional test cases that might not be created with other practices, for example: • Testing a leap year after 2060 (known problems of the past) • Empty sets within input values (a similar application has had errors on this)
EXPERIENCE-BASED TECHNIQUES • Test Case Design • The tester must dispose of applicable experience or knowledge • Intuition – Where can defects be hiding? • Intuition characterizes a good tester • Experience – What defects were encountered where in the past? • Knowledge based on experience • An alternative is to set up a list of recurring defects • Knowledge / Awareness – Where are specific defects expected? • Specific details of the project are incorporated • Where will defects be made due to time pressure and complexity ? • Are inexperienced programmers involved?
EXPERIENCE-BASED TECHNIQUES • Intuitive Test Case Design –possible sources • Test results and practical experience with similar systems • Possibly a predecessor of the software or another system with similar functionality • User experience • Exchange of experience with the system as a user • Focus on deployment • What parts of the system will be used the most? • Development problems • Are there any weak points out of difficulties in the development process?
EXPERIENCE-BASED TECHNIQUES • Error Guessing in practice • Check defect lists • List possible defects • Weight factors depending on risk and probability of occurrence • Test case design • Creating test cases aimed at producing the defects on the list • Prioritizing test cases by their risk value • Update the defect list along testing • Iterative procedure • A structured collection of experience is useful when repeating the procedure in future projects
EXPERIENCE-BASED TECHNIQUES • Exploratory Testing • Test case design procedure especially suitable when the information basis is weakly structured • Also useful when time for testing is scarce • Procedure: • Examine the single parts of the test object • Execute few test cases, exclusively on the parts to be tested, • • applying error guessing Analyze results, develop a rough model of how the test object functions Iteration: Design new test objects applying the knowledge recently acquired Thereby focusing on conspicuous areas and on exploring further characteristics of the test object Capture tools may be useful for logging test activities
EXPERIENCE-BASED TECHNIQUES • Exploratory Testing – Principles • Choose small objects and/or concentrate on particular aspects of a test object • A single iteration should not take more than 2 hours • The results of the iteration form, the information basis of the following iteration • Additional test cases are derived from the particular test situation • Modelling takes place during testing • Test design, test execution, test logging and learning, based on a test charter containing test objectives are concurrent and carried out within time-boxes • Preparing further tests • Herewith, knowledge can be gained to support the appropriate choice of test case design methods
EXPERIENCE-BASED TECHNIQUES • Experience-based versus Specification-based Techniques • Intuitive test case design is a good complement to systematical approaches • It should still be treated as a complementary activity • It cannot give proof of completeness – the number of test cases can vary considerably • Tests are executed in the same way as with systematically defined test cases • The difference is the way in which the test cases were designed/identified • Through intuitive testing defects can be detected that may not be found through systematical testing methods
EXPERIENCE-BASED TECHNIQUES • Summary • Experience based techniques complement systematical techniques to determine test case • They depend strongly individual ability of the tester • Error guessing and Explorative testing are two of the more widely used techniques of experience based testing
Outlines 1. The test development process 2. Categories of Test Design Techniques 3. Specification-based or Black-box Techniques 4. Structure-based or White-box Techniques 5. Experience-based Techniques 6. Choosing Test Techniques
CHOOSING TEST TECHNIQUES • Criteria for choosing the appropriate Test Techniques / 1 • State of information about the test object • Can white-box tests be made at all (source code available)? • Is there sufficient specification material to define black-box tests, or are explorative tests needed to start with? • Predominant test goals • Are functional tests explicitly requested? • Which non-functional test are needed? • Are structural tests needed to attain the test goals? • Risk aspects • Is serious damage expected from hidden defects? • How high is the sequence of usage for the test object? • Are there contractual or legal standards on test execution and test coverage that have to be met?
CHOOSING TEST TECHNIQUES • Criteria for choosing the appropriate Test Techniques / 2 • Project preconditions • • How much time and who is planed for testing? How high is the risk, that testing might not be completed as planed? Which software development method are used? What are the weak points of the project process? • Characteristics of the test object • What possibilities for testing does the test object offer? • What is the availability of the test object? • Contractual and client requirements • Were there any specific agreements made with the client / originator of the project about the test procedures? • What documents are to be handed over at the time of deployment of the system?
CHOOSING TEST TECHNIQUES • Criteria for choosing the appropriate Test Techniques / 3 • Best practice • Which approaches have proven to be appropriate on similar structures? • What experience was gained with which approaches in the past? • Test levels • At which test levels should tests be done? • Further criteria should be applied depending on the specific situation!
CHOOSING TEST TECHNIQUES • Different interests cause different test design approaches • Interest of the project manager: • To create software of ordered quality • meeting time and budget restrictions • Interests of the client / initiator of the project • To receive software of best quality (functionality, reliability, usability, efficiency, portability and maintainability) • Meeting time and budget restrictions • Interests of the test manager: • Sufficient and intensive testing / adequate deployment of the required techniques, from the testing point of view • To assess the quality level that the project has reached • To allocate and use the resources planed for tests in an optimal way
CHOOSING TEST TECHNIQUES • Summary • Testers generally use a combination of test techniques including process, rule and data-driven techniques to ensure adequate coverage of the object under test • Criteria for choosing the appropriate test case design approach: • • • Test basis (Information basis about the test objects) Testing goals (What conclusions are to be reached through testing) Risk aspects Project framework / preconditions Contractual / client requirements
Скачать презентацию Politehnica University of Timisoara Mobile Computing Sensors Network
963ea57029810bfea7b3a681bee79545.ppt